
This page intentionally left blank
Pharmacogenetics and pharmacogenomics are areas of rapidly growing importance at the interface of molecular genetics and psychopharmacology, with implications for drug development and clinical practice. This book provides a conceptual framework for understanding and studying the pharmacogenetics of psychotropic drugs; it reviews advances in the field and describes the findings that have already emerged.
Coverage extends to antipsychotics, antidepressants, mood-stabilizing, cognitive-enhancing, and anxiolytic drugs. Chapters also examine the interface of pharmacogenetics with substance dependence and brain imaging, and consider the impact of pharmacogenetics on the biotechnology and pharmaceutical industries.
This book defines the young field of pharmacogenetics as it applies to psychotropic drugs and is, therefore, an essential reference for all clinicians and researchers working in this field.
Bernard Lerer is Professor of Psychiatry and Director of the Biological Psychiatry Laboratory at Hadassah-Hebrew University Medical Center in Jerusalem. He is Director of the National Institute for Psychobiology in Israel, and Editor-in-Chief of the International Journal of Neuropsychopharmacology.


Pharmacogenetics of Psychotropic Drugs
Edited by
Bernard Lerer
Hadassah-Hebrew University Medical Center, Jerusalem, Israel


The Pitt Building, Trumpington Street, Cambridge, United Kingdom
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia Ruiz de Alarcón 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
© Cambridge University Press 2004
First published in printed format 2002 ISBN 0-511-04173-X eBook (netLibrary)
ISBN 0-521-80617-8 hardback
Every effort has been made in preparing this book to provide accurate and up-to-date information which is in accord with accepted standards and practice at the time of publication. Nevertheless, the authors, editors and publishers can make no warranties that the information contained herein is totally free from error, not least because clinical standards are constantly changing through research and regulation. The authors, editors and publisher therefore disclaim all liability for direct or consequential damages resulting from the use of material contained in this book. Readers are strongly advised to pay careful attention to information provided by the manufacturer of any drugs or equipment that they plan to use.
The publisher has used its best endeavours to ensure that the URLs for external websites referred to in this book are correct and active at the time of going to press. However, the publisher has no responsibility for the websites and can make no guarantee that a site will remain live or that the content is or will remain appropriate.
Part I
1
Part II
2
3
4 5
Part III
6
Contents
List of contributors
Introduction
Genes and psychopharmacology: exploring the interface
Bernard Lerer
Clinical background and research design
From pharmacogenetics to pharmacogenomics of psychotropic drug response
Anil K. Malhotra
Neuropsychopharmacology: the interface between genes and psychiatric nosology
Thomas A. Ban
Methodological issues in psychopharmacogenetics
Sheldon H. Preskorn
Statistical approaches in psychopharmacogenetics
Fabio Macciardi
Molecular background
page viii
3
21
36 57 72
  
The psychopharmacogenetic–neurodevelopmental interface in serotonergic
gene pathways 95 K. Peter Lesch, Jens Benninghoff, and Angelika Schmitt
v
vi Contents
7
Part IV
8 9
Part V
10
11 12 13
14
15 16 17
RNA processing regulation and interindividual variation 127 Colleen M. Niswender and Linda K. Hutchinson
Pharmacokinetics
Pharmacogenetics of psychotropic drug metabolism 157 Vural Ozdemir, Angela D.M. Kashuba, Vincenzo S. Basile, and James L. Kennedy
Pharmacogenetics of chiral psychotropic drugs 181 Pierre Baumann and Chin B. Eap
Specific psychotropic drugs and disorders
Clozapine response and genetic variation in neurotransmitter receptor
targets 217 David A. Collier, Maria J. Arranz, Sarah Osborne, Katherine J. Aitchison, Janet Munro,
Dalu Mancama, and Robert W. Kerwin
Genetic factors underlying drug-induced tardive dyskinesia 245 Ronnen H. Segman and Bernard Lerer
Functional gene-linked polymorphic regions in pharmacogenetics 267 Marco Catalano
Alternative phenotypes and the pharmacogenetics of mood and anxiety
disorders 283 Emanuela Mundo and James L. Kennedy
Pharmacogenetics of anxiolytic drugs and the GABA-benzodiazepine
receptor complex 300 Smita A. Pandit, Spilios V. Argyropoulos, Patrick G. Kehoe, and David J. Nutt
Genetic factors and long-term prophylaxis in bipolar disorder 320 Martin Alda
Genetic influences on responsiveness to anticonvulsant drugs 333 Thomas N. Ferraro
Apolipoprotein E as a marker in the treatment of Alzheimer’s disease 360 Keith Schappert, Pierre Sevigny, and Judes Poirier
 
vii Contents
18
Part VI
19
Part VII
20 21
Genetic variation and drug dependence risk factors 372 Joel Gelernter and Henry Kranzler
Pharmacogenetics and brain imaging
Brain imaging and pharmacogenetics in Alzheimer’s disease and
schizophrenia 391 Steven G. Potkin, James L. Kennedy, and Vincenzo S. Basile
Industry perspectives
Pharmacogenetics in psychotropic drug discovery and development 401 William Z. Potter, AnnCatherine Van Lone, and Larry Altstiel
High-throughput single nucleotide polymorphism genotyping 420 Anne Shalom and Ariel Darvasi
Index 439
 
Contributors
Martin Alda
Department of Psychiatry, Dalhousie University, Abbie J. Lane Building, 5909 Jubilee Rd, Halifax, Nova Scotia, B3H 2E2, Canada
Larry Altsteil
Schering-Plough Research Institute, 2015 Galloping Hill Rd, Kenilworth, NJ 07033- 1300, USA
Spilios V. Argyropoulos
Psychopharmacology Unit, University of Bristol, University Walk, Bristol BS8 1TD, UK
Maria J. Arranz
Institute of Psychiatry, De Crespigy Park, Denmark Hill, London SE5 8AF, UK
Thomas A. Ban
Vanderbilt University, Nashville, TN, USA
Vincenzo S. Basile
Neurogenetics Section, Centre for Addiction and Mental Health, University of Toronto, Toronto, Canada
Pierre Baumann
Department Universitaire de Psychiatrie Adulte, Adulte Hopital de Cery, CH-1008 Prilly, Switzerland
Jens Benninghoff
Department of Psychiatry, University of Wurzburg, Fuchsleinstr. 15, 97080 Wurzburg, Germany
Marco Catalano
IRCCS H. San Raffaele, Department of Neuropsychiatric Sciences, Milan, Italy
David A. Collier
Institute of Psychiatry, De Crespigy Park, Denmark Hill, London SE5 8AF, UK
Ariel Darvasi
The Life Sciences Institute, The Hebrew University of Jerusalem and IDgene Pharmaceuticals, Jerusalem 91904, Israel
Chin B. Eap
Department Universitaire de Psychiatrie Adulte, Adulte Hopital de Cery, CH-1008 Prilly, Switzerland
Thomas N. Ferraro
Center for Neurobiology and Behavior, Department of Psychiatry, University of Pennsylvania School of Medicine, 415 Curie Blvd, Philadelphia, PA 19104, USA
viii
ix List of contributors
Joel Gelernter
Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06516, USA
Linda K. Hutchinson
Department of Pharmacology, Vanderbilt University, Nashville, TN 37232, USA
Angela D.M. Kashuba
School of Pharmacy, University of North Carolina, Chapel Hill, NC, USA
Patrick G. Kehoe
Psychopharmacology Unit, University of Bristol, University Walk, Bristol, BS8 1TD, UK
James L. Kennedy
Neurogenetics Section, Centre for Addiction and Mental Health, University of Toronto, Toronto, Canada
Robert W. Kerwin
Institute of Psychiatry, De Crespigy Park, Denmark Hill, London SE5 8AF, UK
Henry Kranzler
Department of Psychiatry, University of Connecticut, School of Medicine, Farmington, CT, USA
Bernard Lerer
Biological Psychiatry Laboratory, Department of Psychiatry, Hadassah- Hebrew University Medical Center, Ein Karem, Jerusalem 91120, Israel
K. Peter Lesch
Department of Psychiatry, University of Wurzburg, Fuchsleinstr. 15, 97080 Wurzburg, Germany
Fabio Macciardi
Unit of Biostatistics and Genetic Epidemiology, Neurogenetics Section, Clarke Division, Center for Addiction and Mental Health, R-32, 250 College Street, Toronto, Ontario, M5T 1R8, Canada
Anil K. Malhotra
Unit of Molecular Psychiatry, Hillside Hospital, 75–59 263rd St, Glen Oaks, NY 11004, USA
Dalu Mancama
Institute of Psychiatry, De Crespigy Park, Denmark Hill, London SE5 8AF, UK
Emanuela Mundo
Department of Psychiatry, University of Wurzburg, Fuchsleinstr. 15, 97080 Wurzburg, Germany
Janet Munro
Institute of Psychiatry, De Crespigy Park, Denmark Hill, London SE5 8AF, UK
Colleen M. Niswender
Department of Pharmacology, University of Washington, Seattle, WA 98195, USA
David J. Nutt
Psychopharmacology Unit, University of Bristol, University Walk, Bristol, BS8 1TD, UK
Sarah Osborne
Institute of Psychiatry, De Crespigy Park, Denmark Hill, London SE5 8AF, UK
Vural Ozdemir
Department of Pharmacogenomics, Drug Discovery Division, R.W. Johnson Pharmaceutical Research Institute, Route 202 South, #1000 OMP Building, Raritan, NJ 08869, USA
x List of contributors
Smita A. Pandit
Psychopharmacology Unit, University of Bristol, University Walk, Bristol, BS8 1TD, UK
Judes Poirier
Center for Studies in Aging, 6875 Blvd. Lasalle, Verdun, Quebec, H4H 1R3, Canada
Steven G. Potkin
Brain Imaging Center, University of California, Irvine Hall, Irvine, CA 92697–3960, USA
William Z. Potter
Lilly Research Laboratory, Lilly Corporate Center, Drop Code 0532, Indianapolis, IA 46285, USA
Sheldon H. Preskorn
Psychiatric Research Institute, University of Kansas, 1100 N. St Francis, Wichita, KS 67214, USA
Keith Schappert
Mirador DNA Design, Suite 501, 404 McGill St, Montreal, Quebec H2Y 2G1, Canada
Angelina Schmitt
Department of Psychiatry, University of Wurzburg, Fuchsleinstr. 15, 97080 Wurzburg, Germany
Ronnen H. Segman
Biological Psychiatry Laboratory, Department of Psychiatry, Hadassah- Hebrew University Medical Center, Ein Karem, Jerusalem 91120, Israel
Pierre Savigny
Mirador DNA Design, Suite 501, 404 McGill St, Montreal, Quebec H2Y 2G1, Canada
Anne Shalom
The Life Sciences Institute, The Hebrew University of Jerusalem, Israel
AnnCatherine Van Lone
Lilly Research Laboratory, Lilly Corporate Center, Drop Code 0532, Indianapolis, IA 46285, USA
Part I
Introduction


1
Genes and psychopharmacology: exploring the interface
Bernard Lerer
Biological Psychiatry Laboratory, Department of Psychiatry, Hadassah–Hebrew University Medical Center, Jerusalem, Israel
OVERVIEW
Pharmacogenetics is the study of genetically determined, interindividual differences in ther- apeutic response to drugs and susceptibility to adverse effects. The principal objective of pharmacogenetics is to identify and categorize the genetic factors that underlie these differ- ences and to apply these observations in the clinic. Individualization of drug treatment to the specific patient is thus a core objective of pharmacogenetics. The goal of this book is to provide a basic conceptual framework for the pharmacogenetics of psychotropic drugs, to address major issues in the design and implementation of studies that seek to advance the field and to provide an overview of findings that have emerged so far. In this introductory chapter, the rationale for psychopharmacogenetics is considered, a brief historical perspect- ive is provided, some of the pivotal concepts and terms are defined, important issues in the design and interpretation of pharmacogenetic studies in psychiatry are considered and opti- mistic predictions for the future are evaluated. The chapter concludes with a brief overview introducing the reader to the various sections of the book.
Introduction
For as long as medicine has been practiced, physicians have known that patients respond differently to the therapeutic agents that they are administered, even though there are no obvious differences in the nature or severity of their illnesses. Therefore, individual or illness characteristics that might aid the physician in choosing an appropriate treatment have long been sought. The principal objective of pharmacogenetics is to identify and categorize the genetic factors that underlie differences among individuals in their response to drugs and to apply these obser- vations in the clinic.
The pharmacological treatment of psychiatric disorders has made rapid progress since the 1950s. Psychotropic drugs are among the most widely used pharmacolog- ical agents worldwide and their number has increased exponentially. These devel- opments have occurred in spite of the highly complex clinical characteristics of

3

4 B. Lerer
psychiatric disorders and the absence of biological anchors for their diagnosis. Both these limitations are a consequence of the fact that the pathophysiological basis of most psychiatric disorders has not yet been defined. A role for genetic factors in the pathogenesis of many of the major psychiatric syndromes is well established. The advent of modern molecular techniques has led to an intensive search for suscep- tibility genes, efforts that have yielded intriguing leads but no definitive findings.
This book is positioned at the rapidly developing interface between molecular genetics and psychopharmacology. This area is closely related to the search for sus- ceptibility genes but is also separate from it since genes that affect the therapeutic and adverse effects of the drugs used to treat an illness need not be involved in the pathogenesis of the illness. It is an exciting and increasingly productive interface that holds considerable promise. At the same time, the difficulties and complexities are clearly apparent and will grow more evident as optimistic predictions are submitted to empirical testing. The field holds exceptional fascination and it is still a wide-open frontier in that some of the most fundamental studies have yet to be conducted.
The area of basic concepts, experimental approaches and current findings is too large to be comprehensively addressed in a single volume. Nevertheless, there is a very great need to provide the researcher and clinician with an overview of the pivotal topics and to do so in an integrative way. This chapter is intended to serve as a general introduction to psychopharmacogenetics and to the topics covered in this book.
The rationale for psychopharmacogenetics
Whether medicine is an art or a science may seem an outdated, if not naive, debate at the dawn of the twenty-first century. This impression is bolstered by the exten- sive basis of modern medicine in biomedical science. Powerful diagnostic technol- ogy has immeasurably increased the precision of diagnosis, and evidence-based prescription is rapidly becoming a hallowed cornerstone of therapeutics. Yet, one needs to look no further than pharmacotherapy in order to realize how tenuous the scientific roots of our discipline still are. This is particularly true of the pharmaco- therapy of psychiatric disorders but is by no means limited to this field. The psychi- atrist who initiates drug treatment of a depressed patient is faced with a bewildering set of choices – at least five classes of drug if one applies current definitions, includ- ing tricyclic antidepressants, specific serotonin reuptake inhibitors, monoamine oxidase inhibitors and newer agents that specifically inhibit norepinephrine uptake or enhance synaptic availability of both norepinephrine and serotonin (5- hydroxytryptamine (5-HT)) by mechanisms other than reuptake. Within these classes there are subclasses and within each subclass individual drugs that differ from each other in the intensity or specificity of their pharmacological effects. The

5 Genes and psychopharmacology
psychiatrist has no definitive way of knowing which patient will respond to which drug. There is very little hard evidence on which to base a rational choice, certainly not in the area of therapeutic efficacy. Usually the choice is made on the basis of adverse effect profiles or clinical experience.
The situation described is by no means limited to psychopharmacology. The neurologist treating epilepsy and the internist or pediatrician treating asthma (not- withstanding recent advances in the pharmacogenetics of this disorder) are in the similar position of having to make educated guesses regarding the choice of medi- cation on the basis of a far from comprehensive set of evidence-based criteria. This situation is an inevitable consequence of the fact that modern, evidence-based ther- apeutics is by definition group and not individual oriented. Particularly in psychi- atry, large numbers of patients are required to demonstrate unequivocally a significant drug–placebo difference, and drug–placebo differences are the corner- stone of evidence-based medicine. In cases where a drug is effective in two thirds of subjects (versus one third of subjects for placebo), efficacy may be unequivocally demonstrated, but a less appreciated element of the message is that one in three patients will not respond. Moreover, among responders to the active drug, at least one third might be placebo responders.
It is envisaged that pharmacogenetic predictors of response to drugs and of adverse effects will ultimately serve as the basis for simple diagnostic tests that can be applied in the clinic in order to prescribe the appropriate drug to the appropri- ate patient. In the current era of managed care in medicine and limited resources, this will be a development of major economic importance. As discussed in this chapter, the obstacles on the way to this objective are formidable but not insur- mountable. Another very important application of pharmacogenetic screening will be in clinical trials. Knowledge of genetic predictors of response and/or adverse effects, even if this is not at a level of resolution that permits applicability in regular clinical practice, will permit stratification of patients in clinical trials. This will sub- stantially reduce the cost of drug development and shorten the lengthy lag period that currently elapses until a drug is introduced into the clinic.
With the publication of the draft sequence of the human genome, it has become accepted practice to refer to the current era as “postgenomic.” In this postgenomic era, pharmacogenetics, a discipline with a long and distinguished “pregenomic” history, has come of age. Powerful tools have been placed at its disposal and opti- mism abounds as to its anticipated impact on pharmacotherapy.
Historical perspective
Nebert (1997) suggests that Pythagoras was the first to recognize the basic princi- ple of pharmacogenetics when, around 510 , he noted the predisposition of

6 B. Lerer
some individuals but not others to develop an adverse reaction (hemolytic anemia owing to glucose 6-phosphate dehydrogenase deficiency) after consumption of the fava bean. A century ago, the English physician Archibald Garrod suggested that genetic factors directed chemical transformations in humans. His ideas on the hereditary basis of chemical individuality are extensively discussed in his writings (Garrod, 1902, 1909). A seminal study was that of Snyder (1932), which defined the phenylthiourea nontasting phenomenon as an autosomal recessive trait. According to Propping and Nothen (1995), the relationship between adverse drug reactions and genetically determined variation was first demonstrated by Motulsky (1957). The term “pharmacogenetics” was coined by Vogel in 1959. Another important development was the observation that the antitubercular drug isoniazid could be slowly or rapidly acetylated and that this was under genetic control (Evans et al., 1960). A further milestone was the demonstration by Kalow of an abnormal form of serum cholinesterase that leads to catastrophic adverse reactions to succinylcholine. Kalow also wrote the first systematic account of phar- macogenetics (Kalow, 1962). Polymorphism of the P-450 enzyme now termed CYP2D6 was first observed in the 1970s in healthy volunteers who developed adverse effects when taking the antihypertensive agent debrisoquine. The enzyme was initially named debrisoquine hydroxylase but it was subsequently shown that oxidation of sparteine is by the same enzyme (Mahgoub et al., 1977; Eichelbaum et al., 1979).
For much of its history, the focus of pharmacogenetics has been on drug- metabolizing enzymes. This was because of the availability of techniques to detect phenotypic differences between individuals in the plasma level of drugs and to study their genetic basis. The focus on pharmacodynamic variation is more recent and was given considerable impetus by the advent of techniques to determine the sequence of genes and identify variations.
Definition of terms
Table 1.1 summarizes many of the terms that will be used frequently throughout this book. While pharmacogenetics is defined as the study of genetically determined interindividual differences in response to drugs, pharmacogenomics refers to the use of genome-based technologies in drug development. The fields are closely related and the terms are often used interchangeably. Nevertheless, it is useful to maintain the distinction because of the different starting points and outcomes. Pharmacogenetics is individual based and its body of information is derived from the relationship between drug effects and genetic predisposition in patients or volunteer subjects. Its outcome is rational drug choice in the treatment of patients, based on the evidence that has accumulated. The starting point of

7 Genes and psychopharmacology
Table 1.1. A pharmacogenetics glossary

Term
Pharmacogenetics
Definition
The study of genetically determined interindividual differences in response to drugs
The use of genome-based technologies in drug development
Genetically based differences that influence bioavailability of a drug
Genetically based differences in the proteins at which a drug acts Genetic variation that occurs with a frequency of 1% or more in the population
Differences between individuals in a single base of the genomic sequence; these are the most frequently occurring genetic variation Occur in exons (coding regions) of genes
Occur in introns (noncoding regions) of genes
Occur in 3 or 5 regulatory regions (promoter)
SNPs in coding regions that do not influence the structure of the protein
Alter the structure of the protein but not its function
Alter the function of the protein
Chosen for analysis on the basis of an a priori hypothesis regarding the role of the protein coded by the gene in the phenotype under study
Statistical demonstration of a greater than chance occurrence of a polymorphism in a candidate gene in conjunction with the target phenotype
Statistical association of two alleles at a rate greater than would be predicted by chance (owing to the fact that the two alleles are close enough to each other to not undergo recombination)
A trait that is influenced by a number of different genes each of which contributes a portion of the effect
Genetic variance owing to nonadditive effects of alleles at distinct loci
Both genetic and environmental factors contribute substantially and variably to the phenotype

Pharmacogenomics
Types of pharmacogenetic effects
Pharmacokinetic
Pharmacodynamic Polymorphism
Single nucleotide polymorphisms (SNPs)
Coding Noncoding Regulatory Synonymous
Conservative Functional Candidate gene
Association
Linkage disequilibrium (LD)
Polygenic Epistasis Multifactorial


8 B. Lerer
pharmacogenomics is the human genome sequence and its outcome is the devel- opment of new pharmacological agents. The points of interaction between the two approaches are multiple and they complement each other at many levels, so it is inevitable that the distinction is easily blurred.
There are two broad categories within pharmacogenetics, which derive from the fact that differences between individuals can be attributed to two major factors that are under genetic influence. The first set of factors is pharmacokinetic and they encompass genetically based differences in processes that influence bioavailability of a drug, i.e., the concentration of the drug and its active metabolites that is avail- able at the site of action). The second major category is pharmacodynamics and refers to genetically based differences in the proteins at which the drug acts. Both sets of factors may influence the response of the individual to a given drug and they may interact within the same individual.
Polymorphism is a core term in pharmacogenetics. A polymorphism is a genetic variation that occurs with a frequency of 1% or more in the population. Genetic variations that occur more rarely (mutations) also influence drug response, often dramatically, but these are of less importance on a population-wide basis. Single nucleotide polymorphisms (SNPs) are differences between individuals in a single base of the genomic sequence and are the most frequently occurring genetic vari- ation. SNPs occur throughout the human genome at a density of approximately 1 per 1000 bases (kilobases, kb) of DNA. It is important to stress that the vast majority of SNPs are unlikely to influence either the structure or function of pro- teins. SNPs may be classified by their location, occurring in exons (coding regions) or introns (noncoding regions) of genes and in regulatory regions such as the promoter. SNPs in coding regions need not necessarily influence the struc- ture of the protein and are termed synonymous. SNPs that do alter the structure of the proteins need not have functional consequences and are termed conserva- tive or non-functional. SNPs located in intronic regions can have an impact on the coded protein by influencing splicing (Krawezak et al., 1992). SNPs in regula- tory regions are a focus of considerable interest and can have major effects on expression of the gene. In fact, it has been suggested that regulatory mutations rather than mutations that affect protein structure may be the prime cause of bio- logical differences in humans (Chakravarti, 1999; Chapter 12). Major efforts are now in progress by public and industry-based consortia to generate SNP data- bases that will be an invaluable resource for pharmacogenetics in the very near future.
SNPs are not the only type of DNA variation of relevance to pharmacogenetics. Other types of genetic polymorphism result from the insertion or deletion of a few nucleotides (termed insertion/deletion polymorphisms) and variation in the number of times a sequence is repeated. Variable number tandem repeats (VNTRs or
9 Genes and psychopharmacology
minisatellites) have several hundred base pairs repeated while microsatellites (or simple tandem repeats, STRs) have two to four nucleotides repeated a variable number of times. Allelic variations (i.e., variability in the number of repeats) can be considerable and this type of variation is consequently highly polymorphic.
Traditionally, pharmacogenetic studies have sought an association between a specific gene and the response or adverse effect phenotype under study. This is a classical candidate gene approach, which is based on an a priori hypothesis regard- ing the role of the protein coded by the gene in the particular phenotype. However, an SNP or other genetic variation may be statistically associated with a phenotype without having a direct effect. This phenomenon results from linkage disequilib- rium (LD) and it arises from the fact that the variant examined is close enough to the true predisposing variant that it does not undergo recombination during meiosis and is inherited with it. The marker may be at a different site in the gene itself or may be located outside the gene. Linkage disequilibrium mapping employs dense SNP maps in order to localize genes associated with a phenotype and has been suggested as a powerful approach to identify genes for complex phenotypes (Risch, 2000), although there are dissenting voices that question the feasibility of the strategy (Weiss and Terwilliger, 2000). Linkage disequilibrium approaches are the cornerstone of the large-scale mapping projects in pharmacogenetics that are currently being advocated.
Classical genetics of human disease deals with monogenic disorders in which a single mutation in a single gene is causatively related to the phenotype. This para- digm holds reasonably true in pharmacogenetics for classical “pharmacokinetic” polymorphisms that have a major effect on drug bioavailability (such as the effect of CYP2D6 polymorphism on the metabolism of a variety of psychotropic and other drugs, which is inherited as an autosomal recessive trait). For the most part, however, the accepted view is that pharmacogenetic traits are likely to be polygenic and multifactorial. A polygenic trait is one that is influenced by a number of differ- ent genes each of which contributes a portion of the effect and may do so additively as well as interactively (epistasis). The term “multifactorial” indicates that both genetic and environmental factors contribute substantially and variably to the phe- notype.
Core issues in pharmacogenetic research design
The classical experimental context for determining pharmacogenetic influences on drug response is a comparison of responders and nonresponders to the drug or of individuals who develop adverse effects and those who do not. This appealingly simple case-control design should be readily applicable in psychopharmacology. Data from such studies are amenable to analysis by two approaches. The first is a

10 B. Lerer
categorical approach in which patients are grouped according to the phenotype (responder or nonresponder, develops adverse effect or does not) and the frequency of the genotype of interest in these groups is compared. The second approach uti- lizes the response variable (or adverse effect measure) in a continuous fashion and compares scores at a single time point or over a period of treatment in patients grouped according to genotype. There are already numerous examples of the appli- cation of this approach to psychopharmacology. The paucity of replicable results may be a consequence of a number of relatively elementary factors.
Population effects
It is well known that the frequency of genetic polymorphisms differs markedly among ethnic groups. Therefore, unequal inclusion of individuals of different ethnic backgrounds in groups of subjects being compared can lead to spurious results that reflect ethnic stratification. Methods have been proposed to address these issues (Devlin and Roeder, 1999; Lerer et al., 2001). Nevertheless, it is essen- tial that studies be designed to take them into account prospectively. Another point is that the functional relevance of a particular polymorphism may vary among ethnic groups. Thus, a true finding in a sample from a particular population may not be replicable in a different population.
Demographic variables
Since the basis of interindividual variability in drug response is multifactorial, the impact of demographic variables such as age and gender cannot be ignored. A par- ticular genetic variant may not be functionally relevant without the addition of other factors that also influence the phenotype. An example is the influence of certain 5-HT receptor variants on predisposition to neuroleptic-induced tardive dyskinesia, which was demonstrated in a sample of older patients but was not observed in a sample two decades younger (Segman and Lerer, 2002). It is highly conceivable that gender-specific effects are also operative and that interactions between age and gender will be demonstrated.
Definition and evaluation of the phenotype
Drug response to psychotropic drugs is a phenotype that is very difficult to define for evaluation experimentally. There is an extensive literature that debates how to define a “responder” to an antidepressant drug in the context of a clinical trial. Definitions may be based on percentage improvement on a particular rating scale or by using a specific score on that scale as a threshold. There are conventions that are fairly well accepted for clinical trials of psychopharmacologic agents. It remains to be established whether these conventions can be readily transferred to pharma- cogenetic studies (see Rietschel et al., 1999).
11 Genes and psychopharmacology
Study design
Most recently published studies on the pharmacogenetics of psychotropic drugs are “opportunistic” in that they employ subjects who were previously studied and corre- late the phenotypic measures gathered in these studies with genotypic data that are obtained currently. Alternatively, they may be “add-on” in nature. In studies of this type, drawing of blood samples for DNA is included in the protocol of clinical trials for future correlation of genotypes with drug response or adverse effect phenotypes. In both cases, studies are not “purpose designed” to address pharmacogenetic ques- tions. They address the questions that they were designed to answer and do not take into account issues that are pivotal to pharmacogenetics. Since clinical trials demand large sample sizes, there is usually little consideration given to ethnicity of the sub- jects. The age range may be great and diagnostic boundaries tend to be as wide as pos- sible to insure maximum recruitment. It is likely that even very large samples will not have adequate power to address pharmacogenetic questions when subject groups are stratified in order to control for factors that can spuriously affect the results. A “second generation” of pharmacogenetic studies can be anticipated that are designed to take these issues into account. When this occurs, the issue of placebo control is likely to emerge as a major consideration. On the one hand, it is self-evident that non- placebo-controlled studies greatly impede attempts to differentiate factors associated with “response” from these associated with “response to the active drug.” On the other hand, already existing concerns regarding the ethical validity of placebo-controlled trial in disorders such as schizophrenia or depression will be more intense when the aim of the study is refinement of drug prescription rather than demonstration of effi- cacy. Simulation studies are urgently needed in order to address these issues.
Interaction among multiple loci
Traditionally, pharmacogenetic studies have focused on single genes and on single polymorphisms within these genes. It is becoming clear that this approach is too limited to address the complexity of the situation adequately. Single SNPs may not show an association with treatment effects or with disease susceptibility while com- binations do. These may be within a single gene, as demonstrated for the effect of complex haplotypes of SNPs in the coding region and the promoter of the 2- adrenoreceptor gene on the bronchodilator response to asthma therapy (Drysdale et al., 2000). Interactions (epistasis) may be demonstrated between SNPs in differ- ent genes, as recently observed in a study of genetic susceptibility to sporadic breast cancer (Ritchie et al., 2001). In the context of pharmacogenetics, Segman et al. (2002) have observed an interaction between a polymorphism in the gene for cyto- chrome P17 and the gene for dopamine D3 receptor that is associated with more neu- roleptic-related abnormal involuntary movements in patients who carry both mutant genotypes.
12 B. Lerer
Gene expression
It is not inevitable that the genotype carried by an individual will be expressed func- tionally as anticipated from the structure of the protein predicted by the DNA sequence. This assumption is inherent in pharmacogenetic studies that seek to cor- relate DNA variations with a particular phenotype. The reasons for this are numer- ous and may be found at any point along the long route traversing transcription, translation and post-translational modification between the DNA sequence of a gene and the protein for which it codes. Examples considered in Chapter 7 of this volume are the processes of RNA editing and RNA splicing. Both represent impor- tant mechanisms that ultimately contribute to the expression of specific protein isoforms within a given cell. These considerations are of major importance in phar- macogenetics and demand that the functional implications of a specific DNA variant be assessed before it is assumed that association of a trait with the variant can serve as a predictor of response or of adverse effects to a drug.
An age of optimism
In spite of the methodological concerns outlined in the preceding section, there are very substantial reasons for optimism regarding the potential of pharmacogenetics to fulfill some of the expectations that have been assigned to it, although the time frame could well be longer than anticipated.
The identification of DNA variations that serve as a basis for pharmacogenetic studies is proceeding apace. A number of international consortia supported by government, industry, and academic resources are devoting major efforts to iden- tifying and cataloging SNPs throughout the human genome. Much of the informa- tion that is being amassed is being placed in the public domain and will thus be available to researchers worldwide. This is an invaluable resource. Once the human genome is covered by a dense SNP map, it will be possible to proceed with large- scale studies that consider thousands of potential markers.
Efforts in the laboratory are being matched by parallel developments in bioin- formatics. The enormous body of genetic information that is being created is of little value unless appropriate tools are available for the researcher to access and use this information. Private as well as public efforts are directed at making this pos- sible, and the accessibility of the information can be expected to increase with time. There are also major challenges in data analysis that need to be addressed. Current approaches are inadequate to analyze vast amounts of genotypic information except in the context of sample sizes that are completely unrealistic. Recent reports suggest that novel statistical techniques may allow a number of genetic loci to be considered simultaneously in the context of sample sizes that are within the realm of clinical reality (Ritchie et al., 2001). Another very real problem that has to be

13 Genes and psychopharmacology
adequately addressed is the potential to generate false-positive results from a huge number of repeated tests. Currently applied corrections for this problem run the risk of excluding positive results, while less conservative approaches can lead to the accumulation of nonreplicable chance findings.
To be applicable in the clinic, pharmacogenetic findings will need to cross two thresholds. The first is a replicability threshold that establishes their validity as prognostic markers. The second is that enough of the variance in treatment response must be accounted for by a particular marker (an unlikely scenario) or by a combination of markers to render a test based upon these markers clinically useful. It is clear that this process will take considerably longer than developing the technological means to perform such tests in the clinic. Chip technologies are already available that allow up to several thousand SNPs to be typed at one time. The cost of chips of this type is obviously still very high. It can be anticipated that the costs will drop long before psychopharmacogenetics is able to fill the available slots in a rational way.
In the area of psychopharmacology, there are a few recent pharmacogenetic find- ings that have been reasonably well replicated, although not by all groups. One example is the association of an insertion–deletion polymorphism in the serotonin transporter promoter and response to antidepressant treatment with serotonin reuptake blockers (see Chapter 12). A second example is the association of a serine to glycine polymorphism in the dopamine D3 receptor gene with susceptibility to tardive dyskinesia, a serious adverse effect of classical antipsychotic drugs (Chapter 11). Outside of psychopharmacology, asthma therapy has been a very active research focus in pharmacogenetics, with a number of encouraging results reported (Silverman et al., 2001). In patients with coronary atherosclerosis, it was found that a functional polymorphism in the cholesterol ester transfer protein (CETP), which influenced CETP plasma levels, was associated with a stronger ability of the drug pravastatin to lower plasma cholesterol levels (Kuivenhoven et al., 1998).
Pharmacogenetics of psychotropic drugs: an overview
Parts II–IV of the book deal with a selection of important background topics. Chapter 2 discusses the clinical implications of identifying molecular genetic pre- dictors of drug response and basic design issues in studies that are designed to gen- erate data of this type. Chapter 3 reviews the history of psychiatric nosology and its interactions with psychopharmacology and proposes the use of an empirically derived pharmacologically meaningful classification of mental illness in genetic research. Chapter 4 reviews methodological issues facing pharmacogenetic studies in clinical psychopharmacology, considering the drug development process itself,

14 B. Lerer
the problems posed by poor signal:noise ratio, and practical problems posed in trying to study whether a genetically defined population is at increased risk for tox- icity or other adverse effects at routine drug dosages. Chapter 5 considers optimal approaches to the statistical analysis of pharmacogenetic data with reference to the specific challenges posed and their implications for the design of studies.
Background issues at the molecular level are addressed in Part III. Chapter 6 focuses on serotonergic gene pathways and discusses the potential importance for pharmacogenetics of variation in the structure and expression of genes that play a key role in neurodevelopment of these pathways. Chapter 7 reviews the post- transcriptional processes of RNA editing and alternative splicing and considers how processes such as these, beyond the level of structural DNA variation, can be important in psychopharmacogenetics.
Historically and practically, heritable effects on drug metabolism are of major importance in the pharmacogenetics of psychotropic drugs. Section IV deals with this topic. Chapter 8 reviews and contrasts genetic polymorphisms of the metabolic enzymes CYP2D6 and CYP1A2 and their roles in the pharmacokinetics of psycho- tropic drugs. This chapter also considers gene–gene and gene–environment inter- actions. Chapter 9 focuses on the stereoselectivity of cytochrome P-450 isoenzymes towards chiral substrates and its importance in view of the fact that many psycho- tropic drugs have one or more chiral centers.
Part V of the book focus on specific drug groups and syndromes. Antipsychotic drugs, particularly the atypical drug clozapine, have been fairly extensively studied. In Chapter 10, Collier and colleagues discuss the impact of genetic variation in a number of target receptors on the therapeutic efficacy of clozapine, in the light of their own studies and those other investigators. Chapter 11 covers the genetic predisposition to the movement disorder tardive dyskinesia, the most severe adverse effect of classical antipsychotics, and reviews the possible role of a number of candidate genes.
Chapters 12–14 consider the antidepressants and anxiolytics. Chapter 12 high– lights the potentially important role of genetic polymorphisms that are involved in the regulation of gene expression or the activity of key enzymes, as opposed to the structure of receptor sites, in the pharmacogenetics of a variety of antidepressant treatments. In most pharmacogenetic studies of psychotropic drugs, the target phe- notype is very simply defined on a response/nonresponse continuum. Mundo and Kennedy in Chapter 15 consider the importance of more homogeneous phenotype definition in studies of antidepressant pharmacogenetics and also discuss and review data concerning the role of novel alternative phenotypes such as antidepressant-induced mania. Chapter 14 focuses on the treatment of anxiety dis- orders and reviews genetic variations in the gamma-aminobutyric acid (GABA) type A receptor and its subunits and their implications for antianxiety treatments.
15 Genes and psychopharmacology
Chapter 15 considers the pharmacogenetics of mood-stabilizing drugs such as lithium and anticonvulsants. Alda reviews studies that categorize families segregat- ing bipolar disorder according to the responsiveness of their affected members to lithium treatment. These samples are being used to identify genes associated with treatment response or to map genes for bipolar disorder in families homogeneous for treatment response. Besides lithium, anticonvulsant drugs now have an impor- tant place in the treatment of mood disorders. Chapter 16 reviews the pharmaco- genetics of anticonvulsants, including extensive work on the polymorphism of cytochrome P-450 isoforms and their influence on the metabolism of these drugs and more limited work on the genes that encode their protein targets.
Chapter 17 focuses on Alzheimer’s disease, specifically on the possible role of apolipoprotein E (apoE) genotype as a predictor of response to cholinomimetic treatment. Schappert and colleagues discuss findings in this regard and also a pos- sible role for the butyrylcholinesterase gene in modifying treatment outcome with noncholinergic therapies. Chapter 18 covers the topic of substance abuse and dependence, disorders that involve the ingestion of an exogenous substance that can be expected to interact with pharmacogenetic factors possibly specific for the substance. Examples of such factors are discussed in the broader context of genetic risk factors for substance dependence and the development of pharmacological treatment. Chapter 19 brings brain imaging into the domain of pharmacogenetics. Potkin and colleagues discuss data regarding the regional brain metabolic pheno- typic manifestation of specific genotypes (apoE in Alzheimer’s disease and the dop- amine D3 receptor gene in tardive dyskinesia) and the possible use of brain imaging in pharmacogenetic evaluation.
The final part of the book (Part VII) provides perspectives from the phar- maceutical and biotechnology industries. Chapter 20 describes the major benefits to be derived from the implementation of pharmacogenetically oriented research protocols in industry-supported studies; the authors discuss the problems and lim- itations that have prevented such implementation so far and consider how such initiatives may be developed in the context of collaboration between industry and academia. High throughput genotyping is a key element of large-scale pharmaco- genetic analysis. Chapter 21 reviews a number of technologies that have been devel- oped, considering both their effectiveness and their cost.
Conclusions
The interface between psychopharmacology and molecular genetics is already a focus of considerable research activity and this is likely to intensify as antici- pated technological advances provide the tools for even more sophisticated studies. Pharmacogenetics addresses a core issue in pharmacotherapeutics, the

16 B. Lerer
individualization of drug treatment to the specific patient, and promises to provide the tools for making rational clinical decisions that are based on the patient’s genetic profile. This will be a major advance in therapeutics and will have enormous impact on patient care and also important pharmacoeconomic implications. Furthermore, the complex and lengthy process of new drug development could be considerably shortened, with cost reductions that would be passed on to the consumer. Translating this optimistic scenario into reality is likely to take substantially longer than anticipated by some and will require considerable investment of resources in the design and execution of appropriate clinical studies as well as the development of novel and considerably more efficient approaches to data analysis. We are posi- tioned at the beginning of a crucial and highly intriguing era; there will undoubt- edly be high points, but disappointments and some degree of disillusionment are inevitable given the considerable level of current expectations. It may take longer than originally thought, but ultimately pharmacogenetics and pharmacogenomics will revolutionize the field of clinical psychopharmacology and the development of psychotropic drugs.
REFERENCES
Chakravarti A. (1999). Population genetics – making sense out of sequence. Nat Genet 21, 56–60. Devlin B, Roeder K. (1999). Genomic control for association studies. Biometrics 55,
997–1004.
Drysdale CM, McGraw DW, Stack CB et al. (2000). Complex promoter and coding region 2-
adrenergic receptor haplotypes alter receptor expression and predict in vivo responsiveness.
Proc Natl Acad Sci USA 97: 10483–10488.
Eichelbaum M, Spannbrucher N, Steincke B, Dengler HJ (1979). Defective N-oxidation of spar-
teine and debrisoquine in man: a new pharmacogenetic defect. Eur J Clin Pharmacol 16,
183–187.
Evans DAP, Manley KA, McKusick VA (1960). Genetic control of isoniazid metabolism in man.
Br Med J 2, 485–491.
Garrod AE (1902). The incidence of alcaptonuria: a study in chemical individuality. Lancet ii,
1616–1620.
Garrod AE (1909). Inborn Errors of Metabolism. Oxford: Oxford University Press.
Kalow W (1962). Pharmacogenetics – Heredity and Response to Drugs. Philadelphia: Saunders. Krawezak M, Reiss J, Cooper DN (1992). The mutational spectrum of single base-pair substitu-
tions in mRNA splice junctions of human genes: causes and consequences. Hum Genet 90,
41–54.
Kuivenhoven JA, Jukema JW, Zwinderman AH et al. (1998). The role of a common variant of the
cholesteryl ester transfer protein gene in the progression of coronary atherosclerosis. The Regression Growth Evaluation Statin Study Group. N Engl J Med 338, 86–93.

17 Genes and psychopharmacology
Lerer B, Macciardi F, Segman RH et al. (2001). Variability of 5-HT2C receptor Cys23Ser polymor- phism among European populations and vulnerability to affective disorder. Mol Psychiatry 6, 579–585.
Mahgoub A, Idle JR, Dring LG, Lancaster R, Smith RL (1977). Polymorphic hydroxylation of debrisoquine in man. Lancet ii, 584–586.
Motulsky A (1957). Drug reactions, enzymes and biochemical genetics. J Am Med Assoc 165, 835–837.
Nebert DW (1997). Pharmacogenetics: 65 candles on the cake. Pharmacogenetics 7, 435–440. Propping P, Nothen MM (1995) Genetic variation of CNS receptors – a new perspective for phar-
macogenetics. Pharmacogenetics 5, 318–325.
Rietschel M, Kennedy JL, Macciardi F, Meltzer HY (1999). Application of pharmacogenetics to
psychotic disorders: the first consensus conference. The Consensus Group for Outcome
Measures in Psychoses for Pharmacological Studies. Schizophr Res 37, 191–196.
Risch NJ (2000). Searching for genetic determinants in the new millennium. Nature 405,
847–856.
Ritchie MD, Hahn LW, Roodi N et al. (2001). Multifactor-dimensionality reduction reveals high-
order interactions among estrogen-metabolism genes in sporadic breast cancer Am J Hum
Genet 69, 138–147.
Segman RH, Lerer B (2002). Age and the relationship of dopamine, serotonin 2C and serotonin
2A receptor genes to abnormal involuntary movements in chronic schizophrenia. Mol
Psychiatry 7, 137–139.
Segman RH, Heresco-Levy U, Yakir A et al. (2002). Interactive effect of cytochrome P-450 17-
hydroxylase and dopamine D3 receptor gene polymorphisms on abnormal involuntary move-
ments in chronic schizophrenia. Biol Psychiatry 51, 261–263.
Silverman ES, Ligget SB, Gelfand EW et al. (2001). The pharmacogenetics of asthma: a candidate
gene approach. Pharmacogenom J 1, 27–37.
Snyder LH (1932). Studies in human inheritance 1x. The inheritance of taste deficiency in man.
Ohio J Sci 32, 436–468.
Vogel F (1959). Moderne probleme der humangenetik. Ergebn Inn Med Kinderheilk 12, 52–125. Weiss KM, Terwilliger JD (2000). How many diseases does it take to map a gene with SNPs? Nat
Genet 26, 151–157.

Part II
Clinical background and research design


2
From pharmacogenetics to pharmacogenomics of psychotropic drug response
Anil K. Malhotra
Hillside Hospital, Glen Oaks, New York
OVERVIEW
The marked interindividual variation in response to psychotropic drugs creates a clinical dilemma that can only be resolved through often lengthy empirical drug trials. Recent devel- opments in molecular biology have provided an opportunity to identify molecular genetic predictors of drug response, which may have immediate clinical implications as well as provide critical data for future drug development. In this chapter, we will discuss the evi- dence for a heritable component to psychotropic drug response, review basic methodolog- ical issues in pharmacogenetic studies of psychotropic drug response, and then highlight future directions for pharmacogenetics research in psychiatry.
Introduction
Individual differences in clinical response to psychotropic drugs has long been rec- ognized as a fundamental problem in the treatment of the seriously mentally ill patient. This variability in individual response ranges from patients who experi- ence complete symptom remission to a subset of patients often described as “treat- ment refractory,” as well as a marked variability in susceptibility to adverse drug effects. A priori identification of the patients who will respond well to a particular psychotropic drug, or be at a higher risk for development of adverse side effects, has the potential to help clinicians to avoid lengthy ineffective medication trials and to limit a patient’s exposure to drug side effects. Moreover, enhanced predictability of treatment response early in the course of a patient’s illness may result in enhanced patient compliance and willingness to seek treatment rapidly upon symptom exac- erbation or recurrence.
Initial efforts to identify predictors of psychotropic drug response focused on clinical variables such as age at onset of illness, level of premorbid function, and comorbidity (Grebb and Cancro, 1989). Unfortunately, these findings have often

21

22 A. K. Malhotra
been confounded by methodological issues (e.g., selection bias, retrospective assessments of response) and have been limited in their applicability to clinical practice. As the understanding of the biological basis of psychiatric disorders has improved, efforts to identify biological predictors of individual drug response have intensified (Malhotra and Pickar, 1996). Plasma and cerebrospinal fluid levels of neurotransmitter metabolites, neurohormone levels, and brain imaging measures, amongst others, have been hypothesized to provide informative correlates of drug response (Bowers et al., 1984; Pickar et al., 1984; Buchsbaum et al., 1992; Kahn et al., 1993; Szymanski et al., 1995). Despite some initially promising results, consis- tent data in this regard have remained elusive.
The introduction of molecular genetic techniques into psychiatric research has provided the impetus for renewed investigations into the identification of psycho- tropic drug response predictors. It is now possible to extract DNA efficiently from blood samples of human subjects, amplify targeted molecular regions for analysis with the polymerase chain reaction, and rapidly genotype individuals at ever increasing numbers of loci in a cost-effective manner. Armed with these techniques, a number of psychiatric research groups, as well as pharmaceutical and biotechnol- ogy companies, are currently attempting to determine the genetic basis for the vari- ation in clinical response. This field of inquiry offers the prospect of identification of easily accessible biological predictors of psychotropic drug response and may provide information about the molecular substrates of psychotropic drug efficacy. This chapter reviews the evidence for a heritable component to psychotropic drug response, discusses basic methodological issues in pharmacogenetic studies, and highlights some future directions for pharmacogenetic research in psychiatry.
Heritability of psychotropic drug response
A basic assumption of genetic studies is that the phenotype of interest is signifi- cantly heritable. Although data from family, twin and epidemiological studies suggest that the major psychiatric disorders have a significant genetic component (Plomin et al., 1994), this may not be the case for response to the drugs that treat these disorders. Unfortunately, heritability data for psychotropic drug response are extremely limited because of the difficulties in designing and executing appropri- ately powered heritability studies in this area. For phenotypes such as psychiatric diagnosis, estimates of heritability can be calculated by comparison of concordance rates between monozygotic (MZ) and dizygotic (DZ) twins. The ratio of concord- ance rates between MZ and DZ twins provides an estimate of the contribution of genes, assuming that the environment is shared equally between twin groups (Gottesman, 1991). Alternatively, evidence for heritability can be provided by the study of twins who are separated at birth (adoption studies) and brought up in

23 Pharmacogenomics of psychotropic drug response
distinct environments (Kety et al., 1975). These approaches, however, are difficult to utilize for estimates of the heritability of pharmacogenetic phenotypes. The ascertainment of MZ and DZ twin pairs (or adopted twins) in which one individ- ual is psychiatrically ill is a laborious process in and of itself and the possibility that both members of any one twin pair would first be ill (presumably with the same diagnosis), treated with the same drug, at the same (or similar) dosage, for similar durations of treatment, and that each twin’s clinical response would be measured utilizing similarly valid and reliable methods, is extremely unlikely. Moreover, ped- igree studies of several generations cannot provide much information in this regard because of the continual development of new psychotropic drugs and subsequent alterations in prescription patterns.
The majority of heritability data on psychotropic drugs is from studies of anti- depressant drug response. Angst et al. (1964) examined 41 first-degree relative pairs (parents or siblings of probands) both treated with the tricyclic antidepressant imipramine and reported that 38 pairs were concordant for response. Pare et al. (1962) studied first-degree relatives of 170 depressed patients who had participated in clinical trials of antidepressant drugs to assess concordance rates in the relative pairs with similar treatment trials. In the 12 cases of concordant treatments, both members of the relative pair had similar responses, with an overall response rate of 42%. In a subsequent study (Pare and Mack, 1971), the same group utilized the same approach in a new cohort and found that 10 of 12 cases were concordant for antidepressant response. Moreover, the high level of response concordance between the proband and first-degree relative when treated with an antidepressant of the same class was in contrast to the lack of concordance when drugs of differ- ent classes were used. These data are consistent with a retrospective analysis of two generations of a family with multiple ill relatives with major depression in which all four family members (proband, mother, daughter, aunt, and first cousin) who underwent treatment with the monoamine oxidase inhibitor tranylcypromine responded despite being nonresponsive to conventional treatment (O’Reilly et al., 1994).
In these early studies, the actual antidepressant used, the dosage of antidepress- ant, and duration of treatment were not consistently controlled. In a recent study, however, Franchini et al. (1998) assessed 45 first-degree relative pairs all treated with the same antidepressant, fluvoxamine, at a similar dosage, 200–300 mg per day for at least 6 weeks, and measured response with the structured rating scale, the Hamilton Depression Scale. In this work, 67% of pairs were concordant for response compared with the 50% that would be expected by chance in first-degree relatives. Although these data suggest that heritable factors are involved in antide- pressant drug response, none of these studies was designed to examine whether shared environmental effects in antidepressant-treated relative pairs influence
24 A. K. Malhotra
response. Therefore, these data can only be viewed as preliminary, and more research in this area is warranted.
Evidence for the heritability of antipsychotic drug response is even more limited. There have been several reports that antipsychotic drug response varies by ethnic- ity (Frackiewicz et al., 1997), suggesting a genetic component to response; however, it is unclear whether the response variation may simply result from differences in drug metabolism or from other disparities in the clinical treatment of different ethnic groups.
The antipsychotic drug most extensively studied in psychiatric pharmacogenet- ics is the prototypic atypical agent clozapine. Heritability of clozapine response, however, has not been assessed in a systematic fashion. The only current informa- tion is in a single case report (Vojvoda et al., 1996) of a MZ twin pair with schizo- phrenia concordant for good response to 550 mg per day of clozapine despite prior nonresponse to multiple typical antipsychotic agents. Typical antipsychotics have also received only limited study in psychiatric pharmacogenetics, although DeLisi and Dauphinais (1989) retrospectively assessed 28 schizophrenic sibling pairs and found no more concordance for neuroleptic response than expected by chance. The low overall response rate in this study group (30%) suggests that this was either a particularly nonresponsive group or that the criteria for response were relatively conservative.
Antipsychotic drugs are potent dopamine receptor antagonists. Although herit- ability data on the behavioral response to these drugs are extremely limited, con- cordance in twins for behavioral response to a dopamine receptor agonist has been observed. Nurnberger and colleagues (1982) administered 0.3mg/kg of the dopa- mine receptor agonist dextroamphetamine (dexamfetamine) to healthy volunteer twin pairs and found that induced alterations in behavioral excitation (r 0.70; p 0.003) as well as changes in plasma growth hormone and prolactin levels that were highly correlated in 12 MZ twins but not in three DZ twin pairs. These results were not accounted for by correlation of plasma amphetamine levels. Although these results may suggest heritability of behavioral response to dopaminergic per- turbation, the small sample of DZ twins was not sufficient to conduct accurate heritability estimates. Nevertheless, further research utilizing acute pharmacologic “challenge” paradigms (Malhotra et al., 1998a) may be useful to assess heritability of drug responses if appropriate safeguards are in place.
Methodological issues in the pharmacogenetics of psychotropic drug response
Efforts to find disease-susceptibility genes have utilized two major approaches. The first, whole genome scanning by linkage analysis, is a nonhypothesis-driven approach in which polymorphisms distributed throughout the genome are

25 Pharmacogenomics of psychotropic drug response
genotyped in families to identify alleles that are shared by ill relatives more often than predicted by chance. The basic unit of analysis is one family that contains either multiple affected relatives through generations or affected relative pairs. If successful, a chromosomal region is “linked” to a phenotype. Unfortunately, linked regions in genome scans are usually extensive, encompass hundreds of genes, and the ultimate identification of the specific genetic variant that contributes to the linkage result can be laborious and resource intensive (Gusella et al., 1983; Tsui et al., 1985). Moreover, genome scanning is particularly disadvantageous for psychi- atric pharmacogenetic studies because of the necessity to specify a genetic model of transmission (for traditional lod score analyses), the low likelihood of ascertain- ment of families with multiple ill relatives treated with the same medication, and the reduced power of genome scanning to detect genes of small effect.
For these reasons, pharmacogenetic studies in psychiatry have been conducted using a complementary approach to genome scanning – the case-control associa- tion design. As the basic unit of analysis in case-control studies is the individual, this approach is particularly well suited for pharmacogenetic studies in which unre- lated individuals may be all that are available for study.
There are several basic steps in conducting a pharmacogenetic association study. First, a phenotype of interest is identified and response criteria are established. In psychiatric pharmacogenetics, clinical response to psychotropic drug treatment and adverse effects of drug treatment have been the major focus (Arranz et al., 1995, Masellis et al., 1995, Malhotra et al., 1996, 1998b). Second, a candidate gene is selected for analysis. Candidate genes are identified either by position in a chromosomal region linked to the disorder in a genome scan or upon neurobiolog- ical evidence that the gene product influences the phenotype of interest. Third, can- didate polymorphisms and candidate alleles (or groups of alleles or haplotypes) are selected within or near the candidate gene. The criteria for candidate polymor- phism selection are somewhat arbitrary but potential criteria include allele fre- quency (polymorphisms with low rare allele frequencies may be minimally informative and, therefore, provide limited power), whether the polymorphism results in an amino acid substitution at the protein level, and the functional conse- quences of the polymorphism in in vitro systems (Malhotra and Goldman, 1999). Finally, in its most basic form, the frequency of alleles (and/or genotypes) is com- pared between groups of patients (for instance, between clozapine responders and nonresponders).
The candidate gene pharmacogenetic strategy has been successful in nonpsychi- atric diseases such as asthma. Effective pharmacologic treatments for asthma include 2-adrenoreceptor (B2AR) agonists and 5-lipoxygenase (ALOX5) inhib- itors. However, a significant subset of asthma patients fails to respond to treatment and suffer significant morbidity and mortality. Recently, Kotani and colleagues
26 A. K. Malhotra
(1999) conducted a case-control pharmacogenetic association study assessing the role of a B2AR polymorphism, Arg16Gly, in the airway responsiveness (as meas- ured by spirometry) of Japanese patients with asthma (n 92) treated with the beta-agonist salbutamol. The candidate gene was selected because of the known affinity of salbutamol for the receptor, and the polymorphism was selected because it alters agonist-promoted receptor down regulation. The finding that the patients who were homozygous for Gly16 had significantly lower airway responsiveness suggests that the a priori identification of Gly/Gly homozygotes could increase the overall effectiveness of asthma treatment with beta-agonists. Similarly, Drazen et al. (1999) examined the association between improvements in forced expiratory volume in the first second (FEV1) in a placebo-controlled trial (n 221) of the anti- asthma drug ABT-761 and a promoter region polymorphism in the gene ALOX5. In this case, the candidate gene was selected because of the known affinity of ABT- 761 for ALOX5 and the polymorphism was chosen based upon in vitro data indi- cating an effect on ALOX5 gene transcription. Patient stratification by ALOX5 genotype did not distinguish between the patients’ overall disease severity; however, it did reveal that patients homozygous for the rare allele failed to respond to ABT-761. In fact, the homozygotes’ FEV1 response was no different than the group of patients who received placebo treatment. These data suggest that sequence variation in the genes that code for the targets of a therapeutic agent may influence the clinical variation in drug response and that the case-control methodology can be successfully utilized to identify these relationships.
Limitations of case-control association studies
Although the case-control methodology has several advantages (see below), this approach is subject to distinct limitations. In particular, the selection of candidate genes for case-control psychiatric genetic studies is complicated by the lack of direct biological evidence linking any one protein (and hence, gene) to susceptibility to any psychiatric disorder. Therefore, it has been suggested that association studies of psychiatric phenotypes should be corrected for every gene in the genome that might be expressed in the central nervous system (CNS), thus requiring p values for significance that would range as low as 10 7. This may be an overly conserva- tive threshold for psychiatric pharmacogenetic studies because candidate gene selection in this work can be guided by biological data. For example, the initial targets for many psychotropic drugs and the cascade of molecular events that follow drug binding have been intensively studied (Hyman and Nestler, 1996) and can serve to provide initial candidates for psychotropic drug response. Moreover, phenotypes in psychiatric pharmacogenetics may be more amenable to candidate gene selection than phenotypes based solely upon diagnostic classifications. For

27 Pharmacogenomics of psychotropic drug response
instance, although the list of suitable candidate genes for antipsychotic drug effi- cacy is relatively broad, the initial candidate genes for a phenotype such as antipsy- chotic drug-induced extrapyramidal symptoms could be reasonably restricted to genes within or related to the dopamine system, given the large body of data impli- cating dopaminergic function in extrapyramidal efffects (Farde et al., 1992). Moreover, new methods such as differential display should provide new candidates for psychiatric pharmacogenetics by providing quantitative data on the effects of psychotropic drug administration on CNS gene expression (Nguyen et al., 1992). Taken together, these data suggest that rational candidate gene selection for psychi- atric pharmacogenetic studies can be accomplished based upon biological data. Therefore, the establishment of extreme thresholds to achieve significance may result in inflated type II errors rates and minimize the opportunity to take advan- tage of the genomic information being generated by the Human Genome Project and other large-scale sequencing efforts.
A second important consideration in evaluating case-control association studies is the criteria used to select a candidate polymorphism within the gene of interest. For genes that have not been scanned for genetic variation, techniques including single-strand conformational polymorphism analysis (Orita et al., 1989), denatur- ing gradient gel electrophoresis (Myers et al., 1985) denaturing high perfor- mance liquid chromatography (Underhill et al., 1997) and methods using oligonucleotide-based chip arrays (Chee et al., 1996) can be utilized to detect new variants. However, since most genes contain a number of sequence variants includ- ing single nucleotide polymorphisms (SNPs), the selection of which variant or var- iants to genotype in a case-control association study may be critical. Several issues should be considered. First, the frequency of the rare allele of a variant may be important. Variants with low frequencies of the rare allele may provide little power to detect significant associations unless the study group size is large or the variant has a strong influence on the phenotype. Moreover, the number of alleles at a locus must be considered. Variants with relatively large number of alleles provide a greater number of potential genotypes, increasing the informativeness of the locus but complicating the statistical analysis. For instance, the dopamine D4 receptor gene (DRD4) polymorphism leading to a 16 amino acid repeat may produce more than 15 different genotypes (Chang et al., 1996), leading to relatively small groups of subjects with particular genotypes. For this reason, most studies with this DRD4 polymorphism have arbitrarily grouped subjects with different genotypes. In instances where genotypes are to be grouped, this is optimally done on the basis of similarity of function or evolutionary relationship of alleles in order to maximize the likelihood of relating a functional variant to phenotype (Templeton et al., 1987).
Perhaps the best criterion for candidate allele selection is functionality. Many intronic sequence variants do not alter gene splicing or expression and most coding
28 A. K. Malhotra
sequence variants are synonymous substitutions that do not alter the amino acid sequence of the gene product. Association studies utilizing variants with no func- tional effects, or coding region variants known not to alter protein structure, have a significantly lower prior probability of detecting valid associations than studies using functional variants or “candidate alleles.” Positive findings with nonfunc- tional markers are more likely to be secondary to chance or to inadvertent ethnic stratification of cases and controls. The nonfunctional or silent variant may be in linkage disequilibrium (nonrandom population association) with an as yet undis- covered functional variant. However, a difference in the frequency of a silent variant and the functional variant to which it is linked dilutes the information content of the silent variant. The informativeness of the silent variant is also weak- ened by population differences in variant frequencies and less than perfect linkage disequilibrium. Moreover, the possibility that a silent variant is in linkage disequi- librium with a functional variant should be viewed with caution if the candidate gene has been rigorously scanned for variation in large, ethnically mixed popula- tions. An illustrative example of this issue is provided by the well-publicized asso- ciation between the TaqI A1 allele near the dopamine D2 receptor gene (DRD2) and risk of alcoholism (Blum et al., 1990). The initial report of this association sug- gested that variation within DRD2 influenced susceptibility to alcoholism. However, the Taq I A1 allele is located more than 10kb downstream from the coding region of DRD2 and does not appear to impart any structural or functional effects on the D2 receptor (Grandy et al., 1989). Gejman and colleagues (1994) scanned the coding sequence of DRD2 and were unable to detect additional vari- ants that might account for this association, although a functionally deficient variant giving rise to Ser311Cys, was found but was not linked to alcoholism in a Southwestern American-Indian population (Goldman et al., 1997). Therefore, it is unlikely that the Taq I A1 allele is in linkage disequilibrium with a DRD2 variant that increases the susceptibility to alcoholism
An alternative approach to single candidate variant selection within a gene is to genotype subjects at several variants within the gene and construct haplotypes. Using haplotype estimation–maximization algorithms (Long et al., 1995), it is pos- sible to assign haplotypes objectively. However, if one is testing the actual variant that produces the phenotype, then haplotype analysis adds to the degrees of freedom but not the information content and increases the possibility of a false- negative result. The power and validity of haplotype-based analysis is dramatically increased by the grouping of haplotypes according to their most likely common origin (Templeton et al., 1987). In this way, a group of haplotypes is most informa- tive for functional genetic variants that first appeared on an ancestral haplotype and tends to be found on the descendants of that haplotype.
Finally, perhaps the most discussed limitation of case-control association studies
29 Pharmacogenomics of psychotropic drug response
is the potential for ethnic stratification between subject groups. In the case of psychiatric pharmacogenetic studies, this might arise in a study comparing candi- date allele frequencies between drug responders and non responders. If the candi- date allele frequency varies between ethnic groups, and responders and nonresponders are not ethnically matched, a significant difference in allele frequen- cies could be detected that is not associated with drug response. Unfortunately, ethnic variation in allele frequencies is not uncommon. For example, Chang and colleagues (1996) have found that frequencies for the 4-repeat allele of the DRD4 16 amino acid residue repeat varies between 16 and 96% in different populations. The DRD2 Taq I A1 allele is twice as frequent in African-Americans and four times as frequent in some American Indians compared with American Caucasians (Goldman et al., 1993).
One method to control for population differences is to analyze ethnically homo- geneous samples, or population isolates. Unfortunately, it is difficult to ascertain these samples in Westernized society derived from multiple ethnic groups. Moreover, in pharmacogenetic studies, the phenotypic information is generally collected in the context of a clinical trial in which the objective is to study a wide variety of ethnic groups, and limiting study participation to a sole ethnic group would not be possible. An alternative to seeking ethnically homogeneous data sets is to collect DNA from the parents of probands and analyze data with the transmis- sion disequilibrium test (Spielman and Ewens, 1996). With the family-based asso- ciation approaches, the frequency of transmission of a candidate allele between parent and child is compared with the frequency of nontransmission. As the non- transmitted alleles represent the control alleles for the transmitted alleles, these tests are not sensitive to ethnic stratification. Although with this approach it is more difficult to collect large samples because of the necessity to ascertain family members, recent modifications of the transmission disequilibrium test involving siblings and the introduction of quantitative data analytic strategies (Allison et al., 1997) suggest that family-based association may represent a useful tool in psychi- atric pharmacogenetics.
From pharmacogenetics to pharmacogenomics
The first generation of pharmacogenetic studies have utilized a limited number of polymorphisms in relatively small data sets derived from ethnically heterogenous populations. The next generation of studies, however, may be dramatically differ- ent from these early efforts because of impressive advances in genomic informa- tion, novel statistical genetic methodologies, and marked improvements in genotyping technologies. The following section briefly discusses some of the recent major advances in these domains.

30 A. K. Malhotra
Table 2.1. The SNP Consortium Date
23 November 1999 23 December 1999 30 March 2000
28 April 2000
21 August 2000
1 November 2000
Notes:
SNP, single nucleotide polymorphism.
SNPs
2556
7365 41209 102719 296990 638372
  
Completion of the human genome sequence and identification of single nucleotide polymorphisms
The most important development at the beginning of the twenty-first century was the announcement of the completion of the human genome sequence by the US federally funded Human Genome Project and a private biotechnology firm, Celera Genomics. Although full sequence data are not yet available, the genome sequence that is available is already providing the reference sequence for large-scale efforts aiming to identify the genetic variation among individuals. In particular, there has been a focus on the identification of hundreds of thousands of SNPs within the genome. For example, the SNP Consortium, a group of pharmaceutical companies and academic centers, was formed in 1999 to identify SNPs distributed throughout the genome for use in pharmacogenomic studies. This consortium has, as of this writing, identified over 1 000 000 SNPs, with the pace of SNP identification contin- uing to increase rapidly (Table 2.1). In addition, Celera Genomics has recently reported that it has identified more than 2000000 SNPs within the context of its genome sequencing efforts.
Many of the SNPs that are being identified may be useful in pharmacogenetic studies. Cargill et al. (1999) scanned the coding region of 106 genes, many with potential relevance to the CNS, and identified an average of 3.7 coding region SNPs (cSNPs) per gene, of which nearly 50% were nonsynonymous, and 41% of the non- synonymous cSNPs had a minor allele frequency of greater than 5%. As discussed above, nonsynonymous, coding region SNPs may have a higher a priori probabil- ity of altering function and be the most powerful in pharmacogenetic studies. Therefore, the massive SNP identification efforts, coupled with data demonstrat- ing that a significant subset of them may be useful in pharmacogenetic studies, suggest that a large number of novel SNP targets will soon be available for study in the next generation of pharmacogenetic studies.
31 Pharmacogenomics of psychotropic drug response
Developments in statistical genetic methodologies
Previous pharmacogenetic studies have utilized the case-control design. This approach has three major advantages over family-based designs. First, many of the patients participating in clinical trials of psychotropic drugs are already in their mid to late adult years and, therefore, collecting DNA from family members may be difficult because of lack of availability. Second, it has been suggested that pro- bands selected from trios, or other family-based designs, may be subtly different than the general population of cases, thus reducing the generalizability of findings made with the family-based designs (Bruun and Ewald, 1999). Finally, the case- control design has greater statistical power than family-based approaches. Risch and Teng (1998) examined the relative power of the case-control design versus family-based approaches under a number of different genetic models. Under most models, the case-control design was more efficient than family-based designs, with sample sizes of less than 1000 usually sufficient to detect genes of relatively modest effect, even with extremely conservative significance levels (alpha 5 10 8) and minor allele frequencies as low as 5%. These data suggested that the case-control design would be optimal for pharmacogenetic studies; however, as discussed above, the potential for undetected ethnic stratification has tempered enthusiasm for this design (Paterson, 1997).
Fortunately, “genomic control” methods are under development to help to account for ethnic stratification in case-control studies. Genomic control tech- niques are based upon the idea that study groups (cases versus controls, respond- ers versus nonresponders, etc.) can be assessed for the presence of stratification by assessing the allele frequency of markers, unlinked to the phenotype of interest, in each group. Pritchard and Rosenberg (1999) determined that no more than 40 unlinked markers, and perhaps less, are necessary to achieve 95% probability of detecting stratification in study groups of over 200 subjects. If stratification is detected, subjects can be removed until stratification is not present, or correction factors can be introduced that account for the level of stratification between groups (Devlin and Roeder, 1999). Both approaches reduce study power but minimize the risk that the case-control design results in a false-positive or false-negative result.
Genomic control techniques have not yet been utilized in psychiatric pharma- cogenetic studies. However, the considerable advantages of the case-control approach for pharmacogenetic studies, coupled with the genomic control methods, suggest that it should soon be feasible to conduct large-scale pharmacogenetic studies with increased power and with decreased potential for undetected ethnic stratification. Moreover, the extension of genomic control techniques to assess quantitative phenotypes is underway and should provide another compelling reason to consider these approaches in pharmacogenetic studies involving hetero- genous populations.
32 A. K. Malhotra
Technology
The availability of over 500000 SNPs distributed throughout the genome suggests that whole-genome association studies may soon be feasible – provided appropri- ate high-throughput SNP genotyping technology can be developed. Because SNPs are biallellic, unlike short tandem repeat markers with multiple alleles, they are more amenable to automated genotyping procedures. At the time of this writing, the cost of genotyping a single SNP in a single individual can be a low as 40 cents (US), but even this cost remains prohibitive for studies involving hundreds of patients with tens of thousands of SNPs. For example, to genotype 100 000 SNPs in 1000 individuals at this cost would represent a genotyping expenditure of US$40 million, before including the considerable costs associated with phenotype col- lection and DNA isolation. New technologies are currently under development within the biotechnology industry that promise to reduce genotyping costs further (with some near-term estimates as low as 3–5 cents per genotype) to a level that makes whole genome analyses more plausible. Moreover, the use of DNA pooling strategies provides an efficient method for generating genotype data on large numbers of subjects in an increasingly efficient manner. These developments suggest that it should soon be feasible to genotype comprehensively large study groups of patients characterized for drug response across thousands of SNPs dis- tributed throughout the genome.
Conclusions
With the developments mentioned above, the next generation of molecular genetic studies of psychotropic drug responses should be markedly different from the first- generation studies. It will soon be possible to collect DNA from patients enrolled in large-scale clinical trials of thousands of patients, select SNPs that occur in each gene expressed in the CNS, and compare and contrast SNP frequencies in respond- ers and nonresponders (or patients who develop side effects and those who do not) in large case-control designs. Moreover, as statistical methods are developed to account for undetected ethnic stratification between groups, the amount of strat- ification (if any) can be quantified between groups and accounted for in the anal- yses. Improved bioinformatics tools and statistical methods will be required to deal with the volume of data that will be generated by this approach, but it seems likely that such methods will soon be available given the level of interest in genetic approaches to complex diseases and to phenotypes such as drug response. With these developments in place, the first generation of pharmacogenetic studies that utilized SNPs in relatively limited numbers of candidate genes will be replaced by pharmacogenomic studies in which the complete genome is assessed. Therefore, the prospect for molecular genetic studies of psychotropic drug response seems

33 Pharmacogenomics of psychotropic drug response
bright and there is renewed hope that the first genes associated with psychotropic drug response will soon be identified.
REFERENCES
Allison DB (1997). Transmission–disequilibrium tests for quantitative traits. Am J Hum Genet 60, 676–690.
Angst J (1964). Antidepressiver effekt und genetische faktoren. Arzneimittel-Forschung 14(Suppl), 496–500
Arranz M, Collier D, Sodhi M et al. (1995). Association between clozapine response and allelic variation in 5HT-2A receptor gene. Lancet 346, 281–282
Blum K, Noble EP, Sheridan PJ et al. (1990). Allelic association of human dopamine D2 receptor gene in alcoholism. J Am Med Assoc 263, 2055–2060.
Bowers MB, Swigar ME, Jatlow PI, Goicoechea N (1984). Plasma catecholamine metabolites and early response to haloperidol. J Clin Psychiatry 45, 248–251.
Bruun TG, Ewald H (1999). Selection bias of susceptibility genes possible when using parent-off-spring trios in genetic association studies. Mol Psychiatry 4, 414–416.
Buchsbaum MS, Potkin SG, Marshall JF et al. (1992). Effects of clozapine and thiothixene on glucose metabolic rate in schizophrenia. Neuropsychopharmacology 6, 155–163.
Cargill M, Altshuler D, Ireland J et al. (1999). Characterization of single-nucleotide polymor- phisms in coding regions of human genes. Nat Genet 22, 231–238.
Chang FM, Kidd JR, Livak KJ, Pakstis AJ, Kidd KK (1996). The world-wide distribution of allele frequencies at the human dopamine D4 receptor locus. Hum Genet 98, 91–101.
Chee M, Yang R, Hubbell E et al. (1996). Accessing genetic information with high-density DNA arrays. Science 274, 610–614.
DeLisi LE, Dauphinais D (1989). Neuroleptic responsiveness in siblings concordant for schizo- phrenia. Arch Gen Psychiatry 46, 477.
Devlin B, Roeder, K (1999). Genomic control for association studies. Biometrics 55, 997–1004. Drazen JM, Yandava CN, Dube L et al. (1999). Pharmacogenetic association between ALOX5 promoter genotype and the response to anti-asthma treatment. Nat Genet 22,
168–170.
Farde L, Nordstrom AL, Wiesel FA, Pauli S, Halldin C, Sedvall G (1992). Positron emission tom-
ographic analysis of central D1 and D2 dopamine receptor occupancy in patients treated with classical neuroleptics and clozapine. Relation to extrapyramidal side effects. Arch Gen Psychiatry 49, 538–544.
Frackiewicz EJ, Sramek JJ, Herrera JM, Kurtz NM, Cutler NR (1997). Ethnicity and antipsychotic response. Ann Pharmacother 31, 1360–1369.
Franchini L, Serretti A, Gasperini M, Smeraldi E (1998). Familial concordance of fluvoxamine response as a tool for differentiating mood disorder pedigrees. J Psychiatric Res 32, 255–259.
Gejman PV, Ram A, Gelertner et al. (1994). No structural mutation in the dopamine D2 recep- tor gene in alcoholism and schizophrenia. J Am Med Assoc 271, 204–208.

34 A. K. Malhotra
Goldman D, Brown GL, Albaugh B et al. (1993). DRD2 dopamine receptor genotype, linkage dis- equilibrium and alcoholism in American Indians and other populations. Alcohol Clin Exp Res 17, 199–204.
Goldman D, Urbanek M, Guenther D, Robin R, Long UC (1997). The dopamine D2 receptor: Linkage and association of a functional allele [Ser311Cys] and other D2DR markers to alco- holism, substance abuse and schizophrenia in Southwestern American Indians. Neuropsychiat Genet 74, 386–394.
Gottesman II (1991). Schizophrenia Genesis: The Origins of Madness. New York: WH Freeman. Grandy DK, Litt M, Allen L, Bunzow JR, Marchionni M, Makam H (1989). The human dopa- mine D2 receptor gene is located on chromosome 11 at q22–q23 and identifies a TaqI RFLP.
Am J Hum Genet 45, 889–785.
Grebb JA, Cancro R (1989). Schizophrenia: clinical features. In Kaplan HI, Sadock BJ, eds.
Comprehensive Textbook of Psychiatry. Baltimore, MD: Williams & Wilkins, pp. 457–486. Gusella JF, Wezler NS, Conneally PM et al. (1983). A polymorphic DNA marker genetically linked
to Huntington’s disease. Nature 306, 234–238.
Hyman SE, Nestler EJ (1996). Initiation and adaptation: a paradigm for understanding psycho-
tropic drug action. Am J Psychiatry 153, 151–162.
Kahn RS, Davidson M, Siever L, Gabriel S, Apter S, Davis KL (1993). Serotonin function
and treatment response to clozapine in schizophrenic patients. Am J Psychiatry 150,
1337–1342.
Kety SS, Rosenthal D, Wender PH, Schulsinger F, Jacobsen B (1975). Mental illness in the biolog-
ical and adoptive families of adopted individuals who have become schizophrenic: a prelimi- nary report based on psychiatric interviews. In Fieve RR, Rosenthal D, Brill H (eds) Genetic Research in Psychiatry. Baltimore, MD: Johns Hopkins University Press, pp. 147–165.
Kotani Y, Nishimura Y, Maeda H, Yokoyama M (1999). 2-Adrenergic receptor polymorphisms affect airway responsiveness to salbutamol in asthmatics. J Asthma 36, 583–590
Long JC, Williams RC, Urbanek M (1995). An E–M algorithm and testing strategy for multiple- locus haplotypes. Am J Hum Genet 56, 799–810.
Malhotra AK, Goldman D (1999). Benefits and pitfalls encountered in psychiatric genetic asso- ciation studies. Biol Psychiatry 45, 544–550
Malhotra AK, Pickar D (1996). Biologic predictors of clozapine response in schizophrenia. Psychiat Ann 26, 390–394.
Malhotra AK, Goldman D, Ozaki N, Breier A, Buchanan R, Pickar D (1996). Lack of association between polymorphisms in the 5-HT2A receptor gene and the antipsychotic response to cloza- pine. Am J Psychiatry 153, 1092–1094.
Malhotra AK, Breier A, Goldman D, Picken L, Pickar D (1998a). The apolipoprotein E epsilon 4 allele is associated with blunting of ketamine-induced psychosis in schizophrenia. A prelimi- nary report. Neuropsychopharmacology 19, 445–448.
Malhotra AK, Goldman D, Buchanan RW et al. (1998b). The dopamine D3 receptor (DRD3) Ser9Gly polymorphism and schizophrenia: a haplotype relative risk study and association with clozapine response. Mol Psychiatry 3, 72–75.
Masellis M, Paterson AD, Badri F et al. (1995). Genetic variation of 5-HT2A receptor and response to clozapine. Lancet 346, 1128.
35 Pharmacogenomics of psychotropic drug response
Myers RM, Fischer SG, Lerman LS, Maniatis T (1985). Nearly all single base substitutions in DNA fragments joined to a GC-clamp can be detected by denaturing gradient gel electrophoresis. Nucl Acids Res 13, 3131–3145.
Nguyen TV, Kosofsky BE, Birnbaum R, Cohen BM, Hyman SE (1992). Differential expression of c-fos and zip268 in rat striatum after haloperidol, clozapine, and amphetamine. Proc Natl Acad Sci USA 89, 4270–4274.
Nurnberger JI Jr., Gershon ES, Simmons S et al. (1982). Behavioral, biochemical and neuroendo- crine responses to amphetamine in normal twins and “well-state” bipolar patients. Psychoneuroendocrinology 7, 163–176.
O’Reilly RL, Bogue L, Singh SM (1994). Pharmacogenetic response to antidepressants in a multi- case family with affective disorder. Biol Psychiatry 36, 467–471.
Orita M, Suzuki Y, Seyiya T, Hayashi K (1989). Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics 5, 874–879.
Pare CMB, Mack JW (1971). Differentiation of two genetically specific types of depression by the response to antidepressant drugs. J Med Genet 8, 306–309.
Pare CMB, Rees L, Sainsbury MJ (1962). Differentiation of two genetically specific types of depression by the response to anti-depressants. Lancet ii, 1340–1343.
Paterson AD (1997). Case-control association studies in complex traits – the end of an era? Mol Psychiatry 2, 277–278.
Pickar D, Labaraca R, Linnoila M et al. (1984). Neuroleptic-induced decrease in plasma homovanillic acid and antipsychotic activity in schizophrenic patients. Science 225, 954–957. Plomin R, Owen MJ, McGuffin P (1994). The genetic basis of complex human behaviors. Science
264, 1733–1739.
Pritchard JK, Rosenberg NA (1999). Use of unlinked genetic markers to detect population strat-
ification in association studies. Am J Hum Genet 65, 220–228.
Risch N, Teng J (1998). The relative power of family-based and case-control designs for linkage
disequilibrium studies of complex human diseases I. DNA pooling. Genome Res 8, 1273–1288. Spielman RS, Ewens WJ (1996). The TDT and other family-based tests for linkage disequilibrium
and association. Am J Hum Genet 59, 983–989.
Szymanski S, Lieberman J, Pollack S et al. (1995). Clozapine effects on neuroendocrine response
to apomorhine challenge testing in chronic neuroleptic nonresponsive schizophrenia: prelim-
inary findings. Biol Psychiatry 37, 52–55.
Templeton AR, Boerwinkle E, Sing CF (1987). A cladistic analysis of phenotypic associations with
haplotypes inferred from restriction endonuclease mapping. I. Basic theory of an analysis of
alcohol dehydrogenase activity in Drosophila. Genetics 117, 343–351.
Tsui LC, Buchwald M, Barker D et al. (1985). Cystic fibrosis locus defined by a genetically linked
polymorphic DNA marker. Science 230, 1054–1057.
Underhill PA, Jin L, Lin AA et al. (1997). Detection of numerous Y chromosome biallelic poly-
morphisms be denaturing high-performance liquid chromatography. Genome Res 7,
996–1005.
Vojvoda D, Grimmell K, Sernyak M, Mazure CM (1996). Monozygotic twins concordant for
response to clozapine. Lancet 347, 61
3
Neuropsychopharmacology: the interface between genes and psychiatric nosology
Thomas A. Ban
Vanderbilt University, Nashville TA, USA
OVERVIEW
The observation that mental illness runs in families received substantial support in family, twin and adoption studies. Nonetheless, the heterogeneity within the diagnostic categories of schizophrenia and manic-depressive illness has precluded any meaningful research in the genetics of these disorders. To break the impasse in genetic research of mental illness, con- sideration was given to split psychiatric disorders into simpler biological or behavioral com- ponents. However, all alternative approaches fall short of psychiatric nosology in classifying mental illness in a clinically relevant manner. Neuropsychopharmacology has the unique capability of linking the effect of a psychotropic drug on mental illness with the effect of the substance on brain structures involved in the action mechanism of the drug. Since the primary targets of psychotropic drugs are encoded by genes that have been identified, any form of disease which corresponds with the treatment-responsive population to a psycho- tropic drug is suitable for the generation and testing of genetic hypotheses relevant to mental illness. To provide orientation points about what nosology could offer genetic research, the history of psychiatric nosology is reviewed and the varying constructs that can be used are illustrated. Kraepelin’s (1899) diagnostic concepts of dementia praecox and manic-depressive insanity are artificially derived nosologic constructs, whereas Wernicke’s (1899) classification was based on scientific developments that were to become the foun- dation of neuroscience. Schneider’s (1950) rudimentary classification was the first nosology in which it was recognized that mental pathology is expressed in the mode (form) in which the experience appears (processed) and not in the “content” of the experience. Specially devised diagnostic instruments are described that provide more homogeneous populations for genetic research in mental illness. However, the use of nosologic homotypes, derived by the employment of a specially devised nosologic matrix is recommended for obtaining inter- pretable findings. The data collected by using the nosologic matrix could also serve as the starting point for the development of an empirically derived, pharmacologically meaningful classification of mental illness.
36

37 Genes and psychiatric nosology
Introduction
The observation that mental illness runs in families has been documented since the mid-18th century (Battie, 1758; Chiarugi, 1793–1794). The first genetic theory of mental illness was formulated in the mid-1850s (Shorter, 1997). Morel’s (1857) theory of degeneration is based on the assumption that mental illness is the result of an “innate biological defect” that becomes manifest in increasingly severe mental syndromes in “lineal descents.”
Morel’s (1857) degeneration theory was replaced by Moebius’ (1893) “endogeny theory”, which implied a “constitutionally determined predisposition” for develop- ing mental illness. The dichotomy of “endogenous” and “psychogenic” (Wimmer, 1916) psychoses has been lingering on ever since.
Genetics in psychiatry
Epidemiological genetics
The notion that mental illness runs in families was endorsed by Kraepelin (1896) and Magnan (1896). It received substantial support from “family,” “twin,” and “adoption” studies. The majority of epidemiological genetic studies dealt with Kraepelin’s (1899) diagnostic concepts of “manic-depressive insanity” and “dementia praecox.” Magnan’s (1893) diagnostic concept of “delusional psychoses” received relatively little attention in genetic research.
The risk of developing schizophrenia for relatives of schizophrenics, and manic- depressive (bipolar) illness for relatives of manic depressives, was found to be con- sistently higher than in the general population. The risk of developing the respective illness in both diagnostic groups was found to be higher for first- than for second-degree relatives (Slater and Cowie, 1971; Kay, 1978).
Concordance rates for schizophrenia and manic-depressive illness were found to be consistently higher in “monozygotic” than in “dizygotic” twins, with concord- ance rates in monozygotic twins for schizophrenia ranging from 35 to 85.8%, and in manic-depressive illness from 33 to 93%. Concordance rates in “dizygotic” twins in schizophrenia range from 0 to 17%, and in manic-depressive illness from 0 to 8% (Gottesman and Shields, 1966; Tsuang and Vandermey, 1980).
Children of schizophrenic natural (biological) parents adopted into the homes of nonschizophrenic foster parents were found to develop schizophrenia at a much higher rate than “adopted away” children of normal parents (Heston, 1966; Mendlewicz and Rainer, 1977). Mental illness was also encountered at a much higher rate in the biological than in the adoptive families of adopted schizophrenic and manic-depressive children (Kety et al., 1994; Wender et al., 1986).
 
38 T. A. Ban
Mathematical models of inheritance
In spite of findings that genetic factors play an important role in the etiology of schizophrenia and manic-depressive illness, the mode of transmission of these dis- orders, even with the employment of powerful statistical models, has remained hidden (O’Rourke et al., 1982; Craddock et al., 1995). Based on a comprehensive review of “pedigree and segregation analyses,” the Genetics Workgroup of the US National Institute of Mental Health concluded that a “single major locus” cannot account for a large proportion of the “familial aggregation” of either schizophrenia or bipolar illness (a term used for manic-depressive disorder). The group suggested that the mode of inheritance of these disorders is “complex” and very “likely involves multiple interacting genes” (Moldin, 1999).
Molecular genetics: linkage analysis
Findings in “molecular genetic” studies in schizophrenia and manic-depressive illness did not clarify matters further. Employment of “positional cloning” (“back- ward genetics,” “genome scanning”) yielded inconsistent, conflicting results.
There are numerous publications reporting susceptibility loci for schizophrenia on chromosomes 1q, 3p, 5q, 6p, 8p, 8q, 9p, 10q, 12q, 13p, 14p, 15q, 20p, or 22q, and for manic-depressive illness on chromosomes 4p, 5p, 6p, 10q, 11p, 12q, 16p, 18p, 18q, 20p, 21q, or 22q. However, the findings in one group of patients could not be replicated in others (Gershon et al., 1998; Moldin, 1999).
Different findings in different samples from the same diagnostic category indicate genetic heterogeneity within the diagnosis (Tsuang and Faraone, 1990, 1995). However, the heterogeneity, which precluded meaningful genetic research, has not interfered with the recognition that “genetic anticipation,” the essential feature of the first genetic theory of mental illness, is encountered in some schizo- phrenic families and also in some families with manic-depressive illness. Genetic anticipation may result from “trinucleotide repeat mutations,” an anomaly causally linked with Huntington disease (Petronis and Kennedy, 1995; Faraone et al., 1999).
Molecular genetics: candidate gene approach
Genetic heterogeneity, coupled with heterogeneity in pharmacological responsive- ness, has also precluded meaningful research with the employment of “forward genetics” in schizophrenia and manic-depressive illness. Nonetheless, on the basis of the demonstrable therapeutic effectiveness of some psychotropic drugs in schizophrenia and of others in manic-depressive illness, various genes have been implicated in the pathophysiology of these conditions. These genes encode trans- porters, receptors (e.g., serotonin 5-HT transported and 5-HT type 2A receptor, dopamine transporter 1 and D2–D4 receptors), and enzymes (e.g., monoamine oxidase, dopamine--hydroxylase, catechol-O-methyltransferase) involved in
39 Genes and psychiatric nosology
neuronal transmission. “Association studies,” with the employment of the “candi- date gene approach,” however, have invariably failed to detect significant differences in mutations (polymorphisms) in the implicated genes between normal subjects and schizophrenic or manic-depressive patients (Malhotra and Goldman, 1999; Heiden et al., 2000). If there is a difference in “allelic variations” between the treat- ment responsive form of illness within these diagnostic categories and normal sub- jects, it is covered up by the heterogeneity of the patient samples used in the comparisons. An essential prerequisite for the demonstration of a difference in “allelic variations” would be the identification of the treatment-responsive forms of illness within schizophrenia or manic-depressive illness.
Replication studies also failed to support findings that allelic variations in the serotonin 5-HT2A or 5-HT2C receptor genes are responsible for individual differ- ences in the therapeutic response to clozapine (Arranz et al., 1995; Malhotra et al., 1996). If the clozapine-responsive form of illness within schizophrenia could be identified, and the forms of the illness responsive to serotonin 5-HT2A and dopa- mine D2 demonstrated, the question whether therapeutic responsiveness to cloza- pine depends on “allelic variations” in the genes for these proteins would become a testable hypothesis.
Consensus-based diagnostic classifications
The inconsistent and conflicting findings in molecular genetic research in schizo- phrenia and manic-depressive illness led to a steadily growing dissatisfaction with psychiatric nosology. By the late 1990s, it was recognized that it “would be fool- hardy to think” that diagnostic criteria in classifications such as those in the American Psychiatric Association’s Diagnostic and Statistical Manuals of Mental Disorders (DSM-III, DSM-III-R, and DSM-IV; American Psychiatric Association 1980, 1987, 1994) could “select anything that maps into the genome” (Hyman, 1999). DSM-III and its successors are “consensus-based classifications,” i.e., sets of diagnostic formulations agreed upon by a body of well-informed psychiatrists. To accommodate the different orientations in psychiatry, the diagnostic categories in these classifications are broad; because of the inclusion of different forms of disease in the same category of illness, the populations within the diagnostic categories are heterogeneous (Ban, 2000a).
In depressive illness, for example, Kraepelin’s (1899) “unitary concept of melancholia” – derived by pooling together six distinct syndromes of “melancho- lia” (i.e., “simplex,” “gravis,” “stuporous,” “paranoid,” “fantastic,” and “delirious”) – was adopted in DSM-III virtually unchanged. In spite of findings in psycho- pharmacological, genetic and nosological studies that indicated that depressive illness is heterogeneous (Angst, 1966; Overall et al., 1966), the unitary concept of

40 T. A. Ban
melancholia has been retained in DSM-III-R and DSM-IV. Depressive disorders in DSM-IV consist of one core syndrome, “major depression,” which is so broad that, in a “composite (polydiagnostic) evaluation” in 6 of the 25 diagnostic classifications of the diagnostic instrument, 13.5–35.5% of the patients did not fit any of the depressive diagnoses (Ban, 1992b).
While the diagnosis of “major depression” is eminently suited to reconcile widely different conceptualizations of depressive illness, it is a consensus-based diagnosis that covers up its component diagnoses. Schneider’s (1920) concept of “vital depression” – the diagnosis which allowed Roland Kuhn (1957) to recognize imipramine’s antidepressant effect – is covered up in DSM-IV to the extent that, even if the patient is so severely ill that he/she displays all the possible symptoms and signs considered for the DSM-IV diagnosis of “major depression,” one still would not know whether the patient qualifies for “vital depression.” The anti- depressant-responsive forms of depressive illness are also obscured by the DSM-IV. There is no way to predict, within the framework of DSM-IV, which one (or two, including placebo responders) of three patients with the diagnosis of major depression will respond to treatment with an antidepressant drug (Ban, 1987, 1999).
The same defect applies to the DSM-IV diagnostic concept of “schizophrenia” (Ban, 1987, 1999).
Alternative approaches
Discouraged by the limitations of current diagnostic constructs, van Praag (1992, 2000) argues that today’s “psychiatric taxonomy” presents diagnostic entities of “dubious validity,” and suggests replacing the “nosological disease model” by a “reaction form-based disease model” in which “psychiatric conditions” are clas- sified in eight broad basins: “disturbed reality testing and clear consciousness,” “dis- turbed reality testing and lowered consciousness,” “disturbances in affect regulation,” “disturbed cognition,” “conditions in which social adaptation and affil- iative abilities are disturbed,” “conditions with disturbed impulse regulation,” “syn- dromes characterized by termination pathology,” and “somatic syndromes without manifest somatic pathology.” Others propose to break-up psychiatric disorders into “simpler biological or behavioral components” (Lander, 1988; Lander and Schork, 1994). Still others have suggested reconceptualizing mental illness in terms of “dis- crete neurobiological deficits,” i.e., alternative phenotypes. One of the most inten- sively studied “alternative phenotypes” of schizophrenia is the “abnormality of smooth pursuit eye movement” (Holzman et al., 1988). Another frequently studied “alternative phenotype” is the “P-50 (evoked response) deficit” (Freedman et al., 1999). The former had been linked in genetic studies to a locus on the short arm

41 Genes and psychiatric nosology
of chromosome 6 (Arolt et al., 1996), and the latter to the “7-nicotinic acid recep- tor” on the long arm of chromosome 15 (Freedman et al., 1997). Nevertheless, the usefulness of these “alternative phenotypes” in genetic research in schizo- phrenia is questionable, because both “phenotypes” are encountered several times more frequently in the general population than schizophrenia (Faraone et al., 1999).
It has also been proposed to replace “traditional psychiatric nosology” with a “genetic psychiatric nosology,” which would classify patients into categories that “correspond with the genes” (Faraone et al., 1999). While such a nosology could focus attention on overlaps between certain traits (e.g., depression and anxiety), it would group together individuals with genes for a particular disease who fully qualify for the disease and individuals who, despite carrying the genes for the disease, are symptom free.
Neuropsychopharmacology
To-date, there is no alternative methodology to psychiatric nosology for classifying mental pathology in a clinically relevant manner. In the light of the inadequacy of DSM-IV and the failure to replace traditional psychiatric concepts by empirically derived objective measurements, the identification of suitable forms of illness for genetic research in the different nosologies has become of practical importance for progress in the field.
Neuropsychopharmacology, by its unique capability of linking the effect of a psychotropic drug on mental illness with the effect of the substance on brain struc- tures involved in the mechanism of the drug, offers an adequate methodology for the identification of suitable forms of illness for genetic research (Ban, 1999). Since the primary targets of psychotropic drugs in the brain (e.g., G-protein-coupled receptors, nuclear (hormonal) receptors, ion channels, enzymes, etc.) are all encoded by genes that have been identified, any nosologic entity that corresponds with the treatment-responsive population is suitable for the generation and testing of genetic hypotheses relevant to mental illness.
Psychiatric nosology
To provide orientation points about what nosology could offer, and about the nosologies which might be suitable for use in genetic research, the history of psychiatric nosology, with special reference to some of the influential classifications in psychiatry is reviewed. It can be seen that many terms are used to describe the symptoms/syndromes of mental illness, with different classifications implying different groupings.
 
42 T. A. Ban
Boissier de Sauvages
The origin of psychiatric nosology is in the work of Boissier de Sauvages (1768), who classified diseases, including “mania” (insanity), as if they were “specimens of natural history” by dividing them into 244 “species,” 295 “genera,” and 10 “classes” (Garrison, 1929). His assertion that naturally occurring categories of disease can be identified in a manner which would allow the attribution of each patient to only one class by grouping the symptoms at a particular point (cross-section) in time opened the path for the syndromic classifications of mental illness (Ban, 2000b).
Philippe Pinel
The first clinically employed psychiatric nosology was Pinel’s classification (1799, 1801). It was an empirical, “phenetic” classification, based on “meticulous descrip- tion of the appearance of objects ” in which “mental derangements” were “distrib- uted” into five distinct “species” (syndromes): “melancholia (depression) or delirium (delusions) upon one subject exclusively,” “mania (insanity) without delirium,” “mania with delirium,” “dementia or the abolition of the thinking faculty,” and “idiotism or obliteration of the intellectual faculties.” Pinel’s (1801) classification was modified and further elaborated by Esquirol (1838).
Jean-Etienne Dominique Esquirol
Esquirol (1838) divided insanity into five general forms: “lypemany (or melancholy of the ancient),” “monomania,” “mania,” “dementia,” and “imbecility or idiocy.” In variance with Pinel (1801), he assigned “melancholia” a distinct status, separated “partial insanity” (“monomania”) from “insanity proper” (“mania”), and distin- guished within “partial insanity” three distinct forms: “intellectual,” “affective,” and “instinctive.” Esquirol’s (1838) distinction between “partial insanity” and “insanity proper” was adopted by Kahlbaum (1863).
Karl Kahlbaum
Kahlbaum (1863) classified mental illness into five categories: (i) “vesanias,” corres- ponding with “insanity proper,” in which the syndromic expression of the disease changes through different developmental stages until it reaches dementia; (ii) “vecordias,” corresponding with “partial insanity” (e.g., “paranoia”, “dysthymia”), in which the syndromic expression remains essentially unchanged and restricted to one mental faculty during lifetime; (iii) “dysphrenias,” or symptomatic diseases linked to somatic illness; (iv) “neophrenias,” which are inborn or have an onset shortly after birth; and (v) “paraphrenias,” with an onset at periods of transition in biological development (e.g., puberty, involution). Kahlbaum’s (1874) postulation of a close correspondence between etiology, brain pathology, symptom pattern, and outcome picture had a major impact on nosologic development in psychiatry.
43 Genes and psychiatric nosology
It stimulated Kraepelin (1893) to adopt “Sydenham’s disease model” and shift emphasis in his classification of mental illness from cross-sectional syndromes to progression of clinical manifestations (Ban, 2000b).
Emil Kraepelin
Kraepelin’s (1883, 1886, 1889) syndromic classification, presented in the first three editions of his textbook, was gradually replaced by his disease-oriented classifica- tion. To achieve his objective, Kraepelin (1893), in the fourth edition of his text, brought together three distinct syndromes – “demence precoce” (Morel, 1860), “catatonia” (Kahlbaum, 1874), and “dementia paranoides” (Kraepelin, 1893) under the heading of “psychic degeneration processes”. In the fifth edition (1896) he sub- sumed all psychiatric disorders under two inferential classes: “acquired,” and “con- stitutional.” In the sixth edition (Kraepelin, 1899), the unifying diagnostic concept of “dementia praecox,” for the three syndromes “hebephrenia (Hecker, 1871), “cat- atonia,” and “dementia paranoides” and the all-embracing diagnostic concept of “manic-depressive insanity,” appeared. All the distinct mental syndromes were pooled together and assigned on the basis of their course and outcome to one of these two categories of disease. By the time of the seventh edition (Kraepelin, 1903–1904), the inferential classes of “acquired,” and “constitutional” psychiatric disorders were replaced by 15 disease categories, including “manic-depressive insanity” and “dementia praecox.” Of the remaining 13 disease categories, seven were based on inferences or guesses about their possible causes (“exhaustion psy- choses,” “involutional psychoses,” “paranoia,” “psychogenic neuroses,” “constitu- tional psychopathic states,” “psychopathic personalities,” and “defective mental development”), and six were attributed to organic, including toxic, etiologies (“infection psychoses,” “intoxication psychoses,” “thyrogenous psychoses,” “dementia paralytica,” “organic dementias,” and “epileptic insanity”). In the eighth edition (Kraepelin, 1908–1915), the already broad diagnostic concepts were expanded. “Manic-depressive insanity” incorporated “involutional melancholia” (Dreyfus, 1905) and “all cases of affective excess,” and “dementia praecox” incor- porated Magnan’s (1893) diagnostic concept of “delire chronique” (Kraepelin, 1919, 1921; Pichot, 1983). However, in the same edition, Kraepelin (1908–1915) distinguished between the “paranoid form of dementia praecox” and the “para- phrenias.” By the time Bleuler (1911) coined the term “schizophrenia” to replace the term “dementia praecox,” Kraepelin (1919) recognized 10 different forms of “dementia praecox, i.e., “dementia simplex” (Diem, 1903); “silly deterioration” (replacing the term “hebephrenia”); “depressive deterioration;” “depressive deteri- oration with delusional formation;” “circular,” “agitated,” “periodic,” “catatonic,” and “paranoid” forms; and schizophasia. He also defined nine different end-states of the disease, i.e., “cure”; “cure with defect;” “simple deterioration;” “imbecility
44 T. A. Ban
with confusion of speech;” “hallucinatory deterioration;” “hallucinatory insanity;” “dementia paranoides;” “flighty, silly deterioration;” and “dull, apathetic demen- tia” (Fish, 1962; Hamilton, 1976).
Eugen Bleuler
Kraepelin’s (1903–1904) classification was adopted by Eugen Bleuler (1916) with some minor modifications. By replacing the term “dementia praecox” with the term “schizophrenia,” and redefining schizophrenia as a “group of psychoses” char- acterized “by a specific type of thinking, feeling and relation to the external world” which “appear in no other disease in this particular fashion,” Bleuler (1911) con- solidated the diagnostic concept. His “fundamental” or “basic symptoms” remained for well over 50 years the most extensively employed diagnostic criteria of schizophrenia.
Carl Wernicke
Wernicke’s (1899) “classification of psychoses” appeared in the same year as the sixth edition of Kraepelin’s textbook (1899). It was based on contemporary scien- tific contributions that were to become the structural foundation of neuroscience, e.g., the description of “multipolar cells” in the cerebral cortex (Golgi, 1883), the recognition that the “neuron” is the morphological and functional unit of the nervous system (Ramon y Cajal, 1897–1904), and the demonstration that the “synapse” is the functional site of transmission from one “neuron” to another (Sherrington, 1896–1897, 1906). By adopting Sechenov’s (1866) extension of the concept of the “reflex” as the elementary unit of mental pathology, Wernicke (1900) perceived the different forms of mental illness as “loosening of detachment from the rigid structure of associations,” displayed in “hyperfunctioning,” “hypofunc- tioning,” or “parafunctioning” in the “psychosensory,” “intrapsychic,” or “psycho- motor” component(s) of the “psychic reflex” (Franzek, 1990). Wernicke (1899) was also first to describe “motility psychosis” and “anxiety-happiness psychosis”, to sep- arate memory impairment (“dysmnesia”) from personality deterioration (“demen- tia”), and to divide consciousness into “consciousness of the body” (“somatopsyche”), “consciousness of one’s personality” (“autopsyche”), and “con- sciousness of the external world” (“allopsyche”).
Karl Kleist
Wernicke’s (1900) contributions were further elaborated by Kleist (1921, 1923), who split the diagnostic concept of “schizophrenia” into two groups of diseases: “typical schizophrenias,” which are confined to one neurological system, and “atyp- ical schizophrenias,” which affect many different neurological systems in the central nervous system. Kleist (1928) was first to recognize “cycloid psychoses” as a distinct
45 Genes and psychiatric nosology
nosological category that include Wernicke’s (1900) diagnostic concepts of “anxiety psychosis” and “motility psychosis,” and Kleist’s own diagnostic concept of “confusion psychosis.”
Karl Leonhard
Leonhard (1957) replaced Kleist’s dichotomy (1923) of “typical” and “atypical schizophrenias” with the “polarity”-based dichotomy of “systematic” and “unsystematic schizophrenias.” He adopted Kleist’s (1928) diagnostic concept of “cycloid psychoses” and Neele’s (1949) diagnostic concept of “phasic psychoses” and classified “endogenous psychoses” into five classes of illness: “unipolar phasic psychoses,” “bipolar manic-depressive disease,” “bipolar cycloid psychoses,” “bipolar unsystematic schizophrenias,” and “unipolar systematic schizophrenias.” Furthermore, on the basis of the primarily affected component of the “psychic reflex” – “afferent” (sensory–perceptual–cognitive), “central-intrapsychic” (affec- tive), or “efferent” (motor) – he separated three groups of illnesses within the “systematic schizophrenias” (“paraphrenias,” “hebephrenias,” and “catatonias), the “unsystematic schizophrenias” (“cataphasia” “affect-laden paraphrenia,” and “peri- odic catatonia”), and the “cycloid psychoses (“confusion psychosis,” “anxiety- happiness psychosis,” and “motility psychosis”). He also separated “pure mania” and “pure melancholia,” diseases in which the “mental pathology” extends to all three components of the “psychic reflex,” from the pure “euphorias” and “pure depressions,” in which the “mental pathology” is less pervasive. In Leonhard’s (1936, 1961, 1986) “differentiated nosology” there are 16 “psychopathology”-based syndromes differentiated within the “systematic schizophrenia” group; five “psychopathology”-based syndromes within each of the “pure euphorias” and the “pure depressions;” and three psychopathology-based syndromes within each of the “unsystematic schizophrenias” and the “cycloid psychoses.”
Kurt Schneider
Kurt Schneider’s (1959) rudimentary classification was based on Karl Jaspers’ (1910, 1913, 1962) recognition that mental illness is expressed in the “form” (i.e., the “mode”) in which the experience appears (e.g., “sudden primordial delusional idea put in the mind by hallucinatory voices”) and not in the “content” (e.g, “being persecuted”) of the experience. About two decades prior to Schneider (1950), a group of German psychiatrists at the “Heidelberg Clinic,” began to re-evaluate psychiatric nosology with the employment of Jaspers’ (1913) conceptual frame- work. Their activity ended in 1933 with the removal of Wilmans, the head of the “Clinic” by the Nazi regime. (Gruhle, who was the intellectual leader of the group, left the university and took a position at a provincial psychiatric hospital; Mayer- Gross moved to England, etc.). What was left behind was put together by Schneider
46 T. A. Ban
(1959) into a rudimentary classification in which “developmental anomalies” (i.e., “abnormal variations of mental life”) are separated from the “effects of illness and defective structure.” Included among the “developmental anomalies” are “abnor- mal intellectual endowment,” “abnormal personality,” and “abnormal psychic reac- tion.” Included among the “effects of illness” are the “somatically based psychoses,” “schizophrenia,” and “cyclothymia,” the term Schneider (1959) used for “manic- depressive disease.” Although Schneider (1950) retained the diagnosis of schizo- phrenia, he maintained that “there is nothing to which one can point as a common element in all the clinical pictures that are today christened as schizophrenia.”
Diagnostic instruments for research
There are two diagnostic instruments specially devised to provide more homogen- eous populations for research than the diagnostic categories of consensus-based classifications: these are the Diagnostic Criteria for Research Budapest–Nashville (DCR) and the Composite Diagnostic Evaluation (CODE) system.
DCR Budapest–Nashville
The DCR Budapest-Nashville (Petho et al., 1984, 1988) is an “eclectic classifica- tion,” devised on the basis of theoretical considerations. It is an integration of noso- logic contributions from different schools of psychiatry into a classification in which the original diagnostic concepts are retained.
At the core of the DCR is Leonhard’s (1957) classification of “endogenous psy- choses.” However, the DCR also includes the Scandinavian diagnostic concept of “psychogenic (reactive) psychoses” (Wimmer, 1916; Stromgren, 1974), and a com- posite of the German diagnostic concept of “delusional development” (Gaupp, 1914; Kretschmer, 1927) and the French diagnostic concept of “delusional psycho- ses” (Magnan, 1893; Baruk, 1974).
The decision to adopt Leonhard’s (1957) classification of “endogenous psycho- ses” in the DCR was based on findings in epidemiological genetic and psychophar- macological research that were supportive of Leonhard’s (1957) diagnostic concepts (Ban, 1990; Ban and Ucha Udabe, 1995). The incidence of “nosologic homotypy” was high, 57–77% in the parents and siblings of patients with “cycloid psychoses” (Perris, 1974a; Ungvari, 1985); the concordance for “polarity” was as high as 80% in twin pairs concordant for “mood disorders” (Zerbin-Rudin, 1967; Tsuang and Vandermey, 1980). Morbidity risk for “endogenous psychoses” was higher in the relatives (parents and siblings) of patients with “bipolar affective psy- choses” than in the relatives of patients with “unipolar phasic psychoses” or “cycloid psychoses”; in the relatives of those with “unsystematic schizophrenias” than in the relatives of those with “systematic schizophrenias” or “cycloid psychoses”; and in

47 Genes and psychiatric nosology
the relatives of those with “cycloid psychoses” than in the relatives of those with “systematic schizophrenias” (Trostorff, 1968, 1975; Perris, 1974b). Responsiveness to neuroleptic treatment was higher in the “unsystematic schizophrenias” (79%) than in the “systematic schizophrenias” (22%) (Astrup, 1959; Fish, 1964). The different forms (and subforms) of schizophrenia in Leonhard’s (1957) classification were to become suitable end-points for molecular genetic research, yielding to the demonstration that “periodic catatonia,” one of the three forms of “unsystematic schizophrenia,” is associated with a “major disease locus” that “maps” to chromo- some 15q15 (Stober et al., 2000).
The diagnostic process in the DCR is based on a decision tree model that con- sists of 524 variables organized into 179 diagnostic decision clusters, yielding a total of 213 diagnoses (11 undifferentiated; 37 atypical; 21 tentative; 44 provisional; 45 working; and 55 final). An undifferentiated diagnosis in the DCR implies that patient qualifies for “psychosis” but does not fit any of the DCR diagnoses; an atyp- ical diagnosis implies that the patient fulfills only cross-sectional diagnostic crite- ria of a specific DCR diagnosis.
CODE-system
The CODE system provides a methodology for the detection of the forms of mental illness identified in the different nosologies which are biologically the most homo- geneous. It is a set of diagnostic instruments that can assign simultaneously a diag- nosis from several diagnostic systems to a patient by specially devised algorithms. Each instrument (“CODE”) can provide for a polydiagnostic evaluation in a dis- tinct category of mental illness by employment of an integrated criteria list and standardized data collection. To achieve its purpose, each “CODE” consists of a set of symptoms (“codes”), which yield diagnoses in all the component diagnostic systems; a semistructured interview, suitable for the elicitation of all the symptoms in terms of “present” or “absent”; and diagnostic decision trees, which organize symptoms into distinct psychiatric disorders (Ban, 1991). The CODE system differs from other polydiagnostic systems by the inclusion of all distinct diagnostic formu- lations relevant to the conceptual development of a diagnostic category, and by its capability to provide readily accessible information relevant to the diagnostic process from the lowest to the highest level of decision making.
The prototype of the CODE System is CODE-DD, the CODE for “unipolar depressive disorders” (Ban, 1989). It consists of a 90-item Rating Scale for Depressive Diagnoses (RSDD) with a 40-item subscale (the Rating Scale for the Assessment of Severity in Depressive Disorders); a Semi-Structured Interview for Depressive Disorders, suitable for the elicitation of the presence or absence of the 90 variables of the RSDD; and decision trees, which provide diagnoses within 25 different classifications of depression. Many of the classifications of depression
48 T. A. Ban
included in CODE-DD are empirically derived, e.g., those of Kiloh and Garside (1963) and Winokur (1979). Some are based on the conceptual development of depression in Europe, e.g., Schneider (1920), and others on the conceptual devel- opment of depression in North America, e.g., Robins and Guze (1972). Included also in CODE-DD are consensus-based classifications, e.g., DSM-III-R of the American Psychiatric Association (1987), and diagnostic criteria for research, e.g., Feighner et al. (1972).
One would expect low inter-rater agreement in such a complex system like CODE-DD. However, in the first reliability study there was an 87.8% inter-rater agreement (regarding the presence or absence of) on the 90 RSDD variables (Morey, 1991). In the second, agreement increased to 100% (Ban et al., 1993). In a validation study that included 230 patients with a clinical diagnosis of major depression, there was a 99.6% correspondence between the clinical DSM-III-R and the CODE-DD diagnosis of major depression. In another validation study, which included 322 patients, the correspondence was 97.2% (Ban, 1992).
Nosologic matrix
Whether DCR, CODE-DD, or any other diagnostic instrument in preparation would provide suitable populations for genetic research is not known. Until the time such an instrument becomes available, the use of “nosologic homotypes” is an essential prerequisite for obtaining interpretable findings in research in the genet- ics of mental illness. Nosologic homotypes are identical in “elementary units” of mental illness and are assigned the same position in the “nosologic matrix” con- structed with consideration for the “nosologic organizing principles” of Esquirol (1838), Kraepelin (1896, 1899), and Leonhard (1957) (Ban, 2000b).
The elementary units of mental illness are psychopathologic symptoms (Jaspers, 1913, 1962). Each psychopathologic symptom represents a distinct pathology in the processing of mental events (Wernicke, 1900), and each distinct “psychopatho- logic symptom profile” (syndrome) is a potential “phenotype” of mental disorder. The formal characteristics of the “onset” (sudden or insidious), “course” (episodic or continuous), and “outcome” (recovery or defect) of the mental syndrome reflect the pathological process in its “dynamic totality,” and the “dynamic totality” of the pathological process, together with the “holistic character” (Petho, 1990) of the clinical picture (“monomorphous,” “polymorphous,” “amorphous”), provides a “structure” that is “ determined by the illness” (Ban, 1987). It is in terms of this structure that each mental illness is defined and assigned a distinct place in the “nosologic matrix,” based on the three nosologic organizing principles.
The first organizing principle of psychiatric nosology is the “inclusiveness” of the psychopathological process. Its origin is in Esquirol’s (1838) distinction between

49 Genes and psychiatric nosology
“insanity proper” and “partial insanity.” The prototype of “insanity proper” is Morel’s (1860) “demence precoce,” and the prototype of “partial insanity” is Lasegue’s (1852) “persecutory delusional psychosis.” The concept of “partial insan- ity” (i.e., insanity with preserved personality) was extended to include “abortive insanity.” Patients with “abortive insanity” are fully aware (cognizant) that their thinking, feelings, or actions are pathological. The prototype of “abortive insanity” is Westphal’s (1878) diagnostic concept of “obsessional neurosis.”
The second organizing principle of psychiatric nosology is the “course” and “outcome” of the psychopathological process. Its origin is in Kraepelin’s (1899) separation of “manic-depressive insanity” (an episodic and remitting illness) from “dementia praecox” (a continuous and progressing disease). It is within the frame of reference of the second nosologic organizing principle that “attacks” (episodes with brief duration, from minutes to hours, encountered in “panic disorder”) are distinguished from “phases” (episodes with long duration, from days to years, encountered in the “phasic psychoses”) and from “periods” (phases recurring with regularity, encountered in “seasonal affective disorder”). Similarly, “thrusts” (acute events that yield lasting changes, encountered in the “unsystematic schizophren- ias”) can be distinguished from “continuous process” (chronic events, that yield highly differentiated end-states, encountered in the “systematic schizophrenias”) and from “progressive deterioration” (chronic events that yield increasingly severe “dedifferentiation,” encountered in the “organic dementias”).
The third organizing principle of psychiatric nosology is “polarity.” Its origin is in Leonhard’s (1957) distinction between “polymorphous (multiform) bipolar” and “monomorphous (pure) unipolar” psychiatric disorders. “Bipolar illness” swings between two poles of mood (emotions and motility) and displays a contin- uously changing (variable) clinical picture, whereas “unipolar illness” is restricted to one pole of mood (emotions and motility) and displays the same symptomatol- ogy within and across episodes. Each distinct form of “unipolar illness is charac- terized by a syndrome associated with no other form and not even transitionally related to any other forms.”
While “nosologic homotypes,” based on a specially devised “nosologic matrix,” are biologically more homogeneous populations than any of the diagnostic popu- lation identified by the available diagnostic instruments today, the information col- lected by the use of the “nosologic matrix” would allow completion of the re-evaluation of diagnostic concepts started by the psychiatrists at the “Heidelberg Clinic” in the 1920s. The information collected by the use of the “nosologic matrix” could also serve as the starting point for an empirically derived classification of mental illness.
Considering that “nosologic homotypes” are defined in terms of their effect on “processing of mental events,” and psychotropic drugs are defined in terms of their
50 T. A. Ban
effect on “signal transduction” in the brain (Bloom, 2001), the empirically derived diagnostic categories provide clinical entities which are suitable for testing hypoth- eses relevant to the relationship between “processing of mental events” and “signal transduction “ in the central nervous system. Thus, employment of the “nosologic matrix” could open up a new perspective for the development of a psychiatry in which mental pathology is perceived in terms of pathology in “signal transduction” in the brain, and for the development of a rational pharmacotherapy of mental illness. Within the new frame of reference, genetic research in mental illness would enter a new phase.
Conclusions
The observation that mental illness runs in families received substantial support from findings in epidemiological genetic research. The failure to reveal the mode of inheritance of schizophrenia and manic-depressive disorder with the employ- ment of powerful statistical models led psychiatric geneticists to conclude that the mode of inheritance of these disorders is “complex.” However, the inconsistent, conflicting findings in molecular genetic studies of schizophrenia and manic- depressive illness have brought attention to the fact that populations within these diagnostic categories are genetically heterogenous.
Neuropsychopharmacology provides an adequate methodology for the identifi- cation of suitable, pharmacologically homogeneous forms of illness for genetic research in the different nosologies. Until this is done, the use of “nosologic homo- types,” identified by the employment of a “nosologic matrix,” is an essential pre- requisite in genetic research for obtaining interpretable findings.
The “nosologic matrix” may also serve as the starting point for the development of an empirically derived pharmacologically meaningful classification of mental illness. A classification in which the relationship between mental pathology and the pathology of “signal transduction” in the brain is recognized would open up a new perspective in genetic research and in the pharmacotherapy of mental illness.
REFERENCES
American Psychiatric Association (1980). Diagnostic and Statistical Manual of Mental Disorders, 3rd edn. Washington, DC: American Psychiatric Association.
American Psychiatric Association (1987). Diagnostic and Statistical Manual of Mental Disorders, 3rd edn. Revised. Washington, DC: American Psychiatric Association.
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edn. Washington, DC: American Psychiatric Association.
 
51 Genes and psychiatric nosology
Angst J (1966). Zur Aetiologie und Nosologie endogener depressiver Psychosen. Berlin: Springer. Arolt V, Lencer R, Nolte A et al. (1996). Eye tracking dysfunction is a putative phenotypic sus- ceptibility marker of schizophrenia and maps to a locus on chromosome 6p in families with
multiple occurrence of the disease. Am J Med Genet 67, 564–579.
Arranz M, Collier D, Sodhi M et al. (1995). Association between clozapine response and allelic
variation in 5HT2A receptor gene. Lancet 346, 281– 282.
Astrup C (1959). The effects of ataraxic drugs on schizophrenic subgroups related to experimen-
tal findings. Acta Psychiatr Scand 34(Suppl 136), 388–393.
Ban TA (1987). Prolegomenon to the clinical prerequisite: psychopharmacology and the classifi-
cation of mental disorders. Progr Neuropsychopharmacol Biol Psychiatry 11, 527–580.
Ban TA (1989). CODE-DD Composite Diagnostic Evaluation of Depressive Disorders. Nashville,
TN: JM Productions.
Ban TA (1990). Clinical pharmacology and Leonhard’s classification of endogenous psychoses.
Psychopathology 23, 331–338.
Ban TA (1991). CODE system: theory and practice. In Aguglia E, Ban TA, eds. International
Symposium on Functional Psychoses Today. Rome: John Libbey, pp. 279–296.
Ban TA (1992). Composite diagnostic evaluation: methodology and applications. In Ferrero FP, Haynal AE, Sartorius N, eds. Schizophrenia and Affective Psychoses. Nosology in Contemporary
Psychiatry. Rome: John Libbey, pp. 410–419.
Ban TA (1999). Selective drugs versus heterogenous diagnoses towards a new methodology in
psychopharmacologic research. Psiquiatr Biol 7, 177–189.
Ban TA (2000a). From DSM-III to DSM-IV: progress or standstill? In Franzek E, Ungvari GS,
Ruther E, Beckmann H, eds. Progress in Differentiated Psychopathology. Hong Kong:
Contemporary Development Company, pp. 1–11.
Ban TA (2000b). Nosology in the teaching of psychiatry. J Bras Psiquiatr 49, 39–49.
Ban TA and Ucha Udabe R (1995). Clasificacion de las Psicosis. Buenos Aires: Editorial Salerno. Ban TA, Fjetland OK, Kutcher M, Morey LC (1993). CODE-DD development of a diagnostic
scale for depressive disorders. In Hindmarch I, Stonier PD, eds. Human Psychopharmacology.
Measures and Methods, Vol. 4. Chichester, UK: John Wiley & Sons, pp. 73–85.
Baruk H (1974). Delusions of passion. [Translated by Rohde M.] In Hirsch SR, Shepherd M, eds. Themes and Variations in European Psychiatry. Charlottesville, VA: University Press of Virginia,
pp. 375–383.
Battie W (1758). A Treatise of Madness. London: Whiston.
Bleuler E (1911). Dementia praecox oder gruppe der Schizophrenien. Leipzig: Deuticke.
Bleuler E (1916). Lehrbuch der Psychiatrie. Berlin: Springer.
Bloom F (2001). Personal perpectives. Am Coll Neuropsychopharm Bull 7, 4–8.
Boissier de Sauvages (1768). Nosologia Methodica. Amsterdam: Frat de Tournes.
Chiarugi V (1793–1794). On Insanity and its Classification. (Republished in English, 1989).
Canton, MA: Science History Publishers.
Craddock N, Khodel V, van Eerdewegh P, Reich T (1995). Mathematical limits of multilocus
models: the genetic transmission of bipolar disorder. Am J Hum Genet 57, 690–702.
Diem O (1903). Die einfach demente Form der Dementia Praecox (Dementia Simplex). Arch für
Psychiatr Nervenkrank 37, 111–187.
52 T. A. Ban
Dreyfus GL (1905). Die Melancholie ein Zustandsbild des Manisch-Depressiven Irreseins. Jena: Gustav Fischer.
Esquirol JED (1838). Des Maladies Mentales Considerees Sous les Raports Medical, Hygienique et Medico-Legal. Paris: JP. Baillière.
Faraone SV, Tsuang MT, Tsuang DW (1999). Genetics of Mental Disorders. New York: Guilford Press.
Feighner JP, Robins E, Guze B, Woodruff JP, Winokur G, Munoz R (1972). Diagnostic criteria for use in psychiatric research. Arch Gen Psychiatry 26, 57–63.
Fish F (1962). Schizophrenia. Bristol: John Wright.
Fish F (1964). The influence of the tranquilizer on the Leonhard schizophrenic syndromes.
Encephale 53, 245–249.
Franzek E (1990). Influence of Carl Wernicke on Karl Leonhard’s nosology. Psychopathology 23,
277–281.
Freedman R, Coon H, Myles-Worsley M et al. (1997). Linkage of a neurophysiological deficit in
schizophrenia to a chromosome 15 locus. Proc Natl Acad Sci, USA 94, 587–592.
Freedman R, Adler LE, Leonard S (1999). Alternative phenotypes for the complex genetics of
schizophrenia. Biol Psychiatry 45, 551– 558.
Garrison FH (1929). An Introduction to the History of Medicine, 4th edn (reprinted 1960).
Philadelphia, PA: Saunders.
Gaupp R (1914). Die Wissenschaftliche Bedeutung des Falles Wagner. Munch Med Wschr 61,
633–637.
Gershon ES, Badner J, Goldin LR, Sanders AR, Cravchik A, Detera-Wadleigh SD (1998). Closing in
on genes for manic-depressive illness and schizophrenia. Neuropsychopharmacology 18, 233–242. Golgi C (1883). Recherches sur l’histologie des centres nerveux, Arch Ital Biol Turin 3, 285–317. Gottesman II, Shields J (1966). Contributions of twin studies to perspectives on schizophrenia.
In Maher BA, ed. Progress in Experimental Personality Research. New York: Academic Press, pp.
1–84.
Hamilton M (1976). Fish’s Schizophrenia, 2nd edn. Bristol: John Wright.
Hecker E (1871). Die Hebephrenie. Arch Pathol Anat Physiol Klin Med 52, 394–409.
Heiden A, Schussler P, Itzlinger U et al. (2000). Association studies of candidate genes in bipolar
disorder. Neuropsychobiology 42(Suppl 1), 18–21.
Heston LL (1966). Psychiatric disorders in foster home reared children of schizophrenic mothers.
Br J Psychiatr 112, 819– 825.
Holzman PS, Kringlen E, Mathysse S et al. (1988). A single dominant gene can account for eye
tracking dysfunctions and schizophrenia in offspring of discordant twins. Arch Gen Psychiatry
45, 641–647.
Hyman SE (1999). Introduction to the complex genetics of mental disorders. Biol Psychiatry 45,
518–521.
Jaspers K (1910). Eifersuchtswahn: Entwicklung einer Persoenlichkeit oder Prozess. Z Ges Neurol
Psychiatr 1, 567–637.
Jaspers K (1913). Allgemeine Psychopathologie. Springer: Berlin.
Jaspers K (1962). General Psychopathology. [Translated from the German 7th edn by Hoenig J,
Hamilton MW.] Manchester, UK: Manchester University Press.
53 Genes and psychiatric nosology
Kahlbaum KL (1863). Die Gruppierung der psychischen Krankheiten und die Enteilung der Seelenstoerungen. Danzig: AW Kaufman.
Kahlbaum K (1874). Die Katatonie oder das Spannungsirresein. Berlin: Hirschwald.
Kay DWK (1978). Assessment of familial risks in the functional psychoses and their application
in genetic counselling. Br J Psychiatr 133, 385–403.
Kety SS, Wender PH, Jacobsen B et al. (1994) Mental illness in the biological and adoptive rela-
tives of schizophrenic adoptees. Replication of the Copenhagen Study in the rest of Denmark.
Arch Gen Psychiatry 51, 442–455.
Kiloh LG, Garside RF (1963). The independence of neurotic depression and endogenous depres-
sion. Br J Psychiatry 109, 451–463.
Kleist K (1921). Autochtone Degenerationspsychosen. Z Ges Neurol Psychiatr 69, 1–11.
Kleist K (1923). Die Auffassung der Schizophrenien als Systemkrankheiten. Klin J Wochenschr 2,
962–965.
Kleist K (1928). Uber zykloide, paranoide und epileptoide psychosen und uber die Frage der
Degenerationspsychosen. Schweiz Arch Neurol Psychiatr 23, 1–35.
Kraepelin E (1883). Compendium der Psychiatrie. Leipzig: Barth.
Kraepelin E (1886). Compendium der Psychiatrie, 2nd edn. Leipzig: Barth.
Kraepelin E (1889). Compendium der Psychiatrie, 3rd edn. Leipzig: Barth.
Kraepelin E (1893). Psychiatrie, Ein Lehrbuch für Studierende und Arzte, 4th edn. Barth: Leipzig. Kraepelin E (1896). Psychiatrie, Ein Lehrbuch für Studierende und Arzte, 5th edn. Barth:
Leipzig.
Kraepelin E (1899). Psychiatrie, Ein Lehrbuch für Studierende und Arzte, 6th edn. Barth: Leipzig. Kraepelin E (1903–1904). Psychiatrie, Ein Lehrbuch für Studierende und Arzte, 7th edn. Barth:
Leipzig.
Kraepelin E (1908–1915). Psychiatrie, Ein Lehrbuch für Studierende und Arzte, 8th edn. Barth:
Leipzig.
Kraepelin E (1919). Dementia Praecox and Paraphrenia. [Translated by Barclay RM.] Edinburgh:
Livingstone.
Kraepelin E (1921). Manic Depressive Insanity and Paranoia. [Translated by Barclay RM.]
Edinburgh: Livingstone.
Kretschmer E (1927). Der sensitive Beziehungswahn. Berlin: Springer.
Kuhn R (1957). Uber die Behandlung depressives Zustande mit einem iminodibenzyl-derivat
(G22355). Schweiz Med Wochenschr 87, 1135–1140.
Lander ES (1988). Splitting schizophrenia. Nature 336, 105–106.
Lander ES, Schork NI (1994). Genetic dissection of complex traits. Science 265, 2037–2038. Lasegue EC (1852). Du delire de persecution. Arch Gen Med 28, 129–150.
Leonhard K (1936). Die Defektschizophrenen Krankheitsbilder. Leipzig: Thieme.
Leonhard K (1957). Aufteilung der endogenen Psychosen. Berlin: Akademie-Verlag.
Leonhard K (1961). Cycloid psychoses – endogenous psychoses which are neither schizophrenic
nor manic-depressive. J Ment Sci 107, 633–648.
Leonhard K (1986). Aufteilung der endogenen Psychosen und ihre differenzierte Aetiologie, 6th edn
revised. Berlin: Akademie-Verlag.
Magnan V (1893). Leconc Cliniques sur les Maladies Mentales, 2nd edn. Paris: Bataille.
54 T. A. Ban
Magnan V (1896). Considerations generales sur la folie (des hereditaires ou degeneres). Prog Med 2, 108–110.
Malhotra AK, Goldman D (1999). Benefits and pitfalls encountered in psychiatric genetic asso- ciation studies. Biol Psychiatry 45, 544– 550.
Malhotra AK, Goldman D, Mazzani C, Clifton A, Breier A, Pickar D (1996). Lack of association between polymorphism in the 5-HT2A receptor gene and the antipsychotic response to cloz- apine. Am J Psychiatry 153, 1092–1094.
Mendlewicz J, Rainer JD (1977). Adoption study supporting genetic transmission in manic- depressive illness. Nature 268, 326–329.
Moebius JP (1893). Abriss der Lehre im den Nervenkrankheiten. Leipzig: Barth.
Moldin S (1999). Report of the National Institute of Mental Health’s Genetic Workgroup.
Summary of research. Biol Psychiatry 45, 573–602.
Morel BA (1857). Traite de Degenerescences Physiques, Intellectuelles et Morales de l’Espace
Humaine. Paris: Baillière.
Morel BA (1860). Traite des Maladies Mentales. Paris: Mason.
Morey LC (1991). Reliability considerations in the development of CODE-DD. In Aguglia E,
Ban TA, eds. International Symposium on Functional Psychoses Today. Rome: John Libbey, pp.
297–304.
Neele E (1949). Die Phasischen Psychosen. Leipzig: JA Barth.
O’Rourke DH, Gottesman II, Suarez BK, Rice JP, Reich T (1982). Refutation of the general single
locus model for the etiology of schizophrenia. Am J Hum Genet 34, 630–649.
Overall J, Hollister LE, Johnson M, Pennington V (1966). Nosology of depression and differen-
tial response to drugs. J Am Med Assoc 195, 946–948.
Perris C (1974a). A study of cycloid psychoses. Acta Psychiatr Scand 50 (Suppl), 253.
Perris C (1974b). The genetics of affective disorders. In Mendels J, ed. Biological Psychiatry. New
York: John Wiley, pp. 385–415.
Petho B (1990). Development and structure of the DCR Budapest–Nashville. Psychopathology 23,
316–330.
Petho B, Ban TA, Kelemen A et al. (1984). KDK Budapest Kutatasi Diagnosztikai Kriteriumok
funkcionalis psychosisok korismezesehez. Ideggyogyaszati Szemle 37, 102–131.
Petho B, Ban TA in collaboration with Kelemen A, Ungvari G, Karczag I, Bitter I, Tolna Y (Semmelweis Medical University, Budapest), Jarema M, Ferrero F, Aguglia E, Zurria GL, Fjetland OK (Vanderbilt University, Nashville) (1988). DCR Budapest–Nashville in the
Diagnosis and Classification of Functional Psychoses. Basel: Karger.
Petronis A, Kennedy JL (1995). Unstable genes – unstable mind. Am J Psychiatry 152,
164–172.
Pichot P (1983). A Century of Psychiatry. Paris: Editions Roger Dacosta.
Pinel P (1799). Nosographie Philosophique ou La Methode de l’Analyse Appliquee a la Médicine.
Paris: Brosson.
Pinel P (1801). Traite Médico-Philosphique sur l’Alienation Mentale ou la Manie. Paris:
Brosson.
Ramon y Cajal S (1897–1904). Histology of the Nervous System of Man and Vertebrates. [Original
in Spanish; republished 1995 in English.] New York: Oxford University Press.
55 Genes and psychiatric nosology
Robins E, Guze SB (1972). Classification of affective disorders: the primary–secondary, endoge- nous–reactive and the neurotic–psychotic concepts. In Williams A, Katz MM, Shields JA, eds. Recent Advances in Psychobiology of the Depressive Illnesses. Washington, DC: US Government Printing Office, pp. 283–293.
Schneider K (1920). Die Schichtung des emotionalen Lebens und der Aufbau der Depressions zustande. Z Ges Neurol Psychiatr 59, 281–285.
Schneider K (1950). Klinische Psychopathologie, 3rd edn. Stuttgart: Thieme.
Schneider K (1959). Clinical Psychopathology. [Translated by Hamilton MW from the 5th revised
edition of the German original.] New York: Grune & Stratton.
Sechenov IM (1866). Refleksy Golovnogo Mozga. St Petersburg.
Sherrington CS (1896–1897). On reciprocal innervation of antagonistic muscles. Proc Roy Soc
Lond 9, 414–417.
Sherrington CS (1906). The Integrative Action of the Nervous System. London: Scribner.
Shorter E (1997). A History of Psychiatry. New York: John Wiley.
Slater E, Cowie V (1971). The Genetics of Mental Disorder. London: Oxford University Press. Stober G, Saar K, Ruschendorf F et al. (2000). Splitting schizophrenia: periodic catatonia – sus-
ceptibility locus on chromosome 15q15. Am J Hum Genet 67, 1201–1207.
Stromgren E (1974). Psychogenic psychosis. In Hirsch S, Shepherd M, eds. Themes an Variations
in European Psychiatry. Charlottesville: University Press of Virginia, pp. 97–117.
Trostorff S (1968). Uber die hereditare Belastung bei den zykloiden Psychosen und den unsystematischen und systematischen Schizophrenien. Psychiatr Neurol Med Psychol 20,
98–106.
Trostorff S (1975). Verlauf und psychose in der verwandtshaft beiden systematischen und
unsystematischen Schizophrenien und den zykloiden Psychosen. Psychiatr Neurol Med Psychol
(Leipzig) 27, 80–100.
Tsuang MT, Faraone SV (1990). The Genetics of Mood Disorders. Baltimore, MD: Johns Hopkins
University Press.
Tsuang T, Faraone SV (1995). The case for heterogeneity in the etiology of schizophrenia.
Schizophr Res 17, 161–175.
Tsuang MT, Vandermey R (1980). Genes and the Mind. Inheritance of Mental Illness. Oxford:
Oxford University Press.
Ungvari G (1985). A contribution to the validity of Leonhard’s classification of endogenous psy-
choses. Acta Psychiatr Scand 72, 144–149.
van Praag HM (1992). Reconquest of the subjective . Against the waning of psychiatric diagnos-
ing. Br J Psychiatry 160, 266–271.
van Praag HM (2000). Nosologomania: a disorder of psychiatry. World Biol Psychiatry 1,
151–158.
Wender PH, Kety SS, Rosenthal D, Schulsinger F, Ortman J, Lunde I (1986). Psychiatric disorders
in the biological adoptive families of adopted individuals with affective disorders. Arch Gen
Psychiatry 43, 923–929.
Wernicke C (1899). Uber die Klassifikation der Psychosen. Breslau: Schlettersche
Buchhandlung.
Wernicke C (1900). Grundriss der Psychiatrie in Klinischen Vorlesungen. Leipzig: Thieme.
56 T. A. Ban
Westphal C (1878). Ueber Zwangsvorstellungen. Arch Psychiatr Nervenkrank 8, 734–750. Wimmer A (1916). Psykogene Syndssygdomsformer. Copenhagen: Lunds.
Winokur G (1979). Unipolar depression. Is it divisible into autonomous subtypes. Arch Gen
Psychiatry 36, 47–52.
Zerbin-Rudin E (1967). Endogene Psychosen. In Becker PE, ed. Humangenetik ein Kurzes
Handbuch, Vol. 2. Stuttgart: Thieme, pp. 446–577.
4
Methodological issues in psychopharmacogenetics
Sheldon H. Preskorn
Psychiatric Research Institute, University of Kansas School of Medicine, Wichita, Kansas, USA
OVERVIEW
This chapter will discuss methodological issues relevant to conducting pharmacogenetic studies in clinical psychopharmacology. First, basic pharmacologic principles will be reviewed, followed by a discussion of the drug development process. Last, the research linking cytochrome P-450 CPY2D6 poor metabolizers with increased risk of toxicity on routine doses of tricyclic antidepressants will be reviewed as an example of the methodo- logical issues encountered when conducting such studies.
Introduction
As explained in the introductory chapter, pharmacogenetics is the study of geneti- cally determined interindividual differences in the response to a drug.
This chapter will consider methodological issues that can impinge on the success of pharmacogenetic studies of psychiatric medications. As background for this dis- cussion, a number of elementary pharmacological principles will be reviewed in terms of their importance to both the drug development process and the clinical use of psychiatric medications. For additional information of these topics, the reader is referred to http://www.preskorn.com.
The effect of any drug is a function of three variables, as illustrated in Equation 4.1:
Effect Affinity for site of action Drug concentration
Biological variance among individuals (4.1)
The potential effect of a drug (i.e., the product of the three variables in Equation 4.1) can be divided into three major categories: toxicity, intolerability, or efficacy. For any investigational drug to reach the market, the manufacturer (i.e., the phar- maceutical company) and the responsible regulatory agency (e.g., the Food and Drug Administration (FDA) in the USA) must decide that the efficacy of the drug

57

58 S. H. Preskorn
outweighs its potential for causing toxicity or tolerability problems. The drug does not have to be absolutely “safe” but simply its efficacy must outweigh its risk for toxicity. Safety considerations weigh against the risk associated with not treating effectively the underlying disease. It is easy to see why there are different safety con- siderations for a drug that is used to treat a condition that is not life threatening and is self-limited (e.g., seasonal allergies) than for one used in a fulminant and life- threatening condition (e.g., Gram-negative shock or poorly differentiated squa- mous cell carcinoma of the lung). To be marketed, a drug must be sufficiently safe that its likelihood of causing serious toxicity or tolerability problems is outweighed by its ability to produce efficacy in the population with the target disease.
For any drug to be effective in a given disorder, it must affect a site of action rel- evant to the pathophysiology underlying the disorder. For the vast majority of drugs (over 95%), their site of action is a regulatory protein such as a receptor, an enzyme, or an uptake pump. Affinity for the site of action in Equation 4.1 is a nec- essary condition for a drug to be effective in a specific disorder.
However, affinity for site of action is not, in and of itself, a sufficient condition for efficacy; instead an adequate concentration of the drug must reach this target to affect it to a physiologically meaningful degree. That fact is represented by the second variable in Equation 4.1. Of note, the factors that determine drug concen- tration are summarized in Equation 4.2:
Drug concentration (ng/ml) Dosing rate (mg/day)
Clearance (ml/min) (4.2)
Clearance of most drugs is determined by the rate of their oxidative metabolism. That, in turn, is determined by the affinity of the drug for specific drug- metabolizing enzymes, primarily cytochrome P-450 (CYP) enzymes, and the speed and capacity of these enzymes to biotransform the drug into polar metabolites that can be eliminated, usually by the kidneys.
The third variable in Equation 4.1 is the focus of this book. It is the interindivid- ual differences among patients that can shift the dose–response curve for one indi- vidual relative to another making them either “sensitive” (i.e., more effect than expected for the dose given) or “resistant” (i.e., less effect than expected for the dose given) to a specific drug. There are a number of factors that account for interindi- vidual variability in response:
• diagnosis
• age
• intercurrent disease
• concomitantly administered drugs or dietary substances
• genetics
The focus of this book is genetic factors only.

59 Methodological issues in psychopharmacogenetics
Pharmacogenetics and variables affecting drug action
At the beginning of the twenty-first century, conventional clinical trials are primar- ily focused on variables 1 and 2 in Equation 4.1. Drug discovery is focused on var- iable 1. The goal in discovery is to develop a new chemical entity that affects the desired site(s) of action in the desired way (i.e., agonism, antagonism, or inverse agonism) (Preskorn, 2000). (See http://www.preskorn.com for this and related articles.)
The goal of clinical testing is to determine the “usual” dose needed for the “usual” patient in the population enrolled in the study. As illustrated in Equation 4.2, that dose, in turn, determines the “usual” concentration that the “usual” patient enrolled in the trial will achieved. Based on Equation 4.1, that “usual” concentra- tion, in turn, determines the “usual” degree of occupancy or activity of the drug at its site(s) of action. The goal, then, of human testing of an investigational new drug is to determine the “usual” dose in the “usual” patient enrolled in the trial needed to affect the desired site(s) of action to the right degree without encountering tol- erability or safety problems, caused either by affecting the desired site of action to an undesired degree or by affecting an undesired site(s) of action to a physiolog- ically meaningful degree. From this perspective, clinical trials are, in essence, pop- ulation pharmacokinetic studies.
Of relevance to this chapter and book, clinical trials carried out for drug registra- tion have strict inclusion and exclusion criteria. The goal of these criteria is to narrow the population eligible to participate in the study to those who are likely to respond to the treatment and who are not likely to have serious adverse events. From a phar- macogenetic perspective, those criteria are designed to try to make the study popu- lation as homogeneous as is possible using clinical assessment and routine laboratory testing. As a result of these inclusion and exclusion criteria, the “usual” patient enrolled in most registration trials of antidepressants can differ appreciably from the “usual” patient in clinical practice, particularly those being referred to a psychiatrist. The major difference is the exclusion of patients with significant comor- bidity in terms of both psychiatric and general medical disorders. At least some of those comorbid conditions are likely to be in part genetically determined. That raises the possibility that pharmacogenetic differences exist between the clinical trial pop- ulation for drug registration and the populations treated by some practitioners.
For most drugs, regulatory proteins are the important mechanisms determining variables 1 and 2 in Equation 4.1. Most drugs have a regulatory protein as their site(s) of action (variable 1). Most drugs require oxidative drug metabolism by an enzyme (i.e., a regulatory protein) as a necessary step in their clearance from the body (variable 2). The expression of these proteins is dependent on the genetic make-up of the individual. Thus, pharmacogenetic differences may either affect the site of action of the drug or the mechanism mediating its clearance from the body.

60 S. H. Preskorn
100
75
50
25
0
33% (1/3)
Ceiling effect
Floor effect
33% (1/3)
   
Fig. 4.1.
Rule of thirds in antidepressant clinical trials. Response in such trials is commonly defined as a 50% reduction in symptom severity. A common finding is that one third of patients respond on placebo, two thirds on drug, and one third do not respond because of lack of efficacy, loss to follow-up, or early termination because of intolerable adverse effects. The drug-specific response is therefore one out of three patients (i.e., overall drug response minus placebo response). (Reproduced with permission from http://www.preskorn.com)
Stated another way, pharmacogenetically determined differences may be either pharmacokinetic or pharmacodynamic. That is fundamental to understanding the role of pharmacogenetics in determining the response of a specific individual to a drug in relationship to the response of the general population.
33% (1/3)
Placebo response
Drug-specific response
Nonresponse

Efficacy: a poor signal:noise ratio in clinical psychopharmacology
An important methodological issue of relevance to this chapter is the poor signal:noise ratio which plagues most efficacy trials of psychiatric medications. For example, the results of most published antidepressant clinical trials follow the “rule of thirds” (Fig. 4.1): one third of patients enrolled in such studies respond to placebo, one third respond specifically to the antidepressant (i.e., the two third overall response on the drug minus the one third placebo response rate) and one third do not respond regardless of the dose of the antidepressant (Preskorn, 1998a). As a result, the signal:noise ratio in these trials is 1:2: the one third who are respond- ing specifically because of the drug versus the two thirds who are either respond- ing to nondrug factors inherent in the study (e.g., time or supportive psychotherapy (placebo responders)) or who will not respond to the drug. This should be compared with most mature areas of science, where the goal is to have a minimum signal:noise ratio of 10:1.

Response (%)
61 Methodological issues in psychopharmacogenetics
As discussed below, pharmacogenomic approaches to drug development coupled with parallel pharmacogenetic testing to select specifically responsive pop- ulations offers significant promise for improving the signal:noise ratio in clinical psychopharmacology trials in the future. However, the current poor signal:noise ratio means that drugs must have broad effects on brain function to have any chance of showing efficacy relative to placebo. This poor signal:noise ratio is also the reason that two antidepressant clinical trials generally have to be done to get one positive study. Parenthetically, only the positive studies are published; as a result, the existing literature represents a skewed version of the actual data about the efficacy of existing antidepressants. If taken together, a summary of all the clin- ical trials (published and unpublished) would suggest that current antidepressants produce a specific response in less than one third of patients. As discussed below, this is the reason why studies trying to correlate plasma drug level of antidepress- ants versus efficacy are virtually guaranteed to fail: there is not sufficient signal:noise power to demonstrate such a relationship even though it obviously must exist based on Equation 4.1 and the basic pharmacological principles under- lying that equation.
Given the above background, this chapter will now describe methodological issues that must be addressed to realize the promise that deciphering the human genome has for explaining interindividual differences in drug response and for the optimal practice of clinical psychopharmacology.
Drug development: screening out pharmacogenetic differences
Because of the way in which drug development processes work, pharmacogenetic findings are likely to be limited. First, the goal and intent of the inclusions and exclusion criteria is, in part, to make the study population as homogeneous as pos- sible. Second, the drug would likely fail to receive approval if there were relatively sizable (i.e., 10%) subsets of this presumably homogeneous group with genet- ically determined differences in their dose–response curves in terms of efficacy or adverse effects.
As alluded to earlier in this chapter, one area that could be fruitful would be the discovery of a pharmacogenetic difference in patients with one of the comorbid conditions (e.g., alcohol abuse) that is generally an exclusion criterion for partici- pation in a clinical trials of psychiatric medications. The discovery of such a phar- macogenetic difference could have significant implications for the use of the drug in that population because that population would have been excluded from the systematic studies that lead to the approval of the drug. Conceivably, this pharma- cogenetic difference could mark a population that is more or less responsive to either the beneficial or the adverse effects of the drug.

62 S. H. Preskorn
Theoretically, pharmacogenetic studies of existing drugs could also find subsets in the clinical trial population that have reduced or heightened efficacy on the drug. The identification of a genetically determined subset with reduced efficacy is pos- sible, but, would still be a subset of the one third of the patients enrolled in antide- pressant clinical trials who do not respond to current antidepressants. The identification of a subset who are uniquely responsive to the drug is unlikely because of the small fraction of specific drug responders (i.e., one third or less) in the clinical trial population. In fact, drug development as currently practiced is prone to screen out drugs that may be uniquely safe, well tolerated, and/or effica- cious in a genetically determined subset of the general (i.e., wild) population with the disease of interest unless that subset is so large that their unique responsiveness is not washed out by nonresponse in the rest of the population. For example, assume that a researcher is following a “top down” approach and has discovered a variant regulatory protein that he/she believes is etiopathologically important in a subset (less than 15%) of patients suffering from a psychiatric syndrome such as major depression. It is unlikely that this variant regulatory protein will be the site of action of currently marketed drugs because the signal from such a population would have been too small (15:85) to have been detected in the clinical trials that would have been done to gain registration approval for the drug. Ironically, the researcher might have better success screening investigational antidepressants that failed in earlier development attempts for activity at the new target, because the effi- cacy signal of those drugs was insufficient to gain registration approval. An even more fruitful approach would be to develop new drugs stereospecifically to affect this new site of action (variable 1 in Equation 4.1) and also to develop a means of screening for the presence of the variant and then requiring that participants must have this variant to be included in the study (see below).
Another possibility is the identification of genetically determined subpopula- tions less sensitive to the toxicity or tolerability problems caused by the drug. While such a discovery would likely have basic science implications, it is not likely to be clinically useful because this population does not limit the usefulness of the drug or its appropriateness. In other words, the drug is still useful and appropriate for this subpopulation. For that reason, the remainder of this chapter will not consider this scenario but instead will focus on pharmacogenetic studies of existing drugs in which the pharmacogenetic difference causes the drug to be either not useful or not appropriate through heightened risk of toxicity or tolerability problems.
Even in these situations, the results are likely to be limited to a small subset of the clinical trial populations because their heightened sensitivity to tolerability prob- lems and/or toxicity did not prevent the drug from gaining approval. As discussed later in this chapter, the first successful pharmacogenetic finding in psychiatry involved the discovery of just such a subset of the population: individuals who were
63 Methodological issues in psychopharmacogenetics
at increased risk of serious toxicity with tricyclic antidepressants (TCAs) as a result of being genetically deficient in cytochrome P-450 CYP2D6, which preferentially metabolizes these drugs.
Parenthetically, scientists are using this and other related pharmacogenetic dis- coveries to new chemical entities (NCEs) for subsequent drug development. For example, scientists in drug discovery now use high-throughput screening to deter- mine which CYP enzymes mediate the metabolism of their NCEs. If genetically polymorphic CYP enzymes such as CYP2C19 or CYP2D6 metabolizes the NCE, then that NCE will likely be dropped for future development. The drug discoverers will instead add that finding to their structure–activity relationship paradigm and use it to develop other lead compounds that have desired action on the desired site without being metabolized by that polymorphic CYP enzyme. In essence, drug dis- coverers are using pharmacogenetic knowledge to screen out drugs that may pose problems in specific pharmacogenetically determined populations.
By using these techniques to refine the structure of the drug, their goal is to take variable 3 (including pharmacogenetic differences) out of Equation 4.1 as a factor determining the effect of the drug in the general human population. The question is whether they may unwittingly also be screening out drugs with unique efficacy or tolerability features. In essence, pharmacogenetics is limiting drug developments by raising the bar of what characteristics a drug must not have to be developed.
Future psychiatric drug development: screening in pharmacogenetic
differences
The above situation will further evolve when the knowledge of the human genome uncovers variants (i.e., genetic mutations) in proteins that are etiologically respon- sible for disease in a sufficiently large number of people in the population to make drug development a commercially viable proposition. Ironically, some subsets of what are now considered to be common diseases (e.g., major depression) may become “orphan” areas for drug development.
To understand the last statement, consider that major depression is likely many different diseases when understood from an etiopathological point of view. Some of those different conditions may be caused by genetic mutations. In this scenario, what is now a common disease affecting a substantial percentage of the general population may become multiple diseases with each subtype affecting only a small portion of the group presently defined as suffering from clinical depression. Some of those genetically determined subsets may be so small as to make it questionable whether drug development for that subset would be commercially viable. From this perspective, pharmacogenetics has the potential to limit the development of psychiatric drugs for some subtypes of what are now considered common diseases.

64 S. H. Preskorn
Nevertheless, subsets will be found that are sufficiently large to support the com- mercial drug development specifically targeted at that subset. In fact, pharmaco- genetics and pharmacogenomics will facilitate and streamline drug development, thus reducing costs and increasing the likelihood of success. As an example, say that a specific form of clinical depression is mediated by a variant of the serotonin uptake pump. That variant now becomes the site of action in Equation 4.1. The site can be expressed and its three-dimensional configuration deciphered. Armed with that knowledge, medicinal chemists can develop more precise structure–activity relationships to design a NCE which stereospecifically interacts with that variant of the wild-type regulatory protein. That is an example of the so-called top-down approach in which a genetic variant is defined and then one works downward to determine whether it mediates a clinically useful difference in outcome.
In addition to aiding in the drug discovery process, such a pharmacogenetic finding would also help with the signal:noise problem discussed earlier as, at the human testing phase of drug development (i.e., phase I through III), scientists can select the specific subset of the population (i.e., those with the genetic variant) likely to respond to the drug. Using this strategy, the inclusion criteria for the clin- ical trials of this new drug would require that a subject has the genetic variant to be eligible to participate in the study. That would enrich the study sample with indi- viduals likely to be responsive to the drug. That alone could flip the signal:noise ratio around to 2:1, by reducing, if not eliminating altogether, nonresponders. The signal:noise ratio would still be limited by the problem of nonspecific (placebo) response unless this variant population also had a smaller nonspecific response rate. Thus, the discovery of such a pharmacogenetic variant could decrease the cost and speed up the drug development process by focusing the studies and reducing the numbers needed to prove efficacy relative to placebo compared with those needed in current studies with a signal:noise ratio of 1:2.
Pharmacogenetic research in psychiatry could further enhance the signal:noise ratio by finding a genetically determined subset of patients suffering from a psychi- atric syndrome who are prone to placebo response. That finding might well have some basic science etiopathological implications. More germane to this chapter, that finding would provide a genetic test to exclude the placebo-responsive patients from enrollment.
Consequently, pharmacogenetic tests could be used to reduce both ends of the current signal:noise ratio problem in clinical psychopharmacology trials. One test could be used as an inclusion criterion, requiring that those who enter the study have the variant that predicts specific responsiveness to the drug. Another test could be done to exclude the subset of the population genetically predisposed to placebo response. That could make drug development less costly, faster, and less of a gamble. At the same time, these tests and population limitations would be reflected
65 Methodological issues in psychopharmacogenetics
in the package insert for the drug and could become part of clinical practice as a condition of prescribing the drug. The question would remain for the developers whether the new product would be commercially viable with these limitations on its potential market size. For example, the development of a test that identifies the one third of the clinical trial population who respond to placebo would provide a test to exclude drug therapy for this segment of the population with the syndrome. If that one third was translated to the general population in clinical practice who take antidepressants, it would have a significant effect on the size of the market.
Pharmacogenetic studies and existing drugs
As mentioned at the beginning of this chapter, there are three main categories of drug outcome (i.e., effect in Equation 4.1). The drug may be toxic, intolerable, or efficacious. The pharmacogenetic problem with the drug may be mediated by its pharmacodynamics (variable 1 in Equation 4.1) or its pharmacokinetics (variable 2 in Equation 4.1). Pharmacogenetic differences mediating variable 2 are much easier to study at the present time than those mediating variable 1.
The reason is twofold. First, the pharmacogenetic differences underlying vari- able 2 are better understood. The mechanisms accounting for variable 2 (i.e., reg- ulatory proteins such as CYP enzymes) are fewer in number and easier to study than those accounting for variable 1, which for psychiatric medications are a vast number of regulatory proteins in the brain, an organ that is still difficult to study mechanistically.
Variable 2 mechanisms lend themselves to relatively simple in vitro techniques (Beaune and Guengerich, 1988). The results of such in vitro studies are readily extrapolated to in vivo predictions that can then be readily tested in humans (von Moltke et al., 1998; Schmider et al., 1999). For in vitro studies of CYP enzymes, the drug in question can be screened against all known human enzymes by high throughput methods to determine the affinity of the drug for the different enzymes and their speed and capacity for biotransforming the drug. That information can be used to predict which CYP enzyme(s) will likely metabolize it under usual dosing conditions. If the rate-limiting enzyme mediating the oxidative metabolism of the drug is polymorphic, then the drug is likely to have altered clearance in that subset of the population. That can then be tested by measuring the clearance of the drug in a sample of the wild-type population and a sample of the population with the variant enzyme. In this way, the human study becomes a direct extension of the in vitro study. Such screening is now being used to screen out NCEs from further drug development if they may require a change in dosage for the poor metaboliz- ing subpopulation and hence the drug will not meet the desired criteria of having one dose for all patients.

66 S. H. Preskorn
CYP2D6 as an example
CYP2D6 is a polymorphic enzyme that has played a major role in the understand- ing of the potential impact of pharmacogenetics on the effect of psychiatric medi- cations. This CYP enzyme was a classic example of the bottom up approach in which an aberrant response (i.e., increased sensitivity to the dose-dependent adverse effects of a specific drug) was identified, leading to a search for the mechanism that accounted for the increased sensitivity. In the case of CYP2D6, several examples of aberrant responses were observed. One was the increased hypotensive effect of debrisoquine (Maghous et al., 2001). Another was the increased toxicity of TCAs (Sjoqvist and Bertilsson, 1986). In both instances, these increased effects were asso- ciated with increased accumulation of the drug (variable 2 in Equation 4.1) on routine doses owing to the slow clearance resulting from the genetically determined deficiency in enzymatic activity.
In this case, the individuals genetically deficient in enzyme activity were analo- gous to the results of the blank test tube (i.e., no enzyme) in the in vitro studies when the substrate is added (i.e., essentially no metabolism at least via that pathway). This genetic deficiency was easily identified by simple blood sampling and measuring the concentration of the drug. This value could be put into Equation 4.1 and the outcome predicted. In this scenario, the pharmacogenetic difference is like giving a higher dose of the drug to the general population or, in other words, is the same as a left shift in the dose–response curve for toxicity and tolerability problems in the CYP2D6-deficient individuals. Parenthetically, they should for the same reason be more sensitive to the antidepressant effects of TCAs (i.e., responding to lower doses).
However, it is interesting to note that psychiatrists did not readily embrace this idea despite its simplicity. Consider that almost as soon as TCAs were marketed the dose-dependent nature of their propensity to cause serious cardiotoxicity was rec- ognized. After all, the toxicity of these drugs was fulminant following an acute over- dose and was a leading cause of drug overdose death in the USA and other countries for over 30 years. Shortly after these drugs were marketed, it was recognized that 5–10% of Caucasians, particularly those of northern European descent, cleared these drugs much slower than the rest of the population and that this phenomenon was genetically determined (Alexanderson et al., 1969). Moreover, it was recog- nized that these individuals developed blood concentrations many times higher than the general population and could develop toxic levels on routine dosage. Moreover, this phenomenon could be readily detected by monitoring blood levels in these patients.
In essence, therapeutic drug monitoring (TDM) of TCAs was a means of testing for a pharmacogenetically determined difference in oxidative drug metabolism

67 Methodological issues in psychopharmacogenetics
(variable 2) that could seriously shift the dose–response curve in a sizable number, albeit a minority, of patients (5–10%). Despite those simple facts, there was great resistance to the adoption of TDM in clinical practice. Arguments were raised that it was too costly and that physicians could do a better job by simply clinically adjust- ing the dose. Parenthetically, and somewhat ironically, the same arguments were raised two decades later when it was discovered that some popular drugs (e.g., flu- oxetine and paroxetine) could produce phenocopies of genetic deficiency of this enzyme in 50 to 80% of CYP2D6 normal or extensive metabolizers (Preskorn, 1998b). As late as 1984, the task established by the American Psychiatric Association to develop guidelines for the use of laboratory procedures in psychi- atry took a “conservative” posture on the use of TDM for TCAs in psychiatry (Task Force on the Use of Laboratory Tests in Psychiatry, 2001).
The reason behind the slow acceptance of TDM for TCAs reveals some of the methodological problems that will likely face future pharmacogenetic discoveries in psychiatry. The first problem was that the focus of the debate was on the wrong aspect of the concentration–response curve: the relationship between concentra- tion and antidepressant efficacy rather than toxicity. The second problem was who was going to do a study to demonstrate that the more relevant problem was increased risk for serious toxicity?
Efficacy as a difficult target for study in psychiatry
At the time the genetic difference in the metabolism of TCAs was being identified, TDM studies were being conducted to examine the potential relationship between drug level and antidepressant efficacy. It should have been recognized then, and certainly now, that such studies are almost certainly doomed to failure. The reason is the poor signal:noise ratio in such studies. The problem is that a typical conven- tional TDM efficacy study includes all the same patients as a drug registration clin- ical trial: (a) the one third or less who are drug-specific responders, (b) the one third who are placebo responders, and (c) the one third who are nonresponders. However, the only population who can show a concentration–efficacy relationship is the one third or less who are drug-specific responders. Of course, the placebo responders and drug nonresponders still achieve plasma drug concentrations but their response or lack of response is independent of the drug and hence its con- centration. Figure 4.2 illustrates what happens when the concentration–response data from these two groups is superimposed on a clear-cut, plasma concentration– antidepressant response curve from the drug-specific responders: no relationship is found.
The results of such studies lead to the astounding conclusion that drug concen- tration was meaningless. Ironically, that conclusion was being championed at the

68 S. H. Preskorn

Fig. 4.2.
The inability to detect a drug level–effect relationship because of a poor signal-to-noise ratio. In the above two graphs, the y-axis is response and the x-axis is plasma drug concentration. In the left-hand curve, there is a 1 to 1 correlation between response and drug concentration in patients who are specifically drug responsive. In the right-hand curve, the same relationship in shown but with the superimposed results from the one third of patients who are placebo responders and the one third who are not responsive to that drug. Since response and nonresponse in these patients is independent of drug treatment, they are independent of drug concentration. To generate these results, typical response and nonresponse scores in such patients were randomly assigned to expected plasma drug concentrations and plotted on top of the predetermined relationship shown in the left-hand graph. (Reproduced with permission from http://www.preskorn.com)
same time that physicians were being berated for using too low a dosage of TCAs as if somehow dose but not concentration mattered.
The fruitful area to study was the relationship between drug concentration and toxicity. Here the signal was clear: seizures, cardiac arrhythmias, and sudden death. However, who would propose doing a study prospectively to define the risk of these serious adverse effects as a function of either dose or interindividual differences in sensitivity (e.g., slow clearance because of CYP2D6 deficiency) – and would any institutional review board let such a study be carried out? Of course, the answer to these two questions are no one and no. Here then is a dilemma in pharmacogenetic studies. A genetically determined population that is likely to be more susceptible to the adverse effects of a drug is defined, but how can it be determined that they are more susceptible if it is unethical to expose to the potential increased risk. One approach would be to use a surrogate marker for the increased risk. That was done with the TCAs. A number of studies were done showing that poor metabolizers developed substantially higher levels of the drugs and asymptomatic prolongation
69 Methodological issues in psychopharmacogenetics
of the QT interval (Preskorn and Fast, 1991). However, drug levels were kept suffi- ciently low to avoid toxicity and the QT prolongation was asymptomatic. (Additionally, asymptomatic QT prolongation was easier to ignore at that time than it is today.) The absence of clinically obvious toxicity reinforced the idea that drug concentrations could be ignored. A similar situation occurred with studies in the 1990s testing the effects of substantial CYP2D6 inhibitors (e.g., fluoxetine and paroxetine) on CYP2D6 substrates (Preskorn, 1998b). In these studies, the dose of the CYP2D6 substrate was purposely kept low to avoid the risk of serious toxicity. Because no toxicity occurred, some dismissed the finding that fluoxetine and paroxetine could cause 500% increases in the plasma concentrations of drugs such as desipramine (Preskorn, 1999).
While using drug levels as a surrogate for toxicity was met with limited success and skepticism, another approach did prove compelling: naturalistic studies of tox- icity under usual dosing conditions (Preskorn and Jerkovich, 1990; Preskorn and Fast, 1991). Here the failure to employ TDM rationally to guide the dose adjust- ment of the TCAs put a predictable percentage of the vulnerable genetically deter- mined subset of the population at risk for toxicity. When an adverse outcome occurred, plasma drug levels were obtained and found to be unusually high as a result of slowed clearance. Eventually, an upper threshold of 450ng/ml (450g/l) was established beyond which the likelihood of serious toxicity outweighed the likelihood of increase efficacy (Fig. 4.2). Hopefully, future pharmacogenetic find- ings will be more readily accepted but the history with TCAs and CYP2D6 illus- trates that scientific findings are not necessarily rapidly accepted into clinical practice, at least in psychiatry.
Conclusions
This chapter has endeavored to review some of the methodological issues facing those interested in conducting pharmacogenetic studies in clinical psychopharma- cology. It began with two equations that illustrate basic pharmacological principles relevant to drug effects in humans and continued to discuss specific methodologi- cal issues. First was the drug development process itself, goal of which is to minimize pharmacogenetic differences as a variable affecting the response to a drug. Second was the problems posed by the poor signal:noise ratio, which plagues current effi- cacy studies in psychiatry. The third issue was the ethical and practical problems posed in trying to study whether a genetically defined population is at increased risk for toxicity or other adverse effects on what is a routine dosage of the drug for the general population. TDM studies with TCAs were used to illustrate this issue.
While the above are all important hurdles in conducting pharmacogenetic studies with existing drugs, this chapter also reviewed the significant potential for

70 S. H. Preskorn
advancing psychiatric drug development as a result of improved understanding of the pharmacogenetics and pharmacogenomics relevant to the effects of drugs on the human brain. This includes the discovery of completely novel sites of action and the discovery of pathophysiologically and etiopathologically distinct subsets of what are now likely heterogeneous syndromes such as major depression. These dis- coveries should have a profound impact on psychiatric drug development in terms of making it more efficient, less costly, and less of a gamble. However, they may also reduce the potential size of the market for any specific agent by better defining what subset actually needs the treatment. In doing so, some subsets of some psychiatric syndromes may be too small to be commercially viable for drug development. Consequently, developments in this area will both promote and restrict psychiatric drug development.
A number of these topics are explored at greater length in other chapters in this book. The interested reader can also find more information at http://www.preskorn.com.
REFERENCES
Alexanderson B, Evans D, Sjoqvist F (1969). Steady-state plasma levels of nortriptyline in twins: influence of genetic factors and drug therapy. Br Med J 4, 764–768.
Beaune PH, Guengerich FP (1988). Human drug metabolism in vitro. Pharmacol Ther 37, 193–211.
Maghous A, Idle JR, Dring LG, Lancaster R, Smith RL (2001). Polymorphic hydroxylation of debrisoquine in man. Lancet ii, 584–586.
Preskorn SH (1998a). Recent dose–effect studies regarding antidepressants. In Balant LP, Benitez J, Dahl SG, Gram LF, Pinder RM, Potter WZ, eds. European Cooperation in the Field of Scientific and Technical Research. Belgium: European Commission, pp. 45–61.
Preskorn SH (1998b). Debate resolved: there are differential effects of serotonin selective reup- take inhibitors on cytochrome P-450 enzymes. J Psychopharmacol 12, S89–S97.
Preskorn SH (1999). Outpatient Management of Depression: A Guide for the Primary-Care Practitioner. Caddo, OK: Professional Communications Inc., pp. 96–97, 256 (this book is also available at http://www.preskorn.com).
Preskorn SH (2000). The stages of drug development and the human genome project: drug dis- covery. J Psychiatr Pract 6, 341–344. (Also available at http://www.preskorn.com)
Preskorn SH, Fast GA (1991). Therapeutic drug monitoring for antidepressants: efficacy, safety and cost effectiveness. J Clin Psychiatry 52, 23–33.
Preskorn SH, Jerkovich GS (1990). Central nervous system toxicity of tricyclic antidepressants: phenomenology, course, risk factors and role of therapeutic drug monitoring. J Clin Psychopharmacol 10, 88–95.
Schmider J, von Moltke LL, Shader RI, Harmatz JS, Greenblatt DJ (1999). Extrapolating in vitro data on drug metabolism to in vivo pharmacokinetics: evaluation of the pharmacokinetic interaction between amitriptyline and fluoxetine. Drug Metab Rev 31, 545–560.

71 Methodological issues in psychopharmacogenetics
Sjoqvist F, Bertilsson L (1986). Slow hydroxylation of tricyclic antidepressants – relationship to polymorphic drug oxidation. In Kalow W, Goedde HW, Agarwal DP, eds. Ethnic Differences in Reaction to Drugs and Xenobiotics. New York: Liss, pp. 169–188.
Task Force on the Use of Laboratory Tests in Psychiatry (2001). Tricyclic antidepressants–blood level measurements and clinical outcome. Am J Psychiatry 144, 155–162.
von Moltke LL, Greenblatt DJ, Schmider J, Wright CE, Harmatz JS, Shader RI (1998). In vitro approaches to predicting drug interactions in vivo. Biochem Pharmacol 55, 113–122.
5
Statistical approaches in psychopharmacogenetics
Fabio Macciardi
Center for Addiction and Mental Health, University of Toronto. Toronto, Canada
OVERVIEW
The statistical analysis of data in psychopharmacogenetics is a key factor in the evaluation of the importance of a gene or of a set of genes in controlling for the response to a given drug or to explain the emergence of a side effect as a consequence of the administered drug. A pharmacogenetic trait is frequently controlled by more than one gene, with different con- tributions from genes coding for pharmacokinetic traits (absorption, distribution, metabo- lism, elimination), the ultimate drug target (e.g., receptors, transporters or enzymes), and endogenous ligands. As such, a pharmacogenetic characteristic can be considered a complex trait, with a multifaceted etiology. The best approach to evaluate the genetic bases of a pharmacogenetic trait is the association strategy, where each gene contributing to the expression of the trait conveys only a portion of the overall variability of the characteristic itself. The association strategy conceptually entails the candidate gene paradigm, which is based on a “forward genetics” design. The re-emergence of this analytical approach has become possible as a result of the completion of the sequence for the human genome and the possibility of knowing a priori which genes are involved in the biochemical mechanisms that lead to the drug function. However, several issues must be considered to perform a correct analysis, including a proper definition of the phenotype, a detailed knowledge of the extent and dimension of the genetic variation that is present within and across populations, and the choice of the statistical design.
Introduction
Pharmacogenetics has been defined as the study of variability in drug response owing to heredity, based on the original definition of the term coined by F. Vogel (1959). Indeed, it is a common medical experience to observe that the effect of the same drug, when administered to several patients, can be advantageous in some subjects, deleterious in others and of no consequences in a third group. Understanding the mechanisms responsible for variability in drug response, as well as in the origin of side effects, is the major object of pharmacogenetics, considering

72

73 Statistical approaches in psychopharmacogenetics
how to relate a drug response unequivocally (phenotype) to the genotype (Roses, 2000). To accomplish this task, it is necessary to define a quantifiable clinical response and then to detect the variant(s) of the gene(s) implicated in modulating the response to the drug. Statistics is the tool to prove the relationship between the phenotype and the genotype.
In recent years, the term pharmacogenomics has also been introduced; while the terms pharmacogenetics and pharmacogenomics do not substantially differ from each other in that each recognizes that genes play a substantial role in controlling for the response to the administration of a given drug, there are also conceptual differences. The term pharmacogenomics emphasizes the accepted view that most, if not all, of the “response” traits are potentially controlled by more than a single unique gene and are somewhat influenced by the whole human genome. Under this perspective, several different genetic loci are required to explain the effect, or the lack of effect, of a given drug treatment. This implies that pharmacogenetic traits are not simple Mendelian characteristics and that several genes play a causal role in the genesis of a pharmacogenetic trait. Two nonmutually exclusive alternative pos- sibilities exist: each gene has only a small quantitative effect on the trait (“poly- genic” model) or more than one “trait” gene must be simultaneously present in order to produce the phenotypic effect (“epistatic” model). This implicitly means that, taken separately, the individual gene effects are only modest and that the geno- type relative risk (i.e., the increased chance that an individual with a particular genotype for a particular allele has the trait) is low. Despite the small effects of these genes, the magnitude of their attributable risk may be large, since they must be quite frequent in the population owing to the high prevalence of the characteristic.
The following sections will show how to analyze the genetic architecture of a pharmacogenetic trait. First, a definition of a pharmacogenetic trait will be pre- sented, with the consequential methods of analysis, followed by some considera- tions for defining a phenotype suitable for investigation. Then, our present understanding of genetic polymorphisms will be described, with a particular focus on single nucleotide polymorphisms (SNPs). A detailed discussion will follow about ethnic and interindividual differences in the distribution of genetic variants and how they can affect any investigation.
The pharmacogenetic trait: definition and methods of analysis
A pharmacogenetic trait is a complex, multidetermined characteristic, not a disease but rather a “difference” in the expression of a genetically determined mechanism involved in drug metabolism (Nebert, 1999). The distinction is subtle but substan- tial. At variance with a common/complex disorder, we do not have to detect which genes are relevant for the trait to be manifested, and then to detect their

74 F. Macciardi
pathological variants. Rather, it should be known, more or less, from our know- ledge of biology, pharmacology, or neuroscience, that a given set of “proteins” are implicated and responsible for the drug to function. These proteins comprise drug- metabolizing enzymes, transporters, receptors, drug-binding proteins and other enzymes. The goal is then to detect which specific variant(s) of any one of the genes coding for the specific protein(s) relevant for the drug function may be responsible for the lack of drug efficacy, either acting alone or – more probably – in some kind of combined effect (interaction). The focus of the investigation has, therefore, shifted from the identification of a gene as a causative effect of the trait, as when analyzing the genetic bases of a complex disease, toward the identification of a set of variant(s) that across many genes already known will generate the observed phe- notype.
In other words, a pharmacogenetic character is the ultimate effect of an “envi- ronmental” stimulus acting on the specific genetic make-up of an individual. The consequence of such a definition of a pharmacogenetic trait is that traditional genetic analytical strategies like linkage or segregation analyses do not apply. Instead, association or linkage disequilibrium (LD) techniques are the methods of choice, privileging an epidemiological biometrical approach over a Mendelian scheme and using a population genetics strategy. Population genetics is a well- developed science that studies how, where, and why the variation in genes is differ- ent across different groups or human populations. Population genetics has been active for more than 100 years; however, its methods have only begun to be applied to the variation across different individuals in their response to (or in their suscep- tibility to develop side effects with) the same medication, i.e., in pharmacogenetics.
Given the number of biochemical mechanisms involved in controlling psycho- tropic drug actions, the pharmacogenetic “profile” of a given medication can be defined as a “complex trait” that cannot be explained by a simple genetic mechan- ism like a single major locus. The association strategy has the advantage of being able to detect not only major gene effects but, minor gene effects also; the criteria for a positive association rely on detecting those alleles of the candidate or marker genes under analysis that are preferentially transmitted together with the trait. This provides indirect proof that a still as yet unidentified allele of the locus responsible for the trait cosegregates with the corresponding associated allele of the candi- date/marker gene, giving rise to a haplotype (combination of alleles in a chromo- some). Under these conditions, association is equivalent to linkage analysis (Baur and Knapp, 1997; Ott, 1999). In extreme favorable circumstances, the marker and the trait locus are identical, and only the individuals with the trait present a given allele at the marker locus.
In recent years there has been a critical debate about the most appropriate design to study genetic associations. Case-control is the simplest method, but it was
75 Statistical approaches in psychopharmacogenetics
considered a weak approach because of the difficulty of appropriately selecting a group of controls and because it had the potential for failing to deal adequately with issues of admixture and stratification within the population studied. Consequently, techniques that used “internal” – or family-based – controls were preferred, as in the transmission disequilibrium test (TDT: Spielman et al., 1993). The TDT is based on the detection of unequal transmission of high- versus low-risk alleles from heterozygous parents to affected children. The Mendelian expectation under the null hypothesis of no linkage or association is that either allele carried by a heterozygote has 50:50 odds of being transmitted to an affected child. If the allele is causally related to the disorder, however, then the odds of its being transmitted should exceed 50%. Compared with other within-family association tests (e.g., the affected family based control and haplotype-based haplotype relative risk methods) (Schaid and Sommer, 1993; Spielman and Ewens, 1996), the TDT pro- vides a test of linkage as well as association, robustness against artifacts induced by population stratification, greater statistical power, and the ability to include multi- ple affected siblings from a family in a test of linkage (but not association) without having to correct for dependence (Spielman et al., 1993; Schaid and Sommer, 1994; Ewens and Spielman, 1995; Spielman and Ewens, 1996, 1998).
Generalizations of the TDT approach are presently available (Monks and Kaplan, 2000) for any kind of family (trios, nuclear, extended, sib-pair, etc.) even when one parent is not available (Spielman and Ewens, 1998). Despite many expectations, however, the TDT has not definitively proven to be superior to other association designs. Moreover, it is also less efficient than the case-control design, being limited to information only from heterozygous parents, and, since the collection of an appropriate sample size is much more time consuming, ultimately leads to a loss of power. Consequently, unrelated case-controls are now re-emerging as a more pow- erful and efficient approach, in spite of their lack of robustness for possibly inflat- ing type I errors (Risch, 2000). However, it is not possible to avoid false-positive results, and different strategies to deal with this issue are presented in the following paragraphs. In addition to specific strategies, a few simple cautionary approaches, like a priori setting a conservative p value to define a significant finding, help to increase the robustness of the case-control design. The re-emergence of case- control studies in human genetics has also been stimulated by the potential iden- tification of all (or most) of the genes in the human genome, as well as by the identification and cataloguing of the functional variation that occurs naturally within them in human populations. This information will also lead to application of a “forward-genetics” method, rather than a reverse-genetics approach, given the opportunities for studying the impact of those variants on the phenotypic outcome of interest, thus strengthening the candidate gene approach (Risch and Merikangas, 1996; Risch, 2000). However, to avoid the risk of generating biased results, the
76 F. Macciardi
researcher must adopt a strict methodological strategy that involves a relevant defi- nition of the phenotype and a correct use of the genetic polymorphisms, including the knowledge of their variability within and across populations.
How to define a phenotype in psychopharmacogenetics
The objective of any clinical pharmacogenetic analysis is to consider how to corre- late drug response or the emergence of a side effect to the genotype in an unequiv- ocal way. Defining the phenotype of response to a given drug is a critical issue since the expression of the phenotype can be identified only in those subjects who were administered the specific compound. It is obvious that subjects unexposed to a given drug are dormant for the phenotype identification and it will never be known whether those subjects will be responsive or not to the administration of the drug. This makes ascertainment difficult in the general population. Moreover, especially in pharmacogenetics of psychotropic drugs, even the definition of the “response” to a treatment is a challenging task, with serious issues to consider. Apart from some medications and under unusual circumstances, as for example in sophisti- cated clinical trials, the researcher does not have biological or laboratory indices of response to a given psychotropic drug.
In its simplest form, the classification of responder/nonresponder can be consid- ered a dichotomous trait, with a clear separation between the two classes. In reality, however, the classification is based upon a threshold set up by the researcher. Methods to position the threshold are largely drawn from classical pharmacology guidelines as derived from clinical trials or from epidemiological investigations. Despite the appar- ent simplicity of the concept and the appeal of getting a simple yes/no answer, the nominal definition of responder entails a large set of information that need to be crit- ically evaluated before being synthesized into the proposed classification.
Alternatively, the responder/nonresponder phenotype can be measured as a quantitative variable. This allows a more flexible definition of response without the need to impose a rigid bipartition of the phenotype. In this case the degree of responsiveness is measured along a continuous scale, which may be arbitrary or more probably represents a summary measure of various clinical domains. The correspondence between the degree of responsiveness and the genetic make-up can be evaluated using the full range of responsiveness in a straightforward way, but alternatives are possible. One option is to use only the extremes of the scale, that is subjects lying in the areas that contain the “best” and the “worst” responders, usually constituting the upper and lower 5% of a continuous Gaussian distribution (Nebert, 1999). Comparison between extreme discordant subjects in looking for the role of a significant effect of a genetic component is a powerful technique that has already been proven successful (Lifton, 1996; Halushka et al., 1999).

77 Statistical approaches in psychopharmacogenetics
Ultimately, the classification of patients into responder/nonresponder mostly relies on the clinical judgement of the physician who evaluates the improvement of the patient following the administration of the therapeutic compound. Criteria about the reliable definition of responder and nonresponder must, therefore, be agreed among investigators and must reasonably guarantee against the risk of sta- tistical errors in future genetic studies for an associated genetic polymorphism. For example, if responders are defined as those patients who show a 20% improvement over the baseline of their psychopathological symptoms, measured with a certain scale, this will eventually give rise to a sample size of patients where the proportion of responders will be much bigger than if the trait definition had to have a 30% improvement threshold. The consequence of this is that, when studying a possible association with a polymorphic variant with high frequency (e.g., 10–20%) in the general population, the possibility of a positive finding are a priori larger in the “20% improvement” phenotype, irrespective of the real meaning of the association, unless stringent significance levels are used to define a positive association. In fact, relaxing the criteria for the identification of the phenotype increases the risk of including false-positive responders and leads to an unknown corresponding incre- ment of type I error. Unfortunately, in the absence of objective criteria, the pheno- typic definition of responder is frequently based upon loose clinical criteria, and researchers are naturally inclined to prefer phenotype definitions that maximize the efficiency of the analysis. In this case, a “broad” formal definition of the phenotype allows a relatively small sample of patients to be recruited. Even though this small sample size can be justified by a power analysis showing that there is enough power to detect an association, the risk of drawing erroneous conclusions is high. The error results from incorrect premises rather than incorrect procedures. To avoid these kinds of error and since the definition of the phenotype is entirely arbitrary, it is best that one group (responder) be unequivocally separated from the other group (non- responder). It must be also pointed out that virtually all the “common” traits – either diseases or pharmacogenetic characters – represent multiplex phenotypes in the sense that they are polygenic (derived from the contribution of two or more genes). Therefore, the best possible approach in humans is to quantify the pharma- cogenetic phenotype in much the same way as geneticists have done in studying hypertension: instead of looking at subjects as hypertensive/not hypertensive, the identification of relevant genes was possible when researchers looked at the quan- titative phenotype “blood pressure” (Jacob et al., 1991; Schork et al., 1995; Brown et al., 1996; Halushka et al., 1999). Various examples of alternative phenotypic defini- tions useful for psychopharmacogenetic analyses can be found in other chapters (e.g., Chs. 10–13) and the interested reader can find more specific details there.
In principle, the definition of a “side effect” phenotype is not as difficult as the responder/nonresponder phenotype. Many side effects present with objective
78 F. Macciardi
clinical signs, as in tardive dyskinesia during long-term administration of antipsy- chotic medications, or are measurable with laboratory tests, as in the decrease in white blood cells count in patients treated with clozapine or thioridazine. In these and other similar cases, the recognition of the phenotype is relatively easy, not pre- senting the same difficulties as the clinical-response phenotype. However, here too the possibility exists to define the phenotype as a qualitative or a quantitative char- acteristic. In some cases, the selection of one or the other method might induce a difference in the way the phenotype is considered, and the researcher should clearly establish the focus of the analysis at the beginning. For example, if the phenotype is defined as present/absent, an established correlation with a given gene points to this gene as one of the “causes” of the phenotype. The same conclusion does not necessarily apply if the phenotype is defined quantitatively, since – depending upon the quantitative dimension of the phenotype that is being considered – the meaning of the “phenotype” in this case might be different from the simple concept of presence/absence and could be related to other characteristics of the phenotype itself, like the degree of severity. Consequently, given that the trait has a multigene mechanism, a qualitative or a quantitative definition of the phenotype could, in principle, pinpoint different genetic components.
Association and linkage disequilibrium methods
To understand the genetic architecture of complex traits and to dissect the genetic component of the trait into its basic elements is a considerable challenge, despite the remarkable advances in knowledge and the sophistication of contemporary technology. A general view of complex genetic traits is that the contributing genes are genes of small or minor effect, as opposed to major genes characteristic of simple Mendelian traits (Ott and Hoh, 2000); however, opinions differ as to the best way to describe the contribution that each small gene exerts on the trait.
The assumption that complex traits follow a similar genetic model to that seen, for example, in late-onset Alzheimer’s disease, which is associated with the apoE4 allele (Corder et al., 1993), or in type I diabetes mellitus, which is associated with class I alleles of the insulin gene minisatellite (Bennet et al., 1995), has not yet been demonstrated, suggesting that not all predisposing alleles to complex traits are common. In Alzheimer’s disease and diabetes, common alleles/variants of genes implicated in the etiology of the diseases can be found with medium to high fre- quencies in all populations. Under this scenario, the common trait–common variant hypothesis (Risch and Merikangas, 1996; Collins et al., 1997; Chakravarty, 1999), the allele variants are not pathological per se and represent a risk factor for a given disease only when they are present in a particular multigene combination. In other words it is only when they appear concurrent with other “risk” variants at

79 Statistical approaches in psychopharmacogenetics
other loci that the combination produces the ultimate genetic risk. An alternative theory suggests that a large pool of alleles with low population frequency and varying effects on risk (the common trait–rare variant hypothesis) could explain the genetic component of common traits (Terwilliger and Weiss, 1998; Wright et al., 1999; Weiss and Terwilliger, 2000; Pritchard, 2001).
It is probable that a more realistic model lies between these two extreme hypoth- eses, where predisposing alleles of varying population frequencies may represent a true allelic heterogeneity in some cases or a locus heterogeneity in other circum- stances, and alleles would act either independently or epistatically to influence trait outcome (van Hauwe et al., 1999; Johnson and Todd, 2000). It is clear that if we are going to maximize the chances of success in complex trait mapping, we must design our analytical strategies so that we can detect subtle genetic effects under a variety of genetic models.
Polymorphic variants in the human genome
The sequence of human DNA is characterized by a high number of variants (poly- morphisms) that are widely distributed across the various chromosomes and are generally termed markers. A detailed description of the various types of polymor- phism is beyond the scope of this chapter, but it is nonetheless important to intro- duce some definitions and describe in simple details the most widespread form of genetic variants: the SNPs. By polymorphisms, geneticists mean any variant of the sequence of a locus that is represented in the general population with a frequency 1%, while the most represented form of a gene is called the wild type. Another usual term to define the different form that a gene can take is allele: a gene can be present with any number of alleles, from 1, when the gene is not polymorphic, to several hundreds, like in the loci controlling for the HLA complex. It is common for genes to be polymorphic, and this simple fact does not mean that a given variant is pathological. On the contrary, mutations are generally rare variants of a gene, usually 1%, and the present use of the term mutation indicates a pathological form of a gene, generally leading to a disorder.
Among the many variants present in DNA, the most common is the result of a variation in a single base pair or a simple insertion/deletion of a small number of base pairs; these polymorphisms are defined as single nucleotide polymorphisms (SNPs) (Collins et al., 1999). At present there are already more than 2.5 million SNPs mapped into the human DNA, by the same group of investigators that have sequenced the human genome (International SNP Map Working Group, 2001) and by a consortium of private and public researchers (Venter et al., 2001). A rough esti- mate of the frequency of SNPs is that there could be a variant of the sequence every 500/1000 base pairs (Cargill et al., 1999; Kruglyak, 1999), generating a potential number of at least 3 to 6 million SNPs. SNPs are further subdivided according to
80 F. Macciardi
their “location” with respect to the coding sequence of a gene. SNPs within the coding region of a gene can lead to nonconservative alterations (type I), conserva- tive amino acid substitutions (type II) or synonymous substitutions (type III). Noncoding SNPs have been classified into 5 untranslated (UTR) region (type IV), 3 UTR (type V), and other noncoding regions (type VI). There is already the pro- posal that type VI SNPs be divided into further subtypes, depending whether they are found at exon/intron boundaries, in putative intronic AP1 sites, etc., while type IV SNPs might comprise also putative promoter regions. Therefore, it appears very likely that multiple SNPs will be found in and around each and every gene in the human genome, and each locus would be potentially highly polymorphic and con- sequently amenable to a detailed investigation. It is also important to note that changes in the coding sequence (SNP type I to III) are not the only candidates for functional variation, and that SNPs in perigenic regulatory regions can also have a large phenotypic impact (Halushka et al., 1999), as well as having a role in evolu- tion (Beaumont, 1999; Chakravarty, 1999).
SNPs are less frequent in coding regions than in noncoding regions, particu- larly those that lead to amino acid substitutions (type I) and, therefore, have func- tional (phenotypic) implications; also SNPs lying in noncoding regions but having relevant functional effect, like those SNPs in the putative promoter regions of genes (type IV), are infrequent. The majority of these SNPs (type I and IV) usually have a low allelic frequency for the less-represented allele, usually 15% and frequently 5%, with obvious consequences in sampling and for the statisti- cal analyses.
Sample size required for candidate gene studies
The number of patients required to find a statistically significant association between a SNP and a trait depends on a number of factors, including the frequency of the drug response, the proportion of patients having the SNP allele, the minimum detectable effect, the level of statistical significance (p value) and the power. The “effect size” of the association is the likelihood of response to a drug in individuals with the susceptibility allele (O1) compared with those without the allele (O0). The magnitude of the effect in a typical case-control study is measured by the odds ratio (OR O1/ O0), which is expected to be low, 1.5–4.0, given the multigene characteristic of pharmacogenetic traits. If some of the factors that affect the possibility of finding a significant association are known or can be guessed, then we can also calculate the sample size needed to perform a meaningful study.
For example, if it is known, or can be postulated, that (i) the proportion of response to a drug for patients having a given allele is 41% (O1), (ii) the magnitude of the effect that is attributed to the gene is around 2 (OR), and (iii) the frequency of the allele of interest is 0.46, we can calculate the required sample size (N) using
81 Statistical approaches in psychopharmacogenetics
traditional formulae such as those in Fliess (1981) or Agresti (1966). A simple way to calculate the number of cases and controls is (Agresti, 1966):
Ncases Ncontrols (z/2 z)2 [O1(1 O1) O0(1 O0)]/(O1 O0)2
where z/2 is the statistics corresponding to the p value (if p 0.05, then z/2 1.96) and z is the statistical value corresponding to the type II error, equivalent to 1 (i.e., power: 0.84 for power 80%). In this example, O0 22% (O1/1.95, if O1 41%) and N 91. If the frequency of drug response is assumed to be 30% in the general population, approximately 300 subjects must be screened to reach N. It is evident that this scenario can widely vary, depending upon the various assump- tions. If the response to the drug is as low as 10% of all the treated subjects, then approximately 900 patients must be screened. Other than the rate of drug response, another critical factor in determining the sample size is the relative fre- quency of the allele of interest. As already noted, the occurrence of the less fre- quent alleles for type I and type IV SNPs may be much lower than for noncoding SNPs: less than 10–15%. In this case, the sample size required to detect a differ- ence between responders and nonresponders in relation with the given polymor- phism may become very high, beyond the range attainable by a single independent investigator. Table 5.1 shows an example of this range of computations. For a quantitative phenotype, the “effect size” of the association is relative to the per- centage difference of the response measure for the drug treatment in subjects with or without the susceptibility allele. Specific formulae for continuous measures, under determined type I and II errors, give the required sample size to detect what fraction of the variability can be attributed to the allelic effect, in a similar way to the case for qualitative outcomes.
A potential alternative to improve the low “power” of coding SNPs is to use them together with other noncoding SNPs that are present within the same gene. Instead of using a single polymorphism, an array of SNPs are used that are linearly arranged within a short chromosomal area, usually within 1–2 cM, and that are close enough to each other to be not randomly associated, constituting a haplotype. This specific combination of alleles (SNPs) at different loci that are physically close to, or within, a particular gene variant is more likely to be inherited together with that gene variant than alleles that are further apart and are said to be in LD. LD is dependent on the actual physical distance between the polymorphisms and the evolutionary time when the polymorphisms originated (Chakravarty, 1999). LD decays over time as a function of the recombination fraction between loci, but for very close loci, equivalent to very high LD, the decay is negligible. Haplotypes can mark the chromosomal position where the variant of interest is located more efficiently than individual SNPs and haplotype/LD mapping has been already proved to be a useful technique to map genes for several complex traits (Hästbacka et al., 1994;
82 F. Macciardi
Table 5.1. Determination of sample size for the association of a single nucleotide polymorphism (SNP) with a drug response in six different scenariosa
Study population characteristics
Freq of A1 0.46; OR 1.85
Freq of A1 0.23; OR 1.30
3 4
Freq of A1 0.10; OR 1.11
5 6
with varying general population
response rates 1
2
  
General response 40% Responder with A1 (%) Responders without A1 (%) N cases requiredb
N cases to be screenedc
General response 30% Responder with A1 (%) Responders without A1 (%) N cases requiredb
N cases to be screenedc
General response 20% Responder with A1 (%) Responders without A1 (%) N cases requiredb
N cases to be screenedc
General response 10% Responder with A1 (%) Responders without A1 (%) N cases requiredb
N cases to be screenedc
41 30 41 22 16 32 91 142 404
228 356 1009
41 30 41 22 16 32 91 142 404
304 474 1346
41 30 41 22 16 32 91 142 404
457 712 2019
41 30 41 22 16 32 91 142 404
913 1423 4037
30 41 30
23 37 27 638 2214 3546 1596 5535 8866
30 41 30
23 37 27 638 2214 3546 2128 7380 11821
30 41 30
23 37 27 638 2214 3546 3192 11071 17731
30 41 30
23 37 27 638 2214 3546 6383 22141 35463
Notes:
OR, odds ratio.
. a ThescenarioscoverthreedifferentfrequenciesfortheSNPalleleA1(1,2;3,4;and5,6);eachfrequencyis
considered with two different percentage responses for those with and without the allele of interest (1,3,5
and 2,4,6). Details of the calculations needed to compute the sample size are reported in the text.
. b The p value () is set at 0.05 and the power (1 ) at 80% for all scenarios.
. c Number that must be screened to find N responders.
Nickerson et al., 1998; McPeek and Strahs, 1999; Rieder et al., 1999; Jorde, 2000; Martin et al., 2000). Association studies with SNPs are not heavily conditioned by allelic heterogeneity, as may be the case for microsatellite polymorphisms, given the biallelic nature of SNPs, but the finding of the association can be affected by locus heterogeneity or by the presence of a yet undetected genetic interaction. However, it is very difficult to address these issues, and thus avoid speculating with theoretical
83 Statistical approaches in psychopharmacogenetics
hypotheses, until a first initial result makes it easier to understand the specific genetic architecture of the trait.
Methods of analysis: study design
The analysis of a case-control genetic association is straightforward using tradi- tional statistical techniques, like the chi-square and associated measures for qual- itative data (e.g., log-linear or logistic modeling) or the t-test and ANOVA (analysis of variance)/regression for quantitative phenotype definitions (Agresti, 1966; Sokal and Rohlf, 1995). A review of association methods in statistical genetics can be found in Sham (1998). Methods have been developed to account for the lack of robustness in case-control association studies, either by using extensions of the spe- cific techniques, like the Mantel–Haenszel test (1958), for qualitative data or the use of covariates for continuous measure. Moreover, simply defining a conservative p value (i.e., p 0.05) to conclude a significant positive finding may help to avoid spurious associations (Risch, 2000).
Other methods that correct for the risk of confounding, mostly from the chance of population admixture in case-control studies, make use of alternative strategies. One example is the method of genomic control (Devlin and Roeder, 1999), which uses the genome itself to determine appropriate corrections for population-based association tests. The test accounts for nonindependence caused by population stratification and cryptic relatedness by genotyping multiple loci unlikely to affect liability. These are called null loci because they are assumed to have no effect on the disease under study. For a case-control analysis of candidate genes, genomic control computes chi-square test statistics for independence for both null and candidate loci. By means of the variability and magnitude of the test statistics observed at the null loci, which are inflated by the impact of population stratification and cryptic relatedness, a multiplier is derived to adjust the critical value for significance tests for candidate loci. In this way, genomic control permits analysis of stratified case- control data without an increased rate of false positives. If population stratification and cryptic relatedness are not detected from null loci, then genomic control is identical to a standard test of independence for a case-control design (Bacanu et al., 2000). Other similar methods have been proposed to test for stratification (Pritchard and Rosenberg, 1999) and these methods are now being successfully applied to genetic association studies.
A statistical design, however, tests whether the observed distribution of a given allele/genotype in responders and nonresponders deviates from the statistical expectation. The test does not consider any population genetic information from the population-specific background, nor can it usually handle more complicated patterns such as those represented by the analysis of haplotypes. In these cases, a LD strategy is the design of choice. LD is based on the principle that most individuals
84 F. Macciardi
with a specific phenotype in a population carry a polymorphic variant (or a set of variants/alleles) responsible for that phenotype and that such an allele is derived from a single ancestor. For LD to exist, we assume that a given locus A is a disease gene with two alleles Aj and Ai where the allele Ai is the “affected” variant; another locus B is close to A and is polymorphic, with two variants i and j. The Bi variant originated at some point in evolution and on the same chromosome where the Ai variant was already present. Given the close proximity of loci A and B, the two alleles Ai and Bi tend to be transmitted together across many generations, creating an AiBi haplotype. Only in very rare cases a crossover occurs between loci A and B, exchanging Bi with Bj and leading to an AiBj haplotype. Therefore, the majority of affected individuals in the population present with the (Ai)Bi haplotype. Since we can genotype the B locus in real subjects, finding a disproportionate higher rate of Bi alleles in “affected” subjects implies that the yet unobserved Ai allele responsible for the trait is located in the molecular vicinity of locus B. The measure of LD is a complex topic, but in its simplest formulation LD can be measured by D (Lewontin, 1964), which represents the difference from the product of allelic frequencies at two loci with the actual frequency observed in the data:
D p(AiBi) p(Ai)p(Bi)
LD can vary from 0, equivalent to no LD, to 1, when the alleles at the two loci are in complete LD and are always simultaneously co-occurring. A positive value for D less than 1 is indicative of a certain degree of LD. The magnitude of the observed dis- equilibrium is dependent upon a number of different factors, most notably the dis- tance between loci A and B. Despite the fact that the recombination events that lead to this condition are not observable – differing from linkage observed in families – they can nonetheless be inferred by analyzing the degree of DNA shared segments on the actual chromosomes. Compared with recombination, LD is a property of populations and depends extensively on their demographic and social histories, in short, on the evolutionary history of the populations (Chakravarty, 1999; Wright et al., 1999). Isolated populations like Finns and Ashkenazi Jews show extensive LD around rare disease mutations. The degree to which the same will be true for higher- frequency variants is uncertain, although as a general rule the disequilibrium is likely to decline with increasing allele frequency owing to an older coalescence time (Kingman, 2000). Consequently, the advantage of using such populations depends not only on the prevalence of the trait in the population but also on unknown factors, such as the age and frequency of the trait allele(s). For example, the fre- quency of the NAT2 rapid acetylator trait is more than 90% in Japanese, but only about 8% in Middle-Eastern Arabs, owing to striking allelic differences in the gene polymorphisms (Kalow and Bertilsson, 1994). It is speculated that the age of origin of the allelic variant is different between the two populations.
85 Statistical approaches in psychopharmacogenetics
Other than theoretical studies (Lio and Goldman, 1998), practical approaches have been developed to infer the age of a given allele at least at a marker locus, i.e., at a SNP (Bertorelle and Rannala, 1998), thus leading to the possibility of studying LD via evolutionary trees (Lam et al., 2000; Seltman et al., 2001). However, because LD reflects the history of recombinations, populations with different demographic histories will often display different LD patterns (Goddard et al., 2000; Pritchard and Przeworki, 2001). In particular, most studies demonstrate higher levels of LD in recently founded populations than in “older” populations such as those in Africa (e.g., Kidd et al., 1998; Tishkoff et al., 1998). In recent founded groups, such as the Finns, LD may be seen for loci separated by several centimorgans or more, suggest- ing that younger populations may be most useful for the initial detection of a trait locus via LD at large distances. Subsequently, older populations, in which more recombinants have accumulated, may be more useful for the fine-scale LD mapping of the trait locus, assuming that the ages of trait-causing alleles are correlated with the age of a population. In addition, for complex traits, it assumes that the relative effect of each susceptibility locus will be roughly similar in diverse populations (Ott, 2000).
LD can be computed using different mathematical methods (e.g., Devlin and Risch, 1995; Terwilliger, 1995; Devlin et al., 1996; Xiong and Guo, 1997; Lazzeroni, 1998) for single and multilocus models. The incorporation of information from multiple loci – haplotype – can enhance the power and accuracy of LD mapping, mostly if there is the possibility to accommodate for multiple founder mutations and locus heterogeneity, as in Lazzeroni (1998). The multilocus LD mapping involves the analysis of haplotype regions shared in affected cases, taking into account the relationships among groups of markers (Service et al., 1999; Mander and Clayton, 2000). LD can be measured both in large and small (triad) families and in unrelated individuals with numerous ad hoc programs, thus making an LD analysis feasible under many different sampling designs (e.g., Kerem et al., 1989). For unrelated individuals, the reconstruction of the haplotype is based upon prob- abilistic models, since the “phase” of the linked alleles at their sites is usually unknown.
The growing relevance of this approach, and of its associated issues, is high- lighted by the recent finding of a positive association between the response to sal- butamol and the interaction of multiple SNPs within a haplotype (Drysdale et al., 2000) and between a haplotype at the -opiod receptor gene (OPRM1) and heroin addiction (Hoehe et al., 2000). In these cases, only the LD-haplotype analyses sup- ported these findings. Other researchers have already questioned the relevance of the findings, considering that only individual SNP variants, and not a haplo- type, are important and can be considered the “real” causative factors (Davidson, 2000).
86 F. Macciardi
Gene interactions
In complex trait genetics, a large majority of investigators are beginning to think that almost every trait is determined by the joint and synergistic effect of more than one gene. However, as yet a complete theory of gene–gene interaction has not been developed, as indicated by the imprecision of the terminology used to describe these phenomena (Phillips, 1998). The analysis of a gene interaction is theoretically clear and simple for any biallelic system and mostly depends on whether the outcome phenotype is represented by a quantitative or by a qualitative variable. For a single gene controlling for a quantitative variable, the additive genotypic value is defined as half the difference between the genotypic values for the two homozy- gotes[a (G22 G11)/2],whilethedominancegenotypicvalueisgivenbythedevi- ation of the heterozygote from the point midway between the two homozygotes [d G12 (G11 G22)/2]. Using a regression approach, it is possible to estimate values for a and d, using G11 1.0, G12 0 and G22 1.0. Extending this nota- tion to multiple loci, the estimate of an epistatic effect is straightforward: for two interacting loci A and B, there are nine genotypic values (for example, Ga11b11) and by simply adding an additional locus values for a and d can be calculated for the two-locus model. Here, there will be four epistatic genotypic values, referred to as additive-by-additive (aa), additive-by-dominant (ad), dominant-by-additive (da) and dominant-by-dominant (dd). Again, a regression approach allows for the esti- mation of these four parameters (Cheverud, 2000).
For a qualitative phenotype, the case-control association or the TDT must be represented in the form of a logistic regression (Sham, 1998) to allow for the inter- action to be evaluated. The simple coding of genotypes at both loci A and B as dom- inant (e.g., G11 G12 1.0, G22 0.0), codominant (G11 1.0, G12 0.5, G22 0.0) or recessive (G11 1.0, G12 G22 0.0) and the consequent test of the nine possible models (e.g., dominant-by-dominant, dominant-by-codominant, dominant-by- recessive, etc.) allows the estimation of the presence of a specific form of epistatic interaction (Phillips, 1998; Goodnight, 2000).
Conclusions
The present brief review of the methods existing and under development for statist- ical analysis in pharmacogenetics should emphasize the great efforts that geneticists and statisticians are making to improve the ability to detect the complex genetic architecture of pharmacogenetic traits. Indeed, there are already the theoretical backgrounds and the practical techniques to detect the relevance of a given gene for the expression of a complex trait. The methodological approaches must take into account a large set of issues, ranging from the proper estimation of the required sample size to the extent of LD in the particular population under analysis, but, in

87 Statistical approaches in psychopharmacogenetics
principle, a positive significant result is predictable in many cases. At present, there are still major limitations when there are not obvious candidate genes or candidate regions to explore in order to relate the pharmacogenetic trait to its genetic deter- minants. In this case, the researcher is faced with the enormous task of performing a LD mapping across the human genome. In fact, to detect a gene with a LD approach without any a priori assumption requires genotyping each person in the study for every one of 60000 to 500000 SNPs. These are the figures estimated to cover a map of the human genome with a SNP every 100 to 10 kb: such a number of polymorphisms allows the detection of LD in most regions but still may be insuffi- cient to detect genes in regions where LD is less extensive. Although the existing geno- typing technologies are powerful, the current average cost of approximately US$1 per genotype could make a large-scale SNP genotyping study prohibitively expen- sive. It is possible to predict different scenarios with a dramatic reduction of costs achieved through new technologies, such as DNA microchips, but even at 1 cent per genotype, a large-scale LD mapping for a single patient would cost about $5000. If the number of subjects required to detect a gene in a complex pharmacogenetic trait is considered, it becomes obvious that even in favorable circumstances a pharma- cogenetic study represents a notable economic challenge. Among the solutions pro- posed to deal with the cost of these studies, the proposal to pool patient DNA samples (e.g., Barcellos et al., 1997) has gained favor. However, pooling methods rule out the possibilities of subgrouping and performing haplotype analyses, in addition to presenting technical difficulties. In these cases we are still awaiting the significant advances that will make extensive genotyping a standard component of the investigations of complex traits.
Acknowledgement
The work described in this chapter has been partially supported by a NARSAD Investigator Initiated 2000 Award.
REFERENCES
Agresti A (1966). Introduction to Categorical Data Analysis. New York: John Wiley.
Bacanu SA, Devlin B, Roeder K (2000). The power of genomic control. Am J Hum Genet 66,
1933–1944.
Barcellos LF, Klitz W, Field LL et al. (1997). Association mapping of disease loci by use of a pooled
DNA genomic screen. Am J Hum Genet 61, 876–884.
Baur M, Knapp M (1997). Association studies in genetic epidemiology. In Pawlowitzki IH,
Edwards JH, Thompson EA, eds. Genetic Mapping of Disease Genes. San Diego: Academic Press, pp. 159–172.
 
88 F. Macciardi
Beaumont M (1999). Detecting population expansion and decline using microsatellites. Genetics 153, 2013–2029.
Bennett ST, Lucassen AM, Gough SCL et al. (1995). Susceptibility to human type 1 diabetes at IDDM2 is determined by tandem repeat variation at the insulin gene minisatellite locus. Nat Genet 9, 284–292.
Bertorelle G, Rannala B (1998). Using rare mutations to estimate population divergence times: a maximum likelihood approach Proc Natl Acad Sci USA 95, 15452–15457.
Brown DM, Provoost AP, Daly MJ, Lander ES, Jacob HJ (1996). Renal disease susceptibility and hypertension are under independent genetic control in the fawn-hooded rat. Nat Genet 12, 44–51.
Cargill M, Altschuler D, Ireland J et al. (1999). Characterization of single-nucleotide polymor- phisms in coding regions of human genes. Nat Genet 22, 231–248.
Chakravarty A (1999). Population genetics – making sense out of sequence. Nat Genet 21, 56–60. Cheverud JM (2000). Detecting epistasis in quantitative trait loci. In: Wolf JB, Brodie ED III, Wade MJ, eds. Epistasis and the Evolutionary Process. New York: Oxford University Press, pp.
58–81.
Collins A, Longjou C, Morton NE (1999) Genetic epidemiology of single-nucleotide polymor-
phisms. Proc Natl Acad Sci USA 96, 15173–15177.
Collins FS, Guyer MS, Chakravarty A (1997). Variation on a theme: cataloguing human DNA
sequence variation. Science 278, 1580–1581.
Corder EH, Saunders AM, Strittmatter WJ et al. (1993). Gene dose of apolipoprotein E type 4
allele and the risk of Alzheimer’s disease in late onset families. Science 261, 921–923. Davidson S (2000). Research suggests importance of haploytpes over SNPs. Nat Biotech 18,
1134–1135.
Devlin B, Risch N (1995). A comparison of linkage-disequilibrium measures for fine-scale
mapping. Genomics 29, 311–322.
Devlin B, Roeder K (1999). Genomic control for association studies. Biometrics 55, 997–1004. Devlin B, Risch N, Roeder K (1996). Disequilibrium mapping: composite likelihood for pairwise
disequilibirum. Genomics 36, 1–16.
Drysdale CM, McGraw DW, Stack CB et al. (2000). Complex promoter and coding region 2-
adrenergic receptor haplotypes alter receptor expression and predict in vivo responsiveness.
Proc Natl Acad Sci USA 97, 10483–10488.
Ewens WJ, Spielman RS (1995). The transmission/disequillibrium test: history, subdivision and
admixture. Am J Hum Genet 57, 455–464.
Fliess JL (1981). Statistical Methods for Rates and Proportions. New York: John Wiley. Goodnight CJ (2000). Modeling gene interaction in structure populations. In Wolf JB, Brodie ED
III, Wade MJ, eds. Epistasis and the Evolutionary Process. New York: Oxford University Press,
pp. 129–145.
Halushka MK, Fan JB, Bentley K et al. (1999). Patterns of single-nucleotide polymorphisms in
candidate genes for blood-pressure homeostasis. Nat Genet 22, 239–247.
Hästbacka J, de la Chapelle A, Mahtani MM et al. (1994). The diastrophic dysplasia gene encodes a novel sulfate transporter: positional cloning by fine-structure linkage disequilibrium
mapping. Cell 78, 1073–1087.
89 Statistical approaches in psychopharmacogenetics
Hoehe MR, Köple K, Wendel B et al. (2000). Sequence variability and candidate gene analysis in complex disease: association of opioid receptor gene variation with substance dependence. Hum Mol Genet 19, 2895–2908.
International SNP Map Working Group (2001). A map of human genome sequence variation containing 1.42 million single nucleotide polymorphisms Nature 409, 928–923.
Jacob HJ, Lindpaintner K, Lincoln SE et al. (1991). Genetic mapping of a gene causing hyperten- sion in the stroke-prone spontaneously hypertensive rat. Cell 67, 213–224.
Johnson GCL, Todd JA (2000). Strategies in complex disease mapping. Curr Opin Genet Dev 10, 330–334.
Jorde LB (2000). Linkage disequilibrium and the search for complex disease genes. Genome Res 10, 1035–1044.
Goddard KAB, Hopkins PJ, Hall JM, Witte JS (2000). Linkage disequilibrium and allele- frequency distributions for 114 single-nucleotide polymorphisms in five populations. Am J Hum Genet 66, 216–234.
Kalow W, Bertilsson L (1994). Interethnic factors affecting drug response. Adverse Drug Res 23, 1–53.
Kerem B, Rommens JM, Buchanan JA et al. (1989). Identification of the cystic fibrosis gene: genetic analysis. Science 245, 1073–1080.
Kidd KK, Morar B, Castiglione CM et al. (1998). A global survey of haplotype frequencies and linkage disequilibrium at the DRD2 locus. Hum Genet 103, 211–227.
Kingman JFC (2000). Origin of the colascent: 1974–1982. Genetics 156, 1461–1463.
Kruglyak L (1999). Prospects for whole-genome linkage disequilibrium mapping of common
disease genes. Nat Genet 22, 139–144.
Lam JC, Roeder K, Devlin B (2000). Haplotype fine mapping by evolutionary trees. Am J Hum
Genet 66, 659–673.
Lazzeroni LC (1998). Linkage disequilibrium and gene mapping: an empirical least-squares
approach. Am J Hum Genet 62, 159–170.
Lewontin RC (1964). The interaction of selection and linkage. 1. General considerations; hetero-
tic models. Genetics 49, 49–67.
Lifton RP (1996). Molecular genetics of human blood pressure variation. Science 272,
676–680.
Lio’ P, Goldman N (1998). Models of molecular evolution and phylogeny. Genome Res 8,
1233–1244.
Mander A, Clayton D (2000). Haplotype frequency estimation using an EM algorithm and log-
linear modeling. Stat Tech Bull 57, 5–7.
Mantel N, Haenszel W (1958). Statistical aspects of the analysis of data from retrospective studies
of disease. J Nat Cancer Inst 4, 719–748.
Martin ER, Gilbert JR, Lai EH et al. (2000). Analysis of association at single nucleotide polymor-
phisms in the APOE region. Genomics 63, 7–12.
McPeek MS, Strahs A (1999). Assessment of linkage disequilibrium by the decay of haplotype
sharing, with application to fine-scale genetic mapping. Am J Hum Genet 65, 858–875. Monks SA, Kaplan NL (2000). Removing the sampling restrictions from family-based tests of
association for a quantitative-trait locus. Am J Hum Genet 66, 576–592.
90 F. Macciardi
Nebert DW (1999). Pharmacogenetics and pharmacogenomics: why is this relevant to the clini- cal geneticist? Clin Genet 56, 247–258.
Nickerson DA, Taylor SL, Weiis KM et al. (1998). DNA sequence diversity in a 9.7-kb region of the human lipoprotein lipase gene. Nat Genet 19, 233–240.
Ott J (1999). Analysis of Human Genetic Linkage, 3rd edn. Baltimore: Johns Hopkins University Press.
Ott J (2000). Predicting the range of linkage disequilibrium. Proc Natl Acad Sci, USA 97, 2–3. Ott J, Hoh J (2000). Statistical approaches to gene mapping. Am J Hum Genet 67, 289–294. Phillips PC (1998). The language of gene interaction. Genetics 149, 1167–1171.
Pritchard JK (2001). Are rare variants responsible for susceptibility to complex diseases? Am J
Hum Genet 69, 124–137.
Pritchard JK, Przeworki M (2001). Linkage disequilibrium in humans: models and data. Am J
Hum Genet 69, 1–14.
Pritchard JK, Rosenberg NA (1999). Use of unlinked genetic markers to detect population strat-
ification in association studies. Am J Hum Genet 65, 220–228.
Rieder MJ, Taylor SL, Clark AG, Nickerson DA (1999). Sequence variation in the human angio-
tensin converting enzyme. Nat Genet 22, 59–62.
Risch N (2000). Searching for genetic determinants in the new millennium. Nature 405,
847–856.
Risch N, Merikangas K (1996). The future of genetic studies of complex human diseases. Science
273, 1561–1517.
Roses AD (2000). Pharmacogenetics and the practice of medicine. Nature 405, 857–865.
Schaid DJ, Sommer SS (1994). Comparison of statistics for candidate-gene association studies
using cases and parents. Am J Hum Genet 55, 402–409.
Schaid DJ, Sommer SS (1993). Genotype relative risks: methods for design and analysis of can-
didate-gene association studies. Am J Hum Genet 53, 1115–1126.
Schork NJ, Krieger JE, Trolliet MR et al. (1995). A biometrical genome search in rats reveals the
multigenic basis of blood pressure variation. Genome Res 5, 164–172.
Seltman H, Roeder K, Devlin B (2001). Transmission/disequilibrium test meets measured hap-
lotype analysis: family-based association analysis guided by evolution of haplotypes. Am J
Hum Genet 68, 1250–1263.
Service SK, Lang DW, Freimer NB, Sandkuijl LA (1999). Linkage-disequilibrium mapping of
disease genes by reconstruction of ancestral haplotypes in founder populations. Am J Hum
Genet 64, 1728–1738.
Sham PC (1998). Statistical methods in psychiatric genetics. Stat Meth Med Res 7, 279–300. Sokal RR, Rohlf FJ (1995). Biometry, 3rd edn. San Francisco: Freeman.
Spielman RS, Ewens WJ (1996). Invited editorial: the TDT and other family-based tests for
linkage test for linkage disequilibrium and association. Am J Hum Genet 59, 983–989. Spielman RS, Ewens WJ (1998). A sibship test for linkage in the presence of association: the sib
transmission/disequilibrium test. Am J Hum Genet 62, 450–458.
Spielman RS, McGinnis RE, Ewens WJ (1993). Transmission test for linkage disequilibrium: the
insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 56, 777–787.
91 Statistical approaches in psychopharmacogenetics
Terwilliger JD (1995). A powerful likelihood method for the analysis of linkage-disequilibrium between trait loci and one or more polymorphic markers. Am J Hum Genet 56, 777–787.
Terwilliger JD, Weiss KM (1998). Linkage disequilibrium mapping of complex disease: fantasy or reality? Curr Opin Biotechnol 9, 578–594.
Tishkoff SA, Goldman A, Calafell F et al. (1998). A global haplotype analysis of the myotonic dys- trophy locus: implications for the evolution of modern humans and for the origin of myotonic dystrophy mutations. Am J Hum Genet 62, 1389–1402.
van Hauwe P, Coucke PJ, Declau F et al. (1999). Deafness linked to DFNA2: one locus but how many genes? Nat Genet 21, 363.
Venter J et al. (2001). The sequence of human genome. Science 291, 1304–1351.
Vogel F (1959). Moderne probleme der Humangenetik. Ergebn Inn Med Kinderheilk 12, 52–125. Weiss KM, Terwilliger JD (2000). How many diseases does it take to map a gene with SNPs? Nat
Genet 26, 151–157.
Wright AF, Carothers AD, Pirastu M (1999). Population choice in mapping genes for complex
diseases. Nat Genet 23, 397–404.
Xiong M, Guo SW (1997). Fine-scale genetic mapping based on linkage disequilibrium: theory
and application. Am J Hum Genet 60, 629–640.

Part III
Molecular background


6
The psychopharmacogenetic– neurodevelopmental interface in serotonergic gene pathways
K. Peter Lesch, Jens Benninghoff and Angelika Schmitt Department of Psychiatry and Psychotherapy, University of Wuerzburg, Wuerzburg, Germany
OVERVIEW
Individual differences in drug effects and treatment response are relatively enduring, con- tinuously distributed, and substantially heritable; they are, therefore, likely to result from an interplay of multiple genomic variations with environmental influences. As the etiology and pathogenesis of behavioral and psychiatric disorders are genetically complex, so is the response to drug treatment. Psychopharmacological drug response depends on the struc- ture and functional expression of gene products, which may be direct drug targets or may indirectly modify the development and synaptic plasticity of neural networks critically involved in drug response. While formation and integration of these neural networks is dependent on the action of manifold proteins, converging lines of evidence indicate that genetically controlled variability in the expression of genes critical to the development and plasticity of distinct neurocircuits influences a wide spectrum of quantitative traits including treatment response. During brain development, neurotransmitter systems (e.g., the seroto- nergic system), which are frequently targeted by psychotropic drugs, control neuronal spec- ification, differentiation, and phenotype maintenance. The formation and maturation of these neurotransmitter systems, in turn, is directed by an intrinsic genetic program. Based on the notion that complex gene–gene and gene–environment interactions in the regula- tion of brain plasticity contribute to interindividual differences in drug response, the concept of developmental psychopharmacogenetics is emerging. This chapter appraises prototypi- cal genomic variation with impact on gene expression, and complementary studies of gene and environmental effects on brain development and synaptic plasticity in the mouse model. Although special emphasis is given to molecular mechanisms of neurodevelopmen- tal genetics, relevant conceptual and methodological issues pertinent to the dissection of the psychopharmacogenetic–neurodevelopmental interface are also considered.
Introduction
Psychopharmacogenetics is an emerging scientific discipline examining the genetic basis for individual variations in response to psychotherapeutic drugs (Catalano,

95

96 K. P. Lesch, J. Benninghoff, and A. Schmitt
1999). While pharmacogenetics cannot improve the efficacy of a given drug, it may help in selecting patients who are likely to respond well. Psychopharmacological drug response depends on the structure and functional expression of a large number of gene products representing either pharmacokinetic or pharmacody- namic determinants. For most drugs, variations in drug response have, until recently, been considered a result of pharmacokinetic rather than pharmacody- namic differences. However, it now seems that pharmacodynamic variability in humans is large, reproducible, and usually more pronounced than pharmacoki- netic variability. Many drug targets (e.g., receptors, transporters, enzymes) that contribute to the pharmacodynamics of drug response are not only key players in the regulation of neurotransmitter systems but also directly or indirectly modify the development and plasticity of neural networks involved in drug effects. Both variation of structure (which is rare) and variation of expression (which is common) influence gene product availability and function.
Gene expression involves many facets of regulation ranging from transcription, transcript processing, translation, post-translational modification and intracellu- lar trafficking. It is controlled by multiple regulatory proteins (Lesch and Heils, 2000). These regulatory proteins represent gene products themselves and have been expressed sequentially following an intrinsic genetic program. This developmental blueprint controls the hierarchical cascade of gene expression whose different stages subject to similar regulatory principles. There is now considerable evidence that variability of expression of genes critical to neurocircuit development and function influences drug responses. This chapter describes fundamental aspects of the genetics of complex traits including drug response, provides an appraisal of quantitative genetic research, and reviews genomic variations with impact on gene expression.
Psychopharmacological drug response as a complex trait
Considerable evidence has been assembled that treatment response is influenced by genetic factors and that the genetic component is highly complex, polygenic, and epistatic. Moreover, treatment response is believed to involve both genetic and environmental factors. Most likely, the contribution of single genes to drug response is modest, if not minimal. However, interactions between different genes could result in a dramatic modification of drug response (additive, nonadditive or synergistic gene effects). The challenge is to identify genes of relative small effect against a background of substantial genetic and environmental variation.
Many genes that influence complex traits and psychopharmacological drug responses are likely to be distributed continuously. Such genes are, therefore, referred to as quantitative trait loci (QTLs). In contrast to the focus of quantitative

97 Serotonergic gene pathways
genetics on phenotypes of interest and on naturally occurring genetic variation responsible for phenotypic differences, molecular genetics focused on genes and techniques to create new mutations in model organisms such as the worm (Caenorhabditis elegans), fruit fly (Drosophila melanogaster), or mouse in order to investigate how genes work. The pace of integration of quantitative and molecular genetics has been accelerated remarkably as a result of the Drosophila, Mouse, and Human Genome Projects, which have opened the postgenomic era in which the genome of humans and other species is known. While several million DNA varia- tions (polymorphisms) have been identified in the genomic sequence, approxi- mately 30 000–60 000 common polymorphisms are located in coding and regulatory regions of genes that are the ultimate causes of the heritability of complex traits (McPherson et al., 2001; Sachidanandam et al., 2001). New technol- ogies, such as DNA microarrays, have made it possible to investigate the role of thousands of DNA variants in complex traits. Because behavioral traits are the most complex traits of all, response to psychopharmacological drugs, which modify behavior, is likely to profit from this integration. Moreover, behavioral pharmacogenetics will make a major contribution to functional genomics. Although functional genomics has been equated with a bottom-up approach that begins with genes and their protein products in cells – and the importance of this level of analysis is not doubted – the phenotypical level of analysis is also crucial and should provide benefits in terms of diagnosing, treating, and preventing behavioral and psychiatric disorders. The term pharmacogenomics emphasizes the importance of the behavioral level of analysis in understanding how genes work at the developmental interface between the organism and its environment. Bottom- up molecular approaches will eventually meet top-down behavioral research in the brain, the ultimate target for functional genomic analysis of behavior and treat- ment response.
Several genes related to monoaminergic neurotransmission are currently under investigation, including the dopaminergic, noradrenergic, and serotonergic systems, which may contribute to the genetic variance of response to antidepress- ant, anxiolytic, and antipsychotic treatment. However, molecular genetics has so far failed to identify a genomic variation that can consistently contribute to psycho- pharmacological drug responses (Stoltenberg and Burmeister, 2000). Since compo- nents of the monoaminergic neurotransmitter system not only participate in the function and plasticity of the adult brain but also represent key players in the devel- opment of the brain, genetic variation in neurodevelopment further complicates assessment of the interplay of genotype and drug efficacy. Emerging genetic and genomic techniques now offer the prospect of identifying functionally relevant gene variants that participate in the development and adult plasticity of the brain relating to psychopharmacological drug responses.
98 K. P. Lesch, J. Benninghoff, and A. Schmitt
Variation of structure and expression of pharmacodynamic factors
The pace of discovery of genes associated with complex traits, such as drug response, is currently increasing as systematic QTL scans can be conducted using functional DNA variants in the brain that affect coding regions or gene regulation. Of a total approximately 40000 genes, approximately 50% are expressed in the brain. However, a substantial proportion of these are housekeeping genes, which provide structure or maintain basic physiological functions of cells. Consequently, fewer than 15000 tissue-specific genes may be expressed in the brain. Identifying the estimated 8000–16000 functional DNA variants in coding regions of these genes and the DNA variants that regulate the expression of these genes is a high pri- ority for pharmacogenetic research because these genes are likely to be the source of the heritable influence on psychopharmacological drug response. Several classes of genomic variation in coding and regulatory regions of genes are of particular interest. These range from single nucleotide polymorphisms (SNPs) to microdele- tions (or insertions), and polymorphic simple sequence repeats (SSRs) of 2 to 50 nucleotides in length (Fig. 6.1).
Although many features of drug targets, their protein components or subunits, and their spliced or edited isotypes have been revealed, less is known about the genetic elements and transcription factors involved in the regulation of gene expression. Even less still is known about the impact of genomic variation within regulatory regions and elements on gene expression. Nevertheless, there is consen- sus that the basal promoter and associated regulatory elements are located upstream of a transcribed gene, although adjunct regulatory units may be located across the entire region of a gene. In contrast to constitutively active core promot- ers, which effectively drive expression in a large variety of tissues, tissue-selective regulatory elements, which are frequently located upstream of the core promoter unit, confer cell-restricted activity such that the gene is transcribed only in distinct cell types, such as neurons or glial cells (Fig. 6.1).
DNA variants in coding and regulatory regions of genes will be both useful for systematic genome scans, for identifying genes associated with drug response, and for examining integrated systems of gene pathways as an important step on the route to functional genomics. Initiatives focused on genes expressed in the brain and the creation of a brain “transcriptome” map in which gene expression is mapped throughout the brain in mice would greatly facilitate postgenomic research on the links between genes, brain, behavior, and treatment response. Connecting drug response with relevant functional DNA variants in the brain and with differences in gene expression in brain regions represents the ultimate goal for pharmacogenetic research. The serotonergic gene pathways may be a rewarding area of investigation because of the numerous and essential functions of the central

99 Serotonergic gene pathways

Fig. 6.1.
Gene organization, transcriptional control region, and common functionally relevant genomic variations. Frequency of initiation of mRNA synthesis depends on transcription factors that interact with specific elements in the gene promoter, other regulatory sites, and optional locus control regions. The transcription initiation complex comprises multiple factors, including the TATA box-binding protein (TBP) with TBP-associated factors, coactivators, and basal factors. These components associate to a complex with RNA polymerase that are bound to the TATA and enhancer (Enh) motifs. Several structures and motifs characteristic for binding of activators are indicated. Activators bind to enhancer motifs and transduce their signals to coactivators; they control gene activation by increasing transcription rate. Repressors bind to silencer elements (Sil), inhibit activator function and slow down transcription. Coactivators are adapter molecules that integrate signals of activators/repressors and transfer the information on basal factors. Basal factors facilitate binding and activation of RNA polymerase at the transcription start site.
100 K. P. Lesch, J. Benninghoff, and A. Schmitt

Serotonergic neuron
  
Signal transduction
        
Transcription factors
Neurotrophin receptorTrkB

Tryptophan transporter
  
Tryptophan hydroxylase
Serotonin receptors
               
Aromatic L-amino acid decarboxylase
Vesicular monoamine transporter
Monoamine oxidase A
          
Synaptic release apparatus
Serotonin transporter
  
Postsynaptic neuron
Serotonin receptors
Heteroceptors e.g. α2
   
Fig. 6.2.
Serotonergic gene pathways and different components of serotonergic system development, plasticity, and function currently under psychopharmacogenetic investigation.
serotonin (5-HT) system, the striking wealth of drug targets within this system, and the impressive range of serotonergic compounds available in the clinical setting (Fig. 6.2).
Regulators of signal transduction
Representative genomic variants in serotonergic gene pathways
Serotonin receptors
Ligand-binding experiments and the study of functional responses to ago- nists/antagonists initially defined four 5-HT receptor subfamilies, 5-HT1–4. Molecular biology has subsequently both confirmed this classification and also revealed the existence of novel 5-HT receptor subtypes for which few pharmaco- logical or functional data exist (5-HT1E/F, 5-HT3A/B, 5-HT5A/B, 5-HT6, and 5-HT7) (Hoyer and Martin, 1997; Barnes and Sharp, 1999). In the genes for 5-HT2–7 recep- tors, the coding region is interupted by introns, whereas the genes for 5-HT1A-F

Neurotrophins and growth factors
101 Serotonergic gene pathways
receptors contain no introns. The genes for 5-HT2B, 5-HT4, 5-HT6, and 5-HT7 receptors are alternatively spliced, and RNA editing of 5-HT2C receptor subtype in the second intracellular loop confers differential receptor functionality, thus increasing the complexity of the 5-HT receptor superfamily (Gerald, 1995; Canton et al., 1996; Heidmann, 1997; Olsen et al., 1999). The present challenge is to deter- mine the physiological relevance of these gene products, establish their functional- ity as endogenous receptors, find selective ligands, and determine potential therapeutic application of these compounds.
The molecular characterization of different 5-HT receptor families has sim- plified the elucidation of gene transcription, mRNA processing and translation, intracellular trafficking, and post-translational modification relevant to synaptic and postreceptor signaling (for review see Lesch and Heils (2000)). Transcriptional control regions have been cloned for several 5-HT receptor subtypes, and func- tional promoter mapping data are available for the genes for 5-HT1A, 5-HT2A, 5-HT2C, and 5-HT3 receptors. The analysis of genomic regulatory regions of 5-HT receptor genes and modeling variable 5-HT receptor gene function in genetically engineered mice (constitutive and conditional knockout) provides critical know- ledge regarding the respective role of these receptors in neurodevelopment, synap- tic plasticity, and behavior (Bonasera and Tecott, 2000). For example, behavioral evaluation of 5-HT1B knockout mice has implicated this 5-HT autoreceptor in alcohol-seeking and aggressive behavior (Brunner et al., 1999). Mice with targeted disruption of the 5-HT2C gene display anxiety-related behavior, hyperphagia- evoked weight gain, and deficits in spatial learning (Tecott and Barondes, 1996).
Several potentially functional variations in genes of 5-HT receptors have recently been associated with behavioral traits, psychopathological conditions, and psycho- pharmacological drug response. Since this area has recently been reviewed compre- hensively (for example, in Cichon et al. (2000) and Veenstra-VanderWeele et al. (2000)), only a few prototypical polymorphisms with evidence for functional impact on gene expression and function will be discussed here. Studies in a Finnish population and a southwestern American Indian tribe revealed that the silent Gly861Cys (G861C) polymorphism in the 5-HT1B receptor is associated with anti- social alcoholism. In a family-based study, a polymorphism in the 5 -regulatory region of the gene for the 5-HT2A (Ala1438Gly (A1438G)) was associated with schizophrenia (Spurlock et al., 1997). Analysis of the A1438G polymorphism revealed no effect of genotype on basal or cyclic adenosine monophosphate (cAMP) and protein kinase C-induced gene transcription in a cell model and no difference in lymphocyte 5-HTR2A mRNA expression between 1438G/G and A/A homozygotes. However, a preliminary autopsy study demonstrated higher prefron- tal 5-HT2A binding in subjects with the 1438A allele (Turecki et al., 1999). The common polymorphism Cys23Ser (C23S) in the coding region of the 5-HT2C
102 K. P. Lesch, J. Benninghoff, and A. Schmitt
receptor gene shows only weak associations with schizophrenic psychosis but influ- ences psychotic symptoms, clinical course (including duration of hospitalization) and drug response (Cichon et al., 2000). A recently identified polymorphic com- pound dinucleotide repeat in the 5 -regulatory region of the 5-HT2C gene (J. Meyer and K.P. Lesch, unpublished data), which is unique to humans and nonhuman pri- mates, contributes to the predictive power of several variants in serotonergic gene pathways (i.e., 5-HT2A, 5-HT2C, 5-HT transporter (5-HTT)) for clozapine response in schizophrenia (Arranz et al., 2000), whereas no association with panic disorder was detected (Deckert et al., 2000).
Serotonin transporter
In humans, transcriptional activity of the gene for the transporter 5-HTT is mod- ulated by a polymorphic repetitive element (5-HTT gene-linked polymorphic region, 5-HTTLPR) located upstream of the transcription start site. Additional variations have been described in the 5 untranslated region (5 -UTR) through alternative splicing of exon 1B (Bradley and Blakely, 1997), in intron 2 (a 16/17 var- iable number tandem repeat, VNTR-17) (Lesch et al., 1994), and in the 3 UTR (Battersby et al., 1999). Comparison of different mammalian species confirmed the presence of the 5-HTTLPR in simian primates but not in prosimian primates and other mammals (Lesch et al., 1997). The majority of alleles are composed of either 14- or 16-repeat elements in humans (short (s) and long (l) allele, respectively), while alleles with 15, 18, 19, 20, or 22 repeat elements and variants with single-base insertions/deletions or substitutions within individual repeat elements are rare. A predominantly Caucasian population displayed allele frequencies of 57% for the l allele and 43% for the s allele, with a 5-HTTLPR genotype distribution of 32% l/l, 49% l/s, and 19% s/s (Lesch et al., 1996). Different allele and genotype distributions were found in other populations (Gelernter et al., 1997; Ishiguro et al., 1997; Kunugi et al., 1997).
The unique structure of the 5-HTTLPR gives rise to the formation of DNA sec- ondary structure that has the potential to regulate the transcriptional activity of the associated 5-HTT gene promoter. When fused to a luciferase reporter gene and transfected into human 5-HTT-expressing cell lines, the s and l 5-HTTLPR vari- ants differentially modulate transcriptional activity of the 5-HTT gene promoter (Lesch et al., 1996). The effect of 5-HTTLPR length variability on 5-HTT function was determined by studying the relationship between 5-HTTLPR genotype, 5-HTT gene transcription, and 5-HT uptake activity in human lymphoblastoid cell lines. Cells homozygous for the l variant of 5-HTTLPR produced higher concen- trations of 5-HTT mRNA than cells containing one or two copies of the s form. Membrane preparations from l/l lymphoblasts showed higher inhibitor binding than did s/s cells. Furthermore, the rate of specific 5-HT uptake was more than
103 Serotonergic gene pathways
twofold higher in cells homozygous for the l form of the 5-HTTLPR than in cells carrying one or two copies of the s variant of the promoter. The association of the s form with lower 5-HTT expression and function is supported by studies of 5-HTT promoter activity in other cell lines (Mortensen et al., 1999), mRNA concentrations in the raphe complex of human postmortem brain (Little et al., 1998), platelet 5- HT uptake and content (Hanna et al., 1998; Greenberg et al., 1999; Nobile et al., 1999) and in vivo SPECT (single photon emission computed tomography) imaging of human brain 5-HTT (Heinz et al., 1999).
The secondary structure of the 5-HTTLPR is also likely to precipitate a 381bp somatic deletion in the promoter region (del(17)(q11.2)) of the 5-HTT gene, observed in 20–60% of genomic DNA isolated from human brain and mononu- clear cells (Lesch and Mössner, 1999). The localization of the deletion breakpoints adjacent to identical putative signal sequences suggests a recombinase-like rear- rangement event. This suggests that mosaicism of the 5-HTT gene promoter- associated deletion is likely to be regulated by brain region-selective and possibly 5-HTTLPR-dependent mechanisms. This extraordinary feature provides further evidence for complex 5-HTT gene organization and regulation.
A growing body of evidence suggests a role of 5-HTTLPR-dependent allelic vari- ation in 5-HTT expression and function in anxiety-, depression-, and aggression- related personality traits and syndromal dimensions of various psychiatric disorders (for review see Lesch, 2001a). The finding that individuals with reduced 5-HTT function are at risk to develop affective illness would seem at odds with the fact that 5-HT reuptake inhibitors (SRIs), which competitively block 5-HT uptake, are effective in anxiety disorders and depression. The regional variation of 5-HTT expression and the complex autoregulatory processes of 5-HT function that are operational in different brain areas may lead to a plausible hypothesis to explain this apparent contradiction (Routledge and Middlemiss, 1996). The impaired ability for rapid 5-HT clearance associated with the s-allele of 5-HTTLPR follow- ing 5-HT release into the synaptic cleft, may elicit acute increases of 5-HT in the vicinity of serotonergic cell bodies and dendrites in the raphe complex and may exert a somatodendritic 5-HT1A receptor-mediated negative feedback that leads to an overall decrease of 5-HT neurotransmission. By comparison, chronic SRI treat- ment induces adaptive changes in the 5-HTT/5-HT1A receptor-modulated negative feedback regulation that eventually leads to an overall enhancement of terminal 5- HT. In this regard it has been proposed that concurrent antagonism of this autore- ceptor during 5-HT reuptake blockade may have the potential to accelerate the antidepressant effect of 5-HTT inhibition, which could be particularly advanta- geous in 5-HTTLPR genotype-related SRI nonresponders (Blier and de Montigny, 1997; Smeraldi et al., 1998).
Based on theoretical consideration, a complex interaction between genotype,
104 K. P. Lesch, J. Benninghoff, and A. Schmitt

behavioral or syndromal dimensions, and drug response has been predicted (Catalano, 1999). A given genetic predisposition, such as allelic variation in 5-HTT function, may lead to increased susceptibility to anxious or depressive features and to less favorable antidepressant responses in patients affected by mood disorders. Impaired 5-HTT function apparently confers only a very modest susceptibility, if any, to depressed states, because adaptive mechanisms are likely to compensate for the deficiency, while more robust alterations of 5-HT turnover observed during antidepressant treatment may reveal 5-HTTLPR genotype effects that lead to var- iable SRI efficacy. Smeraldi and associates (1998) investigated whether the 5- HTTLPR genotype is related to the antidepressant response to the SRI fluvoxamine and/or to augmentation with the 5-HT1A receptor antagonist pindolol. Their study included patients with major depression with psychotic features (n 102) who had been randomly assigned to treatment with a fixed dose of fluvoxamine and either placebo or pindolol for 6 weeks. Both l/l homozygotes and l/s heterozygotes showed a better response to fluvoxamine than for 5-HTTLPR s/s homozygotes. In the group treated with fluvoxamine plus pindolol, all the genotypes acted like l/l homozygotes treated with fluvoxamine alone. Thus, SRI efficacy in delusional depression seems to be related, in part, to allelic variation within the promoter of the 5-HTT gene. This 5-HTTLPR genotype effect on antidepressant response has recently been rep- licated in an independent sample of depressed patients treated with the SRI parox- etine (Zanardi et al., 2000). Furthermore, an interaction between 5-HTTLPR genotype and therapeutic efficacy of the antimanic/antibipolar agent lithium, which is assumed to act via serotonergic mechanisms, was demonstrated (Del Zompo et al., 1999). Finally, drug-free patients with bipolar depression who were l/l homozygotes for 5-HTTLPR showed better mood improvement after total sleep deprivation than those with the s/l and s/s genotypes (Benedetti et al., 1998). These findings support the notion that 5-HTTLPR genotyping may represent a useful pharmacogenetic tool to individualize treatment of depression and that 5-HTT function is critical for the antidepressant mechanism of action of sleep deprivation.
Monoamine oxidase
Monoamine oxidase A (MAO-A) oxidizes 5-HT, norepinephrine, and dopamine. It is expressed in a cell type-specific manner. Abnormalities in MAO-A activity have been implicated in a wide range of behavioral and psychiatric disorders (for review see Lesch and Merschdorf (2000)). Interestingly, deficiency in MAO-A owing to a hemizygous chain termination mutation in its gene has been shown to be asso- ciated with impulsive aggression and hypersexual behavior in affected males from a single large family. Transgenic mice lacking the gene for MAO-A exhibit aggres- sive behavior in adult males.
While there is considerable controversy regarding the site where MAO-A mRNA
105 Serotonergic gene pathways
synthesis is initiated, tissue-specific length variability of the 5 -UTR has been reported, with multiple transcription start sites clustered primarily around an initiator element, which may also act as a negative cis element (Denney et al., 1994; Zhu et al., 1994; Zhu and Shih, 1997). The core promoter region contains two 90 bp repeat sequences, which are further divided into four imperfect tandem repeats, each containing an Sp1 binding site in reversed orientation. A polymorphic 30bp repeat was recently identified in the promoter region of the MAO-A that differen- tially modulated gene transcription (Sabol et al., 1998; Deckert et al., 1999). Variation in the number of repeats (3 to 5) of this MAO-A gene-linked polymor- phic region (MAO-A-LPR) resulted in different transcriptional efficiencies when the gene was fused to a luciferase reporter gene and transfected into cell lines. The transcriptional efficiency of the 3-repeat allele was twofold lower than those with longer repeats (3a, 4, and 5). Interestingly, there was an up to eightfold higher MAO-A activity in human male skin fibroblasts associated with the 4-repeat MAO- A-LPR genotype ((Denney et al., 1999)). Jonsson and associates (2000)) found increased cerebrospinal fluid (CSF) levels of hydroxyindoleacetic acid (5-HIAA) and homovanillic acid in women with at least one copy of the 3.5- or 4-repeat pro- moter allele, but not in men. This finding further supports the gender-specific effect of the MAO-A-LPR previously reported in panic disorder (Deckert et al., 1999).
The MAO-A-LPR is, therefore, an attractive genomic variation to investigate behavioral disorders that are associated with abnormalities in monoaminergic transmission as well as response to psychopharmacological drugs affecting these neurotransmitter systems. Recent studies showed associations of MAO-A-LPR length variation in alcoholics with dissocial behavior and in female patients with panic anxiety (Deckert et al., 1999; Samochowiec et al., 1999). The findings suggest that the 3-repeat allele of the MAO-A-LPR confers vulnerability to dissocial behav- ior rather than alcohol-seeking behavior in alcohol-dependent males. Among females with panic disorder, the longer alleles (3a, 4, and 5) were more frequent than in the corresponding control populations. Together with the observation that inhibition of MAO-A is clinically effective in the treatment of panic disorder, par- ticularly in women, these results suggest that increased MAO-A activity is a risk factor for panic disorder in females.
Brain development and serotonin
While being important pharmacodynamic factors themselves, components of monoaminergic neurotransmitter systems such as receptors, transporters, and modifying enzymes also participate in brain development and thus set the stage for brain (dys)function and influence presence as well as function of drug targets and many other pharmacodynamically important factors.

106 K. P. Lesch, J. Benninghoff, and A. Schmitt

Phases of cortical development
Arrival and maturity of major cortical afferent systems
Maturation

Migration Neurogenesis
                        
E13 14
15 16
ACh
17 18 19
Ectopia
20 birth
Dysplasia
PND14 21 28 60 Developmental age
  
Fig.6.3.
Neurotransmitters in brain development and plasticity. Timeline for different phases of cortical development and for the arrival and maturity of major cortical afferents. ACh, acetylcholine; DA, dopamine; NE, norepinephrine; 5HT, 5-hydroxytryptamine (serotonin); PND, postnatal day. (Modified from Berger-Sweeney and Hohmann, 1997.)
Development of the brain, including the neocortex, involves a complex series of rigorously timed stages that are subdivided into generation, migration, and differ- entiation of neurons and glia (Berger-Sweeney and Hohmann, 1997) (Fig. 6.3). These events occur during distinct time windows that span the perinatal period and create intricate neural circuits and networks critical in the integration of sensory and cognitive functions and aspects of behavioral responses. Evidence indicates that variation in gene function as well as environmental factors at certain stages of neurodevelopment may lead to circumscribed alterations including hypoplasia (reduced cell number), ectopia (abnormality in migration), and dysplasia (altera- tion in number and structure). Harmonizing these sequential stages and reaching developmental milestones is critical to guide assembly of neuronal cells from differ- ent regions of the brain at the appropriate times and locations to form functional circuits. Several models of manipulations that affect different stages of cortical development, particularly those that compromise ontogeny of neurotransmitter- defined afferent systems, have implicated acetylcholine, norepinephrine, dopa- mine, and 5-HT (Fig. 6.3). Given the complexity of the processes that will eventually be regulated by these circuits, it is conceivable that neurodevelopment is exquisitely prone to allelic variation in functional gene expression.
Hypoplasia
DA

NE

5HT
      
107 Serotonergic gene pathways
There are several lines of evidence that 5-HT system homeostasis is critical to the genesis, differentiation, and maturation of neuronal cells and networks in brain regions controlling sensory inputs, stimulus processing, and motor output. 5-HT is a mitogenic and morphogenetic factor as well as a differentiation signal in corti- cal development. For instance, 5-HT modulates neurite outgrowth of thalamocor- tical glutamatergic neurons in culture, which involves both 5-HT1A and 5-HT1B receptors (Lavdas et al., 1997). Furthermore, it promotes differentiation of cortical glutamatergic neurons via 5-HT1B (Lieske et al., 1999), increases the probability of long-term potentiation in the visual cortex by activation of 5-HT2C (Kojic et al., 2000), and stimulates hippocampal neurogenesis resulting in antidepressant effects, which is likely to be mediated by 5-HT1A (Gould, 1999; Malberg et al., 2000).
Serotonergic projections to cortical regions including sensory areas come almost exclusively from the dorsal and median raphe nuclei of the brainstem. Raphe neurons are generated in mice at embryonic days (E) 11–15 and arrive at cortical areas at about E17. At birth, serotonergic fibers infiltrate all cortical layers and display a transient profuse distribution pattern; this is pruned considerably to the pattern of the mature brain around postnatal week 3. Subtle alterations in cortical morphogenesis have been reported following perinatal pharmacologically induced 5-HT depletion. In the rat, emergence of the thalamocortical innervation pattern is delayed and whisker representation in the somatosensory cortex is reduced, although maturation (i.e., overall somatotropic organization of thalamocortical afferents) is eventually achieved. Other indirect evidence further supports a role for 5-HT in cortical development and behavior. Treatment of neonatal rats with MAO inhibitors reduces serotonergic projections to the cortex and impairs passive avoid- ance in juvenile rats. In mice, prenatal hypoxia delays serotonergic and other fiber ingrowth into the parietal cortex and induces hyperactivity and impairment in attention and spatial memory in adult mice. In general, mouse models have become essential as behavioral pharmacogenetics enters the postgenomics era and as the field moves beyond identifying genes associated with drug response toward understanding how they accomplish their effect.
Mouse models and the emerging concept of developmental pharmacogenetics
Mouse models increasingly contribute to pharmacogenomic research, especially in terms of brain mechanisms and circuits that mediate genetic effects. In addition to the ability to manipulate the genome through transgenic and knockout mice, the models also make it possible to control and to manipulate the environment, which will facilitate elucidation of gene–gene and gene–environment interactions. Although mouse models already play an important role in behavioral pharmaco- genetics, there is consensus that these models are quite limited behaviorally (Lesch,

108 K. P. Lesch, J. Benninghoff, and A. Schmitt
2001b). In other areas such as emotionality and psychopathology, few mouse models are available. What is especially needed are batteries of multiple tasks that can be used to assess a latent construct relatively free of test-specific factors (Crawley and Paylor, 1997; Crawley, 1999). Nonetheless, mice provide a practical model to study the impact of gene knockouts on development and the plasticity of the brain, including regionalization and connectivity of the cerebral cortex and subcortical structures. Obviously, mouse models are essential in the dissection of the psychopharmacogenetic–neurodevelopmental interface.
It has been suggested that the differentiation of cortical areas is controlled by an interplay of intrinsic genetic programs (sequential activation of transcription factors and expression of cell adhesion molecules, such as cadherins) and extrinsic mechanisms including those mediated by thalamocortical afferents (Nakagawa et al., 1999). Both transcription factors and cell adhesion molecules display graded and areal expression patterns. Experimental embryological and developmental genetics has made considerable progress in the elucidation of the molecular events regulating the generation of distinct neuronal cell types at precise locations and in appropriate numbers in the neural tube, which is separated into transverse and lon- gitudinal domains at early embryonic stages. The forebrain neuroepithelium is sub- divided into six so-called prosomeric units, which prevent progenitor cell mixing across their boundaries (Puelles and Rubenstein, 1993). The six transverse domains of the prosomeres can be further subdivided longitudinally by the restricted pattern of a wide spectrum of regulatory proteins. In the embryonic telencephalon, which constitutes cortical areas dorsally and the ventral basal ganglia, homeobox genes Emx and Lhx, and paired homeobox Pax genes, are expressed exclusively in dorsal progenitor cells, whereas ventral progenitors express homeobox genes of the Nkx, Otx, Gbx, and Dlx families (Rubenstein et al., 1998; Cecchi et al., 2000; Simeone, 2000). Knockout studies have demonstrated that thalamocortical connections also require several transcription factors (e.g., Tbr1, Gbx1, Pax6) as well as secreted molecules and cell surface axon guidance proteins (Rubenstein et al., 1999).
Morphological analyses revealed a crucial role of 5-HT in the formation and plasticity of neocortical and subcortical structures, suggesting that it acts as a differ- entiation signal in brain development. Moreover, the timing of serotonergic inner- vation coincides with pronounced growth and synaptogenesis in the cortex, and perinatal manipulations of 5-HT affect cortical 5-HT receptors (Fig. 6.3). The period for 5-HT action corresponds to the period when incoming axons begin to establish synaptic interactions with target neurons and to elaborate a profuse branching pattern (Cases et al., 1996). Investigations of 5-HT participation in neo- cortical development and plasticity have concentrated on the rodent somatosen- sory cortex (SSC), because of its one-to-one correspondence between each whisker and its cortical barrel-like projection area (Killackey et al., 1995) (Fig. 6.4a). The
109 Serotonergic gene pathways

(b)
(a)
Fig. 6.4.
Serotonin in the development and plasticity of the somatosensory cortex (SSC) in rodents. (a) One-to-one correspondence between each whisker and its cortical barrel-like projection area. (b) SSC in wild-type mice and in knockout mice for the serotonin transporter. (c) The deleterious effects of excess serotonin are mediated by the serotonin transporter and by the serotonin receptor subtype 1B (see text for details). GG, ganglion gasseri; V SENS, sensory nerve V; VB, ventrobasal thalamus; BCX, barrelfield cortex; 5HT, serotonin; 5HT1B/2A serotonin receptor subtypes; Glu, glutamate; mGluR5, metabotropic glutamate receptor 5; XYZ, unknown receptors.
processes underlying patterning of projections in the SSC have been intensively studied, with a widely held view that the formation of somatotropic maps does not depend on neural activity (Katz and Shatz, 1996). While pharmacologically induced 5-HT depletion at birth yields smaller barrels but does not prevent the for- mation of the barrel pattern itself (Bennett-Clarke et al., 1994; Osterheld-Haas et al., 1994) excess of extracellular 5-HT (present for example in mice with an inacti- vation of the gene for MAO-A) results in the complete absence of cortical barrel patterns (Cases et al., 1996). Additional evidence for a role of 5-HT in the develop- ment of neonatal rodent SSC derives from the transient barrel-like distribution of
(c)
110 K. P. Lesch, J. Benninghoff, and A. Schmitt
5-HT, 5-HT1B and 5-HT2A receptors, and of the 5-HTT (Fuchs, 1995; Lebrand et al., 1996; Mansour-Robaey et al., 1998). The transient barrel-like 5-HT pattern vis- ualized in layer IV of the SSC of neonatal rodents apparently stems from 5-HT uptake and vesicular storage in thalamocortical neurons, which express both the 5- HTT and the vesicular monoamine transporter (VMAT2) at this developmental stage (Lebrand et al., 1996).
Inactivation of the gene for 5-HTT profoundly disturbs formation of the SSC, with altered cytoarchitecture of cortical layer IV, the layer that contains synapses between thalamocortical terminals and their postsynaptic target neurons (Fig. 6.2). Brains of 5-HTT knockout mice display no or only very few barrels (Bengel et al., 1998; Persico et al., 1999). Cell bodies as well as terminals, typically more dense in barrel septa, appear homogeneously distributed in layer IV of adult 5- HTT knockout brains. Injections of a 5-HT synthesis inhibitor within a narrow time window of 2 days postnatally completely rescued formation of SSC barrel fields. Of note, heterozygous knockout mice develop all SSC barrel fields but fre- quently present irregularly shaped barrels and less-defined cell gradients between septa and barrel hollows. These findings demonstrate that excessive concentrations of extracellular 5-HT are deleterious to SSC development and suggest that tran- sient 5-HTT expression in thalamocortical neurons is responsible for barrel pat- terns in neonatal rodents; its permissive action is required for normal barrel pattern formation, presumably by maintaining extracellular 5-HT concentrations below a critical threshold. Because there is normal synaptic density in SSC layer IV of 5-HTT knockout mice, it is more likely that 5-HT affects SSC cytoarchitecture by promoting dendritic growth toward the barrel hollows and by modulating cytokinetic movements of cortical granule cells, similar to concentration-depen- dent 5-HT modulation of cell migration described in other tissues (Moiseiwitch and Lauder, 1995; Choi et al., 1997; Tamura et al., 1997). Since the reduction in 5- HTT availability in heterozygous knockout mice, which leads to a modest delay in 5-HT uptake but distinctive irregularities in barrel and septum shape, is similar to that reported in humans carrying the low-activity allele of 5-HTTLPR, it may be speculated that allelic variation in 5-HTT function also affects the human brain during development, with due consequences for disease liability and therapeutic response.
Two key players of serotonergic neurotransmission appear to mediate the dele- terious effects of excess 5-HT: the 5-HTT and the 5-HT1B receptor (Fig. 6.4c). Both molecules are expressed in primary sensory thalamic nuclei during the period when the segregation of thalamocortical projections occurs (Bennett-Clarke et al., 1996; Lebrand et al., 1996; Hansson et al., 1998). 5-HT is internalized via 5-HTT in thalamic neurons and is stored in axon terminals (Lebrand et al., 1996; Cases et al., 1998). The presence of VMAT2 within the same neurons allows internalized 5-
111 Serotonergic gene pathways
HT to be stored in vesicles and used as a cotransmitter of glutamate. Lack of 5-HT degradation in MAO-A knockout mice as well as severe impairment of 5-HT clear- ance in mice with an inactivation of the 5-HTT results in an accumulation of 5-HT and overstimulation of 5-HT receptors all along thalamic neurons (Cases et al., 1998). Since 5-HT1B receptors are known to inhibit the release of glutamate in the thalamocortical somatosensory pathway (Rhoades et al., 1994), excessive activation of 5-HT1B receptors could prevent activity-dependent processes involved in the patterning of afferents and barrel structures (Fig. 6.4c). This hypothesis is sup- ported by a recent study using a strategy of combined knockout of MAO-A, 5-HTT, and 5-HT1B receptor genes (Salichon et al., 2001). While only partial disruption of the patterning of somatosensory thalamocortical projections was observed in 5- HTT knockout, MAO-A/5-HTT double knockout mice showed that 5-HT accu- mulation in the extracellular space causes total disruption of the patterning of these projections. Moreover, the removal of 5-HT1B receptors in MAO-A and 5-HTT knockout as well as in MAO-A/5-HTT double knockout mice allowed a normal segregation of the somatosensory projections. These findings point to an essential role of the 5-HT1B receptor in mediating the deleterious effects of excess 5-HT in the somatosensory system. If 5-HT and serotonergic gene expression is involved in a myriad of processes during brain development as well as in synaptic plasticity in adulthood, thus setting the stage for brain (dys)function, complex behavior and psychopharmacological drug response are likely to be influenced by genomic variations within the serotonergic gene pathway.
Regulatory genes of neurodevelopment
Developmental neurogenetics has generally taken a reductionist approach to eluci- date how the brain is built and functions, focusing on neurite outgrowth and axonal pathfinding as well as synaptogenesis and synapse function. Although these studies have not been carried out in ways that would likely identify genes that specify neural networks, investigations aimed at dissecting various sensory modal- ities, such as the somatosensory systems, have unquestionably advanced the under- standing of the basic workings of neurocircuits. A complementary approach to studies of expression and function of proteins that regulate the function of brain neurotransmitter systems encompasses investigation of genes and their protein products implicated in the specification of monoaminergic neurocircuits. Regulatory genes, such as transcription factors, neurotrophins, and other growth or axon guidance factors, functioning in hierarchies across time and space are required for particular aspects of development to occur (Ragsdale and Grove, 2001; Rhinn and Brand, 2001). Analogous to, and frequently in concert with, other morphogenetic factors, neurotrophins act via specific receptors at the cell

112 K. P. Lesch, J. Benninghoff, and A. Schmitt
membrane, and their intracellular signal tranduction pathways converge at the level of gene expression by activation of transcription factors.
The development of distinct neurons is defined by the unique profiles of genes that these neurons express. Transcriptional control regions and their associated cis and trans regulatory elements are targets for modulation of gene expression (Fig. 6.1). It is widely accepted that neuronal genes are initially regulated at the point of transcription initiation, although other mechanisms of gene regulation including alternative splicing, mRNA editing, and mRNA elongation may also be crucial. A large variety of transcription factors has been identified and characterized that acti- vate or repress the transcription of specific genes in the brain, resulting in changes of the neuronal phenotype.
Transcription factors bind to specific motifs in the regulatory sequence of a target gene, resulting in spatio-temporal alterations in promoter-driven gene transcrip- tion. Furthermore, they have the potential to be highly specific in their target recog- nition, and transcription factors from different families may interact with each other when bound to DNA at composite response elements. This has two striking conse- quences: ubiquitous factors can affect cell specificity, and closely related factors from a given family can produce very different regulatory patterns. Therefore, transcrip- tion factors play a key role in the control of development and differentiation.
Different stimuli lead to the activation or repression of different transcription factors, and circumstantial evidence indicates that transcription factors may be tar- geted by psychoactive drugs such as 5-HT2A and 5-HT2C receptor agonists and antagonists or lithium (for review see Lesch and Heils (2000)). The mechanisms by which they exert therapeutic effects still require to be studied in detail, but indu- cible transcription factors such as Fos and Zif268 are believed to mediate between receptor-activated second messenger systems and the transcriptional apparatus of genes involved in the complex functions of neuronal cells (Herdegen and Leah, 1998). Taken together, a large number of genes coding for transcriptional regula- tors, which are either indirect drug targets or pharmacodynamically important factors, play a key role in the control of neuronal development, differentiation, and phenotype maintenance.
Developmental specification and differentiation of serotonergic neurons
Despite the widespread importance of the central serotonergic neurotransmitter system, little is known about the molecular mechanisms controlling the develop- ment of 5-HT neurons. Several regulatory genes (including those for transcription factors, other morphogenetically relevant regulators of gene expression, neuro- trophins, and growth factors as well as 5-HT itself) contribute to the specification, differentiation, and phenotype maintenance of the raphe serotonergic system.

113 Serotonergic gene pathways

Fig. 6.5.
Genetic pathways in the development of the raphe serotonin (5HT) system. Wnt1, Pax2, Pax5, Otx2, Glx2, En1, En2, Shh, Fgf8, Glx1, Pet1, Foggy, Eagle, and GATA3 are transcription factors. BMP (bone morphogenetic protein), TGF (transforming growth factor ), CNTF (ciliary neurotrophic factor), and BDNF (brain-derived neurotrophic factor) are neurotrophins and other growth factors. TrkB, neurotrophin receptor; MAPK, MAP kinase; AD, adenylyl cyclase; PKA, protein kinase A; TPH, tryptophan hydroxylase; R, receptor. (Modified for 5HT from Burbach, 2000.)
Induction of the floor plate at the ventral midline of the neural tube is one of the earliest events in the establishment of dorsoventral polarity in the vertebrate brain. Fibroblast growth factors (FGF8) and “Sonic hedgehog” (Shh) signals control serotonergic and dopaminergic cell fate in the anterior neural plate (Ye et al., 1998; Hynes et al., 2000). While the transcription factor Gli2 is also required for induc- tion of floor plate and adjacent cells, including serotonergic (and dopaminergic) neurons throughout the midbrain, hindbrain and spinal cord (van Doorninck et al., 1999), expression of Pet1, a ETS (electrical transcranial stimulation) domain transcription factor, is restricted to the hindbrain and closely associated with devel- oping serotonergic neurons in the raphe nuclei (Hendricks et al., 1999) (Fig. 6.5). Moreover, consensus Pet1-binding motifs are present in the 5 -regulatory regions of genes for both the human and murine 5-HT1A receptor, 5-HTT, tryptophan hydroxylase, and aromatic L-amino acid decarboxylase genes; this expression
114 K. P. Lesch, J. Benninghoff, and A. Schmitt
profile is characteristic of the serotonergic neuron phenotype, i.e., 5-HT synthesis, release, uptake, and metabolism. These findings identify Pet1 as a critical regulator of serotonergic system specification. Even within the relatively circumscribed serotonergic raphe complex, gene expression in discrete subsystem appears to be differentially controlled by transcriptional regulators. The transcription factor GATA3 plays a critical role in the development of the serotonergic neurons of the caudal raphe nuclei and thus in locomotor performance (Matise et al., 1998). Beyond the point of transcription initiation, the role of mRNA elongation and other mechanisms of neural gene regulation are increasingly attracting systematic scrutiny. Foggy, a phosphorylation-dependent, dual regulator of transcript elonga- tion, affects development of 5-HT (and dopamine)-containing neurons in the zebrafish (Guo et al., 2000). In the fruit fly, Eagle, a zinc finger transcription factor with homology to the steroid receptor family, is required for the specification of 5-HT neurons (Lundell and Hirsh, 1998).
Finally, several neurotrophins and growth factors also modulate the phenotype of serotonergic neurons (Fig. 6.2). These include family members of the neuro- trophins, such as FGF, transforming growth factor-, bone morphogenetic protein, and neurokines (e.g., ciliary neurotrophic factor) (Galter and Unsicker, 2000a, b). 5-HT itself regulates the serotonergic phenotype of neurons by sequential activa- tion of the 5-HT1A receptor, brain-derived neurotrophic factor and its receptor TrkB, as well as a wide spectrum of signal transduction pathways. In particular, transcriptional regulation appears to be dependent on stimulation of the adenylyl cyclase/protein kinase A signaling pathway mediated by a family of cAMP- responsive nuclear factors (including cAMP response element B, cAMP response element modulator, and activating transcription factor 1) (Herdegen and Leah, 1998). These factors contain the basic domain/leucine zipper motifs and bind as dimers to cAMP-responsive elements (CREs). Galter and Unsicker (2000a, b) have, therefore, proposed the neurotrophin receptor TrkB as the master control protein that integrates a diverse array of signals which elicit and maintain serotonergic differentiation. Interestingly, heterozygous mice deficient in brain-derived neuro- trophic factor develop intermale aggressiveness and hyperphagia in conjunction with decreased forebrain 5-HT concentrations and fiber densities (Lyons et al., 1999). Based on the increasing body of evidence that genetically driven variability of transcriptional regulators, neurotrophins, and growth factors is associated with complex behavioral traits (Okladnova et al., 1998; Stoeber et al., 1998; Krebs et al., 2000; Kunugi et al., 2001), investigations are increasingly examining the molecular basis of gene regulation in psychopharmacological drug response. Although the remarkable impact of serotonergic neurons on brain development and behavioral functions is clear, the mechanisms underlying their developmental genetics are only beginning to emerge. Unraveling the interactions of these determinants of
115 Serotonergic gene pathways
development remains a daunting task but it may eventually lead to the discovery of novel drug targets or factors relevant to the genetics of psychotherapeutic drug response.
Are epistatic interactions relevant to brain development?
Although still in its infancy, investigation of gene–gene and gene–environment interactions support the view that both genetic and environmental factors influ- ence brain development, neuroplasticity, and drug response. Despite evidence for a substantial contribution of genetic and environmental factors to the formation of synaptic connections in the brain during development, adult life, and old age, detailed knowledge of the molecular mechanisms is only beginning to accumulate. The resolution of epistatic interactions that are operative in this fine-tuning process is indeed among the last frontiers of genetic research.
Ebstein and coworkers (1998) investigated the behavioral effects of VNTR poly- morphism in exon 3 of the dopamine D4 receptor (DRD4), which had previously been linked to the personality trait of novelty seeking, and the 5-HTTLPR, which seems to influence neuroticism and harm avoidance, in two week-old neonates. Neonate temperament and behavior were assessed using a neonatal assessment scale. In addition to a significant association of the DRD4 polymorphism across four behavioral clusters relevant to temperament, including orientation, motor organization, range of state and regulation of state, an interaction was also observed between the DRD4 polymorphism and 5-HTTLPR. The presence of the 5-HTTLPR s/s genotype decreased the orientation score for the group of neonates carrying the long (l) allelic variant of DRD4. The DRD4 polymorphism–5-HTT-LPR interac- tion was also assessed in a sample of adult subjects. Interestingly, there was no sig- nificant effect of l DRD4 genotype in those homozygotes with 5-HTTLPR when they were grouped by the 5-HTTLPR, whereas in the group without the homozy- gous genotype the effect of l DRD4 was significant and represented 13% of the var- iance in novelty-seeking scores between groups. Temperament and behavior of these infants were psychometrically re-examined at 2 months (Auerbach et al., 1999). There were significant negative correlations between neonatal orientation and motor organization at 2 weeks and negative emotionality, especially distress in daily situations, at 2 months of age. Furthermore, grouping of the infants by DRD4 polymorphism and 5-HTTLPR revealed significant main effects for negative emo- tionality and distress. Infants with l DRD4 alleles had lower scores on negative emo- tionality and distress than infants with DRD4 s alleles. In contrast, infants homozygous for the 5-HTTLPR s allele had higher scores on negative emotionality and distress than infants with the l/s or l/l genotypes. Infants with the s/s genotype who also were lacking the novelty seeking-associated DRD4 l alleles showed most

116 K. P. Lesch, J. Benninghoff, and A. Schmitt
negative emotionality and distress, traits that possibly contribute to the predispo- sition for emotional instability. These findings represent a benchmark example for future research on gene–gene interaction in behavioral development.
Following up on the landmark studies concerning early environmental regula- tion of spatial learning and memory, hippocampal synaptogenesis, and stress reac- tivity by Meaney and associates (Meaney et al., 1996, 2000; Francis et al., 1999; Liu et al., 2000), evidence for a role of gene–environment interactions in brain devel- opment also comes from studies of rhesus macaques, a nonhuman primate species that, like humans, carries the 5-HTT gene-associated polymorphism (rh5- HTTLPR). Previous work in rhesus monkeys has shown that early adverse experi- ences have long-term consequences for the functioning of the central 5-HT system, as indicated by robustly altered CSF 5HIAA levels, in monkeys deprived of their parents at birth and raised only with peers (Higley et al., 1991, 1992). Association between central 5-HT turnover and rh5-HTTLPR genotype was recently tested in rhesus monkeys with well-characterized environmental histories (Bennett et al., 2001). The monkeys’ rearing fell into one of the following four categories: mother- reared, either with the biological mother or cross-fostered, or peer-reared, either with a peer group of three or four monkeys or with an inanimate surrogate and daily contact with a playgroup of peers. Peer-reared monkeys were separated from their mothers, placed in the nursery at birth, and given access to peers at 30 days of age either continuously or during daily play sessions. Mother-reared and cross- fostered monkeys remained with the mother, typically within a social group. At roughly 7 months of age, mother-reared monkeys were weaned and placed together with their peer-reared cohort in large, mixed-gender social groups.
Since the monkey population comprised two groups that received dramatically different social and rearing experience early in life, the interactive effects of envir- onmental experience and the rh5-HTTLPR on cisternal CSF 5HIAA levels and 5-HT-related behavior was assessed. CSF 5HIAA concentrations were significantly influenced by genotype for peer-reared but not for mother-reared subjects. Peer- reared rhesus monkeys with the low-activity rh5-HTTLPR s allele had significantly lower concentrations of CSF 5HIAA than their homozygous l/l counterparts. Low 5-HT turnover in monkeys with the s allele is congruent with in vitro studies that show reduced binding and transcriptional efficiency of the 5-HTT gene associated with the 5-HTTLPR s allele (Lesch et al., 1996). This suggests that the rh5-HTTLPR genotype is predictive of CSF 5HIAA concentrations, but that early experiences make unique contributions to variation in later 5-HT functioning. This finding provides evidence of an environment-dependent association between a polymor- phism in the 5 -regulatory region of the 5-HTT gene and a direct measure of 5-HT functioning, cisternal CSF 5HIAA concentration, thus revealing an interaction between rearing environment and rh5-HTTLPR genotype.
117 Serotonergic gene pathways
Similar approaches have also been applied to the neonatal period to facilitate investigation of the contribution of genotype and rearing environment to the development of behavioral traits. Rhesus macaque infants heterozygous for the s and l variants of the rh5-HTTLPR displayed higher behavioral stress reactivity compared with infants homozygous for the long variant of the allele (l/l) (Champoux et al., 1999). Mother-reared and peer-reared monkeys were assessed on a standardized primate neurobehavioral test designed to measure orienting, motor maturity, reflex functioning, and temperament. Main effects of genotype and, in some cases, interactions between rearing condition and genotype were demonstrated for items indicative of orienting, attention, and temperament. In general, heterozygotes demonstrated diminished orientation, lower attentional capabilities, and increased affective responding relative to l/l homozygotes. However, the genotype effects were more pronounced for monkeys raised in the neonatal nursery than for those reared by their mothers.
Taken together, these findings provide evidence of an association between allelic variation of 5-HTT expression and central 5-HT function. In addition, they dem- onstrate the contributions of rearing environment and genetic background, and their interaction, in a nonhuman primate model of neural and behavioral develop- ment. The developmental and behavioral results of deleterious early experiences of social separation are consistent with the notion that the 5-HTTLPR may influence the risk for affective spectrum disorders and the response to treatment. Nonhuman primate studies are, therefore, useful to help to identify environmental factors that either compound the vulnerability conferred by a particular genetic makeup or, conversely, act to influence the therapeutic outcome associated with that genotype.
Conclusions
Pharmacogenomics, in general, will not improve the efficacy of a given drug, but pharmacogenetic profiling may assist the selection of patients who are likely to respond favorably. Thus, pharmacogenomics provides a view of drug behavior and sensitivity useful for improving the efficacy of drug development and utilization. Progress in developmental psychopharmacogenetics is currently accelerated by closer integration of behavioral, developmental, and genetic approaches. Integration of emerging tools and technologies for genetic analysis will provide the groundwork for an advanced stage of gene identification and functional studies in pharmacogenetics. The documented heterogeneity of both genetic and environ- mental constituents of brain development and function suggests the futility of searching for unitary determinants of psychopharmalogic drug response. This vista should, therefore, encourage the pursuit of quantitative approaches to pharmaco- genetics.

118 K. P. Lesch, J. Benninghoff, and A. Schmitt
Several refined concepts should be adopted regarding future behavioral pharma- cogenetic research. First, pharmacogenetic studies require to employ randomized, double-blind clinical trial methodology, and, in order to detect a small gene effects, a dimensional, quantitative approach to behavioral phenotypes and treatment effects arising from standardized psychometric trait and response assessment is needed. Given the limitation of the diagnostic and psychometric approach, future studies will require extended, homogeneous, and ethnically matched samples. In order to control for nonindependence within case-controlled samples, and thus to minimize the risk of population stratification bias, rigorous methods of “genomic control” have been designed. These statistical strategies are based on the assessment of 60 SNPs or genotypes of 100 unlinked microsatellite markers spread throughout the genome to adjust the significance level of a candidate gene polymorphism (Bacanu et al., 2000; Pritchard et al., 2000). With recent advances in molecular genetics, the rate-limiting step in identifying candidate genes has become defini- tion of phenotype and therapeutic outcome.
Second, more functionally relevant polymorphisms in genes within a single neu- rotransmitter system, or in genes that constitute a developmental and functional unit in their concerted actions, need to be identified and assessed in large associa- tion studies to avoid stratification artifacts and to elucidate complex epigenetic interactions of multiple loci. Although great strides have been made in understand- ing the diversity of the human genome, such as the frequency, distribution, and type of genetic variation that exists, the feasibility of applying this information to uncover useful pharmacogenomic markers remains uncertain. Based on the first draft sequence of the human genome, more than 1.4 million SNPs in the human genome have been identified (Consortium, 2001; Sachidanandam et al., 2001). The health care industry is heavily relying on the commercialized access to SNP data- bases for use in research in the hope of revolutionizing the drug development process. However, the reality of using SNPs to uncover drug response markers is rarely addressed; this requires considerations such as patient sample size, SNP density and genome coverage, SNP functionality, and data interpretation, which will be important for determining the suitability of pharmacogenomic informa- tion. Success will depend on the availability of SNPs in the coding or regulatory regions of a large number of candidate genes as well as knowledge of the average extent of linkage disequilibrium between SNPs, the development of high- throughput technologies for genotyping SNPs, identification of protein-altering SNPs by DNA and protein microarray-assisted expression analysis, and collection of DNA from well-assessed patients. As more and more appreciation of the poten- tial for polymorphisms in gene regulatory regions to impact gene expression is gained, knowledge of novel functional variants is likely to emerge.
Third, genetic influences are not the only pathway that lead to individual
119 Serotonergic gene pathways
differences in personality dimensions, behavior, psychopathology, and drug response. Complex traits are most likely to be generated by a complex interaction of environmental and experiential factors with a number of genes and their prod- ucts. Even pivotal regulatory proteins of cellular pathways and neurocircuits will have only a very modest impact, while noise from nongenetic mechanisms may obstruct identification of relevant gene variants. Although current methods for the detection of gene–gene and gene–environment interaction in behavioral genetics are largely indirect, the most relevant consequence of gene identification for behav- ioral traits and psychopharmacological drug response may be that it will provide the tools required to clarify systematically the effects of gene–environment interac- tion on brain development and plasticity.
Finally, future benefits will stem from the development of techniques involving molecular cell biology, transgenics, and gene transfer technology, which could facilitate novel drug design. In a postgenomics world, behavioral pharmacogenet- ics research will require integration of research on genomics, DNA variants, gene expression, proteomics, brain development, structure and function, and behavior in a wide spectrum of species. Although bioinformatic resources are evolving in most of these areas, integration of these resources from the perspective of psycho- pharmacogenomics will greatly facilitate research.
REFERENCES
Arranz M, Nunro J, Birkett J et al. (2000). Pharmacogenetic prediction of clozapine response. Lancet 355, 1615–1616.
Auerbach J, Geller V, Lezer S et al. (1999). Dopamine D4 receptor (DRD4) and serotonin trans- porter promoter (5-HTTLPR) polymorphisms in the determination of temperament in 2- month-old infants. Mol Psychiatry 4, 369–373.
Bacanu SA, Devlin B, Roeder K (2000). The power of genomic control. Am J Hum Genet 66, 1933–1944.
Barnes NM, Sharp T (1999). A review of central 5-HT receptors and their function. Neuropharmacology 38, 1083–1152.
Battersby S, Ogilvie AD, Blackwood DH et al. (1999). Presence of multiple functional polyaden- ylation signals and a single nucleotide polymorphism in the 3 untranslated region of the human serotonin transporter gene. J Neurochem 72, 1384–1388.
Benedetti F, Serretti A, Colombo C et al. (1998). Influence of a functional polymorphism within the promoter of the serotonin transporter gene on the effects of total sleep deprivation in bipolar depression. Am J Psychiatry 156, 1450–1452.
Bengel D, Murphy DL, Andrews AM et al. (1998). Altered brain serotonin homeostasis and loco- motor insensitivity to 3, 4-methylenedioxymethamphetamine (“Ecstasy”) in serotonin transporter-deficient mice. Mol Pharmacol 53, 649–655.

120 K. P. Lesch, J. Benninghoff, and A. Schmitt
Bennett AJ, Lesch KP, Heils A et al. (2002). Early experience and serotonin transporter gene vari- ation interact to influence primate CNS function. Mol Psychiatry, 7, 118–122.
Bennett-Clarke CA, Leslie MJ, Lane RD, Rhoades RW (1994). Effect of serotonin depletion on vibrissa-related patterns of thalamic afferents in the rat’s somatosensory cortex. J Neurosci 14, 7594–7607.
Bennett-Clarke CA, Chiaia NL, Rhoades RW (1996). Thalamocortical afferents in rat transiently express high-affinity serotonin uptake sites. Brain Res 733, 301–306.
Berger-Sweeney J, Hohmann CF (1997). Behavioral consequences of abnormal cortical develop- ment: insights into developmental disabilities. Behav Brain Res 86, 121–142.
Blier P, de Montigny C (1997). Current psychiatric uses of drugs acting on the serotonin system. In Baumgarten HG, Göthert M, eds. Serotonergic Neurons and 5-HT Receptors in the CNS. Berlin: Springer, pp. 727–751.
Bonasera SJ, Tecott LH (2000). Mouse models of serotonin receptor function: toward a genetic dissection of serotonin systems. Pharmacol Ther 88, 133–142.
Bradley CC, Blakely RD (1997). Alternative splicing of the human serotonin transporter gene. J Neurochem 69, 1356–1367.
Brunner D, Buhot MC, Hen R, Hofer M (1999). Anxiety, motor activation, and maternal–infant interactions in 5-HT1B knockout mice. Behav Neurosci 113, 587–601.
Burbach JP (2000). Genetic pathways in the developmental specification of hypothalamic neuro- peptide and midbrain catecholamine systems. Eur J Pharmacol 405, 55–62.
Canton H, Emeson RB, Barker EL et al. (1996). Identification, molecular cloning, and distribu- tion of a short variant of the 5-hydroxytryptamine 2C receptor produced by alternative splic- ing. Mol Pharmacol 50, 799–807.
Cases O, Vitalis T, Seif I, De Maeyer E, Sotelo C, Gaspar P (1996). Lack of barrels in the somato- sensory cortex of monoamine oxidase A-deficient mice: role of a serotonin excess during the critical period. Neuron 16, 297–307.
Cases O, Lebrand C, Giros B et al. (1998). Plasma membrane transporters of serotonin, dopa- mine, and norepinephrine mediate serotonin accumulation in atypical locations in the devel- oping brain of monoamine oxidase A knock-outs. J Neurosci 18, 6914–6927.
Catalano M (1999). The challenges of psychopharmacogenetics. Am J Hum Genet 65, 606–610.
Cecchi C, Mallamaci A, Boncinelli E (2000). otx and emx homeobox genes in brain development. Int J Dev Biol 44, 663–778.
Champoux M, Bennett A, Lesch KP et al. (1999). Serotonin transporter gene polymorphism and neurobehavioral development in rhesus monkey neonates. Soc Neurosci Abstr 25, 69.
Choi D, Ward S, Messaddeq N, Launay J, Maroteaux L (1997). 5-HT2B receptor-mediated sero- tonin morphogenetic functions in mouse cranial neural crest and myocardial cells. Development 124, 1745–1755.
Cichon S, Nothen MM, Rietschel M, Propping P (2000). Pharmacogenetics of schizophrenia. Am J Med Genet 97, 98–106.
Consortium IHGS (2001). Initial sequencing and analysis of the human genome. Nature 409, 860–921.
121 Serotonergic gene pathways
Crawley JN (1999). Behavioral phenotyping of transgenic and knockout mice: experimental design and evaluation of general health, sensory functions, motor abilities, and specific behav- ioral tests. Brain Res 835, 18–26.
Crawley JN, Paylor R (1997). A proposed test battery and constellations of specific behavioral paradigms to investigate the behavioral phenotypes of transgenic and knockout mice. Horm Behav 31, 197–211.
Deckert J Catalano M, Syagailo YV et al. (1999). Excess of high activity monoamine oxidase A gene promoter alleles in female patients with panic disorder. Hum Mol Genet 8, 621–624.
Deckert J, Meyer J, Catalano M et al. (2000). Novel 5 -regulatory region polymorphisms of the 5-HT2C receptor gene: association study with panic disorder. Int J Neuropsychopharmacol 3, 321–325.
Del Zompo M, Ardau R, Palmas M, Bocchetta A, Reina A, Piccardi M (1999). Lithium response: association study with two candidate genes. Mol Psychiatry 4(Suppl), 66–67.
Denney RM, Sharma A, Dave SK, Waguespack A (1994). A new look at the promoter of the human monoamine oxidase A gene: mapping transcription initiation sites and capacity to drive luciferase expression. J Neurochem 63, 843–856.
Denney RM, Koch H, Craig IW (1999). Association between monoamine oxidase A activity in human male skin fibroblasts and genotype of the MAO-A promoter-associated variable number tandem repeat. Hum Genet 105, 542–551.
Ebstein RP, Levine J, Geller V, Auerbach J, Gritsenko I, Belmaker RH (1998). Dopamine D4 recep- tor and serotonin transporter promoter in the determination of neonatal temperament. Mol Psychiatry 3, 238–46.
Francis D, Diorio J, Liu D, Meaney MJ (1999). Nongenomic transmission across generations of maternal behavior and stress responses in the rat. Science 286, 1155–1158.
Fuchs JL (1995). Neurotransmitter receptors in developing barrel cortex. In Jones EG, Diamond IT, ed. Cerebral Cortex. New York: Plenum Press, pp. 375–409.
Galter D, Unsicker K (2000a). Brain-derived neurotrophic factor and TrkB are essential for cAMP-mediated induction of the serotonergic neuronal phenotype. J Neurosci Res 61, 295–301.
Galter D, Unsicker K (2000b). Sequential activation of the 5-HT1A serotonin receptor and TrkB induces the serotonergic neuronal phenotype. Mol Cell Neurosci 15, 446–455.
Gelernter J, Kranzler H, Cubells JF (1997). Serotonin transporter protein (SLC6A4) allele and haplotype frequencies and linkage disequilibria in African- and European-American and Japanese populations and in alcohol-dependent subjects. Hum Genet 101, 243–246.
Gerald C (1995). The 5-HT4 receptor: molecular cloning and pharmacological characterization of two splice variants. EMBO J 14, 2806–2815.
Gould E (1999). Serotonin hippocampal neurogenesis. Neuropsychopharmacology 21, 46S-51S.
Greenberg BD, Tolliver TJ, Huang SJ, Li Q, Bengel D, Murphy DL (1999). Genetic variation in the serotonin transporter promoter region affects serotonin uptake in human blood platelets. Am J Med Genet 88, 83–87.
Guo S, Yamaguchi Y, Schilbach S et al. (2000). A regulator of transcriptional elongation controls vertebrate neuronal development. Nature 408, 366–369.
122 K. P. Lesch, J. Benninghoff, and A. Schmitt
Hanna GL, Himle JA, Curtis GC et al. (1998). Serotonin transporter and seasonal variation in blood serotonin in families with obsessive-compulsive disorder. Neuropsychopharmacology 18, 102–111.
Hansson SR, Mezey E, Hoffman BJ (1998). Serotonin transporter messenger RNA in the devel- oping rat brain: early expression in serotonergic neurons and transient expression in non- serotonergic neurons. Neuroscience 83, 1185–1201.
Heidmann DE (1997). Four 5-hydroxytryptamine 7 (5-HT7) receptor isoforms in human and rat produced by alternative splicing: species differences due to altered intron–exon organization. J Neurochem 68, 1372–1381.
Heinz A, Jones DW, Mazzanti C et al. (1999). A relationship between serotonin transporter genotype and in vivo protein expression and alcohol neurotoxicity. Biol Psychiatry 47, 643–649.
Hendricks T, Francis N, Fyodorov D, Deneris ES (1999). The ETS domain factor Pet-1 is an early and precise marker of central serotonin neurons and interacts with a conserved element in serotonergic genes. J Neurosci 19, 10348–10356.
Herdegen T, Leah JD (1998). Inducible and constitutive transcription factors in the mammalian nervous system: control of gene expression by Jun, Fos and Krox, and CREB/ATF proteins. Brain Res Rev 28, 370–490.
Higley JD, Suomi SJ, Linnoila M (1991). CSF monoamine metabolite concentrations vary according to age, rearing, and sex, and are influenced by the stressor of social separation in rhesus monkeys. Psychopharmacology 103, 551–556.
Higley JD, Suomi SJ, Linnoila M (1992). A longitudinal assessment of CSF monoamine metabo- lite and plasma cortisol concentrations in young rhesus monkeys. Biol Psychiatry 32, 127–145. Hoyer D, Martin G (1997). 5-HT receptor classification and nomenclature: towards a harmon-
ization with the human genome. Neuropharmacology 36, 419–428.
Hynes M, Ye W, Wang K et al. (2000). The seven-transmembrane receptor smoothened cell-
autonomously induces multiple ventral cell types. Nat Neurosci 3, 41–46.
Ishiguro H, Arinami T, Yamada K, Otsuka Y, Toru M, Shibuya H (1997). An association study between a transcriptional polymorphism in the serotonin transporter gene and panic disorder
in a Japanese population. Psychiatr Clin Neurosci 51, 333–335.
Jonsson EG, Norton N, Gustavsson JP, Oreland L, Owen MJ, Sedvall GC (2000). A promoter
polymorphism in the monoamine oxidase A gene and its relationships to monoamine metabo-
lite concentrations in CSF of healthy volunteers. J Psychiatr Res 34, 239–244.
Katz LC, Shatz CJ (1996). Synaptic activity and the construction of cortical circuits. Science 274,
1133–1138.
Killackey HP, Rhoades RW, Bennett-Clarke CA (1995). The formation of a cortical somatotopic
map. Trends Neurosci 18, 402–407.
Kojic L, Dyck RH, Gu Q, Douglas RM, Matsubara J, Cynader MS (2000). Columnar distribution
of serotonin-dependent plasticity within kitten striate cortex. Proc Natl Acad Sci USA 97,
1841–1844.
Krebs MO, Guillin O, Bourdell MC et al. (2000). Brain derived neurotrophic factor (BDNF) gene
variants association with age at onset and therapeutic response in schizophrenia. Mol Psychiatry 5, 558–562.
123 Serotonergic gene pathways
Kunugi H, Hattori M, Kato T et al. (1997). Serotonin transporter gene polymorphisms: ethnic difference and possible association with bipolar affective disorder. Mol Psychiatry 2, 457–462. Kunugi H, Ueki A, Otsuka M et al. (2001). A novel polymorphism of the brain-derived neuro- trophic factor (BDNF) gene associated with late-onset Alzheimer’s disease. Mol Psychiatry 6,
83–86.
Lavdas AA, Blue ME, Lincoln J, Parnavelas JG (1997). Serotonin promotes the differentiation of
glutamate neurons in organotypic slice cultures of the developing cerebral cortex. J Neurosci
17, 7872–80.
Lebrand C, Cases O, Ådelbrecht C et al. (1996). Transient uptake and storage of serotonin in
developing thalamic neurons. Neuron 17, 823–835.
Lesch KP (2001a). Molecular foundation of anxiety disorders. J Neural Transm 108, 717–746. Lesch KP (2001b). Serotonin transporter: from genomics and knockouts to behavioral traits and
psychiatric disorders. In Briley M, Sulser F, eds. Molecular Genetics of Mental Disorders.
London: Martin Dunitz, pp. 221–267.
Lesch KP, Heils A (2000). Serotonergic gene transcriptional control regions: targets for antide-
pressant drug development? Int J Neuropsychopharmacol 3, 67–79.
Lesch KP, Merschdorf U (2000). Impulsivity, aggression, and serotonin: a molecular psychobio-
logical perspective. Behav Sci Law 18, 581–604.
Lesch KP, Mössner R (1999). 5-HT1A receptor inactivation: anxiety or depression as a murine
experience. Int J Neuropsychopharmacol 2, 327–331.
Lesch KP, Balling U, Gross J et al. (1994). Organization of the human serotonin transporter gene.
J Neural Transm Gen Sect 95, 157–162.
Lesch KP, Bengel D, Heils A et al. (1996). Association of anxiety-related traits with a polymor-
phism in the serotonin transporter gene regulatory region. Science 274, 1527–1531.
Lesch KP, Meyer J, Glatz K et al. (1997). The 5-HT transporter gene-linked polymorphic region (5-HTTLPR) in evolutionary perspective: alternative biallelic variation in rhesus monkeys. J
Neural Transm 104, 1259–1266.
Lieske V, Bennett-Clarke CA, Rhoades RW (1999). Effects of serotonin on neurite outgrowth
from thalamic neurons in vitro. Neuroscience 90, 967–974.
Little KY, McLaughlin DP, Zhang L et al. (1998). Cocaine, ethanol, and genotype effects on human
midbrain serotonin transporter binding sites and mRNA levels. Am J Psychiatry 155, 207–213. Liu D, Diorio J, Day JC, Francis DD, Meaney MJ (2000). Maternal care, hippocampal synapto-
genesis and cognitive development in rats. Nat Neurosci 3, 799–806.
Lundell MJ, Hirsh J (1998). Eagle is required for the specification of serotonin neurons and other
neuroblast 7–3 progeny in the Drosophila CNS. Development 125, 463–472.
Lyons WE, Mamounas LA, Ricaurte GA et al. (1999). Brain-derived neurotrophic factor- deficient mice develop aggressiveness and hyperphagia in conjunction with brain serotonergic
abnormalities. Proc Natl Acad Sci USA 96, 15239–15244.
Malberg JE, Eisch AJ, Nestler EJ, Duman RS (2000). Chronic antidepressant treatment increases
neurogenesis in adult rat hippocampus. J Neurosci 20, 9104–9110.
Mansour-Robaey S, Mechawar N, Radja F, Beaulieu C, Descarries L (1998). Quantified distribu-
tion of serotonin transporter and receptors during the postnatal development of the rat barrel field cortex. Dev Brain Res 107, 159–163.
124 K. P. Lesch, J. Benninghoff, and A. Schmitt
Matise MP, Epstein DJ, Park HL, Platt KA, Joyner AL (1998). Gli2 is required for induction of floor plate and adjacent cells, but not most ventral neurons in the mouse central nervous system. Development 125, 2759–2770.
Meaney MJ, Diorio J, Francis D et al. (1996). Early environmental regulation of forebrain gluco- corticoid receptor gene expression: implications for adrenocortical responses to stress. Dev Neurosci 18, 49–72.
Meaney MJ, Diorio J, Francis D et al. (2000). Postnatal handling increases the expression of cAMP-inducible transcription factors in the rat hippocampus: the effects of thyroid hormones and serotonin. J Neurosci 20, 3926–3935.
McPherson JD, Marra M, Hillier L et al. (2001). A physical map of the human genome. Nature 409, 934–941.
Moiseiwitch JRD, Lauder JM (1995). Serotonin regulates mouse cranial neural crest migration. Proc Natl Acad Sci USA 92, 7182–7186.
Mortensen OV, Thomassen M, Larsen MB, Whittemore SR, Wiborg O (1999). Functional anal- ysis of a novel human serotonin transporter gene promoter in immortalized raphe cells. Brain Res Mol Brain Res 68, 141–148.
Nakagawa Y, Johnson JE, O’Leary DD (1999). Graded and areal expression patterns of regulatory genes and cadherins in embryonic neocortex independent of thalamocortical input. J Neurosci 19, 10877–10885.
Nobile M, Begni B, Giorda R et al. (1999). Effects of serotonin transporter promoter genotype on platelet serotonin transporter functionality in depressed children and adolescents. J Am Acad Child Adolesc Psychiatry 38, 1396–1402.
Okladnova O, Syagailo YV, Tranitz M et al. (1998). A promoter-associated polymorphic repeat modulates Pax-6 expression in human brain. Biochem Biophys Res Commun 248, 402–405.
Olsen MA, Nawoschik SP, Schurman BR et al. (1999). Identification of a human 5-HT6 receptor variant produced by alternative splicing. Brain Res Mol Brain Res 64, 255–263.
Osterheld-Haas MC, van der Loos H, Hornung JP (1994). Monoaminergic afferents to cortex modulate structural plasticity in the barrelfield of the mouse. Dev Brain Res 77, 189–202.
Persico AM, Revay RS, Mössner R et al. (1999). Barrel pattern formation in somatosensory cor- tical layer IV requires serotonin uptake by thalamocortical endings, while vesicular mono- amine release is necessary for development of supragranular layers. J Neurosci 21, 6862–6873.
Pritchard JK, Stephens M, Rosenberg NA, Donnelly P (2000). Association mapping in structured populations. Am J Hum Genet 67, 170–181.
Puelles L, Rubenstein JL (1993). Expression patterns of homeobox and other putative regulatory genes in the embryonic mouse forebrain suggest a neuromeric organization. Trends Neurosci 16, 472–479.
Ragsdale CW, Grove EA (2001). Patterning the mammalian cerebral cortex. Curr Opin Neurobiol 11, 50–58.
Rhinn M, Brand M (2001). The midbrain–hindbrain boundary organizer. Curr Opin Neurobiol 11, 34–42.
Rhoades RW, Bennett-Clarke CA, Shi MY, Mooney RD (1994). Effects of 5-HT on thalamocor- tical synaptic transmission in the developing rat. J Neurophysiol 72, 2438–2450.
125 Serotonergic gene pathways
Routledge C, Middlemiss DN (1996). The 5HT hypothesis of depression revisited. Mol Psychiatry 1, 437.
Rubenstein JL, Shimamura K, Martinez S, Puelles L (1998). Regionalization of the prosence- phalic neural plate. Annu Rev Neurosci 21, 445–477.
Rubenstein JL, Anderson S, Shi L, Miyashita-Lin E, Bulfone A, Hevner R (1999). Genetic control of cortical regionalization and connectivity. Cereb Cortex 9, 524–532.
Sabol SZ, Hu S, Hamer D (1998). A functional polymorphism in the monoamine oxidase A gene promoter. Hum Genet 103, 273–279.
Sachidanandam R, Weissman D, Schmidt SC et al. (2001). A map of human genome sequence variation containing 1.42 million single nucleotide polymorphisms. Nature 409, 928–933.
Salichon N, Gaspar P, Upton AL et al. (2001). Excessive activation of serotonin (5-HT) 1B recep- tors disrupts the formation of sensory maps in monoamine oxidase A and 5-HT transporter knock-out mice. J Neurosci 21, 884–896.
Samochowiec J, Lesch KP, Rottmann M et al. (1999). Association of a regulatory polymorphism in the promoter region of the monoamine oxidase A gene with antisocial alcoholism. Psychiatry Res 86, 67–72.
Simeone A (2000). Positioning the isthmic organizer where Otx2 and Gbx2 meet. Trends Genet 16, 237–240.
Smeraldi E, Zanardi R, Benedetti F, Di Bella D, Perez J, Catalano M (1998). Polymorphism within the promoter of the serotonin transporter gene and antidepressant efficacy of fluvoxamine [see comments]. Mol Psychiatry 3, 508–511.
Spurlock G, Heils A, Holmans P et al. (1997). A family based association study of the T102C poly- morphism in 5HT2A and schizophrenia plus identification of new polymorphisms in the pro- moter. Mol Psychiatry 3, 42–49.
Stoeber G, Jatzke S, Meyer J et al. (1998). Short CAG repeats within the hSKCa3 gene associated with schizophrenia: results of a family-based study. Neuroreport 9, 3595–3599.
Stoltenberg SF, Burmeister M (2000). Recent progress in psychiatric genetics – some hope but no hype. Hum Mol Genet 9, 927–935.
Tamura K, Kanzaki T, Saito Y, Otabe M, Saito Y, Morisaki N (1997). Serotonin (5-hydroxytrypt- amine, 5-HT) enhances migration of rat aortic smooth muscle cells through 5-HT2 receptors. Atherosclerosis 132, 139–143.
Tecott LH, Barondes SH (1996). Genes and aggressiveness. Behav Genet Curr Biol 6, 238–240.
Turecki G, Briere R, Dewar K et al. (1999). Prediction of level of serotonin 2A receptor binding by serotonin receptor 2A genetic variation in postmortem brain samples from subjects who did or did not commit suicide. Am J Psychiatry 156, 1456–1458.
van Doorninck JH, van der Wees J, Karis A et al. (1999). GATA-3 is involved in the development of serotonergic neurons in the caudal raphe nuclei. J Neurosci 19, RC12.
Veenstra-VanderWeele J, Anderson GM, Cook EH Jr. (2000). Pharmacogenetics and the seroto- nin system: initial studies and future directions. Eur J Pharmacol 410, 165–181.
Ye W, Shimamura K, Rubenstein JL, Hynes MA, Rosenthal A (1998). FGF and Shh signals control dopaminergic and serotonergic cell fate in the anterior neural plate. Cell 93, 755–766.
126 K. P. Lesch, J. Benninghoff, and A. Schmitt
Zanardi R, Benedetti F, Di Bella D, Catalano M, Smeraldi E (2000). Efficacy of paroxetine in depression is influenced by a functional polymorphism within the promoter of the serotonin transporter gene. J Clin Psychopharmacol 20, 105–107.
Zhu Q, Shih JC (1997). An extensive repeat structure down-regulates human monoamine oxidase A promoter activity independent of an initiator-like sequence. J Neurochem 69, 1368–1373.
Zhu QS, Chen K, Shih JC (1994). Bidirectional promoter of human monoamine oxidase A (MAO A) controlled by transcription factor Sp1. J Neurosci 14, 7393–7403.
7
RNA processing regulation and interindividual variation
Colleen M. Niswender1 and Linda K. Hutchinson2
1 The Department of Pharmacology, University of Washington, Seattle, USA 2 The Department of Pharmacology, Vanderbilt University, Nashville, USA
OVERVIEW
In the search for causes of human disease and variability of drug response, the study of inter- individual differences in RNA processing has lagged substantially behind analyses at the DNA level. The processes of RNA editing and RNA splicing represent important mechanisms that ultimately contribute to the expression of specific protein isoforms within a given cell. Moreover, these events are subject to complex regulation that differs with each cell’s make- up, permitting intricate regulation of the cellular protein repertoire. This review will focus upon the post-transcriptional processes of RNA editing and alternative splicing and consider the contribution of aberrations within these events to the efficacy of pharmacotherapy for psychiatric diseases. Specific examples of RNA processing defects within receptors for various neurotransmitters such as dopamine, glutamate, and serotonin will be presented. In addition, mechanisms involved in the regulation of RNA editing and splicing will be addressed as contributors to disease etiology and treatment. It is anticipated that studies of RNA processing regulation will enhance our understanding of disease pathology and even- tually improve the rational design of therapeutic compounds.
Introduction
The sequencing of the human genome will almost certainly unlock the secrets of a multitude of human diseases. For the first time, both exonic and intronic regions of the genetic code will be available for analyses of potential disease-causing muta- tions. For disorders such as depression and schizophrenia, the identification of individual genetic variability will almost certainly provide information as to their etiology and treatment. Despite the answers that the human genome may provide, the possibility remains that dysregulation of post-transcriptional processing events could contribute to the cause or the response to treatment of different diseases. In this review, defects in the RNA processing steps of splicing and editing will be con- sidered as potential causes of disease and as mechanisms complicating the efficacy of pharmacotherapy.

127

128 C. M. Niswender and L. K. Hutchinson
Overview of post-transcriptional RNA processing
Messenger RNA (mRNA) transcripts are generated by the transcriptional activity of RNA polymerase II (Pol II). Increasing evidence suggests that processing and transcription of RNA transcripts are closely coupled. Indeed, many of the factors necessary for RNA processing have been shown to bind to, or be closely associated with, the highly phosphorylated C-terminal domain of Pol II. Transcript RNAs may undergo several processing events prior to translation, including 5 -end capping, RNA editing, constitutive and alternative splicing, and polyadenylation.
Capping is the earliest processing event to take place on the nascent RNA tran- script. Three enzymes, a phosphatase, a guanyl transferase, and a methylase, add 7- methyl guanine to the trisphosphate end of the transcript (Proudfoot, 2000). This modification occurs before the transcript is ~30 nucleotides in length and may mark the switch from transcription initiation to elongation. Transcripts without the 5 -end protective cap are sensitive to exonuclease attack and subsequent degra- dation.
Chronologically, RNA editing is probably the next processing event to occur. RNA editing is defined as an RNA processing event (excluding splicing) that gen- erates a transcript with a primary nucleotide sequence that differs from its gene (Simpson and Emeson, 1996; Smith et al., 1997). RNA editing events can be divided into two major categories: base modification and insertion/deletion. In mammals, cytidine-to-uridine (C-to-U) and adenosine-to-inosine (A-to-I) base modifications are the predominant forms of editing that have been identified. This review will focus on A-to-I editing and its role in the processing of several RNA transcripts in the central nervous system, including glutamate receptor subunits, the serotonin (5-HT) 2C receptor (5-HT2CR), and one of the enzymes catalyzing A-to-I modifications, adenosine deaminase (ADA), which acts on RNA 2 (or ADAR2).
In an A-to-I RNA editing event, adenosine is converted to inosine by hydrolytic deamination at the C-6 position of the purine ring and this is accomplished by the coordinate action of enzymes of the ADAR family (Rueter and Emeson, 1998). ADAR1 and ADAR2, the best-characterized members of this enzyme family, contain a catalytic deaminase domain and three or two double-stranded (ds) RNA- binding domains, respectively. Their activity is dependent on the presence of an RNA duplex structure in the substrate RNA (Rueter and Emeson, 1998). In all of the A-to-I editing events characterized to-date, this duplex is formed between inverted repeat sequences by base pairing of exonic and intronic regions of the transcript.
Because the double-stranded RNA structures necessary for editing depend on the coordinate interaction of exonic and intronic sequences, editing must precede
    
129 RNA processing regulation and interindividual variation
splicing in the processing of the RNA transcript. The spliceosome, a dynamic macromolecular complex of small nuclear ribonucleoprotein particles (snRNPs) and extrinsic (non-snRNP) proteins that assemble on the 5 – and 3 -splice sites, is responsible for pre-mRNA splicing. Introns are excised in two transesterification reactions in which the upstream exon is cleaved from the intron and ligated to the downstream exon (Fig. 7.1).
Assembly of the spliceosome begins with recruitment of U1 snRNP to the 5 splice donor site and U2 snRNP to the branchpoint of the 3 acceptor site (Fig. 7.1). The consensus sequence for the 5 donor site is GURAGU, in which the initial GU dinu- cleotide is almost invariant, and the consensus sequence for the 3 acceptor site is (Y)nXCAG (Mount, 1982; Jackson, 1991). Three other snRNPs, U4, U5, and U6, as well as up to 50 other proteins, are involved in the splicing reaction (Murray and Jarrell, 1999). These additional proteins include splicing regulatory (SR) proteins, CTD (C-terminal domain)-associated SR-like protein (CASP), and an SR-like CTD- associated factor (SCAF) (Proudfoot, 2000). The SR proteins typically contain one or two RNA recognition motifs (RRM) that bind RNA as well as an arginine-and- serine-rich domain (RS domain). The RS domain appears to serve as a molecular “glue”, allowing RS–RS interactions that promote spliceosome assembly and target- ing to regulatory sequences. The assembly of all of these proteins with the target RNA transcript is critical for the proper splicing events to occur in the maturation process.
Polyadenylation and 3 -cleavage are likely the final events of transcript matura- tion. At least six factors are required for polyadenylation and 3 -cleavage of mam- malian transcripts, ensuring the correct length of the poly-A tail and cleavage at the appropriate sites (Minvielle-Sebastia and Keller, 1999). Following complete tran- script maturation, the mRNA is released from the Pol II complex and exported to the cytoplasm.
RNA processing defects in psychiatric disorders
While the examination of genomic DNA can identify disease-segregating muta- tions in regulatory elements necessary for post-transcriptional processing (such as mutations within splice sites), the generation of the expressed protein often involves events “hidden” from genomic DNA. These may include alternative splice site selection or the occurrence of editing within an RNA transcript. The discovery that a growing number of transcripts expressed in the brain undergo alternative splicing and RNA editing events that alter protein function suggest that interindi- vidual variation in RNA processing may play a critical role in the response to drugs. A few candidate substrates have been examined directly for aberrant processing in psychiatric disorders; these will be reviewed as examples of the types of event that may affect drug function.


130 C. M. Niswender and L. K. Hutchinson (a)
b
(b)
131 RNA processing regulation and interindividual variation
Aberrant alternative splicing of RNA transcripts in psychiatric disorders
RNA transcripts
A small number of RNA transcripts have been reported to exhibit alterations in RNA splicing patterns in various psychiatric populations. Included among these are RNAs encoding several receptors for neurotransmitters such as glutamic acid, -aminobutyric acid (GABA), and dopamine. Because of the ability of antipsy- chotic and antidepressant drugs to modulate neurotransmitter receptor function, defects within the splicing of these RNAs may affect the efficacy of drug treatment as well as the manifestation of the disease.
The NMDA R1 subunit of ionotropic glutamate receptor

Fig. 7.1.
The N-methyl--aspartate (NMDA) receptor is a multisubunit ion channel that gates sodium, potassium, and calcium ions. Two subunits, arising from five distinct genes, form the basis of the NMDA receptor channel. The NR1 subunit is common to all functional NMDA receptors and is encoded by a single gene (Moriyoshi et al., 1991). The NR2 subunits (NR2A, NR2B, NR2C, NR2D) are generated from differ- ent genes (Ikeda et al., 1992; Kutsuwada et al., 1992; Meguro et al., 1992; Monyer et al., 1992; Ishii et al., 1993) and can substitute for each other within the channel. Receptors are only active in the presence of NR1 (Monyer et al., 1994). Mice that are homozygous null for the NR1 gene die at postnatal day 0, presumably from res- piratory arrest (Forrest et al., 1994). Interestingly, mice expressing a hypomorphic allele of NR1 that results in 5–10% of normal protein levels exhibit behaviors that resemble aspects of human schizophrenia, such as stereotypy and inappropriate social behaviors (Mohn et al., 1999).
The involvement of NMDA receptors in schizophrenia has been based largely
Mechanisms of RNA splicing. This diagram represents a simplistic model of RNA splicing. Many other proteins, not depicted, are involved in the reaction. (a) In the first step of RNA splicing, U1 snRNP (small nuclear ribonucleoprotein; shown as a gray oval) is recruited to the 5 donor site of an upstream exon (black). U2 snRNP (white oval) is then recruited to the branchpoint (UACUAAC) of the 3 acceptor site of the downstream exon (gray box). A U4/U5/U6 trimer binds with U1, recognizing the 5 site of the upstream exon and U6 binds to U2. U1 is then released and U5 shifts from the exon to the intron. (b) The removal of the intron occurs by two transesterification reactions. The first step, catalyzed by U6 snRNP, occurs when the branchpoint adenosine “attacks” the 5 donor, resulting in the formation of an intronic “lariat” structure. The second reaction occurs when the 3 end of the upstream exon attacks the acceptor site of the downstream exon. Subsequent cleavage at the 3 site results in final ligation of the exons and release of the matured RNA transcript.
132 C. M. Niswender and L. K. Hutchinson

Fig. 7.2.
Alternative splice variants of the NR1 subunit of N-methyl-d-aspartate (NMDA receptor). The topology of the rat NR1 subunit is shown with three putative transmembrane domains and one membrane-associated domain. The amino acid sequences of the alternatively spliced regions are shown. The N-terminal cassette represents an inclusion or exclusion of exon 3 (Hollmann et al., 1993). C1 is encoded by exon 21 (Hollmann et al., 1993). Cassettes C2 and C2 reside within the same exon (exon 22; Hollmann et al., 1993) and encode alternative splice acceptor sequences; if C2 is skipped, the normally 3 untranslated C2 region is converted into 22 translated amino acid residues. The postsynaptic density protein 95 interaction sequence (STVV) within the C2 cassette is indicated in bold.
upon the observations that noncompetitive receptor antagonists, such as phency- clidine (PCP) and ketamine, can induce psychotic symptoms. It has been proposed that excess glutamatergic signaling, induced by hypofunction of the NMDA recep- tor and a subsequent disinhibition of normal NDMA receptor-mediated negative feedback, can produce psychotic symptoms (Olney et al., 1999). Studies examining the RNA expression levels of various NMDA subunits have shown alterations in the brains of schizophrenics (Akbarian et al., 1996; Humphries et al., 1996; Sokolov, 1998; Grimwood et al., 1999; Le Corre et al., 2000). In the case of NR1, mRNA levels have been observed to be substantially decreased in the superior temporal gyrus of patients with cognitive impairment (Humphries et al., 1996) and in the frontal cortex of an elderly subset of schizophrenic individuals (Sokolov, 1998). These results have suggested that defects in NR1 mRNA expression or regulation might underlie or result from disease pathology in certain subjects.
NR1 RNA can give rise to at least eight distinct protein isoforms, which are gen- erated by alternative splicing at the N- and C-terminal ends of the protein (Hollmann et al., 1993) (Fig. 7.2). Several splicing events generate alternative C- terminal tails of NR1; these isoforms have been shown to exhibit differences in surface expression and calcium influx in a tissue culture system (Okabe et al.,
133 RNA processing regulation and interindividual variation
1999). Le Corre and colleagues (2000) found that the splice variant of NR1 that lacks both the C1 and C2 terminal cassettes (Fig. 7.2) was elevated in the superior temporal gyrus of schizophrenics by 22%, resulting in a 15% increase in total NR1 mRNA levels. These data conflict with reports of a downregulation of total NR1 mRNA expression in selected subjects (Humphries et al., 1996; Sokolov, 1998). They substantiate other studies, however, showing an increase in NR1 ligand binding in the superior temporal gyrus of middle-aged schizophrenics (Grimwood et al., 1999).
Le Corre and coworkers (2000) suggest that these discrepancies may represent distinct phases of the disease, with increased NR1 levels early and decreased amounts as the disease progresses. It is interesting to note that the isoform found to be upregulated in this study is one that, in contrast to several of the other splice variants, targets efficiently to the cell surface in vitro (Okabe et al., 1999). This NR1 variant also contains a motif necessary for interacting with postsynaptic density protein 95 (PSD-95) (Kornau et al., 1995), a clustering molecule important for the integration of cellular signaling (Fig. 7.2). Based on these observations, it is pos- sible that the amount of signaling-competent receptor expressed at the cell surface is changed in susceptible patients, altering glutamatergic activity in selected brain regions. In regard to the NDMA receptor “hypofunction” model (Olney et al., 1999), the increase in surface expression of this NR1 variant may reflect an attempt at compensation for altered NMDA receptor signaling, perhaps transiently during an affected individual’s lifetime. It may also reveal a time period in which NMDA receptor-based therapeutics may be most effective during the course of illness.
The GABA-A receptor
GABA is the major inhibitory neurotransmitter in the mammalian central nervous system. Pharmacological agents that target the GABAergic system have therapeutic roles in anxiety, depression, and schizophrenia. Previous studies have suggested that GABAergic dysfunction occurs in various schizophrenic populations, either at the level of the transmitter (Simpson et al., 1989; Reynolds et al., 1990; Sherman et al., 1991; Akbarian et al., 1995a; Ohnuma et al., 1999) or the receptor (Squires et al., 1993). These observations have prompted further examination of GABA- mediated neurotransmission in psychiatric disorders.
GABA-A receptors are composed of subunits termed , , and . The 2 subunit is responsible for high-affinity benzodiazepine binding (Pritchett et al., 1989; von Blankenfeld et al., 1990; Wafford et al., 1991) and is regulated by alternative splic- ing to produce a long and short form (2L and 2S, respectively) that differ in length by eight amino acid residues (Whiting et al., 1990; Kofuji et al., 1991) (Fig. 7.3). The long form, 2L, contains an additional phosphorylation site for protein kinase C (PKC). While both the 2L and 2S subunits are negatively regulated by PKC
134 C. M. Niswender and L. K. Hutchinson
                                        
Fig. 7.3.
Alternative splice variants of the -aminobutyric acid (GABA) 2 subunit. The predicted topology of a GABA 2 subunit protein is shown. Alternative splicing within 2 RNA results in the production of two variants, the 2S and the 2L receptors. The 2L isoform is produced by the inclusion of 24 additional nucleotides, leading to the inclusion of eight additional amino acid residues. The additional protein kinase C phosphorylation site within the 2L receptor is indicated in bold.
phosphorylation, the additional site within the 2L protein renders channels con- taining this subunit particularly susceptible to PKC inhibition (Krishek et al., 1994). It would be predicted that a biased inclusion of the 2L protein into func- tional GABA-A receptors would further reduce channel ion conductance. Individual differences in PKC phosphorylation might also affect the activity of these receptors, possibly enhancing or blunting the effect of 2L-containing recep- tors.
Studies by Akbarian et al. (1995b) and Huntsman et al. (1998) demonstrated a selective 50% reduction in 2S mRNA expression in the prefrontal cortex of a subset of five schizophrenic individuals compared with control subjects, resulting in an increased ratio of 2L:2S RNA. Further knowledge of the expression pat- terns of these isoforms in patient populations might enable more effective drug treatment. For example, the predicted dampened activity of 2L-containing receptors may prevent GABAergic agonists from functioning as effectively at these receptors, implying that subjects with high 2L receptor levels may respond better to alternative drug treatments rather than therapeutics targeting the GABA system.
135 RNA processing regulation and interindividual variation
The dopamine D2 and D3 receptors
Dopamine receptors belong to the G-protein coupled receptor superfamily; D2 and D3 receptors are targets for many clinically active antipsychotics. Both D2 and D3 receptor RNAs undergo alternative splicing in regions of the receptor that are important for interaction with cellular signaling machinery. D2 receptor RNA has previously been shown to undergo alternative splicing to produce two major RNA isoforms, termed D2long and D2short, that differ by 29 amino acid residues within the third intracellular loop of the receptor protein (Fig. 7.4a) and show distinct expres- sion patterns (Giros et al., 1989; Monsma et al., 1989). It has been reported that the short form of the receptor exhibits increased sensitivity to certain classes of anti- psychotic agent (Castro and Strange, 1993; Malmberg et al., 1993). In these studies, benzamide-substituted antipsychotics such as raclopride and remoxipride, as well as the atypical antipsychotic clozapine, exhibited higher affinity for the D2short form of the receptor. In contrast, Leysen and colleagues (1993) found equivalent affin- ities for a large number of antipsychotics at these receptor variants. The differences in expression systems and the radioligands chosen for competition assays may have contributed to the observed discrepancies. These studies are suggestive, however, of potential differences in the antipsychotic response of these alternatively spliced D2 receptor isoforms. Recently, mice have been generated that express only the D2short form of the receptor (Usiello et al., 2000; Wang et al., 2000b). While these investigators report discrepancies in the locomotor behaviors observed in these animals, both studies agree that the D2short mice show a blunted response to the cat- alepsy induced by haloperidol administration. These results suggest that the D2long form of the receptor may mediate important aspects of the extrapyramidal side effects of haloperidol-like antipsychotics as well as define the presynaptic versus postsynaptic locations of these two receptor populations (Usiello et al., 2000).
D3 receptor RNA undergoes a number of alternative splicing events. One event results in the deletion of 98 nucleotide bases within the third cytoplasmic loop of the receptor, causing a frameshift and predicted truncation of the D3 receptor protein (Schmauss et al., 1993; Liu et al., 1994). Previous studies (Schmauss et al., 1993) had shown that this alternatively spliced RNA transcript, termed D3nf, was expressed in the prefrontal cortex of schizophrenics, whereas full-length D3 RNA was lost. In an expansion of these findings, D3nf RNA was found to be produced by an unusual splicing event in which the deleted 98 nucleotide bases were recognized as an atypical intron (Schmauss, 1996) (Fig. 7.4b,c). Determination of the relative levels of D3 RNA to D3nf RNA revealed a loss of full-length D3 RNA with a concom- itant increase in D3nf message in the anterior cingulate cortex of an examined schizophrenic population (Schmauss, 1996). These results suggest that, in chronic schizophrenia, there may be aberrant RNA processing events that change the

136 C. M. Niswender and L. K. Hutchinson

(a)

(b) (c)
Fig. 7.4.
Alternative splice variants of the D2 and D3 dopamine receptors. (a) Topology of the D2 receptor with the position of insertion of the 29 amino acid residues specific to the D2Long receptor indicated. (b) Splicing of exon 1 to exon 2 results in the formation of full-length D3 receptor RNA, encoding the complete seven-transmembrane spanning receptor.
(c) Use of an alternative splice acceptor within exon 1 promotes the removal of a 98 nucleotide base minor class intron and results in the formation of D3nf RNA. Translation of this alternative splice product results in a receptor variant with a frameshift mutation; it has been hypothesized that this variant may contain a unique C-terminal transmembrane- spanning region (gray) (Elmhurst et al., 2000).

137 RNA processing regulation and interindividual variation
splicing of the primary D3 transcript, ultimately reducing the level of message that is competent for normal protein generation. D3nf RNA is translated into protein (Liu et al., 1994) and it has been shown that D3 and D3nf receptors can heterodi- merize (Liu et al., 1994). This interaction has been shown to decrease the dopamine-binding ability of the D3 receptor (Elmhurst et al., 2000), suggesting that an enhanced expression of D3nf protein may exert a dominant negative effect and “antagonize” D3 receptor signaling.
Potential mechanisms for splicing alterations
Protein isoforms expressed from alternatively spliced RNA may affect drug respon- siveness in a number of ways. For example, alterations that affect the production of a functional receptor, or the manner in which that receptor binds ligand and trans- mits a signal, are ways in which appropriate RNA processing contributes to an effec- tive response of agonists or antagonists. There are several means by which genetic differences among individuals may affect alternative splicing. These mechanisms fall into two broad categories: differences that occur within cis-acting elements and those within trans-acting factors. Cis-acting elements are defined as the nucleotide sequences and structures of the RNA molecule itself, whereas trans-acting factors are the numerous proteins and other RNA components that assemble and form the cellular splicing machinery.
Many genomic changes that affect RNA splicing occur within the splice donor and acceptor sequences, causing or contributing to disease (Krawczak et al., 1992). Often, however, a mutation may be linked to disease manifestation but not occur within known splicing elements (Krawczak et al., 1992; Cooper and Mattox, 1997; Valentine, 1998). Novel cis-active splicing elements, termed exonic splicing enhancers (ESEs) and exonic splicing suppressors (ESSs), are gaining recognition for their potential role in aberrant splicing (Cooper and Mattox, 1997; Blencowe, 2000; Philips and Cooper, 2000). ESE sequences are interaction sites for SR proteins involved in modulating splicing specificity and efficacy (Blencowe, 2000). Point mutations within ESEs, therefore, could alter SR protein binding and affect selec- tion and efficiency of splicing from a given site. RNAs involved in several disorders, such as Becker muscular dystrophy (Shiga et al., 1997) and spinal muscular atrophy (Lorson et al., 1999), have recently been shown to undergo inappropriate exon skipping or exon inclusion as a result of ESE or ESS mutations. Importantly, these mutations may occur within coding regions of the gene and result in relatively minor changes at the amino acid level (for example, silent mutations at wobble positions). A sequence that has been considered a “silent” polymorphism may actu- ally affect splice site choice, possibly contributing to disease or the effectiveness of drugs that interact with the protein product.
Individual differences in cis-active elements within one gene might be predicted
138 C. M. Niswender and L. K. Hutchinson
to limit altered processing to that particular transcript. Mutations affecting the trans-acting factors of the splicing machinery, however, would be predicted to produce more global effects (Philips and Cooper, 2000). While mutation of a ubi- quitous splicing factor might prove lethal, alteration of an alternative splicing protein might affect only a subset of RNAs and only in cells where the factor is expressed. In addition, some alternatively spliced sites are regulated by the compe- tition of splicing proteins for specific sites on the RNA; relatively small changes in the level of one protein may disrupt this delicate balance. All of these factors con- tribute to the final repertoire of spliced proteins and the effects those proteins have on cellular function.
Editing of RNA transcripts implicated in psychiatric disorders RNA transcripts
Glutamate receptors
The AMPA (-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) and kainate subtypes of glutamate receptors are multisubunit ion channels permeable to sodium and potassium (Bettler et al., 1990; Boulter et al., 1990; Keinanen et al., 1990; Nakanishi et al., 1990; Sakimura et al., 1990; Hollmann et al., 1991; Dingledine et al., 1999). Certain subunit combinations of AMPA and kainate receptors have also been shown to conduct calcium, a property previously attrib- uted solely to the NMDA subtype of glutamate receptors (Mayer and Westbrook, 1987; Mayer et al., 1987; Collingridge and Lester, 1989). Calcium influx into the postsynaptic neuron is critical for normal neurophysiological processes and may underlie activity-dependent changes necessary for the initiation and maintenance of memory (Collingridge and Singer, 1990). Glutamate receptors may also be involved in chronic and acute neurodegenerative disorders, including epilepsy, amyotrophic lateral sclerosis, Parkinson’s disease and stroke (Choi and Rothman, 1990; Olney, 1990). The observation that AMPA and kainate receptors may con- tribute to neuronal calcium influx has led to further examination of their biologi- cal roles and regulation of subunit composition.
Four subunits, GluR1, GluR2, GluR3, and GluR4, in homomeric or heteromeric combinations, make up the AMPA subtype of glutamate receptor (Boulter et al., 1990; Keinanen et al., 1990; Nakanishi et al., 1990; Hollmann et al., 1991; Sakimura et al., 1990). Channels containing the GluR2 subunit display negligible permeabil- ity to calcium, a phenotype dependent on a specific arginine residue located in second hydrophobic domain of the GluR2 protein (Hume et al., 1991; Mishina et al., 1991; Verdoorn et al., 1991). Greater than 99% of GluR2 transcripts contain an arginine (CGG) codon at this position although the genomic sequence encodes a

139 RNA processing regulation and interindividual variation
glutamine residue (CAG). It has subsequently been shown that RNA editing gen- erates the altered base at this position (the Q/R site) through an A-to-I conversion in these transcripts (Sommer et al., 1991; Burnashev et al., 1992) (Fig. 7.5). This RNA editing event is unique to the GluR2 subunit.
While Q/R site editing does not occur in the other subunits of the AMPA recep- tor, another RNA editing event occurs within the GluR2, GluR3, and GluR4 sub- units (Lomeli et al., 1994). Conversion of A to I at this site alters an arginine (AGA) codon to a glycine (GGA) codon. This site is referred to as the R/G site and is imme- diately upstream of an alternative splice site preceding the fourth hydrophobic domain (TM4) (Fig. 7.5). The alteration of this single amino acid increases the rate of recovery from receptor desensitization, presumably enhancing receptor activity (Lomeli et al., 1994).
The kainate receptor family consists of five subunits, KA1, KA2, GluR5, GluR6, and GluR7 (Dingledine et al., 1999). GluR5 and GluR6 are edited at a site analo- gous to the GluR2 Q/R site in 40 and 80% of transcripts in adult rat brain, respec- tively (Sommer et al., 1991) (Fig. 7.5). The GluR6 subunit RNA can undergo two additional editing events in the first predicted transmembrane domain (TM1), resulting in the conversion of an isoleucine codon to a valine codon (I/V site) and a tyrosine codon to a codon for cysteine (Y/C site) (Egebjerg and Heinemann, 1993; Kohler et al., 1993) (Fig. 7.5). Calcium permeability of GluR6-containing kainate receptors is dependent on the pattern of editing at all three sites, indicating that both TM1 and TM2 are involved in the ion permeation and electrophysiological properties of the kainate receptors.
Mice have been generated with a deletion of essential intronic sequences forming the RNA duplex region required for editing of the Q/R site of GluR2 (Brusa et al., 1995; Sprengel et al., 1999). Heterozygous mice carrying this editing-incompetent allele develop epileptic seizures and die by 3 weeks of age. It is interesting to note the distinct phenotype of these animals compared with mice that lack the GluR2 protein completely (Jia et al., 1996). GluR2-null animals, while exhibiting increased calcium permeability in principal neurons and defects in long-term potentiation similar to GluR2 editing-incompetent mice, do not exhibit seizures and death at postnatal day 21. These results indicate that the presence of nonedited GluR2 within an AMPA receptor is distinguishable from a receptor assembled from combinations of the other three subunits, even though both of these scenarios gen- erate calcium-permeable receptors.
The above mouse model suggests that GluR2 Q/R site editing may be involved in the disease pathology of epilepsy and other disorders in humans (Sprengel et al., 1999). A recent study of the surgically excised hippocampi of patients with refrac- tory epilepsy showed that some of these patients had decreased editing at the Q/R site of GluR2 RNA (Grigorenko et al., 1998). Other studies have revealed a
140 C. M. Niswender and L. K. Hutchinson
  
R/G site flip/flop cassette
                                                       
I/V site
Y/C site
Q/R site
I/V
GluR2 genomic cDNA
GluR3 SSLR genomic TCCGCATTAAGA
       
cDNA
TCC GCA TTA GGA
SSLG
Y/C Q/R R/G
FMQQ SSLR TTTATGCAGCAA TCCTCATTAAGA TTTATGCGGCAA TCC TCA TTA GGA
FMRQ SSLG
GluR4 SSLR
genomic cDNA
GluR5 LMQQ
TCCTCATTAAGA TCC TCA TTA GGA
SSLG
genomic cDNA
GluR6 genomic cDNA
Fig. 7.5.
CTC ATG CAG CAA CTC ATG CGG CAA
LMRQ
YILLAYL LMQQ TAT ATT CTG CTG GTC TAC TTG CTC ATG CAG CAA TAT GTT CTG CTG GTC TGC TTG CTC ATG CGG CAA
YVLLACL LMRQ
RNA editing events within glutamate receptor (GluR) subunits. The predicted topology of GluR subunits is shown with three putative transmembrane domains and one membrane- associated domain. The location of the editing sites is indicated by the white circles. The I/V and Y/C sites of GluR5 and GluR6 are located within the first transmembrane domain. The Q/R site lies near the end of the membrane-associated domain, and the R/G site is located “near” the third transmembrane domain. Below the structure, genomic DNA and cDNA sequences are aligned to show the consequences of RNA editing at these positions within the GluR1–GluR6 subunits. Adenosine-to-inosine editing results in adenosine to guanosine discrepancies between genomic DNA and cDNA sequences and changes the coding potential of the edited region of the receptor. I, isoleucine; V, valine; Y, tyrosine; C, cysteine.
141 RNA processing regulation and interindividual variation
significant elevation in the ratio of unedited to edited GluR2 in the prefrontal cortex of individuals with Alzheimer’s disease or schizophrenia and in the striatum of subjects with Huntington’s disease when compared with age-matched controls (Akbarian et al., 1995b). The increase in unedited (Q) GluR2 subunits in these dis- eases may have neurotoxic consequences, causing increased levels of intracellular calcium, which in turn trigger events that lead to structural damage and potential cell death. Whether the decrease in Q/R editing leads to these disorders or is a con- sequence of the disease in these patients remains to be explored.
The serotonin 2C receptor
The serotonin 2C receptor (5-HT2C receptor or 5-HT2CR) is a G-protein-coupled receptor that is linked to activation of phospholipase C and the production of inos- itol phosphates. A-to-I editing within human 5-HT2CR RNA occurs at five major positions, termed the A, B, C, D, and E sites (Niswender et al., 1998, 1999; Fitzgerald et al., 1999), resulting in the alteration of amino acids within the second intracel- lular loop of the receptor (Fig. 7.6). Different combinations of adenosine and inosine at these five positions are predicted to generate as many as 32 distinct RNA species encoding 24 different protein isoforms. Editing levels vary among brain areas (Burns et al., 1997; Fitzgerald et al., 1999; Wang et al., 2000b; Niswender et al., 2001), resulting in tissue-specific isoform expression. It has been reported that the human receptor variant that is edited at all five positions, termed the 5-HT2C–VGVR, undergoes a unique splicing event in the amygdala to produce a frameshifted and inactive protein (Wang et al., 2000b). This observation places into question the expression of this isoform as a full-length receptor in certain regions of the brain.
The amino acid alterations that occur as a result of editing within 5-HT2CR RNA result in differential G-protein coupling profiles of the receptors (Burns et al., 1997; Fitzgerald et al., 1999; Herrick-Davis et al., 1999; Niswender et al., 1999; Wang et al., 2000a). It has been proposed that these distinct coupling profiles result from differential ability of the receptors to isomerize spontaneously into a constitutively active conformation, giving each receptor isoform a unique “signal to noise” ratio (Herrick-Davis et al., 1999; Niswender et al., 1999). As a result of the alterations in G-protein coupling exhibited by the edited isoforms, agonists show unique potencies when interacting with distinct receptor variants. For example, the hallu- cinogens ( )-1-(4-iodo-2,5-dimethoxyphenyl)-2-aminopropane (DOI) and N,N– dimethyltryptamine (DMT) stimulate phosphoinositide turnover when interacting with the completely nonedited form of the receptor (termed the 5- HT2C-INIR). When interacting with the fully edited receptor, or 5-HT2C-VGVR, these drugs showed an approximately 50- to 100-fold reduction in potency, relative to the potency at the 5-HT2C-INIR (Niswender et al., 1999). Unique among agonists exam- ined thus far, ( )-lysergic acid diethylamide (LSD) stimulates phosphoinositide

142
C. M. Niswender and L. K. Hutchinson
Fig. 7.6.
RNA editing events within the serotonin 2C receptor (5-HT2CR). The nonedited RNA sequence of the 5-HT2CR is shown at the top with the position of the five editing sites indicated in bold. Below is shown the predicted topology of the 5-HT2CR protein with the amino acid changes introduced by adenosine-to-inosine editing. Specific combinations of editing can induce distinct amino acid changes at certain sites; for example, editing at the A position alone or the A and B positions concurrently introduces a valine at position 156 of the human receptor. Editing at the B site alone, however, results in the inclusion of a methionine at this site, leading to three amino acids that may be encoded at this position (isoleucine, valine, or methionine) depending upon editing status. I, isoleucine; N, asparagine; M, methionine; V, valine; D, aspartic acid; S, serine; G, glycine.
production when interacting with the 5-HT2C-INIR but appears completely inactive at the 5-HT2C-VGVR (Backstrom et al., 1999; Fitzgerald et al., 1999; Niswender et al., 2001). In addition, pretreatment of 5-HT2C-VGVR-expressing cells with LSD results in a blockade of subsequent activation by 5-HT, suggesting that LSD may function as an antagonist at certain 5-HT2CR isoforms (Niswender et al., 2001). These results indicate that some hallucinogenics may interact efficiently only with distinct 5- HT2CR edited isoforms and also suggests that individual responses to hallucinogens may be determined by the isoforms expressed.
Many drugs used in the treatment of schizophrenia and depression interact with high affinity at the 5-HT2CR. Several of these agents act as simple antagonists of the system, blocking the activity of agonists. Other drugs, such as clozapine, also inhibit the basal, or constitutive, activity of the 5-HT2CR (Barker et al., 1994;
143 RNA processing regulation and interindividual variation
Westphal and Sanders-Bush, 1994); these drugs are termed inverse agonists. The affinity of agonists and inverse agonists for a G-protein-coupled receptor is regu- lated by G-protein interaction. Agonists exhibit higher affinity for the G-protein- coupled form of the receptor, and inverse agonists preferentially interact with the uncoupled form. Recently, several inverse agonists have been shown to exhibit higher affinity at the 5-HT2C-INI isoform compared with the 5-HT2C-VGVR (Niswender et al., 2001). These results again suggest that the isoforms have differ- ential abilities to “precouple” to G-proteins, with the 5-HT2C-INIR residing predom- inantly in the precoupled state and the 5-HT2C-VGVR existing preferentially in an uncoupled form. Several groups have now assessed the ability of a large number of antipsychotics and antidepressants to block constitutive activity of 5-HT2CRs, finding that some agents are effective inverse agonists while others behave as neutral antagonists (Herrick-Davis et al., 2000; Rauser et al., 2001; Weiner et al., 2001). The inverse agonist or neutral antagonist profile of a drug also depends upon the edited receptor examined because of the distinctions in basal activity between variants (Niswender et al., 2001; Rauser et al., 2001). It is intriguing to speculate that the clinical efficacy of certain agents may be related to their ability to block con- stitutive receptor activation and that failure rates for certain antipsychotics may involve interindividual differences in the edited receptor repertoire.
Because of the observed differences in interaction of both hallucinogenic drugs and antipsychotic medications with various 5-HT2CRs, it is possible that changes in editing status may be involved in psychiatric diseases. Examination of 5-HT2CR RNA editing levels from the prefrontal cortex of 13 normal individuals, 13 subjects with major depression, and 13 individuals with schizophrenia revealed a small but significant increase in the level of editing at the A position in patients who had committed suicide (Niswender et al., 2001). Recently, Sodhi et al. (2001) also per- formed a detailed sequencing analysis on RNA samples derived from control and schizophrenic cortex, finding a significant increase in the expression of the 5-HT2C-INIR and a decreased expression of the 5-HT2C-VNVR and 5-HT2C-VSVR isoforms. These results suggest that these patients express unique repertoires of edited receptors, possibly resulting in changes in 5-HT-mediated signaling through the involvement of this area of the receptor in coupling efficiency (Wong et al., 1990; Moro et al., 1993; Pin et al., 1994; Arora et al., 1995, 1997; Blin et al., 1995; Verrall et al., 1997; Ballesteros et al., 1998). These results, coupled with the poten- tial differences in interaction of psychiatric medications with various 5-HT2CRs, suggest that detailed analyses from large numbers of affected and control individ- uals are warranted.
Potential mechanisms for editing alterations
As is the case for splicing, both cis- and trans-active elements are critical for the maintenance of efficient editing. A-to-I RNA editing relies upon the presence of a
144 C. M. Niswender and L. K. Hutchinson
dsRNA structure formed by base pairing between inverted repeat sequences within the RNA (Rueter and Emeson, 1998). In the examples of A-to-I editing identified so far, one half of the repeat resides within the intron immediately downstream of the editing sites. The folding pattern of this RNA template helps to direct editing target choice; only a few adenosines are edited within a large sequence of RNA. Interindividual polymorphisms or mutations may affect folding of this RNA struc- ture, altering editing choice or efficiency.
The trans-acting factors responsible for A-to-I editing are a family of dsRNA- specific ADARs. Both ADAR1 and ADAR2 appear to be expressed in most tissues and tissue culture cell lines (Wagner et al., 1990) but exhibit differing substrate specificity. In vitro, ADAR1 can efficiently edit the R/G site of GluR subunits, the Q/R site of GluR5 RNA (Herb et al., 1996; Maas et al., 1996) and the A and B sites of the 5-HT2CR transcripts (Burns et al., 1997). ADAR2 can also edit the R/G site of GluR subunits but, unlike ADAR1, is able efficiently to edit the Q/R site of GluR2 RNA (Melcher et al., 1996) and the C and D sites of 5-HT2CR transcripts (Burns et al., 1997).
These enzymes undergo alternative splicing events that may control the level of editing activity in the cell. Two isoforms of ADAR1 have been described (Patterson and Samuel, 1995). One is a 150kDa interferon (IFN)-inducible variant expressed in both the cytoplasm and nucleus of human amnion and neuroblastoma cell lines. The other, a 110 kDa isoform, is constitutively expressed exclusively in the nucleus. Two promoters, one IFN inducible and the other not, initiate transcription of the gene ADAR1 (George and Samuel, 1999). Alternative splicing of unique exon 1 structures to a common exon 2 junction results in expression of either the 110 kDa ADAR1 protein or the IFN-induced 150 kDa ADAR1 protein (George and Samuel, 1999). Three variants of the IFN-inducible human ADAR1 have been described, termed ADAR1a, ADAR1b, and ADAR1c (Liu et al., 1997) (Fig. 7.7a). All three iso- forms exhibit similar dsRNA ADAR activities for a synthetic RNA duplex, although mutagenic analyses of the three dsRNA-binding domains suggest that the role of each in the activity or specificity of spliced ADAR1 isoforms may not be identical (Liu et al., 1997).
ADAR2 may also undergo several alternative splicing events (Fig. 7.7b). Compared with ADAR2a, ADAR2b contains an insertion of 120 nucleotide bases within the deaminase domain (Gerber et al., 1997; Lai et al., 1997). Comparison of the activities of recombinant ADAR2a and ADAR2b reveal that ADAR2a is approx- imately twice as active as ADAR2b for all RNA substrates (Gerber et al., 1997). Further ADAR2 diversity is generated by alternative splicing at the 3 end of the coding region, generating four isoforms. ADAR2c and ADAR2d, which have a truncated C-terminus, exhibit negligible editing activity for the Q/R and R/G sites of GluR2 pre-mRNA (Lai et al., 1997). It has been suggested that these protein
145 RNA processing regulation and interindividual variation
isoforms may serve as competitive inhibitors of the other editing-competent iso- forms or that the altered C-terminal structure allows them to modify adenosine res- idues within currently unidentified RNA targets (Lai et al., 1997).
ADAR2 expression is further regulated by “autoediting”, which directs the pro- duction of two additional isoforms (2e and 2f) (Rueter et al., 1999). In this case, ADAR2 edits its own RNA to generate a new splice acceptor site, resulting in the inclusion of an inactivating 47 nucleotide base cassette (Fig. 7.7c). Inclusion of this alternatively spliced region only occurs after the sequence AA has been modified to AI, effectively mimicking a functional AG splice acceptor. Addition of these 47 nucleotide bases alters the reading frame and results in truncation of the protein. It has been hypothesized that “autoediting” may serve as a mechanism to regulate the levels of functional ADAR2 protein. For example, transgenic mice designed to overexpress a cytidine deaminase (APOBEC-1) involved in the C-to-U editing of apolipoprotein B transcripts have been shown to develop liver tumors as a result of editing of substrates that are not normal targets in control animals (Yamanaka et al., 1995). Extrapolation of these findings to the ADAR enzymes suggests that cel- lular expression may be tightly regulated to ensure editing of the correct nucleo- tides.
Mice were recently generated that are homozygous for a targeted null allele of ADAR2 (Higuchi et al., 2000). Editing in these mice is reduced at many of the described editing positions in several transcripts. The mutant mice die between postnatal days 0 and 20, becoming progressively seizure prone after postnatal day 12. Interestingly, this phenotype can be rescued by replacing both GluR2 alleles with alleles encoding the edited (R) version of GluR2. While these studies indicate that ADAR2 expression and GluR2 Q/R site editing is critical for viability, the con- sequences of reduced editing at other sites has not been addressed. It is anticipated that differences in ADAR splice variant production, as well as potential polymor- phisms/mutations with cis-active editing elements, will determine the amounts and patterns of edited proteins expressed by a given individual.
Conclusions
Understanding the regulation of RNA processing is certain to enhance our ability to treat disease. The capacity of drugs to interact uniquely with certain splice vari- ants or edited proteins represents an opportunity to tailor pharmacotherapy to a given individual. Knowledge that patients with a specific disease express an altered subset of edited 5-HT2CRs, for example, may allow the rational design or use of drugs that target these receptor variants. It is also important to consider that changes in alternative splicing/editing events may represent a cellular compensa- tory mechanism; using drugs that antagonize a cell’s attempts to compensate for

146 C. M. Niswender and L. K. Hutchinson
(a) Adenosine deaminase that acts on RNA 1
1
1
1 1c
502
502
502
573 613 684 725 796
Double-stranded RNA-binding domains
573 613 684 725 796
908 1036 1226
Deaminase domain
882 1010 1200
     
1a
1b
NLS
     
807 VTPVTGASLRRTMLLLSRSPEAQPKT 832
573 613 684 712 777 863 991 1181
     
695 EGMISESLDNLESMMPNK 712 (b) Adenosine deaminase that acts on RNA 2
(b)
2a
76 146 230 300 392 526 673701
1
1
1
1
1 2e
47 NT insert
Genomic DNA RNA
1. No acceptor editing; cassette skipping
2. Acceptor editing; cassette inclusion

NLS
Do
uble-str
anded
Deaminase
RNA-binding domains domain
76 146 230 300 392 466505 566 713741
GSRSYTQAGVQWCNMGSLQPRPPGLLSDPSTSTFQGAGTT 76 146 230 300 392 466505 566 714
VH 76 146 230 300 392 526 674
Premature stop codon
ARLFTAFIKAGLGAWVEKPTEQDQFSLTP
                              
2b
2c
2d
                                                                                                                 
2f (c)

(c)
CCACAG
TTACAA
CCACAG
TTACAA
CCACAG
TTACAI
                     
147 RNA processing regulation and interindividual variation
disease-induced alterations may prove ineffective. In addition, changes in RNA processing may result from drug treatment. For example, drugs that affect IFN pro- duction may influence the activity of ADAR1, possibly changing editing efficiency. The identification of RNA processing events that affect psychotropic drug activity will continue to be challenging. Understanding these processes, however, will cer- tainly contribute to effective disease treatment.
Acknowledgements
The authors would like to thank Dr Ron Emeson, Dr Elaine Sanders-Bush, Ray Price, and T. Renee Dawson for critical reading of the manuscript. This work was supported by an NRSA fellowship (CMN).
REFERENCES
 
Fig. 7.7.
Akbarian S, Huntsman MM, Kim JJ et al. (1995a). GABAA receptor subunit gene expression in human prefrontal cortex: comparison of schizophrenics and controls. Cereb Cortex 5, 550–560.
Akbarian S, Smith MA, Jones EG (1995b). Editing for an AMPA receptor subunit RNA in pre- frontal cortex and striatum in Alzheimer’s disease, Huntington’s disease and schizophrenia. Brain Res 699, 297–304.
Akbarian S, Sucher NJ, Bradley D et al. (1996). Selective alterations in gene expression for NMDA receptor subunits in prefrontal cortex of schizophrenics. J Neurosci 16, 19–30.
Arora KK, Sakai A, Catt KJ (1995). Effects of second intracellular loop mutations on signal trans- duction and internalization of the gonadotropin-releasing hormone receptor. J Biol Chem 270, 22820–22826.
RNA processing events within human RNA-specific adenosine deaminases (ADAR) 1 and 2. Schematic representations of ADAR1 (a) and ADAR2 (b) protein isoforms are shown, with the locations of the nuclear localization signal (NLS) (black boxes), double-stranded RNA-binding domains (light gray boxes) and the adenosine deaminase domain (dark gray boxes) indicated. The amino acid residues deleted in the ADAR1b and ADAR1c isoforms are indicated with the one-letter amino acid code. Also indicated are the amino acids inserted in the deaminase domain of ADAR2b and ADAR2c as well as at the C-termini of the ADAR2 isoforms. The location of the 47 nucleotide insertion in ADAR2e and ADAR2f is also indicated. (c) The genomic sequence and the sequence of nonedited and edited ADAR2 RNA are shown. If editing at the upstream splice acceptor does not occur, splicing to the downstream acceptor is preferred and results in the formation of RNA encoding full-length ADAR2. Adenosine-to-inosine editing of the upstream acceptor results in the inclusion of the 47 nucleotide base inactivating cassette, causing a frameshift and truncation of the protein.
148 C. M. Niswender and L. K. Hutchinson
Arora KK, Cheng Z, Catt KJ (1997). Mutations of the conserved DRS motif in the second intra- cellular loop of the gonadotropin-releasing hormone receptor affect expression, activation, and internalization. Mol Endocrinol 11, 1203–1212.
Backstrom JR, Chang MS, Chu H, Niswender CM, Sanders-Bush E (1999). Agonist-directed sig- naling of serotonin 5-HT2C receptors: differences between serotonin and lysergic acid diethy- lamide (LSD). Neuropsychopharmacology 21, 77S–81S.
Ballesteros J, Kitanovic S, Guarnieri F et al. (1998). Functional microdomains in G-protein- coupled receptors. The conserved arginine-cage motif in the gonadotropin-releasing hormone receptor. J Biol Chem 273, 10445–10453.
Barker EL, Westphal RS, Schmidt D, Sanders-Bush E (1994). Constitutively active 5-hydroxytryptamine 2C receptors reveal novel inverse agonist activity of receptor ligands. J Biol Chem 269, 11687–11690.
Bettler B, Boulter J, Hermans-Borgmeyer I et al. (1990). Cloning of a novel glutamate recep- tor subunit, GluR5: expression in the nervous system during development. Neuron 5, 583–595.
Blencowe BJ (2000). Exonic splicing enhancers: mechanism of action, diversity and role in human genetic diseases. Trends Biochem Sci 25, 106–110.
Blin N, Yun J, Wess J (1995). Mapping of single amino acid residues required for selective activa- tion of Gq/11 by the M3 muscarinic acetylcholine receptor. J Biol Chem 270, 17741–17748. Boulter J, Hollmann M, O’Shea-Greenfield A et al. (1990). Molecular cloning and functional
expression of glutamate receptor subunit genes. Science 249, 1033–1037.
Brusa R, Zimmermann F, Koh DS et al. (1995). Early-onset epilepsy and postnatal lethality asso-
ciated with an editing-deficient GluR-B allele in mice. Science 270, 1677–1680.
Burnashev N, Monyer H, Seeburg PH, Sakmann B (1992). Divalent ion permeability of AMPA
receptor channels is dominated by the edited form of a single subunit. Neuron 8, 189–198. Burns CM, Chu H, Rueter SM et al. (1997). Regulation of serotonin-2C receptor G-protein cou-
pling by RNA editing. Nature 387, 303–308.
Castro SW, Strange PG (1993). Differences in the ligand binding properties of the short and long
versions of the D2 dopamine receptor. J Neurochem 60, 372–375.
Choi DW, Rothman SM (1990). The role of glutamate neurotoxicity in hypoxic-ischemic neuro-
nal death. Annu Rev Neurosci 13, 171–182.
Collingridge GL, Lester RA (1989). Excitatory amino acid receptors in the vertebrate central
nervous system. Pharmacol Rev 41, 143–210.
Collingridge GL, Singer W (1990). Excitatory amino acid receptors and synaptic plasticity. Trends
Pharmacol Sci 11, 290–296.
Cooper TA, Mattox W (1997). The regulation of splice-site selection, and its role in human
disease. Am J Hum Genet 61, 259–266.
Dingledine R, Borges K, Bowie D, Traynelis SF (1999). The glutamate receptor ion channels.
Pharmacol Rev 51, 7–61.
Egebjerg J, Heinemann SF (1993). Ca2 permeability of unedited and edited versions of the
kainate selective glutamate receptor GluR6. Proc Natl Acad Sci, USA 90, 755–759.
Elmhurst JL, Xie Z, BF OD, George SR (2000). The splice variant D3nf reduces ligand binding to the D3 dopamine receptor: evidence for heterooligomerization. Res Mol Brain Res 80, 63–74.
149 RNA processing regulation and interindividual variation
Fitzgerald LW, Lyer G, Conklin DS et al. (1999). Messenger RNA editing of the human serotonin 5-HT2C receptor. Neuropsychopharmacology 21, 82S–90S.
Forrest D, Yuzaki M, Soares HD et al. (1994). Targeted disruption of NMDA receptor 1 gene abol- ishes NMDA response and results in neonatal death. Neuron 13, 325–338.
George CX, Samuel CE (1999). Human RNA-specific adenosine deaminase ADAR1 transcripts possess alternative exon 1 structures that initiate from different promoters, one constitutively active and the other interferon inducible. Proc Natl Acad Sci, USA 96, 4621–4626.
Gerber A, O’Connell MA, Keller W (1997). Two forms of human double-stranded RNA-specific editase 1 (hRED1) generated by the insertion of an Alu cassette. RNA 3, 453–463.
Giros B, Sokoloff P, Martres MP, Riou JF, Emorine LJ, Schwartz JC (1989). Alternative splicing directs the expression of two D2 dopamine receptor isoforms. Nature 342, 923–926.
Grigorenko EV, Bell WL, Glazier S, Pons T, Deadwyler S (1998). Editing status at the Q/R site of the GluR2 and GluR6 glutamate receptor subunits in the surgically excised hippocampus of patients with refractory epilepsy. Neuroreport 9, 2219–2224.
Grimwood S, Slater P, Deakin JF, Hutson PH (1999). NR2B-containing NMDA receptors are up- regulated in temporal cortex in schizophrenia. Neuroreport 10, 461–465.
Herb A, Higuchi M, Sprengel R, Seeburg PH (1996). Q/R site editing in kainate receptor GluR5 and GluR6 pre-mRNAs requires distant intronic sequences. Proc Natl Acad Sci, USA 93, 1875–1880.
Herrick-Davis K, Grinde E, Niswender CM (1999). Serotonin 5-HT2C receptor RNA editing alters receptor basal activity: implications for serotonergic signal transduction. J Neurochem 73, 1711–1717.
Herrick-Davis K, Grinde E, Teitler M (2000). Inverse agonist activity of atypical antipsychotic drugs at human 5-hydroxytryptamine 2C receptors. Pharmacol Exp Ther 295, 226–232.
Higuchi M, Maas S, Single FN et al. (2000). Point mutation in an AMPA receptor gene rescues lethality in mice deficient in the RNA-editing enzyme ADAR2. Nature 406, 78–81.
Hollmann M, Hartley M, Heinemann S (1991). Ca2 permeability of KA-AMPA-gated glutamate receptor channels depends on subunit composition. Science 252, 851–853.
Hollmann M, Boulter J, Maron C et al. (1993). Zinc potentiates agonist-induced currents at certain splice variants of the NMDA receptor. Neuron 10, 943–954.
Hume RI, Dingledine R, Heinemann SF (1991). Identification of a site in glutamate receptor sub- units that controls calcium permeability. Science 253, 1028–1031.
Humphries C, Mortimer A, Hirsch S, de Belleroche J (1996). NMDA receptor mRNA correlation with antemortem cognitive impairment in schizophrenia. Neuroreport 7, 2051–2055.
Huntsman MM, Tran BV, Potkin SG, Bunney WE Jr., Jones EG (1998). Altered ratios of alterna- tively spliced long and short gamma2 subunit mRNAs of the gamma-aminobutyrate type A receptor in prefronal cortex of schizophrenics. Proc Natl Acad Sci, USA 8, 15066–10571.
Ikeda K, Nagasawa M, Mori H et al. (1992). Cloning and expression of the epsilon 4 subunit of the NMDA receptor channel. FEBS Lett 313, 34–38.
Ishii T, Moriyoshi K, Sugihara H et al. (1993). Molecular characterization of the family of the N– methyl--aspartate receptor subunits. J Biol Chem 268, 2836–2843.
Jackson IJ (1991). A reappraisal of non-consensus mRNA splice sites. Nucl Acids Res 19, 3795–3798.
150 C. M. Niswender and L. K. Hutchinson
Jia Z, Agopyan N, Miu P et al. (1996). Enhanced LTP in mice deficient in the AMPA receptor GluR2. Neuron 17, 945–956.
Keinanen K, Wisden W, Sommer B et al. (1990). A family of AMPA-selective glutamate recep- tors. Science 249, 556–560.
Kofuji P, Wang JB, Moss SJ, Huganir RL, Burt DR (1991). Generation of two forms of the gamma- aminobutyric acid A receptor gamma 2-subunit in mice by alternative splicing. J Neurochem 56, 713–715.
Kohler M, Burnashev N, Sakmann B, Seeburg PH (1993). Determinants of Ca2 permeability in both TM1 and TM2 of high affinity kainate receptor channels: diversity by RNA editing. Neuron 10, 491–500.
Kornau HC, Schenker LT, Kennedy MB, Seeburg PH (1995). Domain interaction between NMDA receptor subunits and the postsynaptic density protein PSD-95. Science 269, 1737–1740.
Krawczak M, Reiss J, Cooper DN (1992). The mutational spectrum of single base-pair substitu- tions in mRNA splice junctions of human genes: causes and consequences. Hum Genet 90, 41–54.
Krishek BJ, Xie X, Blackstone C, Huganir RL, Moss SJ, Smart TG (1994). Regulation of GABAA receptor function by protein kinase C phosphorylation. Neuron 12, 1081–1095.
Kutsuwada T, Kashiwabuchi N, Mori H et al. (1992). Molecular diversity of the NMDA receptor channel. Nature 358, 36–41.
Lai F, Chen CX, Carter KC, Nishikura K (1997). Editing of glutamate receptor B subunit ion channel RNAs by four alternatively spliced DRADA2 double-stranded RNA adenosine deami- nases. Mol Cell Biol 17, 2413–2424.
Le Corre S, Harper CG, Lopez P, Ward P, Catts S (2000). Increased levels of expression of an NMDARI splice variant in the superior temporal gyrus in schizophrenia. Neuroreport 11, 983–986.
Leysen JE, Gommeren W, Mertens J et al. (1993). Comparison of in vitro binding properties of a series of dopamine antagonists and agonists for cloned human dopamine D2S and D2L recep- tors and for D2 receptors in rat striatal and mesolimbic tissues, using [125I] 2 -iodospiperone. Psychopharmacology (Berl) 110, 27–36.
Liu K, Bergson C, Levenson R, Schmauss C (1994). On the origin of mRNA encoding the trun- cated dopamine D3-type receptor D3nf and detection of D3nf-like immunoreactivity in human brain. J Biol Chem 269, 29220–29226.
Liu Y, George CX, Patterson JB, Samuel CE (1997). Functionally distinct double-stranded RNA- binding domains associated with alternative splice site variants of the interferon-inducible double-stranded RNA-specific adenosine deaminase. J Biol Chem 272, 4419–4428.
Lomeli H, Mosbacher J, Melcher T et al. (1994). Control of kinetic properties of AMPA receptor channels by nuclear RNA editing. Science 266, 1709–1713.
Lorson CL, Hahnen E, Androphy EJ, Wirth B (1999). A single nucleotide in the SMN gene reg- ulates splicing and is responsible for spinal muscular atrophy. Proc Natl Acad Sci, USA 96, 6307–6311.
Maas S, Melcher T, Herb A et al. (1996). Structural requirements for RNA editing in glutamate receptor pre-mRNAs by recombinant double-stranded RNA adenosine deaminase. J Biol Chem 271, 12221–12226.
151 RNA processing regulation and interindividual variation
Malmberg A, Jackson DM, Eriksson A, Mohell N (1993). Unique binding characteristics of anti- psychotic agents interacting with human dopamine D2A, D2B, and D3 receptors. Mol Pharmacol 43, 749–754.
Mayer ML, Westbrook GL (1987). Permeation and block of N-methyl--aspartic acid receptor channels by divalent cations in mouse cultured central neurones. J Physiol 394, 501–527.
Mayer ML, MacDermott AB, Westbrook GL, Smith SJ, Barker JL (1987). Agonist- and voltage- gated calcium entry in cultured mouse spinal cord neurons under voltage clamp measured using arsenazo III. J Neurosci 7, 3230–3244.
Meguro H, Mori H, Araki K et al. (1992). Functional characterization of a heteromeric NMDA receptor channel expressed from cloned cDNAs. Nature 357, 70–74.
Melcher T, Maas S, Herb A, Sprengel R, Seeburg PH, Higuchi M (1996). A mammalian RNA editing enzyme. Nature 379, 460–464.
Minvielle-Sebastia L, Keller W (1999). mRNA polyadenylation and its coupling to other RNA processing reactions and to transcription. Curr Opin Cell Biol 11, 352–357.
Mishina M, Sakimura K, Mori H et al. (1991). A single amino acid residue determines the Ca2 permeability of AMPA-selective glutamate receptor channels. Biochem Biophys Res Commun 180, 813–821.
Mohn AR, Gainetdinov RR, Caron MG, Koller BH (1999). Mice with reduced NMDA receptor expression display behaviors related to schizophrenia. Cell 98, 427–436.
Monsma FJ Jr., McVittie LD, Gerfen CR, Mahan LC, Sibley DR (1989). Multiple D2 dopamine receptors produced by alternative RNA splicing. Nature 342, 926–929.
Monyer H, Sprengel R, Schoepfer R et al. (1992). Heteromeric NMDA receptors: molecular and functional distinction of subtypes. Science 256, 1217–1221.
Monyer H, Burnashev N, Laurie DJ, Sakmann B, Seeburg PH (1994). Developmental and regional expression in the rat brain and functional properties of four NMDA receptors. Neuron 12, 529–540.
Moriyoshi K, Masu M, Ishii T, Shigemoto R, Mizuno N, Nakanishi S (1991). Molecular cloning and characterization of the rat NMDA receptor. Nature 354, 31–37.
Moro O, Lameh J, Hogger P, Sadee W (1993). Hydrophobic amino acid in the i2 loop plays a key role in receptor–G protein coupling. J Biol Chem 268, 22273–22276.
Mount SM (1982). A catalogue of splice junction sequences. Nucl Acids Res 10, 459–72.
Murray HL, Jarrell KA (1999). Flipping the switch to an active spliceosome. Cell 96, 599–602. Nakanishi N, Shneider NA, Axel R (1990). A family of glutamate receptor genes: evidence for
the formation of heteromultimeric receptors with distinct channel properties. Neuron 5,
569–581.
Niswender CM, Sanders-Bush E, Emeson RB (1998). Identification and characterization of RNA
editing events within the 5-HT2C receptor. Ann NY Acad Sci 861, 38–48.
Niswender CM, Copeland SC, Herrick-Davis K, Emeson RB, Sanders-Bush E (1999). RNA editing of the human serotonin 5-hydroxytryptamine 2C receptor silences constitutive activ-
ity. J Biol Chem 274, 9472–9478.
Niswender CM, Herrick-Davis K, Gilley GE et al. (2001). RNA editing of the human serotonin
5-HT2C receptor: alterations in suicide and implications for serotonergic pharmacotherapy. Neuropsychopharm 24, 478–491.
152 C. M. Niswender and L. K. Hutchinson
Ohnuma T, Augood SJ, Arai H, McKenna PJ, Emson PC (1999). Measurement of GABAergic parameters in the prefrontal cortex in schizophrenia: focus on GABA content, GABA(A) receptor alpha-1 subunit messenger RNA and human GABA transporter-1 (HGAT-1) messen- ger RNA expression. Neuroscience 93, 441–448.
Okabe S, Miwa A, Okado H (1999). Alternative splicing of the C-terminal domain regulates cell surface expression of the NMDA receptor NR1 subunit. J Neurosci 19, 7781–7792.
Olney JW (1990). Excitotoxic amino acids and neuropsychiatric disorders. Annu Rev Pharmacol Toxicol 30, 47–71.
Olney JW, Newcomer JW, Farber NB (1999). NMDA receptor hypofunction model of schizo- phrenia. J Psychiatr Res 33, 523–533.
Patterson JB, Samuel CE (1995). Expression and regulation by interferon of a double-stranded- RNA-specific adenosine deaminase from human cells: evidence for two forms of the deami- nase. Mol Cell Biol 15, 5376–5388.
Philips AV, Cooper TA (2000). RNA processing and human disease. Cell Mol Life Sci 57, 235–49.
Pin JP, Joly C, Heinemann SF, Bockaert J (1994). Domains involved in the specificity of G protein activation in phospholipase C-coupled metabotropic glutamate receptors. EMBO J 13, 342–348.
Pritchett DB, Sontheimer H, Shivers BD et al. (1989). Importance of a novel GABAA receptor subunit for benzodiazepine pharmacology. Nature 338, 582–585.
Proudfoot N (2000). Connecting transcription to messenger RNA processing. Trends Biochem Sci 25, 290–293.
Rauser L, Savage JE, Meltzer HY, Roth BL (2001). Inverse agonist actions of typical and atypical antipsychotic drugs at the human 5-hydroxytryptamine 2C receptor. J Pharmacol Exp Ther 299, 83–89.
Reynolds GP, Czudek C, Andrews HB (1990). Deficit and hemispheric asymmetry of GABA uptake sites in the hippocampus in schizophrenia. Biol Psychiatry 27, 1038–1044.
Rueter SM, Emeson RB (1998). Adenosine-to-inosine conversion in mRNA. In Grosjean H, Benne R, eds. Modification and Editing of RNA. Washington, DC: ASM Press, pp. 343–361. Rueter SM, Dawson TR, Emeson RB (1999). Regulation of alternative splicing by RNA editing.
Nature 399, 75–80.
Sakimura K, Bujo H, Kushiya E et al. (1990). Functional expression from cloned cDNAs of glu-
tamate receptor species responsive to kainate and quisqualate. FEBS Lett 272, 73–80. Schmauss C (1996). Enhanced cleavage of an atypical intron of dopamine D3-receptor pre-
mRNA in chronic schizophrenia. J Neurosci 16, 7902–9790.
Schmauss C, Haroutunian V, Davis KL, Davidson M (1993). Selective loss of dopamine D3-type
receptor mRNA expression in parietal and motor cortices of patients with chronic schizophre-
nia. Proc Natl Acad Sci, USA 90, 8942–8946.
Sherman AD, Hegwood TS, Baruah S, Waziri R (1991). Deficient NMDA-mediated glutamate
release from synaptosomes of schizophrenics. Biol Psychiatry 30, 1191–1198.
Shiga N, Takeshima Y, Sakamoto H et al. (1997). Disruption of the splicing enhancer sequence within exon 27 of the dystrophin gene by a nonsense mutation induces partial skipping of the
exon and is responsible for Becker muscular dystrophy. J Clin Invest 100, 2204–2210.
153 RNA processing regulation and interindividual variation
Simpson L, Emeson RB (1996). RNA editing. Annu Rev Neurosci 19, 27–52.
Simpson MD, Slater P, Deakin JF, Royston MC, Skan WJ (1989). Reduced GABA uptake sites in
the temporal lobe in schizophrenia. Neurosci Lett 107, 211–215.
Smith HC, Gott JM, Hanson MR (1997). A guide to RNA editing. RNA 3, 1105–1123.
Sodhi MS, Burnet PW, Makoff AJ, Kerwin RW, Harrison PJ (2001). RNA editing of the 5-HT(2C)
receptor is reduced in schizophrenia. Mol Psychiatry 6, 373–379.
Sokolov BP (1998). Expression of NMDAR1, GluR1, GluR7, and KA1 glutamate receptor
mRNAs is decreased in frontal cortex of “neuroleptic-free” schizophrenics: evidence on rever-
sible up-regulation by typical neuroleptics. J Neurochem 71, 2454–2464.
Sommer B, Kohler M, Sprengel R, Seeburg PH (1991). RNA editing in brain controls a determi-
nant of ion flow in glutamate-gated channels. Cell 67, 11–19.
Sprengel R, Higuchi M, Monyer H, Seeburg PH (1999). Glutamate receptor channels: a possible
link between RNA editing in the brain and epilepsy. Adv Neurol 79, 525–534.
Squires RF, Lajtha A, Saederup E, Palkovits M (1993). Reduced [3H]flunitrazepam binding in cin- gulate cortex and hippocampus of postmortem schizophrenic brains: is selective loss of glu-
tamatergic neurons associated with major psychoses? Neurochem Res 18, 219–223.
Usiello A, Baik J-H, Rouge-Pont F et al. (2000). Distinct functions of the two isoforms of dopa-
mine D2 receptors. Nature 408, 199–203.
Valentine CR (1998). The association of nonsense codons with exon skipping. Mutat Res 411,
87–117.
Verdoorn TA, Burnashev N, Monyer H, Seeburg PH, Sakmann B (1991). Structural deter-
minants of ion flow through recombinant glutamate receptor channels. Science 252,
1715–1718.
Verrall S, Ishii M, Chen M, Wang L, Tram T, Coughlin SR (1997). The thrombin receptor second
cytoplasmic loop confers coupling to Gq-like G proteins in chimeric receptors. Additional evi- dence for a common transmembrane signaling and G protein coupling mechanism in G protein-coupled receptors. J Biol Chem 272, 6898–6902.
von Blankenfeld G, Ymer S, Pritchett DB et al. (1990). Differential benzodiazepine pharmacol- ogy of mammalian recombinant GABAA receptors. Neurosci Lett 115, 269–273.
Wafford KA, Burnett DM, Leidenheimer NJ et al. (1991). Ethanol sensitivity of the GABAA receptor expressed in Xenopus oocytes requires 8 amino acids contained in the gamma 2L subunit. Neuron 7, 27–33.
Wagner RW, Yoo C, Wrabetz L et al. (1990). Double-stranded RNA unwinding and modify- ing activity is detected ubiquitously in primary tissues and cell lines. Mol Cell Biol 10, 5586–5590.
Wang Q, O’Brian PJ, Chen CX, Cho DS, Murray JM, Nishikura K (2000a). Altered G protein- coupling functions of RNA editing isoform and splicing variant serotonin 2C receptors. J Neurochem 74, 1290–1300.
Wang Y, Xu R, Sasaoka T, Tonegawa S, Kung M-P, Sankoorikal EB (2000b). Dopamine D2 long receptor-deficient mice display alterations in striatum-dependent function. J Neurosci 20, 8305–8314.
Weiner DM, Burstein ES, Nash N et al. (2001). 5-Hydroxytryptamine 2A receptor inverse ago- nists as antipsychotics. J Pharmacol Exp Ther 299, 268–276.
154 C. M. Niswender and L. K. Hutchinson
Westphal RS, Sanders-Bush E (1994). Reciprocal binding properties of 5-hydroxytryptamine type 2C receptor agonists and inverse agonists. Mol Pharmacol 46, 937–942.
Whiting P, McKernan RM, Iversen LL (1990). Another mechanism for creating diversity in gamma-aminobutyrate type A receptors: RNA splicing directs expression of two forms of gamma 2 phosphorylation site. Proc Natl Acad Sci, USA 87, 9966–9970.
Wong SK, Parker EM, Ross EM (1990). Chimeric muscarinic cholinergic: beta-adrenergic recep- tors that activate Gs in response to muscarinic agonists. J Biol Chem 265, 6219–6224.
Yamanaka S, Balestra ME, Ferrell LD et al. (1995). Apolipoprotein B mRNA-editing protein induces hepatocellular carcinoma and dysplasia in transgenic animals. Proc Natl Acad Sci, USA 92, 8483–8487.
Part IV
Pharmacokinetics


8
Pharmacogenetics of psychotropic drug metabolism
Vural Ozdemir1, 2, Angela D.M. Kashuba3, Vincenzo S. Basile1, and James L. Kennedy1
1 Centre for Addiction and Mental Health, University of Toronto, Canada 2 Department of Pharmacology, University of Toronto, Canada
3 School of Pharmacy, University of North Carolina, Chapel Hill, USA
OVERVIEW
Drug metabolism is a critical determinant of therapeutic and adverse effects of many psycho- tropic drugs. Research since the 1960s has firmly established that genetic factors play a prominent role in marked person-to-person variability in drug metabolism. Pharmacogenetics is the study of the hereditary basis of individual differences in drug response. The present chapter reviews the basic concepts and definitions pertinent to phar- macogenetics of psychotropic drug metabolism from a clinical psychiatry perspective. The focus of pharmacogenetic investigations has traditionally been unusual and extreme drug metabolism phenotypes resulting from a single gene effect. To this end, we discuss the CYP2D6 genetic polymorphism as a classic example of a monogenic variation in drug metabolism and as a high-affinity and low-capacity elimination route. This is contrasted with CYP1A2, a nonpolymorphic variation in drug metabolism under polygenic control. In addi- tion to genetic contribution to interindividual differences in drug metabolism, we review various sources of intraindividual variations and gene–environment interactions of relevance for psychopharmacology. For example, we describe the extent and mechanistic basis of drug and nutraceutical (e.g., St. John’s wort) interactions and disease influences (e.g., human immunodeficiency virus (HIV) infection) on drug metabolism. Lastly, we highlight some of the future research directions in psychotropic drug metabolism. In particular, we emphasize the need to evaluate genetic variability in drug metabolism in conjunction with other genes encoding drug transporters, receptors and ion channels, which can all influence an individ- ual’s risk for adverse drug reactions or therapeutic failure. It is anticipated that these phar- macogenetic inquiries in the first decade of the twenty-first century will provide a framework for the rational choice and dosage of psychotropic drugs.
Introduction
Rational drug treatment in psychiatry has a relatively brief history. The introduc- tion of lithium in 1949 and chlorpromazine in the 1950s represents the first

157

158 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
examples where pharmacological adjuncts were used to alleviate mental health dis- orders. At that time, it also became evident that therapeutic doses of psychotropic medications vary markedly from person to person, with some patients failing to respond despite treatment with high dosages. Early on, it was proposed that genetic factors were likely to explain differences between individuals and populations in response to drugs and other foreign chemicals (Motulsky, 1957; Kalow, 1962). The term pharmacogenetics was coined to establish a new medical subspecialty devoted to study of the hereditary basis of variability in drug effects (Vogel, 1959; Kalow, 1962). Subsequently, a series of seminal twin studies in the 1970s firmly established that heredity is indeed a key determinant of a drug’s pharmacological effects (Endrenyi et al., 1976; Vesell, 1978). More recently, advances in molecular genetics put on record numerous examples of genetic polymorphisms in drug-metabolizing enzymes (DMEs), with important therapeutic and toxicological ramifications (Ingelman-Sundberg et al., 1999; Kalow, 1999).
Among many factors that may influence drug response, hepatic function and drug metabolism are of great importance for nearly all medications used in psychi- atry (Bertilsson and Dahl, 1996; Brøsen, 1996; Cohen and DeVane, 1996). After more than four decades of research, it is interesting to point out that relatively little effort has been made to translate the established knowledge-base in pharmacoge- netics to discrete treatment guidelines for rational choice of drugs and dose- titration regimens. The lack of recognition of functional variability in DMEs and the “one-dose-fits-all” approach to pharmacotherapy are important shortcomings of the current patient care in psychiatry (Chou et al., 2000). For example, the inci- dence of serious and fatal adverse drug reactions caused by such nonoptimal treat- ment regimens in North America ranks between the fourth to sixth leading cause of death, ahead of pneumonia and diabetes (Lazarou et al., 1998).
The purpose of this chapter is to provide an overview of the basic concepts per- tinent to pharmacogenetics of psychotropic drug metabolism, with an emphasis on clinical applications and future research directions. In addition to differences between individuals, we also discuss intraindividual (within person) variations in drug metabolism capacity and gene–environment interactions. An adequate knowledge of these factors is important for a balanced interpretation of findings from pharmacogenetic studies and individualized therapeutics in psychiatry.
Phase I and phase II drug metabolism
Many of the psychotropic drugs are lipophilic in nature. Hence, they usually undergo oxidative biotransformation to form more water-soluble metabolites to facilitate their disposition from the body. Drug metabolism is generally divided into two phases. Phase I reactions involve oxidative, reductive, and hydrolytic reac- tions that unmask or introduce a functional group (e.g., a hydroxyl moiety) in the
159 Pharmacogenetics of psychotropic drug metabolism
parent compound. This results in an increase in polarity of medications. Phase II reactions involve conjugation (e.g., with glucuronic acid, glutathione, or acetate) of the metabolite produced in phase I reactions, or the parent compound, to more hydrophilic and usually less-toxic metabolites. Phase I reactions are mediated prin- cipally by the cytochrome P-450 (CYP) enzymes, which are mostly found attached to the smooth endoplasmic reticulum of hepatocytes. By contrast, most phase II metabolism takes place in the cytosol (e.g., by glutathione S-transferases) with certain exceptions, such as glucuronidation by membrane-bound microsomal UDP-glucuronosyl transferases (Eaton and Bammler, 1999). Ideally, phase I and phase II biotransformation should be evaluated in concert for a thorough under- standing of drug disposition and its clinical significance. Notably, however, the majority of studies on psychotropic drug metabolism have been conducted with a focus on CYP enzymes.
Physicochemical properties and nomenclature for cytochrome P-450s
CYP enzymes are heme-containing membrane-bound proteins (Omura and Sato, 1964a). The heme moiety consists of a protoporphyrin IX molecule and an iron atom and serves as the prosthetic group for the CYP apoprotein. An important characteristic of the CYP enzymes is that they exhibit a unique spectral feature: when the heme iron is in its reduced ferrous form (Fe2 ) and bound to carbon monoxide, a maximum absorption is observed at a wavelength of 450nm. This spectral feature is the basis for the CYP name and is also used to measure the total CYP content of a given tissue in solution (Omura and Sato, 1964a, b).
The current nomenclature for CYPs is based on the differences in amino acid sequence homology (Nelson et al., 1996). A CYP gene is named starting with the italicized CYP, followed by an Arabic number signifying the gene family, an upper case letter signifying the gene subfamily, and another Arabic number for the indi- vidual gene (e.g., CYP2D6). The same letters and numbering in nonitalicized form are used to name the corresponding gene products (i.e., mRNA, cDNA and protein). A CYP enzyme from one gene family has 40% homology with the amino acid sequence of a CYP enzyme from any other family. In other words, in a given family, there is 40% homology in amino acid sequences of its members. However, a point to keep in mind is that belonging to the same family or subfamily does not necessarily imply similarities in substrate preference, catalytic functions, or regulation of gene expression. Further information on CYP nomen- clature and new CYP alleles can be obtained from the worldwide web (http://www.imm.ki.se/CYPalleles/).
The extent of metabolism of a drug by a given CYP isoform is determined by both the affinity of the substrate–enzyme complex and the relative abundance of the particular CYP protein in relation to the total CYP content (Eaton, 2000).
 
160 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
Table 8.1. Relative content of individual CYP isoforms determined immunochemically in relation to total CYP content in the human livera

CYP isoform Percentage of total P-450 content
Coefficient of variation (%)b
49
80
150
37
87
44
36
values from 60 samples.
[(SD/mean) 100] is a measure of interindividual variability in

CYP1A2 12.7 CYP2A6 4.0 CYP2B6 0.2 CYP2C 18.2 CYP2D6 1.5 CYP2E1 6.6 CYP3A 28.8
Notes:
a Data are presented as mean b The coefficient of variation
CYP content.
Source: Shimada et al., 1994.

Approximately 72% of the total CYP content in the human liver can be attributed to seven subfamilies (Table 8.1). Although the CYP3A and CYP2C subfamilies appear to account for almost half of the total CYP content, it is notable that a sig- nificant portion of the total CYP content (i.e., 28%) is attributable to other uniden- tified enzymes in the CYP superfamily (Shimada et al., 1994). Since these data reflect the CYP content, that is the amount of enzyme protein, caution is needed when evaluating their functional (catalytic) and clinical importance.
Genetic variability in drug metabolism
CYP2D6 polymorphism: a classical example of monogenic control in drug metabolism
The CYP2D6 genetic polymorphism (debrisoquine/sparteine) was discovered in late 1970s and represents one of the most intensively studied monogenic variations in drug metabolism (Nebert, 1999; Sjöqvist, 1999). Approximately 7% of Caucasians are poor metabolizers (PMs) of CYP2D6 substrates while the rest are considered as extensive metabolizers (EMs) (Bertilsson et al., 1992). CYP2D6 enzyme has particular significance for clinical psychiatry as it is involved in clear- ance of many psychotropic medications such as tricyclic antidepressants (e.g., desipramine), selective serotonin reuptake inhibitors (SSRIs, e.g., paroxetine), clas- sical antipsychotics (e.g., perphenazine), some of the atypical antipsychotic agents (e.g., risperidone), drugs of abuse, and codeine (Bertilsson and Dahl, 1996; Brøsen, 1996).

In vivo, CYP2D6 activity can vary up to 1000-fold in the population (Bertilsson
161 Pharmacogenetics of psychotropic drug metabolism
((a) NORTRIPTYLINE 60
NUMBER OF 50      FUNCTIONAL
CYP2D6 GENES 40  0
(b) 10-HYDROXYNORTRIPTYLINE 250
                             
30 20 10
1
2
200
150
100
NUMBER OF FUNCTIONAL CYP2D6 GENES
13
3
                            
3 502 1
                                          
13 00
         
Fig. 8.1.
0
0 24 48 72 0 24 48
Time (hours)
Average plasma concentrations of nortriptyline (a) and 10-hydroxynortriptyline (b) after a 25mg single oral dose in Caucasian healthy volunteers with 0, 1, 2, 3, and 13 functional copies of CYP2D6. Note that the concentration of nortriptyline and its metabolite 10- hydroxynortriptyline are inversely affected by the number of functional CYP2D6 copies. (Reprinted with permission from Dalén et al., 1998.)
et al., 1992; Meyer and Zanger, 1997). This is a consequence of the presence of more than 50 CYP2D6 alleles which encode enzymes with inactive, decreased, increased or normal catalytic function. Genotype–phenotype correlation studies indicate that the number of functional CYP2D6 predicts drug and metabolite concentra- tions in the plasma (Fig. 8.1) (Dalén et al., 1998). From a clinical perspective, as a general rule, PMs are at risk for drug toxicity during treatment with standard doses (Pollock et al., 1995). Hence, PMs may present with poor compliance early in the course of drug treatment. For prodrugs that require activation (e.g., codeine), PM phenotype may lead to treatment resistance (Sindrup and Brøsen, 1995). At the other extreme of CYP2D6 activity, in ultrarapid metabolizers with multiple copies of CYP2D6, conventional dose titration regimens may cause delays in therapeutic response and lead to prolonged inpatient hospital stay and institutionalization. CYP2D6 genotyping may be clinically invaluable to differentiate ultrarapid metabolizers with unusually low plasma drug concentrations from patients who do not comply with drug treatment (Bertilsson et al., 1985; Dalén et al., 1998). Interestingly, approximately 29% of Ethiopians carry duplicated or multidupli- cated functional CYP2D6 alleles (Aklillu et al., 1996). Identification of such popu- lations with ultrarapid drug metabolism is important as it may provide a mechanistic basis for treatment resistance to some psychotropic drugs.
PLASMA CONCENTRATION (nmol/l) PER 25 mg NT DOSE
162 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
A notable and well-established interethnic difference in CYP2D6 expression occurs between Caucasians and Asians (Bertilsson et al., 1992; Lin et al., 1996). The prevalence of PM phenotype is only 1% in Asians. Another frequently overlooked difference is that the distribution of CYP2D6 activity is significantly shifted towards lower values in Asian EMs. The molecular genetic basis of a slower CYP2D6 activity in Asian EMs was shown to be the result of a cytosine-to-thymine change at position 188 (C188T) in exon 1, leading to Pro34Ser amino acid substi- tution in a highly conserved region (Pro-Pro-Gly-Pro) of the CYP2D6 enzyme (Johansson et al., 1994; Bertilsson, 1995). Importantly, this allele (CYP2D6*10) has a high frequency in Asians (51% in Chinese) and causes a 10-fold decrease in cat- alytic activity in vivo (Johansson et al., 1994). Therefore, interindividual differ- ences in therapeutic/adverse effects of psychotropics in Asian populations may be explained in part by the presence of CYP2D6*10 allele. Further information on interethnic differences in CYP2D6 expression is available elsewhere (Bertilsson, 1995).
CYP2D6 with identical pharmacological and molecular properties was also identified in the brain (Britto and Wedlund, 1992). Hence, it is conceivable that CYP2D6 may contribute to local clearance of psychotropics at the site of action. CYP2D6 in the brain is functionally associated with the dopamine transporter and shares similarities in substrates and inhibitors (e.g., d-amphetamine), suggesting a role in dopaminergic neurotransmission (Niznik et al., 1990). Differences in per- sonality traits between EMs and PMs were noted in both Swedish and Spanish healthy Caucasian subjects, further suggesting that there may be an endogenous substrate for CYP2D6 (Llerena et al., 1993). At present, little is known on the bio- logical significance and regulation of drug metabolism in the brain, but this is a topic of considerable relevance for personalized therapeutics in psychiatry (Miksys et al., 2000).
Although the CYP2D6 polymorphism has been known since the 1980s, routine application of this information in clinical practice was hampered, in part, by the lack of cost-effectiveness analyses for pharmacogenetic testing. Recently, Chou et al. (2000) estimated that the annual cost of treating patients at extremes for CYP2D6 activity (PMs and ultrarapid metabolizers) is on average US$4000–$6000 greater than for the rest of the population. Moreover, the total duration of hospi- tal stay appears to be more pronounced in PMs, presumably because of a higher incidence of adverse drug events (Chou et al., 2000). In the near future, further pharmacoeconomic evaluations of genetic variability in drug metabolism and its impact on health outcomes may convince the insurers and other third-party payers to incorporate pharmacogenetic testing into the list of routinely available diagnos- tic tests. Examples of genetic polymorphisms in other DMEs and their major variant alleles are presented in Table 8.2.
    
Table 8.2. Human polymorphic cytochrome P-450 enzymes and the global distribution of their major variant alleles
Allele frequencies (%)
Enzyme CYP2A6
Major variant alleles
Mutation
Consequences for enzyme function
Caucasians Asians
Black Ethiopians and Africans Saudi Arabians
CYP2C9
CYP2A6*2 CYP2A6*del CYP2C9*2
Leu160His Gene deletion Arg144Cys
Inactive enzyme
No enzyme
Reduced affinity for P-450 oxidoreductase
Altered substrate specificity Inactive enzyme
Inactive enzyme
Increased enzyme activity Inactive enzyme
No enzyme
Unstable enzyme
Reduced affinity for substrates
1–3 0 115 8–13 0
ND ND ND ND ND ND
CYP2C19 CYP2D6
CYP2C9*3 CYP2C19*2 CYP2C19*3 CYP2D6*2xN CYP2D6*4 CYP2D6*5 CYP2D6*10 CYP2D6*17
Ile359Leu
Aberrant splice site
Premature stop codon
Gene duplication or multiduplication Defective splicing
Gene deletion
Pro34Ser, Ser486Thr
Thr107Ile, Arg296Cys, Ser486Thr
6–9 13
0
1–5 12–21 2–7 1–2
2–3 23–32 6–10
ND ND 13 14–15 ND 0–2
Notes:
ND, not determined.
Source: Reprinted with permission from Ingelman-Sundberg et al., 1999.
0 ND
34 3–9
51
0–2 1
6
2 10–16 2 1–4 4 1–3 6 3–9
164 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
Nonpolymorphic drug metabolism with polygenic control: an example with CYP1A2
Historically, CYP2D6 and other genetically polymorphic DMEs have received much research attention in psychopharmacology. Although CYP2D6 is clearly important for the metabolism of many psychotropic drugs, it accounts for only 2% of the total CYP content in the human liver. Consequently, CYP2D6 is considered a “high affinity–low capacity” metabolic clearance pathway (Murphy et al., 2000). Consistent with this, the contribution of CYP2D6 to disposition of substrates is comparatively reduced during multiple- versus single-dose drug administration (Sindrup et al., 1992). Therefore, it is likely that metabolism by pathways other than CYP2D6 may also contribute to disposition of psychotropic drugs, especially under steady-state conditions during multiple-dose treatment. A related point to keep in mind is that CYP2D6 is practically not inducible by environmental factors such as smoking and cruciferous vegetable consumption, while such factors may upregu- late the expression of certain nonpolymorphic forms of CYPs (e.g., CYP1A2) and thereby increase their contribution to drug clearance.
There is a growing recognition that metabolic routes without a clear-cut poly- morphic pattern in distribution of their activity may be subject to “polygenic” reg- ulation (Casley et al., 1999; Kurth, 2000). For example, CYP1A2 enzyme is constitutively expressed in human liver and accounts, on average, for 13% of the total CYP content (Shimada et al., 1994). CYP1A2 activity varies greatly (up to 130- fold) among individuals in many populations and contributes to disposition of several important psychotropic medications, including clozapine, olanzapine and tacrine (Bertilsson et al., 1994; Eaton et al., 1995; Ring et al., 1996; Fontana et al., 1998). Studies in monozygotic and dizygotic twins indicate that genetic factors play a prominent role in regulation of CYP1A2 activity (h1 0.83, with 3-methylxanthine formation from theophylline) (Miller et al., 1984). Using a genome-wide interval mapping approach in a mouse model, Casley et al. (1999) recently provided intriguing data that three independent loci on chromosomes 1, 4, and 9 (the last colocalizes with the murine CYP1A2 locus) explain 63.2% of var- iability in caffeine N-3 demethylation (an index of CYP1A2 activity). This is a notable landmark study because it is the first application of a genome-wide inter- val mapping approach to characterize polygenic variation in drug metabolism (Casley et al., 1999). In the near future, it is likely that similar genome-wide searches will provide further insights into polygenic control of drug metabolism.
Genetic contribution to interindividual variability in drug metabolism
Relevance for the design and interpretation of genetic studies of complex diseases
An important corollary of genetic variability in DMEs is that it may potentially lead to a selection bias in candidate gene studies of complex diseases such as

165 Pharmacogenetics of psychotropic drug metabolism
schizophrenia. For example, patients who are at risk for drug toxicity (e.g., PMs of CYP2D6) may use inpatient services more commonly and, thus, may have a greater chance of being included in research studies (see discussion by Steen et al. (1997)). In other words, some of the false-positive as well as false-negative genetic studies on complex diseases may be attributed to such population stratification secondary to pharmacogenetic reasons – and not a real disease-causing genetic factor. In future pharmacogenetic investigations, it is advisable, therefore, to stratify and match control and patient populations not only with respect to known confound- ers such as gender and ethnicity, but also for inpatient and outpatient status. Alternatively, the robustness of conclusions concerning the genetic basis of psychi- atric diseases may be ascertained with use of “inpatient status” as another poten- tially relevant covariate during data analysis.
Intraindividual variations in drug metabolism and gene–environment interactions
Traditionally, genetic factors have mainly been evaluated in relation to interindi- vidual (between person) variations in psychotropic drug metabolism and its clini- cal significance. Relatively less attention has been given to intraindividual (within person) differences in DME activity, but this variability is also important to con- sider, particularly for medications that have narrow therapeutic ranges (a high risk of toxicity or therapeutic failure after slight changes in plasma concentrations) and when using compounds for chronic conditions, or for extended periods of time. In these situations, it is important to recognize whether the required dose of a medi- cation administered today will be the same required by the patient 6 months, 1 year, or 5 years from now (e.g., consider the treatment of patients for chronic major depression or schizophrenia). Moreover, intraindividual variability may potentially mimic some of the hereditary drug metabolism phenotypes (Preskorn, 1997; Alfaro et al., 1999). For example, during the course of treatment with potent CYP2D6 inhibitors (e.g., paroxetine, quinidine), EMs may clinically appear as PMs (Alfaro et al., 2000). Thus, drug-induced or physiological variations in drug metab- olism may confound the interpretation and conclusions of pharmacogenetic studies (Meyer et al., 1996).
Intraindividual variability in DME activity can be quantified with phenotyping. Although more cumbersome than determining genotype, phenotyping may be clinically more useful as it describes an individual’s enzyme activity at any one point in time and accounts for the combined inductive or inhibitory effects of genetic, environmental, and physiological influences. This method involves the oral or intravenous administration of a carefully selected “biomarker compound,” and subsequently examining urine, plasma, or saliva concentrations of parent drug and/or metabolites (Carrillo et al., 2000; Streetman et al., 2000a).
Short-term (2 to 6 months) intraindividual variability data in healthy volunteers
166 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
exist for a number of DMEs (Kashuba et al., 1998a–c; Labbe et al., 2000). When repeated phenotyping measures are performed in healthy volunteers, and coeffi- cients of variation are calculated, intraindividual variability in CYP1A2 phenotype ranges from approximately 5 to 50%, CYP2D6 phenotype varies from 12 to 140%, CYP3A4 phenotype varies from 5 to 21%, N-acetyltransferase 2 (NAT2) phenotype varies from 2 to 27%, and xanthine oxidase phenotype varies from 2 to 13%. A pro- portion of this intraindividual variability may be ascribed to experimental varia- tion in sample collection and assay techniques, and the variability in other DMEs that may contribute to the biomarker compounds’ metabolism. In general, short- term temporal variability in DME activity in diet-restricted, drug-free healthy vol- unteers appears to be relatively stable. However, environmental and physiological influences on drug–drug interactions in patient populations can all significantly affect intraindividual variability in DME activity.
The following is a synopsis of the frequently encountered environmental and physiological influences that may modify genetically determined variations in drug metabolism. Further information on gene–environment interactions of relevance for drug metabolism are available elsewhere (Eaton and Klaassen, 1996; Costa, 2000).
Dietary influence on intraindividual variability
Certain dietary and environmental factors can significantly alter intestinal and hepatic cytochrome P-450 activity (Walter-Sack and Klotz, 1996). Studies clearly demonstrate that consumption of charcoal-broiled and smoked foods (containing polycyclic aromatic hydrocarbons) can increase the activity of the CYP1A subfam- ily of isozymes in the intestinal epithelium and hepatocytes (Conney et al., 1976; Kappas et al., 1978; Kall and Clausen, 1995). Cruciferous vegetables such as brus- sels sprouts, cabbage, broccoli, cauliflower, kale, spinach, and watercress can also alter the activity of selected CYP isozymes: indole-containing vegetables (cabbage, cauliflower) upregulate CYP1A (Pantuck et al., 1989), and isothiocyanate- containing vegetables (watercress) can inhibit CYP2E1 (Kim and Wilkinson, 1996). Organosulfur compounds in garlic, like diallylsulfide, have been found to be inhib- itors of CYP2E1 and inducers of the CYP1A, CYP3A, and phase II biotransforma- tion enzymes (Wilkinson, 1997).
Several components of grapefruit juice acting on enterocytes can increase the bioavailability of CYP3A (and in some cases, CYP1A) substrates by four- to five- fold (Bailey et al., 1994; Dresser et al., 2000). A number of phytochemicals found in grapefruit juice that may be responsible for CYP3A (and possibly CYP1A) inhi- bition and degradation have been investigated. Although the exact compound or compounds is/are currently unknown, ingestion of grapefruit juice can lead to large apparent intraindividual variability, primarily in the pharmacokinetics of
167 Pharmacogenetics of psychotropic drug metabolism
CYP3A compounds. Additionally, certain vitamins and spices have been implicated in altering DME activity (Wilkinson, 1997).
Although the diet contains many more potentially active compounds than have been formally investigated, it is important to recognize that the body is exposed to a constantly changing variety of chemicals that are capable of modulating DMEs in both the intestine and the liver. Many of these may have opposing effects, resulting in complex and often unpredictable interactions, which can contribute to intrain- dividual variability in drug metabolism and response.
Drug–drug interactions and nutraceutical influences on drug metabolism
Large intraindividual variability in drug metabolism can also be caused by con- comitant medication therapy. A wide variety of medications are known to induce or inhibit DME activity profoundly, such as the rifamycins, anticonvulsants, macrolide antibiotics, azole antifungal agents, nefazodone, and certain selective serotonin reuptake inhibitors (SSRIs). The effects of medications on DME activ- ities have been extensively reviewed elsewhere (Lin and Lu, 1998; Fang and Gorrow, 1999; Tanaka and Hisawa, 1999; Flockhart and Oesterheld, 2000). Patients initiat- ing or terminating a medication or compound that can profoundly influence DME activity may experience altered responses to concomitant medications.
For centuries, herbs and other dietary compounds have played a significant role in the treatment of disease in many traditional cultures. In developed countries, there is now increasing interest in the use of the active constituents of these prod- ucts for medical benefit. Along with this increased use has come the knowledge that these are not innocuous compounds. The detrimental effects of extracts of St. John’s wort (Hypericum perforatum), used for mild to moderate states of depression, on drug availability have recently been described. Kerb et al. (1997) first demonstrated that taking 300mg of St. John’s wort three times a day for 14 days significantly enhanced urinary 6-hydroxycortisol excretion, a marker of CYP3A activity. After 14 days of therapy with St. John’s wort, the protease inhib- itor indinavir had a mean 60% decrease in exposure (area under the concentration-versus-time curve: AUC) and an 80% decrease in trough concen- trations (Cmin) (Piscitelli et al., 2000), indicative of large CYP3A induction effects on hepatic and intestinal activity. In women taking St. John’s wort and oral contraceptives, breakthrough bleeding with potential loss of contraceptive efficacy has also been reported (Ernst, 1999). Another recent report demonstrated that ciclosporin (cyclosporine) concentrations decreased in one kidney transplant re- cipient sporadically taking St. John’s wort, necessitating a 46% increase in dose (Mai et al., 2000).
Mechanisms of these nutraceutical effects are under investigation. Wentworth et al. (2000) found that St. John’s wort enhanced the transcriptional activity of the
168 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
steroid X receptor, a member of the nuclear receptor superfamily that is activated by drugs such as rifampin, which are potent inducers of hepatic cytochrome P-450 CYP3A enzyme activity. St. John’s wort also appears to alter P-glycoprotein (P-gp) activity. P-gp is a transmembrane efflux pump that removes drugs from the cell and deposits them into the extracellular space (Kashuba and Bertino, 2000; von Moltke and Greenblatt, 2000). It has been suggested that CYP3A and P-gp might play com- plementary roles in drug disposition by biotransformation and antitransport. This is especially pertinent in enterocytes, where CYP3A and P-gp can act synergistically to decrease drug bioavailability (Wacher et al., 1995). Currently, little is known about the significance of P-gp-mediated transport of psychotropics during distri- bution. Johne et al. (1999) administered 300mg dried Hypericum extract three times per day, along with digoxin (a P-gp substrate) 0.25mg once per day in five healthy volunteers. After 10 days of treatment with St. John’s wort, digoxin expo- sure (AUC) decreased 25% and trough concentrations decreased 20–30%. The authors postulated that P-gp activity, transporting digoxin from enterocytes into the intestinal tract, might be activated upon exposure to St. John’s wort.
The effects of St. John’s wort on both CYP3A and P-gp activity are not com- pletely unexpected. It has been determined that most, but not all (e.g., midazolam), compounds that are metabolized by CYP3A are substrates for P-gp. Most recently, Piscitelli et al. (2001) have demonstrated that, in the presence of garlic supplemen- tation taken twice daily for 3 weeks, steady-state AUC and trough concentrations of saquinavir (a CYP3A and P-gp substrate) both decrease approximately by 50%. This effect appears to be exceedingly prolonged, as these parameters returned to only 60–70% of baseline values 10 days after stopping the garlic supplementation. Much work is still needed to examine the effects of nutraceuticals on psychotropic drug disposition, and how significantly they contribute to intraindividual variabil- ity in DME activity and therapeutic response.
Influence of aging on drug metabolism
Beginning in the third decade of life, both liver blood flow and liver volume decline linearly over time. By the time an individual reaches 80–90 years of age, these values are approximately 20–40% less (Durnas et al., 1990). In rodents and other animal models, aging is often associated with impaired drug metabolism; however, extrap- olation of these findings to humans is difficult (Schmucker, 1989). In addition, problems such as large interindividual variability and small sample sizes hinder the extrapolation of in vitro human data to the clinical situation (Wilkinson, 1997).
Nonetheless, the available in vitro data suggest that age-related effects on DME activity are modest. Generally, it appears that CYP1A2, CYP2C9/10, CYP2C18/19, and CYP3A4/5 are moderately reduced in the elderly. There are no data on intesti- nal CYP activity changes in the elderly (Kinirons and Crome, 1997). Although the
169 Pharmacogenetics of psychotropic drug metabolism
extent of change is unpredictable with respect to specific drug and particular indi- vidual, it appears to be greatest with drugs exhibiting significant ( 80%) first-pass effects in young subjects. Fortunately, very few currently marketed drugs possess this characteristic.
Influence of disease on variability in drug metabolism
In addition to genetics, diet, concomitant medications/nutraceuticals and physio- logic effects such as aging, the presence of physical diseases can also be a major determinant of variability in drug disposition and clinical response. However, non- psychiatric disease states are frequently overlooked in patients with mental health problems and during their treatment with psychotropic drugs.
As early as the 1960s, acute inflammation and infection were demonstrated to affect the metabolism of drugs and toxins in animals, thereby modulating phar- macological and toxicological effects (Wooles and Borzelleca, 1966; Renton and Mannering, 1976). The earliest report of infection altering human DME activity occurred a decade later, with quinine concentrations consistently elevated in sub- jects experimentally infected with Plasmodium falciparum malaria (Trenholme et al., 1976). Since that time, numerous reports have described alterations in drug metabolism with viral and bacterial infections (Brockmeyer et al., 1998; Shedlofsky et al., 1994), and with traumatic events such as surgery and bone marrow trans- plantation (Gidal et al., 1996). The effects of inflammation and infection on P-450 activity are ascribed to stimulation of the cellular immune response (Renton, 2000). Although many different mediators may be involved, there has been partic- ular focus on the major proinflammatory cytokines interleukins 1 and 6 and tumor necrosis factor (Watkins et al., 1995; Haas, 2000).
Very little information exists concerning the influence of chronic infection on DME activity. Recently, the effects of HIV infection on drug metabolism have been investigated (Lee et al., 1993; Gotzkowsky et al., 2000; O’Neil et al., 2000). The most comprehensive data were reported by Gotzkowsky et al. (2000), who examined 17 HIV-infected individuals (with disease stage 1A of the Centers for Disease Control and Prevention classification) taking no medications known to alter DME activity. Most notably it was demonstrated that, compared with age- and sex-matched healthy volunteers, CYP3A and CYP2D6 activities were significantly decreased by 30–90%, and 25% of HIV-infected individuals exhibited genotype–phenotype dis- cordance for CYP2D6 (all were EMs by genotype, but PMs by phenotype). These findings indicate that the concurrent presence of HIV infection may limit the utility of genotype-based approaches for the assessment of DME activity in patients who use psychotropic mediations.
Liver disease can also modify blood flow and reduce the activity of DMEs. In acute disease, the major alteration is in hepatocellular function, but in chronic liver
170 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
disease (cirrhosis), the major abnormality lies with liver blood flow and possibly alteration in liver function. The effects of liver disease appear to be highly variable and difficult to predict (Morgan and McLean, 1995). Several studies with liver biopsy samples have found reduced protein concentrations and/or catalytic activ- ity associated with CYP1A (Bechtel et al., 2000), CYP2D6, CYP2E1, and CYP3A (George et al., 1995; Wilkinson, 1997), although the results were not uniformly consistent in other investigations (Adedoyin et al., 1998). Generally, hepatic impairment appears to be greater in cirrhotic livers than in patients with less severe forms of hepatic dysfunction, including chronic active hepatitis and cholestasis (Vesell, 1984; Wilkinson, 1997).
Genetic variability in drug metabolism and therapeutic drug monitoring
While there is considerable interest in identification of patients with unusual drug metabolism traits through genetic testing, this also raises an important question. Should metabolism by a genetically polymorphic pathway be a sufficient reason to restrict prescription of drugs in certain subpopulations?
The answer is intimately related to how we view the variability in drug response. A contribution from a polymorphic metabolic pathway implies considerable uncertainty in clearance and pharmacological effects of a drug. This uncertainty can take the form of drug toxicity but also can involve lack of efficacy owing to sub- therapeutic concentrations in association with ultrarapid metabolism. On the basis of such therapeutic uncertainties, the initial temptation is to avoid this variability by limiting the prescription of compounds that undergo genetically polymorphic metabolism. By contrast, this strategy may prevent the development of newer and efficacious drugs. This has particular significance for therapeutic fields where there are no well-established efficacious medicines (e.g., Alzheimer’s disease and other neurodegenerative disorders). If a new compound demonstrates therapeutic effi- cacy either alone or in combination, and prolongs the lives of patients, then vari- ability in its metabolism should not be a constraint for clinical development. For example, some of the classical antipsychotics (e.g., perphenazine) developed more than two decades ago are still widely used in many countries and demonstrate effi- cacy in approximately 70% of patients with major psychosis. Yet many of them are eliminated by CYP2D6-mediated oxidative metabolism. Such compounds can be used safely, provided adequate dose adjustments are made before drug administra- tion. Moreover, as discussed previously, the implicit assumption that nonpolymor- phic drug metabolism is associated with less variability is not always valid (e.g., consider the case with CYP1A2). In short, variability in drug metabolism is the rule, rather than the exception (Okey, 1990).

An alternative to avoidance of variability in drug metabolism would be to
171 Pharmacogenetics of psychotropic drug metabolism
characterize medications early in the course of their development with respect to primary phase I and phase II biotransformation enzymes responsible for their dis- position. Since the DMEs will continue to exist and play an important role in clear- ance of drugs and other xenobiotics, this approach needs to be complemented by advances in our understanding of the regulation of expression of DMEs. It is a common clinical practice to adjust the doses of medications eliminated predomin- antly by the kidney in patients with acute or chronic renal failure. Interestingly, marked interindividual differences in drug metabolism typically receive much less attention in routine clinical practice. This dual metabolic profiling strategy, that is of medications as well as patients, should allow a more optimal use of psychotropic drugs that are metabolized by a genetically polymorphic pathway. There is evidence that drug metabolism is already beginning to play a significant role in psychotropic drug development. For example, a recent survey conducted by the US Food and Drug Administration found that nearly 70% of the drugs approved by the Division of Neuropharmacological Products contained in vitro drug interaction and metab- olism studies (Yuan et al., 1999).
Therapeutic drug monitoring through genotyping and/or phenotyping of DMEs offer several advantages over the traditional approach based on measure- ment of plasma drug concentrations (Freeman and Oyewumi, 1997). The know- ledge of major DMEs that contribute to clearance of a drug is useful to forecast its pharmacokinetic variability in the general population, since the distribution of cat- alytic activity of most DMEs is established in different populations. Further, this information may provide mechanistic insights to predict inhibitory or inductive drug–drug interactions. In patient populations with known compliance problems or in those exposed to inducers of DMEs (e.g., tobacco smoke), information on individual drug metabolism capacity may help to establish a quantitative index of compliance to drug treatment at any given dosage. Lastly, certain DMEs are expressed both in the liver and the brain (e.g., CYP2D6 and CYP1A2) (Farin and Omiecinski, 1993). Therefore, genetic profiling of DMEs may provide an estimate of not only plasma drug concentrations but also psychotropic drug disposition at the site of action.
Conclusions
The theoretical foundations of genetic variability in drug metabolism have been firmly established since the 1960s (Kalow, 1997; Weber, 1997; Grant, 1999). The earlier pioneering pharmacogenetic studies on drug metabolism were conducted mainly in healthy volunteers or in small numbers of selected patients who present with unusual adverse drug reactions (Pfost et al., 2000). Recent advances in molec- ular biology, along with the declining costs and increased throughput of genetic

172 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
analyses, now make prospective validation of pharmacogenetic hypotheses entirely feasible in most clinical settings. Notably, pharmacogenetic inquiries are no longer solely dependent on observations of unusual drug metabolism phenotypes and can be initiated directly at the molecular genetic level (Nebert, 1999). The next few years will likely witness a proliferation in the number and scope of correlative “genotype versus clinical outcome” studies in psychopharmacology.
In the midst of this exciting molecular genetic revolution, some cautionary restraint and reflection over past accomplishments as well as methodological short- comings are needed to gain a balanced context for the future. Until recently, much emphasis was placed on monogenic control in drug metabolism (e.g., CYP2D6 genetic polymorphism), with little attention to multigenic regulation of gene expression (Cichon et al., 2000; Nebert, 2000). A case in point is the CYP3A4 enzyme, the most abundant CYP in the human liver. Twin and repeated drug administration studies indicate that heredity plays a prominent role in regulation of CYP3A4 expression (Penno et al., 1981; Ozdemir et al., 2000). Yet, despite inten- sive research efforts, few polymorphisms could be found within the coding or the immediate promotor regions of CYP3A4. It is likely that the marked interindivid- ual variability in CYP3A4 activity is attributable to multigenic control, but little is known on the identity and precise physical location of such putative regulatory loci in the genome. With the completion of the Human Genome Project in 2001, vir- tually all genes in the human genome will be available to examine novel gene–gene interactions and molecular genetic mechanisms for regulation of CYP3A4 and other DMEs. Assumption-free genome-wide association studies and linkage anal- yses may prove to be very useful tools to discover unprecedented regulatory genetic loci of relevance for drug metabolism that may not be readily predicted by the exist- ing list of plausible candidate genes (Casley et al., 1999; Goodman, 1999; Malhotra and Goldman, 1999).
It is estimated that the human genome contains more than 1000000 single nucleotide polymorphisms (SNPs) and these may explain differences in drug metabolism between individuals and populations. However, a point to keep in mind is that not all SNPs will influence the function of the encoded proteins, even though they may alter the amino acid sequences. Hence, it is conceivable that one of the critical determinants of success in pharmacogenetic research in the postge- nomic period will be the ability to perform gene expression and functional genomic studies. These investigations will provide the much needed mechanistic rationale to choose the battery of most relevant candidate genes and SNPs in subsequent clinical pharmacogenetic “validation” studies in humans (Evans and Relling, 1999).
The final pharmacological effects of medications are determined by a complex interplay of numerous genes and environmental factors (Masellis et al., 2000). Dietary factors, smoking, physiological and disease states, drug interactions, and
173 Pharmacogenetics of psychotropic drug metabolism
nutraceutical interactions may all modify or cause phenocopying of drug metabo- lism. Hence, these factors can mimic some of the hereditary drug metabolism traits (e.g., PM versus EM). An adequate knowledge of gene–environment interactions is central to proper interpretation of pharmacogenetic studies in psychiatry.
The contribution of genetic factors to drug metabolism has been mainly sought in the context of constitutive person-to-person variability, in the absence of inhib- itors or inducers of DMEs. Recently, some progress has been made in understand- ing the genetic basis of individual differences in DME induction. For example, smoking has been known for a long time as a potent inducer of CYP1A2 activity. Although the extent of exposure to tobacco smoke may explain part of the variabil- ity in CYP1A2 induction, our knowledge of the putative genetic factors that may influence CYP1A2 induction has been limited. A recent study in a sample of Caucasian volunteers identified a single nucleotide polymorphism (cytosine to adenosine) in intron 1 of CYP1A2 (CYP1A2*1F) at position 734 downstream from the transcription initiation site (Sachse et al., 1999). The CYP1A2*1F allele report- edly accounts for 18% of variability in CYP1A2 activity in Caucasian smokers and appears to influence caffeine disposition and the risk for tardive dyskinesia (Sachse et al., 1999; Basile et al., 2000). Clearly, the design of future pharmacogenetic studies may benefit from further explorations of similar gene–environment inter- actions.
Although genotyping is central to any psychiatric pharmacogenetic study, phe- notyping of DMEs with biomarker compounds (e.g., caffeine) will continue to play a significant role in validation of observations on associations between a particular DME genotype and clinical end-points. Progress has also been made in the devel- opment of rapid phenotyping procedures or “probe cocktails” for convenient measurement of multiple DME activities in tandem (Svensson and Bertilsson, 1999; Carrillo et al., 2000; Streetman et al., 2000b).
Pharmacogenetics provides a rational framework for evaluation of genetic vari- ation in DMEs as well as in genes encoding drug transporters, receptors, and ion channels, which can all influence an individual’s risk for adverse drug reactions or therapeutic failure (Arranz and Kerwin, 2000; Meyer, 2000; Sjöqvist, 2000). It can be anticipated that psychiatric pharmacogenetics will reduce the uncertainty and trial-and-error in choice and dosing of medications in the near future, thereby importantly contributing to development of personalized treatment guidelines in clinical psychiatry.
Acknowledgements
The authors thank Drs Werner Kalow, Laszlo Endrenyi, and Allan B. Okey for many insightful discussions and support.

174 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
REFERENCES
Adedoyin A, Arns PA, Richards WO, Wilkinson GR, Branch RA (1998). Selective effect of liver disease on the activities of specific metabolizing enzymes; investigation of cytochromes P-450 2C19 and 2D6. Clin Pharmacol Ther 64, 8–17.
Aklillu E, Persson I, Bertilsson L, Johansson I, Rodrigues F, Ingelman-Sundberg M (1996). Frequent distribution of ultrarapid metabolizers of debrisoquine in an Ethiopian population carrying duplicated and multiduplicated functional CYP2D6 alleles. J Pharmacol Exp Ther 278, 441–446.
Alfaro CL, Lam YW, Simpson J, Ereshefsky L (1999). CYP2D6 status of extensive metabolizers after multiple-dose fluoxetine, fluvoxamine, paroxetine, or sertraline. J Clin Psychopharmacol 19, 155–163.
Alfaro CL, Lam YW, Simpson J, Ereshefsky L (2000). CYP2D6 inhibition by fluoxetine, paroxe- tine, sertraline, and venlafaxine in a crossover study: intraindividual variability and plasma concentration correlations. J Clin Pharmacol 40, 58–66.
Arranz MJ, Kerwin RW (2000). Neurotransmitter-related genes and antipsychotic response: pharmacogenetics meets psychiatric treatment. Ann Med 32, 128–133.
Bailey DG, Arnold JMO, Spence JD (1994). Grapefruit juice and drugs – how significant is the interaction? Clin Pharmacokinet 26, 91–98.
Basile VS, Ozdemir V, Masellis M et al. (2000). A functional polymorphism of the cytochrome P-450 1A2 (CYP1A2) gene: association with tardive dyskinesia in schizophrenia. Mol Psychiatry 5, 410–417.
Bechtel YC, Haffen E, Lelouet H et al. (2000). Relationship between the severity of alcoholic liver cirrhosis and the metabolism of caffeine in 226 patients. Int J Clin Pharmacol Ther 38, 467–475. Bertilsson L (1995). Geographical/interracial differences in polymorphic drug oxidation. Current state of knowledge of cytochromes P-450 (CYP) 2D6 and 2C19. Clin Pharmacokinet
29, 192–209.
Bertilsson L, Dahl ML (1996). Polymorphic drug oxidation. Relevance to the treatment of psychi-
atric disorders. CNS Drugs 5, 200–223.
Bertilsson L, Aberg-Wistedt A, Gustafsson LL, Nordin C (1985). Extremely rapid hydroxylation
of debrisoquine: a case report with implication for treatment with nortriptyline and other tri-
cyclic antidepressants. Ther Drug Monit 7, 478–480.
Bertilsson L, Lou YQ, Du YL et al. (1992). Pronounced differences between native Chinese and
Swedish populations in the polymorphic hydroxylations of debrisoquin and S-mephenytoin.
Clin Pharmacol Ther 51, 388–397.
Bertilsson L, Carrillo JA, Dahl ML et al. (1994). Clozapine disposition covaries with CYP1A2
activity determined by a caffeine test. Br J Clin Pharmacol 38, 471–473.
Britto MR, Wedlund PJ (1992). Cytochrome P-450 in the brain. Potential evolutionary and ther- apeutic relevance of localization of drug-metabolizing enzymes. Drug Metab Dispos 20,
446–450.
Brockmeyer NH, Barthel B, Mertins L, Goos M (1998). Changes of antipyrine pharmacokinet-
ics during influenza and after administration of interferon-alpha and -beta. Int J Clin Pharmacol Ther 36, 309–311.

175 Pharmacogenetics of psychotropic drug metabolism
Brøsen K (1996). Drug-metabolizing enzymes and therapeutic drug monitoring in psychiatry. Ther Drug Monit 18, 393–396.
Carrillo JA, Christensen M, Ramos SI et al. (2000). Evaluation of caffeine as an in vivo probe for CYP1A2 using measurements in plasma, saliva, and urine. Ther Drug Monit 22, 409–417.
Casley WL, Menzies JA, Whitehouse LW, Moon TW (1999). Detection of quantitative trait loci affecting caffeine metabolism by interval mapping in a genome-wide scan of C3H/HEJ APN F2 mice. Drug Metab Dispos 27, 1375–1380.
Cohen LJ, DeVane CL (1996). Clinical implications of antidepressant pharmacokinetics and pharmacogenetics. Ann Pharmacother 3, 1471–1480.
Chou WH, Yan FX, de Leon J et al. (2000). Extension of a pilot study: impact from the cyto- chrome P-450 2D6 polymorphism on outcome and costs associated with severe mental illness. J Clin Psychopharmacol 20, 246–251.
Cichon S, Nöthen MM, Rietschel M, Propping P (2000). Pharmacogenetics of schizophrenia. Am J Med Genet 97, 98–106.
Conney AH, Pantuck EJ, Hsiao KC et al. (1976). Enhanced phenacetin metabolism in human subjects fed charcoal broiled beef. Clin Pharmacol Ther 20, 633–642.
Costa LG (2000). The emerging field of ecogenetics. Neurotoxicology 21, 85–90.
Dalén P, Dahl ML, Ruiz ML, Nordin J, Bertilsson L (1998). 10-Hydroxylation of nortriptyline in white persons with 0, 1, 2, 3, and 13 functional CYP2D6 genes. Clin Pharmacol Ther 63,
444–452.
Dresser GK, Spence JD, Bailey DG (2000). Pharmacokinetic–pharmacodynamic consequences
and clinical relevance of cytochrome P-450 3A4 inhibition. Clin Pharmacokinet 38, 41–57. Durnas C, Loi CM, Cusack BJ (1990). Hepatic drug metabolism and aging. Clin Pharmacokinet
19, 359–389.
Eaton DL (2000). Biotransformation enzyme polymorphism and pesticide susceptibility.
Neurotoxicology 21, 101–111.
Eaton DL, Bammler TK (1999). Concise review of the glutathione S-transferases and their sig-
nificance to toxicology. Toxicol Sci 49, 156–164.
Eaton DL, Klaassen CD (1996). Principles of toxicology. In Klaassen CD, ed. Casarett and Doull’s
Toxicology: The Basic Science of Poisons, 5th edn. New York: McGraw-Hill, pp. 13–33.
Eaton DL, Gallagher EP, Bammler TK, Kunze KL (1995). Role of cytochrome P-4501A2 in chem- ical carcinogenesis: implications for human variability in expression and enzyme activity.
Pharmacogenetics 5, 259–274.
Endrenyi L, Inaba T, Kalow W (1976). Genetic studies of amobarbital elimination based on its
kinetics in twins. Clin Pharmacol Ther 20, 701–714.
Ernst E (1999). Second thoughts about safety of St. John’s wort. Lancet 254, 2014–2016.
Evans WE, Relling MV (1999). Pharmacogenomics: translating functional genomics into ratio-
nal therapeutics. Science 286, 487–491.
Fang J, Gorrow JW (1999). Metabolism, pharmacogenetics, and metabolic drug–drug interac-
tions of antipsychotic drugs. Cell Mol Neurobiol 19, 491–510.
Farin FM, Omiecinski CJ (1993). Regiospecific expression of cytochrome P-450s and microso-
mal epoxide hydrolase in human brain tissue. J Toxicol Environ Health 40, 317–335.
176 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
Flockhart DA, Oesterheld JR (2000). Cytochrome P-450-mediated drug interactions. Child Adolesc Psychiatr Clin N Am 9, 43–76.
Fontana RJ, deVries TM, Woolf TF et al. (1998). Caffeine based measures of CYP1A2 activity cor- relate with oral clearance of tacrine in patients with Alzheimer’s disease. Br J Clin Pharmacol 46, 221–228.
Freeman DJ, Oyewumi LK (1997). Will routine therapeutic drug monitoring have a place in cloz- apine therapy? Clin Pharmacokinet 32, 93–100.
George J, Murray M, Byth K, Farrell GC (1995). Differential alterations of cytochrome P-450 pro- teins in livers from patients with severe chronic liver disease. Hepatology 21, 120–128.
Gidal BE, Reiss WG, Liao JS, Pitterle ME (1996). Changes in interleukin-6 concentrations follow- ing epilepsy surgery: potential influence on carbamazepine pharmacokinetics. Ann Pharmacother 30, 545–546.
Goodman L (1999). Hypothesis-limited research. Genome Res 9, 673–674.
Gotzkowsky SK, Kim J, Tonkin J et al. (2000). Altered drug metabolizing enzyme activity and genotype-phenotype discordance in HIV-infected subjects. 40th Interscience Conference on
Antimicrobial Agents and Chemotherapy, Toronto, Canada, abstract 1169:301.
Grant DM (1999). Pharmacogenomics and the changing face of clinical pharmacology. Can J
Clin Pharmacol 6, 131–132.
Haas CE (2000). Drug–cytokine interactions. In Piscitelli SC, Rodvold KA, eds. Drug Interactions
in Infectious Diseases. Totowa NJ: Humana Press, pp. 287–310.
Ingelman-Sundberg M, Oscarson M, McLellan RA (1999). Polymorphic human cytochrome P-
450 enzymes: an opportunity for individualized drug treatment. Trends Pharmacol Sci 20,
342–349.
Johansson I, Oscarson M, Yue QY, Bertilsson L, Sjöqvist F, Ingelman-Sundberg M (1994).
Genetic analysis of the Chinese cytochrome P-4502D locus: characterization of variant CYP2D6 genes present in subjects with diminished capacity for debrisoquine hydroxylation. Mol Pharmacol 46, 452–459.
Johne A, Brockmoller J, Bauer S, Maurer A, Langheinrich M, Roots I (1999). Pharmacokinetic interaction of digoxin with an herbal extract from St Johns wort (Hypericum perforatum). Clin Pharmacol Ther 66, 338–345.
Kall MS, Clausen J (1995). Dietary effect on mixed function P-450 1A2 activity assayed by esti- mation of caffeine metabolism in man. Hum Exp Toxicol 14, 801–807.
Kalow W (1962). Pharmacogenetics: Heredity and the Response to Drugs. Philadelphia, PA: Saunders, pp. 1–231.
Kalow W (1997). Pharmacogenetics in biological perspective. Pharmacol Rev 49, 369–379. Kalow W (1999). Pharmacogenetic research: a revolutionary science. J Psychiatr Neurosci 24,
139–140.
Kappas A, Alvares AP, Anderson KE et al. (1978). Effects of charcoal-broiled beef on antipyrine
and theophylline metabolism. Clin Pharmacol Ther 23, 445–450.
Kashuba ADM, Bertino JS Jr. (2000). Mechanisms of drug interactions. In Piscitelli SC, Rodvold
KA, eds. Drug Interactions in Infectious Diseases. Totowa, NJ: Humana Press, pp. 13–38. Kashuba AD, Bertino JS Jr., Kearns GL et al. (1998a). Quantitation of three-month intra- individual variability and influence of sex and menstrual. cycle phase on CYP1A2,
177 Pharmacogenetics of psychotropic drug metabolism
N-acetyltransferase-2, and xanthine oxidase activity determined. with caffeine phenotyping.
Clin Pharmacol Ther 63, 540–551.
Kashuba AD, Bertino JS Jr., Rocci ML Jr., Kulawy RW, Beck DJ, Nafziger AN (1998b).
Quantification of 3-month intraindividual variability and the influence of sex and menstrual cycle phase on CYP3A activity as measured by phenotyping with intravenous midazolam. Clin Pharmacol Ther 64, 269–277.
Kashuba AD, Nafziger AN, Kearns GL, et al. (1998c). Quantification of intraindividual variabil- ity and the influence of menstrual cycle phase on CYP2D6 activity as measured by dextro- methorphan phenotyping. Pharmacogenetics 8, 403–410.
Kerb R, Bauer S, Brockoller J, Roots I (1997). Urinary 6-hydroxycortisol excretion rate is affected by treatment with hypericum extract [abstract]. Eur J Clin Pharmacol 52(Suppl), A186.
Kim RB, Wilkinson GR (1996). Watercress inhibits human CYP2E1 activity in vivo as measured by chlorzoxazone 6-hydroxylation. Clin Pharmacol Ther 59, 170.
Kinirons MT, Crome P (1997). Clinical pharmacokinetic considerations in the elderly: an update. Clin Pharmacokinet 33, 302–312.
Kurth JH (2000). Pharmacogenomics: future promise of a tool for identifying patients at risk. Drug Info J 34, 223–227.
Labbe L, Sirois C, Pilote S, et al. (2000). Effect of gender, sex hormones, time variables and physiological urinary pH on apparent CYP2D6 activity as assessed by metabolic ratios of marker substrates. Pharmacogenetics 10, 425–438.
Lazarou J, Pomeranz BH, Corey PN (1998). Incidence of adverse drug reactions in hospitalized patients: meta-analysis of prospective studies. J Am Med Assoc 279, 1200–1205.
Lee BL, Wong D, Benowitz NL, Sullam PM (1993). Altered patterns of drug metabolism in patients with acquired immunodeficiency syndrome. Clin Pharmacol Ther 53, 529–535.
Lin JH, Lu AY (1998). Inhibition and induction of cytochrome P-450 and the clinical implica- tions. Clin Pharmacokinet 35, 361–90.
Lin KM, Poland RE, Wan Y-JY, Smith MW, Lesser IM (1996). The evolving science of pharma- cogenetics: clinical and ethnic perspectives. Psychopharmacol Bull 32, 205–217.
Llerena A, Edman G, Cobaleda J, Benitez J, Schalling D, Bertilsson L (1993). Relationship between personality and debrisoquine hydroxylation capacity. Suggestion of an endogenous neuroactive substrate or product of the cytochrome P-4502D6. Acta Psychiatr Scand 87, 23–28.
Mai I, Kruger H, Budde K et al. (2000). Hazardous pharmacokinetic interaction of Saint John’s wort (Hypericum perforatum) with the immunosuppressant cyclosporin. Int J Clin Pharmacol Therap 38, 500–502.
Malhotra AK, Goldman D (1999). Benefits and pitfalls encountered in psychiatric genetic asso- ciation studies. Biol Psychiatry 45, 544–550.
Masellis M, Basile VS, Ozdemir V, Meltzer HY, Macciardi FM, Kennedy JL (2000). Pharmacogenetics of antipsychotic treatment: lessons learned from clozapine. Biol Psychiatry 47, 252–266.
Meyer UA (2000). Pharmacogenetics and adverse drug reactions. Lancet 356, 1667–1671. Meyer UA, Zanger UM (1997). Molecular mechanisms of genetic polymorphisms of drug metab-
olism. Annu Rev Pharmacol Toxicol 37, 269–296.
178 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
Meyer UA, Amrein R, Balant LP et al. (1996). Antidepressants and drug-metabolizing enzymes – expert group report. Acta Psychiatr Scand 93, 71–79.
Miksys S, Rao Y, Sellers EM, Kwan M, Mendis D, Tyndale RF (2000). Regional and cellular dis- tribution of CYP2D subfamily members in rat brain. Xenobiotica 30, 547–564.
Miller M, Opheim KE, Raisys VA, Motulsky AG (1984). Theophylline metabolism: variation and genetics. Clin Pharmacol Ther 35, 170–182.
Morgan DJ, McLean AJ (1995). Clinical pharmacokinetic and pharmacodynamic considerations in patients with liver disease: an update. Clin Pharmacokinet 29, 370–391.
Motulsky AG (1957). Drug reactions, enzymes and biochemical genetics. J Am Med Assoc 165, 835–837.
Murphy MP, Beaman ME, Clark LS et al. (2000). Pharmacogenetics 10, 583–590.
Nebert DW (1999). Pharmacogenetics and pharmacogenomics: why is this relevant to the clini-
cal geneticist? Clin Genet 56, 247–258.
Nebert DW (2000). Suggestions for the nomenclature of human alleles: relevance to ecogenetics,
pharmacogenetics and molecular epidemiology. Pharmacogenetics 10, 279–290.
Nelson DR, Koymans L, Kamataki T et al. (1996). P-450 superfamily: update on new sequences,
gene mapping, accession numbers and nomenclature. Pharmacogenetics 6, 1–42.
Niznik HB, Tyndale RF, Sallee FR et al. (1990). The dopamine transporter and cytochrome P450IID1 (debrisoquine 4-hydroxylase) in brain: resolution and identification of two distinct
[3H]GBR-12935 binding proteins. Arch Biochem Biophys 276, 424–432.
Okey AB (1990). Enzyme induction in the cytochrome P-450 system. Pharmacol Ther 45,
241–298.
Omura T, Sato R (1964a). The carbon monoxide-binding pigment of liver microsomes. 1.
Evidence for its hemoprotein nature. J Biol Chem 239, 2370–2378.
Omura T, Sato R (1964b). The carbon monoxide-binding pigment of liver microsomes. 2.
Solubilization, purification, and properties. J Biol Chem 239, 2379–2385.
O’Neil WM, Bilfix BM, Markoglou N, Di Girolamo A, Tsoukas CN, Wainer IW (2000). Genotype and phenotype of cytochrome P-450 2D6 in human immunodeficiency virus-positive patients and patients with acquired immunodeficiency syndrome. Eur J Clin Pharmacol 56,
231–240.
Ozdemir V, Kalow W, Tang BK et al. (2000). Evaluation of the genetic contribution to CYP3A4
activity in vivo: a repeated drug administration method. Pharmacogenetics 10, 373–388. Pantuck EJ, Pantuck CB, Min BH et al. (1989). Stimulatory effect of brussels sprouts and cabbage
on human drug metabolism. Clin Pharmacol Ther 35, 88–95.
Penno MB, Dvorchik BH, Vesell ES (1981). Genetic variation in rates of antipyrine metabolite
formation: a study in uninduced twins. Proc Natl Acad Sci, USA 78, 5193–5196.
Pfost DR, Boyce-Jacino MT, Grant DM (2000). A SNP shot: pharmacogenetics and the future of
drug therapy. Trend Biotechnol 18, 334–338.
Piscitelli SC, Burstein AH, Chaitt D, Alfaro RM, Falloon J (2000). Indinavir concentrations and
St. John’s wort. Lancet 12, 547–548.
Piscitelli SC, Burstein AH, Welden N, Gallicano K, Falloon J (2001). Garlic supplements decrease
saquinavir plasma concentrations. Abstracts of the 8th Annual Conference on Retroviruses and Opportunistic Infections, Chicago, IL.
179 Pharmacogenetics of psychotropic drug metabolism
Pollock BG, Mulsant BH, Sweet RA, Rosen J, Altieri LP, Perel JM (1995). Prospective cytochrome P-450 phenotyping for neuroleptic treatment in dementia. Psychopharmacol Bull 31, 327–331. Preskorn SH (1997). Clinically relevant pharmacology of selective serotonin reuptake inhibitors. An overview with emphasis on pharmacokinetics and effects on oxidative drug metabolism.
Clin Pharmacokinet 32(Suppl 1), 1–21.
Renton KW (2000). Hepatic drug metabolism and immunostimulation. Toxicology 142, 173–178. Renton KW, Mannering GJ (1976). Depression of the hepatic cytochrome P-450 mono-oxyge-
nase system by administered tilorone (2, 7-bis(2-(diethylamino)ethoxy)fluoren-9-one dihy-
drochloride). Drug Metab Dispos 4, 223–231.
Ring BJ, Catlow J, Lindsay TJ et al. (1996). Identification of the human cytochromes P-450
responsible for the in vitro formation of the major oxidative metabolites of the antipsychotic
agent olanzapine. J Pharmacol Exp Ther 276, 658–666.
Sachse C, Brockmoller J, Bauer S, Roots I (1999). Functional significance of a C→A polymor-
phism in intron 1 of the cytochrome P-450 CYP1A2 gene tested with caffeine. Br J Clin
Pharmacol 47, 445–449.
Schmucker DL (1989). Is the rat a suitable model for ageing research? In Woodhouse KW, Yelland
C, James OFW, eds. The Liver: Metabolism and Aging. Rijswijk, the Netherlands: Eurage, pp. 3–18. Shedlofsky SI, Israel BC, McClain CJ, Hill DB, Blouin RA (1994). Endotoxin administration to humans inhibits hepatic cytochrome P-450-mediated drug metabolism. J Clin Invest 94,
2209–2214.
Shimada T, Yamazaki H, Mimura M et al. (1994). Interindividual variations in human liver cyto-
chrome P-450 enzymes involved in the oxidation of drugs, carcinogens and toxic chemicals: studies with liver microsomes of 30 Japanese and 30 Caucasians. J Pharmacol Exp Ther 270, 414–423.
Sindrup SH, Brøsen K (1995). The pharmacogenetics of codeine hypoalgesia. Pharmacogenetics 5, 335–346.
Sindrup SH, Brøsen K, Gram LF et al. (1992). The relationship between paroxetine and the spar- teine oxidation polymorphism. Clin Pharmacol Ther 51, 278–287.
Sjöqvist F (1999). The past, present and future of clinical pharmacology. Eur J Clin Pharmacol 55, 553–557.
Sjöqvist F (2000). Drug safety in relation to efficacy: the view of a clinical pharmacologist. Pharmacol Toxicol 86(Suppl 1), 30–32.
Steen VM, Lovlie R, MacEwan T et al. (1997). Dopamine D3-receptor gene variant and suscepti- bility to tardive dyskinesia in schizophrenic patients. Mol Psychiatry 2, 139–145.
Streetman DS, Bertino JS Jr., Nafziger AN (2000a). Phenotyping of drug-metabolizing enzymes in adults: a review of in-vivo cytochrome P-450 phenotyping probes. Pharmacogenetics 10, 187–216.
Streetman DS, Bleakley JF, Kim JS et al. (2000b). Combined phenotypic assessment of CYP1A2, CYP2C19, CYP2D6, CYP3A, N-acetyltransferase-2, and xanthine oxidase with the “Cooperstown cocktail”. Clin Pharmacol Ther 68, 375–383.
Svensson JO, Bertilsson L (1999). Rapid high-performance liquid chromatographic method for determination of debrisoquine and 4-hydroxy-debrisoquine in urine for CYP2D6 phenotyp- ing. Pharmacogenetics 9, 529–531.
180 V. Ozdemir, A. D. M. Kashuba, V. S. Basile, and J. L. Kennedy
Tanaka E, Hisawa S (1999). Clinically significant pharmacokinetic drug interactions with psycho- active drugs: antidepressants and antipsychotics and the cytochrome P-450 system. J Clin Pharm Ther 24, 7–16.
Trenholme GM, Williams RL, Rieckmann KH, Frischer H, Carson PE (1976). Quinine disposi- tion during malaria and during induced fever. Clin Pharmacol Ther 19, 459–467.
Vesell ES (1978). Twin studies in pharmacogenetics. Hum Genet Suppl 1, 19–30.
Vesell ES (1984). Noninvasive assessment in vivo of hepatic drug metabolism in health and
disease. Ann NY Acad Sci 428, 293–307.
Vogel F (1959). Moderne Probleme der Human-genetik. Ergebn Inn Med Kinderheilk 12, 52–125. von Moltke LL, Greenblatt DJ (2000). Drug transporters in psychopharmacology – are they
important? J Clin Psychopharmacol 20, 291–294.
Wacher VJ, Wu CY, Benet LZ (1995). Overlapping substrate specificities and tissue distribution
of cytochrome P-450 3A and P-glycoprotein: implications for drug delivery and activity in
cancer chemotherapy. Mol Carcinogen 13, 129–134.
Walter-Sack I, Klotz U (1996). Influence of diet and nutritional status on drug metabolism. Clin
Pharmacokinet 31, 47–64.
Watkins LR, Maier SF, Goehler LE (1995). Immune activation: the role of pro-inflammatory
cytokines in inflammation, illness responses, and pathological pain states. Pain 63, 289–302. Weber WW (1997). Pharmacogenetics. New York: Oxford University Press.
Wentworth JM, Agostini M, Love J, Schwabe JW, Chatterjee VKK (2000). St John’s wort, a herbal
antidepressant, activates the steroid X receptor. J Endocrinol 166, R11–R16.
Wilkinson GR (1997). The effects of diet, aging, and disease-states on presystemic elimination
and oral drug bioavailability in humans. Adv Drug Deliv Rev 27:129–159.
Wooles WR, Borzelleca JF (1966). Prolongation of barbiturate sleeping time in mice by stimula-
tion of the reticuloendothelial system (RES). J Reticuloendothel Soc 3, 41–47.
Yuan R, Parmelee T, Balian JD et al. (1999). In vitro metabolic interaction studies: experience of
the food and drug administration. Clin Pharmacol Ther 66, 9–15.
9
Pharmacogenetics of chiral psychotropic drugs
Pierre Baumann and Chin B. Eap
University Department of Adult Psychiatry, Prilly-Lausanne, Switzerland
OVERVIEW
A recent editorial on metabolism and chirality in psychopharmacology stated that many invest- igations do not consider that many psychotropic drugs have one or more chiral centers, intro- ducing steric factors that may make important contributions to their overall pharmacological or toxicological profile (Baker et al., 1994). Moreover, there is an increasing awareness that, besides environmental factors, genetic factors regulate the fate of drugs in the organism, and that polymorphic enzymes such as some cytochrome P-450 isozymes display stereoselectiv- ity toward chiral substrates or in the formation of chiral metabolites from achiral parent com- pounds. Present knowledge about the pharmacogenetics of metabolism of psychotropic drugs is based mainly on the study of the polymorphic enzymes CYP2D6 and CYP2C19. This chapter summarizes present knowledge on the pharmacology, metabolism, pharmacokinetics, and pharmacogenetics of antidepressants, antipsychotics, and methadone.
Introduction
Numerous psychotropic drugs are chiral and have been introduced as racemates (Table 9.1). Chiral drugs are defined as having one or several asymmetric centers: at least one carbon (or sulfur) atom of their molecule has four different atoms or groups attached to it. For one asymmetric carbon atom, two enantiomers may exist and they are mirror images of each other. Sometimes, confusing terms, signs, and letters are used for their denomination. Polarized light is deviated by such a mole- cule and the direction of rotation is called levorotatory ( ; l) (or anticlockwise) or dextrorotatory ( ; d) (or clockwise). However, because rotation may depend on the solvent used, the absolute configuration of the chiral molecule relative to the enantiomers of glyceraldehyde as standards is indicated by the symbols L and R. Increasingly, for drugs, the symbols R and S are used, according to the Cahn–Ingold–Prelog system for the designation of absolute configurations (Wainer, 1993). Clarification of these terms is essential because, in many of the reports cited in this chapter, different terms are used to designate enantiomers.

181

182 P. Baumann and C. B. Eap
Table 9.1. Metabolism of chiral antidepressants by CYP2D6 and CYP2C19
Eutomer: substrateb of Generic name Drug introduced as enantiomer (eutomer)a CYP2D6 CYP2C19
Pharmacologically active

Escitalopram Citalopram Fluoxetine Paroxetine Sertraline Mianserin Mirtazapine Trimipramine Bupropion Milnacipran Reboxetine Venlafaxine Viloxazine
Enantiomer
Racemate
Racemate 3S–trans-Paroxetine ( )-cis-(1S,4S)-Sertraline Racemate
Racemate
Racemate
Racemate
Racemate
(R,R)- and (S,S)-Reboxetine Racemate
Racemate
(S)-Citalopram
(S)-Citalopram
(S)- and (R)-Fluoxetine 3S–trans-Paroxetine (+)-cis-(1S,4S)-Sertraline – (S)- and (R)-Mianserin (S)- and (R)-Mirtazapine l-Trimipramine
–
– (S,S)-Reboxetine –
(R)-Venlafaxine
(?)
Notes:
. a Someofthedistomersmayalsohavesomeclinicallyrelevantpharmacologicalactivity,whichcandiffer
from that of the eutomer quantitatively or qualitatively.
. b Substratemetabolism:–,notbyCYP2D6; ,minorconversionbyCYP2D6; ,CYP2D6(orCYP2C19)
alone or with other forms of cytochrome P-450 plays a major role (some of the distomers may also be substrates of CYP2D6).
Generally, one of the enantiomers is considered to be the eutomer, i.e., the phar- macologically (more) active compound, in contrast to the distomer, which displays a lower pharmacological activity. Stereoselectivity is often expressed as the eudismic ratio (ratio of activities of a pair of enantiomers). However, a particular enantiomer may be an eutomer for one type of pharmacological activity, and the distomer for another type. This classification, therefore, neglects situations in which the enan- tiomers have either an overlapping pharmacological profile or distinct pharmaco- logical properties that can all contribute to the overall clinical activity, as has been demonstrated for mirtazapine. This is explained by the fact that, in order to exert maximal pharmacological activity, drugs have to fit optimally into receptor mole- cules, taking account of many steric conditions. Enantiomers of chiral psychotropic drugs may differ by their pharmacokinetic, pharmacogenetic, and pharmacody- namic properties (Baumann and Rochat, 1995; Lane and Baker, 1999; Baumann and Eap, 2001). Knowledge of the activity of enantiomers is, therefore, important for a better understanding of their mechanism of action. Moreover, in the clinical context, it could be relevant to use stereoselective assays for therapeutic drug monitoring.
183 Pharmacogenetics of chiral psychotropic drugs
Most psychotropic drugs have active metabolites (Rudorfer and Potter, 1997, 1999; Sanchez and Hyttel, 1999), and many achiral drugs give rise to chiral metabo- lites, the formation of which may be stereoselective and depend on the genotype of the patients.
Although genetic polymorphisms have been described for numerous enzymes implicated in drug metabolism (Ingelman-Sundberg, 1998), clinically relevant data about the pharmacogenetics of the metabolism of psychotropic drugs have only been reported in relation to CYP2D6 and CYP2C19, which are isozymes of cytochrome P-450 (Bertilsson and Dahl, 1996; Bertilsson et al., 1997; Ingelman-Sundberg et al., 1999; Coutts and Urichuk, 1999; Dahl and Sjöqvist, 2000; see also Ch. 8). For these two enzymes, genotyping methods are available, but subjects can also be phenotyped for CYP2D6 activity with dextromethorphan, debrisoquine or sparteine, and for CYP2C19 activity with mephenytoin. For CYP2D6, gene amplification has been described, giving rise to high enzyme activity in subjects presenting this genotype. They are generally considered to be ultrarapid metabolizers (UMs) (Bertilsson et al., 1993; Johansson et al., 1993; Lundqvist et al., 1999). Phenotyping procedures do not allow to discriminate between extensive metabolizers (EMs) and UMs.
Cytochrome P-450 enzyme systems implicated in conjugation of drugs and other enzymes have been shown to have steric preferences toward drugs, and this explains differences in the metabolism of enantiomers. The aim of this chapter is to present the pharmacology, metabolism, pharmacokinetics, and pharmacogenet- ics of chiral psychotropic drugs. The relative lack of adequate studies on their phar- macogenetics may partly be explained by technical difficulties in the stereoselective analysis of these types of compounds (Marzo and Balant, 1996), but also by a lack of availability of pure enantiomers of drugs or their metabolites. Drugs that have been introduced as pure enantiomers (e.g., paroxetine, sertraline) will not be dis- cussed here, but data concerning their pharmacogenetics are summarized else- where (see Ch. 8).
Chiral antidepressants
Tricyclic antidepressants and related compounds

Mianserin
Mianserin, a chiral antidepressant, is known for its antagonistic effects at serotonin and presynaptic 2-adrenoreceptors (Pinder, 1985). (S)-( )-Mianserin is more potent than the (R)-( )-enantiomer in displacing 1- and 2-ligands, and in inhibiting norepinephrine reuptake in the brain, while both enantiomers have a negligible effect on serotonin (5-HT) and dopamine reuptake. While (R)-( )- mianserin has no effect on potassium-evoked release of norepinephrine in cortical
184 P. Baumann and C. B. Eap
slices, (S)-( )-mianserin exerts a potent increase in potassium release. After chronic administration, (S)-( )-mianserin but not (R)-( )-mianserin produces functional supersensitivity at 2-autoreceptors. The antidepressant effect of (S)- ( )-mianserin, in contrast to that of the (R)-enantiomer, is also evident from animal experiments using behavioral tests. Interestingly, (S)-( )-mianserin has a higher affinity for 5-HT1 and 5-HT2 receptors than (R)-( )-mianserin, but stereo- selectivity is inversed with regard to their affinity for 5-HT3 receptors.
These observations suggest that (S)-mianserin is probably the eutomer. In humans, it is more readily 8-hydroxylated and N-oxidized than the distomer (R)- mianserin, but the latter is preferentially N-desmethylated (Koyama et al., 1996; Chow et al., 1999). 8-Hydroxylation of mianserin is controlled by CYP2D6, CYP2B6, CYP3A4, and CYP1A2, and N-desmethylation by CYP2B6, CYP2C19, CYP3A4, CYP1A2, and CYP2D6 (Koyama et al., 1996). CYP3A4 does not appear to metabolize preferentially the enantiomers of mianserin (Eap et al., 1999). CYP2D6 more readily 8-hydroxylates (S)-mianserin but it also N-desmethylates (R)-mianserin (Chow et al., 1999). In Japanese patients (of whom probably none was a poor metabolizer (PM)) treated with mianserin, the mean ratio of (S)-/(R)- mianserin was 1.9, but it varied interindividually between 0.5 and 4.8 (Tybring et al., 1995). In EMs of dextromethorphan (CYP2D6), the ratio of (S)-/(R)-mianserin and (S)-/(R)-desmethylmianserin was 1–4.6 and 0.19–0.64, respectively (Eap et al., 1994). In CYP2D6-phenotyped healthy volunteers treated with a single oral dose of mianserin, the area under the plasma concentration versus time curve (AUC) of (S)-( )-mianserin but not (R)-( )-mianserin correlated significantly with the metabolic ratio of debrisoquine; the ratio of the AUCs of (S)-/(R)-mianserin was higher in PMs than in EMs (Dahl et al., 1994). Desmethylmianserin is also metab- olized by CYP2D6 but none of these compounds is apparently a substrate of CYP2C19. However, in clinical steady-state conditions, concentrations of the enan- tiomers of mianserin and desmethylmianserin did not differ spectacularly between homozygous and heterozygous EMs (CYP2D6), but the only PM had the highest ratio (S)-/(R)-mianserin in plasma (Eap et al., 1998). The hypothesis that (S)- mianserin rather than (R)-mianserin is a substrate of CYP2D6 is strengthened by an interaction study (Yasui et al., 1997) with depressive patients treated with mian- serin and thioridazine, which is a strong CYP2D6 inhibitor (Baumann et al., 1992). Coadministration of thioridazine (40 mg/day) for 1 week doubled plasma concen- trations of (S)-mianserin but was without effect on (R)-mianserin. Interestingly, thioridazine increased also both (S)-desmethylmianserin and (R)-desmethylmi- anserin. In conclusion, (S)-mianserin and (R)-desmethylmianserin, but not (R)- mianserin, have to be considered to be stereoselectively metabolized by CYP2D6.
In carriers of the CYP2D6*10 allele, CYP2D6 activity is decreased. This explains why Japanese carriers of a CYP2D6*10 allele had higher plasma concentrations of
185 Pharmacogenetics of chiral psychotropic drugs
(S)-mianserin. Interestingly, they were more likely to respond to the drug than depressive patients with a wild type/wild type genotype (Mihara et al., 1997).
Mirtazapine
Mirtazapine (ORG 3770) (Holm and Markham, 1999) differs from mianserin only by a N-atom replacing a C-atom in the ring structure. It is also a potent 2- adrenoceptor antagonist, with little affinity for 1-adrenoceptors and it does not block norepinephrine reuptake. (S)-( )-Mirtazapine is several times more potent than (R)-( )-mirtazapine as an 2-autoreceptor antagonist, as shown both by neurochemical and by behavioral experiments. Only (S)-( )- but not (R)-( )-mirtazapine increases serotonergic raphe cell firing and 5-HT release (David and Wilde, 1996). During a chronic treatment, (R)-( )-mirtazapine had a more pronounced effect on 5-HT1A receptor function than does the (S)-( )-enantiomer. Mirtazapine also acted as an antagonist at 5-HT2- and 5-HT3-receptors (Holm and Markham, 1999). (R)-( )-Mirtazapine is thought to be responsible for 2-adrenergic heteroreceptor and 5-HT3-receptor blockade (Kooyman et al., 1994; McGrath et al., 1998). Chronic treatment with (R)-( )- but not with (S)-( )-mirtazapine led to a decrease in the density of 1-adrenoceptors in frontal cortex, as already described for many other antidepressant treatments, and both enantiomers decreased the density of 5-HT2-receptors (McGrath et al., 1998). However, the authors conclude that, in their model, none of the enantiom- ers seems to be more active than racemic mirtazapine.
In patients of unknown phenotype who were treated with mirtazapine, plasma ( )-mirtazapine and ( )-mirtazapine concentrations varied interindividually between 5 and 69ng/ml and 13 and 88ng/ml, respectively; generally, trough levels of ( )-mirtazapine were one third or half of those of ( )-mirtazapine (Dodd et al., 2000). Dahl et al. (1997b) summarized the published studies on the pharmacokinetics of mirtazapine and its metabolism by cytochrome P-450. In vitro, using microsomes from cells expressing single cytochrome P-450 isoforms, the formation of 8-hydroxymirtazapine occurs mainly by CYP2D6, and to some extent by CYP1A2, while CYP3A4 is mainly implicated in N-2 desmethylation and N-2 oxidation. More recently, it was calculated, on the basis of in vitro experiments in human liver microsomes, that the overall metabolism of mirtazapine was 55% by 8-hydroxylation, 35% by N-desmethylation, and 10% by N-oxidation, at an extrapolated in vivo mirtazapine concentration of 2mol/l in liver. In these condi- tions, CYP2D6 contributed 65% of the hydroxylation of mirtazapine, CYP1A2 30% (Störmer et al., 2000a), while CYP2C8 and CYP2C9 contributed less than 10% to the overall mirtazapine biotransformation (Störmer et al., 2000a). At higher concentrations of mirtazapine (250mol/l), the contribution of CYP2D6 decreased to 20% and that of CYP1A2 increased to 50% (Störmer et al., 2000b).
186 P. Baumann and C. B. Eap
Recently, an in vitro study with cDNA transfected human lymphoblast microsomes expressing CYP isoforms showed a preferential metabolism of ( )-mirtazapine by CYP2D6 in comparison with ( )-mirtazapine, while CYP1A2 and CYP3A4 show low activity towards both mirtazapine enantiomers. Moreover, CYP2C9 and CYP2C19 do not seem to be involved in the metabolism of this antidepressant (Dodd et al., 2001). It was, therefore, unexpected to find, in a panel study with healthy human volunteers, that the pharmacokinetics of mirtazapine and its N- desmethylated metabolite, as examined by an achiral analytical procedure, did not differ between PMs and EMs of debrisoquine (Dahl et al., 1997a). In humans (S)- ( )-mirtazapine has a half-life of 9.9 ± 3.1h and is mainly metabolized by 8- hydroxylation followed by glucuronidation, while (R)-( )-mirtazapine has a half-life of 18.0 ± 2.5h and is preferentially and reversibly N-glucuronidated as a quaternary ammonium glucuronide. After admininistration of racemic mirtaza- pine, the elimination of (R)-mirtazapine occurs at a similar rate in EMs and PMs (CYP2D6), but (S)-mirtazapine has a longer half-life in PMs (18.8 ± 4.7h) than in EMs (13.2 ± 4. h), as observed using a stereoselective method (Delbressine et al., 1998; Timmer et al., 2000).
Trimipramine
The pharmacology of the chiral antidepressant trimipramine differs from that of other typical tricyclic drugs in that its activity as a norepinephrine and serotonin reuptake inhibitor is weak, but its affinity for 5-HT2- and dopamine receptors is pronounced. In this sense, trimipramine presents some similarities with atypical antipsychotics such as clozapine. -Trimipramine rather than -trimipramine has to be considered as the pharmacologically active compound, as shown in in vitro studies. -Trimipramine displaced ligands at dopamine D1, D2 binding sites in stri- atal areas in pig brain and 5-HT2 binding sites in cerebral cortex of rat more potently than did -trimipramine (Gross et al., 1991). -Trimipramine, inhibited potassium-induced uptake of calcium in rat brain synaptosomes more potently than did -trimipramine (Beauchamp et al., 1992); however, depending on the model used and the brain region examined there is a loss of stereoselectivity (Lavoie et al., 1994). Preliminary studies suggest that in accordance with its pharmacolog- ical profile, trimipramine may have some therapeutic effect in acute schizophrenia (Eikmeier et al., 1990; Berger and Gastpar, 1996).
Trimipramine is N-desmethylated to desmethyltrimipramine and both com- pounds are hydroxylated. Measurements of plasma steady-state concentrations of the enantiomers of trimipramine and its metabolites suggest that the metabolism of this antidepressant occurs stereoselectively in patients (Eap et al., 1992a). The demonstration that trimipramine is a CYP2D6 substrate was made in a study with healthy volunteers in whom plasma half-life of trimipramine was doubled after
187 Pharmacogenetics of chiral psychotropic drugs
coadministration of the strong CYP2D6 inhibitor quinidine. Moreover, electro- encephalograph modifications produced by trimipramine were more pro- nounced and longer lasting after addition of quinidine (Eap et al., 1992b). CYP2D6 hydroxylates trimipramine stereoselectively, as shown in a study with depres- sive patients presenting a CYP2D6 and CYP2C19 EM phenotype and treated with 300–400mg/day trimipramine for 5 weeks. CYP2D6 was implicated in the 2-hydroxylation of -trimipramine, -desmethyltrimipramine and -desmethyltrimipramine, but not of -trimipramine; CYP2C19 with some stereoselectivity responsible for N-desmethylation of -trimipramine. The only patient with a genetic deficiency of CYP2D6 (PM of dextromethorphan) had the highest dose- and weight-corrected concentrations of - and -desmethyltrim- ipramine, while the only PM of mephenytoin (CYP2C19 deficiency) had the highest - and -trimipramine concentrations (Eap et al., 2000a). CYP3A4 con- tributes also to the biotransformation of trimipramine.
Selective serotonin reuptake inhibitors
Citalopram
The chiral antidepressant drug citalopram is a potent and the most selective sero- tonin reuptake inhibitor (SSRI) available (Hyttel et al., 1995; Sanchez and Hyttel, 1999) (Fig. 9.1). In vitro 5-HT uptake inhibition studies with rat brain synapto- somes and behavioral experiments demonstrated that both the (S)-( )-enan- tiomers of citalopram and, to a minor extent, its main metabolite desmethylcitalopram have to be considered as the pharmacologically active agents (Hyttel et al., 1992; Owens et al., 2001). Eudismic ratios for 5-HT uptake inhibition are 167 and 6.6 for citalopram and desmethylcitalopram, respectively. (S)-( )- Citalopram (escitalopram), in contrast to (R)-( )-citalopram, potentiated the effect of -5-hydroxytryptophan (-5-HTP) in mice (Hyttel et al., 1992). In depres- sive patients treated with citalopram, (S)-citalopram plasma concentrations were generally lower than those of (R)-citalopram (Rochat et al., 1995a,b; Bondolfi et al., 1996, 2000; Sidhu et al., 1997). Mean serum elimination half-life was found to be 47 h and 35 h for (R)-( )- and (S)-( )-citalopram, respectively, in healthy subjects with an EM phenotype (CYP2D6) (Sidhu et al., 1997). Escitalopram represents the first example in psychopharmacology of a “chiral switch,” i.e., the introduction of a single enantiomeric form of a drug previously available in a racemate form (Tucker, 2000). Some clinical studies confirm the clinical efficacy of escitalopram in depression (Montgomery et al., 2001).
In vitro studies with human liver microsomes and complementary DNA (cDNA)-expressed human cytochrome P-450 isoforms, combined with in vitro interaction studies, showed that CYP2C19, CYP3A4, and CYP2D6 catalyzed
 
Fig.9.1. Metabolic pathways of citalopram. MAO, monoamine oxidase; CYP, cytochrome P-450; AO, aldehyde oxidase.
189 Pharmacogenetics of chiral psychotropic drugs
N-desmethylation of citalopram to N-desmethylcitalopram (Kobayashi et al., 1997), while CYP1A2, CYP2A6, CYP2B6, CYP2C9, CYP2D6, and CYP2E1 were not involved (Fig. 9.1). In a panel study with CYP2C19 and CYP2D6 phenotyped healthy subjects, CYP2C19 controlled citalopram N-desmethylation to desmethyl- citalopram while CYP2D6 was involved in desmethylation of the latter to dides- methylcitalopram (Sindrup et al., 1993). In steady-state conditions, plasma concentration ratios of citalopram/desmethylcitalopram were found to be significantly higher in patients presenting a PM phenotype than in EMs (CYP2C19) (Baumann et al., 1996). However, in vitro, CYP2C19 and CYP3A4 show stereoselectivity towards (S)-citalopram, while some authors do not agree about a stereoselective activity of CYP2D6 towards citalopram (Rochat et al., 1997; Olesen and Linnet, 1999). Recent in vitro studies using human liver microsomes and expressed cytochromes carried out with escitalopram confirm the important role of CYP2C19, CYP2D6 and CYP3A4 in its N-demethylation (von Moltke et al., 2001).
Monoamine oxidase A and B (MAO-A and MAO-B) isoforms in human liver (Rochat et al., 1998) and in human brain (Kosel et al., 2002) metabolized citalo- pram stereoselectively (Fig. 9.1), but the pharmacogenetics of the metabolism of citalopram by polymorphic MAO has not yet been studied (Shih et al., 1999).
Fluoxetine
The chiral antidepressant fluoxetine and its metabolite, norfluoxetine, are both selective 5-HT uptake inhibitors. However, (S)-norfluoxetine (seproxetine) (Robertson et al., 1991) is considerably more potent than (R)-norfluoxetine as an eudismic ratio of 22 was found for 5-HT uptake inhibition in in vitro synaptoso- mal preparations (Wong et al., 1993). (S)-Norfluoxetine was also more potent than the (R)-enantiomer in blocking the in vivo depletion of 5-HT by p-chloroamphet- amine (Fuller et al., 1992) and it had a higher affinity than (R)-fluoxetine for opioid 1 receptors (Narita et al., 1996). (S)- and (R)-Fluoxetine, in contrast, do not notably differ in their inhibition of paroxetine binding at 5-HT uptake carrier sites (Wong et al., 1993) and in 5-HT uptake inhibition in vitro, in vivo, and ex vivo (Wong et al., 1990a). In contrast, (R)-norfluoxetine is more potent than the (S)-enantiomer in displacing ligands to the 5-HT1C and 5-HT2-receptors (Wong et al., 1993). As an anorectic agent in animals, (S)-fluoxetine is slightly more potent than (R)-fluoxetine (Wong et al., 1990b). Therefore, (S)- and (R)-fluoxetine and (S)-norfluoxetine, by their 5-HT uptake inhibition potency, are considered as the pharmacologically relevant enantiomers, in contrast to the less potent (R)-norfluoxetine.
After administration of the enantiomers of norfluoxetine, (S)-norfluoxetine was eliminated faster from rat brain than (R)-norfluoxetine (Fuller et al., 1992).
190 P. Baumann and C. B. Eap
Separate injections (10 mg/kg intraperitoneally) of the enantiomers of fluoxetine in mice led to a longer persistance of 5-HT uptake inhibition by (S)-fluoxetine, despite the fact that its metabolite (S)-norfluoxetine is eliminated more rapidly than (R)-norfluoxetine after (R)-fluoxetine administration (Fuller and Snoddy, 1993). This is explained by the relative weakness of (R)-norfluoxetine as a 5-HT uptake inhibitor.
In vitro studies, using achiral analytical procedures, of N-desmethylation of flu- oxetine to norfluoxetine in human liver microsomes and in microsomes from transfected cell lines expressing different cytochrome P-450 isoforms suggest that CYP2C9 is the main enzyme implicated. CYP2C19 and CYP3A4 play a minor role, while CYP2D6 is seemingly not implicated (von Moltke et al., 1997). However, in vitro, there is a significant correlation between microsomal immunodetectable CYP2D6 in human liver and N-desmethylation rates of (R)- and (S)-fluoxetine to their respective norfluoxetine metabolites. By comparison, CYP2D6 concentra- tions correlated better with bufuralol 1 -hydroxylation, which is a specific indica- tor for CYP2D6 activity, and quinidine only partly inhibited desmethylation of fluoxetine enantiomers. This suggests that other enzymes in addition to CYP2D6 largely contribute to the formation of norfluoxetine (Stevens and Wrighton, 1993).
Different results were obtained in vivo, using an achiral method, in a panel study with healthy human volunteers who were administered a single dose of fluoxetine. In this study, the terminal half-life of fluoxetine was considerably longer in PMs (76h) than in EMs (24h) of debrisoquine. The observation that partial metabolic clearance of fluoxetine into norfluoxetine was 10 times smaller in PMs than in EMs suggests that N-desmethylation of fluoxetine is at least partially controlled by CYP2D6 (Hamelin et al., 1996). Another panel study demonstrated clear evidence of a stereoselective elimination of fluoxetine and norfluoxetine by CYP2D6. A single oral dose of 60mg fluoxetine was administered to six EMs and six PMs of debrisoquine and the elimination kinetics of fluoxetine, norfluoxetine and their enantiomers were measured over a period of 6 weeks. Oral clearance (plasma half- life) of (R)- and (S)-fluoxetine was 3.0 l/h (9.5 days) and 17 l/h (6.1 days), respec- tively, in the PMs, while the corresponding figures were 36 l/h (2.6 days) and 40 l/h (1.1 days) in the EMs (Fjordside et al., 1999). The plasma elimination half-lives of (R)- and (S)-norfluoxetine were the same in EMs (5.5 days), but were 6.9 days and 17.4 days, respectively in PMs (Fjordside et al., 1999). These data suggest that, at least partly, CYP2D6 controls the biotransformation of (R)- and of (S)-fluoxetine, and (S)-norfluoxetine, but not of (R)-norfluoxetine. However, a recent study with patients during a 3-week treatment with fluoxetine (20 mg/day) showed that only plasma concentrations of (S)-fluoxetine (Fig. 9.2a) and (S)-norfluoxetine (Fig. 9.3a) differed significantly between PMs and EMs, whereas those of (R)-fluoxetine (Fig. 9.2b) and (R)-norfluoxetine (Fig. 9.3b) did not (Eap et al., 2001a).
191 Pharmacogenetics of chiral psychotropic drugs
(a) (a)
100
80
60
40
20
0
25
20
15
10
5
0
NS (p = 0.068)
  
PM
EM
HT EM HM EM
p = 0.014
                            
p = 0.037
                                                                                                                                          
(b) (b)
0 7 14 23 Time (days)
 
NS
      
NS
           
NS
                                                                                                      
Fig. 9.2.
Trough plasma concentrations of (S)-fluoxetine (a) and (R)-fluoxetine (b) measured in three CYP2D6 poor metabolizers (PM), eight extensive metabolizers (EM), two heterozygous extensive metabolizers (HT EM) and six homozygous extensive metabolizers (HM EM) after administration of 20mg racemic fluoxetine during 7, 14, and 23 days. The p values are given for comparison of poor metabolizers with extensive metabolizers; NS, not significant.
0 7 14 23 Time (days)
(R)-Fluoxetine (ng/ml) (R)-Fluoxetine (ng/ml)
192 P. Baumann and C. B. Eap
(a) (a)
120
 
EM 100            HT EM
p = 0.038
PM
    
HM EM
                     
80 60 40 20
0
60
50
40
30
20
10
0
p = 0.014
                      
p = 0.036
                                                                   
(b) (b)
NS
0 7 14 23 Time (days)
 
NS
                                                          
NS
                                                             
Fig. 9.3.
Trough plasma concentrations of (S)-norfluoxetine (a) and (R)-norfluoxetine (b) measured in three CYP2D6 poor metabolizers (PM), eight extensive metabolizers (EM), two heterozygous extensive metabolizers (HT EM) and six homozygous extensive metabolizers (HM EM) after administration of 20 mg racemic fluoxetine during 7, 14, and 23 days. The p values are given for comparison of poor metabolizers with extensive metabolizers; NS, not significant.
0 7 14 23 Time (days)
(R)-Norfluoxetine (ng/ml)
(S)-Norfluoxetine (ng/ml)
193 Pharmacogenetics of chiral psychotropic drugs
Recent antidepressants
Venlafaxine
Venlafaxine is an antidepressant characterized by its reuptake inhibiting proper- ties for both 5-HT and norepinephrine in rat brain synaptosomes. The (R)-( )-enantiomer is somewhat more potent than (S)-( )-venlafaxine in inhib- iting 5-HT reuptake, while there is little stereoselectivity with regard to norepi- nephrine reuptake inhibition. In this regard, nothing is known about the stereoselectivity concerning the active metabolite O-desmethylvenlafaxine, the plasma concentrations of which may exceed those of the parent compound in clin- ical steady-state conditions (Holliday and Benfield, 1995). After a single dose of venlafaxine in healthy volunteers, there was apparently no stereoselective elimina- tion of venlafaxine enantiomers (Wang et al., 1992), but other studies are needed to confirm this finding (see below).
In vivo evidence for the metabolism of venlafaxine by CYP2D6 was shown in a panel study with healthy volunteers, who were treated with a single dose of venla- faxine. Mean oral clearance of venlafaxine was four times lower in PMs (23 8 l/h) than in EMs (100 62l/h). Coadministration of the CYP2D6 inhibitor did not modify venlafaxine clearance in PMs but reduced it in EMs to the value observed in PMs (Lessard et al., 1999). In a plasma concentrations versus clinical effective- ness study with venlafaxine in CYP2D6-genotyped depressive patients, none of the three PMs (CYP2D6) were responders (Veefkind et al., 2000). In comparison with EMs, they showed some decreased biotransformation of venlafaxine to O-desmethylvenlafaxine and an increase of N-desmethylation to the minor metabolite N-desmethylvenlafaxine. In vitro studies with human liver microsomes confirmed that CYP2D6 and CYP3A4 control venlafaxine O-desmethylation and venlafaxine N-desmethylation, respectively (Otton et al., 1996). However, the ques- tion of stereoselectivity of this biotransformation appears to be complicated by the fact that, at high saturating concentrations, (R)-( )-venlafaxine is mainly O-desmethylated, while at lower, non-saturating concentrations, the stereoselect- ivity is reversed. At any dose studied, (S)-( )-venlafaxine is faster N-desmethylated than (R)-( )-venlafaxine (Otton et al., 1996). In a clinical case study with an EM patient medicated with venlafaxine and other drugs differing by their CYP2D6- inhibiting potency, a similar dose-dependent influence of CYP2D6 in the stereose- lective biotransformation of venlafaxine was observed (Eap et al., 2000b). Recent studies suggest that, besides CYP2C9 and CYP3A4, CYP2C19 may also contribute to N-desmethylation of venlafaxine, but no observations concerning stereoselectiv- ity of this reaction are available (Fukuda et al., 2000).
194 P. Baumann and C. B. Eap
Other recent chiral antidepressants
For other chiral antidepressants, such as bupropion, milnacipran, viloxazine, very few data are available with regard to a possible role of CYP2D6 or CYP2C19 in their metabolism or to some stereoselectivity in their fate (Rotzinger et al., 1999).
Milnacipran does not seem to have active metabolites, and the role of cyto- chrome P-450 in their formation is thought to be negligible (Puozzo and Leonard, 1996). The elimination of -milnacipran seems to be slower than that of -mil- nacipram (Spencer and Wilde, 1998).
Reboxetine, a selective norepinephrine reuptake-inhibiting antidepressant (Holm and Spencer, 1999), carries two asymmetric centers. A mixture of (R,R)- ( )- and (S,S)-( )-reboxetine has been introduced in clinical practice. (S,S)-( )- Reboxetine is more potent in inhibiting norepinephrine reuptake but plasma concentrations of (R,R)-( )-reboxetine are generally twice as high as those of (S,S)-( )-reboxetine, which is eliminated faster, as shown in single dose studies with healthy volunteers (Fleishaker et al., 1999). CYP3A4 is probably the main enzyme regulating their metabolism (Herman et al., 1999), while in vitro (Herman et al., 1999) and in vivo (Avenoso et al., 1999) studies suggest that reboxetine is not a substrate of CYP2D6, nor of other forms of cytochrome P-450.
No data are available on the role of cytochrome P-450 and on the stereoselective metabolism of the enantiomers of the antidepressant viloxazine (Rotzinger et al., 1999).
( )-Bupropion is more potent than ( )-bupropion in uptake inhibition of nor- epinephrine and dopamine (Musso et al., 1993). CYP2B6 (Hesse et al., 2000) but not CYP2D6 (Pollock et al., 1996) appears to be the enzyme responsible for the hydroxylation of bupropion to its main metabolite, as shown by achiral methods.
Achiral antidepressants with chiral metabolites: amitriptyline and nortriptyline
The achiral antidepressant amitriptyline, its N-desmethylated metabolite nortrip- tyline and their 10-hydroxylated metabolites are all norepinephrine and 5-HT reuptake inhibitors (Nordin and Bertilsson, 1995). CYP2D6 is the main enzyme responsible for their hydroxylation, while CYP2C19, CYP3A4, CYP2C9, and CYP2D6 all contribute to N-desmethylation of amitriptyline to nortriptyline (Balant-Georgia et al., 1982; Mellström et al., 1983; Baumann et al., 1986; Breyer- Pfaff et al., 1992; Dahl et al., 1996; Schmider et al., 1996; Coutts et al., 1997; Ghahramani et al., 1997; Olesen and Linnet, 1997).
The introduction of a hydroxy group in the amitriptyline and nortriptyline molecules gives rise to two geometric isomers (E and Z) and four stereoisomers for each compound (Nusser et al., 1988) (Fig. 9.4), which differ in their molecular dynamics (Heimstad et al., 1992). They may undergo stereoselective glucuronida- tion before they are renally excreted, as shown for hydroxynortriptyline
195 Pharmacogenetics of chiral psychotropic drugs
           
C
H CH2
H2C
N
CH3 CH3
Amitriptyline
CYP3A4 CYP2C19 CYP1A2
CYP26D
(mainly for formation of the (–)-(E)-enantiomer)
(±)-Enantiomers of
(Z)- and (E)-10-hydroxyamitriptyline
H OH
  
H OH
                             
CC
    
H CH2 H2C
CH2
H
+ NN
   
CYP2D6
CYP2C9
H2C
  
CYP2D6
(mainly for formation of the (–)-(E)-enantiomer)
CH3
H CH3 H
(±)-Enantiomers of
(E )-10-hydroxynortriptyline
             
C
H CH2
H2C
N CH3 H
Nortriptyline
Desmethylnortriptyline
H OH
H OH
                                         
CC
H CH2 + H CH2
H2C H2C
NN CH3 H CH3 H
(±)-Enantiomers of
(Z )-10-hydroxynortriptyline
             
Fig. 9.4.
Metabolic pathways of nortriptyline to chiral metabolites. CYP, cytochrome P-450.
(Dahl-Puustinen et al., 1989). After administration of nortriptyline or amitripty- line in humans, there is a preferential formation of (E)-( )-10-hydroxymetabo- lites (Mellström et al., 1981; Young et al., 1988; Breyer-Pfaff et al., 1992). Pilot studies suggest that (E)-10-hydroxynortriptyline could be an antidepressant (Nordin et al., 1991). In the rat, (Z)-hydroxynortriptyline seems to be more cardi- otoxic than (E)-hydroxynortriptyline.
196 P. Baumann and C. B. Eap
Several reports demonstrate that, to a major extent, formation of (E)-( )-10- hydroxyamitriptyline and (E)-( )-10-hydroxynortriptyline, but not that of the other hydroxy metabolites, is stereoselectively controlled by CYP2D6 (Mellström et al., 1981; Dahl et al., 1991; Breyer-Pfaff et al., 1992; Pfandl et al., 1992; Nordin and Bertilsson, 1995). As an example, coadministration of the potent CYP2D6 inhibitor quinidine with nortriptyline decreased urinary excretion of (E)-( )-10- hydroxynortriptyline but not that of (E)-( )-10-hydroxynortriptyline, (Z)-( )- and (Z)-( )-10-hydroxynortriptyline in humans (Pfandl et al., 1992).
Antipsychotics
Chiral antipsychotics
Many antipsychotics are substrates of CYP2D6. At higher doses, which generally give rise to higher plasma concentrations, the risk for adverse effects of the extra- pyramidal type increases. Therefore, many recent studies have dealt with the rela- tionship between neuroleptic-induced movement disorders (Spina et al., 1992; Arthur et al., 1995; Andreassen et al., 1997; Armstrong et al., 1997; Chong, 1997; Sajjad, 1997) or a neuroleptic malignant syndrome (Ueno et al., 1996) and the CYP2D6 pharmacogenetic status of the patients.
Thioridazine
Thioridazine is one of the most frequently prescribed chiral antipsychotics, even though its use is related to the highest risks for cardiotoxicity (Reilly et al., 2000). This risk increases with increasing plasma concentrations of both the parent com- pound and some of its metabolites (Hartigan-Go et al., 1996). The metabolism of thioridazine is complicated by the fact that it is a racemic compound and because a second asymmetric center is introduced by its sulfoxidation to the active metabo- lite mesoridazine (thioridazine 2-sulfoxide) (Fig. 9.5). It is then further metabo- lized to sulforidazine (thioridazine 2-sulfone), which, together with thioridazine and mesoridazine, is also introduced as an antipsychotic agent in some countries. Thioridazine-ring sulfoxide (thioridazine 5-sulfoxide) is another important but cardiotoxic metabolite with four enantiomers (Gottschalk et al., 1978). Finally, the tertiary amine thioridazine and the sulfoxidized metabolites are also N-desmethy- lated.
(R)-( )-Thioridazine (Patrick and Singletary, 1991) increased the turnover of dopamine in the striatum of rat brain 4.1 times more potently than (S)-( )-thior- idazine, despite both enantiomers being present in similar concentrations in the brain after peripheral administration of the enantiomers of thioridazine. In con- trast, ( )-thioridazine is slightly more potent than the ( )-enantiomer in produc- ing catalepsy and it seems also to be more toxic at higher doses. However, racemic

197 Pharmacogenetics of chiral psychotropic drugs

Fig. 9.5. Metabolic pathways of thioridazine.
198 P. Baumann and C. B. Eap
thioriazine appeared to be more cataleptogenic than either enantiomer alone, at equimolar concentrations. Both D1 and D2 receptors seem to be involved in this behavioral effect. Indeed, in vitro studies with rat forebrain tissues showed that the Ki-ratios (( )-thioridazine/( )-thioridazine) with regard to the affinity for differ- ent receptors were 2.71, 0.0978, 4.46, and 1.06 for D2 receptors, D1 receptors, 1- adrenoceptors and muscarinic receptors, respectively (Svendsen et al., 1988). From these data, it seems difficult to define the eutomer clearly, as ( )-thioridazine and ( )-thioridazine appeared to be rather selective D2 antagonists and D1 antagonists, respectively. With regard to the risk of cardiotoxic effects, thioridazine 5-sulfoxide rather than thioridazine itself could be the arrhythmogenic compound, but this effect is probably not stereoselective (Hale and Poklis, 1996).
The pharmacogenetics of the metabolism of thioridazine in CYP2D6-pheno- typed subjects was studied, using achiral methods, for the drug and its metabolites mesoridazine, sulforidazine, and thioridazine 5-sulfoxide (von Bahr et al., 1991). According to this study, the formation of mesoridazine, and partly that of the ring sulfoxide from thioridazine, appears to be controlled by CYP2D6. In a clinical study, only 1 of 25 schizophrenic and debrisoquine-phenotyped patients was a PM. They were treated with 400mg/day thioridazine for 10 days, but the patient with a genetic CYP2D6 deficiency suffered from severe adverse effects and was a nonre- sponder. His daily dose of thioridazine was lowered to 100mg, but in comparison with the other patients, his plasma levels of thioridazine were highest and those of mesoridazine and sulforidazine were lowest (Meyer et al., 1990). It was proposed to use the ratio of mesoridazine/thioridazine in plasma as a marker of CYP2D6 activity in subjects “phenotyped” with thioridazine (Llerena et al., 2000).
In a clinical study comparing the clinical effect of moclobemide/placebo versus moclobemide/thioridazine in 21 patients phenotyped with dextromethorphan and mephenytoin, the patients received orally 100 mg/day thioridazine for 14 days (Eap et al., 1996). Thioridazine, thioridazine 2-sulfoxide, thioridazine 2-sulfone, and thioridazine 5-sulfoxide in plasma were determined by a stereoselective high-pres- sure liquid chromatographic method. The results suggested that CYP2D6 is involved in the formation of two of the four enantiomers of mesoridazine, namely (S)-thioridazine 2-sulfoxide (a fast eluting band (FE)) and (R)-thioridazine 2-sulfoxide (a slow eluting band (SE)) and probably also in that of (S)-thioridazine 5-sulfoxide (FE), and (R)-thioridazine 5-sulfoxide (SE). (This denomination of stereoisomers is explained by the unavailability of pure enantiomers (Eap et al., 1991).) Unexpectedly, thioridazine was also found to be metabolized by CYP2C19. CYP2C19 activity, as measured with the mephenytoin test, correlated significantly with plasma concentrations of thioridazine, thioridazine 2-sulfone, thioridazine 2-sulfoxide (FE and SE) and thioridazine 5-sulfoxide (FE and SE).
199 Pharmacogenetics of chiral psychotropic drugs
Sulpiride
Both by affinity studies for the D2 receptor and by behavioral tests in animals, (S)- ( )-sulpiride rather than (R)-( )-sulpiride appears to be the antipsychotically active enantiomer (Rognan et al., 1990). The (S)-( )-enantiomer is eight times more potent than racemic sulpiride in producing stimulation of prolactin in rats (Kakigi et al., 1992). (S)-( )-Sulpiride, also called levosulpiride, has been intro- duced as an antipsychotic drug. After intravenous administration of racemic sul- piride, there is no evidence for stereoselective pharmacokinetics of this drug, the metabolism of which is probably not submitted to a genetically polymorphic metabolism (Wagstaff et al., 1994). After oral administration of sulpiride, the ratio /-sulpiride in serum of schizophrenic patients was 1 (range 0.66–0.97), in steady-state conditions (Müller et al., 2001).
Achiral antipsychotics with chiral metabolites
Haloperidol
The most important metabolic pathways of haloperidol implicate its glucurocon- jugation and formation of reduced haloperidol, which is then partly backoxidized to the parent compound. CYP2D6 and CYP3A4 contribute to hydroxylation and dealkylation of haloperidol and reduced haloperidol (Kudo and Ishizaki, 1999). Reduced haloperidol lacks potent D2 antagonism, but it has a high affinity for - receptors (Walker et al., 1990; Quirion et al., 1992; Bowen et al., 1995). Panel studies with haloperidol-treated PMs and EMs of debrisoquine showed that plasma half-life of haloperidol was significantly longer in PMs and its clearance was lower than in EMs (Llerena et al., 1992a). In the same study, plasma concentrations of reduced haloperidol were found to be higher in PMs than in EMs (Llerena et al., 1992b). Other authors excluded a leading role for CYP2D6 in the reversible inter- conversion between haloperidol and reduced haloperidol (Young et al., 1993), but, apparently, CYP3A4 could be involved in oxidation of reduced haloperidol (Kudo and Odomi, 1998; Pan et al., 1998).
All of these studies do not take account of the fact that reduced haloperidol has a chiral center, the enantiomers of which differ stereoselectively in their in vitro for- mation in human brain, liver, and blood: (S)-( )-reduced haloperidol is formed almost exclusively (about 99%) (Eyles and Pond, 1992). However, in patients taking haloperidol, about 25% of reduced haloperidol excreted in urine is the (R)-enantiomer (Eyles et al., 1998). No data are available on the stereoselective backoxidation of reduced haloperidol to haloperidol, especially with regard to the contribution of cytochrome P-450.
200 P. Baumann and C. B. Eap
Risperidone
The atypical antipsychotic drug risperidone and its chiral metabolite 9-hydroxyrisperidone are both potent D2 and 5-HT2 antagonists. Therefore, the sum of risperidone and 9-hydroxyrisperidone is considered to constitute the “active moiety” (Huang et al., 1993). This hypothesis is supported by proton emis- sion tomographic studies carried out with PMs and EMs (CYP2D6) (Nyberg et al., 1995), but the antipsychotic effect of 9–hydroxyrisperidone given alone has not been evaluated in clinical studies.
In vitro studies with recombinant human cytochrome P-450 forms show that ris- peridone is metabolized both by CYP2D6 and by CYP3A4 to 9-hydroxyrisperidone. At least in the rat, CYP3A4 inducers can increase formation of 9-hydroxyrisperidone severalfold (Fang et al., 1999). Plasma half-life of risperidone was about 3 h and 22 h in EMs and PMs of debrisoquine, respectively, as shown in a small panel study. Plasma half-life of 9-hydroxyrisperidone is very similar in EMs and PMs (20–27h) and that of the “active moiety” is also close to 20 h (Huang et al., 1993). In another study, dose-corrected concentrations of the “active moiety” were not higher in PMs than in EMs (Olesen et al., 1998). Therefore, it is thought that there is little need for adapting risperidone dose as a function of the pharmacogenetic status. However, a recent case study suggested that CYP2D6-deficient patients do not tolerate risperi- done treatment well, despite “normal active moiety” concentrations (Bork et al., 1999). Data collected in UMs showed that dose-corrected plasma concentrations of risperidone were particularly low, compared with PMs and heterozygous EMs (Scordo et al., 1999). None of these studies considered the possibility that there are seemingly differences in the stereoselective hydroxylation of risperidone by CYP2D6 and CYP3A4 (L. Bertilsson et al., unpublished data), and there are no data available on the pharmacological profile of the individual enantiomers of 9-hydroxyrisperi- done. However, in a recent in vitro study with human liver microsomes and recom- binantly expressed enzymes, ( )-hydroxyrisperidone formation was found to be higher than that of the ( )-enantiomer. CYP2D6 preferentially formed ( )- hydroxyrisperidone, while CYP3A4 mainly catalyzed risperidone biotransformation to (8)-hydroxyrisperidone (Yasui-Furukori et al., 2001). In extensive metabolizers by CYP2D6, plasma concentrations of ( )-hydroxyrisperidone were higher than those of the ( )-hydroxy-enantiomer, in patients treated with risperidone.
Tranquillizers and hypnotic agents Oxazepam
Oxazepam, a chiral anxiolytic drug, is mainly metabolized by glucuronidation. A stereoselective analysis of the metabolism of oxazepam appears to be a very difficult task as its enantiomers racemize spontaneously and rapidly (Yang and Lu, 1989).

201 Pharmacogenetics of chiral psychotropic drugs
Zopiclone
Zopiclone is a racemic hypnotic drug, of which the ( )-enantiomer has a 50 times higher affinity for the benzodiazepine receptor-binding site than the ( )-enantiomer (Blaschke et al., 1993). After a single dose of zopiclone, peak con- centrations of ( )-zopiclone in plasma of volunteers appeared to be higher than those of ( )-zopiclone. The former enantiomer is eliminated more slowly than the ( )-enantiomer (Fernandez et al., 1993). No pharmacogenetic data are available, as CYP3A4 and CYP2C8, in contrast to CYP2C19 and CYP2D6, are found to metabolize zopiclone in vitro (Becquemont et al., 1999). The chiral benzodiaze- pines zopiclone and oxazepam represent examples of compounds that, at least to some extent, undergo spontaneous racemization in aqueous solutions (Fernandez et al., 1995).
Stimulants and illicit drugs
The metabolism of dexfenfluramine was examined in a panel study with healthy CYP2D6-phenotyped subjects after administration of a dose of 30mg of this ano- rectic drug (Gross et al., 1996). Mean plasma AUC was about twice as high in PMs than in EMs. The apparent oral clearance was higher in EMs, but there was no difference between the groups for renal clearance. The data suggest that CYP2D6 mediates N-desmethylation of dexfenfluramine to nordexfenfluramine as appar- ent nonrenal clearance of the metabolite was considerably lower in PMs than in EMs. PMs were more likely to experience serotonergic adverse effects, such as nausea and vomiting, than were EMs.
Amphetamine derivatives such as MDMA (Ecstasy) and MDE (Eve) present a chiral structure. Their main metabolic pathways are desmethylation and N-dealkylation. Besides other forms of cytochrome P-450, CYP2D6 contributes to their metabolism (Maurer et al., 2000). Very little is known about the stereoselect- ive metabolism and pharmacogenetics of these and other illicit drugs (Quinn et al., 1997).
Opioid substituents
Methadone, an opioid agonist, is used for the treatment of opioid dependence. In vitro binding experiments have shown that the necessary concentration of (R)- or l-methadone to inhibit by 50% the binding of [3H]-naloxone to whole rat brain homogenate is 10 times less than that of (S)- or d-methadone (Pert and Snyder, 1973). A 10-fold difference of affinity has also been found between the two enan- tiomers for the 1 and 2 receptors (concentrations for 50% inhibition (IC50) of 3.0 and 6.9nmol/l, respectively, for (R)-methadone and 26.4 and 88nmol/l,
 
202 P. Baumann and C. B. Eap
respectively, for (S)-methadone for the 1- and 2-receptors, respectively), using preparations from bovine caudate nucleus (Kristensen et al., 1995). In humans, (R)-methadone is about 50 times as analgesically potent as the (S)-form (Scott et al., 1948) and it accounts for the large majority, if not all, of opioid effects of racemic methadone. Both enantiomers of methadone exhibit similar affinities for the N-methyl--aspartate (Ki of 3.4 and 7.4μmol/l for (R)- and (S)-methadone, respectively) (Gorman et al., 1997).
Methadone is extensively metabolized in the body, mainly at the level of the liver, but probably also by intestinal CYP3A4. Its main metabolite (2-ethylidene-1, 5-dimethyl-3,3-diphenylpyrrolidine or EDDP) is inactive: it is formed by N-desmethylation and subsequent spontaneous cyclization (Sullivan and Due, 1973). However, urinary excretion of methadone plus EDDP only accounted for 17–57% of the given dose and, in addition to methadone, seven metabolites (including EDDP) have been isolated and identified in urine (Ånggard et al., 1975). In vitro and in vivo studies have shown that CYP3A4 (Iribarne et al., 1996, 1998; Moody et al., 1997; Foster et al., 1999) and CYP2D6 (Yue et al., 1995; Eap et al., 1997) are involved in methadone metabolism. Other isoforms, such as CYP1A2 (Yue et al., 1995; Eap et al., 1997), CYP2C9 (Foster et al., 1999), and CYP2C19 (Foster et al., 1999) might also be implicated, but their in vivo relevance has still to be demonstrated. Methadone N-desmethylation, which is mediated principally by CYP3A4, is probably not stereoselective (Foster et al., 1999). However, CYP2D6, which is probably involved in another metabolic pathway, might preferentially metabolize the (R)-enantiomer, as suggested by in vivo inhibition studies with flu- oxetine and paroxetine, two strong CYP2D6 inhibitors, and by a panel study with CYP2D6 PMs and EMs, which showed a significantly lower partial metabolic clear- ance of (R)-methadone in the PMs compared with the EMs (Yue et al., 1995; Eap et al., 1997). In a very recent clinical study, the pharmacokinetic and pharmacoge- netic data obtained suggest that the CYP2D genotype partly codetermines clinical response to methadone. (R)-Methadone plasma concentrations were significantly lower in ultrarapid metabolizers than in poor metabolizers (Eap et al., 2001b). A detailed review on the pharmacokinetics, pharmacodynamics, and pharmacoge- netics of methadone can be found elsewhere (Eap et al., 2001).
Conclusions
Chiral psychotropic drugs represent a useful tool for the study of neuropharmaco- logical mechanisms and of the function of metabolizing enzymes. Many widely used drugs are introduced as racemic compounds, the enantiomers of which clearly differ in their pharmacology, metabolism, and pharmacogenetics. With regard to their fate in the organism, this presentation is centered exclusively on metabolic

203 Pharmacogenetics of chiral psychotropic drugs
aspects. It should be considered, however, that other aspects such as drug plasma protein binding may also be genetically and stereoselectively determined. As an example, the enantiomers of methadone stereoselectively bind to 1-acid glycopro- tein, of which one particular variant (A) preferentially binds basic drugs (Eap et al., 1990; Eap and Baumann, 1990; Jolliet-Riant et al., 1998). This genetically deter- mined variant A may partially control transfer of drugs from blood to brain.
This overview confirms that the use of achiral analytical procedures can lead to erroneous conclusions with regard to the pharmacogenetics of metabolism of psychotropic drugs (Nation, 1994), as shown by the example of mirtazapine. As a consequence, pharmacogenetic studies should include the use of stereoselective methods to analyze the fate of chiral psychotropic drugs in the organism.
Acknowledgements
We gratefully acknowledge the editorial assistance of Mrs C. Bertschi, Mrs K. Powell, and the bibliographic help of Mrs E. Ponce.
REFERENCES
Andreassen OA, MacEwan T, Gulbrandsen A-K, McCreadie RG, Steen VM (1997). Non- functional CYP2D6 alleles and risk for neuroleptic-induced movement disorders in schizo- phrenic patients. Psychopharmacology 131, 174–179.
Armstrong M, Daly AK, Blennerhassett R, Ferrier N, Idle JR (1997). Antipsychotic drug-induced movement disorders in schizophrenics in relation to CYP2D6 genotype. Br J Psychiatry 170, 23–26.
Arthur H, Dahl M-L, Siwers B, Sjöqvist F (1995). Polymorphic drug metabolism in schizophrenic patients with tardive dyskinesia. J Clin Psychopharmacol 15, 211–216.
Avenoso A, Facciolï G, Scordo MG, Spina E (1999). No effect of the new antidepressant reboxe- tine on CYP2D6 activity in healthy volunteers. Ther Drug Monit 21, 577–579.
Ånggard E, Gunne L-M, Holmstrand J, McMahon RE, Sandberg C-G, Sullivan HR (1975). Disposition of methadone in methadone maintenance. Clin Pharmacol Ther 17, 258–266. Baker GB, Coutts RT, Holt A (1994). Metabolism and chirality in psychopharmacology. Biol
Psychiatry 36, 211–213.
Balant-Georgia AE, Schulz P, Dayer P et al. (1982). Role of the oxidation polymorphism on blood
and urine concentrations of amitriptyline and its metabolites in man. Arch Psychiatr
Nervenkrank 232, 215–222.
Baumann P, Eap CB (2001). Enantiomeric antidepressant drugs should be considered on indi-
vidual merit. Hum Psychopharmacol Clin Exp 16, S85–S92.
Baumann P, Rochat B (1995). Comparative pharmacokinetics of selective serotonin reuptake
 
inhibitors: a look behind the mirror. Int Clin Psychopharmacol 10(Suppl 1), 15–21.
204 P. Baumann and C. B. Eap
Baumann P, Jonzier-Perey M, Koeb L, Küpfer A, Tinguely D, Schöpf J (1986). Amitriptyline phar- macokinetics and clinical response: II. Metabolic polymorphism assessed by hydroxylation of debrisoquine and mephenytoin. Int Clin Psychopharmacol 1, 102–112.
Baumann P, Meyer JW, Amey M et al. (1992). Dextromethorphan and mephenytoin phenotyp- ing of patients treated with thioridazine or amitriptyline. Ther Drug Monit 14, 1–8.
Baumann P, Nil R, Souche A et al. (1996). A double-blind, placebo-controlled study of citalo- pram with and without lithium in the treatment of therapy-resistant depressive patients: a clinical, pharmacokinetic, and pharmacogenetic investigation. J Clin Psychopharmacol 16, 307–314.
Beauchamp G, Lavoie P-A, Elie R (1992). Effect of trimipramine on depolarization-induced and Na –Ca2 exchange-induced 45calcium uptake in synaptosomes from the cortex of the rat brain. Neuropharmacology 31, 229–234.
Becquemont L, Mouajjah S, Escaffre O, Beaune P, Funck-Brentano C, Jaillon P (1999). Cytochrome P-450 3A4 and 2C8 are involved in zopiclone metabolism. Drug Metab Dispos 27, 1068–1073.
Berger M, Gastpar M (1996). Trimipramine: a challenge to current concepts on antidepressives. Eur Arch Psychiatry Clin Neurosci 246, 235–239.
Bertilsson L, Dahl M-L (1996). Polymorphic drug oxidation – relevance to the treatment of psychiatric disorders. CNS Drugs 5, 200–223.
Bertilsson L, Dahl M-L, Sjöqvist F et al. (1993). Molecular basis for rational megaprescribing in ultrarapid hydroxylators of debrisoquine. Lancet 341, 63.
Bertilsson L, Dahl M-L, Tybring G (1997). Pharmacogenetics of antidepressants: clinical aspects. Acta Psychiatr Scand 96(Suppl 391), 14–21.
Blaschke G, Hempel G, Müller WE (1993). Preparative and analytical separation of the zopiclone enantiomers and determination of their affinity to the benzodiazepine receptor binding site. Chirality 5, 419–421.
Bondolfi G, Chautems C, Rochat B, Bertschy G, Baumann P (1996). Non-response to citalopram in depressive patients: pharmacokinetic and clinical consequences of a fluvoxamine augmen- tation. Psychopharmacology 128, 421–425.
Bondolfi G, Lissner C, Kosel M, Eap CB, Baumann P (2000). Fluoxetine augmentation in citalo- pram nonresponders: pharmacokinetic and clinical consequences. Int J Neuropsycho- pharmacol 3, 55–60.
Bork JA, Rogers T, Wedlund PJ, de Leon J (1999). A pilot study on risperidone metabolism: the role of cytochromes P-450 2D6 and 3A. J Clin Psychiatry 60, 469–476.
Bowen WD, Bertha CM, Vilner BJ, Rice KC (1995). CB-64D and CB-184: ligands with high sigma(2) receptor affinity and subtype selectivity. Eur J Pharmacol 278, 257–260.
Breyer-Pfaff U, Pfandl B, Nill K et al. (1992). Enantioselective amitriptyline metabolism in patients phenotyped for two cytochrome P-450 isozymes. Clin Pharmacol Ther 52, 350–358.
Chong S-A (1997). Tardive dyskinesia and CYP2D6 polymorphism in Chinese. Br J Psychiatry 171, 586.
Chow T, Hiroi T, Imaoka S, Chiba K, Funae Y (1999). Isoform-selective metabolism of mianse- rin by cytochrome P-450 2D. Drug Metab Dispos 27, 1200–1204.
205 Pharmacogenetics of chiral psychotropic drugs
Coutts RT, Urichuk LJ (1999). Polymorphic cytochromes P-450 and drugs used in psychiatry. Cell Mol Neurobiol 19, 325–354.
Coutts RT, Bach MV, Baker GB (1997). Metabolism of amitriptyline with CYP2D6 expressed in a human cell line. Xenobiotica 27, 33–47.
Dahl M-L, Sjöqvist F (2000). Pharmacogenetic methods as a complement to therapeutic moni- toring of antidepressants and neuroleptics. Ther Drug Monit 22, 114–117.
Dahl-Puustinen ML, Dumont E, Bertilsson L (1989). Glucuronidation of E-10-hydroxynor- triptyline in human liver, kidney, and intestine. Drug Metab Dispos 17, 433–436.
Dahl M-L, Nordin C, Bertilsson L (1991). Enantioselective hydroxylation of nortriptyline in human liver microsomes, intestinal homogenate, and patients treated with nortriptyline. Ther Drug Monit 13, 189–194.
Dahl M-L, Tybring G, Elwin C-E et al. (1994). Stereoselective disposition of mianserin is related to debrisoquine hydroxylation polymorphism. Clin Pharmacol Ther 56, 176–183.
Dahl M-L, Bertilsson L, Nordin C (1996). Steady-state plasma levels of nortriptyline and its 10-hydroxy metabolite: relationship to the CYP2D6 genotype. Psychopharmacology 123, 315–319.
Dahl M-L, Voortman G, Alm C et al. (1997a). In vitro and in vivo studies on the disposition of mirtazapine in humans. Clin Drug Invest 13, 37–46.
Dahl M-L, Voortman G, Alm C et al. (1997b). In vitro and in vivo studies on the disposition of mirtazapine in humans. Clin Drug Invest 13, 37–46.
Dalén P, Dahl M-L, Bernal Ruiz ML, Nordin J, Bertilsson L (1998). 10-Hydroxylation of nortrip- tyline in white persons with 0, 1, 2, 3, and 13 functional CYP2D6 genes. Clin Pharmacol Ther 63, 444–452.
David R, Wilde MI (1996). Mirtazapine – a review of its pharmacology and therapeutic poten- tial in the management of major depression. CNS Drugs 5, 389–402.
Delbressine LPC, Moonen MEG, Kaspersen FM et al. (1998). Pharmacokinetics and biotransfor- mation of mirtazapine in human volunteers. Clin Drug Invest 15, 45–55.
Dodd S, Burrows GD, Norman TR (2000). Chiral determination of mirtazapine in human blood plasma by high-performance liquid chromatography. J Chromatogr B 748, 439–443.
Dodd S, Boulton DW, Burrows GD, de Vane CL, Norman TR (2001). In vitro metabolism of mir- tazapine enantiomers by human cytochrome P450 enzymes. Hum Psychopharmacol 6, 541–544. Eap CB, Baumann P (1990). Pharmacogenetics of drug binding to albumin and alpha1-acid gly- coprotein. In Belpaire F, Bogaert M, Tillement J-P, Verbeeck R, eds. Plasma Binding of Drugs
and its Consequences. Ghent: Academia Press, pp. 69–82.
Eap CB, Cuendet C, Baumann P (1990). Binding of d-methadone, l-methadone, and dl–
methadone to proteins in plasma of healthy volunteers: role of the variants of alpha1-acid gly-
coprotein. Clin Pharmacol Ther 47, 338–346.
Eap CB, Souche A, Koeb L, Baumann P (1991). Light-induced racemization: artifacts in the anal-
ysis of the diastereoisomeric pairs of thioridazine 5-sulfoxide in the plasma and urine of
patients treated with thioridazine. Ther Drug Monit 13, 356–362.
Eap CB, Koeb L, Holsboer-Trachsler E, Baumann P (1992a). Plasma levels of trimipramine and
metabolites in four patients: determination of the enantiomer concentrations of the hydroxy metabolites. Ther Drug Monit 14, 380–385.
206 P. Baumann and C. B. Eap
Eap CB, Laurian S, Souche A et al. (1992b). Influence of quinidine on the pharmacokinetics of trimipramine and on its effect on the waking EEG of healthy volunteers – a pilot study on two subjects. Neuropsychobiology 25, 214–220.
Eap CB, Powell K, Campus-Souche D et al. (1994). Determination of the enantiomers of mian- serin, desmethylmianserin, and 8-hydroxymianserin in the plasma and urine of mianserin- treated patients. Chirality 6, 555–563.
Eap CB, Guentert TW, Schäublin-Loidl M et al. (1996). Plasma levels of the enantiomers of thi- oridazine, thioridazine 2-sulfoxide, thioridazine 2-sulfone, and thioridazine 5-sulfoxide in poor and extensive metabolizers of dextromethorphan and mephenytoin. Clin Pharmacol Ther 59, 322–331.
Eap CB, Bertschy G, Powell K, Baumann P (1997). Fluvoxamine and fluoxetine do not interact in the same way with the metabolism of the enantiomers of methadone. J Clin Psychopharmacol 17, 113–117.
Eap CB, de Mendonça Lima CA et al. (1998). Steady state concentrations of the enantiomers of mianserin and desmethylmianserin in poor and in homozygous and heterozygous extensive metabolizers of debrisoquine. Ther Drug Monit 20, 7–13.
Eap CB, Yasui N, Kaneko S, Baumann P, Powell K, Otani K (1999). Effects of carbamazepine coadministration on plasma concentrations of the enantiomers of mianserin and of its metabolites. Ther Drug Monit 21, 166–170.
Eap CB, Bender S, Gastpar M et al. (2000a). Steady state plasma levels of the enantiomers of tri- mipramine and of its metabolites in CYP2D6-, CYP2C19- and CYP3A4/5-phenotyped patients. Ther Drug Monit 22, 209–214.
Eap CB, Bertel-Laubscher R, Zullino D, Amey M, Baumann P (2000b). Marked increase of ven- lafaxine enantiomer concentrations as a consequence of metabolic interactions: a case report. Pharmacopsychiatry 33, 112–115.
Eap CB, Bondolfi G, Zullino D, Savary L, Powell K, Kosel M, Baumann P (2001a). Concentrations of the enantiomers of fluoxetine and norfluoxetine after multiple dose of flu- oxetine in cytochrome P-4502D6 poor and extensive metabolizers. J Clin Psychopharmacol 21, 330–334.
Eap CB, Broly F, Mino A et al. (2001b). Cytochrome P450 2D6 genotype and methadone steady- state concentrations. J Clin Psychopharmacol 21, 229–234.
Eikmeier G, Muszynski K, Berger M, Gastpar M (1990). High-dose trimipramine in acute schizo- phrenia. Pharmacopsychiatry 23, 212–214.
Eyles DW, Pond SM (1992). Stereospecific reduction of haloperidol in human tissues. Biochem Pharmacol 44, 867–8712.
Eyles DW, McGrath JJ, Stedman TJ, Pond SM (1998). Chirality of reduced haloperidol in humans. Eur Neuropsychopharmacol 8, 127–129.
Fang J, Bourin M, Baker GB (1999). Metabolism of risperidone to 9-hydroxyrisperidone by human cytochromes P-450 2D6 and 3A4. Naunyn-Schmiedeberg’s Arch Pharmacol 359, 147–151.
Fernandez C, Maradeix V, Gimenez F, Thuillier A, Farinotti R (1993). Pharmacokinetics of zop- iclone and its enantiomers in Caucasian young healthy volunteers. Drug Metab Dispos 21, 1125–1128.
207 Pharmacogenetics of chiral psychotropic drugs
Fernandez C, Gimenez F, Mayrargue J, Thuillier A, Farinotti R (1995). Degradation and racemiza- tion of zopiclone enantiomers in plasma and partially aqueous solutions. Chirality 7, 267–271.
Fjordside L, Jeppesen U, Eap CB, Powell K, Baumann P, Brøsen K (1999). The stereoselective metabolism of fluoxetine in poor and extensive metabolizers of sparteine. Pharmacogenetics 9, 55–60.
Fleishaker JC, Mucci M, Pellizzoni C, Poggesi I (1999). Absolute bioavailability of reboxetine enantiomers and effect of gender on pharmacokinetics. Biopharm Drug Dispos 20, 53–57.
Foster DJR, Somogyi AA, Bochner F (1999). Methadone N-demethylation in human liver microsomes: lack of stereoselectivity and involvement of CYP3A4. Br J Clin Pharmacol 47, 403–412.
Fukuda T, Nishida Y, Zhou Q, Yamamoto I, Kondo S, Azuma J (2000). The impact of the CYP2D6 and CYP2C19 genotypes on venlafaxine pharmacokinetics in a Japanese population. Eur J Clin Pharmacol 56, 175–180.
Fuller RW, Snoddy HD (1993). Drug concentrations in mouse brain at pharmacologically active doses of fluoxetine enantiomers. Biochem Pharmacol 45, 2355–2358.
Fuller RW, Snoddy HD, Krushinski JH, Robertson DW (1992). Comparison of norfluoxetine enantiomers as serotonin uptake inhibitors in vivo. Neuropharmacology 31, 997–1000.
Ghahramani P, Ellis SW, Lennard MS, Ramsay LE, Tucker GT (1997). Cytochromes P-450 medi- ating the N-demethylation of amitriptyline. Br J Clin Pharmacol 43, 137–144.
Gorman AL, Elliott KJ, Inturrisi CE (1997). The d– and l-isomers of methadone bind to the non- competitive site on the N-methyl--aspartate (NMDA) receptor in rat forebrain and spinal cord. Neurosci Lett 223, 5–8.
Gottschalk LA, Dinovo E, Biener R, Nandi BR (1978). Plasma concentrations of thioridazine metabolites and EEG abnormalities. J Pharm Sci 67, 155–157,
Gross G, Xin X, Gastpar M (1991). Trimipramine: pharmacological reevaluation and compari- son with clozapine. Neuropharmacology 30, 1159–1166.
Gross AS, Phillips AC, Rieutord A, Shenfield GM (1996). The influence of the sparteine/debris- oquine genetic polymorphism on the disposition of dexfenfluramine. Br J Clin Pharmacol 41, 311–317.
Hale P, Poklis A (1996). Thioridazine cardiotoxicity. Clin Toxicol 34, 127–128.
Hamelin BA, Turgeon J, Vallée F, Bélanger P-M, Paquet F, Lebel M (1996). The disposition of flu- oxetine but not sertraline is altered in poor metabolizers of debrisoquine. Clin Pharmacol Ther
60, 512–521.
Hartigan-Go K, Bateman DN, Nyberg G, Martensson E, Thomas SHL (1996). Concentration-
related pharmacodynamic effects of thioridazine and its metabolites in humans. Clin
Pharmacol Ther 60, 543–553.
Heimstad E, Edvardsen O, Dahl SG (1992). Molecular structure and dynamics of the four 10-
hydroxynortriptyline isomers. Neuropsychopharmacology 6, 137–144.
Herman BD, Fleishaker JC, Brown MT (1999). Ketoconazole inhibits the clearance of the enan-
tiomers of the antidepressant reboxetine in humans. Clin Pharmacol Ther 66, 374–379. Hesse LM, Venkatakrishnan K, Court MH et al. (2000). CYP2B6 mediates the in vitro hydroxy- lation of bupropion: potential drug interactions with other antidepressants. Drug Metab
Dispos 28, 1176–1183.
208 P. Baumann and C. B. Eap
Holliday SM, Benfield P (1995). Venlafaxine – a review of its pharmacology and therapeutic potential in depression. Drugs 49, 280–294.
Holm KJ, Spencer CM (1999). Reboxetine – a review of its use in depression. CNS Drugs 12, 65–83.
Holm KJ, Markham A (1999). Mirtazapine – a review of its use in major depression. Drugs 57, 607–631.
Huang M-L, van Peer A, Woestenborghs R et al. (1993). Pharmacokinetics of the novel antipsy- chotic agent risperidone and the prolactin response in healthy subjects. Clin Pharmacol Ther 54, 257–268.
Hyttel J, Arnt J, Sanchez C (1995). The pharmacology of citalopram. Rev Contemp Pharmacother 6, 271–285.
Hyttel J, Bøgesø KP, Perregaard J, Sanchez C (1992). The pharmacological effect of citalopram resides in the (S)-( )-enantiomer. J Neural Transm 88, 157–160.
Ingelman-Sundberg M (1998). Functional consequences of polymorphism of xenobiotic metabolising enzymes. Toxicol Lett 102 – 103, 155–160.
Ingelman-Sundberg M, Oscarson M, McLellan RA (1999). Polymorphic human cytochrome P-450 enzymes: an opportunity for individualized drug treatment. Trends Pharmacol Sci 20, 342–349.
Iribarne C, Berthou F, Baird S et al. (1996). Involvement of cytochrome P-450 3A4 enzyme in the N-demethylation of methadone in human liver microsomes. Chem Res Toxicol 9, 365–373.
Iribarne C, Picart D, Dréano Y, Berthou F (1998). In vitro interactions between fluoxetine or flu- voxamine and methadone or buprenorphine. Fundam Clin Pharmacol 12, 194–199.
Johansson I, Lundqvist E, Bertilsson L, Dahl M-L, Sjöqvist F, Ingelman-Sundberg M (1993). Inherited amplification of an active gene in the cytochrome P-450 CYP2D locus as a cause of ultrarapid metabolism of debrisoquine. Proc Natl Acad Sci, USA 90, 11825–11829.
Jolliet-Riant P, Boukef MF, Duché J-C, Simon N, Tillement JP (1998). The genetic variant A of human alpha 1-acid glycoprotein limits the blood to brain transfer of drugs it binds. Life Sci 62, PL219–PL226.
Kakigi T, Maeda K, Tanimoto K, Kaneda H, Shintani T (1992). Effect of substituted benzamides on prolactin secretion in the rat. Biol Psychiatry 31, 827–831.
Kelder J, Funke C, de Boer T, Delbressine L, Leysen D, Nickolson V (1997). A comparison of the physicochemical and biological properties of mirtazapine and mianserin. J Pharm Pharmacol 49, 403–411.
Kobayashi K, Chiba K, Yagi T et al. (1997). Identification of cytochrome P-450 isoforms involved in citalopram N-demethylation by human liver microsomes. J Pharmacol Exp Ther 280, 927–933.
Kooyman AR, Zwart R, van der Heijden PML, van Hooft JA, Vijverberg HPM (1994). Interaction between enantiomers of mianserin and ORG3770 at 5-HT(3) receptors in cultured mouse neuroblastoma cells. Neuropharmacology 33, 501–507.
Kosel M, Gnerre C, Voirol P et al. (2002). In vitro biotransformation of the selective serotonin reuptake inhibitor citalopram, its enantiomers and demethylated metabolites by monoamine oxidase in rat and human brain preparations. Mol Psychiatry 7, 181–188.
209 Pharmacogenetics of chiral psychotropic drugs
Koyama E, Chiba K, Tani M, Ishizaki T (1996). Identification of human cytochrome P-450 iso- forms involved in the stereoselective metabolism of mianserin enantiomers. J Pharmacol Exp Ther 278, 21–30.
Kristensen K, Christensen CB, Christrup LL (1995). The mu1, mu2, delta, kappa opioid recep- tor binding profiles of methadone stereoisomers and morphine. Life Sci 56, 45–50.
Kudo S, Ishizaki T (1999). Pharmacokinetics of haloperidol – an update. Clin Pharmacokinet 37, 435–456.
Kudo S, Odomi M (1998). Involvement of human cytochrome P-450 3A4 in reduced haloperi- dol oxidation. Eur J Clin Pharmacol 54, 253–259.
Lane RM, Baker GB (1999). Chirality and drugs used in psychiatry: nice to know or need to know? Cell Mol Neurobiol 19, 355–372.
Lavoie P-A, Beauchamp G, Elie R (1994). Absence of stereoselectivity of some tricyclic antide- pressants inhibition of depolarization-induced calcium uptake in rat cingulate cortex synap- tosomes. J Psychiatry Neurosci 19, 208–212.
Lessard E, Yessine M-A, Hamelin BA, O’Hara G, LeBlanc J, Turgeon J (1999). Influence of CYP2D6 activity on the disposition and cardiovascular toxicity of the antidepressant agent venlafaxine in humans. Pharmacogenetics 9, 435–443.
Llerena A, Alm C, Dahl M-L, Ekqvist B, Bertilsson L (1992a). Haloperidol disposition is depen- dent on debrisoquine hydroxylation phenotype. Ther Drug Monit 14, 92–97.
Llerena A, Dahl M-L, Ekqvist B, Bertilsson L (1992b). Haloperidol disposition is dependent on the debrisoquine hydroxylation phenotype – increased plasma levels of the reduced metabo- lite in poor metabolizers. Ther Drug Monit 14, 261–264.
Llerena A, Berecz R, de la Rubia A, Norberto M-J, Benitez J (2000). Use of the mesoridazine/thi- oridazine ratio as a marker for CYP2D6 enzyme activity. Ther Drug Monit 22, 397–401.
Lundqvist E, Johansson I, Ingelman-Sundberg M (1999). Genetic mechanisms for duplication and multiduplication of the human CYP2D6 gene and methods for detection of duplicated CYP2D6 genes. Gene 226, 327–338.
Marzo A, Balant LP (1996). Investigation of xenobiotic metabolism by CYP2D6 and CYP2C19: importance of enantioselective analytical methods. J Chromatogr B: Biomed Appl 678, 73–92.
Maurer HH, Bickeboeller-Friedrich J, Kraemer T, Peters FT (2000). Toxicokinetics and analytical toxicology of amphetamine-derived designer drugs (“Ecstasy”). Toxicol Lett 112, 133–142.
McGrath C, Burrows GD, Norman TR (1998). Neurochemical effects of the enantiomers of mir- tazapine in normal rats. Eur J Pharmacol 356, 121–126.
Mellström B, Bertilsson L, Säwe J, Schulz H-U, Sjöqvist F (1981). E– and Z-10-hydroxylation of nortriptyline: relationship to polymorphic debrisoquine hydroxylation. Clin Pharmacol Ther 30, 189–193.
Mellström B, Bertilsson L, Sjöqvist F (1983). Amitriptyline metabolism: relationship to polymor- phic debrisoquine hydroxylation. Clin Pharmacol Ther 34, 516–520.
Meyer JW, Woggon B, Baumann P, Meyer UA (1990). Clinical implication of slow sulphoxida- tion of thioridazine in a poor metabolizer of the debrisoquine type. Eur J Clin Pharmacol 39, 613–614.
210 P. Baumann and C. B. Eap
Mihara K, Otani K, Tybring G, Dahl M-L, Bertilsson L, Kaneko S (1997). The CYP2D6 genotype and plasma concentrations of mianserin enantiomers in relation to therapeutic response to mianserin in depressed Japanese patients. J Clin Psychopharmacol 17, 467–471.
Montgomery SA, Loft H, Sanchez C, Reines EH, Papp M (2001). Escitalopram (S-enantiomer of citalopram): clinical efficacy and onset of action predicted from a rat model. Pharmacol Toxicol 88, 282–286.
Moody DE, Alburges ME, Parker RJ, Collins JM, Strong JM (1997). The involvement of cyto- chrome P-450 3A4 in the N-demethylation of L--acetylmethadol (LAAM), norlaam, and methadone. Drug Metab Dispos 25, 1347–1353.
Müller MJ, Härtter S, Köhler D, Hiemke C (2001). Serum levels of sulpiride enantiomers after oral treatment with racemic sulpiride in psychiatric patients: a pilot study. Pharmacopsychiatry 34, 27–32.
Musso DL, Mehta NB, Soroko FE, Ferris RM, Hollingsworth EB, Kenney BT (1993). Synthesis and evaluation of the antidepressant activity of the enantiomers of bupropion. Chirality 5, 495–500.
Narita N, Hashimoto K, Tomitaka S, Minabe Y (1996). Interactions of selective serotonin reup- take inhibitors with subtypes of receptors in rat brain. Eur J Pharmacol 307, 117–119.
Nation RL (1994). Chirality in new drug development – clinical pharmacokinetic considerations. Clin Pharmacokinet 27, 249–255.
Nordin C, Bertilsson L (1995). Active hydroxymetabolites of antidepressants: Emphasis on E-10- hydroxynortriptyline. Clin Pharmacokinet 28, 26–40.
Nordin C, Bertilsson L, Dahl M-L, Resul B, Toresson G, Sjöqvist F (1991). Treatment of depres- sion with E-10-hydroxynortriptyline – a pilot study on biochemical effects and pharmacoki- netics. Psychopharmacology 103, 287–290.
Nusser E, Nill K, Breyer-Pfaff U (1988). Enantioselective formation and disposition of (E)- and (Z)-10-hydroxynortriptyline. Drug Metab Dispos 16, 509–511.
Nyberg S, Dahl M-L, Halldin C (1995). A PET study of D2 and 5-HT2 receptor occupancy induced by risperidone in poor metabolizers of debrisoquin and risperidone. Psychopharmacology 119, 345–348.
Olesen OV, Linnet K (1997). Metabolism of the tricyclic antidepressant amitriptyline by cDNA- expressed human cytochrome P-450 enzymes. Pharmacology 55, 235–243.
Olesen OV, Linnet K (1999). Studies on the stereoselective metabolism of citalopram by human liver microsomes and cDNA-expressed cytochrome P-450 enzymes. Pharmacology 59, 298–309. Olesen OV, Licht RW, Thomsen E, Bruun T, Viftrup JE, Linnet K (1998). Serum concentrations and side effects in psychiatric patients during risperidone therapy. Ther Drug Monit 20,
380–384.
Otton SV, Ball SE, Cheung SW, Inaba T, Rudolph RL, Sellers EM (1996). Venlafaxine oxidation
in vitro is catalysed by CYP2D6. Br J Clin Pharmacol 41, 149–156.
Owens MJ, Knight DL, Nemeroff CB (2001). Second-generation SSRIs: human monoamine
transporter binding profile of escitalopram and R-fluoxetine. Biol Psychiatry 50, 345–350. Pan LP, de Vriendt C, Belpaire FM (1998). In-vitro characterization of the cytochrome P-450 iso- enzymes involved in the back oxidation and N-dealkylation of reduced haloperidol.
Pharmacogenetics 8, 383–389.
211 Pharmacogenetics of chiral psychotropic drugs
Patrick KS, Singletary JL (1991). Relative configuration of thioridazine enantiomers. Chirality 3, 208–211.
Pert CB, Snyder SH (1973). Opiate receptor: demonstration in nervous tissue. Science 179, 1011–1014.
Pfandl B, Morike K, Winne D, Schareck W, Breyer-Pfaff U (1992). Stereoselective inhibition of nortriptyline hydroxylation in man by quinidine. Xenobiotica 22, 721–730.
Pinder RM (1985). Adrenoreceptor interactions of the enantiomers and metabolites of mianse- rin: are they responsible for the antidepressant effect? Acta Psychiatr Scand 72(Suppl 320), 1–9. Pollock BG, Everett G, Perel JM (1992). Comparative cardiotoxicity of nortriptyline and its iso-
meric 10-hydroxymetabolites. Neuropsychopharmacology 6, 1–10.
Pollock BG, Sweet RA, Kirshner M, Reynolds CF, III (1996). Bupropion plasma levels and
CYP2D6 phenotype. Ther Drug Monit 18, 581–585.
Puozzo C, Leonard BE (1996). Pharmacokinetics of milnacipran in comparison with other anti-
depressants. Int Clin Psychopharmacol 11(Suppl 4), 15–27.
Quinn DI, Wodak A, Day RO (1997). Pharmacokinetic and pharmacodynamic principles of
illicit drug use and treatment of illicit drug users. Clin Pharmacokinet 33, 344–400.
Quirion R, Bowen WD, Itzhak Y et al. (1992). Meeting report. A proposal for the classification
of sigma binding sites. Trends Pharmacol Sci 13, 85–86.
Reilly JG, Ayis SA, Ferrier IN, Jones SJ, Thomas SHL (2000). QTc-interval abnormalities and
psychotropic drug therapy in psychiatric patients. Lancet 355, 1048–1052.
Ring BJ, Eckstein JA, Gillespie JS, Binkley SN, Vandenbranden M, Wrighton SA (2001). Identification of the human cytochromes P450 responsible for in vitro formation of R– and S–
norfluoxetine. J Pharmacol Exp Ther 297, 1044–1050.
Robertson DW, Krushinski JH, Bymaster FP, Reid LR, Wong DT (1991). Seproxetine: absolute
configurations and stereospecific interactions of the enantiomers of norfluoxetine with the
serotonin uptake carrier. Pharmacologist 33, 219.
Rochat B, Amey M, Baumann P (1995a). Analysis of enantiomers of citalopram and its demeth-
ylated metabolites in plasma of depressive patients using chiral reverse-phase liquid chroma-
tography. Ther Drug Monit 17, 273–279.
Rochat B, Amey M, van Gelderen H, Testa B, Baumann P (1995b). Determination of the enan-
tiomers of citalopram, its demethylated and propionic acid metabolites in human plasma by
chiral HPLC. Chirality 7, 389–395.
Rochat B, Amey M, Gillet M, Meyer UA, Baumann P (1997). Identification of three cytochrome
P-450 isozymes involved in N-demethylation of citalopram enantiomers in human liver
microsomes. Pharmacogenetics 7, 1–10.
Rochat B, Kosel M, Boss G, Testa B, Gillet M, Baumann P (1998). Stereoselective biotransforma-
tion of the selective serotonin reuptake inhibitor, citalopram, and its demethylated metabo-
lites by monoamine oxidases in human liver. Biochem Pharmacol 56, 15–23.
Rognan D, Sokoloff P, Mann A et al. (1990). Optically active benzamides as predictive tools for
mapping the dopamine D2 receptor. Eur J Pharmacol 189, 59–70.
Rotzinger S, Bourin M, Akimoto Y, Coutts RT, Baker GB (1999). Metabolism of some “second”
– and “fourth”- generation antidepressants: iprindole, viloxazine, bupropion, mianserin, maprotiline, trazodone, nefazodone and venlafaxine. Cell Mol Neurobiol 19, 427–442.
212 P. Baumann and C. B. Eap
Rudorfer MV, Potter WZ (1997). The role of metabolites of antidepressants in the treatment of depression [review]. CNS Drugs 7, 273–312.
Rudorfer MV, Potter WZ (1999). Metabolism of tricyclic antidepressants. Cell Mol Neurobiol 19, 373–409.
Sajjad SHA (1997). CYP2D6 genotype and tardive dyskinesia. Br J Psychiatry 170, 580.
Sanchez C, Hyttel J (1999). Comparison of the effects of antidepressants and their metabo- lites on reuptake of biogenic amines and on receptor binding. Cell Mol Neurobiol 19,
467–489.
Schmider J, Greenblatt DJ, Harmatz JS, Shader RI (1996). Enzyme kinetic modelling as a tool to
analyse the behaviour of cytochrome P-450 catalysed reactions: application to amitriptyline
N-demethylation. Br J Clin Pharmacol 41, 593–604.
Scordo MG, Spina E, Facciola G, Avenoso A, Johansson I, Dahl ML (1999). Cytochrome P-450
2D6 genotype and steady-state plasma levels of risperidone and 9-hydroxyrisperidone.
Psychopharmacology 147, 300–305.
Scott CC, Robbins EB, Chen KK (1948). Pharmacological comparison of the optical isomers of
methadone. J Pharmacol Exp Ther 92, 282–286.
Shih JC, Chen K, Ridd MJ (1999). Monoamine oxydase: from genes to behavior. Ann Rev Neurosci
22, 197–217.
Shin JG, Kane K, Flockhart DA (2001). Potent inhibition of CYP2D6 by haloperidol metabolites:
stereoselective inhibition by reduced haloperidol. Br J Pharmacol 51, 45–52.
Sidhu J, Priskorn M, Poulsen M, Segonzac A, Grollier G, Larsen F (1997). Steady-state pharma- cokinetics of the enantiomers of citalopram and its metabolites in humans. Chirality 9,
686–692.
Sindrup SH, Brøsen K, Hansen MGJ, Aaes-Jørgensen T, Fredricson Overø K, Gram LF (1993).
Pharmacokinetics of citalopram in relation to the sparteine and the mephenytoin oxidation
polymorphisms. Ther Drug Monit 15, 11–17.
Smith DF (ed.) (1984). CRC Handbook of Stereoisomers: Drugs in Psychopharmacology. Boca
Raton, FL: CRC Press, pp. 1–475.
Spencer CM, Wilde MI (1998). Milnacipran – a review of its use in depression. Drugs 56,
405–427.
Spina E, Ancione M, Di Rosa AE, Meduri M, Caputi AP (1992). Polymorphic debrisoquine oxi-
dation and acute neuroleptic-induced adverse effects. Eur J Clin Pharmacol 42, 347–348. Stevens JC, Wrighton SA (1993). Interaction of the enantiomers of fluoxetine and norfluoxetine
with human liver cytochromes P-450. J Pharmacol Exp Ther 266, 964–971.
Störmer E, von Moltke LL, Greenblatt DJ (2000a). Scaling drug biotransformation data from cDNA-expressed cytochrome P-450 to human liver: a comparison of relative activity factors and human liver abundance in studies of mirtazapine metabolism. J Pharmacol Exp Ther 295,
793–801.
Störmer E, von Moltke LL, Shader RI, Greenblatt DJ (2000b). Metabolism of the antidepressant
mirtazapine in vitro: contribution of cytochromes P-450 1A2, 2D6, and 3A4. Drug Metab
Dispos 28, 1168–1175.
Sullivan HR, Due SL (1973). Urinary metabolites of dl-methadone in maintenance subjects. J
Med Chem 16, 909–913.
213 Pharmacogenetics of chiral psychotropic drugs
Svendsen CN, Froimowitz M, Hrbek C et al. (1988). Receptor affinity, neurochemistry and behavioral characteristics of the enantiomers of thioridazine: evidence for different stereose- lectivities at D1 and D2 receptors in rat brain. Neuropharmacology 27, 1117–1124.
Timmer CJ, Sitsen JMA, Delbressine LP (2000). Clinical pharmacokinetics of mirtazapine. Clin Pharmacokinet 38, 461–474.
Tucker GT (2000). Chiral switches. Lancet 355, 1085–1087.
Tybring G, Otani K, Kaneko S, Mihara K, Fukushima Y, Bertilsson L (1995). Enantioselective
determination of mianserin and its desmethyl metabolite in plasma during treatment of
depressed Japanese patients. Ther Drug Monit 17, 516–521.
Ueno S, Otani K, Kaneko S et al. (1996). Cytochrome P-450 2D6 gene polymorphism is not asso-
ciated with neuroleptic malignant syndrome. Biol Psychiatry 40, 72–74.
Veefkind AH, Haffmans PMJ, Hoencamp E (2000). Venlafaxine serum levels and CYP2D6 geno–
type. Ther Drug Monit 22, 202–208.
von Bahr C, Movin G, Nordin C et al. (1991). Plasma levels of thioridazine and metabolites
are influenced by the debrisoquin hydroxylation phenotype. Clin Pharmacol Ther 49,
234–240.
von Moltke LL, Greenblatt DJ, Duan SX et al. (1997). Human cytochromes mediating N-
demethylation of fluoxetine in vitro. Psychopharmacology 132, 402–407.
von Moltke LL, Greenblatt DJ, Giancarlo GM, Granda BW, Harmatz JS, Shader RI (2001). Escitalopram (S-citalopram) and its metabolites in vitro: cytochromes mediating biotransfor- mation, inhibitory effects, and comparison to R-citalopram. Drug Metab Dispos 29,
1102–1109.
Wagstaff AJ, Fitton A, Benfield P (1994). Sulpiride – a review of its pharmacodynamic and phar-
macokinetic properties, and therapeutic efficacy in schizophrenia. CNS Drugs 2, 313–333. Wainer IW (1993). Drug Stereochemistry. Analytical Methods and Pharmacology, 2nd edn, revised
and expanded. New York: Marcel Dekker.
Walker JM, Bowen WD, Walker FO, Matsumoto RR, De Costa B, Rice KC (1990). Sigma recep-
tors: biology and function. Pharmacol Rev 42, 355–403.
Wang CP, Howell SR, Scatina J, Sisenine SF (1992). The disposition of venlafaxine enantiomers
in dogs, rats, and humans receiving venlafaxine. Chirality 4, 84–90.
Wong DT, Fuller RW, Robertson DW (1990a). Fluoxetine and its two enantiomers as selective
serotonin uptake inhibitors. Acta Pharm Nord 2, 171–179.
Wong DT, Reid LR, Threlkeld PG (1990b). Suppression of food intake in rats by fluoxetine: com-
parison of enantiomers and effects serotonin antagonists. Pharmacol Biochem Behav 31,
475–479.
Wong DT, Bymaster FP, Reid LR, Mayle DA, Krushinski JH, Robertson DW (1993).
Norfluoxetine enantiomers as inhibitors of serotonin uptake in rat brain. Neuropsycho-
pharmacology 8, 337–344.
Wood MD, Thomas DR, Watkins CJ, Newberry NR (1993). Stereoselective interaction of mian-
serin with 5-HT3 receptors. J Pharm Pharmacol 45, 711–714.
Yang SK, Lu X-L (1989). Racemization kinetics of enantiomeric oxazepams and stereoselective
hydrolysis of enantiomeric oxazepam 3-acetates in rat liver microsomes and brain homogen- ate. J Pharm Sci 78, 10, 789–795.
214 P. Baumann and C. B. Eap
Yasui N, Tybring G, Otani K et al. (1997). Effects of thioridazine, an inhibitor of CYP2D6, on the steady-state plasma concentrations of the enantiomers of mianserin and its active metabolite, desmethylmianserin, in depressed Japanese patients. Pharmacogenetics 7, 369–374.
Yasui-Furukori N, Hidestrand M, Spina E, Facciola G, Scordo MG, Tybring G (2001). Different enantioselective 9-hydroxylation of risperidone by the two human CYP2D6 and CYP3A4 enzymes. Drug Metab Dispos 29, 1263–1268.
Young RC, Dhar AK, Kutt H, Alexopoulos GS (1988). Isomers of 10-hydroxynortriptyline in ger- iatric depression. Ther Drug Monit 10, 164–167.
Young D, Midha KK, Fossler MJ et al. (1993). Effect of quinidine on the interconversion kinetics between haloperidol and reduced haloperidol in humans. Implications for the involvement of cytochrome P-450IID6. Eur J Clin Pharmacol 44, 433–438.
Yue QY, Svensson JO, Bertilsson L, Säwe J (1995). Racemic methadone kinetics in relation to the debrisoquine hydroxylation polymorphism in man. Therapie Suppl, 88.
Part V
Specific psychotropic drugs and disorders


10
Clozapine response and genetic variation in neurotransmitter receptor targets
David A. Collier1, Maria J. Arranz2, Sarah Osborne2, Katherine J. Aitchison2, Janet Munro2, Dalu Mancama2, and Robert W. Kerwin2 Section of Genetics1 and Section of Clinical Neuropharmacology2, Institute of Psychiatry, London, UK
OVERVIEW
Clozapine is an atypical antipsychotic drug with unique clinical features, particularly in treat- ment resistant schizophrenia. Between 10 and 60% of patients resistant or intolerant to treatment with other antipsychotic drugs respond to clozapine. Reliable genetic prediction of which patients will respond would have major economic, clinical, and safety implications. Candidate genes of potential utility in pharmacogenetic tests can be identified from cloza- pine’s neurotransmitter targets. Clozapine has a complex pattern of interaction with neuro- transmitter receptors, with high affinity for dopamine D4, serotonin (5-HT) 1A, 2A, 2C, 6, and 7, histamine H1, muscarinic M1 and 1- and 2-adrenoceptors. Exactly which of these recep- tors mediates clozapine’s clinical efficacy is unknown, but the involvement of 5-HT is very likely. Despite the lack of formal evidence from genetic epidemiology, it is reasonable to hypothesize that genetic variation in these receptors influences clinical response to cloza- pine by altering receptor function in some way. At most of these receptors, clozapine is a neutral antagonist, but at others, such as 5-HT2C, it has inverse agonist activity. Further com- plexity comes from the unusual regulation of several of these receptors, with evidence for functional alternative splicing of 5-HT2C, polymorphic imprinting of 5-HT2A, and unusual downregulation of at least three 5-HT receptors through clozapine-induced cellular internal- ization. Most of clozapine’s receptor targets have been tested in some way for association with clinical response using a case-control allelic association design. Polymorphisms in two genes, for 5-HT2A and dopamine D3, have been implicated in more than one study. However, there is a general lack of large, powerful prospective studies using multiple measures of response, and many polymorphisms in potentially relevant genes, such as that for - aminobutyric acid type A (GABA-A), have not yet been examined. A novel epistatic approach, simultaneously examining genetic variation in multiple genes that encode clozapine’s neu- rotransmitter targets has been piloted, but needs replication in a prospective study. To predict fully clinical response to clozapine treatment in schizophrenia, it will be necessary to take account of many variables in addition to genetic influence on clozapine’s pharma- cokinetics, including social, demographic, and clinical factors (e.g., compliance, social support, and history of birth trauma); genetic vulnerability to side effects such as weight gain
217

218 D. A. Collier et al.
and sedation; and the influence of genetic susceptibility factors for psychosis, which may define subtype and symptomatology. This chapter will concentrate on the genetics of response to the antipsychotic drug clozapine, focusing mainly on pharmacodynamic factors related to clozapine’s receptor binding.
Introduction
Pharmacogenetics
There are two broad reasons for treatment failure: adverse reaction (i.e., “one which is noxious and unintended and which occurs at doses used in man for prophylaxis and treatment” (World Health Organization)) and lack of efficacy (Price-Evans, 1993). Drugs may fail to elicit a clinical response because of failure to reach a sus- tained therapeutic concentration (pharmacokinetics) or fail to elicit a response because of altered receptor binding or coupling (pharmacodynamics). In addition, differences in the pathophysiology of heterogeneous and complex diseases such as schizophrenia may affect response.
Historically, pharmacogenetics has not used classical epidemiological methods such as twin studies to analyze the heritability of drug effects, but instead has relied on biochemical genetics. Biotransformation was initially identified as being genet- ically variable because of the interindividual variability in the effects of certain drugs, with classical bi- or trimodal differences in drug breakdown across individ- uals. This indicates the existence of simple genetic effects such as a defective or overactive enzyme. Since the 1960s, progress has been much faster in understand- ing the genetics of drug metabolism than in the study of the genetic phenomena of drug receptors (Price-Evans, 1993), presumably because of the relative simplicity of understanding families of related enzymes rather than individual (and often unknown) drug targets. The field of receptor-based pharmacogenetics is just re-emerging.
Clozapine

Clozapine is a tricyclic dibenzodiazepine (Fig. 10.1) first developed in the 1960s and 1970s. It was known to have an atypical profile with a low incidence of extrapyra- midal side effects (EPS) (Gerlach et al., 1974; Nair et al., 1977) but was withdrawn in 1975 after a cluster of cases of agranulocytosis in Finland (Amsler et al., 1977). It was later reintroduced with strict hematological monitoring because of its effec- tive antipsychotic, antiaggressive and anxiolytic properties. In most of the world, clozapine is used in treatment-refractory or intolerant patients, i.e., those who are unresponsive to two or more classes of antipsychotic or are intolerant of their neurological side effects. In this group, about 37% are responsive to clozapine treat- ment after 6 weeks and about 81% after 1 year (Meltzer, 1992), although in one
219 Clozapine response and neurotransmitter receptor targets
   
Cl
N NCH3 N
N H
         
Fig. 10.1.
The chemical structure of clozapine, a tricyclic dibenzodiazepine.
study of strictly defined treatment-refractory schizophrenia or schizoaffective dis- order only 10% responded (Simpson et al., 1999). Clozapine is used as a first line of treatment in several countries, notably Switzerland and the People’s Republic of China. Treatment response to clozapine in first-onset psychosis exceeds 80%.
Clozapine is given at various doses from 50 to 900 mg per day, depending on how well the patient tolerates the drug. Dose is usually titrated to about 150mg over 2 weeks and then in 25–50mg increments to a maximum of 900mg per day over several weeks (although usual clinical doses are lower). Although early research appeared to indicate that clozapine serum level was not important in response (reviewed by Fitton and Heel (1990)), more recent research indicates that a plasma serum concentration of clozapine of at least 350g/l and as high as 420g/l is required to optimize response (reviewed in Lieberman et al. (1995a)). For example Potkin et al. (1994) found that after 12 weeks 73% of patients with serum levels 420g/l responded to clozapine compared with only 29% of those with 420g/l. Its clinical efficacy is thought to be associated principally with its action at 5-HT neurotransmitter receptors, particularly 5-HT1A, 5-HT2A, and 5-HT2C, but also at dopamine. It also has affinity for a variety of other neurotransmitter recep- tors, including high-affinity binding to muscarinic receptors and -adrenoceptors.
What influences clinical response to antipsychotic medication?
The factors that influence response to antipsychotic medication are a largely unknown mixture of genetic and environmental factors. Clinical and social factors associated with good response to antipsychotics include compliance, family support during treatment, less severe delusions and hallucinations and better attention at baseline, male sex, fewer obstetric complications, less than 9 years of illness, and later age at onset (Wilcox and Nasrallah, 1987; Robinson et al., 1999).
The development of side effects influences treatment response, since a propor- tion of patients will be unable to tolerate a particular drug. For typical antipsychot- ics, these will include the development of EPS, and for clozapine side effects include neutropenia, sedation, seizures, and tachycardia. These are, in part, a result of her- itable variability in drug metabolism: poor metabolizers of a given drug will develop dose-dependent side effects more easily since peak plasma concentrations
220 D. A. Collier et al.
of drugs will be much higher for a given dose. Other reasons may include genetic susceptibility to specific side effects.
However, the major reason for antipsychotic treatment failure is lack of efficacy for unknown reasons. There is almost no information from twin or family studies on the heritability of clinical responsiveness (i.e., improvement in disease symp- toms) to antipsychotic drugs, aside from a few case reports of concordant twins (Vojvoda et al., 1996; Mata et al., 2001). While it is reasonable to assume that there is a pharmacogenetic influence on response to clozapine, there is an urgent need to clarify this by genetic epidemiological study.
Clinical response to clozapine might be mediated by variation in receptor density (through changes in gene expression, protein stability, or transport), altered kinetics of dimer or multimer formation, variation in efficiency of coupling to second messenger systems, or alterations in ligand affinity or allosteric modula- tor binding. It is important to note, however, that a pharmacogenetic effect is dependent on functional genetic variation; for some genes, there may be no genetic variation that affects clozapine’s action.
Research methodology
What is response to clozapine?
Deciding on the criteria most useful for judging response to clozapine is difficult, since there are multiple ways to judge response ranging from specific neurophysio- logical measures such as P50 sensory gating through to quality-of-life measures. Furthermore, patients taking this drug in most parts of the world are likely to be resistant or intolerant to treatment with other antipsychotics, have worse illness, and a complex treatment history. Consequently, patients taking clozapine may have more severe symptoms, more chronic illness, and be more sensitive to side effects than schizophrenic patients in general. These factors must be borne in mind when generalizing pharmacogenetic findings to other antipsychotic drugs when cloza- pine has been used as a model.
The two most important considerations for judging clinical response are the instrument used and the time scale of measurement. The use of multiple but com- patible measures has been recommended (Masellis et al., 2000) and consensus methods to judge clinical response have been proposed. Clearly, a double-blind prospective study of clozapine response, in the same manner as a drug trial, is the best method of analysis since good baseline data and detailed response criteria can be collected over a defined time period. Retrospective analysis, where a judgement of response is made based on case notes and/or the opinion of the treating medical staff, cannot be as good since it will necessarily produce less detailed clinical data and cannot follow patients from a defined baseline at the commencement of clozapine therapy. However retrospective analysis has practical advantages since

221 Clozapine response and neurotransmitter receptor targets
substantial numbers of patients can be assessed relatively quickly. Retrospective methodology may be necessary for twin or family study of response since it will be difficult to ascertain a significant number of twins prospectively.
The Global Assessment Scale (known as GAS, GAF, or GAFS) (Endicott et al., 1976; see also Tress and Patton, 1994) is an improvement of the Health Rating Scale of Luborsky (1962) and the simplest and quickest scale used to assess global response to antipsychotic drugs. It is scored from 1 to 100 (individuals with minimal psychopathology score above 71), with ten anchor points divided into deciles. It has been used in a series of retrospective analyses of clozapine response (Arranz et al., 1998a). Response is usually defined as a 20-point increase in GAS score but can also be used as a continuous or quantitative scale in regression anal- ysis. The Brief Psychiatric Rating Scale (BPRS) (Overall and Gorham, 1962; see also Tress and Patton, 1994), and the Positive and Negative Symptom Scale (PANSS) (Kay et al., 1992) are more detailed and the most often used for prospective studies of clozapine response (see Masellis et al., 1998, 2001). The PANSS measures posi- tive and negative symptoms of schizophrenia and general psychopathology. Response according to both BPRS and PANSS is usually defined as a 20% decrease in score, but some clinical trials use a 40% decrease. An alternative measure, the Clinical Global Impression scale (CGI) (Guy, 1976), measures the impression of the severity of illness on a seven category scale. The CGI Improvement scale (CGI-I; 0–7) is widely used in clinical trials but has only been used in one pharmacogenetic study of clozapine response to augment BPRS criteria (Masellis et al., 1998). Improvement on at least one category (CGI-I score of 1 or 2) has been widely used to define response. Other methods used include the clinical classification of patients into four response groups, taking into account minimum clozapine dose (Noethen et al., 1995) or into three groups (Kohn et al., 1997). The Bunney–Hamburg Global Psychosis rating scale has also been used (Rao et al., 1994).
Other approaches, such as the measurement of specific symptomatology or quality of life, have been considered. Scales which measure specific symptomatol- ogy in more detail, such as PANSS may be useful, as severity of delusions and hal- lucinations predicts responsiveness and may be more sensitive measures for pharmacogenetic studies. Quality-of-life scales, such as the Lehman Quality of Life Interview and the Heinrichs–Carpenter scale (reviewed in Lehman et al. (1993)) or the WHOQOL-100 may also be useful. The WHOQOL-100 (Skevington and Wright, 2001) appears closely correlated with the effectiveness of antidepressant drugs, with 96% of facets responsive to perceived change in clinical depression. Quality of life may be the most important treatment outcome, for the patient, as well as being a sensitive indicator of drug action.
The use of neuropsychology or neurophysiology to assess treatment response in schizophrenia is a new area of investigation. Neuropsychological deficits appear to
222 D. A. Collier et al.
be established at the onset of illness, and indeed in childhood (Meltzer and McGurk, 1999; Bilder et al., 2000). A large generalized neuropsychological deficit and more subtle executive and attentional deficits are seen in first episode psycho- sis, mark the more severely ill patients, and have been postulated to predict poor outcome (Bilder et al., 2000). The use of neuropsychological tests to analyze treat- ment response may be particularly powerful as there is evidence that atypical drugs such as clozapine have the ability to improve various aspects of cognitive function, which is not seen for typical drugs (Meltzer and McGurk, 1999). For example, cloz- apine appears to improve verbal ability, attention, and some types of executive function, and these variables appear different in comparison with other drugs such as olanzapine (Meltzer and McGurk, 1999). Neurophysiological measures may also be powerful; clozapine corrects the P50 sensory gating deficit seen in the majority of schizophrenics (Nagamoto et al., 1996; Light et al., 2000).
How long does the response take?
The time needed to see the full spectrum of clinical response to clozapine is unclear, but, in general, 37% of patients are thought to respond after 6 weeks and 81% after 1 year (Meltzer, 1992). Evidence suggests that any prospective study of clozapine response should use a minimum time period of at least 12 weeks, since several trials have shown that this time length is necessary for response in most patients (Breier et al., 1994; Lieberman et al., 1994a, b; Potkin et al., 1994). Some patients may take 12 months to show good response, but the probable optimum is between 12 and 24 weeks (Lieberman et al., 1994b). Positive symptoms have been seen to improve after 8 weeks (Tandon et al., 1993) and improvement in negative symptoms may lag about 7 weeks behind (Lieberman et al., 1994b). Up to 9 months has been sug- gested as necessary to see the full benefits of clozapine (Meltzer, 1995), but some observers suggest that all of clozapine’s differential beneficial effects (i.e., advan- tages over typical drugs) occur during the first 6 weeks (Rosenheck et al., 1999). However, it is important to note that clozapine is titrated up to a therapeutic dose and this may delay response.
Defining candidates genes: the pharmacology of clozapine and results of
genetic analysis
The ability of clozapine to improve the symptoms of psychosis is dependent on its antagonist and reverse agonist activities at neurotransmitter receptors. Clozapine’s affinity for some of its most important neurotransmitter receptors is shown in Table 10.1. The serum level of clinical doses of clozapine is estimated at 20nmol (Seeman, 1992), so receptors with Ki values substantially more than this will not be fully occupied. Its affinity for the dopamine D2 receptor is relatively weak compared

223 Clozapine response and neurotransmitter receptor targets
Table 10.1. The affinity of antipsychotic drugs for neurotransmitter receptors
Binding constant Ki (nmol/l)
Receptor Haloperidol Clozapine Risperidone Olanzapine Ziprazidone Quetiapine

Dopamine
D1 120 141 75 31
130 3.1
7.2 32
2.5 0.39 0.72
455 160 940
2200 1000
295 1000 1000 4100
11 120
7 87
D2 1.383
3.1 11 9.6 50 7 27
488 1000 0.16 5
25.8 11.3
. D3 3.2
. D4 2.3
Serotin
5-HT1A 1000 5-HT2A 78 5-HT2C 1000 5-HT3 1000 5-HT6 6000
Histamine
H1 1000 Muscarinic
M1 1000
Adrenergic
1 46 2 360
200 20
6.46 2.5 8.6
95 1000 11 2000
23 155
1.9 1000
57 – 10 76
7 47
1.9 5100 19 13
39 2
11.6 3 228 310
with the affinity of typical antipsychotics such as haloperidol, which may provide some of the explanation for its reduced propensity to cause movement disorders. Its affinity for other dopamine receptors is also unexceptional (Shaikh et al., 1997), but it is a potent antagonist for 5-HT receptors (Meltzer, 1999), particularly 5- HT1A, 5-HT2A, and 5-HT2C. The ratio D2:5-HT affinity may be a key feature of atyp- ical antipsychotic action. D4 receptor binding (particularly D2:D4 and D4:5-HT2A binding ratios) may also be important for clozapine’s action (Seeman, 1992; Shaikh et al., 1997; Kulkarni and Ninan, 2000; Kapur and Remington, 2001).
Genetic variation in clozapine’s receptor targets is a potential source of pharma- codynamic influence on drug response, by altering drug action. It is not clear exactly which of clozapine’s numerous receptor targets is responsible for its clini- cal action. Proof that genetic variation in a particular receptor affects clozapine’s clinical efficacy is proof of that receptor’s role in drug action. This would represent an important pharmacological finding in itself, in addition to the potential for genetic testing and the individualization of treatment.
224 D. A. Collier et al.
Dopamine receptors
Dopamine D1 and D5 receptors
Clozapine is a weak antagonist at both D1 and D5 receptors (Table 10.1), with an approximately twofold lower affinity for D1 and D5 compared with D2. Consequently, the D1 receptor has not been thought to be a significant mediator of clozapine’s clinical effects. However, it is the most abundant dopamine receptor and is far more prevalent than D2 receptors in areas that have been implicated in schizophrenia, such as the prefrontal cortex and nucleus accumbens (Missale et al., 1998). The D5 receptor is expressed at much lower levels in the brain, principally in hippocampus, cerebral cortex and lateral thalamus. Both D1 and D5 receptors have pre- and postsynaptic localization, and although they have similar pharmacology, they are not functionally redundant because of differences in expression patterns and cellular localization (Missale et al., 1998).
Both D1 and D5 receptors are downregulated in the prefrontal cortex but not the striatum by treatment with many commonly used antipsychotic drugs, including clozapine, which reduces cortical D1 and D5 expression by 30–40% (Lidow et al., 1997). There is evidence that this regional downregulation may be a consequence of D2 antagonism rather than direct binding to D1/D5. This effect may be impor- tant for clinical efficacy of antipsychotics, including clozapine, as D1-specific antag- onists have been reported to improve negative symptoms in schizophrenia. However, D1 blockade may diminish working memory (Williams and Goldman- Rakic, 1995). There is some evidence that clozapine is, in fact, a direct D1 agonist (Ahlenius, 1999). Clozapine binds weakly to D5 receptors with a Ki of 320.
The gene DRD1 is not highly polymorphic, with no common missense variants in the coding region (Cichon et al., 1996). Several variants are known in the 5 region and promoter ( 48A/G, 2218T/C, 2102C/A, 2030T/C, 1992G/A, 1251G/C, and 800T/C) (Cichon et al., 1996) but these do not appear to be in regions critical for gene regulation. The DRD5 coding region is more polymorphic (Sobell et al., 1995; Feng et al., 1998), producing Leu88Phe in the putative second transmembrane domain, Ala269Val in the third intracellular 100p, Pro330Gln in the third extracellular loop, Asn351Asp in the seventh transmembrane domain, Ser453Cys in the C-terminal domain, Cys335Stop in the third extracellular loop, and several silent polymorphisms. Most of these are uncommon and they have not been examined in clozapine response.
Dopamine D2 receptor
The “D2-like” receptors D2, D3 and D4 differ from D1/D5 in that they have three introns, a long third cytoplamic loop, and a short C-terminal tail; they also inhibit rather than stimulate the enzyme adenylyl cyclase. Clozapine and other atypical
225 Clozapine response and neurotransmitter receptor targets
antipsychotic drugs are notable because their clinical efficacy is not dependent on high affinity for the dopamine D2 receptor, unlike D2-specific drugs such as halo- peridol (Snyder, 1981; Roth et al., 1994). Clozapine has affinity for D2 receptors that is at least an order of magnitude lower than that for 5-HT1A, 5-HT2A, or 5-HT2C receptors, and this is perhaps the reason atypical antipsychotics such as clozapine act without causing EPS. However, it is likely that D2 binding, albeit weak, is impor- tant for clozapine’s efficacy (Meltzer, 1994) and there is evidence for limbic over striatal selectivity based on assessment of D2 occupancy using positron emission tomography neuroimaging (Pilowsky et al., 1997). The affinity of antipsychotics for D2 receptors is dependent on the rate at which they dissociate from the recep- tor (Koff; Kapur and Seeman, 2000). Drugs such as clozapine with a higher Koff will dissociate from receptors faster and, once the receptors are unblocked, will provide more access to surges in dopamine transmission. A higher Koff for the D2 receptor has been proposed by Kapur and Seeman (2000) as a mechanism for “atypical” antipsychotic effect.
There are multiple polymorphisms in DRD2, including -241A/G, -141C, the “TaqI A” polymorphism, and that giving rise to Ser311Cys. One of these, the func- tional promoter variant 141C, has been examined in a retrospective sample of clozapine responders and nonresponders and was associated with response in one study (Malhotra et al., 1998) but not in a second (Arranz et al., 1998b).
Dopamine D3 receptor
Clozapine has relatively weak affinity for the dopamine D3 receptor (Ki 200 nmol/l) and is likely to have low occupancy of this receptor in vivo, suggesting that this receptor may not be of major importance to clozapine’s action. The D3 receptor is about 100-fold less abundant than D2 receptors in the brain, but the expression of D3 in limbic areas, especially the nucleus accumbens, makes an attractive target for antipsychotic drugs (Schwartz et al., 1993). Malmberg et al. (1998) have shown that clozapine is a partial inverse agonist at D3.
The D3 receptor is coupled to several signaling pathways, including Gi/G0- proteins and the c-fos system, but intracellular responses are relatively weak (Sokoloff et al., 1992). The ability of clozapine to induce c-fos mRNA, an index of receptor coupling, is not altered in transgenic mice lacking the DRD3, also indicat- ing that D3 is not an important mediator of clozapine’s clinical effects. However D3 has attracted interest as a potential genetic risk factor for schizophrenia, and ele- vated D3 mRNA has been seen in lymphocytes from patients with schizophrenia (Ilani et al., 2001).
There are four known polymorphisms in DRD3: noncoding 712G/C and 205A/G and coding polymorphisms resulting in Ser9Gly and Ala38Thr (Crocq et al., 1992; Ishiguro et al., 2000). Homozygosity for one of these, giving Ser9Gly,
226 D. A. Collier et al.
has been associated with schizophrenia (Crocq et al., 1992; Shaikh et al., 1996) and tardive dyskinesia during treatment with typical neuroleptics (Steen et al., 1997; Basile et al., 1999; Segman et al., 1999, 2000; Eichhammer et al., 2000; Lovlie et al., 2000; Ozdemir et al., 2001). It has also been implicated as a pharmacogenetic factor for clozapine response in a retrospectively assessed sample using GAFS (Shaikh et al., 1996). Although this association with clozapine response was not replicated in subjects from a double-blind clinical trial of clozapine (Malhotra et al., 1998), a subsequent study in Pakistani patients showed a significant association with the allele producing Gly9 (Scharfetter et al., 1999). A combined analysis of these three reports was also significant (p 0.004). Therefore, despite the low affinity of cloz- apine for D3 receptors, there is now good evidence for a pharmacogenetic influence on clozapine response.
Dopamine D4 receptor
The D4 receptor has similar pharmacology to the D2 and D3 receptors except that its affinity for clozapine is 10–20-fold higher (van Tol et al., 1991). This switch in specificity appears to be the result of a change in just a few amino acid residues in the second, third, and seventh transmembrane segments, which form a binding site crevice (Simpson et al., 1999). DRD4 is expressed in regions of the brain particu- larly relevant to schizophrenia, especially the frontal cortex, amygdala, and hippo- campus, and also the hypothalamus and mesencephalon, but at low levels in the basal ganglia (reviewed in Missale et al. (1998)).
Based on estimates that the therapeutic plasma concentration of clozapine is 20 nmol/l, D4 is expected to be the only dopamine receptor occupied at physiolog- ical concentration (Seeman, 1992). However it is not clear whether D4 mediates any of clozapine’s therapeutic actions: in a clinical trial the D4-specific antagonist L- 745870 did not improve psychotic symptoms and, if anything, made them worse (reviewed in Bristow et al. (1997)).
DRD4 is unusually variable. A 48bp tandem repeat in exon 3 of the gene, with alleles ranging from 1 to 12 repeats, was hypothesized to influence clozapine response since it appeared to modulate clozapine’s affinity for the receptor (van Tol et al., 1992). However a series of genetic association studies of clozapine response, using both prospective and retrospective methods, failed to find association between DRD4 VNTR (variable number tandem repeats) alleles and clozapine response (Shaikh et al., 1993, 1995; Rao et al., 1994; Rietschel et al., 1996; Kohn et al., 1997; Cohen et al., 1999; Kaiser et al., 2000). It subsequently became evident that there were only minor pharmacological differences between different alleles of the 48bp VNTR (Jovanovic et al., 1999), although an effect from the extensive internal sequence variation of the repeats themselves cannot be ruled out (Lichter et al., 1993).
227 Clozapine response and neurotransmitter receptor targets
A number of other variants also occur in DRD4, including a 11C/T in the 5 untranslated region; a 12 bp repeat which adds four amino acid residues and a 21 bp deletion which deletes seven amino acid residues, both in exon 1; base changes giving rise to Gly11Arg and Val194Gly; and a 13bp nonsense mutation in the region encoding transmembrane region 2 which results in a non-functional protein. None of these has been associated with clozapine response (reviewed in Shaikh et al. (1997)). More recently, a series of polymorphisms in the promoter region ( 1217G ins/del, 809G/A, 616C/G, 603T ins/del, 602(G)8–9, and 521C/T) has been identified (Okuyama et al., 1999). The polymorphism –521C/T is functional in vitro in that it reduces receptor expression by 40% in in vitro assays. These polymorphisms have not been tested in clozapine response.
Serotonin receptors
Serotonin 5-HT1A receptors
Because of its expression in regions of the brain thought to be relevant to schizo- phrenia, such as the hippocampus dorsal raphe nucleus and the neocortex, and its pharmacological properties, 5-HT1A is a strong candidate for a pharmacogenetic influence on clozapine’s clinical efficacy (Masellis et al., 2001). In contrast to other receptors, at which clozapine is an antagonist, clozapine acts as a partial agonist at 5-HT1A receptors, along with other atypical drugs such as ziprasidone (Newman- Tancredi et al., 1998). There is good evidence to suggest that 5-HT1A agonism is an important component of the therapeutic action of clozapine; 5-HT1A receptor stimulation appears to increase dopamine release in the prefrontal cortex (Kuroki et al., 1999; Ichikawa and Meltzer, 2000) and also to be partly responsible for a reduction in 5-HT efflux in the rat ventral hippocampus (Bengtsson et al., 1998). Agonism at 5-HT1A receptors appears to have many of the same effects as 5-HT2A antagonism (Meltzer, 1999). Antagonism at 5-HT2A receptors (or D2 antagonism) concurrent with 5-HT1A agonism has a synergistic effect in animal models of antipsychotic activity (Wadenberg and Ahlenius, 1991; Zifa and Fillion, 1992).
There a several known polymorphisms in 5-HT1A: 51T/C, 152C/G, 321G/C, 480delA, 581C/A. and 1018C/G in the 5 untranslated region and 294G/A, and 549C/T plus changes resulting in Gly272Asp and Pro16Leu in the coding region (Erdmann et al., 1995; Nakhai et al., 1995; Bergen et al., 1996; Kawanishi et al., 1998; Wu and Comings, 1999). Most of these polymorphisms are uncommon and consequently not informative; one of these polymorphisms (Pro16Leu) originally detected in Japan, has been tested for association with cloz- apine response in Americans but was not polymorphic in that population (Masellis et al., 2001).
228 D. A. Collier et al.
Serotonin 5-HT2A and 5-HT2C receptors
Clozapine and other atypical antipsychotics are potent ligands for 5-HT2A (Ki 2.5) and 5-HT2C (Ki 8.6) and consequently these receptors have attracted the most atten- tion as mediators of clozapine’s clinical action (Roth et al., 1994). However, there is no clear correlation between 5-HT2A and 5-HT2C affinity and atypical antipsy- chotic action, and some typical drugs, such as chlorpromazine, have high affinity for 5-HT2C. Both 5-HT2A and 5-HT2C receptors are linked to the stimulation of intracelluar inositol phosphate levels via G-protein coupling.
The 5-HT2A receptor is highly expressed in many areas of the cortex, the neocor- tex, parts of the limbic system, and the basal ganglia. Detailed studies of the macaque cortex revealed that the most intense 5-HT2A expression is in pyramidal neurons of the cortex, where expression is both pre- and postsynaptic (Jakab and Goldman-Rakic, 1998). The “hot-spot” for expression is in the apical dendrite field proximal to the pyramidal cell soma, where the receptors may participate in sensory gating to regulate working memory and the related process latent inhibi- tion. Both of these are relevant to dysfunctional states in psychosis, and clozapine may act specifically to improve sensory gating. The 5-HT2C receptor is expressed in the globus pallidus and substantia nigra of the basal ganglia, the choroid plexus and limbic system (reviewed in Hoyer (1994)).
Constitutively active forms of both 5-HT2A and 5-HT2C receptors have been pro- duced by site-directed mutagenesis (Herrick-Davies et al., 1997; Egan et al., 1998) and this has led to the reclassification of many drugs, including atypical drugs, as 5-HT2A/5-HT2C inverse agonists (Egan et al., 1998; Herrick-Davies et al., 1998, 2000). The mRNA for 5-HT2C, but not 5-HT2A, undergoes RNA editing to produce multiple isoforms differing at just a few amino acids (Burns et al., 1997). These iso- forms are widely distributed in the brain and differ in the level of constitutive activ- ity, with RNA editing tending to silence it (Herrick-Davies et al., 1999; Niswender et al., 1999). In this model, the unedited 5-HT2C receptor exists in conformational equilibrium shifted toward the active form in the absence of ligand. Atypical anti- psychotics such as clozapine shift the equilibrium to favor the inactive conforma- tion, rather than being classical neutral antagonists. Thus atypical antipsychotics have unique functional effects on 5-HT2C; these effects mediate some of the clini- cal efficacy of clozapine, then regulation of RNA editing may be an important factor in the pharmacogenetics of clozapine response.
Both 5-HT2A and 5-HT2C receptors are also unusual in that they have another unusual mode of regulation, since both agonists and antagonists (including cloza- pine) induce their post-translational downregulation, mainly by internalization of the receptor (Newton and Elliot, 1997; Bhatnagar et al., 2001). Clozapine induces 5-HT2A internalization via an arrestin-independent endosome-mediated pathway (Bhatnagar et al., 2001). This internalization process may be central to clozapine’s clinical efficacy.
229 Clozapine response and neurotransmitter receptor targets
The gene 5-HT2A has multiple polymorphic variants in both the coding region (102T/C, 516C/T and giving rise to Thr25Asn, Ala447Val, and His452Tyr; Warren et al., 1993; Erdmann et al., 1996) and the 5 sequence/promoter region ( 1438A/G, 1421C/T; Spurlock et al., 1998). An initial report of association between a 5-HT2A variant, 102T/C, and clozapine response (Arranz et al., 1995) was not replicated in several independent studies (Masellis et al., 1995, 1998; Noethen et al., 1995; Malhotra et al., 1996a). However, a meta-analysis showed a trend of association between this polymorphism and response (Arranz et al., 1998a). In addition, Arranz et al. (1998c) and Masellis et al. (1998) found associa- tion between His452Tyr and clozapine response. The meta-analysis by Arranz et al. (1998a) also investigated the relation between His452Tyr polymorphism and cloz- apine response, confirming the apparent association. These results constitute the strongest evidence of association between receptor variants and treatment response and suggest an important role for the 5-HT2A receptors in mediating the therapeu- tic activity of clozapine.
5-HT2C has two polymorphisms, one giving rise to Ser23Cys and the other a complex GT/CT repeat region (Deckert et al., 1997; Arranz et al., 2000a). Both of these have been examined for association with clozapine response. The Ser23- producing allele has been associated with good clozapine response in a retrospec- tive analysis (Sodhi et al., 1995) but this has not been replicated (Malhotra et al., 1996b; Rietschel et al., 1997; Masellis et al., 1998). The promoter GT/CT repeat has also been associated with clozapine response (Arranz et al., 2000a); however, since it is in strong linkage disequilibrium with the polymorphism giving Ser23Cys, this may reflect linkage disequilibrium between the two polymorphisms.
Serotonin 5-HT3 receptors
Clozapine is a functional antagonist for 5-HT3 receptors and binds with moderate affinity (Watling et al., 1990; Brunello et al., 1995) with a Ki value of about 123 nmol/l (Hermann et al., 1996). Unlike other 5-HT receptors, 5-HT3 is a ligand- gated ion channel with structural features in common with GABA-A and glycine and nicotinic acetylcholine receptors (Maricq et al., 1991). It is expressed in the area postrema and mesolimbic system in the brain, which acts as an interface between limbic and motor structures. It has been suggested that 5-HT3 can mediate anxio- lytic (Broekkamp et al., 1989) and atypical antipsychotic properties (Warburton et al., 1994). There is also neurochemical evidence against clozapine’s antipsychotic effects being dependent on 5-HT3 (Squires and Saederup, 1999). Two different 5- HT3 isoforms, 5-HT3A and 5-HT3B, encoded by separate genes, have been cloned, both of which map close together on chromosome 11q23 (Weiss et al., 1995; Davies et al., 1999). 5-HT3A has been screened for polymorphisms in schizophrenic and bipolar patients but these are rare and not associated with disease (Niesler et al., 2001). One study has examined 5-HT3B in clozapine response using a retrospectively
230 D. A. Collier et al.
assessed GAFS score (Arranz et al., 2000a). Two silent polymorphisms, 178C/T and 1596G/A, were examined but no association was found.
Serotonin 5-HT5A receptors
The 5-HT5A receptor does not have a high affinity for antipsychotic drugs such as clozapine (Waeber et al., 1998). It is expressed exclusively in the brain, mainly in the cerebral cortex (especially layers II–III and V–VI of the neocortex), hippocam- pus (dentate gyrus and pyramidal cell layer of CA1 and CA3), and cerebellum. This distribution indicates that 5-HT5A may be involved in higher cortical and limbic functions (Pasqualetti et al., 1998), and the receptor is thus a good candidate for etiological involvement in psychiatric disorders. One study has examined 5-HT5A receptors in clozapine response using a retrospectively assessed GAFS score (Birkett et al., 2000). Two silent polymorphisms, 19G/C and 12A/T, were examined but no association was found.
Serotonin 5-HT6 receptors
Clozapine is a high-affinity antagonist for the 5-HT6 receptor (Roth et al., 1994) and in rat brain the 5-HT6 receptor may represent up to 40% of specific receptor binding (Glatt et al., 1995). The 5-HT6 receptor is widely distributed in the human brain and is expressed most prominently in the caudate nucleus, at significant levels in the hippocampus and the amygdala, and at lower levels in the thalamus, subthal- amic nucleus, and substantia nigra (Kohen et al., 1996). It stimulates adenylyl cyclase through Gs-protein (Barnes and Sharp, 1999). The human 5-HT6 receptor is structurally (89% amino acid identity) and functionally similar to that in the rat, and on the basis of the affinity of the receptor for clozapine is expected to be sub- stantially occupied at therapeutic doses in humans (Kohen et al., 1996). Clozapine appears to have a potent effect on 5-HT6 expression as it can reduce 5-HT6-binding sites (Bmax) by as much as 62% in serum-free HeLa cells transfected with the rat receptor (Zhukovskaya and Neumaier, 2000). Downregulation of a receptor by an antagonist is unusual but has also been seen for the 5-HT2A and 5-HT2C receptors, where internalization is triggered by atypical antipsychotic drugs including cloza- pine (Willins et al., 1999). This indicates that 5-HT2A, 5-HT2C, and 5-HT6 may be similar in that they share “paradoxical” downregulation by clozapine. The high affinity of clozapine for 5-HT6 and its effect on expression indicates that it may be an important site for clinical action.
The gene 5-HT6 has six polymorphic variants (Vogt et al., 2000), 126G/T, 267C/T, 873 30C/T, 873 128A/C, 1128G/C, and 1376T/G, none of which change the amino acid sequence of the protein. One of these (267C/T) has been examined for association with clozapine response, and a weak association was found (Yu et al., 1999), but this was not replicated in a sample from the USA (Masellis et al., 2001).
231 Clozapine response and neurotransmitter receptor targets
Serotonin 5-HT7 receptors
Clozapine is also a high-affinity agonist for the 5-HT7 receptor (Roth et al., 1994). This receptor is expressed in midline, thalamic, and limbic structures in the brain (Eglen et al., 1997; Vanhoenacker et al., 2000) where it is thought to regulate cranial vasodilatation, emotion, and circadian rhythms. In contrast to 5-HT6, it is not downregulated by clozapine (Zhukovskaya and Neumaier, 2000). One polymor- phism is known in 5HT7, giving rise to Pro279Leu (Pesonen et al., 1998) and this has not been associated with clozapine response (Masellis et al., 2001).
Serotonin transporter
The transporter 5-HT (5-HTT) has also been postulated to influence response to clozapine, as functional variation in the gene may affect the efficacy of antipsy- chotic drugs that act through the 5-HT system (Lesch, 1998). 5-HTT gene knock- out results in marked desensitization of 5-HT1A autoreceptors in the dorsal raphe nucleus without altering postsynaptic 5-HT1A receptor functioning in the hippo- campus (Cour et al., 2001). One study has examined two polymorphisms in 5-HTT (a polymorphic repetitive element upstream of the transcription start site (5- HTTLPR) and a VNTR in intron 2) for association with clozapine response (Arranz et al., 2000b), one of these, the functional 5-HTTLPR, showed a trend toward asso- ciation with clozapine response but failed to reach statistical significance.
Histamine receptors
With the recent discovery of a novel histamine receptor gene HHR4 (Nakamura et al., 2000; Oda et al., 2000; Liu et al., 2001; Morse et al., 2001; Nguyen et al., 2001), there are now four known receptors (H1–H4) encoded by four separate genes (Hough, 2001). All are G0 protein-coupled receptors, transducing signals via Gq (H1), Gs (H2), and Gi/0 (H3, H4). Both H1 and H2 receptors are expressed in the caudate, putamen, neocortex, and hippocampus, whereas H3 receptors are expressed in the basal ganglia and globus pallidus and H4 receptors in peripheral tissues (eosinophils and bone marrow) but not the brain (Nakamura et al., 2000). However RNAse protection assays indicate that some H4 receptors may, in fact, be present in the brain (Liu et al., 2001). Clozapine is an agonist at all of these except the H3 receptor (Ki 10000nmol/l), with strongest affinity for H1 (Ki 2.8nmol/l) and only moderate affinity for H2 (Ki 100nmol/l), and H4 (Ki 510–693nmol/l). H1 is the only receptor likely to be significantly occupied at physiological concen- trations of clozapine.
H1 and H2 genes have been screened for polymorphisms and tested for an influ- ence on clozapine response (Mancama et al., 2000). Five polymorphisms (giving Lys19Asn, Asp349Glu, A1068G, Phe358 and Leu449Ser) are known in H1 and only one (G543A) in H2. However, no evidence of a role in influencing clozapine response was found.
232 D. A. Collier et al.
Cholinergic muscarinic receptors
There are five known cholinergic muscarinic receptors, encoded by five separate intronless genes, CHRM1 through to CHRM5. Clozapine binds with high affinity to all of these receptors in in vitro binding studies (Bymaster et al., 1999), with Ki values 1.4nmol/l (M1) to 10nmol/l (M2), very close to those for olanzapine. However, there is evidence from binding studies in intact cells that these measures may be overestimates, with the true range nearer to 31nmol/l (M1) to 204nmol/l (M2) (Bymaster and Falcone, 2000). Risperidone and typical drugs such as halo- peridol have low muscarinic affinity.
Historically, the cholinergic system was one of the first to be investigated in the pharmacological treatment of schizophrenia, where the use of anticholinesterase agents such as physostigmine and cholinergic stimulation with acetylcholine and arecoline showed some efficacy. A role for hypercholinergia has been postulated for psychosis, and there is also evidence that atypical antipsychotic action may be mediated by muscarinic receptors. For example the M1/M4 agonist xanomeline, which has little or no affinity for dopamine receptors, has antipsychotic-like activ- ity in rats and mice (Shannon et al., 2000).
Genetic variation within the five muscarinic receptors has been proposed to con- tribute toward the differences in treatment response observed amongst patients. To-date four polymorphic loci have been identified within CHMR1: 267C/A, 1044G/A, 1221C/T and 1353C/T, while single polymorphisms have been identified in CHMR3 (193G/A) and CHMR4 (1338C/T). However, none of these have been found to influence clozapine response (D. Mancama, personal communication).
Alpha-adrenoceptors
There are seven -adrenoceptors, encoded by separate genes (ADRA1A–1D and ADRA2A–2C). Pharmacologically, these receptors are broadly divided into 1- and 2-adrenoceptors, and clozapine is a high-affinity antagonist for both these classes, with a Ki of between 4 and 12nmol/l. There is little information on the affinity of clozapine for individual receptor subtypes.
Central 1-noradrenergic neurotransmission has been shown to be an import- ant complement of dopaminergic transmission in the control of motor activity. The 1A-adrenoceptor is widespread throughout the rat central nervous system, with high levels in olfactory regions, hypothalamic nuclei, brainstem, and spinal cord, particularly in areas related to motor function. The mRNA for 1B-receptors is highly expressed in the pineal gland, most thalamic nuclei, lateral nucleus of the amygdala, and dorsal and median raphe nuclei, with moderate expression levels noted throughout the cerebral cortex and other regions. The 1D-adrenoceptor is expressed in the olfactory bulb, cerebral cortex, hippocampus, reticular thalamic nucleus, regions of the amygdala, motor nuclei of the brainstem, inferior olivary
233 Clozapine response and neurotransmitter receptor targets
complex and spinal cord. From studies of transgenic mice, the 1B-adrenoceptor has been implicated in memory consolidation and fear-motivated exploratory behaviors (Knauber and Muller, 2000), which are both relevant to the cognitive and social deficits in schizophrenia, as well as having peripheral involvement in vascu- lar and blood pressure responses (Cavalli et al., 1997).
The 2-adrenoceptor subtypes inhibit adenylyl cyclase, are G-protein coupled and are widely expressed in human tissues, including brain (Eason and Liggett, 1993). Both the 2A- and 2C-adrenoceptors are required for normal presynaptic control of neurotransmitter release from central noradrenergic neurons (Hein et al., 1999). Antagonism of 2-adrenoceptors enhances adrenergic transmission and reinforces frontocortical dopaminergic transmission, whereas blockade of 1- adrenoceptors inhibits dorsal raphe-derived serotonergic pathways. This profile of activity may contribute to the antipsychotic properties of atypical drugs such as clozapine.
One study has examined adrenoreceptor genes in clozapine response (Bolonna et al., 2000). One polymorphism in ADRA1A (Arg492Cys) and two polymor- phisms in the promoter region of ADRA2A ( 1291C/G and 261G/A) were invest- igated in a sample of clozapine-treated schizophrenic patients scored retrospectively for response using the GAS. However, there were no differences between the responder and nonresponder groups, suggesting that these polymor- phisms do not play an important role in determining antipsychotic response.
GABA receptors
Although dopamine, 5-HT, cholinergic muscarinic, histamine and adreno recep- tors are thought to be the major receptors for clozapine in the brain, other systems may play an important role in clozapine’s action. Clozapine is a weak antagonist at GABA-A receptors (Squires and Saederup, 1997, 1998), can reverse the inhibitory effect of GABA on the binding of an inhibitory antagonist at GABA receptors (Squires and Saederup, 1991) and shows a preference for specific combinations of - and -subtypes (Korpi et al., 1995). This evidence suggests that clozapine can reduce GABA receptor activation; more recent work suggests that functional antag- onism occurs at physiological concentrations of the drug (Michel and Trudeau, 2000). However, GABA receptors have not been tested for association with cloza- pine response as yet.
Interaction between genes
Clozapine’s complex array of receptor binding makes it likely that multiple recep- tors contribute to the clinical action of this drug. It is, therefore, easy to imagine that genetic variation in several of the genes encoding these receptors will influ- ence clinical response to clozapine. Additive genetic analysis of combinations of
234 D. A. Collier et al.
these polymorphisms should, therefore, improve the pharmacogenetic prediction of clozapine response over that considering only single gene variants. Arranz et al. (2000a) have attempted to test this approach using a series of polymorphisms from genes previously tested in a retrospectively assessed sample for association with clozapine. From 19 polymorphisms tested, a combination of six (5-HT2C 102T/C and His452Tyr; 5-HT2C 330G/T/ 244C/T and Ser23Cys; 5-HTTLPR; and H2 1018G/A) were selected on the basis that they showed a p value of 0.09 for asso- ciation with response to clozapine. This combination gave a positive “predictive” value of 0.76 and a negative predictive value of 0.82, with a p value overall of 0.0001. A criticism of this approach is that it is a post hoc analysis of a series of polymor- phisms not all of which have been shown to be individually associated with cloza- pine response but which have been selected for p values below an arbitrary level (0.09). This finding should be regarded as exploratory and needs careful replica- tion in a prospectively assessed sample. However, it points the way forward to the prospect of combined analysis of multiple polymorphisms to provide accurate pre- dictive information on clozapine response.
Conclusions
The aim of pharmacogenetic research is to define composite genetic tests that can be used to predict the outcome of drug treatment. This outcome can be defined in several ways and includes clinical response to treatment (reduction or elimination of disease symptomatology), vulnerability to side effects, or improvement in more specific disease-related measures such as cognitive performance. Such tests will provide benefits to the patient by identifying treatments likely to be successful early on in the disease process; early intervention appears to be important for long-term prognosis in schizophrenia. In addition, drug safety will be improved, as those patients most vulnerable to side effects can be identified in advance and either care- fully monitored or given alternative therapy. Finally, these tests will have economic benefit in a climate of treatment rationing as the rate of unsuccessful treatment with costly drugs such as atypical antipsychotics will be avoided.
Clozapine therapy has been used as a model system for pharmacogenetic research into antipsychotic drugs. This is largely for the historical reasons that it has high affinity for the dopamine D4 receptor, which shows extensive genetic variabil- ity, but also because its pharmacology is well understood. Although polymor- phisms in two genes, 5-HT2A and dopamine D3, have been implicated in more than one study as influencing response to clozapine, research into its pharmacogenetics is still in its infancy, particularly with respect to genetic influence on side effects.
A novel epistatic approach, simultaneously examining genetic variation in multi- ple genes that encode clozapine’s neurotransmitter targets, has been performed,

235 Clozapine response and neurotransmitter receptor targets
and this points the way to analysis of the future. However, there is still a general lack of large, powerful prospective studies using multiple measures of response, and many polymorphisms in potentially relevant genes, such as GABA-A, have not yet been examined.
It is also important to note that pharmacogenetic tests will not be clinically useful if they only apply to a single drug such as clozapine. A genetic test will be most val- uable in choosing which drug will be, on balance, most efficacious and least risky from a panel of suitable choices. This will require extensive pharmacogenetic anal- ysis of a variety of antipsychotic drugs, both typical and atypical.
To predict fully clinical response to antipsychotic treatment in schizophrenia, it will be necessary to take account of many variables in addition to genetic influence on clozapine’s pharmacokinetics and pharmacodynamics, including social, demo- graphic, and clinical factors (e.g., compliance, social support, and history of birth trauma); genetic vulnerability to side effects such as weight gain and sedation; and the influence of genetic susceptibility factors for psychosis, which may define subtype and symptomatology. When this is achieved, a simple and reliable process for the individualization of treatment will be attainable.
REFERENCES
Ahlenius S (1999). Clozapine: dopamine D1 receptor agonism in the prefrontal cortex as the code to decipher a Rosetta stone of antipsychotic drugs. Pharmacol Toxicol 84, 193–196.
Amsler HA, Teerenhovi L, Barth E, Harjula K, Vuopio P (1977). Agranulocytosis in patients treated with clozapine. A study of the Finnish epidemic. Acta Psychiatr Scand 56, 241–248. Arranz M, Collier D, Sodhi M et al. (1995). Association between clozapine response and allelic
variation in 5-HT2A receptor gene. Lancet 346, 281–282.
Arranz MJ, Munro J, Sham P et al. (1998a). Meta analysis of studies on variation in 5HT2A recep-
tors and clozapine response. Schizophr Res 32, 93–99.
Arranz MJ, Li T, Munro J et al. (1998b). Lack of association between a polymorphism in the pro-
moter region of the dopamine-2 receptor gene and clozapine response. Pharmacogenetics 8,
481–484.
Arranz MJ, Munro J, Owen MJ et al. (1998c). Evidence for association between polymorphisms
in the promoter and coding regions of the 5-HT2A receptor gene and response to clozapine.
Mol Psychiatry 3, 61–66.
Arranz MJ, Munro J, Birkett J et al. (2000a). Pharmacogenetic prediction of clozapine response.
Lancet 355, 1615–1616.
Arranz MJ, Bolonna AA, Munro J, Curtis CJ, Collier DA, Kerwin RW (2000b). The serotonin
transporter and clozapine response. Mol Psychiatry 5, 124–125.
Barnes NM, Sharp T (1999). A review of central 5-HT receptors and their function.

Neuropharmacology 38, 1083–1152.
236 D. A. Collier et al.
Basile VS, Masellis M, Badri F et al. (1999). Association of the MscI polymorphism of the dopa- mine D3 receptor gene with tardive dyskinesia in schizophrenia. Neuropsychopharmacology 21, 17–27.
Bengtsson HJ, Kullberg A, Millan MJ, Hjorth S (1998). The role of 5-HT1A autoreceptors and alpha1-adrenoceptors in the modulation of 5-HT release – III. Clozapine and the novel puta- tive antipsychotic S 16924. Neuropharmacology 37, 349–356.
Bergen A, Wang CY, Nakhai B, Goldman D (1996). Mass allele detection (MAD) of rare 5-HT1A structural variants with allele-specific amplification and lectrochemiluminescent detection. Hum Mutat 7, 135–143.
Bhatnagar A, Willins DL, Gray JA, Woods J, Benovic JL, Roth BL (2001). The dynamin-depen- dent, arrestin-independent internalization of 5-hydroxytryptamine 2A (5-HT2A) serotonin receptors reveals differential sorting of arrestins and 5-HT2A receptors during endocytosis. J Biol Chem 276, 8269–8277.
Bilder RM, Goldman RS et al. (2000). Neuropsychology of first-episode schizophrenia: initial characterization and clinical correlates. Am J Psychiatry 157, 549–559.
Birkett JT, Arranz MJ, Munro J, Osborne S, Kerwin RW, Collier DA (2000). Association analysis of the 5-HT5A gene in depression, psychosis and antipsychotic response. Neuroreport 11, 2017–2020.
Bolonna AA, Arranz MJ, Munro J et al. (2000). No influence of adrenergic receptor polymor- phisms on schizophrenia and antipsychotic response. Neurosci Lett 280, 65–68.
Breier A, Buchanan RW, Irish D, Carpenter, WT Jr. (1993). Clozapine treatment of outpatients with schizophrenia: outcome and long-term response patterns. Hosp Community Psychiatry 44, 1145–1149.
Bristow LJ, Kramer MS, Kulagowski J, Patel S, Ragan CI, Seabrook GR (1997). Schizophrenia and L-745, 870, a novel dopamine D4 receptor antagonist. Trend Pharmacol Sci 18, 186–188.
Broekkamp CL, Berendsen HH, Jenck F, van Delft AM (1989). Animal models for anxiety and response to serotonergic drugs. Psychopathology 22(Suppl 1), 2–12.
Brunello N, Masotto C, Steardo L, Markstein R, Racagni G (1995). New insights into the biology of schizophrenia through the mechanism of action of clozapine. Neuropsychopharmacology 13, 177–213.
Burns CM, Chu H, Rueter SM et al. (1997). Regulation of serotonin-2C receptor G-protein cou- pling by RNA editing. Nature 387, 303–308.
Bymaster FP, Falcone JF (2000). Decreased binding affinity of olanzapine and clozapine for human muscarinic receptors in intact clonal cells in physiological medium. Eur J Pharmacol 390, 245–248.
Bymaster FP, Nelson DL, DeLapp NW et al. (1999). Antagonism by olanzapine of dopamine D1, serotonin 2, muscarinic, histamine H1 and alpha1-adrenergic receptors in vitro. Schizophr Res 37, 107–122.
Cavalli A, Lattion AL, Hummler E et al. (1997). Decreased blood pressure response in mice defi- cient of the alpha1b-adrenergic receptor. Proc Natl Acad Sci, USA 94, 11589–11594.
Cichon S, Nothen MM, Stober G et al. (1996). Systematic screening for mutations in the 5 -reg- ulatory region of the human dopamine D1 receptor (DRD1) gene in patients with schizophre- nia and bipolar affective disorder. Am J Med Genet 67, 424–428.
237 Clozapine response and neurotransmitter receptor targets
Cohen BM, Ennulat DJ, Centorrino F et al. (1999). Polymorphisms of the dopamine D4 recep- tor and response to antipsychotic drugs. Psychopharmacology (Berl) 141, 6–10.
Cour CM, Boni C, Hanoun N, Lesch KP, Hamon M, Lanfumey L (2001). Functional conse- quences of 5-HT transporter gene disruption on 5-HT1A receptor-mediated regulation of dorsal raphe and hippocampal cell activity. J Neurosci 21, 2178–2185.
Crocq MA, Mant R, Asherson P et al. (1992). Association between schizophrenia and homozy- gosity at the dopamine D3 receptor gene. J Med Genet 29, 858–860.
Davies PA, Pistis M, Hanna MC et al. (1999). The 5-HT3B subunit is a major determinant of sero- tonin-receptor function. Nature 397, 359–363.
Deckert J, Catalano M, Heils A et al. (1997). Functional promoter polymorphism of the human serotonin transporter: lack of association with panic disorder. Psychiatr Genet 7, 45–47.
Eason MG, Liggett SB (1993). Human alpha-2-adrenergic receptor subtype distribution: wide- spread and subtype-selective expression of alpha-2–C10, alpha-2–C4, and alpha-2–C2 mRNA in multiple tissues. Mol Pharmacol 44, 70–75.
Egan C, Herrick-Davis K, Teitler M (1998). Creation of a constitutively activated state of the 5- HT2A receptor by site-directed mutagenesis: revelation of inverse agonist activity of antago- nists. Ann NY Acad Sci 861, 136–139.
Eglen RM, Jasper JR, Chang DJ, Martin GR (1997). The 5-HT7 receptor: orphan found. Trend Pharmacol Sci 18, 104–107.
Eichhammer P, Albus M, Borrmann-Hassenbach M et al. (2000). Association of dopamine D3- receptor gene variants with neuroleptic induced akathisia in schizophrenic patients: a gener- alization of Steen’s study on DRD3 and tardive dyskinesia. Am J Med Genet 96, 187–191.
Endicott J, Spilzer RL, Fless JL, Cohen J (1976). The global assessment scale. Arch Gen Psychiatry 33, 766–772.
Erdmann J, Shimron-Abarbanell D, Cichon S et al. (1995). Systematic screening for mutations in the promoter and the coding region of the 5-HT1A gene. Am J Med Genet 60, 393–399.
Erdmann J, Shimron-Abarbanell D, Rietschel M et al. (1996). Systematic screening for mutations in the human serotonin-2A (5-HT2A) receptor gene: identification of two naturally occurring receptor variants and association analysis in schizophrenia. Hum Genet 97, 614–619.
Feng J, Sobell JL, Heston LL, Cook EH Jr., Goldman D, Sommer SS (1998). Scanning of the dop- amine D1 and D5 receptor genes by REF in neuropsychiatric patients reveals a novel missense change at a highly conserved amino acid. Am J Med Genet 81, 172–178.
Fitton A, Heel RC (1990). Clozapine: a review of its pharmacological properties, and therapeu- tic use in schizophrenia. Drugs 40, 722–747.
Gerlach J, Koppelhus P, Helweg E, Monrad A (1974). Clozapine and haloperidol in a single-blind cross-over trial: therapeutic and biochemical aspects in the treatment of schizophrenia. Acta Psychiatr Scand 50, 410–424.
Glatt CE, Snowman AM, Sibley DR, Snyder SH (1995). Clozapine: selective labeling of sites resembling 5HT6 serotonin receptors may reflect psychoactive profile. Mol Med 1, 398–406. Guy W (1976). ECDEU Assessment Manual for Psychopharmacology, revised edn. Bethesda, MD:
US Department of Health, Education and Welfare, pp. 217–222.
Hein L, Altman JD, Kobilka BK (1999). Two functionally distinct alpha-2-adrenergic receptors
regulate sympathetic neurotransmission. Nature 402, 181–184.
238 D. A. Collier et al.
Hermann B, Wetzel CH, Pestel E, Zieglgansberger W, Holsboer F, Rupprecht R (1996). Functional antagonistic properties of clozapine at the 5-HT3 receptor. Biochem Biophys Res Commun 225, 957–960.
Herrick-Davis K, Egan C, Teitler M (1997). Activating mutations of the serotonin 5-HT2C recep- tor. J Neurochem 69, 1138–1144.
Herrick-Davis K, Grinde E, Gauthier C, Teitler M (1998). Pharmacological characterization of the constitutively activated state of the serotonin 5-HT2C receptor. Ann NY Acad Sci 861, 140–145.
Herrick-Davis K, Grinde E, Niswender CM (1999). Serotonin 5-HT2C receptor RNA editing alters receptor basal activity: implications for serotonergic signal transduction. J Neurochem 73, 1711–1717.
Herrick-Davis K, Grinde E, Teitler M (2000). Inverse agonist activity of atypical antipsychotic drugs at human 5-hydroxytryptamine 2C receptors. J Pharmacol Exp Ther 295, 226–232.
Hough LB (2001). Genomics meets histamine receptors: new subtypes, new receptors. Mol Pharmacol 59, 415–419.
Hoyer D (1988). Molecular pharmacology and biology of 5-HT1C receptors. Trends Pharmacol Sci 9, 89–94.
Ichikawa J, Ishii H, Bonaccorso S, Fowler WL, O’Laughlin IA, Meltzer HY (2001). 5-HT(2A) and D(2) receptor blockade increases cortical DA release via 5-HT(1A) receptor activation: a pos- sible mechanism of atypical antipsychotic-induced cortical dopamine release. J Neurochem 76, 1521–1531.
Ichikawa J, Meltzer HY (2000). The effect of serotonin (1A) receptor agonism on antipsychotic drug-induced dopamine release in rat striatum and nucleus accumbens. Brain Res 858, 252–263.
Ilani T, Ben-Shachar D, Strous RD et al. (2001). A peripheral marker for schizophrenia: increased levels of D3 dopamine receptor mRNA in blood lymphocytes. Proc Natl Acad Sci, USA 98, 625–628.
Ishiguro H, Okuyama Y, Toru M, Arinami T (2000). Mutation and association analysis of the 5 region of the dopamine D3 receptor gene in schizophrenia patients: identification of the Ala38Thr polymorphism and suggested association between DRD3 haplotypes and schizo- phrenia. Mol Psychiatry 5, 433–438.
Jakab RL, Goldman-Rakic PS (1998). 5-Hydroxytryptamine 2A serotonin receptors in the primate cerebral cortex: possible site of action of hallucinogenic and antipsychotic drugs in pyramidal cell apical dendrites. Proc Natl Acad Sci. USA 95, 735–740.
Jovanovic V, Guan HC, van Tol HH (1999). Comparative pharmacological and functional anal- ysis of the human dopamine D4.2 and D4.10 receptor variants. Pharmacogenetics 9, 561–568. Kaiser R, Konneker M, Henneken M et al. (2000). Dopamine D4 receptor 48-bp repeat polymor- phism: no association with response to antipsychotic treatment, but association with catatonic
schizophrenia. Mol Psychiatry 5, 418–424.
Kapur S, Remington G (2001). Atypical antipsychotics: new directions and new challenges in the
treatment of schizophrenia. Annu Rev Med 52, 503–517.
Kapur S, Seeman P (2000). Antipsychotic agents differ in how fast they come off the dopamine
D2 receptors: implications for atypical antipsychotic action. J Psychiatr Neurosci 25, 161–166.
239 Clozapine response and neurotransmitter receptor targets
Kawanishi Y, Harada S, Tachikawa H, Okubo T, Shiraishi H (1998). Novel mutations in the pro- moter and coding region of the human 5-HT1A receptor gene and association analysis in schizophrenia. Am J Med Genet 81, 434–439.
Kay SR, Opler LA, Fiszbein A (1992). Positive and Negative Symptom Scale (PANSS). Niagra Falls, NY: Multi-Health Systems.
Knauber J, Muller WE (2000). Decreased exploratory activity and impaired passive avoidance beha- viour in mice deficient for the alpha(1b)-adrenoceptor. Eur Neuropsychopharmacol 10, 423–427. Kohen R, Metcalf MA, Khan N et al. (1996). Protein, nucleotide, OMIM cloning, characteriza- tion, and chromosomal localization of a human 5-HT6 serotonin receptor. J Neurochem 66,
47–56.
Kohn Y, Ebstein RP, Heresco-Levy U et al. (1997). Dopamine D4 receptor gene polymorphisms:
relation to ethnicity, no association with schizophrenia and response to clozapine in Israeli
subjects. Eur J Neuropsychopharmacol 7, 39–43.
Korpi ER, Wong G, Luddens H (1995). Subtype specificity of gamma-aminobutyric acid type A
receptor antagonism by clozapine. Naunyn-Schmiedeberg Arch Pharmacol 352, 365–373. Kulkarni SK, Ninan I (2000). Dopamine D4 receptors and development of newer antipsychotic
drugs. Fund Clin Pharmacol 14, 529–539.
Kuroki T, Meltzer HY, Ichikawa J (1999). Effects of antipsychotic drugs on extracellular dopa-
mine levels in rat medial prefrontal cortex and nucleus accumbens. J Pharmacol Exp Ther 288,
774–781.
Lehman AF, Postrado LT, Rachuba LT (1993). Convergent validation of quality of life assessments
for persons with severe mental illnesses. Qual Life Res 2, 327–333.
Lesch KP (1998). Hallucinations: psychopathology meets functional genomics. Mol Psychiatry 3,
278–281.
Lichter JB, Barr CL, Kennedy JL, van Tol HH, Kidd KK, Livak KJ (1993). A hypervariable segment
in the human dopamine receptor D4 (DRD4) gene. Hum Mol Genet 2, 767–773.
Lidow MS, Elsworth JD, Goldman-Rakic PS (1997). Down-regulation of the D1 and D5 dopa- mine receptors in the primate prefrontal cortex by chronic treatment with antipsychotic drugs.
J Pharmacol Exp Ther 281, 597–603.
Lieberman JA, Kane JM, Safferman AZ et al. (1994a). Predictors of response to clozapine. J Clin
Psychiatry 55(Suppl B), 126–128.
Lieberman JA, Safferman AZ, Pollack S et al. (1994b). Clinical effects of clozapine in chronic
schizophrenia: response to treatment and predictors of outcome. Am J Psychiatry 151,
1744–1752.
Light GA, Geyer MA, Clementz BA, Cadenhead KS, Braff DL (2000). Normal P50 suppression in
schizophrenia patients treated with atypical antipsychotic medications. Am J Psychiatry 157,
767–771.
Liu C, Wilson SJ, Kuei C, Lovenberg TW (2001). Comparison of human, mouse, rat, and guinea
pig histamine H4 receptors reveals substantial pharmacological species variation. J Pharmacol
Exp Ther 299, 121–130.
Lovlie R, Daly AK, Blennerhassett R, Ferrier N, Steen VM (2000). Homozygosity for the Gly-9
variant of the dopamine D3 receptor and risk for tardive dyskinesia in schizophrenic patients. Int J Neuropsychopharmacol 3, 61–65.
240 D. A. Collier et al.
Luborsky L (1962). Clinician’s judgement of mental health. Arch Gen Psychiatry 7, 407–417. Malhotra AK, Goldman D, Ozaki N, Breier A, Buchanan R, Pickar D (1996a). Lack of associa- tion between polymorphisms in the 5-HT2A receptor gene and the antipsychotic response to
clozapine. Am J Psychiatry 153, 1092–1094.
Malhotra AK, Goldman D, Ozaki N et al. (1996b). Clozapine response and the 5HT2C Cys23Ser
polymorphism. Neuroreport 7, 2100–2102.
Malhotra AK, Goldman D, Buchanan RW et al. (1998). The dopamine D3 receptor (DRD3)
Ser9Gly polymorphism and schizophrenia: a haplotype relative risk study and association with
clozapine response. Mol Psychiatry 3, 72–75.
Malmberg A, Mikaels A, Mohell N (1998). Agonist and inverse agonist activity at the dopamine
D3 receptor measured by guanosine 5 -gamma-thio-triphosphate–35S-binding. J Pharmacol
Exp Ther 285, 119–126.
Mancama D, Arranz MJ, Munro J, Makoff A, Kerwin R (2000). The histamine H1 and histamine
H2 genes: candidates for schizophrenia and clozapine drug response. GeneScreen 1, 29–34. Maricq AV, Peterson AS, Brake AJ, Myers RM, Julius D (1991). Primary structure and functional
expression of the 5HT3 receptor, a serotonin-gated ion channel. Science 254, 432–437. Masellis M, Paterson AD, Badri F et al. (1995). Genetic variation of 5-HT2A receptor and response
to clozapine. Lancet 346, 1108.
Masellis M, Basile V, Meltzer HY et al. (1998). Serotonin subtype 2 receptor genes and clinical
response to clozapine in schizophrenia patients. Neuropsychopharmacology 19, 123–132. Masellis M, Basile VS, Ozdemir V, Meltzer HY, Macciardi FM, Kennedy JL (2000). Pharmacogenetics of antipsychotic treatment: lessons learned from clozapine. Biol Psychiatry
47, 252–266.
Masellis M, Basile VS, Meltzer HY et al. (2001). Lack of association between the T → C267 sero-
tonin 5-HT6 receptor gene (HTR6) polymorphism and prediction of response to clozapine in
schizophrenia. Schizophr Res 47, 49–58.
Mata I, Madoz V, Arranz MJ, Sham P, Murray RM (2001). Olanzapine: concordant response in
monozygotic twins with schizophrenia. Br J Psychiatry 178, 86.
Meltzer HY (1992). Dimensions of outcome with schizophrenia. Br J Psychiatry 160(Suppl 17),
46–53.
Meltzer HY (1994). An overview of the mechanism of action of clozapine. J Clin Psychopharmacol
55 (Suppl B), 47–52.
Meltzer HY (1995). The role of serotonin in schizophrenia and the place of serotonin-dopamine
antagonist antipsychotics. J Clin Psychopharmacol 15(Suppl 1), 2S-3S.
Meltzer HY (1999). The role of serotonin in antipsychotic drug action. Neuropsychopharmacology
2(Suppl), 106S–115S.
Meltzer HY, McGurk SR (1999). The effects of clozapine, risperidone, and olanzapine on cogni-
tive function in schizophrenia. Schizophr Bull 25, 233–255.
Michel FJ, Trudeau LE (2000). Clozapine inhibits synaptic transmission at GABAergic synap-
ses established by ventral tegmental area neurones in culture. Neuropharmacology 39,
1536–1543.
Missale C, Nash SR, Robinson SW, Jaber M, Caron MG (1998). Dopamine receptors: from struc-
ture to function. Physiol Rev 78, 189–225.
241 Clozapine response and neurotransmitter receptor targets
Morse KL, Behan J, Laz TM et al. (2001). Cloning and characterization of a novel human hista- mine receptor. J Pharmacol Exp Ther 296, 1058–1066.
Nagamoto HT, Adler LE, Hea RA, Griffith JM, McRae KA, Freedman R (1996). Gating of audi- tory P50 in schizophrenics: unique effects of clozapine. Biol Psychiatry 40, 181–188.
Nair NP, Zicherman V, Schwartz G (1977). Dopamine and schizophrenia. A reappraisal in the light of clinical studies with clozapine. Can Psychiatr Assoc J 22, 285–293.
Nakamura T, Itadani H, Hidaka Y, Ohta M, Tanaka K (2000). Protein, nucleotide molecular cloning and characterization of a new human histamine receptor, HH4R. Biochem Biophys Res Commun 279, 615–620.
Nakhai B, Nielsen DA, Linnoila M, Goldman D (1995). Two naturally occurring amino acid sub- stitutions in the human 5-HT1A receptor: glycine 22 to serine 22 and isoleucine 28 to valine 28. Biochem Biophys Res Commun 210, 530–536.
Newman-Tancredi A, Gavaudan S, Conte C et al. (1998). Agonist and antagonist actions of anti- psychotic agents at 5-HT1A receptors: a [35S]GTPgammaS binding study. Eur J Pharmacol 355, 245–256.
Newton RA, Elliott JM (1997). Mianserin-induced down-regulation of human 5-hydroxytrypt- amine 2A and 5-hydroxytryptamine 2C receptors stably expressed in the human neuroblas- toma cell line SH-SY5Y. J Neurochem 69, 1031–1038.
Nguyen T, Shapiro DA, George SR et al. (2001). Discovery of a novel member of the histamine receptor family. Mol Pharmacol 59, 427–433.
Niesler B, Weiss B, Fischer C et al. (2001). Serotonin receptor gene HTR3A variants in schizo- phrenic and bipolar affective patients. Pharmacogenetics 11, 21–22.
Niswender CM, Copeland SC, Herrick-Davis K, Emeson RB, Sanders-Bush E (1999). RNA editing of the human serotonin 5-hydroxytryptamine 2C receptor silences constitutive activ- ity. J Biol Chem 274, 9472–9478.
Noethen MM, Rietschel M, Erdmann J et al. (1995). Genetic variation of the 5-HT2A receptor and response to clozapine. Lancet 346, 908–909.
Oda T, Morikawa N, Saito Y, Masuho Y, Matsumoto S (2000). Molecular cloning and character- ization of a novel type of histamine receptor preferentially expressed in leukocytes. J Biol Chem 275, 36781–36786.
Okuyama Y, Ishiguro H, Toru M, Arinami T (1999). A genetic polymorphism in the promoter region of DRD4 associated with expression and schizophrenia. Biochem Biophys Res Commun 258, 292–295.
Overall JE, Gorham DR (1962). The brief psychiatric rating scale. Psychol Rep 10, 799–812. Ozdemir V, Basile VS, Masellis M, Kennedy JL (2001). Pharmacogenetic assessment of antipsy- chotic-induced movement disorders: contribution of the dopamine D3 receptor and cyto-
chrome P-450 1A2 genes. J Biochem Biophys Meth 47, 151–157.
Pasqualetti M, Ori M, Marazziti D, Castagna M, Nardi I (1998). Distribution of 5-HT2C and
5-HT5A receptor mRNA in human brain. Ann NY Acad Sci 861, 245.
Pesonen U, Koulu M, Bergen A et al. (1998). Mutation screening of the 5-hydroxytryptamine 7
receptor gene among Finnish alcoholics and controls. Psychiatry Res 77, 139–145.
Pilowsky LS, Mulligan RS, Acton PD, Ell PJ, Costa DC, Kerwin RW (1997). Limbic selectivity of
clozapine. Lancet 350, 490–491.
242 D. A. Collier et al.
Potkin SG, Bera R, Gulasekaram B et al. (1994). Plasma clozapine concentrations predict clinical response in treatment-resistant schizophrenia. J Clin Psychiatry 55(Suppl B), 133–136.
Price-Evans DA (1993). Genetic Factors in Drug Therapy: Clinical and Molecular Pharmacogenetics. Cambridge, UK: Cambridge University Press.
Rao PA, Pickar D, Gejman PV, Ram A, Gershon ES, Gelernter J (1994). Allelic variation in the D4 dopamine receptor (DRD4) gene does not predict response to clozapine. Arch Gen Psychiatry 51, 912–917.
Rietschel M, Naber D, Oberlander H et al. (1996). Efficacy and side-effects of clozapine: testing for association with allelic variation in the dopamine D4 receptor gene. Neuropsycho- pharmacology 15, 491–496.
Rietschel M, Naber D, Fimmers R, Moller HJ, Propping P, Nothen MM (1997). Efficacy and side- effects of clozapine not associated with variation in the 5-HT2C receptor. Neuroreport 8, 1999–2003.
Robinson DG, Woerner MG, Alvir JM et al. (1999). Predictors of treatment response from a first episode of schizophrenia or schizoaffective disorder. Am J Psychiatry 156, 544–549.
Rosenheck R, Evans D, Herz L et al. (1999). How long to wait for a response to clozapine: a com- parison of time course of response to clozapine and conventional antipsychotic medication in refractory schizophrenia. Schizophr Bull 25, 709–719.
Roth BL, Craigo SC, Choudhary MS et al. (1994). Binding of typical and atypical antipsychotic agents to 5-hydroxytryptamine-6 and 5-hydroxytryptamine-7 receptors. J Pharmacol Exp Ther 268, 1403–1410.
Scharfetter J, Chaudhry HR, Hornik K et al. (1999). Dopamine D3 receptor gene polymorphism and response to clozapine in schizophrenic Pakistani patients. Eur Neuropsychopharmacol 10, 17–22.
Schwartz JC, Levesque D, Martres MP, Sokoloff P (1993). Dopamine D3 receptor: basic and clin- ical aspects. Clin Neuropharmacol 16, 295–314.
Seeman P (1992). Dopamine receptor sequences. Therapeutic levels of neuroleptics occupy D2 receptors, clozapine occupies D4. Neuropsychopharmacology 7, 261–284.
Segman R, Neeman T, Heresco-Levy U et al. (1999). Genotypic association between the dopa- mine D3 receptor and tardive dyskinesia in chronic schizophrenia. Mol Psychiatry 4, 247–253. Segman RH, Heresco-Levy U, Finkel B et al. (2000). Association between the serotonin 2C recep- tor gene and tardive dyskinesia in chronic schizophrenia: additive contribution of 5-HT2Cser
and DRD3gly alleles to susceptibility. Psychopharmacology (Berl) 152, 408–413.
Shaikh S, Collier D, Kerwin RW et al. (1993). Dopamine D4 receptor subtypes and response to
clozapine. Lancet 341, 116.
Shaikh S, Collier DA, Sham P et al. (1995). Analysis of clozapine response and polymorphisms
of the dopamine D4 receptor gene (DRD4) in schizophrenic patients. Am J Med Genet 60,
541–545.
Shaikh S, Collier DA, Sham PC et al. (1996). Allelic association between a Ser-9–Gly polymor-
phism in the dopamine D3 receptor gene and schizophrenia. Hum Genet 97, 714–719.
Shaikh S, Makoff A, Collier DA, Kerwin RW (1997). Dopamine D4 receptors. Potential therapeu-
tic implications in the treatment of schizophrenia. CNS Drugs 8, 1–11.
243 Clozapine response and neurotransmitter receptor targets
Shannon HE, Rasmussen K, Bymaster FP et al. (2000). Xanomeline, an M(1)/M(4) preferring muscarinic cholinergic receptor agonist, produces antipsychotic-like activity in rats and mice. Schizophr Res 42, 249–259.
Simpson GM, Josiassen RC, Stanilla JK, de Leon J et al. (1999). Double-blind study of clozapine dose response in chronic schizophrenia. Am J Psychiatry 156, 1744–1750.
Skevington SM, Wright A (2001). Changes in the quality of life of patients receiving antidepress- ant medication in primary care: validation of the WHOQOL-100. Br J Psychiatry 178, 261–267.
Snyder SH (1981). Dopamine receptors, neuroleptics, and schizophrenia. Am J Psychiatry 138, 460–446.
Sobell JL, Lind TJ, Sigurdson DC et al. (1995). The D5 dopamine receptor gene in schizophrenia: identification of a nonsense change and multiple missense changes but lack of association with disease. Hum Mol Genet 4, 507–514.
Sodhi MS, Arranz MJ, Curtis D et al. (1995). Association between clozapine response and allelic variation in the 5-HT2C receptor gene. Neuroreport 7, 169–172.
Sokoloff P, Andrieux M, Besancon M et al. (1992). Pharmacology of human D3 receptor expressed in a mammalian cell line: comparison with D2 receptor. Eur J Pharmacol 225, 331–337.
Spurlock G, Heils A, Holmans P et al. (1998). A family based association study of T102C poly- morphism in 5HT2A and schizophrenia plus identification of new polymorphisms in the pro- moter. Mol Psychiatry 3, 42–49.
Squires RF, Saederup E (1991). A review of evidence for GABAergic predominance/glutamater- gic deficit as a common etiological factor in both schizophrenia and affective psychoses: more support for a continuum hypothesis of “functional” psychosis. Neurochem Res 16, 1099–1111.
Squires RF, Saederup E (1997). Clozapine and some other antipsychotic drugs may preferentially block the same subset of GABA(A) receptors. Neurochem Res 22, 151–116.
Squires RF, Saederup E (1998). Clozapine and several other antipsychotic/antidepressant drugs preferentially block the same “core” fraction of GABA(A) receptors. Neurochem Res 23, 1283–129.
Squires RF, Saederup E (1999). Clozapine’s antipsychotic effects do not depend on blockade of 5-HT3 receptors. Neurochem Res 24, 659–667.
Steen VM, Lovlie R, MacEwan T, McCreadie RG (1997). Dopamine D3-receptor gene variant and susceptibility to tardive dyskinesia in schizophrenic patients. Mol Psychiatry 2, 139–145.
Tandon R (1993). Neuropharmacologic basis for clozapine’s unique profile. Arch Gen Psychiatry 50, 158–159.
Tress KH, Patton G (1994). The assessment of change in psychopathology. In Barnes TRE, Nelson HE, eds. The Assessment of Psychosis: A Practical Handbook. London: Chapman & Hall.
van Tol HH, Bunzow JR, Guan HC et al. (1991). Cloning of the gene for a human dopamine D4 receptor with high affinity for the antipsychotic clozapine. Nature 350, 610–614.
van Tol HH, Wu CM, Guan HC et al. (1992). Multiple dopamine D4 receptor variants in the human population. Nature 358, 149–152.
Vanhoenacker P, Haegeman G, Leysen JE (2000). 5-HT7 receptors: current knowledge and future prospects. Review. Trend Pharmacol Sci 21, 70–77.
244 D. A. Collier et al.
Vogt IR, Shimron-Abarbanell D, Neidt H et al. (2000). Investigation of the human serotonin 6 [5-HT6] receptor gene in bipolar affective disorder and schizophrenia. Am J Med Genet 96, 217–221.
Vojvoda D, Grimmell K, Sernyak M, Mazure CM (1996). Monozygotic twins concordant for response to clozapine. Lancet 61, 347.
Wadenberg ML, Ahlenius S (1991). Antipsychotic-like profile of combined treatment with rac- lopride and 8-OH-DPAT in the rat: enhancement of antipsychotic-like effects without cata- lepsy. J Neural Transm Genet Sect 83, 43–53.
Waeber C, Grailhe R, Yu XJ et al. (1998). Putative 5-HT5 receptors: localization in the mouse CNS and lack of effect in the inhibition of dural protein extravasation. Ann NY Acad Sci 861, 85–90. Warburton EC, Joseph MH, Feldon J, Weiner I, Gray JA (1994). Antagonism of amphetamine- induced disruption of latent inhibition in rats by haloperidol and ondansetron: implications
for a possible antipsychotic action of ondansetron. Psychopharmacology (Berl) 114, 657–664. Warren JT Jr., Peacock ML, Rodriguez LC, Fink JK (1993). An MspI polymorphism in the human serotonin receptor gene (HTR2): detection by DGGE and RFLP analysis. Hum Mol Genet 2,
338.
Watling KJ, Beer MS, Stanton JA, Newberry NR (1990). Interaction of the atypical neuroleptic
clozapine with 5-HT3 receptors in the cerebral cortex and superior cervical ganglion of the rat.
Eur J Pharmacol 182, 465–472.
Weiss B, Mertz A, Schrock E, Koenen M, Rappold G (1995). Assignment of a human homolog of
the mouse htr3 receptor gene to chromosome 11q23.1–q23.2. Genomics 29, 304–305.
Wilcox JA, Nasrallah HA (1987). Perinatal insult as a risk factor in paranoid and nonparanoid
schizophrenia. Psychopathology 20, 285–287.
Williams GV, Goldman-Rakic PS (1995). Modulation of memory fields by dopamine D1 recep-
tors in prefrontal cortex. Nature 376, 572–575.
Willins DL, Berry SA, Alsayegh L et al. (1999). Clozapine and other 5-hydroxytryptamine-2A
receptor antagonists alter the subcellular distribution of 5-hydroxytryptamine-2A receptors in
vitro and in vivo. Neuroscience 91, 599–606.
Wu S, Comings DE (1999). A common C 1018G polymorphism in the human 5-HT1A recep–
tor gene. Psychiatr Genet 9, 105–106.
Yu YW, Tsai SJ, Lin CH, Hsu CP, Yang KH, Hong CJ (1999). Serotonin-6 receptor variant (C267T)
and clinical response to clozapine. Neuroreport 10, 1231–1233.
Zhukovskaya NL, Neumaier JF (2000). Clozapine downregulates 5-hydroxytryptamine 6 (5-
HT6) and upregulates 5-HT7 receptors in HeLa cells. Neurosci Lett 288, 236–240. Zifa E, Fillion G (1992). 5-Hydroxytryptamine receptors. Pharmacol Rev 44, 401–458.
11
Genetic factors underlying drug-induced tardive dyskinesia
Ronnen H. Segman and Bernard Lerer
Department of Psychiatry, Hadassah-Hebrew University Medical Center, Jerusalem, Israel
OVERVIEW
Tardive dyskinesia (TD) affects about one fifth of schizophrenia patients following chronic exposure to dopamine receptor antagonist drugs. Spontaneous dyskinesia has been reported in unmedicated schizophrenia patients and patients with TD have been reported to show dis- tinct clinical features, suggesting a common underlying phenotype that may bear a distinct geneticpredisposition.Drug-andpatient-relatedriskfactorsforthedevelopmentofTDhave received much research attention but appear to predict only a minor part of the variance in the incidence of TD. Genetically determined individual variability in factors affecting drug levels, as well as compensatory responses to chronic dopaminergic antagonism, may account for a major portion of the variance in the incidence of TD. To-date, however, there has been a conspicuous lack of studies exploring a genetic predisposition to TD. Current data stem from sporadic clinical observations and from supportive, albeit indirect, evidence from rodent studies showing strain differences in behavioral and phamacodynamic models for drug- inducedTD.Despitethelackofanestablishedgeneticcontributionormodeofinheritance for vulnerability to develop TD, in recent years a number of groups have ventured to examine directly a possible contribution of specific candidate genes to TD, employing case-control association design in chronically medicated schizophrenia patients. Such studies report direct association of candidate polymorphic genes with drug-induced TD, providing further support for the existence of a genetic contribution and suggesting the likelihood of a polygenic, multi- factorial inheritance for such vulnerability. These studies open a window for improved under- standing of drug-induced extrapyramidal reactions, as well as a more selective schizophrenia phenotype. The current chapter will review this rapidly developing field, incorporating recently available molecular genetic tools for a renewed exploration of the biological basis of TD.
Introduction
Human evidence supporting a genetic basis for tardive dyskinesia
Tardive dyskinesia (TD) is an irreversible, debilitating adverse drug reaction char- acterized by the delayed appearance of involuntary choreoathetotic movements in

245

246 R. H. Segman and B. Lerer
patients chronically exposed to dopamine receptor antagonists. Prevalence esti- mates are confounded by factors such as a considerable rate of spontaneous dyski- nesias, as well as masking of TD by concurrent neuroleptic administration (Koller, 1988). Kane and Smith (1982) reviewed 50 prevalence studies conducted over a 20- year period and calculated a mean prevalence rate of 20%. Risk factors for the devel- opment of TD include older age and female gender. Other factors, less consistently confirmed, include duration and intensity of prior neuroleptic exposure, organic brain abnormalities and the existence of affective disorder (Burke et al., 1989). As drug-related variables predict only a minor part of the variance in the incidence of TD, a predominant pharmacogenetic component appears plausible. To-date, however, there has been a conspicuous lack of studies exploring a genetic predisposition to TD.
In a retrospective comparison of patients with chronic schizophrenia patients, those exhibiting TD showed a significantly increased rate of family history of psychiatric disorder, as well as a reduced rate of obstetric complications, suggest- ing the importance of a genetic predisposition (O’Callaghan et al., 1990). Also, there are case reports of concordance for TD in siblings exposed to neuroleptics, pointing to the importance of conducting systematic family studies (Weinhold et al., 1981; Muller et al., 1998a).
Three studies examined the correlation of TD with human leukocyte antigen (HLA) subtypes. HLA-B44 was associated with increased incidence of TD in one study (Canoso et al., 1986), but this finding was not replicated in two other studies (Meltzer et al., 1990; Brown and White, 1991). HLA-DR4 was found to increase the relative risk for TD to 3.04 in a group of 53 patients with chronic schizophrenia (Meltzer et al., 1990).
Recently, increasing research efforts have directly examined the genetic make-up of patients exhibiting spontaneously occurring extrapyramidal disorders, such as Parkinson’s disease and the dystonias. Whereas single major loci have been impli- cated with familial forms of Parkinson’s disease (Gasser, 2000) and dystonia (Muller et al., 1998b), the more prevalent form of sporadic Parkinson’s disease has been hypothesized to relate to a multifactorial polygenetic etiology where multiple small-effect loci may increase the risk for neurotoxin-induced nigral degeneration depending on combined effects of environmental exposure and interactions with other genetic susceptibility factors. Importantly, such loci were found to contrib- ute a small proportion to the risk for incurring spontaneous sporadic Parkinson’s disease (Tan et al., 2000), pointing to the relevance of a candidate gene association design in such a context.
Animal data supporting a genetic basis for tardive dyskinesia
Studies in rats chronically exposed to neuroleptics found different strains to show significant differences in animal behavioral and biochemical models of TD, suggesting the importance of a genetic component. Repetitive jaw movement
247 Genetics of drug-induced tardive dyskinesia
following withdrawal of chronic neuroleptic treatment was found to be signifi- cantly different in different strains of selectively bred rats (Rosengarten et al., 1994). Similarly, the appearance of vacuous chewing movements following chronic halo- peridol administration was found to differ markedly in various rat strains (Tamminga et al., 1990). One of the seminal theories regarding the pathophysio- logy of TD proposes underlying dopamine receptor supersensitivity in the basal ganglia (Klawans et al., 1977). The mean rise in caudate dopamine receptor popu- lations in 10 different mouse strains chronically exposed to haloperidol was found to be significantly different, again pointing to the importance of genetic factors mediating this pharmacodynamic property in mice (Belmaker et al., 1981).
Genetic strategies
General considerations
While the level of exposure to the offending agent (the cumulative dosage and duration of chronic neuroleptic exposure) appears to bear some correlation with the incidence of these idiosyncratic reactions, as yet undefined, genetically based patient factors are likely to play an important role. It is likely that heritable individ- ual variability in factors affecting drug levels, as well as those regulating endog- enous dopamine levels and compensatory responses to dopaminergic antagonism, may account for a major portion of the variance in the incidence of TD.
Association approach for locating actual genes involved
The fact that prolonged exposure to D2 receptor antagonist drugs is necessary to unravel the clinical phenotype limits the usefulness of classical family studies for determining the degree of heritability and mode of inheritance of drug-induced TD. Similarly, the use of linkage-based studies for locating involved genes is limited by the difficulty in ascertaining family members sharing a similar exposure to neuroleptics.
A candidate gene association design is a useful alternative. Genes coding for known drug targets can be directly explored for involvement in TD susceptibility. Allelic association is a widely used strategy for the detection of quantitative trait loci (QTL), as it provides the statistical power required for detection of genes accounting for a relatively small percentage of the variance (Owen and McGuffin, 1993). Association studies are most meaningful when they employ candidate genes that make a priori sense based on current knowledge of brain chemistry. The majority of polymorphic regions and mutations described in genes coding for key proteins in neurotransmitter systems are silent (nonfunctional) intronic sites or rare mutations. The focus should be on polymorphic regions exhibiting prevalent allelic variants and possessing functional significance. Common functional variability resulting from polymorphic sites in genes encoding key proteins in

248 R. H. Segman and B. Lerer
relevant candidates can be directly explored for increased prevalence among patients with TD compared with schizophrenic patients not expressing TD despite a similar neuroleptic exposure.
Candidate genes for drug-induced tardive dyskinesia
Examples of such candidates include relevant neurotransmitter systems, such as dopamine, serotonin (5-HT), and -aminobutyric acid (GABA), and enzymes involved in drug metabolism, which may both influence the level of exposure to drugs and neutralize neurotoxic metabolites (see below). Interaction between allelic variation in such small-effect genetic loci may increase or decrease the risk for TD in a cumulative or epistatic manner. Such risk-modifying loci may act through several putative mechanisms, as detailed in the following examples.
Dopaminergic enzymes and receptors
Dopaminergic enzymes and receptors may be implicated in several ways. Genetic polymorphisms in key synthetic and metabolic enzymes responsible, respectively, for the production and degradation of dopamine may influence individual vari- ability in endogenous dopamine levels, as well as the capacity for compensatory adaptation of dopamine levels in response to chronic dopamine receptor antag- onist administration. Genetic variability in dopamine receptors may predict func- tional differences in response patterns to neuroleptic agents and may be relevant to the expression of idiosyncratic reactions.
Other neurotransmitters
Other neurotransmitter systems may be implicated as risk-modifying factors. The reported lack of TD with clozapine treatment has been attributed, among other theories, to its potent serotonergic action as a 5-HT2A/C antagonist at A-9 neurons. Genetic variation in the functional status of these receptors may result in individ- ual differences in their response to endogenous serotonergic tone.
Metabolizing enzymes
Metabolizing enzymes may act through modifying the risk for neural damage fol- lowing exposure to environmental agents such as toxins and drugs. In the case of drug-induced TD, polymorphic forms of the cytochrome P-450/debrisoquine hydroxylation system, resulting in individual variability in the metabolic clearance of neuroleptic drugs, may be directly implicated.
Analytic strategies and phenotype definition
Association of polymorphic candidate genes with risk for TD can be pursued by comparing unrelated ethnically matched patients exposed to neuroleptic
249 Genetics of drug-induced tardive dyskinesia
drugs, some of whom develop TD and some of whom remain unaffected, despite similar exposure. Two alternative analytic strategies can be employed in the context of such a model. The first strategy assigns a dichotomous threshold divi- sion to patients as affected or nonaffected and examines the relative prevalence of alleles and genotypes in the two groups. The other strategy investigates contin- uous measures of dyskinesia in all drug-exposed patients grouped according to allelic status. As pointed out by Plomin et al. (1991), phenotypic associations with QTL can be utilized to explore the relationship between continuous variation in the clinical phenotype and the putative pathogenic genotype. This approach does not force reliance solely on a single, dichotomous diagnostic category but util- izes all available information, which may correlate more readily with candidate QTL.
Diagnostic assessment for genetic studies
A major hindrance to the application of a case-control, association design to the study of drug-induced TD is the verification of the clinical status of both affected and control subjects. Individuals harboring a genetic make-up that may predispose to TD may be included in the control group because they have not been exposed to a dosage and duration of neuroleptic treatment sufficient for phenotypic expres- sion. Similarly, individuals suffering from milder manifestations of TD may be missed, owing to symptom masking by current neuroleptic administration. A case- control design must, therefore, employ well-evaluated patients demonstrating established TD, compared with ethnically matched patients who do not exhibit TD despite a similar heavy neuroleptic exposure in terms of both dosage and duration. Importantly, drug exposure must be comparable in terms of D2 receptor affinity for different neuroleptic drugs. It is customary to compare dosages against a conven- tional reference drug in terms of chlorpromazine equivalent units. In order to avoid common false-positive diagnoses of TD, assessment must take place following a prolonged period of stability in both antipsychotic and anticholinergic drug dose to avoid confounding transient withdrawal dyskinesia. In the elderly, pseudodys- kinetic manifestations of edentulousness must be ruled out.
Candidate gene studies of drug-induced tardive dyskinesia
Starting with the study of Arthur et al. (1995), a number of groups chose to explore specific genetic contributions to TD directly. Given the above considerations, these studies have employed a candidate genetic association approach, bypassing the difficult to apply steps of establishing heritability through family studies. Following is a review of the current state of data regarding principal genes investi- gated to date.

250 R. H. Segman and B. Lerer
Dopaminergic genes
Dopamine D3 receptor gene
The gene for the dopamine D3 receptor (DRD3) has recently attracted considerable interest as a site of action of antipsychotic drugs and for its potential role in the pathogenesis of schizophrenia (Schwartz et al., 2000). The human DRD3 gene has been localized to chromosome 3q13.3 by in situ hybridization. The D3 receptor readily binds classical and also atypical antipsychotic drugs but differs from other dopamine receptor subtypes in that it is primarily localized to limbic regions, which may be particularly important in the regulation of emotions and in the pathogen- esis of schizophrenia (Sokoloff et al., 1990; Zahm and Brog, 1992). DRD3 contains a polymorphic site in the first exon that gives rise to a serine to glycine substitution in the N-terminal extracellular domain of the receptor protein (Lannfelt et al., 1993). The in vivo functional significance of the polymorphic Ser9Gly site is unknown. DRD3 receptor-binding analysis of Chinese hamster ovary (CHO) cells infected with Semliki Forest virus to express the wild-type complementary DNA (cDNA), a recombinant DNA producing Ser9Gly, or both showed similar phar- macological properties for several D3 receptor ligands. However, the DRD3gly-gly homozygote showed a significantly higher binding affinity for dopamine whereas heterozygotes (i.e., doubly infected cells) were not significantly different from the wild type. In addition, both DRD3 gly-gly homozygotes and DRD3 ser-gly hetero- zygotes showed significantly higher binding affinity for the selective D3 ligand GR99841 compared with the wild-type receptor (Lundstrom and Turpin, 1996). While these results do not allow a straightforward extrapolation with regard to the biological significance of heterozygote versus homozygote status, they support either examining DRD3gly-gly homozygotes against all other genotypes or group- ing DRD3gly-gly homozygotes and DRD3ser-gly heterozygotes against wild-type homozygotes. Such extrapolations should be made with caution, however, as these results were obtained in an in vitro recombinant system not necessarily reflecting in vivo receptor status in the brain. A clearer understanding of the functional sig- nificance of the Ser9Gly site must await more detailed studies employing in vivo radioligand binding, as well as postmortem brain and in vitro receptor studies.
Steen et al. (1997) first reported association of DRD3 allele 2 (DRD3gly) and TD in schizophrenic patients and a significant excess of alleles 2/2 (DRD3gly-gly) homozygotes. This finding was subsequently supported by the observations of Segman et al. (1999), Basile et al. (1999), and Lovlie et al. (2000) but not by Rietschel et al. (2000). In line with a small-effect cumulative polygenetic model described above, a significant small increase in the Abnormal Involuntary Movement Scale (AIMS) total and subscores can be seen among schizophrenic patients carrying the DRD3gly allele as opposed to those homozygotic for DRD3ser (Fig. 11.1).
251 Genetics of drug-induced tardive dyskinesia
7 6 5 4 3 2 1 0
Fig. 11.1.
Total
Orofacial
Distal
Trunk
Incapacitation
                                   
Abnormal Involuntary Movement Scale (AIMS) total and subscale scores in schizophrenia patients grouped according to the structure of the polymorphic site in the dopamine D3 receptor: DRD3ser–ser ( )or DRD3ser–gly/ser–gly( ). A small but significant (p 0.05) increase in AIMS total score and in trunk and incapacitation subscale scores can be seen among schizophrenia patients carrying the DRD3-gly allele compared with the DRD3-ser homozygotes among schizophrenia patients. (Data from Segman et al., 1999.)
The findings of Steen et al. (1997) suggest that previous reports of an association of DRD3 with schizophrenia may have been confounded by variable rates of TD- prone subjects in the samples examined. This explanation would be particularly relevant to studies that found an association between schizophrenia and allele 2 of DRD3 (DRD3gly) (Kennedy et al., 1995; Ebstein et al., 1997). An association between neurocognitive deficits and TD has been described, and it has also been suggested that patients with more pronounced negative symptoms may be more vulnerable to TD, and that these features may represent associated aspects of a par- ticular schizophrenic subtype (Kane, 1995). Whereas our results (Segman et al., 1999) do not show a direct association of DRD3 genotype with Positive and Negative Syndrome Scale (PANSS) score, AIMS and PANSS scores were signifi- cantly correlated only in patients with the DRD3ser-gly or DRD3gly (alleles 1/2 and alleles 2/2, respectively) genotypes. The correlation was not limited to negative or cognitive symptoms but encompassed positive features as well. Among patients with the DRD3ser-ser (alleles 1/1) genotype, AIMS total was not significantly cor- related with PANSS total nor with any of the PANSS subscales. Although prelimi- nary, these results may be interpreted to support a genetic contribution of DRD3 genotype to a subtype of schizophrenia characterized by greater severity across a range of symptoms as well as predisposition to TD. Previous inconclusive results of attempts to associate DRD3 genotype and schizophrenia may have been con- founded by differing rates of patients prone to TD in the samples, as proposed by Steen et al. (1997). A useful approach to address this possibility further would be
AIMS score
252 R. H. Segman and B. Lerer
to dissect schizophrenic patient samples clinically according to DRD3 genotype, with our findings predicting higher TD and PANSS scores in patients carrying the DRD3gly (2) allele.
Dopamine D2 receptor gene
The gene for the dopamine D2 receptor (DRD2) is the main target for classical anti- psychotic drugs and a natural candidate for investigation of its role as a putative risk modifier of drug-induced TD. The only published study to-date, examined a Taq-I restriction polymorphism at the DRD2 locus located on 11q23. A1 allelic status at this polymorphic site has been linked with lower DRD2 binding potential in vivo in positron emission tomographic (Pohjalainen et al., 1998) and single photon emission computed tomographic (Thompson et al., 1997) studies, although not by all studies (Laruelle et al., 1998). If the A1 allele is associated with reduced DRD2 binding, it may be through linkage disequilibrium with another functional polymorphism in the DRD2 gene. Chen et al. (1997) examined the rel- ative frequency of Taq-I A alleles among 93 patients with TD compared with 84 controls without TD. They found increased frequency of A2 homozygotes and car- riers among TD patients, although A2 allelic frequency was significantly increased only for female patients with TD in the sample divided according to gender. Independent replication of these preliminary findings as well as investigation of other, functional polymorphic sites in DRD2 are warranted.
Serotonergic genes
Serotonin 2A receptor gene
The 5-HT2A receptor is a site of action for atypical antipsychotic agents and has been implicated with their added efficacy as well as their reduced extrapyramidal side effect profile (Meltzer, 1999). Atypical antipsychotic agents such as clozapine (Tamminga et al., 1994; Casey, 1998) and olanzapine (Glazer, 2000) have been shown to induce much lower rates of TD compared with conventional antipsy- chotic drugs, and this has been attributed to a protective effect of their high 5-HT2 receptor-blocking activity relative to D2 receptor blockade (Glazer, 2000). The 5- HT2A receptor is distributed in striatal brain areas that modulate motor activity (Marazziti et al., 1997). This receptor has been shown to interact with dopaminer- gic neurotransmission in brain regions relevant to antipsychotic drug action (Schmidt et al., 1995). Rodent studies suggest that high occupation of 5-HT2A receptors with lower D2 receptor occupancy might be involved in the absence of upregulation of D2 receptors after treatment with atypical antipsychotic drugs (Kusumi et al., 2000). This may be relevant to the documented lower rates of TD with atypical agents, given the classical hypothesis implicating D2 receptor
253 Genetics of drug-induced tardive dyskinesia
supersensitivity with pathogenesis of drug-induced TD (Rubovits and Klawans, 1972). Furthermore, pretreatment with atypical antipsychotic agents has been shown to reduce repetitive jaw movements in a rat model for TD (Rosengarten et al., 1999) and 5-HT2A receptor antagonists have been shown to attenuate apomor- phine-induced stereotypic oral movements in rats (Barwick et al., 2000). Finally, long-term elevations in 5-HT2A receptor binding and mRNA expression in neostri- atal regions have been documented in response to ontogenetic loss of dopamine neurons following 6-hydroxydopamine administration (Kostrzewa et al., 1998), suggesting that such changes may similarly be relevant to pathogenetic processes related to neuroleptic-induced TD. Taken together, the above clinical and basic data, suggest a protective role for 5-HT2A receptor blockade in TD. We hypothesized that a genetically variant 5-HT2A receptor may alter risk for TD in schizophrenia patients following chronic exposure to antipsychotic drugs.
The gene for 5-HT2A receptor (5-HTR2A) is located on chromosome 13q14–21 (Hsieh et al., 1990). A number of polymorphic sites have been described in 5- HTR2A (Waren et al., 1993; Erdmann et al., 1996; Ozaki et al., 1996; Collier et al., 1997; Ohara et al., 1997). Of these, three sites, including the 102T/C, 1438A/G, and His452Tyr, are common variations in the population. A number of studies have shown a small but significant contribution of the 102T/C and, more recently, the A-1438G polymorphisms to risk for schizophrenia (Tay et al., 1997; Williams et al., 1997; Spurlock et al., 1998) but several other studies could not replicate these findings (Hawi et al., 1997; Shinkai et al., 1998; Kouzmenko et al., 1999; Ohara et al., 1999). A number of studies, including a meta-analysis, have found association of the His452Tyr polymorphism with response to clozapine in schizophrenic patients (Arranz et al., 1998; Masellis et al., 1998) whereas both positive (Arranz et al., 1998) and negative (Malhotra et al., 1996; Masellis et al., 1998) results have been found for the 102T/C polymorphism. Given their respective localization, the A- 1438G polymorphism may be expected to alter the transcriptional activity of the receptor whereas the silent T102C site is likely to reflect the impact of the promoter site through linkage disequilibrium; however, A-1438G was not found to affect pro- moter activity in HeLa cells (Spurlock et al., 1998) and correlations with receptor density are inconclusive (Kouzmenko et al., 1999; Turecki et al., 1999). Further studies are required to establish the functional impact of these polymorphic sites.
Given the above, we examined the possibility that these common 5-HTR2A poly- morphic sites influence the tendency to express TD following prolonged antipsy- chotic drug exposure in schizophrenic patients (Segman et al., 2001). Two polymorphisms were examined in the coding region of 5–HTR2A, the conservative T102C and that producing His452Tyr, and the A-1438G in the promoter. The T102C and the A-1438G polymorphisms were in complete linkage disequilibrium. There was a significant excess of 102C and 1438G alleles (62.7%) in the patients
254 R. H. Segman and B. Lerer
with TD compared with the patients without TD (41.1%) and matched normal control subjects (45.9%; 2 12.8; df 2; p 0.002; odds ratio (OR) 2.41; 95% con- fidence interval (CI) 1.43–3.99) and of 102CC and 1438GG genotypes (schizo- phrenia with TD 42.4%, schizophrenia without TD 16.1%; controls 20.8%; 2 13.3; df 4, p 0.01). The 102CC and the 1438GG genotypes were associated with significantly higher AIMS trunk dyskinesia scores (F 3.9; df 2, 116; p 0.02) and more incapacitation (F 5.0; df 2, 115; p 0.006). In contradistinction, the His452Tyr polymorphism showed no association at all with TD. These find- ings suggest that 5-HTR2A is significantly associated with susceptibility to TD in patients with chronic schizophrenia (Segman et al., 2001). Previously reported association of the T102C and A-1438G polymorphisms in 5-HTR2A with schizo- phrenia may reflect association of a subgroup of patients with a susceptibility to abnormal involuntary movements related to antipsychotic drug exposure.
It is noteworthy that Basile et al. (2001) did not find an association of 5-HTR2A with TD. Their clinical sample of schizophrenic patients was approximately 20 years younger than that of Segman et al. (2001). Based on a further analysis of their data, Segman and Lerer (2001) demonstrated that the association of 5-HTR2A with abnormal involuntary movements is age related. Older patients ( 49 years) showed a significant association while younger patients did not. This effect could explain the discrepant findings of Segman et al. (2001) and Basile et al. (2001) in regard to the association of TD with 5-HTR2A.
Serotonin 2C receptor gene
The 5-HT2C receptor is distributed in nigrostriatal brain areas that modulate motor activity (Abramowski et al., 1995). Functional studies on serotonin-mediated phosphoinositide hydrolysis in the caudate nucleus in rats demonstrated a promi- nent role for this receptor subtype, different from that of 5-HT2A receptor- mediated activity (Wolf and Schultz, 1997). Additionally, 5-HT2C agonists were shown to decrease locomotor activity, whereas 5-HT2A agonists increased such activity (Wolf and Schultz, 1997). The 5-HT2C receptor has been shown to interact with dopaminergic neurotransmission in basal ganglia in rodents; 5-HT2C recep- tors are upregulated following 6-hydroxydopamine lesioning in rats, and 5-HT2C antagonist administration reduces dyskinetic movements in these rats (Fox et al., 1988). Interestingly, excessive stimulation of 5-HT2C receptors in the caudate and subthalamic nuclei elicits orofacial dyskinesias resembling those seen in patients following chronic neuroleptic exposure (Erble-Wang et al., 1996). Both systemic and local subthalamic infusion of the 5-HT2C receptor agonist, m– chlorophenylpiperazine (mCPP), has been shown to elicit orofacial dyskinesias, and this effect was blocked by 5-HT2C antagonism (Erble-Wang et al., 1996). Based on the above data, it has been suggested that the 5-HT2C receptor may be
255 Genetics of drug-induced tardive dyskinesia
responsible for some of the dyskinetic effects of antipsychotic drugs (Wolf and Schultz, 1997).
The human 5-HT2C receptor gene (HTR2C) has been localized to chromosome Xq24. It contains a common C–G transversion at nucleotide 68 of the coding sequence that gives rise to a cysteine to serine substitution in the N-terminal extra- cellular domain of the receptor protein (Lappalainen et al., 1995). The pharmaco- logical profile and in vivo significance of the mutant receptor protein is unknown; however, preliminary reports point to an altered binding affinity to mCPP in vitro (Lappalainen et al., 1995). To-date only two studies have examined some functional aspects of the 5-HT2C serine/cysteine polymorphic site in humans, both reporting a functional significance. Higher concentrations of cerebrospinal fluid 3-methoxy- 4-hydroxyphenylglycol were reported in Finnish, alcoholic violent offenders and population control males who were hemizygotes for the serine genotype (Lappalainen et al., 1999). This finding is of interest to the extent that it can be gen- eralized to other populations. Also, unaltered endocrine/thermic responses to mCPP but a reduced hypophagic response were observed in a small group of normal female subjects comprising 5-HT2C serine/cysteine heterozygotes and serine homozygotes compared with wild-type cysteine homozygote female subjects (Quested et al., 1999).
Given the above, we sought to examine the possibility that the HTR2C polymor- phic site influences the tendency to manifest TD following prolonged antipsychotic drug exposure in schizophrenia patients (Segman et al., 2000). We found a signifi- cant excess of 5-HTR2Cser alleles in schizophrenic patients with TD (27.2%) com- pared with patients without TD (14.6%) and normal controls (14.2%; 2 6.4; df 2; p 0.03), which was owing to the female patients ( 2 8.6; df 2; p 0.01). Among the female TD patients, there was an excess of Cys-Ser and Ser-Ser geno- types ( 2 11.9; df 4; p 0.02). Analysis of covariance controlling for age at first antipsychotic treatment revealed a significant effect of 5-HTR2C genotype on oro- facial dyskinesia scores (F 3.47; df 2; p 0.03). In a stepwise multiple regression analysis, 5-HTR2C and DRD3 genotype (5-HTR2Cser and DRD3gly allele car- riage), respectively, contributed 4.2% and 4.7% to the variance in orofacial dyski- nesia scores. These findings support a small but significant contribution of the HTR2C and DRD3 to susceptibility to TD, which is additive in nature.
Serotonin transporter gene
The serotonin transporter (5-HTT) is the rate-limiting factor in terminating the synaptic action of 5-HT, and is the molecular target for selective serotonin reup- take inhibitors (SSRIs). SSRIs have been reported to cause TD (Dubovsky and Thomas, 1996; Leo, 1996). Nigrostriatal dopaminergic neurons are tonically inhib- ited by dorsal raphe serotonergic projections (Nedergaard et al., 1988). SSRIs have
256 R. H. Segman and B. Lerer
been suggested to act via these connections in reducing the release of synaptic dop- amine in the nigrostriatal system. Chronic decreases in dopaminergic transmission may, in turn, facilitate hypersensitivity of the postsynaptic dopamine receptors, thereby giving rise to TD (Chong et al., 2000). An insertion/deletion polymor- phism has been reported in the promoter region of the 5-HTT gene (5-HTT linked polymorphic region: 5-HTTLPR) (Heils et al., 1996) localized to 17q11.1–12. The short variant of the polymorphism has been shown to reduce the in vitro transcrip- tional efficiency of the 5-HTT gene promoter, resulting in decreased 5-HTT expres- sion and 5-HT uptake in lymphoblasts (Heils et al., 1996). Chong et al. (2000) examined the 5-HTTLPR site in 188 schizophrenic patients and found no correla- tion of allelic status with AIMS score or with TD diagnosis. We have observed similar results (R. H. Segman et al., unpublished data).
Metabolic enzymes
Cytochrome P-450 system
Cytochrome P-450 (CYP) enzymes may be related to the pathogenesis of TD either through a pharmacokinetic effect of altering neuroleptic drug levels or through neutralization of putative secondary neurotoxic factors.
Cytochrome P-450 2D6
The metabolism of many drugs including several common neuroleptic drugs is influenced by polymorphisms of CYP2D6 located on 22q13.1, which encodes the CYP2D6 enzyme debrisoquine/sparteine hydroxylase. Approximately 5–10% of European Caucasians lack the enzyme CYP2D6 and are designated poor metabol- izers (PM). Given standard drug therapy, PM subjects are likely to achieve higher than average concentrations of neuroleptic drugs in plasma, with an increased risk of extrapyramidal side effects, possibly including TD. A number of studies investi- gated association of CYP2D6 alleles with TD. Arthur et al. (1995) first investigated CYP2D6 allele frequencies among 16 schizophrenic patients with TD and found only one PM subject, suggesting no increased rate of PM compared with normal population rates. Armstrong et al. (1997) examined 76 schizophrenic patients and reported a trend for increased movement disorder ratings and TD among the five PM subjects in the group. Andreasen et al. (1997) examined a group of 100 schizo- phrenia patients 10% of whom were PM. They reported a nonsignificant trend for increased rate of TD among PM subjects. Ohmori et al. (1999) found no correla- tion of CYP2D6 alleles with AIMS scores or TD diagnosis among 99 schizophrenic patients. Taken together, these studies do not support the likelihood of a major con- tribution of CYP2D6 alleles to risk for neuroleptic-induced TD. However, the small number of TD subjects in each study is insufficient to exclude a small effect of this locus, given the low population prevalence of the PM allele.
257 Genetics of drug-induced tardive dyskinesia
Cytochrome P-450 1A2
Although CYP1A2 has a lower affinity for typical antipsychotic agents than the high affinity CYP2D6, it is much more abundant in the liver and is further induced by smoking, making it a clinically important metabolic clearance pathway for typical antipsychotics. CYP1A2 activity in vivo shows an over 10-fold variability and is under significant genetic control (Kalow and Tang, 1991) . A C-to-A substitution has been described in intron 1 of CYP1A2 located on 15q22–qter (Sachse et al., 1999). Among healthy Caucasians, smokers with the C/C genotype had, on average, 40% lower CYP1A2 activity as measured by a caffeine metabolic ratio ([1, 7- dimethylxanthine 1, 7-dimethyluric acid]/[caffeine]), in comparison with those homozygous for the variant A. In contrast, the C-to-A genetic polymorphism in CYP1A2 did not importantly contribute to variability in CYP1A2 activity in non- smokers (Sachse et al., 1999). Basile et al. (2000) investigated a sample of 85 schizo- phrenic patients assessed for TD severity using AIMS for association with the C-to-A polymorphism in CYP1A2. The mean AIMS score in patients with the C/C genotype (associated with reduced CYP1A2 inducibility) was 2.7- and 3.4-fold greater than in those with the A/C or A/A genotype, respectively (F 7.4; df 2, p 0.0007). A subanalysis in the 44 known smokers in the sample revealed a more pro- nounced effect. The mean AIMS score in smokers was 5.4– and 4.7-fold greater in C/C homozygotes when compared with heterozygotes and A/A homozygotes, respectively (F 3.7; df 2, 41; p 0.008). These data suggest that the C-to-A genetic polymorphism in CYP1A2 may serve as a genetic risk factor for the devel- opment of TD in patients with schizophrenia. Independent replication of these findings is warranted.
Manganese superoxide dismutase
Impaired free radical detoxification has been previously implicated in schizophre- nia (Reddy et al., 1991). In addition, data supporting increased free radical burden have been reported following neuroleptic treatment (Pall et al., 1987), and increased free radical activity in patients with TD has also been reported (Lohr et al., 1990). Based on the above, Hori et al. (2000) investigated a role in TD for man- ganese superoxide dismutase (MnSOD), the gene for which is located on 6q25.3. They examined a biallelic polymorphism (Ala9Val) in the mitochondrial targeting sequence of human MnSOD, which was previously reported to associate with risk for Parkinson’s disease (Shimoda-Matsubayashi et al., 1996). The study included 192 schizophrenic patients, 39 with TD and 153 without TD, and 141 normal con- trols. No significant differences were found in the allelic or genotypic distribution between schizophrenics and controls. A significantly reduced rate of the rare 9Ala carriers (homozygote and heterozygotes) was found among patients with TD com- pared with schizophrenia patients without TD (p 0.03), and decreased 9Ala allele was found among patients with TD (p 0.02; OR 0.29; 95% CI 0.10–0.83). They
258 R. H. Segman and B. Lerer
conclude that the allele giving 9Ala may correlate with higher MnSOD activity and may play a role in protecting against susceptibility to TD in schizophrenics.
Additive small gene effects
From the data reviewed above, a polygenic contribution of several small-effect genetic loci is suggested. In such case, the additive contribution of multiple such loci in an individual will dictate the consequent response to the external event (e.g., long-term neuroleptic exposure, in terms of the propensity to express TD. In theory, such contributing loci may have an additive contribution operating to increase the risk in a simple additive or in a synergistic pattern. Conversely, some loci may instead operate to lower the risk, if they play a protective effect.
The first evidence for such additive effect of different loci in TD has been reported for the 5-HTR2C and DRD3 loci (Segman et al., 2000). In a stepwise multiple regression analysis, 5-HTR2C and DRD3 genotypes (5-HTR2Cser and DRD3gly allele carriage, respectively) contributed independently 4.2% and 4.7% to the variance in orofacial dyskinesia scores. In Fig. 11.2, the additive significance of allele carriage in the two contributing loci can be seen.
Such an additive effect is well based in current neurochemical understanding of dopamine–5-HT interactions. As noted above, the 5-HT2C receptor has been shown to interact with dopaminergic neurotransmission in basal ganglia. The 5- HT2C receptors are upregulated following 6-hydroxydopamine lesioning in rats and 5-HT2C antagonist administration reduces dyskinetic movements in these rats (Fox et al., 1988). It may be hypothesized that 5-HT2C receptor upregulation may also occur in response to neurotoxic damage, which may be mediated by the accumu- lation of neurotoxins such as 6-hydroxydopamine following prolonged exposure to dopamine receptor antagonist drugs. The additive effect of genetically determined endogenous differences in 5-HT2C and D3 receptor reactivity may, therefore, contribute in a cumulative manner to clinical differences in resultant expression of TD.
Conclusions
Genetic association studies of TD have proven to be a fruitful approach for locat- ing predisposing molecular mechanisms, despite the lack of family-based evidence for heritability and mode of inheritance (Table 11.1). This is largely because targets of drug action pose direct hypotheses implicating candidate genes that may be directly tested for involvement in risk determination. The DRD3 gene is by far the most robust finding to date, with four of five independent studies confirming asso- ciation. Other promising findings, including the reported associations with
 
259 Genetics of drug-induced tardive dyskinesia
6
5 4* 3
2
1
0
*#
          
Fig. 11.2.
Adjusted mean Abnormal Involuntary Movement Scale-Oro-Facial-Digital (AIMS-OFD) scores (derived from ANCOVA) of patients with chronic schizophrenia carrying varying alleles for the serotonin (5-HT) 2C receptor (5-HT2C) and the dopamine D3 receptor (DRD3). 1, both wild type (WT); 2, WT plus mutant; 3, mutant WT; 4, mutant plus mutant. By ANCOVA with age at first antipsychotic treatment as covariate: main effect 5- HT2C, F 5.38 (df 1: 110), p 0.02; main effect DRD3, F 7.28 (df 1:110), p 0.008; interaction, F 10.06 (df 1:110), p 0.81. * p 0.05 compared with alleles 1; p 0.05 compared with alleles 2 (univariate planned comparisons). An additive contribution can be seen where pathogenic allele carriers for both loci (i.e. 5–HT2Cser and DRD3gly) show highest AIMS-OFD scores. Wild-type carriers for both loci (i.e., 5-HT2Ccys and DRD3ser) show the lowest scores, and carriers of a pathogenic allele in just one of the loci show an intermediate AIMS score. (With permission, from Segman et al., 2000.)
5-HTR2A and 5-HTR2C and MnSOD genetic polymorphic sites, require indepen- dent replications. Such findings open a window for understanding the molecular basis of drug-induced TD, and possibly other extrapyramidal disorders. In addi- tion they may aid in the prediction of susceptibility to neuroleptic-induced TD before drug exposure, as well as help to direct the development of safer and more specific antipsychotic drugs. Genetic susceptibility to TD may mark a more specific phenotype of schizophrenia, with selective prognostic course and drug-response profile, and may prove instrumental for improving phenotype definition in genetic studies of schizophrenia.
Future research of the genetic causes of TD may employ a number of methodo- logical strategies. The currently utilized case-control candidate genetic association design may be served by larger samples of phenotypically well-defined TD patients and controls sharing similar drug exposure. A superior approach entails the use of transmission disequilibrium analysis, employing parental genotypes of TD sub- jects, as this methodology is immune both to the problem of ethnic stratification
1: 5-HT2Ccys- DRD3ser
2: 5-HTCcys- DRD3gly
3: 5-HTCser- DRD3ser
4: 5-HTCser- DRD3gly
AIMS-OFD score
260 R. H. Segman and B. Lerer
Table 11.1. Genetic association findings in tardive dyskinesia
Chromosomal
Gene polymorphism location Association Replication Nonreplication
DRD3 Ser9Gly
DRD2 Taq-I
5-HTR2A: T102C/A-, 1438G, his452tyr
5-HTR2Ccys-ser 5-HTT promoter I/D CYP2D6 (PM allele)
CYP1A2 intron 1 C/A MnSOD Ala9Val
3q13.3 Positive: Steen et al., 1997
11q23 Positive in female patients with TD:
Chen et al., 1997
13q14–21 Segman et al., 2000; positive for
T102C/A and 1438G; negative for His452Tyr
Xq24 Positive: Segman et al., 2000
17q11.1–12 Negative: Chong et al., 2000
22q13.1 Trend; Arthur et al., 1995; Andreasen
et al., 1997
15q22-qter Positive: Basile et al., 2000
6q25.3 Hori et al., 2000
Segman et al., 1999; Basile et al., 1999; Lovlie et al., 2000
(trend)
NA NA
NA NA NA
NA
NA
Rietschel et al., (2000)
NA
Basile et al., 2001
(Note: discrepancy may be explained by age effect – Segman and Lerer, 2002)
NA NA
Armstrong et al., 1997; Ohmori et al., 1999
NA
NA
Notes:
NA, not available; TD, tardive dyskinesia; PM, poor metabolizer.
among patients and controls and of false-negative diagnoses of TD status among medicated control schizophrenic subjects, who may either develop TD upon a more prolonged exposure or have latent TD masked by the concurrent use of neuroleptics at the time of assessment.
The alternative approach of exploration of unknown genes through genome scan can be implemented through one of two routes. Routine inclusion of antipsy- chotic drug exposure and response quantification and dyskinesia assessments in large-scale linkage studies for schizophrenia would allow utilization of pharmacog- enetic information for sharpening of the biological schizophrenia phenotype, pos- sibly improving its diagnostic specificity as well as allowing a linkage-based search
261 Genetics of drug-induced tardive dyskinesia
for genetic loci predisposing for TD. Alternatively, single nucleotide polymorphism (SNP) genome mapping may be employed for larger groups of unrelated with TD patients and medicated schizophrenia controls. Finally, comparative global expres- sion patterns of mRNA transcripts may be utilized for matching relevant brain areas from autopsy brain specimens of patients with TD and medicated schizo- phrenia controls.
TD constitutes a sharply defined pharmacogenetic phenotype, which may con- tinue to serve as a model for implementing accumulating genetic molecular tools for locating genotype–phenotype correlations in complex phenotypes.
Acknowledgements
This work was supported in part by grants from the Yisumi Fund of Hadassit Research and Development Corporation and the National Institute for Psychobiology in Israel to RS. The Biological Psychiatry Laboratory, Hadassah Hebrew University Medical Center, is supported by the Harry Stern Family Foundation.
REFERENCES
Abramowski D, Rigo M, Duc D, Hoyer D, Staufenbiel M (1995). Localization of the 5-hydroxytryptamine 2C receptor protein in human and rat brain using specific antisera. Neuropharmacology 34, 1635–1645.
Andreasen OA, MacEwan T, Gulbrandsen AK, McCreadie RG, Steen VM (1997). Non-functional CYP2D6 alleles and risk for neuroleptic-induced movement disorders in schizophrenic patients. Psychopharmacology (Berl) 131, 174–179.
Armstrong M, Daly AK, Blennerhassett R, Ferrier N, Idle JR (1997). Antipsychotic drug-induced movement disorders in schizophrenics in relation to CYP2D6 genotype. Br J Psychiatry 170, 23–26.
Arthur H, Dahl ML, Siwers B, Sjoqvist F (1995). Polymorphic drug metabolism in schizophrenic patients with tardive dyskinesia. J Clin Psychopharmacol 15, 211–216.
Arranz MJ, Munro J, Sham P (1998). Meta-analysis of studies on genetic variation in 5-HT2A receptors and clozapine response. Schizophr Res 32, 93–99.
Barwick VS, Jones DH, Richter JT, Hicks PB, Young KA (2000). Subthalamic nucleus micro- injections of 5-HT2 receptor antagonists suppress stereotypy in rats. Neuroreport 11, 267–270. Basile VS, Masellis M, Badri F et al. (1999). Association of the MscI polymorphism of the dopa- mine D3 receptor gene with tardive dyskinesia in schizophrenia. Neuropsychopharmacology 21,
17–27.
Basile VS, Ozdemir V, Masellis M et al. (2000). A functional polymorphism of the cytochrome
P-450 1A2 (CYP1A2) gene: association with tardive dyskinesia in schizophrenia. Mol Psychiatry 4, 410–417.
 
262 R. H. Segman and B. Lerer
Basile VS, Ozdemir V, Masellis M et al. (2001). Lack of association between serotonin-2A recep- tor gene (HTR2A) polymorphisms and tardive dyskinesia in schizophrenia. Mol Psychiatry 6, 230–234.
Belmaker RH, Bannet J, Brecher-Fried E, Yanai J, Ebstein RP (1981). The effect of haloperidol feeding on dopamine receptor number in ten mouse strains. Clin Genet 19, 353–356.
Brown KW, White T (1991). HLA and TD. Br J Psychiatry 158, 270–272.
Burke RE, Kangh UJ, Jankovic J (1989). Tardive akathisia: an analysis of clinical features and
response to open therapeutic trials. Mov Disord 4, 157–175.
Canoso RT, Romero JA, Yunis EJ (1986). Immunogenetic markers in chlorpromazine induced
TD. J Neuroimmunol 12, 247–252.
Casey DE (1998). Effects of clozapine therapy in schizophrenic individuals at risk for tardive dys-
kinesia. J Clin Psychiatry 59 (Suppl 3), 31–37.
Chen CH, Wei FU, Koong FJ, Hsiao KJ (1997). Association of Taq-I A polymorphism of
dopamine D2 receptor gene and tardive dyskinesia in schizophrenia. Biol Psychiatry 41,
827–829.
Chong SA, Tan EC, Tan CH, Mahendren R, Tay AHN, Chua HC (2000). Tardive dyskinesia is not
associated with the serotonin gene polymorphism (5-HTTLPR) in Chinese. Am J Med Genet
(Neuropsychiatr Genet) 96, 712–715.
Collier DA, Arranz MJ, Li T, Mupita D, Brown N, Treasure J (1997). Association between 5-HT2A
gene promoter polymorphism and anorexia nervosa. Lancet 350, 412.
Dubovsky SL, Thomas M (1996). Tardive dyskinesia associated with fluoxetine. Psychiatr Serv 47,
991–993.
Ebstein RP, Macciardi F, Blaine D, Serretti A, Gur E, Verga M (1997). Evidence for an association
between the dopamine D3 receptor gene (DRD3) and schizophrenia. Hum Hered 47, 6–16. Erble-Wang K, Lucki I, Chesselet MF (1996). A role for the subthalamic nucleus in the 5-HT2C
induced oral dyskinesia. Neuroscience 72, 117–128.
Erdmann J, Shimron-Abarbanell D, Rietschel M (1996). Systematic screening for mutations in
the human serotonin-2A (5-HT2A) receptor gene: identification of two naturally occurring
receptor variants and association analysis in schizophrenia. Hum Genet 97, 614–619.
Fox SH, Moser B, Brotchie M (1988). Behavioral effects of 5-HT2C receptor antagonism in the substantia nigra zona reticulata of the 6-hydroxydopamine-lesioned rat model of Parkinson’s
disease. Exp Neurol 151, 35–49.
Glazer WM (2000). Expected incidence of tardive dyskinesia associated with atypical antipsy-
chotics. J Clin Psychiatry 61, 21–26.
Gasser T (2000). Autosomal-dominantly inherited forms of Parkinson’s disease. J Neural Transm
Suppl 58, 31–40.
Hawi Z, Myakishev MV, Straub RE (1997). No association or linkage between the 5-
HT2A/T102C polymorphism and schizophrenia in Irish families. Am J Med Genet 74,
370–373.
Heils A, Teufel A, Petri S et al. (1996). Allelic variation of human serotonin transporter gene
expression. J Neurochem 66, 2621–2624.
Hori H, Ohmori O, Shinkai T (2000). Manganese superoxide dismutase dene polymorphism and
schizophrenia: relation to tardive dyskinesia. Neuropsychopharmacology 23, 170–177.
263 Genetics of drug-induced tardive dyskinesia
Hsieh CL, Bowcock AM, Farrer LA et al. (1990). The serotonin receptor subtype 2 locus HTR2 is on chromosome 13 near genes for esterase D and retinoblastoma and on mouse chromo- some 14. Somat Cell Genet 16, 567–574.
Kalow W, Tang BK (1991). Use of caffeine metabolite ratios to explore CYP1A2 and xanthine oxidase activities. Clin Pharmacol Ther, 50, 508–519.
Kane JM (1995). Tardive dyskinesia: epidemiological and clinical presentation. In Bloom FE, Kupfer DJ, eds. Psychopharmacology: The 4th Generation of Progress. Raven Press: New York, pp. 1485–1495.
Kane JM, Smith JM (1982).Tardive dyskinesia. Arch Gen Psychiatry 37, 473–481.
Kennedy JL, Billett E, Macciardi FM, Verga M, Parsons T, Meltzer HY (1995). Association study of the dopamine D3 receptor gene and schizophrenia. Am J Med Genet (Neuropsychiatr Genet)
60, 558–562.
Klawans HL, Hitri A, Nausieda PA, Wiener WL (1977). In Hanin IR, Usdin E, eds. Animal Models
of Dyskinesia in Psychiatry and Neurology, Oxford, Pergamon Press, pp. 351–363.
Koller WC (1988). Idiopathic oral facial dyskinesia. Adv Neurol 149, 225–237.
Kostrzewa RM, Reader TA, Descarries L (1998). Serotonin neural adaptations to ontogenetic loss
of dopamine neurons in rat brain. J Neurochem 70, 889–898.
Kouzmenko AP, Scaffidi A, Pereira AM, Hayes WL, Copolov DL, Dean B (1999). No correlation
between A( 1438)G polymorphism in 5-HT2A receptor gene promoter and the density of
frontal cortical 5-HT2A receptors in schizophrenia. Hum Hered 49, 103–105.
Kusumi I, Takahashi Y, Suzuki K, Kameda K, Koyama T (2000). Differential effects of subchronic treatments with atypical antipsychotic drugs on dopamine D2 and serotonin 5-HT2A receptors
in the rat brain. J Neur Transm 107, 295–302.
Lannfelt L, Sokoloff P, Martres MP et al. (1993). Amino acid substitution in the dopamine D3
receptor as a useful polymorphism for investigating psychiatric disorders. Psychiatr Genet 2,
249–256.
Lappalainen J, Zhang L, Dean M et al. (1995). Identification expression and pharmacology of a
Cys23–Ser23 sustitution in the human 5-HT2C receptor gene (HTR2C). Genomics 27, 274–279. Lappalainen J, Long JC, Virkkunen M, Ozaki N, Goldman D, Linnoila M (1999). HTR2C Cys23Ser polymorphism in relation to CSF monoamine metabolite concentrations and DSM-
III-R psychiatric diagnoses. Biol Psychiatry 46, 821–826.
Laruelle M, Gelernter J, Innis RB (1998). D2 receptors binding potential is not affected by Taq1
polymorphism at the D2 receptor gene. Mol Psychiatry 3, 261–265.
Leo RJ (1996). Movement disorders associated with the serotonin selective reuptake inhibitors. J
Clin Psychiatry 57, 449–454.
Lohr JB, Kuczenski R, Bracha HS, Moir M, Jeste DV (1990). Increased indices of free radical activ-
ity in the cerebrospinal fluid of patients with tardive dyskinesia. Biol Psychiatry 28, 535–539. Lovlie R, Daly AK, Blennerhassett R, Ferrier N, Steen VM (2000). Homozygosity for Gly-9 variant of the dopamine D3 receptor and risk for tardive dyskinesia in schizophrenic patients.
Int J Neuropsychopharmacol 3, 61–66.
Lundstrom K, Turpin MP (1996). Proposed schizophrenia related gene polymorphism: expres-
sion of the Ser9Gly mutant human dopamine D3 receptor with the Semliki Forest virus system. Biochem Biophys Res Commun 225, 1068–1072.
264 R. H. Segman and B. Lerer
Malhotra AK, Goldman D, Ozaki N, Breier A, Buchanan R, Pickar D (1996). Lack of association between polymorphisms in the 5-HT2A receptor gene and the antipsychotic response to cloz- apine. Am J Psychiatry 153, 1092–1094.
Marazziti D, Rossi A, Palego L (1997). [3H]Ketanserin binding in human brain postmortem. Neurochem Res 22, 753–757.
Masellis M, Basile V, Meltzer HY (1998). Serotonin subtype 2 receptor genes and clinical response to clozapine in schizophrenia patients. Neuropsychopharmacology 19, 123–132.
Meltzer HY (1999). The role of serotonin in antipsychotic drug action. Neuropsychopharmacology 21(Suppl), 106–115.
Meltzer WS, Newton JE, Steele RW (1990). HLA antigens in tardive dyskinesia. J Neuroimmunol 26, 179–181.
Muller DJ, Ahle G, Alfter D et al. (1998a). Familial occurrence of tardive dyskinesia. Am J Med Genet 81, 527.
Muller U, Steinberger D, Nemeth AH (1998b). Clinical and molecular genetics of primary dys- tonias. Neurogenetics 1, 165–177.
Nedergaard S, Bolam JP, Greenfield SA (1988). Facilitation of a dendritic calcium conductance by 5-hydroxytryptamine in the substantia nigra. Nature 333, 174–177.
O’Callaghan E, Larkin C, Kinsella A, Wadington JL (1990). Obstetric complications, the putative familial sporadic distinction and TD in schizophrenia. Br J Psychiatry 157, 578–584.
Ohara K, Ino A, Ishigaki T, Tani K, Tsukamoto T, Nakamura Y (1997). Analysis of the 5 -flanking promoter region of the 5-HT2A receptor gene in schizophrenia. Neuropsychopharmacology 17, 274–278.
Ohara K, Nagai M, Tani K, Tsukamoto T (1999). Schizophrenia and the serotonin-2A receptor promoter polymorphism. Psychiatr Res 85, 221–224.
Ohmori O, Kojima H, Shinkai T, Terao T, Suzuki T, Abe K (1999). Genetic association analysis between CYP2D6*2 allele and tardive dyskinesia in schizophrenic patients Psychiatr Res 87, 239–244.
Owen MJ, McGuffin P (1993). Association and linkage: complementary strategies for complex disorders. Am J Med Genet 30, 578–584.
Ozaki N, Rosenthal NE, Pesonen U (1996). Two naturally occurring amino acid substitutions of the 5-HT2A receptor: similar prevalence in patients with seasonal affective disorder and con- trols. Biol Psychiatry 40, 1267–1272.
Pall HS, Williams AC, Blake DR, Lunec J (1987). Evidence of enhanced lipid peroxidation in the cerebrospinal fluid of patients taking phenothiazines. Lancet ii, 596–599.
Plomin R, McClearn GE, Gora-Maslak G, Neiderhiser JM (1991). Use of recombinant inbred strains to detect quantitative trait loci associated with behavior. Behav Genet 21, 99–116.
Pohjalainen T, Rinne JO, Nagren K et al. (1998). The A1 allele of the human D2 dopamine recep- tor gene predicts low D2 receptor availability in healthy volunteers. Mol Psychiatry 3, 256–260. Quested DJ, Whale R, Sharpley AL et al. (1999). Allelic variation in the 5-HT2C receptor and functional response to the 5-HT2c receptor antagonist mCPP. Psychopharmacology 144,
306–307.
Reddy R, Sahebarao MP, Mukherjee S, Murthy JN (1991). Enzymes of the antioxidant defense
system in chronic schizophrenic patients. Biol Psychiatry 30, 409–412.
265 Genetics of drug-induced tardive dyskinesia
Rietschel M, Krauss H, Muller DJ et al. (2000). Dopamine D3 receptor variant and tardive dyski- nesia. Eur Arch Psychiatry Clin Neurosci 250, 31–35.
Rosengarten H, Schiweitzter JW, Friedhoff AJ, (1994). Possible genetic factors underlying the pathophysiology of TD. Pharmacol Biochem Behav 49, 633–667.
Rosengarten H, Schweitzer JW, Friedhoff AJ (1999). The effect of novel antipsychotics in rat oral dyskinesia. Prog Neuropsychopharmacol Biol Psychiatry 23, 1389–1404.
Rubovits R, Klawans HL Jr. (1972). Implications of amphetamine induced stereotyped behavior as a model for tardive dyskinesia. Arch Gen Psychiatry 27, 502–507.
Sachse C, Brockmoller J, Bauer S, Roots I (1999). Functional significance of a C→A polymor- phism in intron 1 of the cytochrome P-450 CYP1A2 gene tested with caffeine. Br J Clin Pharmacol 47, 445–449.
Schmidt CJ, Sorensen SM, Kehne JH, Carr AA, Palfreyman MG (1995). The role of 5-HT2A recep- tors in antipsychotic activity. Life Sci 56, 2209–2222.
Schwartz JC, Diaz J, Pilon C, Sokoloff P (2000). Possible implications of the dopamine D(3) receptor in schizophrenia and in antipsychotic drug actions. Brain Res Rev 2–3, 277–287.
Segman RH, Lerer B (2002). Age and the relationship of dopamine D3, serotonin 2C and serot- onin 2A receptor genes to abnormal involuntary movements in chronic schizophrenia. Mol Psychiatry 6, 225–229.
Segman R, Neeman T, Heresco-Levy U et al. (1999). Genotypic association between the dopa- mine D3 receptor gene and tardive dyskinesia in chronic schizophrenia. Mol Psychiatry 4, 247–253.
Segman R, Heresco-Levy U, Finkel B et al. (2000). Association between the serotonin 2C recep- tor gene and tardive dyskinesia in chronic schizophrenia: additive contribution of 5-HT2Cser and DRD3gly to susceptibility. Psychopharmacology (Berl) 152, 408–413.
Segman RH, Heresco-Levy U, Finkel B et al. (2001). Association between the serotonin 2A recep- tor gene and tardive dyskinesia in chronic schizophrenia. Mol Psychiatry 6, 225–229.
Shimoda-Matsubayashi S, Matsumine H, Kobayashi T, Naka-gawara-Hattori Y, Shimizu Y, Mizuno Y (1996). Structural dimorphism in the mitochondrial targeting sequence in the human man- ganese superoxide dismutase gene. Biochem Biophys Res Commun 226, 561–565.
Shinkai T, Ohmori O, Kojima H, Terao T, Suzuki T, Abe K (1998). Negative association between T102C polymorphism of the 5-HT2A receptor gene and schizophrenia in Japan. Hum Hered 48, 212–215.
Sokoloff P, Giros B, Martres MP, Bouthenet ML, Schwartz JC (1990). Molecular cloning and char- acterization of a novel dopamine receptor (D3) as a target for neuroleptics. Nature 347, 146– 151.
Spurlock G, Heils A, Holmans P et al. (1998). A family based association study of T102C poly- morphism in 5HT2A and schizophrenia plus identification of new polymorphisms in the pro- moter. Mol Psychiatry 3, 42–49.
Steen VM, Lovelie R, MacEwan T, McCreadie RG (1997). Dopamine D3-receptor gene variant and susceptibility to tardive dyskinesia in schizophrenic patients. Mol Psychiatry 21, 139–145. Tamminga CA, Dale JM, Goodman L, Kaneda H, Kaneda N (1990). Neuroleptic induced vacuous chewing movements as an animal model of tardive dyskinesia; a study in 3 rat strains.
Psychopharmacology 102, 474–478.
266 R. H. Segman and B. Lerer
Tamminga CA, Thaker GK, Moran M, Kakigi T, Gao XM (1994). Clozapine in tardive dyskine- sia: observations from human and animal model studies. J Clin Psychiatry 55(Suppl B), 102–106.
Tan EK, Khajavi M, Thornby JI, Nagamitsu S, Jankovic J, Ashizawa T (2000). Variability and validity of polymorphism association studies in Parkinson’s disease. Neurology 55, 533–538. Tay AH, Lim LC, Lee WL, Wong KE, Wong LY, Tsoi WF (1997). Association between allele 1 of
T102C polymorphism, 5-hydroxytryptamine 2A receptor gene and schizophrenia in Chinese
males in Singapore. Hum Hered 47, 298–300.
Thompson J, Thomas N, Singleton A et al. (1997). D2 dopamine receptor gene (DRD2) Taq1 A
polymorphism: reduced dopamine D2 receptor binding in the human striatum associated with
the A1 allele. Pharmacogenetics 7, 479–484.
Turecki G, Briere R, Dewar K (1999). Prediction of level of serotonin 2A receptor binding by
serotonin receptor 2A genetic variation in postmortem brain samples from subjects who did
or did not commit suicide. Am J Psychiatry 156, 1456–1458.
Waren JT, Peacock ML, Rodriguez LC, Fink JK (1993). An MspI polymorphism in the human
serotonin receptor gene: detection by DGGE and RFLP analysis. Hum Mol Genet 2, 338. Weinhold P, Wegner JT, Kane JM (1981). Familial occurrence of tardive dyskinesia. J Clin
Psychiatry 42, 165–166.
Williams J, McGuffin P, Nothen M, Owen MJ (1997). Meta-analysis of association between the
5-HT2A receptor T102C polymorphism and schizophrenia. European Multicentre Association
Study of Schizophrenia Collaborative Group [letter]. Lancet 349, 1221.
Wolf A, Schultz LJ (1997). The serotonin 5-HT2C receptor is a prominent serotonin receptor in basal ganglia: evidence from functional studies on serotonin mediated phosphoinositide
hydrolysis. J Neurochem 69, 1449–1458.
Zahm DS, Brog JS (1992). On the significance of subterritories in the “accumbens” part of the
rat ventral striatum. Neuroscience 50, 761–767.
12
Functional gene-linked polymorphic regions in pharmacogenetics
Marco Catalano
IRCCS H. San Raffaele, Department of Neuropsychiatric Sciences, Milan, Italy
OVERVIEW
Starting from the complexity of neural pathways and their close integration, this chapter focuses on the possible advantages of investigating polymorphisms involved in the regula- tion of gene expression or in variation of enzymatic activity, using examples related to neu- rotransmitter pathway modulation (namely, uptake and metabolism). These examples (i.e., serotonin transporter-linked polymorphic region, monoamine oxidase A (MAO-A) promoter region, and catechol-O-methyltransferase (COMT)) are also used to support the hypothesis that quantitative variations of expression and functional levels would be better than structu- ral changes at single receptor sites to identify differences in both treatment response and, likely, psychopathology. The possible influence of the serotonin (5-HT) transporter variants on the efficacy of fluvoxamine, paroxetine, pindolol augmentation, and on the effects of total sleep deprivation, is described in depression. The possible importance of this functional poly- morphism in obsessive-compulsive disorder (OCD), stress reactivity and panic disorder (PD) is also discussed. Variations in MAO-A activity in female patients are discussed in relation to the pharmacogenetics of panic disorder, together with some hypotheses regarding the chromosomal location of the gene. Some preliminary results are described linking a func- tional polymorphism in the coding sequence of the gene COMT and antidepressant response in unipolar depression and bipolar mood disorder. Finally, the expected impact of new approaches (i.e., orphan receptor research, nucleic acid chips, and single nucleotide poly- morphisms) is discussed in terms of the advantages and pitfalls (e.g., many new exciting data but also a stronger need for very careful replications and analysis). Ethical considerations are also outlined in that pharmacogenomics undoubtedly bears scientific promise but also raises ethical concerns for the conduct of research with human subjects, particularly with respect to confidentiality, risk–benefit analysis, DNA banking, and pharmacoeconomic issues.
Introduction
Neurotransmission and neuromodulation represent the final result of a complex series of events involving transmitters, receptors, transporters, transducers,

267

268 M. Catalano
enzymes, and other molecules for the fine tuning of many (if not all) physiological processes in most living organisms.
In Vertebrata, moreover, all neural pathways are closely integrated; consequently, an event at a given point can have consequences at many other points sideways, upwards or downwards. Early biological psychiatry was compelled by the know- ledge of those times into restricted, simplistic views and limited to the “synecdo- chaic fallacy.” For many years, the broad and rough concept of neurotransmitter pathway dysfunction (“one receptor–one dysfunction–one disorder”) as the most important causal factor for psychopathology has been a leading “dogma” in biolog- ical psychiatry. The therapeutic effect of many psychotropic drugs was exclusively attributed to their ability to block or enhance these altered pathways, thus support- ing, in turn, a sort of crystallized view (i.e., one pathway–one disease) (Hyman and Nestler, 1996). This, for example, gave rise to the dopaminergic hypothesis of schizophrenia and the noradrenergic–serotonergic dichotomy of mood disorders. More recently, serotonergic hypotheses have gained a great popularity for a (maybe too) wide spectrum of psychiatric disorders, ranging from mood to anxiety, to OCDs.
The advent of molecular biology techniques, and their application to the complex issue of central nervous system (CNS) functioning, has contributed to a substantial modification of these limited views. The many findings they have allowed have highlighted the limits of traditional psychiatric nosology and have had impact on knowledge and clinical practice. Consequently, diagnostic research has not seemed to keep pace with advances in molecular neurobiology, and the rate-limiting step in identifying susceptibility genes for psychiatric disorders has become phenotype definition. Therefore, the success of psychiatric genetics may require the development of a “genetic nosology” that can classify individuals in terms of the heritable aspects of psychopathology (van Praag 1997; Smoller and Tsuang, 1998).
Molecular techniques, indeed, have played a fundamental role in revealing some aspects of the real complexity of the CNS. They have enabled, for example, a more precise definition of receptor families and subtypes and allowed the development of more sophisticated genome-based pharmacogenetic approaches, thus offering more opportunities than views and studies based only on peripheral phenotypes. Furthermore, the growing interest in regulatory regions and post-transcriptional regulatory mechanisms is offering not only interesting findings but also new incen- tives for a more open-minded molecular psychiatry. Quantitative variations of expression (regardless of their pre- or post-transcriptional origin) could explain the complexity of CNS functioning better and may be better for explaining both psychopathology and clinical response to psychotropic drugs than structural vari- ations that resemble a type of “on/off model.” These advances can justify increased
269 Functional gene-linked polymorphic regions
optimism about the future of psychopharmacogenetics, and it is reasonable to believe that clinical applications will follow in the next few years, albeit limited to some psychiatric disorders.
The serotonin transporter
The serotonin transporter (5-HTT) plays a pivotal role in regulating serotonergic transmission by removing 5-HT from the synaptic cleft. 5-HTT represents a prime target for the widely used serotonin reuptake inhibitors, both selective and not (SSRIs and SRIs, respectively). The 5-HTT is encoded by a single gene located on chromosome 17 and organized in 14 exons spanning about 35 kilobases (kb). A mutation screening study of the whole coding region of the gene gave no exciting result (Di Bella et al., 1996) but a functional polymorphism in the promoter site (the 5-HTT gene-linked polymorphic region (5-HTTLPR)) has opened the doors to some interesting studies. This polymorphism consists of the insertion or dele- tion of a 44bp sequence (Heils et al., 1996) and gives rise to two variants: long (l), and short (s). These variants showed different transcriptional efficiencies, resulting in different 5-HTT protein expression (the basal transcriptional activity of the l variant being more than twice that of the s variant) and functional levels. In fact, in vitro studies showed that the differences in 5-HTT mRNA synthesis resulted in different 5-HTT expressions and 5-HT cellular uptake (Lesch et al., 1996). Moreover, 5-HTTLPR genotypes seem to affect significantly 5-HTT-binding sites and mRNA levels in postmortem human brain tissue (Little et al., 1998). Therefore, the 5-HTT locus could represent a good candidate for psychiatric genetics.
Many studies have been performed on the possible role of this locus in confer- ring genetic susceptibility for different psychiatric disorders, with interesting (even if sometimes controversial) results suggesting, as a whole, a possible link to the anxiety–depression spectrum (for a review see Lesch and Mössner (1998)). However, since psychiatric disorders are likely multifactorial and oligo- or poly- genic, a gene (or genes) with a minor or even null contribution in conferring sus- ceptibility to a given disease could play a significant role in determining the response to treatment. Briefly, it could influence pharmacokinetic and/or pharma- codynamic parameters, act directly at a target site, or condition onset and/or inten- sity of side effects (Kalow, 1997; Iyer and Ratain, 1998; Nebert, 1999; Masellis et al., 2000). Indeed, interesting results have been produced using this polymorphism in pharmacogenetic approaches.
Two consecutive studies by the same research group have suggested a relation- ship between the antidepressant efficacy of two SSRIs, fluvoxamine (Smeraldi et al., 1998) and paroxetine (Zanardi et al., 2000), and allelic variation within the 5- HTTLPR, at equivalent drug plasma levels. In both studies, l/l homozygotes

270 M. Catalano

(a)
(b)
Fig. 12.1.
Patterns of symptomatological change as shown by HDRS (Hamilton Depression Rating Scale) score during treatment with fluvoxamine alone (a) or fluvoxamine plus pindolol
(b) in groups carrying long (l) or short (s) variants of the serotonin transporter linked polymorphic region (5-HTTLPR). Genotypes were l/l ( ; n 15), l/s (❇; n 23), and s/s
( ; n 15 (a), n 8 (b) (see text)). s/s versus l/l and l/s: *p 0.05; ** p 0.0001. s/s treated with fluvoxamine plus placebo (a) versus s/s treated with fluvoxamine plus pindolol (b): p 0.01; , p 0.0001. l/s treated with fluvoxamine plus placebo (a) versus l/s treated with fluvoxamine plus pindolol (b): † p 0.0001. l/l treated with fluvoxamine plus placebo (a) versus l/l treated with fluvoxamine plus pindolol, p 0.01. (With permission from Smeraldi et al. (1998). Polymorphism within the promoter of the serotonin transporter gene and antidepressant efficacy of fluvoxamine. Molecular Psychiatry 3, 508–511.)
showed a more prompt and better antidepressant response than l/s and s/s individ- uals (Figs. 12.1 and 12.2). Even more interestingly, the former study was performed on a particular type of depression, delusional depression. This is a very severe form of mood disorder, often also characterized by refractoriness to antidepressant treat- ments. In these cases, it is common to use augmentation therapy with the addition of pindolol (a mixed -adrenergic and 5-HT1A antagonist) to boost antidepressant effect by preventing inhibition of serotonergic firing by activation of somatoden- dritic autoreceptors. Unfortunately, side effects are not uncommon, mainly because of the -adrenergic antagonism. Consequently, the possibility of identify- ing a subset of patients who may benefit from pindolol augmentation could
271 Functional gene-linked polymorphic regions

Fig. 12.2.
Paroxetine plasma levels in patients with different genotypes for the serotonin transporter in the linked polymorphic region 5-HTTLPR: l/l ( ; n 16); l/s (❇, n 26); s/s ( , n 16).
A two-way repeated measures ANOVA (MANOVA) on HDRS (Hamilton Depression Rating Scale) scores (with time as the within subjects factor, genotype as the between subjects factor, and paroxetine plasma levels as covariate) showed a highly significant effect of genotype (F 10.62; df 2, 54; p 0.0001), and of genotype time interaction (F 9.06; df 8, 220; p 0.0001). Within-cell regression showed no significant effects of paroxetine plasma levels. Post hoc comparisons (Newman–Keuls test) showed that group l/l had significantly better scores than group l/s and s/s from week 2 to week 4; while group l/s had significantly better scores than group s/s from week 2 to week 4. (IRCCS Hospital San Raffaele, Department of Neuropsychiatric Science, unpublished figure. Ref. paper: Zanardi et al., 2000.)
represent a significant advance in the treatment of this severe form of depression. In fact, our finding showed that the addition of pindolol abolished the genotype effect on fluvoxamine response. In other words, in the group of patients treated with fluvoxamine plus pindolol, all the genotypes acted like the l/l homozygotes treated with fluvoxamine alone, suggesting that only the s/s homozygotes should be particularly amenable to pindolol addition.
These findings have been recently confirmed by a submitted study performed on an independent sample of 155 depressed patients using the same design (Serretti, personal communication). In this case too, l/l homozygotes showed a better response to fluvoxamine, and pindolol augmentation ameliorated the outcome in
272 M. Catalano
5-HTTLPR s/s homozygotes, again reducing the difference in the response rate among the genotypes. Another study (Pollock et al., 2000) obtained interesting results using paroxetine in a group of elderly depressed patients. The findings sug- gested a prompter antidepressant effect only in subjects bearing the l/l alleles sug- gesting a significant relationship between 5-HTTLPR and acute response to SSRI therapy. SSRI plasma levels were also used as variable in the data analysis in this study. All these results suggest that genotyping at 5-HTTLPR could represent a promising approach to individualize the treatment of depression.
However, a recent study in patients of Asiatic descent indicated a better antide- pressant response in s/s homozygotes (Kim et al., 2000). This raises the possibility that other factors, such as ethnic differences or population stratification, could act as confounders in pharmacogenetic protocols. This last study differed from the ones suggesting a better response for l/l homozygotes in several ways. The main difference resided in the assessment protocol, as the authors used an end-point analysis instead of repeated measures: using the dichotomous variable “respon- der/nonresponder” instead of a time course assessment of depressive symptoms. There was also the possible presence of placebo responders (to be suspected also in the study by Pollock and associates (2000)). This could be an important issue, given the possibility that 5-HTTLPR variants might have significant effects on personal- ity features (Hamer et al., 1999; Greenberg et al., 2000; Osher et al., 2000), which in turn, could condition a different psychological attitude toward medication. Last, but not least, results in the study of Kim et al. (2000) were not corrected for the esti- mated bioavailability of the drug. This is a difficult issue in that even the exclusion of outliers for plasma levels of the given SSRI is not a guaranteed method, although it would protect results from influences owing to extreme variations in bioavailabil- ity. Unfortunately, to the best of my knowledge, only one study exists on the plasma/CNS ratio of SSRIs (Strauss et al., 1998). Nevertheless, despite these differ- ences, all the studies seem to confirm the possible importance of the 5-HTTLPR in the pharmacogenetics of serotonergic antidepressants.
SRIs and SSRIs are also used to treat OCD and PD. In OCD, two studies seem to exclude any significant influence of 5-HTT variant on the response rate to usual SRIs treatment protocols (Billet et al., 1997; Catalano et al., 2000), suggesting that in this disorder other etiopathogenetic factors could predominate. In other words, it cannot be excluded that different biochemical pathways could subtend or condi- tion the efficacy of SRIs in different disorders, and there may be different etiopa- thogenetic mechanisms. This hypothesis seems to fit with the well-known differences between depression and OCD in latency of therapeutic effect and dose regimen (Montgomery, 1994; Tollefson et al., 1994; Greist et al., 1995; Blier and de Montigny, 1998).
By comparison, a preliminary result from an Italian sample of PD seems to
273 Functional gene-linked polymorphic regions
suggest a significant relationship between the 5-HTTLPR l allele and positive response to treatment with SSRIs (citalopram and paroxetine) but not with SRIs (clomipramine). This study also suggested the presence of a gender difference with regard to both response and some clinical features, as indicated in a previous study on PD that will be discussed later in this chapter (Deckert et al., 1999). A higher fre- quency of the l allele was significantly associated with a positive response to treat- ment, as well as with the absence of agoraphobia, only in the female subsample (Bosi, 2000). This last finding seems to be in agreement with results from Wichems and associates (2000), who found increases in anxiety and stress responses to be more prominent in female than in male 5-HTT knockout mice.
Two further studies on 5-HTTLPR deserve mention as they suggest other pos- sible roles of 5-HTT. In the first, cerebrospinal fluid (CSF) 5-hydroxyindoleacetic acid (5-HIAA) and blood pressure plus heart rate were used as indices of 5-HT turnover and stress reactivity, respectively. All these parameters showed a signifi- cant increase in subjects bearing one or two copies of the 5-HTTLPR l allele (Williams et al., 2001). The second study described the effects of total sleep depri- vation (TSD) in bipolar patients and showed a significantly better response (expressed as higher perceived mood levels after TSD) in l/l homozygotes for 5- HTTLPR (Benedetti et al., 1999) (Fig. 12.3). Taken together, these results seem to support the hypothesis of 5-HT as a “wide-spectrum” modulating neurotransmit- ter (as also suggested by its multiple pre- and postsynaptic receptor subtypes) with different roles in different backgrounds. In other words, if there was a significant influence of 5-HTTLPR variants it would indicate a more direct role of 5-HT in a given situation (for example, depression and treatment response); a nonsignificant or negligible effect could suggest a more indirect role or no role at all.
Obviously, these data must be cautiously interpreted, and their further replica- tion is critical. However, they seem to confirm the possible importance of this func- tional polymorphism not only as a pharmacogenetic tool but also as an auxiliary tool to develop a more precise genetic nosology in psychiatry.
Neurotransmitter metabolism
Neurotransmitter metabolism is another “hot-spot” both for psychiatric genetics and psychopharmacogenetics. Differences in enzymatic activity can condition different rates of neurotransmitter degradation, which, in turn, can condition different functional consequences. Two classes of enzyme play a fundamental role in neurotransmitter catabolism: the MAOs and COMT.
Two MAO isozymes are known, A and B, which differ in molecular weight and immunological properties. Two adjacent genes, mapped on the 11.23–11.4 region of the X chromosome, encode the isozymes. Norepinephrine and 5-HT are the

274 M. Catalano

Fig. 12.3.
The effect of genotype for the serotonin transporter in the linked polymorphic region (5-HTTLPR) on the response to total sleep deprivation in bipolar patients. l/l ( ; n 20); l/s (❇; n 35); s/s ( ; n 13). Self-ratings of mood (scores 0, 50, 100, correspond to depression, euthymia, and euphoria) before and after sleep deprivation were analyzed using a two-way repeated measures ANCOVA, with time (p 0.0008) and genotype (not significant) as independent factors (time genotype interaction: p 0.05) and age at onset as covariate (not significant). (IRCCS. Hospital San Raffaele, Department of Neuropsychiatric Science unpublished figure. Ref. paper: Benedetti et al., 1999.)
preferential substrates of MAO-A, while MAO-B preferentially oxidizes phenyl- ethylamine and dopamine.
Monoamine oxidase A
Inhibitors of MAO-A can be both reversible and irreversible in action and have antidepressant properties; they have been shown to be effective in the treatment of PD. Therefore, the gene MAO-A might be considered a candidate for genetic or pharmacogenetic studies.
A novel repeat polymorphism in the promoter of MAO-A has been examined for association with PD in two independent samples of different descent (German and Italian). This polymorphism consists of a 30bp motif that repeats two (very rare) to five times. An additional allele, named “3a” was shown to contain three full repeats plus 18bp of the repeated motif. Functional characterization of the poly- morphism using luciferase assay showed that constructs containing the longer repeats (3a, 4, and 5) had a higher transcriptional activity than the shorter (i.e.,
275 Functional gene-linked polymorphic regions
3 repeats, since the 2 repeats variant was not tested, given its rareness), thus sug- gesting increased MAO-A activity in subjects bearing the longer alleles. There was a significantly higher frequency of the longer alleles among females with PD than among control females, whereas no significant difference was observed between male patients with PD and male controls (Deckert et al., 1999). The finding could simply suggest that increased MAO-A activity is a risk factor for PD in female patients, perhaps also explained by epigenetic hormonal influences on the disease; however, the chromosomal location of the gene made the situation more intri- guing. In fact, in females, one allele should have been randomly silenced because of X-inactivation. Accordingly, “long/long” females should have behaved like male hemizygotes bearing only a long allele. It is possible that the differences observed were by chance because of the size differences between the two subsamples. If this is not the case then several hypotheses could be invoked to explain it. It is possible that genomic imprinting has occurred in which the imprint silences the allele from one parent. In this case, the males with PD might bear a silenced MAO-A gene. However, the issue of X-linked imprinting is still very controversial, and the pos- sibility of a living organism without any MAO-A activity would actually be surpris- ing. A further possibility is skewed random X inactivation, which has been reported, as has the tendency for male offspring of skewed mothers to inherit alleles from the inactive X chromosome in the region from Xp11 to Xq22 (Naumova et al., 1995). Alternatively, as X inactivation silences most but not all of the genes on one of the two X chromosomes in mammalian females, women with PD could bear two active MAO-A alleles.
Albeit interesting, these hypotheses are largely speculative. Nevertheless, a differ- ent study on clinical psychopharmacology using an artificial neural network (Politi et al., 1999) has suggested a possible profile of a good responder to moclobemide (a reversible MAO-A inhibitor) as a female with prominent anxiety symptoms. While these results should be assessed cautiously, these data could suggest MAO-A inhibitors as a good choice in the treatment of females with PD.
Interestingly, a sexually dimorphic effect for MAO-A has been shown in two studies, although involving different disorders and the opposite direction. The first study, by Karayiorgou and associates (1999) used a different, intragenic polymor- phism but suggested an association between a high activity MAO-A allele and male OCD probands. The second, very preliminary, showed a similar finding (i.e., a higher frequency of more active alleles in males) in a group of bipolar patients using the MAO-A promoter polymorphism (Casorati, 1999).
Catechol-O-methyltransferase
COMT, along with MAO, represents the major mammalian enzymatic system involved in the degradation of catecholamines. A gene mapped to 22q11.1–q11.2 encodes the enzyme (Grossman et al., 1992). Variation in COMT enzyme activity
276 M. Catalano
is a common finding in humans. This variation is attributable to a functional poly- morphism within the coding region of the gene. It consists of a G-to-A transition at codon 158 of the gene, and gives rise to two variants with huge differences in enzymatic activity (i.e., a three- to fourfold variation between the high- and low- activity alleles) (Lachmann et al., 1996a).
This metabolic role, together with findings regarding behavioral and psychiatric consequences of the 21q11 microdeletion (Shprintzen et al., 1992; Pulver et al., 1994; Lachmann et al., 1996b; Papolos et al., 1996), has suggested COMT as a can- didate gene and has prompted several psychiatric genetic studies, all with contrast- ing results (Kirov et al., 1998, 1999; Mynett-Johnson et al., 1998; de Chaldee et al., 1999; Wei and Hemmings, 1999; Geller and Cook, 2000; Henderson et al., 2000; Li et al., 2000). Consequently, it is still unclear whether COMT has any role in confer- ring susceptibility to psychiatric disorders. However, some findings suggesting a possible interaction between 5-HTTLPR and COMT polymorphisms (Benjamin et al., 2000 a,b) prompted a study of COMT in relation to fluvoxamine antidepress- ant response despite the affinity profile of the enzyme. In fact, COMT does not degrade 5-HT, having mainly dopamine and norepinephrine as substrates.
A very recent unpublished study on 94 unipolar depressed and 65 bipolar patients in which the time course of response to fluvoxamine was assessed by COMT genotype suggested a less favorable response to the antidepressant only in unipolar depressed patients homozygous for the higher activity variant of the gene (Tentoni, 2000). Obviously, the possibility of a chance finding, despite statistical significance, cannot be discarded a priori, and application to clinical practice is clearly premature. Moreover, as ethnic differences in both erythrocyte COMT activity and COMT gene alleles have been previously demonstrated (McLeod et al., 1998; Palmatier et al., 1999), population-linked effects may produce false-positive results. Nevertheless, this finding could support proposed biological differences between bipolar and unipolar disease and, if confirmed, could offer a new pharma- cogenetic tool for the treatment of depression. First, it could allow the identifica- tion of unipolar patients who might be particularly amenable to fluvoxamine (or other SSRIs) treatment and, second, even suggest COMT inhibitors as adjuvant drugs (as in Parkinson’s disease).
Conclusions
Regulatory regions and metabolism of neurotransmitters seem to have something to offer psychopharmacogenetics. Indeed, quantitative variations of expression provide better explanations of both psychopathological variations and response to treatment than do mutations leading to alterations at receptor sites. Pharmacogenetic studies are also providing data to support the hypothesis that

277 Functional gene-linked polymorphic regions

Fig. 12.4.
Influence of catechol-O-methyltransferase (COMT) genotypes (HH ( ), HL ( ), and LL ( ); where H indicates higher activity and L lower activity) on response to fluvoxamine in unipolar depression. A two-way repeated measures ANOVA (MANOVA) on HDRS (Hamilton Depression Rating Scale) scores (with time as the within subjects factor, genotype as the between subjects factor) showed a significant effect of genotype (F 3.41; df 2, 91; p 0.0372), of time (F 247.86; df 6, 546; p 0.0001), and of genotype time interaction (F 2.20; df 12, 546; p 0.0107). Post hoc analysis (Newman–Keuls test) showed significant statistical differences starting from weeks 2 (HH versus HL) and 3 (HH versus LL). (Adapted from Tentoni, 2000.)
response to the same drug does not always mean that the treated disorders share identical pathogenetic pathways. For example, SRIs are advantageous in treating both depression and OCD, but a direct influence of 5-HTTLPR variant seems to be specifically restricted to depression. Similarly, a lack of responsiveness to a given drug does not always mean a misdiagnosis, since refractory patients are a reality, and now we may hope to identify them in advance. Similar reasoning can be applied to adverse reactions, as the real benefit of a treatment is always the fruit of a careful balance between efficacy and side effects.
278 M. Catalano
Obviously, genes with products that are prime targets of specific drugs are far more interesting, and screening for polymorphisms should include regulatory regions and/or sequences. On the pharmacological side, this means that, when pos- sible, drugs that display a narrower range of targets are preferable starting points. Also, it is worth recalling that, since psychiatric disorders are likely multifactorial and oligo- or polygenic, a gene with a minor contribution in conferring suscepti- bility could play a significant role in determining the response to treatment. In other words, a single gene could contribute only a very minor part of the patholog- ical variance but, at the same time, encode a protein of pivotal importance for the response to a given treatment. It could, for example, exert a significant influence on the metabolism of the drug, or represent a prime target for the drug.
Like all experimental results, psychopharmacogenetic findings need to be cau- tiously regarded and confirmed by independent replication. This holds particularly true now that new approaches can potentially offer significant new information together with a large number of data requiring explanation. For example, investi- gation of orphan receptors could lead to the discovery of many neuropeptides, which, in turn, could represent the starting points of new drug discovery programs (Civelli et al., 1999) and DNA and RNA chip technology could accelerate molecu- lar research by allowing the simultaneous screening of thousand of genes.
Single nucleotide polymorphisms (SNPs) are another new area that has gener- ated great excitement and SNPs are rapidly becoming the main focus of “new phar- macogenetics” (for a review see Pfost et al. (2000)). Prudent skepticism, at least for complex psychiatric disorders, is warranted. For many years, new possible tools and related promises and perspectives have excited interest, from restriction fragment length polymorphisms, through variable number of tandem repeats, high- throughput sequencing, microsatellites, etc. Although these techniques have allowed an enormous number of experiments worldwide and have required high investments, the practical results are quite unbalanced and thus disappointing. It is true that the increasing number of known SNPs could give psychopharmacogenet- ics new chances, but SNP discovery is not enough. SNP scoring is needed, and we know that correlating genotypes with clinical phenotypes is never easy in psycho- pharmacology. Nevertheless, if correctly managed, this approach could be helpful, particularly with regard to polymorphisms involved in post-transcriptional gene regulatory mechanisms, the importance of which has been described by Day and Tuite (1998) and in Chapter 7.
Finally, some ethical consideration and attention to doubts and criticism of so- called pharmacogenomics are needed, as these developments, although bearing sci- entific promise, raise ethical concerns for the conduct of research with human subjects, particularly with respect to confidentiality, risk–benefit analysis, DNA banking, and pharmacoeconomic issues (Issa, 2000).
279 Functional gene-linked polymorphic regions
Acknowledgements
The author gratefully acknowledges his colleagues at DSNP for collaboration and support, and Professor Bernard Lerer for valuable discussion and friendly advice.
REFERENCES
Benedetti F, Serretti A, Colombo C et al. (1999). Influence of a functional polymorphism within the promoter of the serotonin transporter gene on the effects of total sleep deprivation in bipolar depression. Am J Psychiatry 156, 1450–1452.
Benjamin J, Osher Y, Kotler M (2000a). Association between tridimensional personality ques- tionnaire (TPQ) traits and three functional polymorphisms: dopamine receptor D4 (DRD4), serotonin transporter promoter region (5-HTTLPR) and catechol O-methyltransferase (COMT). Mol Psychiatry 5, 96–100.
Benjamin J, Osher Y, Lichtenberg P et al. (2000b). An interaction between the catechol O-methyltransferase and serotonin transporter promoter region polymorphisms contributes to tridimensional personality questionnaire persistence scores in normal subjects. Neuropsychobiology 41, 48–53.
Billett EA, Richter MA, King N, Heils A, Lesch KP, Kennedy JL (1997). Obsessive compulsive dis- order, response to serotonin reuptake inhibitors and the serotonin transporter gene. Mol Psychiatry 2, 403–406.
Blier P, de Montigny C (1998). Possible serotonergic mechanisms underlying the antidepressant and anti-obsessive-compulsive disorder responses. Biol Psychiatry 44, 313–323.
Bosi M (2000). Applicazione di tecniche molecolari alla genetica e farmacogenetica del disturbo di panico. Postgraduate Thesis, University of Milan School of Medicine.
Casorati P (1999). Applicazione di tecniche molecolari ai problemi della nosologia psichiatrica: ruolo di un polimorfismo funzionale nella regione regolatoria del gene codificante per le MAO-A. MD Graduate Thesis, University of Milan, School of Medicine.
Catalano M, Di Bella D, Erzegovesi S, Bellodi L (2000). Serotonin transporter polymorphisms and response to SSRIs in different psychiatric disorders: a mirror of etiopathogenetic differ- ences? Int J Neuropsychopharmacol 3, S26.
Civelli O, Reinscheid RK, Nothacker H-P (1999). Orphan receptors, novel neuropeptides and reverse pharmaceutical research. Brain Res 848, 63–65.
Day DA, Tuite MF (1998). Post-transcriptional gene regulatory mechanisms in eukaryotes: an overview. J Endocrinol 157, 361–371.
de Chaldee M, Laurent C, Thibaut F et al. (1999). Linkage disequilibrium on the COMT gene in French schizophrenics and controls. Am J Med Genet 88, 452–457.
Deckert J, Catalano M, Syagailo YV et al. (1999). Excess of high activity monoamine oxidase A gene promoter alleles in female patients with panic disorder. Hum Mol Genet 8, 621–624.
Di Bella D, Catalano M, Balling U, Smeraldi E, Lesch K-P (1996). Systematic screening for muta- tions in the coding region of the human serotonin transporter (5-HTT) gene using PCR and DGGE. Am J Med Genet 67, 541–545.
 
280 M. Catalano
Geller B, Cook EH Jr. (2000). Ultradian rapid cycling in prepubertal and early adolescent bipo- larity is not in transmission disequilibrium with Val/Met COMT alleles. Biol Psychiatry 47, 605–609.
Greenberg BD, Li Q, Lucas FR et al. (2000). Association between the serotonin transporter pro- moter polymorphism and personality traits in a primarily female population sample. Am J Med Genet 96, 202–216.
Greist J, Chouinard G, DuBoff E et al. (1995). Double-blind parallel comparison of three dosages of sertraline and placebo in outpatients with obsessive-compulsive disorder. Arch Gen Psychiatry 52, 289–295.
Grossman MH, Emanuel BS, Budarf ML (1992). Chromosomal mapping of the human catechol- O-methyltransferase gene to 22q11.1–q11.2. Genomics 12, 822–825.
Hamer DH, Greenberg BD, Sabol SZ, Murphy DL (1999). Role of the serotonin transporter gene in temperament and character. J Pers Disord 13, 312–327.
Heils A, Teufel A, Petri S et al. (1996). Allelic variation of human serotonin transporter gene expression. J Neurochem 66, 2621–2624.
Henderson AS, Korten AE, Jorm AF et al. (2000). COMT and DRD3 polymorphisms, environ- mental exposures, and personality traits related to common mental disorders. Am J Med Genet 96, 102–107.
Hyman SE, Nestler EJ (1996). Initiation and adaptation: a paradigm for understanding psycho- tropic drug action. Am J Psychiatry 153, 151–162.
Issa AM (2000). Ethical considerations in clinical pharmacogenomics research. Trend Pharm Sci 21, 247–249.
Iyer L, Ratain MJ (1998). Pharmacogenetics and cancer chemotherapy. Eur J Cancer 34, 1493–1499.
Kalow W (1997). Pharmacogenetics in biological perspective. Pharm Rev 49, 369–379. Karayiorgou M, Sobin C, Blundell ML et al. (1999). Family-based association studies support a sexually dimorphic effect of COMT and MAOA on genetic susceptibility to obsessive-
compulsive disorder. Biol Psychiatry 45, 1178–1189.
Kim DK, Lim SW, Lee S et al. (2000). Serotonin transporter gene polymorphism and antidepress-
ant response. Neuroreport 11, 215–219.
Kirov G, Murphy KC, Arranz MJ et al. (1998). Low activity allele of catechol-O-methyltransferase
gene associated with rapid cycling bipolar disorder. Mol Psychiatry 3, 342–345.
Kirov G, Jones I, McCandless F, Craddock N, Owen MJ (1999). Family-based association studies of bipolar disorder with candidate genes involved in dopamine neurotransmission: DBH,
DAT1, COMT, DRD2, DRD3 and DRD5. Mol Psychiatry 4, 558–565.
Lachmann HM, Papolos DF, Saito T, Yu YM, Szumlanski CL, Weinshilboum RM (1996a).
Human catechol-O-methyltransferase pharmacogenetics: description of a functional poly- morphism and its potential applications to neuropsychiatric disorders. Pharmacogenetics 6, 243–250.
Lachmann HM, Morrow B, Shprintzen R et al. (1996b). Association of codon 108/158 catechol- O-methyltransferase gene polymorphism with the psychiatric manifestations of velo-cardio- facial syndrome. Am J Med Genet 67, 468–472.
281 Functional gene-linked polymorphic regions
Lesch K-P, Mössner R (1998). Genetically driven variation in serotonin uptake: is there a link to affective spectrum, neurodevelopmental, and neurodegenerative disorders? Biol Psychiatry 44, 179–192.
Lesch K-P, Bengel D, Heils A et al. (1996). Association of anxiety-related traits with a polymor- phism in the serotonin transporter gene regulatory region. Science 274, 1527–1531.
Li T, Ball D, Zhao J et al. (2000). Family-based linkage disequilibrium mapping using SNP marker haplotypes: application to a potential locus for schizophrenia at chromosome 22q11. Mol Psychiatry 5, 77–84.
Little KY, McLaughlin DP, Zhang L et al. (1998). Cocaine, ethanol, and genotype effects on human midbrain serotonin transporter binding sites and mRNA levels. Am J Psychiatry 155, 207–213. Masellis M, Basile VS, Ozdemir V et al. (2000). Pharmacogenetics of antipsychotic treatment:
lessons learned from clozapine. Biol Psychiatry 47, 252–266.
McLeod HL, Syvanen AC, Githang’a J et al. (1998). Ethnic differences in catechol O-methyl-
transferase pharmacogenetics: frequency of the codon 108/158 low activity allele is lower in
Kenyan than Caucasian or Southwest Asian individuals. Pharmacogenetics 8, 195–199. Montgomery SA (1994). Pharmacological treatment of obsessive-compulsive disorder. In Hollander E, Zohar J, Marazziti D, Olivier B, eds. Current Insights in Obsessive Disorder. New
York: John Wiley, pp. 215–226.
Mynett-Johnson LA, Murphy VE, Claffey E, Shields DC, McKeon P (1998). Preliminary evidence
of an association between bipolar disorder in females and the catechol-O-methyltransferase
gene. Psychiatr Geneti 8, 221–225.
Naumova AK, Olien L, Bird LM et al. (1995). Transmission-ratio distortion of X-chromosomes
among male offspring of females with skewed X-inactivation. Dev Genet 17, 198–205. Nebert DW (1999). Pharmacogenetics and pharmacogenomics: why is this relevant to the clini-
cal geneticist? Clin Genet 56, 247–258.
Osher Y, Hamer D, Benjamin J (2000). Association and linkage of anxiety-related traits with a
functional polymorphism of the serotonin transporter gene regulatory region in Israeli sibling
pairs. Mol Psychiatry 5, 216–219.
Palmatier MA, Kang AM, Kidd KK (1999). Global variation in the frequencies of functionally
different catechol-O-methyltransferase alleles. Biol Psychiatry 46, 557–567.
Papolos DF, Faedda GL, Veit S et al. (1996). Bipolar spectrum disorders in patients diagnosed with velo-cardio-facial syndrome: does a hemizygous deletion of chromosome 22q11 result in
bipolar affective disorder? Am J Psychiatry 153, 1541–1547.
Pfost DR, Boyce-Jacino MT, Grant DM (2000). A SNPshot: pharmacogenetics and the future of
drug therapy. Trend Biotechnol 18, 334–338.
Politi E, Balduzzi C, Bussi R, Bellodi L (1999). Artificial neural networks: a study in clinical
psychopharmacology. Psychiatry Res 87, 203–215.
Pollock BG, Ferrell RE, Mulsant BH et al. (2000). Allelic variation in the serotonin transporter
promoter affects onset of paroxetine treatment response in late-life depression. Neuropsycho-
pharmacology 23, 587–590.
Pulver AE, Nestadt G, Goldberg R et al. (1994). Psycotic illness in patients diagnosed with velo-
cardio-facial syndrome and their relatives. J Nerv Mental Disord 182, 476–478.
282 M. Catalano
Shprintzen RJ, Goldberg RB, Golding-Kushner KJ, Marion RW (1992). Late onset psycosis in velo-cardio-facial syndrome. Am J Med Genet 42, 141–142.
Smeraldi E, Zanardi R, Benedetti F, Di Bella D, Perez J, Catalano M (1998). Polymorphism within the promoter of the serotonin transporter gene and antidepressant efficacy of fluvoxamine. Mol Psychiatry 3, 508–511.
Smoller JW, Tsuang MT (1998). Panic and phobic anxiety: defining phenotypes for genetic studies. Am J Psychiatry 155, 1152–1162.
Strauss WL, Layton ME, Dager SR (1998). Brain elimination half-life of fluvoxamine measured by 19F magnetic resonance spectroscopy. Am J Psychiatry 155, 380–304.
Tentoni L (2000). Approccio molecolare alla farmacogenetica dei disturbi dell’umore. MD Graduate Thesis, University of Milan School of Medicine.
Tollefson GD, Ramprey AH Jr., Potvin JH et al. (1994). A multicenter investigation of fixed-dose fluoxetine in the treatment of obsessive-compulsive disorder. Arch Gen Psychiatry 51, 559–567. van Praag HM (1997). Over the mainstream: diagnostic requirements for biological psychiatric
research. Psychiatry Res 72, 201–212.
Wei J, Hemmings GP (1999). Lack of evidence for association between the COMT locus and
schizophrenia. Psychiatr Genet 9, 183–186.
Wichems C, Li Q, Andrews A, Lesch KP, Murphy DL (2000). Serotonin transporter knockout
mice show a spontaneous behavioural phenotype of increased “anxiety” and stress responses.
Int J Neuropsychopharmacol 3, S47.
Williams RB, Marchuk DA, Gadde KM et al. (2001). Central nervous system serotonin function
and cardiovascular responses to stress. Psychosomat Med 63, 300–305.
Zanardi R, Benedetti F, Di Bella D, Catalano M, Smeraldi E (2000). Efficacy of paroxetine in depression is influenced by a functional polymorphism within the promoter of serotonin
transporter gene. J Clin Psychopharmacol 20, 105–107.
13
Alternative phenotypes and the pharmacogenetics of mood and anxiety disorders
Emanuela Mundo and James L. Kennedy Centre for Addiction and Mental Health, University of Toronto, Canada
OVERVIEW
The biological mechanisms of action of the main classes of antidepressant compound involve the serotonin (5-HT) system. Consequently, genes of this system have been consid- ered ideal candidates in pharmacogenetic studies of the antidepressant response. There are critical methodological issues created by the complexity of the definition of the phenotypes (i.e., categorical versus dimensional), the involvement of nongenetic factors in determining the clinical effect of antidepressants, and the different genetic strategies available to detect genetic susceptibility in complex traits (e.g., family-based association studies, transmission disequilibrium test for qualitative and quantitative traits). In this chapter, we present and discuss the most recent findings on genetic predictors of the response to antidepressants in mood and anxiety disorders. The need for a more homogeneous phenotype definition (e.g., including phenotypes related to the diagnosis such as rapid cycling course, psychotic symp- toms, atypical features) is pointed out. We also propose and discuss the role of alternative phenotypes (side effects or non-desirable reactions) in pharmacogenetic studies focused on the prediction of the clinical effect of antidepressants. As an example, the phenomenon of antidepressant-induced mania, as an abnormal response to antidepressants, is described. The most recent data on the role of candidate genes (particularly for the 5-HT system, e.g., 5-HTT, 5HT1D, 5HT2A) in contributing to the risk of developing this phenotype are pre- sented and discussed. In the conclusion, the importance of having stable (i.e., genetic) pre- dictors of normal and abnormal responses to antidepressants is pointed out, as well as the role of molecular genetic strategies in the clinical management of psychiatric disorders requiring treatment with antidepressant compounds.
Introduction
Most of the known compounds with antidepressant activity have as a main target the monoaminergic systems. Tricyclic antidepressants (TCAs) block the reuptake

283

284 E. Mundo and J. L. Kennedy
of serotonin (5-HT) and/or norepinephrine, selective serotonin reuptake inhibi- tors (SSRIs) selectively block the reuptake of 5-HT, and monoamine oxidase inhib- itors (MAOIs) interfere with the metabolism of 5-HT, norepinephrine, and dopamine. However, all these compounds, as well as other antidepressant strate- gies (e.g, electroconvulsive treatment and sleep deprivation), have in common as a final effect the enhancement of 5-HT transmission, although mediated via differ- ent mechanisms (Blier and de Montigny, 1998). For this reason, 5-HT system genes have been considered the ideal candidates for the study of the prediction of antide- pressant response.
In a pharmacogenetic perspective, the response to antidepressant treatment should be considered a complex phenotype. This means that it is heterogeneous, that its mode of inheritance is unclear, that it is very likely to be controlled by more than one gene, and that it is influenced also by environmental factors (Hirschfeld et al., 1998). Therefore, a clear and careful definition of the phenotype represents a critical step in designing molecular genetic studies for the prediction of the response to antidepressants.
In comparative clinical trials, this phenotype is usually defined in a quantita- tive/dimensional fashion (i.e., the amount of symptom improvement). However, this kind of evaluation alone may lead to confounding conclusions as the “mean” effect of a drug may derive from the co-occurrence in the same treatment group of dramatic improvements and poor responses (or lack of response). By comparison, the inclusion also of a qualitative/categorical evaluation of the response (i.e., number of responders/nonresponders according to validated and “a priori” crite- ria), taking into consideration the variability within treatment groups, gives more precise information about the efficacy of the compounds tested.
In genetic studies, we need to adapt these definitions of the phenotype to the ana- lytical strategies available to identify gene effects in complex traits. To-date, family- based association study designs are considered the most appropriate to detect genetic susceptibility for complex phenotypes, where single genes are thought to be of small effect (Risch and Merikangas, 1996). In these studies, the best strategy in the definition of the “pharmacological response” phenotype appear to be the com- bined use of a dimensional and a categorical approach (Macciardi and Mundo, 1999). The effect of the use of this strategy on the modeling and on the statistical power of pharmacogenetic studies is extensively addressed in Chapter 5.
Genetic predictors of the response to antidepressants in mood and anxiety
disorders: the controversial role of the serotonin transporter gene
Despite a wide variety of compounds with antidepressant properties being avail- able and commonly used, proserotonergic drugs are the ones that have received the

285 Alternative phenotypes and pharmacogenetics
most consideration in a pharmacogenetic perspective. SSRIs are commonly used to treat major depression (Potter, 1998) and some anxiety disorders, such as obsessive- compulsive disorder (OCD) (Greist et al., 1995; Mundo et al., 1997), panic disor- der, and generalized anxiety disorder (Feighner, 1999; Zohar and Westenberg, 2000). The exact mechanism of action of SSRIs in inducing a clinical response is not completely known and it is very likely to be different in depression and OCD (Blier and de Montigny, 1998). However, all these compounds share the ability of blocking the serotonin transporter (5-HTT), although with different degrees of selectivity and potency (Hyttel, 1994).
The 5-HTT is located on the terminals of serotonergic neurons and its function is the reuptake of 5-HTT from the synaptic cleft into the cell. It is encoded by a gene (SLC6A4) located on chromosome 17, which spans 31 kilobases (kb) and consists of 14 exons. This gene has two known polymorphisms. One is within the promoter region and consists of a 44 base pair (bp) insertion/deletion (the 5-HTT gene- linked polymorphic region, 5-HTTLPR) (Heils et al., 1996; Lesch et al., 1996). According to the results of functional studies (Collier et al., 1996; Heils et al., 1996; Lesch et al. (1996) the long variant (l) generates more gene expression than the short one (s): the uptake of 5-HT is up to 50% less in cells carrying one or two copies of the s allele than in cells homozygous for the l allele, and the l/l cells produce steady-state concentrations of 5-HTT mRNA that are up to 1.7 times the concentrations in both l/s and ss cells. These data support the presence of a domi- nant model, instead of a codominant one, for the effect of the s allele. Moreover, according to a recent report, the s variant of the 5-HTTLPR has been suggested to confer higher stability to 5-HT function. In the study of Hanna et al. (1998), the seasonal fluctuations of blood 5-HT concentrations in OCD families were enhanced in subjects homozygous for the l variant in comparison with subjects heterozygous or homozygous for the s allele.
The other polymorphism known for the SLC6A4 is a variable number of tandem repeats (VNTR) located in the second intron of the gene, with three alleles (Stin2.9, Stin2.10 and Stin2.12) known in humans (Ogilvie et al., 1995).
Given the critical role of the 5-HTT in 5-HT neurotransmission, the functional variants of SLC6A4 have been considered ideal candidates for the prediction of the clinical response to pro-serotonergic compounds in both OCD and mood disor- ders (Table 13.1).
The serotonin transporter gene and the antiobsessional response
The relationship between the 5-HTTLPR and antiobsessional response was first investigated in a paper by Billett et al. (1997) in which 72 patients with OCD treated with pro-serotonergic agents (clomipramine or SSRIs) were subdivided into responders and nonresponders to the medication. The allele and genotype
286 E. Mundo and J. L. Kennedy
Table 13.1. Summary of the main studies investigating the association between 5-HTTLPR variants and the response to antidepressants

Study
Billet et al., 1997 McDougle et al., 1998 Smeraldi et al., 1998 Smeraldi et al., 1998 Benedetti et al., 1999 Zanardi et al., 2000 Pollock et al., 2000 Pollock et al., 2000 Kim et al., 2000
Notes:
Treatment
SSRIs, clomipramine SSRIs
Fluvoxamine Fluvoxamine pindolol Sleep deprivation Paroxetine
Paroxetine Nortriptyline Fluoxetine, paroxetine
Diagnosis
OCD OCD BP, UPa BP, UPa BP
BP, UPa UPb UPb
UP
Association/variant None
Negative/l Positive/l None Positive/l Positive/l Positive/l None Positive/s
 
SSRI, selective serotonin reuptake inhibitor; OCD, obsessive-compulsive disorder; BP, bipolar disorder; UP, unipolar disorder; l, long variant; s, short variant.
a Allpatientshadamajordepressiveepisodewithpsychoticfeatures.
b Patientswithmajordepressiveepisodeswithpsychoticfeatureswereexcluded.
frequencies for the 5-HTTLPR were computed and compared between the two groups. No significant differences were found, apparently excluding the role of this polymorphism in predicting the antiobsessional response. However, no definitive conclusions could be drawn from this preliminary study as the definition of the phenotype (i.e., the evaluation of the pharmacological response) was done retrospectively, in a semiquantitative fashion according to a global clinician’s rating, and without validation criteria.
A more recent study performed on a smaller sample of OCD patients (McDougle et al., 1998) suggested an association between the l allele of the 5-HTTLPR and the lack of antiobsessional response to SSRIs, defined by a more appropriate combina- tion of several validated criteria. In this study, a patient was defined as a “respon- der” if he/she met all the following criteria: (i) an improvement 35% on the scores of the Yale–Brown Obsessive-Compulsive scale (Y-BOCS) (Goodman et al., 1989), which is commonly used to rate the severity of OCD symptoms and the symptom subtypes; (ii) a total Y-BOCS score 16; (iii) a global improvement rated as “much” or “very much” at the Clinical Global Impression (CGI) rating scale (Guy, 1976); and (iv) the consensus of the treating clinician and of two of the inves- tigators that the clinical condition was improved. In addition, in the same study, a critical alternative phenotype was evaluated: the presence of comorbid tic disor- ders. The co-occurrence of chronic tics or Gilles de la Tourette syndrome has been considered one of the most relevant predictors of nonresponse to SSRIs in OCD
287 Alternative phenotypes and pharmacogenetics
(McDougle et al., 1994; McDougle, 1997). Consequently, studies on the genetic predictors of the pharmacological response in this disorder should always take into account this variable as a possible confounding factor. According to the data reported by McDougle et al. (1998), the l variant of the 5-HTTLPR was significantly associated with a lack of response to SSRIs in OCD patients either with or without comorbid tic disorders. This suggests that the predictive role of the gene for the effect of the antiobsessional compounds is independent of the presence of one of the traits that usually confer pharmacological resistance in OCD. However, many other phenotypes alternative to the diagnosis of OCD (e.g., presence/absence of poor insight, specific symptom subtypes) could have accounted for the association found in this study and may be primarily related to the gene effect detected.
Finally, these studies (Billet et al., 1997; McDougle et al., 1998), despite having investigated a critical issue in the clinical management of OCD and having tried to identify stable (i.e., genetic) predictors of the treatment outcome, both present the limitation of having considered as homogeneous a category that may be not homo- geneous. There is growing evidence that OCD is a heterogeneous diagnosis, includ- ing patients with different genetic susceptibilities, clinical characteristics, and different response to medication (McDougle, 1997; Alsobrook et al., 1999; Nestadt et al., 2000). These differences should be taken into account and stratified as pos- sible confounding factors in determining genetic predictors of the antiobsessional response. To investigate this issue further, studies considering more homogeneous subgroups of OCD patients are warranted.
The serotonin transporter gene and antidepressant response
In the first controlled study published of SLC6A4 and mood disorders, patients with major depression and with at least one copy of the l variant appeared to show a better response to the SSRI fluvoxamine than patients with the s variant only (Smeraldi et al., 1998). The investigation of another sample of patients with major depression and treated with paroxetine showed the same association between the l variant of the 5-HTTLPR and the antidepressant response (Zanardi et al., 2000). A similar result was reported in the paper by Pollock et al. (2000), where patients treated with paroxetine showed an association between the l variant of the polymorphism and antidepressant response. In the same study, an independent group of patients treated with the noradrenergic antidepressant nortriptyline showed no association with the variants of 5-HTTLPR, suggesting the specificity of this marker in predicting the response to serotonergic compounds. (See also Ch. 12.)
Similar results were obtained investigating a sample of depressed patients with bipolar disorder (BP) who underwent one night of total sleep deprivation. The mechanism of action of total sleep deprivation is still unclear, but there is evidence
288 E. Mundo and J. L. Kennedy
for a specific role of the 5-HT system (Smeraldi et al., 1999). In the study of Benedetti et al. (1999) the 5-HTTLPR was investigated as a genetic predictor of the antidepressant response to one night of total sleep deprivation. Patients homozy- gous for the l variant showed a better improvement of mood ratings than patients heterozygous or homozygous for the s allele.
However, in a recent report by Kim et al. (2000), a significant association was revealed between either homozygosity for the Stin2.12 allele of the VNTR or homo- zygosity for the s allele of the 5-HTTLPR and the response to SSRIs. In particular, the absence of the Stin2.12 allele was the most powerful predictor of nonresponse to treatment.
At present, the significance of these findings and the relationships between the variants of the 5-HTTLPR and antidepressant response remain unclear. It has been commented (Kelsoe, 1998) that subjects homozygous for the s variant have a lower number of 5-HTT sites and, thus, have higher 5-HT levels in the synaptic cleft. This situation induces a higher inhibition of 5-HT firing and release by 5-HT presynap- tic autoreceptors, with the final result being a reduction of overall serotonergic neurotransmission, which in turn predisposes to a poorer response to SSRIs. This hypothesis is apparently confirmed by the evidence that patients treated with adju- vant pindolol (a selective blocker of the 5-HT1A autoreceptors) show no association between the antidepressant response to SSRIs and either of the 5-HTTLPR variants (Smeraldi et al., 1998; see also Ch. 12).
The rationale of administering pindolol together with SSRIs as an adjuvant treat- ment for major depression is rooted in both the mechanism of action of SSRIs and in the complex regulation mechanisms of 5-HT neurotransmission (Artigas et al., 1994; Blier and de Montigny, 1998). The administration of SSRIs induces, within hours, blockade of the 5-HTT and, as a consequence, increased availability of 5-HT in the synaptic cleft. This happens after the first administration of the compounds, with a latency that mostly depends on the time of absorption of the different SSRIs. The increased availability of 5-HT stimulates the presynaptic autoreceptors (5- HT1A), which, as a compensatory mechanism to maintain homeostasis, reduce the firing of the 5-HT neurons, decreasing 5-HT transmission. This is one of the explanatory hypotheses of the 3–4 week latency of the clinical response to antide- pressant compounds, the time needed for the appropriate downregulation of the 5- HT1A autoreceptors, which has been found to be shortened by the addition of pindolol (Artigas et al., 1994; Zanardi et al., 1998). However, the fact that l/l, l/s, and ss subjects who are given pindolol in addition to fluvoxamine respond in the same way regardless their genotype (Smeraldi et al., 1998) partially contradicts the explanatory hypothesis of the association between the l/l genotype and the response to SSRIs discussed above. In fact, if the presence of the l variant of the 5-HTTLPR, which implies a higher reuptake of 5-HT, was truly associated with a lower
289 Alternative phenotypes and pharmacogenetics
inhibition of the 5-HT1A presynaptic autoreceptors, the addition of pindolol to these subjects should result in a better and much faster response to SSRIs. Moreover, the situation appears to be even more controversial considering the recent report by Kim et al. (2000), which associates a good response to SSRIs alone with the s/s genotype instead of the l/l one. The fact that Kim et al. (2000) used an Asian population may have contributed to the discrepant results.
A critical review of these studies (Table 13.1) suggests that the involvement of genes other than SLC6A4 in determining the variability of the effects of SSRIs is quite likely.
As for all complex phenotypes, it is likely that several genes interact to confer sus- ceptibility to the antidepressant response, in a system in which each gene has a small effect. Consequently, the investigation of single candidates is unlikely to give unequivocal and definitive results. Statistical methods of modeling the way in which several genes may interact to determine the susceptibility to the response to pharmacological compounds are now being developed (Macciardi et al., 1999; see also Ch. 5). In this modeling the interaction of genes (e.g., additive, multiplicative) in influencing a given phenotype can be detected for both qualitative and quanti- tative traits.
Given the fundamental role of dendritic autoreceptors in determining the latency and the quality of the clinical response to pro-serotonergic drugs in mood disorders (Blier and de Montigny, 1998), the investigation of variants of the gene 5-HT1A together with the variants of SLC6A4 in patients who are responder or nonresponder to medication and tests of gene–gene interaction models appear to be the most appropriate strategies to deal with the complexity of the pharmacoge- netics of antidepressant response. It should be also considered that several method- ological issues might have contributed to the failure to obtain unequivocal results in studies carried out so far. All these investigations have examined the role of SLC6A4 in predicting the phenotype of antidepressant response. However, several sources of heterogeneity, as well as interesting alternative phenotypes to the diag- nosis, have not been controlled. First, the antidepressant response has been evalu- ated in different diagnostic groups, BP, unipolar disorder (UP), or both. The diagnostic category “major depression” includes different subgroups of patients with different clinical characteristics and biological substrata, which possibly implies different genetic susceptibilities. Second, some aspects of this heterogeneity appear to be critical in influencing the response to treatment. For instance, the presence of “atypical features” (McGrath et al., 2000), psychotic symptoms (Schatzberg and Rothshild, 1992; Coryell, 1998), “rapid cycling” course (Fujiwara et al., 1998), and some temperamental factors (Tome et al., 1997) have been found to confer different degrees of resistance to the antidepressant treatment. As a con- sequence, these are phenotypes alternative to, but strictly related to, the principal
290 E. Mundo and J. L. Kennedy
diagnosis and they should be considered independently when studying the possible genetic predictors of the antidepressant response.
Finally, in the different studies reviewed here, the response to antidepressants has been defined variably (e.g., with hetero- and self-administered ratings), and this also prevents an adequate integration of the different results. The variability in the definition of the phenotypes as well as in the methods of rating is very likely to have added complexity to a trait that is already intrinsically complex, and this is likely to be reflected in controversial results.
The genetic prediction of antidepressant response using alternative
phenotypes
The methodological accuracy of molecular genetic studies for the prediction of response to antidepressants is largely dependent upon the complexity and the het- erogeneity of the phenotype studied. As already discussed, the response to antide- pressants implies several concurrent sources of complexity and variability. These include the genetic and biological heterogeneity observed across different diag- noses and within the same diagnostic category, and the variability of environmen- tal factors that influence both the clinical features of the illness and the response to the medication. The best way to approach the problem of the complexity of phe- notypes in pharmacogenetic studies on psychotropic drugs is to select simpler phe- notypes, qualitatively defined as present/absent, for which the causal relationship with the drug is clearly detectable and for which the correlation genotype– phenotype could be more clear-cut. The occurrence of side effects (or nondesirable reactions) represents one such strategy, and it appears also to be relevant from a clinical perspective. Given that side effects are complex phenotypes, the investiga- tion of more than one marker and the incorporation of nongenetic factors is advis- able. This strategy of including more than one genetic predictor together with nongenetic variables in modeling the prediction of developing side effects from psychotropic medication has been already used successfully in the investigation of the genetic predictors of tardive dyskinesia in long-term neuroleptic-treated patients (Basile et al., 1999, 2000).
Antidepressant-induced mania and the predictive role of serotonin system genes
Epidemiological and clinical features of antidepressant-induced mania
The induction of mania during antidepressant treatment is not a rare phenome- non. It has been described to occur with different frequencies in patients with BP, UP, and OCD (Wehr and Goodwin, 1987; Solomon et al., 1990; Vieta and Bernardo, 1992; Mundo et al., 1993; Altshuler et al., 1995; Boerlin et al., 1998). In

291 Alternative phenotypes and pharmacogenetics
the past, the frequency of induction of mania during antidepressant treatment has been estimated to be 9.5–33%, varying across different mood disorder diagnostic categories (i.e., UP and BP) and different antidepressant treatments (Bunney, 1978; Lewis and Winkour, 1982). However, the phenomenon of antidepressant-induced mania appears to be strictly related to a diagnosis of BP, and in these patients the switch rate during antidepressant treatment is definitively higher than in UP patients (Angst, 1985). Therefore, in the 1990s, the rate of manic switches has been estimated separately for UP and BP patients, and researchers have been primarily focusing on the occurrence of the phenomenon in BP.
According to the report of Stoll et al. (1994) antidepressant-induced manic/hypomanic episodes appear to be clinically different from the spontaneous ones, usually having a shorter duration and less severe psychotic symptoms. Whether the type of antidepressant treatment can influence the risk of mood switches remains controversial. According to Solomon et al. (1990), a manic switch during antidepressant treatment occurs in approximately 20% of the inpatient admissions with a diagnosis of BP regardless of treatment status (TCAs, MAOIs, or electroconvulsive therapy). Other reports, however, show lower rates of antide- pressant-induced manic switches in BP patients treated with SSRIs than in those treated with TCAs or MAOIs (Peet, 1994; Boerlin et al., 1998).
The impact of the occurrence of antidepressant-induced manic switches on the natural course and on the clinical management of BP is quite high (Wehr and Goodwin, 1987; Goodwin and Jamison, 1990). Several studies have focused on the possible clinical predictors of this phenomenon, and to-date, a higher number of previous manic or hypomanic episodes appears to be the only clinical variable affecting the risk for developing induced mania during antidepressant treatment (Angst, 1985; Boerlin et al., 1998).
Pharmacogenetics of antidepressant-induced mania
The pharmacogenetics of the antidepressant response is a rather new field of inter- est and, to-date, there are no systematic studies that have investigated the role of 5- HT system genes in predicting the induction of mania during antidepressant treatment. However, in a recent pilot report, an interesting association between the 5-HTTLPR and the induction of manic switches during antidepressant treatment was shown (Mundo et al., 2000). In this study, two groups of unrelated patients with BP were selected. The first group comprised patients with a positive history of antidepressant-induced mania, with the following characteristics: (i) a confirmed DSM-IV (American Psychiatric Association, 1994) diagnosis of bipolar I or bipolar II disorder; (ii) at least one depressive episode treated with pro-serotonergic anti- depressants (i.e., fluoxetine, fluvoxamine, paroxetine, sertraline, nefazodone, moclobemide, imipramine, or clomipramine); and (iii) at least one episode
292 E. Mundo and J. L. Kennedy
Fig. 13.1.
Frequency of genotypes (l, long allele; s short allele), for the serotonin transporter in the linked polymorphic region 5-HTTLPR in bipolar patients with ( ; n 18) and without ( ; n 76) antidepressant-induced mania. Statistical analysis: chi-square 12.432;
df 2; p 0.001.
fulfilling the DSM-IV criteria for either mania or hypomania (American Psychiatric Association, 1994) developed during the antidepressant treatment. The second group consisted of patients with the following characteristics: (i) a con- firmed diagnosis of bipolar I or bipolar II disorder; and (ii) no antidepressant- induced manic or hypomanic episodes. For all the subjects the 5-HTTLPR was genotyped under blind conditions with respect to the phenotype of occurrence of induced mania or not.
Allelic association analysis showed that among patients with antidepressant- induced mania there was an excess of the s allele (chi-square 14.333; df 1; p 0.0001). The association analysis performed with the genotypes was also signifi- cant, showing a lower rate of homozygosity for the l variant and a higher rate of homozygosity for the s in the group of patients who experienced manic switches during antidepressant treatments (Fig. 13.1). More recently, these findings on the association between the 5-HTTLPR and antidepressant-induced mania have been replicated in a more controlled study including a larger group of BP subjects with a history of antidepressant-induced mania and a matched group of BP subjects who at no time experienced manic switches during treatment with pro- serotonergic compounds (Mundo et al., 2001).
Two additional genes (for the 5-HT1D and the 5-HT2A receptor) have been investigated as possible predictors of the same phenotype (E. Mundo et al., unpub- lished data). The 5-HT1D receptor is a terminal autoreceptor involved in the reg- ulation of 5-HT release, and it is expressed mostly in the limbic regions and in the striatum. It is encoded by an intronless gene located on chromosome 6 (6q14–15)
100 90 80 70 60 50 40 30 20 10 0
         
l/l l/s s/s
 

Frequency (%)
293 Alternative phenotypes and pharmacogenetics
((a) 100 90 80 70 60 50 40 30 20 10 0
(b) 100 90 80 70 60 50 40 30 20 10 0
GG GC CC
                   
Fig. 13.2.
Genotype frequencies in bipolar patients with ( ) and without ( ) antidepressant- induced mania. (a) G861C polymorphism (homozygotes GG and CC; heterozygote GC) of the gene for the 5-hydroxytryptamine (5-HT) 1DB receptor (5-HT1D) (chi-square 1.331; df 2; p 0.514). (b) T102C polymorphisms (TT and CC homozygotes and TC heterozygote) of the gene for the 5HT2A receptor (chi-square 0.115; df, 2; p 0.925).
(Demchyshyn et al., 1992). The polymorphisms known for this gene are G861C, T 261G, and T371G (Nöthen et al., 1994; Lappalainen et al., 1995). 5-HT2A is located on chromosome 13q14–21 and its variants have been investigated as can- didates in mood disorders, with mixed results (Gutierrez et al., 1995; Zhang et al., 1997; Enoch et al., 1999; Vincent et al., 1999).
In the sample of patients with antidepressant-induced mania described above and in a matched group of BP patients who never experienced manic or hypomanic episodes during antidepressant treatment with pro-serotonergic compounds, the G861C polymorphism of 5-HT1D and the T102C polymorphism of 5-HT2A (Warren et al., 1993) were studied. No allelic or genotypic association (Fig. 13.2)
TT TC CC
Frequency (%)
Frequency (%)
294 E. Mundo and J. L. Kennedy
was found with either of the two variants tested. These additional loci, which did not associate with the occurrence of antidepressant-induced mania, appear to confer specificity to the role of 5-HTTLPR in predicting this phenotype. One of the possible hypotheses to explain this specific finding is based on the functional char- acteristics of the polymorphism. According to the data reported by Lesch et al. (1996), subjects homozygous for the s variant have a lower gene expression and, thus, should have fewer 5-HTT sites. This could imply a higher sensitivity to either the block of 5-HT reuptake or the increase of 5-HT availability. Both these effects are commonly induced by the administration of pro-serotonergic compounds that directly or indirectly act on 5-HT neurotransmission (Blier and de Montigny, 1998).
However, if this hypothesis is true, patients with the s/s genotype should be more likely either to respond to pro-serotonergic antidepressants or to show a shorter latency for the response itself. As discussed in the previous section, most of the studies investigating the role of 5-HTTLPR in predicting the response to SSRIs have found a significant association between the antidepressant effect of these drugs and the l variant of the polymorphism, and not with the s. Further studies are needed to address the complexity of the biological mechanisms underlying the antide- pressant response (Kelsoe, 1998). However, from the data here presented, we can reasonably hypothesize that the normal clinical response (i.e., the remission of depressive symptoms with return to euthymia) and the “abnormal” response (i.e., the induction of manic switches) to antidepressant medication are different phe- nomena, implying different biological mechanisms and different genetic suscept- ibilities.
The identification of specific predictors of these two phenotypes will lead to a substantial improvement in the understanding of the complex biological mecha- nisms that underlie the desirable and the nondesirable clinical effects of antide- pressant compounds. Important information on the pathophysiology of depression and mania will be derived from a better understanding of the mecha- nism of action of antidepressants; moreover, such knowledge will contribute to the design of more specific, and thus more efficient, therapeutic strategies.
Conclusions
During the 1990s, there was a significant advance in understanding of the mecha- nism of action of antidepressants and the pathophysiology of the diseases success- fully treated with these drugs. However, the identification of stable (i.e., genetic) predictors of treatment outcome, on which ideal pharmacological strategies for the clinical management of mood and anxiety disorders could be based, remains a difficult task. The heterogeneity of the definition of the phenotype “antidepressant

295 Alternative phenotypes and pharmacogenetics
response,” the different meaning and clinical expression related to it across the different diagnoses and within the same diagnostic category, the complexity of the biological mechanisms involved in the clinical response itself, and the interaction between environmental and genetic factors in determining it, all prevent the pos- sibility of observing a clear and unequivocal correlation between genotype and phenotype in pharmacogenetic studies. For these reasons it may be helpful to use simpler and more homogeneous alternative phenotypes (e.g., nondesirable side effects) and statistical genetic strategies that allow the modeling of complex inter- actions among different genes and between genetic and nongenetic factors. The utility of pharmacological tools in identifying biological heterogeneity and, thus, alternative phenotypes in complex disorders has been underlined recently by several studies conducted in mood and anxiety disorders. The future of pharma- cogenetic studies oriented towards the identification of stable predictors of the response to antidepressants should follow the same direction, combining clinical pharmacological and genetic approaches. This strategy will be quite valuable to understand the biological basis of the heterogeneous effects of the different antide- pressant compounds and the biological basis of psychiatric disorders and adverse reactions to medication, with the consequence of major improvements in the treat- ment and in the long-term outcome of these disorders.
REFERENCES
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edn. Washington, DC: American Psychiatric Association.
Alsobrook JP II, Leckman JF, Goodman WK, Rasmussen SA, Pauls DL (1999). Segregation anal- ysis of obsessive-compulsive disorder using symptom-based factor scores. Am J Med Genet 88, 669–675.
Altshuler LL, Post RM, Leverich GS, Mikalauskas K, Rosoff A, Ackerman L (1995). Antidepressant-induced mania and cycle acceleration: a controversy revisited. Am J Psychiatry 152, 1130–1138.
Angst J (1985). Switch from depression to mania – a record survey over decades between 1920 and 1982. Psychopathology 18, 140–154.
Artigas F, Perez V, Alvarez E (1994). Pindolol induces a rapid improvement of depressed patients treated with serotonin reuptake inhibitors. Arch Gen Psychiatry 51, 248–251.
Basile VS, Masellis M, Badri F et al. (1999). Association of the MscI polymorphism of the dopa- mine D3 receptor gene with tardive dyskinesia in schizophrenia. Neuropsychopharmacology 21, 17–27.
Basile VS, Ozdemir V, Masellis M et al. (2000). A functional polymorphism of the cytochrome P-450 1A2 (CYP1A2) gene: association with tardive dyskinesia in schizophrenia. Mol Psychiatry 5, 410–417.

296 E. Mundo and J. L. Kennedy
Benedetti F, Serretti A, Colombo C et al. (1999). Influence of a functional polymorphism within the promoter of the serotonin transporter gene on the effects of total sleep deprivation in bipolar depression. Am J Psychiatry 156, 1450–1452.
Billett EA, Richter MA, King N, Heils A, Lesch KP, Kennedy JL (1997). Obsessive-compulsive dis- order, response to serotonin reuptake inhibitors and the serotonin transporter gene. Mol Psychiatry 2, 403–406.
Blier P, de Montigny C (1998). Possible serotonergic mechanisms underlying the antidepressant and the anti-obsessive-compulsive disorder responses. Biol Psychiatry 44, 313–323.
Boerlin HL, Gitlin MJ, Zoellner LA, Hammen CL (1998). Bipolar depression and antidepressant- induced mania: a naturalistic study. J Clin Psychiatry 59, 374–379.
Bunney WE (1978). Psychopharmacology of the switch process in affective process. In Lipton MA, Damascio A, Kellam KF, eds. Psychopharmacology: a Generation of Progress. New York: Raven Press, pp. 1249–1259.
Collier DA, Stober G, Li T et al. (1996). A novel functional polymorphism within the promoter of the serotonin transporter gene: possible role in susceptibility to affective disorders. Mol Psychiatry 1, 453–460.
Coryell W (1998). The treatment of psychotic depression. J Clin Psychiatry 60, 326–335. Demchyshyn L, Sunahara RK, Miller K et al. (1992). A human serotonin 1D receptor variant (5- HT1Dbeta) encoded by an intronless gene on chromosome 6. Proc Natl Acad Sci USA 89,
5522–5526.
Enoch MA, Goldman D, Barnett R, Sher L, Mazzanti CM, Rosenthal NE (1999). Association
between seasonal affective disorder and the 5-HT2A promoter polymorphism, -1438G/A. Mol
Psychiatry 4, 89–92.
Feighner JP (1999). Overview of antidepressants currently used to treat anxiety disorders. J Clin
Psychiatry 60(Suppl 22), 18–22.
Fujiwara Y, Honda T, Tanaka Y, Aoki S, Kuroda S (1998). Comparison of early- and late-onset
rapid cycling affective disorders: clinical course and response to pharmacotherapy. J Clin
Psychopharmacol 18, 282–288.
Goodman WK, Price LH, Rasmussen SA et al. (1989). The Yale–Brown Obsessive-Compulsive
Scale (Y-BOCS): Part I. Development, use, and reliability. Arch Gen Psychiatry 46, 1006–1011. Goodwin FK, Jamison KR (1990). Manic Depressive Illness. New York: Oxford University Press. Greist JH, Jefferson JW, Kobak KA, Katzelnick DJ, Serlin RC (1995). Efficacy and tolerability of
serotonin transport inhibitors in obsessive-compulsive disorder: a meta-analysis. Arch Gen
Psychiatry 52, 53–60.
Gutierrez B, Arranz M, Fananas L et al. (1995). 5-HT2A receptor gene and bipolar affective dis-
order. Lancet 346, 969.
Guy W (1976). ECDEU Assessment Manual for Psychopharmacology. [Publication 76–338.]
Washington, DC: US Department of Health, Education and Welfare.
Hanna GL, Himle GA, Curtis GC et al. (1998). Serotonin transporter and seasonal variation in
blood serotonin in families with obsessive-compulsive disorder. Neuropsychopharmacology 18,
102–111.
Heils A, Teufel A, Petri S et al. (1996). Allelic variation of human serotonin transporter expres-
sion. J Neurochem 66, 2621–2624.
297 Alternative phenotypes and pharmacogenetics
Hirschfeld RM, Russel JM, Delgado PL et al. (1998). Predictors of response to acute treatment of chronic and double depression with sertraline or imipramine. J Clin Psychiatry 59, 669–675.
Hyttel J (1994). Pharmacological characterization of selective serotonin reuptake inhibitors (SSRIs). Int Clin Psychopharmacol 9(Suppl 1), 19–26.
Kelsoe JR (1998). Promoter prognostication: the serotonin transporter gene and antidepressant response. Mol Psychiatry 3, 475–476.
Kim DK, Lim SW, Lee S et al. (2000). Serotonin transporter gene polymorphism and antidepress- ant response. Neuroreport 11, 215–219.
Lappalainen J, Dean M, Charbonneau L, Virkkunen M, Linnoila M, Goldman D (1995). Mapping of the serotonin 5-HT1D beta autoreceptor gene on chromosome 6 and direct anal- ysis for sequence variants. Am J Med Genet 60, 157–161.
Lesch KP, Bengel D, Heils A et al. (1996). Association of anxiety-related traits with a polymor- phism in the serotonin transporter gene regulatory region. Science 274, 1527–1531.
Lewis JL, Winkour G (1982). The induction of mania. Arch Gen Psychiatry 39, 303–306. Macciardi F, Mundo E (1999). Power of quantitative transmission disequilibrium test in evalu- ating the genetic components of pharmacological response in psychiatric disorders. Abstracts of the 8th Annual Meeting of the International Genetic Epidemiology Society, St. Louis, MO,
USA, abstract 96.
Macciardi F, Basile VS, Kennedy JL (1999). Modeling gene–gene interaction in complex traits. A
J Hum Genet 65(Suppl 4), A210.
McDougle CJ (1997). Update on pharmacologic management of OCD: agents and augmenta-
tion. J Clin Psychiatry 58(Suppl 12), 11–17.
McDougle CJ, Goodman WK, Leckman J, Lee N, Heninger G, Price LH (1994). Haloperidol addi-
tion in fluvoxamine-refractory obsessive-compulsive disorder. Arch Gen Psychiatry 51,
302–308.
McDougle CJ, Epperson CN, Price LH, Gelernter J (1998). Evidence for linkage disequilibrium
between serotonin transporter gene (SLC6A4) and obsessive-compulsive disorder. Mol
Psychiatry 3, 270–273.
McGrath PJ, Stewart JW, Petkova E et al. (2000). Predictors of relapse during fluoxetine
continuation or maintenance treatment of major depression. J Clin Psychiatry 61,
518–524.
Mundo E, Ronchi P, Bellodi L (1993). Obsessive-compulsive patients at risk for antidepressant-
induced mania. Hosp Community Psychiatry 44, 689–690.
Mundo E, Bianchi L, Bellodi L (1997). Efficacy of fluvoxamine, paroxetine and citalopram in the
treatment of obsessive-compulsive disorder: a single-blind study. J Clin Psychopharmacol 17,
267–271.
Mundo E, Walker M, Tims H, Macciardi FM, Kennedy JL (2000). The role of serotonin trans-
porter gene in antidepressant-induced mania in bipolar patients. Biol Psychiatry 47(Suppl 8),
135S.
Mundo E, Walker M, Cate T, Macciardi F, Kennedy JL (2001). The role of serotonin transporter
protein gene in antidepressant-induced mania in bipolar disorder: preliminary findings. Arch Gen Psychiatry 58, 539–544.
298 E. Mundo and J. L. Kennedy
Nestadt G, Lan T, Samuels J et al. (2000). Complex segregation analysis provides compelling evi- dence for major gene underlying obsessive-compulsive disorder and for heterogeneity by sex. Am J Hum Genet 67, 1611–1616.
Nötham MM, Erdmann J, Shimron-Abarbanell D, Propping P (1994). Identification of a genetic variation in the human serotonin 1D beta receptor gene. Biochem Biophys Res Commun 205, 1194–1200.
Ogilvie AD, Battersby S, Bubb VJ et al. (1995). Polymorphism in serotonin transporter gene asso- ciated with susceptibility to major depression. Lancet 347, 731–733.
Peet M (1994). Induction of mania with selective serotonin reuptake inhibitors and tricyclic anti- depressants. Br J Psychiatry 164, 549–550.
Pollock BG, Ferrell RE, Mulsant BH et al. (2000). Allelic variation in the serotonin transporter promoter affects onset of paroxetine treatment response in late-life depression. Neuropsychopharmacology 23, 587–590.
Potter WZ (1998). Bipolar depression: specific treatments. J Clin Psychiatry 59(Suppl): 18, 30–36. Risch N, Merikangas K (1996). The future of genetic studies on complex human diseases. Science
273, 1516–1517.
Schatzberg AF, Rothshild AJ (1992). Psychotic (delusional) major depression: should it be
included as a distinct syndrome in DSM-IV? Am J Psychiatry 49, 733–745.
Smeraldi E, Zanardi R, Benedetti F, Di Bella D, Perez J, Catalano M (1998). Polymorphism within the promoter of the serotonin transporter gene and antidepressant efficacy of fluvoxamine.
Mol Psychiatry 3, 508–511.
Smeraldi E, Benedetti F, Barbini B, Campori E, Colombo C (1999). Sustained antidepressant
effect of sleep deprivation combined with pindolol in bipolar depression: a placebo-controlled
trial. Neuropsychopharmacology 20, 380–385.
Solomon R, Rich CL, Darko DF (1990). Antidepressant treatment and occurrence of mania in
bipolar patients admitted for depression. J Affect Disord 18, 253–257.
Stoll Al, Mayer PV, Kolbrener M et al. (1994). Antidepressant-associated mania: a controlled
comparison with spontaneous mania. Am J Psychiatry 151, 1642–1645.
Tome MB, Cloninger CR, Watson JP, Isaac MT (1997). Serotonergic autoreceptor blockade in the reduction of antidepressant latency: personality variables and response to paroxetine and pin-
dolol. J Affect Disord 44, 101–109.
Vieta E, Bernardo M (1992). Antidepressant-induced mania in obsessive-compulsive disorder.
Am J Psychiatry 149, 1282–1283.
Vincent JB, Masellis M, Lawrence J et al. (1999). Genetic association analysis of serotonin system
genes in bipolar affective disorder. Am J Psychiatry 156, 136–138.
Warren JT Jr., Peacock ML, Rodriguez LC, Fink JK (1993). An MspI polymorphism in the human
serotonin receptor gene (HTR2): detection by DDGE and RFLP analysis. Hum Mol Genet 2,
338.
Wehr TA, Goodwin FK (1987). Can antidepressants cause mania and worsen the course of affec-
tive illness? Am J Psychiatry 144, 1403–1411.
Zanardi R, Franchini L, Gasperini M, Lucca A, Smeraldi E, Perez J (1998). Faster onset of action
of fluvoxamine in combination with pindolol in the treatment of delusional depression: a con- trolled study. J Clin Psychopharmacol 18, 441–446.
299 Alternative phenotypes and pharmacogenetics
Zanardi R, Benedetti F, Di Bella D, Catalano M, Smeraldi E (2000). Efficacy of paroxetine in depression is influenced by a functional polymorphism within the promoter of the serotonin transporter gene. J Clin Psychopharmacol 20, 105–107.
Zhang HY, Ishigaki T, Tani K et al. (1997). Serotonin 2A receptor gene polymorphism in mood disorders. Biol Psychiatry 41, 768–773.
Zohar J, Westenberg HG (2000). Anxiety disorders: a review of tricyclic antidepressants and selective serotonin reuptake inhibitors. Acta Psychiatr Scand 403, 39–49.
14
Pharmacogenetics of anxiolytic drugs and the GABA–benzodiazepine receptor complex
Smita A. Pandit, Spilios V. Argyropoulos, Patrick G. Kehoe, and David J. Nutt
Psychopharmacology Unit, School of Medical Sciences, Bristol, UK
OVERVIEW
The benzodiazepines (BDZ) have proven to be both effective and controversial in the treat- ment of anxiety. Whilst giving rapid relief of anxiety, they have undesirable characteristics, such as the development of tolerance, dependence, withdrawal symptoms, and alcohol potentiation. New insights into the genetics of the BD2 receptor system may lead to the development of new drugs that act on the -aminobutyric acid (GABA) receptor complex and are devoid of the problems associated with the classical BDZs. This pharmacotherapeu- tic approach has gathered impetus following the discovery of the various subunits of the GABA receptor, which are thought to play an important role in the regulation of anxiety and the actions of anxiolytics, and which demonstrate differential brain expression. There are a number of genetic variations in the GABA-A receptor and its subunits that can have an effect on the pharmacology of anxiolytic drugs. The various permutations of the subunits in the composition of these receptors impact on the activity and efficacy of these drugs. A second level of complexity is introduced with the genetic variations in each of these subunits. Such variations can result in amino acid replacements, deletions or imperfect subunit construc- tion at a protein level, thereby resulting in different receptor combinations. This chapter dis- cusses the above with reference to recent evidence from animal and human studies, as well as the implications for future anxiolytic treatment strategies.
Introduction
Clinical anxiety is a common cause of morbidity in the general population (Weiller et al., 1998). The current classification systems of mental disorders, ICD-10 (World Health Organization, 1992) and DSM-IV (American Psychiatric Association, 1994), use a categorical/nosological approach and divide the various anxiety states into specific syndromes, the anxiety disorders. Beyond the subjective distress these syndromes may cause, the importance of the associated social and occupational

300

301 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
malfunctioning, as well as the financial implications, have become all too evident in recent years (Rice and Miller, 1998).
Despite the undoubted success of the existing treatments for anxiety, a substan- tial proportion of patients fail to respond to them. Since their introduction in the 1960s, the benzodiazepines (BDZs) have been the mainstay for the treatment of anxiety. More recently the newer serotonin reuptake inhibitor (SSRI) “antidepress- ants” are used increasingly, and cognitive behavior therapy has become the psycho- logical therapy of choice. The available treatments are not without their problems. Long-term use of BDZs have well-documented difficulties, such as tolerance to the therapeutic effect, potential for abuse and dependence, and discontinuation syn- drome (Lader, 1999). Antidepressants are slow to act and are not devoid of side effects. Psychological therapies are not easily transferred from the research context to clinical practice and properly trained therapists are in short supply, especially in the national health services. Therefore, the quest for effective ways of treating anxiety including novel pharmacological approaches to anxiety remains as press- ing as ever.
Pharmacogenetics is one of the exciting new ideas arising from the recent boom in genetic research. The emerging combination of two strands of scientific inquiry – genetics and neurochemistry – into this new discipline promises to revolutionize our ways of developing and prescribing drugs. The first aim of pharmacogenetics is to study how genetic differences between subjects influence their variability in response to drugs. Furthermore, pharmacogenetics, by means of using specific genetic profiles, aims to be able to predict individual patient’s responses to a par- ticular compound (Roses, 2000). However, unlike the pharmacogenetic studies carried out in schizophrenia with clozapine (Arranz et al., 1995), studies in anxiety disorders with anxiolytic drugs are limited in number.
The genetics that underlie anxiety disorders are of a complex nature. Pathways consist of genes and their products (i.e., proteins), each of which is subject to genetic heterogeneity. Therefore, the functioning of such pathways may be mod- ulated by genetic variation. The most obvious example of this is the role of metab- olism and its associated genetic variation in the pharmacology of anxiolytic drugs. These drugs are metabolized by the P-450 enzyme complex in the liver, which is susceptible to genetic variations. Such genetic variations can result in individuals inheriting genes that influence the rates at which they metabolize drugs and thus the bioavailability of anxiolytics can vary among these individuals. Therefore, pharmacokinetic considerations are as important as pharmacodynamic ones. They are treated separately, in more detail, in Chapter 8.
Another target for a pharmacogenetic approach in anxiety is the pharmacody- namics of anxiolytic drugs. The neurochemistry of anxiety deals mainly with monoamine (serotonin (5-HT), norepinephrine, dopamine) neurotransmission,
302 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
and the GABA – BDZ receptor complex (Sandford et al., 2000). Of the existing anx- iolytics, the SSRIs are thought to exert their action through the 5-HT system, and polymorphisms of the gene (SLC6A4) for the serotonin transporter (5-HTT) have been shown to contribute to the expression of anxiety symptoms (Evans et al., 1997). The 5-HT system is also discussed in more detail elsewhere in this book (Chs. 6, 12, and 13). Low dopamine D2 receptor expression and dopamine reup- take site density in the striatum of patients with social phobia has been reported (Tiihonen et al., 1997; Schneier et al., 2000) and this is, in all likelihood, under genetic control.
In this chapter we will focus on developments in the GABA–BDZ receptor field of research. Recent advances in molecular and neuroimaging techniques have given us greater insight into the structure of this receptor and its properties. Most of our understanding about this receptor and the ligands which act on it is based on studies conducted in animals. Based on our current knowledge of this receptor, future studies can be conducted in humans, focusing on the development of new drugs that could pave the way towards effective treatment of anxiety.
GABA, benzodiazepines, and anxiety
GABA is the main inhibitory neurotransmitter in the central nervous system. Depending on the brain region, 20–50% of all central synapses utilize GABA as their transmitter. Three types of GABA receptor have been identified, but most of the GABA activity in the brain involves the fast-signaling, ligand-gated GABA-A receptors, of which the BDZ site is an integral part. Inhibition of the GABAergic system results in enhanced alertness, reduced sleep, anxiety, restlessness and exag- gerated reactions, even to harmless stimuli (Nutt and Malizia, 2001).
The mechanism of action of the BDZs was only discovered in the 1970s. It was shown that there was a highly specific potentiation of GABA by BDZs (Mohler and Okada, 1977; Squires and Braestrup, 1977). In the decade that followed, the binding site for BDZs in the brain, later shown to be a part of the GABA-A recep- tor complex, was discovered (Hafely, 1978), and the GABA-A receptor complex itself was isolated and sequenced (Schofield et al., 1987). The receptor complex was finally visualized by electron microscopy (Nayeem et al., 1994).
The GABA-A receptor is a protein complex made up of five protein subunits (Tallman and Gallager, 1985; Sieghart, 1989; Nayeem et al., 1994). The subunits are arranged like a rosette around a central pore that spans the cell membrane. When GABA binds with the receptor, there is a conformational change of the complex. This results in the opening of the central pore, a net influx to the cell of chloride ions, hyperpolarization of the cell, and inhibition of the excitability of the cell membrane. Other compounds such as barbiturates, anesthetics, alcohol, and

303 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
GABA Antagonists anxiety (bicuculline)
↑ anxiety
  
Agonists (benzodiazepines) ↓ anxiety
Inverse agonists (-carbolines) ↑ anxiety
Benzodiazepine site
GABA-A ++
Cl– channel
Barbiturate
Agonists (barbiturates) ↓ anxiety
Ethanol and chloral hydrate ↓ anxiety
     
Picrotoxin
Neurosteroid site
 
Fig. 14.1.
The -aminobutyric acid (GABA)–benzodiazepine receptor chloride ionophore complex.
neurosteroids also bind to the complex and directly open the chloride channels (Nutt and Malizia, 2001) (Fig. 14.1). BDZs, however, act by allosteric modulation of the GABA-A recognition site, thereby augmenting or diminishing its inhibitory effects without directly opening the chloride channel. This action of BDZs results in their distinct advantage over other drugs like barbiturates, chloral hydrate, and clormethiazole, all of which can, in high concentrations, cause sustained tonic opening of chloride ion channels, which in an overdose can result in respiratory depression and death.
Pentylenetetrazol ↑ anxiety
Pregnenalone ↓ anxiety
Molecular biology of the GABA-A receptor
Molecular cloning studies have shown that the five subunits making up the GABA- A receptor are assembled from a family of at least 17 subunits. Each subunit is encoded by a different gene in mammals, and the complementary DNA (CDNA) and amino acid sequences have been characterized at a molecular level (Doble, 1999). Seven classes of subunit, with multiple isoforms, have been identified by their sequence homology (1–6, 1–3, 1–3, , , , 1,2) (Fritschy and Mohler, 1995;

↑
304 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
Mehta and Ticku, 1999; Whiting, 1999). Each subunit consists of four transmem- brane domains; a long N-terminal exocytic domain with various glycosylation sites and to which ligands probably bind (Cockroft et al., 1990), a short C-terminal exo- cytic chain, and a intracellular loop susceptible to protein kinases. Amino acid sequence homology varies between different subunits from 30% to as much as 70% if they are of the same class.
In situ hybridization histochemistry experiments have shown that there is a var- iable distribution of GABA-A receptor subunits and that they can be quantified by the distribution of their mRNAs (Zhang et al., 1991a, b; Persohn et al., 1992; Wisden et al., 1992). For example, while 1- and 6-subunits are present predom- inantly in the cerebellum the 5-subunit appears to be present only in the hippo- campus, although in lower concentrations than the 1- and 2-subunits.The cerebral cortex has intermediate levels of 1–4–subunits and low levels of 5- subunit. The expression of the -subunits is weaker in the substantia nigra (Sieghart et al., 1987; Sequier et al., 1988; Luddens et al., 1990; McLennan et al., 1991; Malizia and Nutt, 1995). Of the -subunits, the 1-subunit mRNA is pre- dominantly expressed in the amygdala and septum and is probably the only one with this kind of distribution. The 2-subunit, however, is present in abundance in almost every region of the brain while the 3-subunit is present in the cortex and the basal nuclei (Ymer et al., 1990; Wisden et al., 1992; Wang et al., 1998).
Immunoprecipitation studies have shown that the most commonly found sub- units in the brain are 1, 2, and 2 and these are present in 60–70% of GABA-A receptors (Benke et al., 1991a, 1994; Duggan et al., 1992; Ruano et al., 1994), while 2-, 3-, 5-, 3-, and -subunits are moderately present, each being found in 15–30% of GABA-A receptors (Duggan and Stephenson, 1990; Benke et al., 1991b, 1994; McKernan et al., 1991; Endo and Olsen, 1993; Marksitzer et al., 1993; Mertens et al., 1993). From the remaining subunits (i.e., 4, 6, 1, 1 and 3), each is rep- resented in only about 10% of GABA-A receptors (Fritschy et al., 1993; Quirk et al., 1994; Togel et al., 1994). The majority of GABA-A receptors contain at least one 1- subunit variant along with 2-, 3- and 2-subunits (Fritschy and Mohler, 1995); the commonest permutation being 122, which is found in 67% of receptors (Fritschy et al., 1992). Interestingly, the genes encoding for this particular combi- nation of subunits are themselves located as a cluster of genes, all on chromosome 5 in humans and on chromosome 11 in mouse (Table 14.1). This begged the ques- tion whether there are certain proteins that preferentially help cluster specific sub- units together? The proteins rapsyn (Yang et al., 1997) and gephrin (Essrich et al., 1998) were found to cluster some but not all GABA-A receptors. The amino acid sequence on subunits mediating clustering by rapsyn may be used in future to isolate proteins responsible for clustering of native receptors in the brain (Kardos, 1999). This also opens up new ways to look at the postsynaptic receptor structures
305 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
Table 14.1. Chromosomal localization of the genes that encode the various subunits of the -aminobutyric acid GABA-A receptor in humans
Subunit Human chromosome
. 1 5q32–5q33
. 2 4p13–4q11
. 3 Xq28
. 4 4p13–4q11
. 5 15q11–15q13
. 6 5q32–5q33
. 1 4p13–4q11
. 2 5q32–5q33
. 3 15q11–15q13
. 1 4p13–4q11
. 2 5q32–5q33
. 3 15q11–15q13
Source: Adapted with permission from Doble, 1999.
that might use a single clustering protein and to construct different inhibitory and/or excitatory neurotransmitter-gated ion channels enabling the postsynaptic membrane to receive signals from different neurotransmitters (Jonas et al., 1998; Nicoll and Malenka, 1998; Kardos, 1999).
However, gene clustering is not the only factor that controls the assembly of the subunits in a specific receptor. For example, even though 1- and 6-subunits are colocalized in the same gene cluster on chromosome 5, the 6-subunit is only expressed in the cerebellum while the 1-subunit is present throughout the CNS. Therefore, there is some level of control on gene expression (Fritschy and Mohler, 1995), which suggests that the expression of GABA-A receptors may change in response to physiological or pathophysiological stimuli. This is supported by evi- dence that there is an alteration in GABA-A receptor expression in response to behavioral or pharmacological challenges such as stress, chronic alcohol adminis- tration, or visual deprivation (Mhatre et al., 1993; Huntsman et al., 1994; Mhatre and Ticku, 1994). This also opens up an interesting area for further research into the etiology of anxiety disorders and, more specifically, personality factors asso- ciated with the development of anxiety disorders. The expression of the various subunits may be affected by chronic administration of BDZs, thus leading to the phenomenon of dependence and tolerance. In generalized anxiety disorder, there is a tolerance of the sedative effect without the development of tolerance to the anx- iolytic effect (Doble and Martin, 1996) and this could be a result of the differential
  
306 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
expression of the subunits in different areas of the brain during chronic treatment with BDZs.
Putative functions of the GABA-A receptor subunits
GABA binds to the -subunit. The BDZ site is localized at the interface of the - and -subunits of the GABA-A receptor (Wong et al., 1992; Zezula et al., 1996), however, when hypnotics like zaleplon, zopiclone, and zolpidem and anxiolytic and anticonvulsant BDZs (as well as BDZ antagonists and inverse agonists) bind to the receptor, the actual binding is at the -subunit (Sieghart, 1995; Johnstone, 1996). The -subunit is essential to allow classical BZDs to bind to the receptor, but it is the 2-variant that has an influence on their pharmacological profiles (Pritchet et al., 1989a; Angelotti and MacDonald, 1993). However, when the 1- and 3- subunits are present, they result in receptors that have a low affinity for BDZs. GABA-gated chloride conductance is more enhanced when at least one - and one -subunit are present (Siegel et al., 1990; Verdoorn et al., 1990).
The functional properties of the receptors, including their sensitivity for GABA, degree of rectification and rate of desensitization, vary with the subunit combina- tions (Pritchett et al., 1989b; Saxena and MacDonald 1994; Verdoorn, 1994; Fritschy and Mohler, 1995). Thus different subtypes of GABA-A receptors have different sensitivities to BDZ receptor ligands (Doble, 1999). Initial attempts to classify GABA-A/BDZ receptors suggested the presence of two different types of receptor: type I receptors, which are present mainly in the cerebellum and are formed by the coexpression of 1-, 1, 3-, and 2-subunits, and type II receptors, which are present mainly in the cortical structures and are formed by the coexpression of 2, 3, 5-, 1,3-, and 2-subunits (Pritchett et al., 1989b). The receptors carrying an 4-subunit have low affinity for classical BDZs like diazepam, while those carrying 2- and 3-subunits have a higher affinity for zolpidem. The 6-subunit-containing receptor, which is exclusively expressed in cerebellar granule cells, is insensitive to all hypnotic drugs except Ro-154513 (Luddens et al., 1990; Pritchett and Seeburg, 1990; Wisden et al., 1991).
Smith et al. (1998) were able to show that the blockade of the GABA-A receptor 4-subunit gene transcript prevented withdrawal properties of the endogenous neuroactive steroid allopregnanolone, which caused an increased susceptibility to seizures. Increased expression of 4-subunit was seen in the hippocampus and cere- bral cortex after chronic ethanol consumption in rats (Mathews et al., 1998). An understanding of the physiological role of the subunits and the changes in the expression of receptors in pathological conditions will determine which subunit- selective drugs may be developed to treat certain conditions (Kardos, 1999; Whiting, 1999).
307 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
Identification of the role of GABA-A subunits in anxiety, through molecular manipulation
Several approaches, including gene knockout (gene deletion) and gene knockin (gene mutation), have been used to try to establish the role of various subunits and the resulting receptor variants (Mehta and Ticku, 1999). Experiments with knock- out mice involve the prevention of expression of specific subunit genes in order to produce GABA-A receptors that are deficient in specific subunits. One of the first such experiments involved the gene for the 2-subunit (Gunther et al., 1995). Homozygous 2-subunit knockout mice, with both alleles knocked-out, were found to be not viable. By comparison, heterozygotes, with one active allele of the gene, survived to adulthood and bred. Interestingly, the GABA-A–BDZ receptors in these heterozygotes have only half the usual complement of 2-subunits and are less sensitive to BDZs like flunitrazepam and diazepam. The mice also exhibited symptoms of hypervigilance and anxiety. Therefore, this approach produces a system that can be seen as a genetically defined model of trait anxiety which repro- duces molecular, pharmacological, and behavioral features akin to human anxiety disorders (Nutt and Malizia, 2001). Further, these mice exhibited decreased flu- mazenil binding throughout the brain, which was similar to the decreases shown in flumazenil binding found in panic disorder in humans (Malizia et al., 1998; Crestani et al., 1999). These findings also suggest that the GABA-A receptor dys- function could be a causative factor for heightened harm avoidance behavior and a hypersensitivity to negative associations in patients (Crestani et al., 1999).
The knockout technique has been used to produce mice deficient in other types of subunit as well. When the gene for the 3-subunit is knocked out, hyperactive mice demonstrating poor coordination, abnormal reflexes, and spontaneous sei- zures are produced (Homanics et al., 1997). The receptors in these mice have an attenuated response to GABA. Their pharmacological response to loreclazole indi- cate that the 3-subunit has been replaced by a 2-subunit (Krasowski et al., 1998). Reduced sensitivity to anesthetic agents, like halothane and midazolam, was also observed, thus linking these receptors with the effects of these drugs.
Knocking out the gene encoding the 6-subunit produces mice that appear normal. However, there is a complete absence of BDZ-insensitive GABA-A recep- tors in the cerebellum. It also suppresses the expression of -subunit. The 6- subunit was earlier implicated in the mediation of some of the CNS-depressant effects of alcohol (Ticku et al., 1988; Luddens et al., 1990). However, subsequent research has shown that the deletion of the 6-encoding gene does not lead to any changes in sensitivity to alcohol in animals, nor does it have a role to play in the development of alcohol tolerance (Homanics et al., 1998). Despite the above, it has recently been reported that male children of male alcoholics have an increased like- lihood of possessing a genetic variant of the 6-subunit, which may explain their inherited subsensitivity to alcohol and BDZs (Schukit et al., 1999). Recent studies
308 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
have shown that RO-154513 is a good ligand for characterization of 6-containing- receptors in rats and humans by positron emission tomography (Feeney et al., 2000). The use of neuroimaging techniques can thus assist in delineating the role of subunits in humans.
More recently, the knockin technique has been used to produce mice where the 1-subunit, following mutation, has become insensitive to BDZs but is still respon- sive to GABA (see next section). This results in the abolition of the sedative actions of BDZs, with preservation of the anxiolytic and anticonvulsant properties (Rudolph et al., 1999). Subsequent mutation of the two remaining -subunits has shown that the anxiolytic effect is also lost if the 2- but not the 3-subunit is mutated (Low et al., 2000). The 2-subunit is mainly localized to the limbic system. Therefore, the above finding further supports the role of this circuit in anxiety. The growing evidence for the specificity of the various BDZ actions being mediated via different receptor subtypes/subunit clusters is currently driving research into subunit-selective drugs, such as 2- or 3-agonists, as nonsedating anxiolytics and 5 inverse agonists (which act mainly in the hippocampus) as memory enhancers.
Studies of mutation of individual amino acids
Although BDZs bind to the -subunit, photoaffinity labeling experiments led to the hypothesis that this site may carry different binding domains for different chemical classes of drug (Mohler et al., 1980; Mohler, 1982). Other experiments evaluated the effect of treatment of GABA-A receptors with chemical reagents that would irreversibly modify particular amino acid groups on the receptor surface, and the residue histidine in the binding domain of the BDZs was incriminated (Doble et al., 1992). Based on this, a map of the GABA-A receptor surface with a pocket containing an histidyl group was proposed (Doble et al., 1995). In this model, it is proposed that the BDZs bind to the bottom of the pocket whereas zop- iclone, as well as other cyclopyrrolones and probably flumazenil, bind to the rim of the pocket. If drugs bind to different domains on the receptor surface, this may have consequences for the different sorts of conformational change that the receptor can assume upon activation (Doble, 1999).
In an attempt to characterize the molecular determinants of clinically important drug targets, studies involving site-directed mutagenesis of the genes of individual amino acid residues in the receptor have identified essential residues in the binding of different molecules (Table 14.2). Tyrosine (Tyr)157, Threonine (Thr)160, Thr202 and Tyr205 amino acid residues of the -subunit and phenylalanine (Phe)64 of the 1-subunit are important for the binding of GABA in rats. Histidine (His)101 (His102 in humans), and glycine (Gly)200 (Gly201 in humans) of the - subunit and Phe77, methionine (Met)130 and Thr142 of the -subunit are reported to be the key determinants for BDZ binding in the rat brain. Two tyrosine residues, Tyr159 and Tyr209, are also crucial for BDZ binding (Amin et al., 1997). The His102
309 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
Table 14.2. Site-directed mutagenesis of the - and -subunits

Amino acid Histidine 101
Tyrosine 159 Glycine 200 Phenylalanine 77
Methionine 130 Threonine 142
Subunit Effect
Site of photolabeling; indispensable for benzodiazepine binding; irrelevant for flumazenil and zopiclone
binding
Indispensable for benzodiazepine binding
Confers BZ1/BZ2 selectivity
Indispensable for zolpidem and flumazenil binding;
irrelevant for benzodiazepine binding
Indispensable for flumazenil and benzodiazepine binding
Irrelevant for benzodiazepine binding; important determinant of efficacy
 
Source: Adapted with permission from Doble, 1999.
residue of the -subunit interacts directly with the phenyl group of diazepam, flu- nitrazepam, chlordiazepoxide, and other 5-phenyl BDZs. This His residue is absent in the 4-subunit and is replaced by arginine (Arg)100 in the 6-subunit of the human and rat GABA-A receptors. These changes make the GABA-A receptors con- taining 4- and 6-subunits insensitive to conventional BDZs like diazepam, fluni- trazepam, and clonazepam. The replacement of Arg100 by the glutamine (Gln)100 in the 6-subunit of alcohol-sensitive rat cerebellum alters the normal diazepam- insensitive GABA-A receptors into diazepam-sensitive ones. Mutations in the 2- subunit also affect BDZ activity. Thr142 determines efficacy (Mihic et al., 1994) and its mutation to serine increases the efficacy of BDZs and flumazenil but decreases that of imidazopyridines. These results are consistent with the idea that the BDZ binding pocket is in the interface between the - and -subunits (Doble, 1999).
In the rat, variation in alcohol and BDZ sensitivity has been correlated with an inherited variant of the GABA-A 6-receptor. The Pro385Ser genotype was impli- cated and studied in humans by studying saccadic eye movements in children of alcoholics (Iwata et al., 1999). They found that this genotype may play a role in BDZ sensitivity and conditions such as alcoholism and may be correlated with this trait. Similar studies in humans, translating results of animal studies, need to be carried out.
Possible interventions at the GAB-A receptor set-point
The BDZ receptor site is unique in having three different functional classes of ligand binding to it: agonists, inverse agonists, and antagonists. The current treatment of
310 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
            
agonist
                   
agonist
                    
Fig. 14.2.
The -aminobutyric acid (GABA)–benzodiazepine receptor set-point.
anxiety disorders involves the use of BDZs, which are full agonists at most GABA-A receptor subtypes. The inverse agonists decrease the probability of the chloride channel opening in response to GABA and have stimulant and proconvulsant prop- erties. The antagonists (like flumazenil) block the activities of both agonists and inverse agonists. These features indicate that the receptor manifests bidirectional agonism.
The GABA-A–BDZ receptor site has been hypothesized to play a regulatory role so that a protein conformation “fine tunes” GABA function, altering maximal effi- cacy or the rate of desensitization. The BDZ receptor spectrum is not fixed and the “set-point,” where drugs bind, can be altered. Alterations in the set-point have been suggested as a possible mechanism to explain the phenomena of tolerance and dependence seen with BDZ treatment and rebound anxiety and insomnia on their withdrawal. It has been hypothesized that, with prolonged use, the set-point drifts in the inverse agonist direction, thus leading to a change in the receptor sensitivity and subsequent development of the above effects (Little et al., 1988; Nutt et al., 1992) (Fig. 14.2). This change in sensitivity can be explained by exchange of one subunit for another within the receptor. A model has been proposed according to which, with chronic BDZ treatment, the transcription of the gene cluster that encodes the prevalent 1, 2,3-receptor on chromosome 5 is downregulated while transcription of subunits encoded on chromosome 15 are upregulated; as a result the rarer subunits of chromosome 15 replace those of chromosome 5, leading to reduced sensitivity of the receptor (Holt et al., 1996, 1997a). This postulated change of the transcription rates is supported by the repression of the activity of the pro- moter sequence of the 1-subunit gene seen after chronic treatment with loraze- pam (Kang et al., 1994) and the decreased rate of transcription of the 2-subunit seen in diazepam-treated animals (Holt et al., 1997b).
This theory led to the development of variety of new drugs such as abecarnil (Holt et al., 1996), zopiclone, zolpidem (Holt et al., 1997a), and imidazenil (Pesold et al., 1997), which produce more restricted changes in subunit transcription.
311 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
These drugs also have less dependence potential compared with the classical BD2s (Lader, 1997, 1998; Doble, 1999). Also, chronic treatment with the inverse agonist FG7142 leads to opposite changes in GABA-A receptor subunit expression (Primus and Gallager, 1992). This postulated mechanism of transcription changes can thus provide a potentially useful animal model in which experimental treatment regi- mens can be identified to reset the receptor set-point in order to forestall the appearance of or attenuate or abolish established dependence (Doble and Martin, 1996). In epilepsy, studies have been conducted to ‘reset the set-point’ with the antagonist flumazenil. Such an approach reverses the tolerance to the anticonvul- sant effect of clonazepam. While it is too early to say the same about BDZ depend- ence, studies in this area are needed.
Another strategy to prevent the development of dependence and tolerance is to initiate “drug holidays” (Doble and Martin, 1996). This may lead to changes in the set-point as well as changes in the expression of subunits. This hypothesis also requires further investigation, both in animals and humans.
Drug development: subtype selective ligands
The GABA-A receptor shows marked heterogeneity, and specific GABA-A recep- tors in particular brain regions with specific subunit assemblies are expressed to serve specific functions. Based on this hypothesis, subtype-selective ligands are being developed in the hope that they may have more precisely targeted therapeu- tic activity and fewer side effects. The major drawback of this attractive approach is that the physiological and pharmacological profiles of some relevant receptor assemblies are as yet unknown. Once this information becomes available, it would be theoretically possible to synthesize compounds selective for a particular recep- tor assembly so as to get a desired therapeutic effect, while the undesirable effects could be minimized.
Based on our current knowledge of the physiological and pharmacological role of various subunits of GABA-A receptors, the pharmacodynamic action of drugs like triazolopyridazine (CL218872), zolpidem, and abecarnil can be surmised. These drugs have been shown to bind with different affinities at different GABA-A receptors. CL218 872 has a significantly higher affinity for the BDZ type I receptors than for type II receptors. It is thought to be anxiolytic without being sedative. Since type I receptors are present in the cerebellum, and type II receptors are present in the hippocampus, it was hypothesized that anxiolysis was mediated in the cerebellum. However, subsequent research does not support this. Zolpidem has a higher affinity for 122-receptors. It has a 20-fold reduced affinity for the type II 222-receptors and 333-receptor subtypes but not to the type II 522- receptor (Pritchett and Seeburg, 1990). Zolpidem produces sedation at lower doses than those producing anxiolysis, muscle relaxation, and anticonvulsant effects

312 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
(Darcourt et al., 1999). Abecarnil is a partial agonist at the GABA-A 522-recep- tor and a full agonist at 323-receptor. This is reflected in its therapeutic and side effect profile. Animal work with abecarnil has shown that it is an effective anxio- lytic and anticonvulsant, with weak or insignificant effects on motor incoordina- tion and a low propensity to cause dependence (Stephens et al., 1990). Further, it has no amnesic effects (Ozawa et al., 1994). In some (but not all) human studies, abecarnil was significantly better than placebo in the treatment of generalized anxiety disorder, and did not produce withdrawal effects upon abrupt discontinu- ation (Aufdembrinke, 1998).
Neurosteroids
Besides the GABA and BDZ sites, there are at least three other domains with binding sites on the GABA-A receptor: for picrotoxin, barbiturates, and anesthetic steroids (Gee et al., 1988; Turner et al., 1989). Unlike the BDZ site, the modulation of the GABA receptor by neurosteroids requires the presence of a -subunit instead of a -subunit (Puia et al., 1990). The substitution of a 1- by 2-subunit greatly enhances the sensitivity of neuroactive steroids (Puia et al., 1991, 1993). The - and -subunits inhibit neurosteroid modulation of GABA-A receptors.
The natural steroid hormone progesterone is converted in the brain to 3- hydroxy-5-pregnan-20-one (allopregnenolone) (Morrow et al., 1987), which is sedative, decreases anxiety, and controls seizure activity by enhancing GABA func- tion (Holzbauer, 1976; Crawley et al., 1986). During the premenstrual period, when the circulating levels of progesterone fall sharply, anxiety symptoms and sus- ceptibility to seizures is known to occur. In an elegant experiment, Smith et al. (1998) produced a progesterone-withdrawal model in rats. Falling levels of the hormone led to insensitivity to BDZs and increased seizure susceptibility. These effects were mediated through the corresponding falling levels of allopregnenolone, which enhanced the transcription of the 4-subunit. Blockade of the transcription of this subunit prevented these phenomena.
Ganaloxone is a synthetic neurosteroid with improved bioavailability compared with allopregnenolone. It may provide control of seizures that is superior to that of valproate, clonazepam, ethosuximide or diazepam, and it is currently undergoing clinical trials. Animal studies show that it may provide additional benefits by con- trolling anxiety, and mood changes associated with preseizure activity (Carter et al., 1997; Beekman et al., 1998).
Future research
There is a potential for further research into the development of compounds that effectively differentiate between the receptor subtypes of GABA-A receptor and
 
313 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
target different subunits. Furthermore, research looking at modulation of differen- tial expression in receptors, the specific physiological role of each subunit, the mechanisms influencing the clustering of subunit genes and their expression, and the different binding domains of receptors could lead to development of ligands that are effective anxiolytics.
Conclusions
Our understanding of the neurobiology of anxiety disorders and their treatment has been considerably increased in recent years following advances in neurochem- istry and neuroimaging. The GABA-A receptor has been shown to play a pivotal role in mediating anxiety. There are a number of genetic variations in the GABA- A receptor and its subunits that can impact on the pharmacology of anxiolytic drugs. The various permutations of the subunits in the composition of these recep- tors have an effect on the activity and efficacy of these drugs. A second level of com- plexity is introduced with genetic variations in each of these subunits. Such variations can result in amino acid replacements, deletions, and imperfect subunit construction at a protein level, thereby resulting in different receptor combina- tions. Any such variations can strongly interfere with the pharmacokinetics of anx- iolytic drugs. Developments in genetics and molecular biology have opened a window into the role of the GABA-A receptor subunits and the consequences of the heterogeneity of GABA-A receptors in different parts of the brain.
Exciting new research into pharmacogenetics will be based on our improving understanding of different drugs acting at this receptor. Future insights and research would pave the way for designing effective anxiolytic drugs with better safety profile and fewer side effects than the existing ones. If the pharmacogenetic approach were to prove clinically useful in the future, it will also enable us to predict the responsiveness of different patient populations to specific drugs. Then the vague recommendation of old to “tailor” each treatment regimen to the individual patient concerned may, finally, acquire a true meaning.
REFERENCES
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders, 4th edn. Washington, DC: American Psychiatric Association.
Amin J, Brooks-Kayal A, Weiss DS (1997). Two tyrosine residues on the alpha subunit are crucial for benzodiazepine binding and allosteric modulation of gamma-aminobutyric acid-A recep- tors. Mol Pharmacol 51, 833–841.
Angelotti T, MacDonald RL (1993). Assembly of GABA-A receptor subunits 11 and 112 subunits produce unique ion channels with dissimilar single-channel properties. J Neurosci 13, 1429–1440.
 
314 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
Arranz MJ, Collier D, Sodhi M et al. (1995). Association between clozapine response and allelic variation in 5-HT2A receptor gene. Lancet 346, 281–282.
Aufdembrinke B (1998). Abercanil, a new beta-carboline, in the treatment of anxiety disorders. Br J Psychiatry 173(Suppl 34), 55–63.
Beekman M, Ungard JT, Gasior M et al. (1998). Reversal of behavioural effects of pentylenetet- razol by the neuroactive steroid ganoxolone. J Pharmacol Exp Ther 284, 868–877.
Benke DA, Mertens S, Trzeciak A, Gillessen D, Mohler H (1991a). GABA-A receptor display asso- ciation of 2-subunit with 1 and 2/3 subunits. J Biol Chem 266, 4478–4483.
Benke DA, Cicen-Sain A, Mertens S, Mohler H (1991b). Immunochemical identification of the 1 and 3 subunits of the GABA-A receptor in the rat brain. J Receptor Res 11, 407–424.
Benke D, Fritschy JM, Trzeciak A, Bannwarth W, Mohler H (1994). Distribution, prevalence and drug binding profile of gamma aminobutyric acid type A receptor subtypes differing in the beta subunit variant. J Biol Chem 269, 27100–27107.
Carter RB, Wood PL, Wieland S et al. (1997). Characterisation of the anticonvulsant properties of ganoloxone (CCD1042; 3-hydroxy-3beta-methyl-5alpha-pregnan-20-one), a selective high-affinity steroid modulator of the gamma-aminobutyric acid-A receptor. J Pharmacol Exp Ther 280, 1284–1295.
Cockroft VB, Osguthorpe DJ, Barnard EA, Lunt GG (1990). Modelling of agonist binding to the ligand gated ion channel superfamily of receptors. Protein Struct Funct Genet 8, 386G–397G.
Crawley JN, Glowa JR, Majewska MD, Paul SM (1986). Anxiolytic activity of an endogenous adrenal steroid. Brain Res 398, 382–385.
Crestani F, Lorez M, Baer K et al. (1999). Decreased GABA-A receptor clustering results in enhanced anxiety and a bias for threat cues. Nat Neurosci 2, 833–839.
Darcourt G, Pringuey D, Sallière D, Lavoisy J (1999). The safety and tolerability of zolpidem – an update. J Psychopharmacol 13, 81–93.
Doble A (1999). New insights into the mechanism of action of hypnotics. J Psychopharmacol 13 (Suppl 4:1), S11–S20.
Doble A, Martin IL (1996). The GABA/Benzodiazepine Receptor as a Target for Psychoactive Drugs. Austin, TX: RG Landes Company.
Doble A, Canton T, Piot O et al. (1992). The pharmacology of cyclopyrrolone derivatives acting at the GABA-A benzodiazepine receptor. Adv Biochem Psychopharmacol 47, 407–418.
Doble A, Canton T, Malgouris P (1995). The mechanism of action of zopiclone. Eur Psychiatry 10 (Suppl 3), 117S-128S.
Duggan M, Stephenson FA (1990). Biochemical evidence for the existence of gamma-aminobu- tyrate-A receptor iso-oligomers. J Biol Chem 265, 3831–3835.
Duggan M, Pollard S, Stephenson FA (1992). Quantitative immunoprecipitation studies with anti-gammaaminobutyric acid A receptor 2 1–15 Cys antibodies. J Neurochem 58, 72–77.
Endo S, Olsen RW (1993). Antibodies specific for -subunit subtypes of GABA-A receptors reveals brain heterogeneity. J Neurochem 60, 1388–1398.
Essrich C, Lorez M, Benson JA, Fritschy JM, Luscher B (1998). Postsynaptic clustering of major GABA-A receptor subtypes requires the 2 subunit and gephrin. Nat Neurosci 1, 563–571.
315 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
Evans J, Battersby S, Ogilvie AD et al. (1997). Association of short alleles of a VNTR of the sero- tonin transporter gene with anxiety symptoms in patients presenting after deliberate self harm. Neuropharmacology 36, 439–443.
Feeney A, Lingford-Hughes A, Hume S et al. (2000). Characterisation of 11C- RO 154513 as a PET ligand for limbic GABA-benzodiazepine receptors. J Psychopharmacol 14(Suppl), A30, P30. Fritschy JM, Mohler H (1995). GABA-A receptor heterogeneity in the adult rat brain: differen- tial regional and cellular distribution of seven major subunits. J Comp Neurol 359, 154–194.
Fritschy JM, Benke D, Mertens S, Oertel WH, Bachi T, Mohler H (1992). Five subtypes of type A gammaaminobutyric acid receptors identified in neurons by double and triple immunofluo- rescence staining with subunit specific antibodies. Proc Natl Acad Sci, USA 89, 6726–6730.
Fritschy JM, Benke D, Mertens S, Gao B, Mohler H (1993). Immunochemical distinction of GABA-A receptor subtypes differing in drug binding profiles and cellular distribution. Soc Neurosci Abst 19, 476.
Gee KW, Bolger MB, Brinton RE, Coirini H, McEwen BS (1988). Steroid modulation of the chlo- ride ionophore in rat brain: structure–activity requirements, regional dependence and mech- anism of action. J Pharmacol Exp Ther 246, 803–812.
Gunther U, Benson J, Benke D et al. (1995). Benzodiazepine-insensitive mice generated by tar- geted disruption of the gamma2 subunit of gamma-aminobutyric acid type A receptors. Proc Natl Acad Sci, USA 92, 7749–7753.
Hafeley WE (1978). Central action of the benzodiazepines: general introduction. Br J Psychiatry 133, 231–238.
Holt RA, Bateson AN, Martin IL (1996). Chronic treatment with diazepam or abecarnil differen- tially affects the expression of GABA-A receptor subunit mRNAs in the rat cortex. Neuropharmacology 35, 1457–1463.
Holt RA, Bateson AN, Martin IL (1997a). Chronic zolpidem treatment alters GABA-A receptor mRNA levels in the rat cortex. Eur J Pharmacol 329, 129–132.
Holt RA, Martin IL, Bateson AN (1997b). Chronic diazepam exposure decreases transcription of the rat GABA-A receptor gamma-2 subunit gene. Mol Brain Res 48, 164–166.
Holzbauer M (1976). Physiological aspects of steroids with anaesthetic properties. Med Biol 22, 97–102.
Homanics GE, Ferguson C, Quinlan JJ et al. (1997). Gene knockout of the 6 subunit of the GABA type A receptor: lack of effect on responses to ethanol, phenobarbital, and general anaesthetics. Mol Pharmacol 51, 588–596.
Homanics GE, Le NQ, Kist F, Mihalek R, Hart AR, Quinlan JJ (1998). Ethanol tolerance and with- drawal responses in GABA-A receptor alpha 6 subunit null allele mice and in inbred C57BL/6J and strain 129/SvJ mice. Alcohol Clin Expl Res 22, 259–265.
Huntsman NM, Isackson PJ, Jones EG (1994). Lamina-specific expression and activity-depen- dent regulation of seven G receptor subunit mRNAs in monkey visual cortex. J Neurosci 14, 2236–2259.
Iwata N, Cowley DS, Radel M, Roy-Byrne PP, Goldman D (1999). Relationship between a GABAA alpha6 Pro385Ser substitution and benzodiazepine sensitivity. Am J Psychiatry 156, 1447–1449.
Johnstone GA (1996). GABA-A receptor pharmacology. Pharmacol Ther 69, 173–198.
316 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
Jonas P, Bischofberger J, Sandkuhler J (1998). Corelease of two fast neurotransmitters at a central synapse. Science 281, 419–424.
Kang I, Lindquist DG, Kinane TB, Ercolani L, Pritchard GA, Miller LG (1994). Isolation and char- acterization of the promoter of the human GABA-A receptor 1 subunit gene. J Neurochem 62, 1643–1646.
Kardos J (1999). Recent advances in GABA research. Neurochem Int 34, 353–358.
Krasowski MD, Harrison NL, Firestone LL, Homanics GE (1998). A deficit of functional GABA-
A receptors in neurons of beta3 subunit knockout mice. Neurosc Lett 240, 81–84.
Lader M (1997). Zopiclone: is there any dependence and abuse potential? J Neurol 244(Suppl 1),
S18–S22.
Lader M (1998). Withdrawal reactions after stopping hypnotics in patients with insomnia. CNS
Drugs 10, 425–440.
Lader MH (1999). Limitations on the use of benzodiazepines in anxiety and insomnia: are they
justified? Eur Neuropsychopharmacol 9(Suppl 6), S399–S405.
Little HT, Gale R, Sellers N, Nutt DJ, Taylor SC (1988). Chronic benzodiazepine treatment
increases the effects of the inverse agonist FG7142. Neuropharmacology 27, 381–383.
Low K, Crestani F, Keist R et al. (2000). Molecular and neuronal substrate for the selective atten-
uation of anxiety. Science 290, 131–134.
Luddens H, Pritcett DB, Kohler M et al. (1990). Cerebellar GABA-A receptor selective for a
behavioural alcohol antagonist. Nature 346, 648–651.
Malizia AL, Nutt DJ (1995). Psychopharmacology of benzodiazepines – an update. Hum
Psychopharmacol 1, S1–S14.
Malizia AL, Cunningham VJ, Bell CJ, Liddle PF, Jones T, Nutt DJ (1998). Decreased brain GABA-
A-benzodiazepine receptor binding in panic disorder. Preliminary results from a quantitative
PET study. Arch Gen Psychiatry 55, 715–720.
Marksitzer R, Benke D, Fritschy JM, Trzeciak A, Bannwarth W, Mohler H (1993). GABAA-recep-
tors: drug binding profile and distribution of receptors con alpha 2-subunit in situ. J Receptor
Res 13, 467–477.
Mathews DB, Devaud LL, Fritschy JM, Sieghart W, Morrow AL (1998). Differential regulation of
GABA-A receptor gene expression by ethanol in the rat hippocampus versus cerebral cortex. J
Neurochem 70, 1160–1166.
McKernan RM, Quirk K, Prince R et al. (1991). GABA-A receptor subtypes immunopurified
from rat brain with subunit specific antibodies have unique pharmacological properties.
Neuron 7, 667–676.
McLennan A, Brecha N, Khrestchatisky M et al. (1991). Independent cellular and ontogenic
expression of mRNA encoding alpha polypeptides of the rat GABA-A receptor. Neuroscience
43, 369–380.
Mehta AK, Ticku MK (1999). An update on GABA-A receptors. Brain Res Rev 29, 196–217. Mertens S, Benke D, Mohler H (1993). GABA-A receptor populations with novel subunit com-
binations and drug binding profiles identified in brain by 5 and -subunit specific immuno-
purification. J Biol Chem 268, 5965–5973.
Mhatre MC, Ticku MK (1994). Chronic ethanol treatment up-regulates the GABA-A receptor
subunit expression. Mol Brain Res 23, 246–252.
317 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
Mhatre MC, Pena G, Sieghart W, Ticku MK (1993). Antibodies specific for GABA-A receptor subunits reveal that chronic alcohol treatment down-regulates -subunit expression in rat brain regions. J Neurochem 61, 1620–1625.
Mihic SJ, Whiting PJ, Klein RL, Wafford KA, Harris RA (1994). A single amino acid of the human gamma-aminobutyric acid type A receptor gamma 2 subunit determines benzodiazepine effi- cacy. J Biol Chem 269, 32768–32773.
Mohler H (1982). Benzodiazepine receptors: differential interaction of benzodiazepine agonists and antagonists after photoaffinity labeling with flunitrazepam. Eur J Pharmacol 80, 435–436. Mohler H, Okada T (1977). Benzodiazepine receptor: demonstration in the central nervous
system. Science 198, 849–851.
Mohler H, Battersby MK, Richards JG (1980). Benzodiazepine receptor protein identified and
visualised brain tissue by a photoaffinity label. Proc Natl Acad Sci, USA 77, 1666–1670. Morrow AL, Suzdak PD, Paul SM (1987). Steroid hormone metabolites potentiate GABA recep-
tor mediated chloride ion influx with nanomolar potency. Eur J Pharmacol 142, 483–485. Nayeem N, Green TP, Martin IL, Barnard EA (1994). Quaternary structure of the native GABA-
A receptor determined by electron microscopic image analysis. J Neurochem 62, 815–818. Nicoll RA, Malenka RC (1998). A tale of two transmitters. Science 281, 360–361.
Nutt DJ, Malizia AL (2001). New insights into the role of the GABA-A-benzodiazepine receptor
in psychiatric disorder. Br J Psychiatry 179, 390–396.
Nutt DJ, Smith CF, Bennett R, Jackson HC (1992). Investigations on the set-point theory of ben-
zodiazepine receptor function. Adv Biochem Psychopharmacol 47, 419–429.
Ozawa M, Sugimachi K, Nakada-Kometani Y, Akai T, Yamaguchi M (1994). Chronic pharmaco- logical activities of the novel anxiolytic -carboline abecarnil in rats. J Pharmacol Exp Ther
269, 457–462.
Persohn E, Malherbe P, Richards JG (1992). Comparative molecular neuroanatomy of cloned
GABA-A receptor subunits in the rat CNS. J Comp Neurol 326, 193–216.
Pesold C, Carruncho HJ, Impagnatiello F et al. (1997). Tolerance to diazepam and changes in the
GABA-A receptor subunit expression in rat neocortical areas. Neuroscience 79, 477–487. Primus RJ, Gallager DW (1992). GABA-A receptor subunit mRNA levels are differentially influ-
enced by chronic FG 7142 and diazepam exposure. Eur J Pharmacol 226, 21–28.
Pritchett DB, Seeburg PH (1990). GABA-A receptor alpha 5 subunit creates novel type II ben-
zodiazepine receptor pharmacology. J Neurochem 54, 1802–1804.
Pritchett DB, Luddens H, Seeburg PH (1989a). Type I and type II GABA-A–benzodiazepine
receptors produced in transfected cells. Science 245, 1389–1392.
Pritchett DB, Sontheimer H, Shivers B et al. (1989b). Importance of a novel GABA-A receptor
subunit for benzodiazepine pharmacology. Nature 338, 582–585.
Puia G, Santi M, Vicini S et al. (1990). Neurosteroids act on recombinant GABA-A receptors.
Neuron 4, 759–765.
Puia G, Vicini S, Seeburg PH, Costa E (1991). Influence of recombinant GABA-A receptor
subunit composition on the action of allosteric modulators of GABA-gated Cl currents. Mol
Pharmacol 39, 691–696.
Puia G, Ducic S, Vicini S, Costa E (1993). Does neurosteroid modulatory efficacy depend on
GABA-A receptor subunit composition? Receptors Channels 1, 135–142.
318 S. A. Pandit, S. V. Argyropoulos, P. G. Kehoe, and D. J. Nutt
Quirk K, Gillard NP, Ragan CI, Whiting PJ, McKernan RM (1994). GABA-type A receptors in the rat brain can contain both gamma2 and gamma3 subunits, but gamma1 does not exist in com- bination with another gamma subunit. Mol Pharmacol 45, 1061–1070.
Rice DP, Miller LS (1998). Health economics and cost implications of anxiety and other mental disorders in the United States. Br J Psychiatry 173(Suppl 34), 4–9.
Roses AD (2000). Pharmacogenetics and the practice of medicine. Nature 405, 857–865.
Ruano D, Araujo A, Machado A, Deblas AL, Vitorica J (1994). Molecular characterisation of type I GABA-A receptor complex from rat cerebral cortex and hippocampus. Mol Brain Res 25,
225–233.
Rudolph U, Crestani F, Benke D et al. (1999). Benzodiazepine actions mediated by specific
gamma-aminobutyric acid (A) receptor subtypes. Nature 401, 796–800.
Sandford JJ, Argyropoulos SV, Nutt DJ (2000). The psychobiology of anxiolytic drugs. Part I:
basic neurobiology. Pharmacol Ther 88, 197–212.
Saxena NC, MacDonald RL (1994). Assembly of GABA-A receptor subunits: role of the gamma
subunit. J Neurosci 14, 7077–7086.
Schneier FR, Liebowitz MR, Abi-Dargham A, Zea-Ponce Y, Lin SH, Laruelle M (2000). Low dop-
amine D(2) receptor binding potential in social phobia. Am J Psychiatry 157, 457–459. Schofield PR, Darlison MG, Fujita N et al. (1987). Sequence and functional expression of the
GABA-A receptor shows a ligand-gated receptor super-family. Nature 328, 221–227.
Schukit MA, Mazzanti C, Smith TL, Ahmed U, Radel M, Iwata N, Goldman D (1999). Selective genotyping for the role of 5 HT2C and GABA alpha6 receptors and the serotonin transporter
in the level of response to alcohol: a pilot study. Biol Psychiatry 45, 647–651.
Sequier JM, Richards JG, Malherbe P, Price GW, Mathews S, Mohler H (1988). Mapping of brain areas containing RNA homologous to cDNAs encoding the and subunits of the rat GABA-
A receptor. Proc Natl Acad Sci, USA 85, 7815–7819.
Siegel E, Baur R, Trube G, Mohler H, Malherbe P (1990). The effect of subunit composition of
rat brain GABA-A receptors on channel function. Neuron 5, 703–711.
Sieghart W (1989). Multiplicity of GABAA-benzodiazepine receptors. Trend Pharm Sci 10,
407–411.
Sieghart W (1995). Structure and pharmacology of GABA-A receptor subtypes. Pharm Rev 47,
181–234.
Sieghart W, Eichinger A, Richards JG, Mohler H (1987). Photoaffinity labeling of benzodiaze-
pine receptor proteins with the partial inverse agonist [3H]RO 15–4513: a biochemical and
autoradiographic study. J Neurochem 48, 46–52.
Smith SS, Gong QH, Hsu FC , Markowitz RS, French-Mullen JMH, Li X (1998). GABA(A) recep-
tor 4 subunit suppression prevents withdrawal properties of an endogenous steroid. Nature
392, 926–930.
Squires RF, Braestrup C (1977). Benzodiazepine receptors in rat brain. Nature 266, 732–734. Stephens DN, Schneider HH, Kehr W et al. (1990). Abecarnil, a metabolically stable, anxioselec-
tive beta-carboline acting at benzodiazepine receptors. J Pharmacol Exp Ther 253,
334–343.
Tallman JF, Gallager DW (1985). The GABA-ergic system: a locus of benzodiazepine action. Annu
Rev Neurosci 8, 21–44.
319 Anxiolytic drugs and the GABA–benzodiazepine receptor complex
Ticku M, Mehta A, Lehoullier P (1988). Spinal cord cultured neurons: an in vitro model to study GABA synaptic pharmacology. In Biggio G, Costa E eds. Adv Biochem Psychopharmacol: Chloride channels and their Modulation by Neurotransmitters and Drugs, Vol 45. New York: Raven Press, pp. 151–159.
Tiihonen J, Kuikka J, Bergstrom K, Lepola U, Koponen H, Leinonen E (1997). Dopamine reup- take site densities in patients with social phobia. Am J Psychiatry 154, 239–242.
Togel M, Mossier B, Fuchs K, Sieghart W (1994). -Aminobutyric acid A receptors displaying association of 3 subunits with 2/3 and different -subunits exhibit unique pharmacologi- cal properties. J Biol Chem 269, 12993–12998.
Turner DM, Ransom RW, Yang JSJ, Olsen RW (1989). Steroid anaesthetics and naturally occur- ring analogs modulate the gamma-aminobutyric acid receptor complex at a site distinct from barbiturates. J Pharmacol Exp Ther 248, 960–966.
Verdoorn TA (1994). Formation of heteromeric -aminobutyric acid type A receptors contain- ing two different -subunits. Mol Pharmacol 45, 475–480.
Verdoorn TA, Draguhn A, Ymer S, Seeburg PH, Sakmann B (1990). Functional properties of recombinant rat GABA-A receptors depend upon subunit composition. Neuron 4, 919–928.
Wang, JB, Liu ZF, Kofuji P, Burt DR (1998). The GABA-A receptor gamma1 subunit in seizure prone (DBA/2) and resistant (C57BL/6) mice. Brain Res Bulletin 45, 421–425.
Weiller E, Bisserbe J-C, Maier W, Lecrubier Y (1998). Prevalence and recognition of anxiety symptoms in five European primary care settings: a report from the WHO study on psycho- logical problems in general health care. Br J Psychiatry 173(Suppl 34), 18–23.
Whiting PJ (1999). The GABA-A receptor gene family: new targets for therapeutic intervention. Neurochem Int 34, 387–390.
Wisden W, Herb A, Wieland H, Kiennanen K, Luddens H, Seeburg PH (1991). Cloning, phar- macological characteristics and expression pattern of the rat GABA-A receptor 4 subunit. FEBS Lett 289, 227–230.
Wisden W, Laurie DJ, Moyner H, Seeburg PH (1992). The distribution of 13 GABA-A receptor subunit mRNAs in the rat brain. I. Telencephalon, diencephalon, mesencephalon. J Neurosci 12, 1040–1062.
Wong G, Sei Y, Skolnick P (1992). Stable expression of gamma-aminobutyric acid typeA/ben- zodiazepine receptors in transfected cell line. Mol Pharmacol 42, 996–1003.
World Health Organization (1992). The Tenth Revision of the International Classification of Diseases and Related Health Problems (ICD-10). Geneva: World Health Organization.
Yang SH, Armson PF, Cha J, Philips WD (1997). Clustering of GABA-A receptors by rapsyn/43kD protein in vitro. Mol Cell Neuroscience 8, 430–438.
Ymer S, Draguhn A, Wisden W et al. (1990). Structural and functional characterisation of the subunit of GABA/benzodiazepine receptors. EMBO J 9, 3261–3267.
Zezula J, Slany A, Sieghart W (1996). Interaction of allosteric ligands with GABA-A receptors containing one, two, or three different subunits. Eur J Pharmacol 301, 207–214.
Zhang JH, Araki T, Sato M, Tohyma M (1991a). Distribution of GABA-A receptor 1 subunit gene expression in the rat forebrain. Mol Brain Res 11, 239–247.
Zhang JH, Sato M, Tohyma M (1991b). Region specific expression of the mRNAs encoding - subunits (1, 2 and 3) of GABA-A receptor in the rat brain. J Comp Neurol 303, 637–657.
15
Genetic factors and long-term prophylaxis in bipolar disorder
Martin Alda
Department of Psychiatry, Dalhousie University, Halifax, Canada
OVERVIEW
Response to long-term lithium treatment in bipolar disorder appears to be related to the family history of a patient. Conversely, some studies have indicated that treatment response could identify a subtype of bipolar disorder characterized by a stronger role of genetic factors and possibly by major-gene effects. Several research groups have now collected samples from patients treated with lithium and these are being studied by molecular genetic methods. These studies aim to identify genes associated with the treatment response or to map genes for bipolar disorder in homogeneous populations of treatment responders. Preliminary findings with a number of candidate genes have produced mostly negative find- ings, but several promising associations have been also identified, for instance, an associa- tion of lithium-responsive bipolar disorder with the gene for phospholipase C1. Further research will be needed to examine differences between genetic factors involved in treat- ment response as opposed to genes associated with the illness. One possible research strat- egy appears to be a study of pairs of relatives concordant for the illness but discordant for their treatment response. Other important areas of research are phenotype definition, espe- cially definition of long-term prophylactic response, and studies of biochemical phenotypes and alterations in gene expression.
Introduction
Lithium is the standard treatment for bipolar disorder. In recent years, several other treatments have appeared promising, mainly anticonvulsants, but also calcium channel blockers and other drugs.
Response to lithium appears to identify a more homogeneous subtype of bipolar disorder with higher heritability. Recent data also suggest that treatment response itself may be a highly familial trait. Several groups have now conducted molecular genetic studies in order to identify genes contributing to the etiology of bipolar dis- order as well as genes that may influence long-term treatment response.

320

321 Genetic factors and long-term prophylaxis in bipolar disorder
There is limited evidence that the response to mood stabilizers could be individ- ually specific (Post et al., 1984). Therefore, we will focus on individual treatments, in particular on lithium for which the most data are available.
Lithium
Family and family-history studies of lithium response
The relationship between family history and response to lithium has been the subject of investigations since the 1970s. These included studies where diagnostic information on most relatives was obtained indirectly (family-history studies) and studies where diagnostic information was obtained by interviewing in person as many relatives as possible (family studies). Arguably, the distinction is somewhat artificial and many studies used a combination of approaches to gather all available information. These studies are summarized in Table 15.1.
Mendlewicz et al. (1973) published a family study of 36 patients treated with lithium for 3 years. Those patients who did not relapse (24 out of 36) were classified as responders and, significantly more often, had a family history of bipolar disor- der compared with those who relapsed on lithium (12 out of 36). Subsequently, several other studies found an association between prophylactic lithium response and a family history of bipolar disorder. These studies generally fall into one of two categories. Family-history studies often looked for factors predicting response to lithium. Family history would be typically included among other clinical variables. Their results are less clear-cut than those of family studies (discussed below). Some family-history studies found an association between a family history of bipolar dis- order and treatment response (Prien et al., 1974; Svestka and Nahunek, 1975; Svestka, 1979; Maj et al., 1984); some also found an association with a family history of unipolar depression (Abou-Saleh and Coppen, 1986). Several other studies, however, were unable to confirm such findings (Dunner et al., 1976) or found the opposite: an association of family history of bipolar disorder and non- response to lithium (Misra and Burns, 1976; Strober et al., 1988). The latter two studies must be interpreted with caution as the first was based on a case series of seven nonresponders only, with no comparison group, and the second was a study of adolescent bipolar disorders. A more detailed review of these studies can be found in Grof et al. (1994).
Family studies, by comparison, provide a consistent support of the hypothesis of familial differences between responders and nonresponders. In addition to the study of Mendlewicz et al. (1973), several other investigations showed similar results (Zvolsky et al., 1974, 1979; Mendlewicz, 1979; Smeraldi et al., 1984; Grof et al., 1994). In our analysis of 121 families, we found, on the one hand, that responders to lithium had significantly higher rates of bipolar disorder among their

  
Table 15.1. Family/family-history studies and response to lithium
Study Type Mendlewicz et al., 1973 F
Design/outcome
Probands 24R,12NR
Treatment duration 6–37 months
Results
More frequent history of BD in families of R
Zvolsky et al., 1974 F
Prevalence of psychiatric disorders in first-degree relatives
58 BD
2 years 2 years
Higher rates in relatives of complete and partial responders
Prien et al., 1974 FH
Response to treatment Treatment outcome
48 R, 55 R,
43 NR 26 NR
Better response in subjects with FH of BD
Svestka and Nahunek, FH 1975
1–7 years
FH of endogenous psychoses and suicide higher in R
Dunner et al., 1976 FH Misra and Burns, 1977 FH Zvolsky et al., 1979 F
Relapse rate on lithium FHofAD
52 R, 9NR 26 R,
44 NR
1–45 months 2–8 years 7 years
No effect of FH of BD or UD 7outof9NRhadFHofAD
Mendlewicz et al., 1979 F Maj et al., 1984 FH
Concordance rates in twins
24 R, 59 R,
18 NR 41 NR
2 years 2 years
Higher concordance (both MZ and DZ) for R More frequent FH of BD in relatives of R
Smeraldi et al., 1984 F
Morbidity risks of AD in first- degree relatives
53 NR
Affective Morbidity Index (AMI) 1986 during lithium treatment
Abou-Saleh and Coppen, FH Shapiro et al., 1989 FH
Relapse rate on lithium
Relapse rate on lithium or imipramine unrelated to FH
Prevalence of family history of BD and UD
Prevalence of psychiatric disorders in first-degree relatives
17 NR
Higher rates in relatives of R compared with those of NR
Frequency of FH of BD, UD, and other disorders
92 R, 27 BP 117
3 years 2 years
2 years
Higher rates of AD in relatives of R
Lower AMI in patients with FH of BD or UD

 
Grof et al., 1994 Engstrom et al., 1997 Coryell et al., 2000
F F
Morbidity risks of BD, UD, SZ in first-degree relatives
71 R, 50 NR
3–20 years 13 8 years
Increased risk of BD in families of R, increased SZ in families of NR
Notes:
F and FH
Morbidity risks of BD, UD, SZ and alcoholism in relatives of probands with low, medium or high morbidity on Li
62 low,
55 medium, 69 high
11 8 years 7.8 5.8 years
No major differences between the groups
Frequency of episodes on lithium
51 Fam, 47 nFam
0.32 0.44 in nFam patients 0.58 0.70 in Fam patients
F, family study; FH, family-history study; R, responders; NR, nonresponders; BD, bipolar disorder; UD, unipolar depression; AD, affective disorder; SZ, schizophrenia; Fam, familial; nFam, non-familial (sporadic).
5.7 4.1 years 5.5 4.5 years

324 M. Alda
first-degree relatives, while nonresponders had higher rates of schizophrenia in their families (Grof et al., 1994). On the other hand, we did not find any difference between the two groups with respect to family history of alcoholism, and the prev- alence of alcoholism in families of both groups of probands was quite low (Duffy et al., 1998b).
More recently, two studies, in particular, raised questions about the possible link between family history and treatment response. Engstrom et al. (1997) analyzed data from 90 families, 39 with positive and 51 with negative family histories of bipolar disorder. They found that the family-history-negative subjects had fewer episodes on lithium compared with those with positive family histories. This was a family-history study with at least one relative per family interviewed. The treat- ment response was not defined categorically but rather used as a quantitative vari- able (frequency of episodes on and off lithium). Therefore, it is not possible to compare the data with those from earlier studies. For instance, it is not clear how many responders (subjects with no activity of illness while treated) were in the sample.
In the second study, Coryell et al. (2000) reported psychiatric morbidity in rela- tives of probands subdivided according to their frequency of episodes in the course of prophylactic treatment. The diagnostic data on relatives were based on personal interviews with a majority of the relatives, who were also re-interviewed after 6 years. As in the study of Engstrom et al. (1997), Coryell et al. (2000) reported average frequencies of episodes on and off lithium. The study could not find any differences between the rates of bipolar disorder in families of patients divided into three groups according to their episode frequency in the course of lithium treat- ment (high–medium–low). The group with the highest morbidity on lithium had a higher frequency of family history positive for major depression. Again, it is not possible to tell whether the probands had actually responded to the treatment or whether the differences reflect the natural course of the illness with some “respond- ers” being patients with slow cycling disorder. Even the responders had a noticeable activity of illness while in treatment. Finally, the interpretation of the findings is more difficult because the group with a low frequency of episodes on lithium also had significantly lower morbidity before the treatment.
Consequently, it is not clear to what extent these recent results are in conflict with the older ones. The earlier studies usually included subjects with large numbers of pretreatment episodes who, by definition, had to stay free of any recurrence for a specified length of time – a minimum of 3 years in Smeraldi et al. (1984) and in Grof et al. (1994). Finally, we have argued elsewhere that bipolar disorder is now diagnosed more liberally (Grof et al., 1995; Grof and Alda, 2000). This makes it more difficult to compare results obtained within studies that used a more con- servative approach to diagnosis. This is also reflected in the huge increase in
325 Genetic factors and long-term prophylaxis in bipolar disorder
population prevalence of mood disorders, certainly not explained simply by period or cohort effects, and in the increase in morbidity risks in family studies (Grof et al., 1995; Duffy et al., 2000a).
Family studies are complemented by high-risk studies of the children of lithium responders and nonresponders. These studies show that the offspring of respond- ers typically develop disorders in the affective spectrum, with little or no comor- bidity and often high premorbid functioning. Their illness seems to be episodic with complete interepisodic recoveries. The children of nonresponders often develop multiple psychiatric problems that include nonaffective symptoms as well. Their illness seems to follow a more chronic course, characterized by incomplete remissions (Duffy et al., 1998a).
Mode of inheritance
Smeraldi et al. (1984) first applied segregation analysis to family data from responders and nonresponders to lithium. They found support for major-gene effects in the families of responders. Our data from responders are also consistent with a major-gene effect, specifically a gene of a recessive type (Alda et al., 1994). This result was replicated by our group in an independent sample (Alda et al., 1997). As segregation analyses typically test the mode of inheritance by fitting spe- cific genetic models to family data, their results need to be viewed as approximate, providing leads for further research. Their application to lithium-responsive bipolar disorder suggests that this may be a suitable population for gene mapping studies.
Lithium response as a phenotype
It has been argued that lithium response identifies a more homogeneous genetically distinct subtype of bipolar disorder. Conversely, some authors view lithium response mainly as a continuum, determined to a large extent by good compliance and low (or no) comorbidity. It is also possible that lithium response is influenced by genetic factors, partially or completely independent from those contributing to the etiology of the illness.
Therefore, it is of interest to establish whether lithium response itself is familial. Only few studies have examined this issue in the past.
Several investigations of children treated with lithium suggested beneficial effects in a variety of conditions but primarily in affective symptoms (Annell, 1969; Delong, 1978; Youngerman and Canino, 1978; McKnew et al., 1981). One study of children of bipolar lithium responders suggested that they had a response concor- dant with that of their parents (McKnew et al., 1981). Indirect support for the familiality of lithium treatment response in the children of bipolar probands can also be found in a study by Duffy et al. (1998a), who described similarities between
326 M. Alda
parents and children with respect to the episodicity of the clinical course, a hall- mark of the lithium-responsive form of bipolar disorder (Grof et al., 1993). Most recently, in a study of 24 relatives of probands who responded unequivocally to lithium, 16 (67%) showed a clear-cut response (Grof et al., 2000). This is signifi- cantly higher than the response rate in a comparison sample in which only 9 out of 30 (30%) subjects responded according to the same criteria.
Admittedly, none of these results can differentiate between the two main pos- sibilities, namely, that lithium response is determined by an independent genetic factor or that responders represent a subtype of the illness. Both would predict a trend towards familial clustering of the response. For the same reason, the results of association studies comparing responders and nonresponders cannot be inter- preted one way or the other either. Ultimately, a comparison of nonresponders and responders from the same family could help in identifying genes associated with the response but not with the illness. In our experience, such pairs of relatives, with their response (or nonresponse) well established, are difficult to find.
Molecular genetic studies
Several research groups have established clinical samples of patients treated with lithium for the purpose of molecular genetic investigations. Some of these studies are aimed primarily at the identification of those loci influencing the response to the treatment. Other groups use treatment response as an additional phenotypic dimension or homogeneity criterion.
Studies of lithium responders as a homogeneous subtype of bipolar disorder
We have studied patients with bipolar disorder responsive to lithium in collabora- tion with the International Group for Study of Lithium (IGSLI). The group coor- dinates research efforts of centers in Austria, Canada, Czech Republic, Denmark, Germany, and Sweden. The cornerstone of our work has been the assumption that lithium-responsive bipolar disorder is a distinct, more genetically based subtype of bipolar disorder.
The initial association studies focused on the gene for tyrosine hydroxylase (Cavazzoni et al., 1996) and on markers on chromosome 18 (Turecki et al., 1996). Subsequently, candidate genes were analyzed that were thought to be relevant to the etiology of the illness and/or treatment response. Among these, there is a positive association with the gene for phospholipase C1 (Turecki et al., 1998). In a subset of unilineal families there was also suggestive linkage (lod 1.45; p 0.004). Phospholipase C is a promising candidate gene because of its role in the phospho- inositol cycle, a major target of lithium. No association was found with several other candidate genes: monoamine oxidase A (Turecki et al., 1999a), corticotropin- releasing hormone, proenkephalin (Alda et al., 2000), and several genes related to
327 Genetic factors and long-term prophylaxis in bipolar disorder
GABAergic function (Duffy et al., 2000b). The hypothesis of anticipation and unstable trinucleotide repeats was tested by searching for polyglutamate protein sequences in lymphoblasts, using a specific antibody (Turecki et al., 1999c), and by studying association with markers known to contain CAG repeats (Turecki et al., 2000). Neither study supported the anticipation hypothesis.
These candidate gene association studies were usually complemented by linkage analyses of a related sample of moderately large families. Linkage was also tested to chromosome 18, at the time one of the most studied regions in bipolar disorder, with several independent reports of linkage (Berrettini et al., 1994; Stine et al., 1995; Freimer et al., 1996). Using a set of markers spanning the entire chromosome, no linkage could be demonstrated even after separating the families into those with maternal or paternal transmission of the illness (Turecki et al., 1999b). It is possible that linkage to chromosome 18 may be more common in subjects with atypical illness. This would also be consistent with the report of linkage of schizophrenia to chromosome 18 (Schwab et al., 1998).
Following these studies of candidate genes and candidate regions, a full genome scan was completed with several chromosomal regions showing lod scores in the 1.8 to 3.5 range, namely those on chromosomes 6, 7, 15, 21, and 22 (Turecki et al., 2001). The scan included a total of 247 subjects in 31 families identified through probands with bipolar disorder responsive to lithium. A total of 108 subjects in these families were affected with bipolar disorder or recurrent unipolar depression. These conditions were considered as a phenotype for the first set of analyses. For the genome scan, 378 markers spaced at approximately 10cM intervals were used. The data were analyzed by parametric (lod score) and non-parametric (SimIBD) methods. There was evidence for linkage in the 15q14 region (marker ACTC lod 3.46; locus-specific; p 0.000014) and suggestive linkage for another marker on chromosome 7q11.2 (D7S1816; lod 2.68; p 0.00011). In the second set of anal- yses, treatment response was used as a phenotype; unaffected relatives and those not treated with lithium were considered “phenotype unknown” in the linkage analysis. The highest lod score for this phenotype was for the marker D7S1816 (lod 1.53; locus-specific p 0.003).
Studies comparing responders and nonresponders to lithium
Several research groups have initiated searches for genes associated with response (or no response) to lithium using association case-control strategies.
A sample of lithium-treated patients has been collected as a part of the European Collaborative Project on Affective Disorders. In a preliminary analysis of the sample with dopamine receptor genes and tyrosine hydroxylase, Lipp et al. (1997) reported equivocal results, with a trend towards the association of bipolar disorder with the D2 receptor gene among nonresponders to lithium. Part of the sample
328 M. Alda
collected at the University of Milan was analyzed separately and reported on in several papers. The association analyses included genes for dopamine D3 receptor (Serretti et al., 1998), tryptophan hydroxylase (Serretti et al., 1999a), dopamine receptors D2 and D4 and -aminobutyric acid A (GABA-A) 1-subunit (Serretti et al., 1999b), and serotonin receptors 2A, 2C, and 1A (Serretti et al., 2000). These analyses were based on a sample of up to 125 subjects with major affective disor- ders, both bipolar and unipolar. None of the markers tested gave any evidence of association.
The group of V. Steen has been focusing on genes controlling the phosphoinos- itol pathway (Steen et al., 1996; Sjoholt et al., 1997; Lovlie et al., 1999). They iden- tified several polymorphisms in the gene for inositol polyphosphate 1-phosphatase. One of these polymorphisms was associated with lithium response in the Norwegian part of the sample, but not in the subjects from Israel (Steen et al., 1998).
Del Zompo et al. (1999) studied the gene for the serotonin transporter, in par- ticular the polymorphism in the promoter region known to affect the transcrip- tional activity of the gene. They found a trend towards higher frequency of the l allele among lithium nonresponders, which contrasts with the reported association of nonresponse to several different serotonin reuptake inhibitors and the ss geno- type (Smeraldi et al., 1998; Kim et al., 2000; Zanardi et al., 2000; see also Ch. 6.)
Other mood stabilizers
There is a relative paucity of data pertaining to family history and the response to other long-term treatments in bipolar disorder. Post et al. (1984) suggested that the acute antimanic response to carbamazepine was associated with family history neg- ative for bipolar disorder. In the above-mentioned study by Coryell et al. (2000), a small number of subjects (31) took anticonvulsants carbamazepine or valproate alone for at least 26 weeks. The authors could not find any differences between family histories of those who had high, medium, or low activity of the illness during the treatment. They caution, however, that the probability of type II error was quite high because of the low numbers.
Both theoretically interesting and practically important is the question of spec- ificity of prophylactic treatments. Post et al. (1984) suggested that the long-term treatment response could be specific to an individual patient. For example, subjects who respond to an anticonvulsant may not be lithium responders and vice versa. It is unclear, at present, how far this hypothesis can be extrapolated with respect to family history. Greil et al. (1997) conducted a large prophylactic trial of lithium and carbamazepine in bipolar disorder and found that lithium was more effective in typical patients while carbamazepine had an advantage in subjects with atypical illness. It would be reasonable to predict that these two groups would differ with

329 Genetic factors and long-term prophylaxis in bipolar disorder
respect to their family histories. Such a conjecture, however, needs to be tested in a rigorous study. The study by Coryell et al. (2000) included 36 subjects treated with both lithium and anticonvulsants in separate trials. In these patients, it seemed that the morbidity on both treatments was similar. The interpretation of this observa- tion is again obscured by the differences in morbidity before the treatment.
Conclusions
The study of genetic factors in long-term treatment of bipolar disorder has been complicated by methodological difficulties. The results so far are promising, yet not conclusive. New research strategies such as studies of gene expression as well as progress in phenotype definition might provide new insights and lead to results that could have important implications for the management of this severe illness.
Acknowledgements
Parts of this work were supported by grants from the Medical Research Council of Canada and from Ontario Mental Health Foundation. Martin Alda holds an Independent Investigator Award from NARSAD.
REFERENCES
Abou-Saleh M, Coppen A (1986). Who responds to prophylactic lithium? J Affect Disord 10, 115–125.
Alda M, Grof P, Grof E, Zvolsky P, Walsh M (1994). Mode of inheritance in families of patients with lithium-responsive affective disorders. Acta Psychiatr Scand 90, 304–310.
Alda M, Grof E, Cavazzoni P et al. (1997). Autosomal recessive inheritance of affective disorders in families of responders to lithium prophylaxis? J Affect Disord 44, 153–157.
Alda M, Turecki G, Grof P et al. (2000). Association and linkage analyses of CRH and PENK genes in bipolar disorder responsive to lithium: a collaborative IGSLI study. Am J Med Genet (Neuropsychiatr Genet) 96, 178–181.
Annell AL (1969). Manic-depressive illness in children and effect of treatment with lithium car- bonate. Acta Paedopsychiatr 36, 292–301.
Berrettini WH, Ferraro TN, Goldin LR et al. (1994). Chromosome 18 DNA markers and manic- depressive illness: evidence for a susceptibility gene. Proc Natl Acad Sci, USA 21, 5918–5921.
Cavazzoni P, Alda M, Turecki G et al. (1996). Lithium-responsive affective disorders: no associ- ation with the tyrosine hydroxylase gene. Psychiatry Res 64, 91–96.
Coryell W, Akiskal H, Leon AC, Turvey C, Solomon D, Endicott J (2000). Family history and symptom levels during treatment for bipolar I affective disorder. Biol Psychiatry 47, 1034–1042.
  
330 M. Alda
Delong GR (1978). Lithium carbonate treatment of select behavior disorders in children suggest- ing manic-depressive illness. J Pediatr 93, 689–694.
Del Zompo M, Ardau R, Palmas MA, Bocchetta A, Reina A, Piccardi MP (1999). Lithium response: association study with two candidate genes. Mol Psychiatry 4, 66–67.
Duffy A, Alda M, Kutcher S, Fusee C, Grof P (1998a). Psychiatric symptoms and syndromes among adolescent children of parents with lithium-responsive or lithium-nonresponsive bipolar disorder. Am J Psychiatry 155, 431–433.
Duffy A, Grof P, Grof E, Zvolsky P, Alda M (1998b). Evidence supporting the independent inher- itance of primary affective disorders and primary alcoholism in the families of bipolar patients. J Affect Disord 50, 91–96.
Duffy A, Grof P, Robertson C, Alda M (2000a). The implications of genetic studies of major mood disorders for clinical practice. J Clin Psychiatry 61, 630–637.
Duffy A, Turecki G, Grof P et al. (2000b). Association and linkage studies of candidate genes involved in GABAergic neurotransmission in lithium responsive bipolar disorder. J Psychiatry Neurosci 25, 353–358.
Dunner DL, Fleiss JL, Fieve RR (1976). Lithium carbonate prophylaxis failure. Br J Psychiatry 129, 40–44.
Engstrom C, Astrom M, Nordqvist-Karlsson B, Adolfsson R, Nylander PO (1997). Relationship between prophylactic effect of lithium therapy and family history of affective disorders. Biol Psychiatry 42, 425–433.
Freimer NB, Reus VI, Escamilla MA et al. (1996). Genetic mapping using haplotype, association and linkage methods suggests a locus for severe bipolar disorder (BPI) at 18q22–q23. Nat Genet 12, 436–441.
Greil W, Ludwig-Mayerhofer W, Erazo N et al. (1997). Lithium versus carbamazepine in the maintenance treatment of bipolar disorders – a randomised study. J Affect Disord 43, 151–161.
Grof P, Alda M (2000). Discrepancies in the efficacy of lithium. Arch Gen Psychiatry 57, 191. Grof P, Alda M, Grof E, Fox D, Cameron P (1993). The challange of predicting response to stab- ilising lithium treatment. The importance of patient selection. Br J Psychiatry 163(Suppl 21),
16–19.
Grof P, Alda M, Grof E, Zvolsky P, Walsh M (1994). Lithium response and genetics of affective
disorders. J Affect Disord 32, 85–95.
Grof P, Alda M, Ahrens B (1995). Clinical course of affective disorders: were Emil Kraepelin and
Jules Angst wrong? Psychopathology 28(Suppl 1), 73–80.
Grof P, Duffy A, Cavazzoni P et al. (2000). Is response to prophylactic lithium a familial trait? Int
J Neuropsychopharmacol 3(Suppl 1), 339–339.
Kim DK, Lim SW, Lee S et al. (2000). Serotonin transporter gene polymorphism and antidepress-
ant response. Neuroreport 11, 215–219.
Lipp O, Mahieu B, Souery D et al. (1997). Molecular genetics of bipolar disorders: implication
for psychotropic drugs. Eur Neuropsychopharmacol 7(Suppl 2), S112–S113.
Lovlie R, Gulbrandsen AK, Molven A, Steen VM (1999). Genomic structure and sequence anal- ysis of a human inositol polyphosphate 1-phosphatase gene (INPP1). Pharmacogenetics 9,
517–528.
331 Genetic factors and long-term prophylaxis in bipolar disorder
Maj M, Del Vecchio M, Starace F, Pirozzi R, Kemali D (1984). Prediction of affective psychoses response to lithium prophylaxis: the role of socio-demographic, clinical, psychological and biological variables. Acta Psychiatr Scand 69, 37–44.
McKnew DH, Cytryn L, Buchsbaum MS et al. (1981). Lithium in children of lithium-responding parents. Psychiatry Res 4, 171–180.
Mendlewicz J (1979). Prediction of treatment outcome: family and twin studies in lithium pro- phylaxis and the question of lithium red blood cell/plasma ratios. In Cooper TB, Gershon S, Kline NS, Schou M, eds. Lithium: Controversies and Unresolved Issues. Amsterdam: Excerpta Medica, pp. 226–240.
Mendlewicz J, Fieve RR, Stallone F (1973). Relationship between the effectiveness of lithium therapy and family history. Am J Psychiatry 130, 1011–1013.
Misra PC, Burns BH (1976). “Lithium nonresponders” in a lithium clinic. Acta Psychiatr Scand 55, 32–38.
Post RM, Berrettini W, Uhde TW, Kellner C (1984). Selective response to the anticonvulsant car- bamazepine in manic-depressive illness: a case study. J Clin Psychopharmacol 4, 178–185.
Post RM, Uhde TW, Roy-Byrne PP, Joffe RT (1987). Correlates of antimanic response to carba- mazepine. Psychiatr Res 21, 71–83.
Prien RF, Caffey EM, Klett CJ (1974). Factors associated with treatment success in lithium car- bonate prophylaxis: report of the Veterans Administration and National Institute of Mental Health Collaborative Study Group. Arch Gen Psychiatry 31, 189–192.
Schwab SG, Hallmayer J, Lerer B et al. (1998). Support for a chromosome 18p locus conferring susceptibility to functional psychoses in families with schizophrenia, by association and linkage analysis. Am J Hum Genet 63, 1139–1152.
Serretti A, Lilli R, Lorenzi C, Franchini L, Smeraldi E (1998). Dopamine receptor D3 gene and response to lithium prophylaxis in mood disorders. Int J Neuropsychopharmacol 1, 125–129.
Serretti A, Lilli R, Lorenzi C, Gasperini M, Smeraldi E (1999a). Tryptophan hydroxylase gene and response to lithium prophylaxis in mood disorders. J Psychiatr Res 33, 371–377.
Serretti A, Lilli R, Lorenzi C et al. (1999b). Dopamine receptor D2 and D4 genes, GABA(A) alpha- 1 subunit genes and response to lithium prophylaxis in mood disorders. Psychiatry Res 30, 7–19.
Serretti A, Lorenzi C, Lilli R, Smeraldi E (2000). Serotonin receptor 2A, 2C, 1A genes and response to lithium prophylaxis in mood disorders. J Psychiatr Res 34, 89–98.
Shapiro DR, Quitkin FM, Fleiss JL (1989). Response to maintenance therapy in bipolar illness. Arch Gen Psychiatry 46, 401–405.
Sjoholt G, Molven A, Lovlie R, Wilcox A, Sikela JM, Steen VM (1997). Genomic structure and chromosomal localization of a human myo-inositol monophosphatase gene (IMPA). Genomics 45, 113–122.
Smeraldi E, Petroccione A, Gasperini M, Macciardi F, Orsini A, Kidd KK (1984). Outcomes on lithium treatment as a tool for genetic studies in affective disorders. J Affect Disord 6, 139–151. Smeraldi E, Zanardi R, Benedetti F, Di Bella D, Perez J, Catalano M (1998). Polymorphism within the promoter of the serotonin transporter gene and antidepressant efficacy of fluvoxamine.
Mol Psychiatry 3, 508–511.
332 M. Alda
Steen VM, Gulbrandsen AK, Eiken HG, Berle JO (1996). Lack of genetic variation in the coding region of the myo-inositol monophosphatase gene in lithium-treated patients with manic depressive illness. Pharmacogenetics 6, 113–116.
Steen VM, Lovlie R, Osher Y, Belmaker RH, Berle JO, Gulbrandsen AK (1998). The polymorphic inositol polyphosphate 1-phosphatase gene as a candidate for pharmacogenetic prediction of lithium-responsive manic-depressive illness. Pharmacogenetics 8, 259–268.
Stine OC, Xu J, Koskela R, McMahon FJ et al. (1995). Evidence for linkage of bipolar disorder to chromosome 18 with a parent-of-origin effect. Am J Hum Genet 57, 1384–1394.
Strober M, Morrell W, Burroughs J, Lampert C, Danforth H, Freeman R (1988). A family study of bipolar I disorder in adolescence: early onset of symptoms linked to increased familial loading and lithium resistance. J Affect Disord 15, 255–268.
Svestka J (1979). Therapy and prophylaxis of manic-depressive psychosis with lithium carbonate, with respect to possibility of prediction of the therapeutic result. Acta Facult Med Uni Brunensis 65, 25–102.
Svestka J, Nahunek K (1975). The results of lithium therapy in acute phases of affective psycho- ses and some other prognosticator factors of lithium prophylaxis. Activ Nerv Super (Praha) 17, 270–272.
Turecki G, Alda M, Grof P et al. (1996). No association between chromosome-18 markers and lithium responsive affective disorders. Psychiatry Res 63, 17–23.
Turecki G, Grof P, Cavazzoni P et al. (1998). Evidence for a role of phospholipase C-gamma1 in the pathogenesis of bipolar disorder. Mol Psychiatry 3, 534–538.
Turecki G, Grof P, Cavazzoni P et al. (1999a). MAO-A: association and linkage studies with lithium responsive bipolar disorder. Psychiatr Genet 9, 13–16.
Turecki G, Grof P, Cavazzoni P et al. (1999b). Lithium responsive bipolar disorder, unilineality, and chromosome 18: A linkage study. Am J Med Genet (Neuropsychiatr Genet) 88, 411–415. Turecki G, Alda M, Grof P et al. (1999c). Expanded polyglutamine tracts: no evidence of a major
role in bipolar disorder. Mol Psychiatry 4, 220–222.
Turecki G, Alda M, Grof P et al. (2000). Polyglutamine coding genes in bipolar disorder. Lack of
association with selected candidate loci. J Affect Disord 53, 63–68.
Turecki G, Grof P, Grof E et al. (2001). Mapping susceptibility genes for bipolar disorder: a phar-
macogenetic approach based on excellent response to lithium. Mol Psychiatry 6, 570–578. Youngerman J, Canino I (1978). Lithium carbonate use in children and adolescents. A survey of
the literature. Arch Gen Psychiatry 35, 216–224.
Zanardi R, Benedetti F, Di Bella D, Catalano M, Smeraldi E (2000). Efficacy of paroxetine in
depression is influenced by a functional polymorphism within the promoter of the serotonin
transporter gene. J Clin Psychopharmacol 20, 105–107.
Zvolsky P, Vinarova E, Dostal T, Soucek K (1974). Family history of manic-depressive and endo-
genous depressive patients and clinical effect of treatment with lithium. Activ Nerv Super 16,
194–195.
Zvolsky P, Dvorakova M, Soucek K, Vinarova E, Dostal T (1979). Clinical use of lithium salts
from the genetic viewpoint. In Cooper TB, Gershon S, Kline NS, Schou M, eds. Lithium: Controversies and Unresolved Issues. Amsterdam: Excerpta Medica, pp. 152–154.
16
Genetic influences on responsiveness to anticonvulsant drugs
Thomas N. Ferraro
Center for Neurobiology and Behavior, Department of Psychiatry, University of Pennsylvania, Philadelphia, USA
OVERVIEW
This chapter addresses pharmacogenetic issues associated with the clinical use of anticon- vulsant drugs (ACDs). Individual variability in responsiveness has been documented for nearly all ACDs in common use and current perspectives suggest that genes play a signifi- cant causal role. Variation in the DNA sequence of genes that encode drug-metabolizing enzymes is a major factor contributing to differences in drug responsiveness, and several important polymorphisms have been documented to affect the disposition of ACDs, partic- ularly, the influence of CYP2C9 and CYP2C19 on metabolism of phenytoin and mepheny- toin, respectively. Although the metabolism of other ACDs may also involve polymorphisms producing CYP isoforms, additional studies are needed to determine if these polymorphisms affect pharmacokinetics. A much greater need, however, is for studies that focus on gene variations that encode target proteins against which ACDs act. Studies that relate genetic variation to ACD pharmacodynamics will provide insight into specific gene sequences and protein structures that are most important for drug action. Additionally, this experimental strategy will have the potential of identifying previously unknown drug targets and mecha- nisms of action. Together, information on genetic contribution to ACD pharmacokinetics and pharmacodynamics will enable more successful and less complicated use of this class of drugs.
Introduction
Clinical experience has documented that patients with epilepsy respond in differ- ent ways to standard ACDs. Individuals with similar seizure disorders based on diagnostic criteria exhibit variable responses to the same medication, with some patients deriving significant benefit and others none at all. Additionally, up to one third of patients with epilepsy are refractory to all drug therapies (Porter and Rogawski, 1992; Jallon, 1997; Loscher, 1997). The increasing trend involving use of ACDs for treatment of affective disorders, most notably bipolar disorder (Denicoff,

333

334 T. N. Ferraro
et al., 1997; Mishory et al., 2000; Vasudev et al., 2000), will likely result in similar observations in those patient populations. Thus, understanding individual differ- ences in response to ACDs has many important clinical ramifications.
A consequence of large interindividual differences in responsiveness to ACDs is that it has been difficult to generate standardized treatment regimens even for clin- ically well-defined epilepsy subtypes. The development of new ACDs is particularly problematic since the least ACD-responsive patients are included in clinical trials. Individual differences in adverse side effects are a major factor influencing the eval- uation of new drugs and greatly complicate standardized therapy. Individual vari- ability in response to therapeutic agents has many causes, both genetic and environmental. Until recently, however, attempts to understand failure of drug treatment in epilepsy had been focused almost exclusively on environmental vari- ables such as dosing regimens, drug interactions, and dietary factors. Advances in molecular genetic analysis during the 1990s now provide the means to begin to shift focus to identify genetic factors that influence ACD therapy (Evans and Relling, 1999).
The nature of genetic variation
There are a number of different kinds of natural genetic variation that result in functional protein differences and affect drug responsiveness. The type of variation that is most prevalent in the genome and, arguably, the most relevant to common diseases and complex traits is represented by single nucleotide polymorphisms (SNPs) (Kleyn and Vessell, 1998). SNPs are single base substitutions in a given DNA sequence and are estimated to occur every 500–1000 bases throughout the genome (Johnson and Todd, 2000; McCarthy and Hilfiker, 2000). Protein diversity that extends from SNPs can take the form of single amino acid differences, termed “mis- sense mutations” if the base change occurs within the sequence of a gene exon. Such missense mutations can result in a protein product that exhibits a gain or loss of function. SNPs may also introduce premature stop codons and eliminate func- tional protein product. Because of the redundancy of the genetic code, many SNPs are masked at the protein level. They are sometimes referred to as “silent muta- tions.” Noncoding sequences help to govern exon splicing, rate of transcription, and mRNA stability (Lewin, 2000). SNPs that occur in noncoding regions can give rise to different amounts of proteins with altered structures and levels of functional activity. While gross mutations such as inversions, translocations, deletions, or insertions can have dramatic functional consequences, careful analysis of SNPs may be the most likely path to advances in pharmacogenomics and a greater under- standing of the genetic basis of individual variability in drug responsiveness (Kleyn and Vessell, 1998).

335 Genetic influences on responsiveness to anticonvulsant drugs
Genetic variability and drug responsiveness
There are many ways in which genetic variability, the differences in the sequence of specific genes, can lead to variability in drug responsiveness and it is likely that each of these mechanisms is involved in determining the effects of ACDs. Up to now, the focus of pharmacogenetic studies has been on “polymorphic drug metabolism” (Lin et al., 1996; Kalow, 1997). This line of research examines variations in the sequence of genes that code for drug-metabolizing enzymes. SNPs have been found to underlie variability in drug-metabolizing capacity and such variability is often documented as differences in drug and metabolite profiles between patients. It is likely that variation in genes coding for proteins involved in other pharmacokinetic processes, including drug absorption, distribution, and excretion, also help to determine individual differences in responsiveness although these mechanisms are not as well studied regarding genetic influences.
Another way in which genetic variation plays a role in drug responses depends upon polymorphisms in genes that code for proteins that are drug targets or that mediate drug action. For example, if the drug target is a neurotransmitter receptor, it can be hypothesized that subtle differences in the amino acid sequence of the pro- teins that make up the receptor could give rise to important functional differences that affect response. These differences in response apply both to therapeutic effects and adverse effects so that both efficacy and toxicity may be influenced by individ- ual genetic variation. In the future, pharmacogenetic discoveries related to ACDs will make individualization of therapy even more difficult in the sense that the genetic substrate that mediates efficacy may well differ from that which mediates toxicity. As such, two different sets of genes and genotypes will need to be exam- ined in order to match patients with the most appropriate ACD.
In addition to natural variation in the genes that code for primary drug targets or molecules involved in pharmacokinetic processes, another way that genes can affect response to therapy is related to the underlying genetic predisposition to the disease itself. Such is the case in a syndrome called autosomal dominant nocturnal frontal lobe epilepsy, where disease-causing mutations in genes for nicotinic ace- tylcholine receptor subunits can be demonstrated to yield receptors that are abnor- mally sensitive to the electrophysiological effects of the drug carbamazepine (Picard et al., 1999). Thus, natural variation in genes that predispose to common forms of epilepsy could also have an influence on the response to ACDs.
Genetic influences on the pharmacology of anticonvulsant drugs
Although it has expanded during the 1990s, the list of drugs used in clinical prac- tice for the treatment of epilepsy is relatively small. This section first presents
 
336 T. N. Ferraro
general comments relevant to the mechanisms of action of conventional classes of ACD and then addresses how these mechanisms are potentially influenced by genetic variation. Subsequently, commonly used drugs are discussed individually in some depth relative to genes that may affect various aspects of their pharmacol- ogy, including metabolism.
Among the nonsedating standard ACDs, those most frequently prescribed are phenytoin, carbamazepine, and valproic acid with the depressant drugs phenobar- bital and clonazepam also being in common use (McNamara, 1996; Cafiero and Verdone, 1997). Standard ACDs suppress seizure activity in one of three ways: blockade of voltage-dependent sodium channels, blockade of T-type calcium chan- nels, and/or potentiation of -aminobutyric acid (GABA)-mediated postsynaptic inhibition with increased chloride influx (MacDonald and Kelly, 1993). Frequently, ACDs are shown to have more than one cellular action and there can be uncertainty regarding which actions are most relevant to observed anticonvulsant effects.
It is well documented that barbiturates and benzodiazepines facilitate chloride influx into neurons by virtue of specific allosteric binding sites on GABA-A recep- tors (MacDonald and Kelly, 1993). GABA-A subunits vary in their ability to bind ACDs such that recombinant receptors comprising only - and -subunits respond to barbiturates but not benzodiazepines, which require the presence of a -subunit (Mehta and Ticku, 1999; see also Ch. 15). Subtypes of each subunit are encoded by related genes with differing degrees of homology, underscoring the potential influence of genetic variability on overall receptor function since GABA- A receptors comprise multiple subunits (Whiting, 1999). Some of these genes have already been shown to harbor functional SNPs (Cheng et al., 1997; Iwata et al., 2000). Thus, the effect of genetic variability on drug responsiveness is potentially very great. Polymorphisms of - and -subunits of GABA-A receptors may alter responsiveness to barbiturates, whereas genetic variation in -subunits might be predicted to have the greatest impact upon responsiveness to benzodiazepines.
In general, traditional nonsedating ACDs such as phenytoin, carbamazepine, and valproic acid are thought to produce anticonvulsant effects by virtue of their ability to stabilize the inactive form of sodium channels in a voltage-dependent fashion. This action likely underlies the ability of these drugs to block sodium chan- nels in vitro and to block sustained high-frequency action potential firing in vivo (MacDonald and Kelly, 1993). Sodium channels are multimeric complexes encoded by multiple genes. Again, the large number of polymorphisms already dis- covered in sodium channel genes (Ludwig et al., 1998; Persu et al., 1999; Xu et al., 1999; Moulard et al., 2000) suggests that these molecules may have a role to play in determining individual differences in response to certain ACDs. Future studies will be aimed at correlating responsiveness in phenytoin- and carbamazepine-treated patients with specific genotypes and haplotypes associated with these candidate
337 Genetic influences on responsiveness to anticonvulsant drugs
genes. The actions of valproic acid may be even more varied than those of pheny- toin and carbamazepine, including an ability to block calcium channels (Kelly et al., 1990) and interfere with GABA metabolism (Johanessen, 2000). Therefore the potential sources of genetic variability for responsiveness to this agent are increased in number.
In addition to the conventional ACDs described above, compounds exist with mechanisms of action that either differ from standard drugs or are not yet fully characterized. These compounds and their mechanisms will lead eventually to the investigation of specific, possibly novel, genes in relation to drug responsiveness. Gamma-vinyl-GABA potentiates GABAergic transmission through its ability to inactivate irreversibly the GABA degradative enzyme, GABA transaminase (Jung and Palfreyman, 1995). Therefore, the gene that codes for this enzyme would be a logical candidate for influencing responsiveness to this agent. There are a number of drugs still in the process of gaining wide acceptance and these have been approved mainly for adjunctive therapy of partial and secondary generalized sei- zures (Bazil and Pedley, 1998) or are still considered investigational. These include compounds such as MK-801, felbamate, gabapentin, lamotrigine, topiramate, tia- gabine, retigabine, neurontin, zarontin, and riluzole. Since the mechanisms of action of these drugs have been studied less well (MacDonald and Kelly, 1993; Ikeda et al., 1998), candidate genes that may influence individual responsiveness to them are less well defined. As these drugs gain clinical acceptance and become more widely used, greater interest will be generated in determining genetic influences on patient responsiveness.
Hydantoins
A major polymorphism of oxidative metabolism in humans involves hydroxylation of the anticonvulsant drug mephenytoin. This racemic drug is metabolized in a highly stereospecific manner whereby the (s)-enantiomer undergoes rapid and complete oxidation to a p-hydroxylated derivative and the (R)-enantiomer is subject to a much slower pathway involving N-demethylation to an active metabo- lite 5-phenyl-5-ethylhydantoin (Guttendorf and Wedlund, 1992). Early pharma- cokinetic studies uncovered a familial concurrence of poor hydroxylation capacity, and subsequent population and family studies revealed autosomal recessive inher- itance patterns (Kupfer and Preisig, 1984; Wedlund et al., 1984). Lacking the capac- ity for p-hydroxylation of mephenytoin, poor metabolizers detoxify both enantiomers via N-demethylation, leading to very high 5-phenyl-5-ethylhydantoin levels and exaggerated or toxic responses (Kupfer et al., 1984; Ninomiya et al., 2000). Molecular studies have led to the discovery of CYP2C19 as the P-450 isoform responsible for the stereoselective hydroxylation of mephenytoin (Wrighton et al., 1993; Goldstein et al., 1994; Watanabe et al., 1998) and, further, have uncovered
338 T. N. Ferraro
several mutations that lead to the formation of a defective protein and that result in the poor metabolizer phenotype (de Morais et al., 1994; Ferguson et al., 1998; Ibeanu et al., 1998a,b, 1999). There is significant ethnic diversity in the frequency of these polymorphisms, with the poor metabolizer phenotype observed in 2–5% of Caucasian populations from North America and Europe (Kupfer and Preisig, 1984; Wedlund et al., 1984; Goldstein et al., 1997) and in 18–23% of populations from Japan and China (Horai et al., 1989; Goldstein et al., 1997; Xie, 2000). Individuals of other ethnic backgrounds are also characterized by specific CYP2C19 allele frequencies (Goldstein et al., 1997; Herrlin et al., 1998; Bathum et al., 1999; Xie et al., 1999). Even though substantial progress has been made in identifying the relationship between CYP2C19 genetics and effects of mephenytoin, the sequence variants discovered so far do not explain all of the individual variability seen clini- cally, suggesting that other factors exist which influence responsiveness to this drug.
Other drugs that bind to the mephenytoin p-hydroxylase isoenzyme include diazepam, flurazepam, and phenytoin, and drugs with metabolism that cosegre- gates with that of mephenytoin, include mephobarbital, hexobarbital, and diaze- pam (Inaba et al., 1985). Deficient p-hydroxylation of phenytoin has been observed in a small percentage of the population and appears to run in families (Clark, 1985); however, the major P-450 isoform involved in the p-hydroxylation of pheny- toin is CYP2C9, with CYP2C19 serving a secondary role (Levy, 1995; Hashimoto et al., 1996). The poor metabolizer phenotype is associated with increased pheny- toin-induced toxicity, also a familial trait (Gennis et al., 1991), whereas hypermeta- bolism results in treatment failure (Lebrun and Villeneuve, 1983), indicating that beneficial therapeutic effects as well as adverse effects are significantly influenced by genetic factors. CYP2C9 harbors a conservative Ile359Leu polymorphism, which dramatically alters metabolism of phenytoin (Hashimoto et al., 1996; Takanashi et al., 2000) as well as a more radical but less functionally important Arg144Cys polymorphism (Mamiya et al., 1998). The prevalence of these polymor- phisms vary with ethnicity (Bertilsson, 1995) and as a result many studies have focused on identifying polymorphism frequencies in specific populations (Mamiya et al., 1998; Aynacioglu et al., 1999; Bathum et al., 1999) since such data are useful and important for establishing treatment guidelines for individuals of specific ethnic backgrounds. The metabolizing capacity of the CYP2C9 system exhibits a gene dosage effect that distinguishes individuals homozygous for the wild-type allele from those that are heterozygous (Hashimoto et al., 1996). Again, while genetic variation leading to polymorphic metabolism has helped to elucidate the basis of individual differences in response to phenytoin, a significant amount of unexplained response variability still exists.
In order to address residual, nonmetabolic variation in responsiveness to pheny- toin, strategies are required that focus on pharmacodynamic aspects of drug action
339 Genetic influences on responsiveness to anticonvulsant drugs
and the potential genetically based variability in cellular targets related to therapeu- tic effects. Such strategies can be fueled efficiently by studying the sequences of genes with products that are known targets, such as sodium channel genes in the case of phenytoin. The next generation of studies will take advantage of DNA sequence polymorphisms that have already been reported for sodium channel genes (Ludwig et al., 1998; Persu et al., 1999; Xu et al., 1999; Moulard et al., 2000) as well as the many more that will doubtless be identified in the future. Comparison of allele frequencies between carefully matched groups of phenytoin-responsive and phenytoin-unresponsive patients can yield statistical evidence for significant effects of specific genes and specific sequence variants within those genes. This can- didate gene approach is useful when the intended drug targets are known; however, a limitation of this strategy is the possibility that the primary endogenous substrate of drug action has not been unequivocally identified. For example, although phen- ytoin interacts directly with sodium channels (MacDonald and Kelly, 1993), there is evidence that it also interacts with Na /K -ATPase (Guillaume, 1988). It is pos- sible that other as yet unknown targets for this drug also exist. In such cases, alter- native gene identification strategies are required.
Studies using experimental animals offer a means of studying further relevant genes identified in humans and also provide strategies for identifying genes for new or previously unknown drug target molecules. An approach that represents a pow- erful tool in this regard is called quantitative trait locus (QTL) mapping. One strat- egy for QTL mapping involves cross-breeding two strains of an organism that have divergent or opposite phenotypes for a given trait and using their second genera- tion progeny to map the location of genes that contribute to the phenotype through phenotype–genotype correlational analysis (Lander and Botstein, 1989), with the ultimate goal being gene identification. QTL studies of pharmacogenetic relevance can be facilitated by exploiting strain differences in responsiveness to specific drugs (Ferraro and Berrettini, 1996). In terms of ACDs, one line of previous work has focused on mapping genes influencing sensitivity or resistance to specific toxic effects, particularly those associated with fetal abnormalities. Strain differences in sensitivity to craniofacial malformation have been reported in mice exposed to phenytoin in utero (Hansen and Hodes, 1983; Brown et al., 1985) and have facili- tated mapping genes related to this trait (Karolyi et al., 1990). Although no genes have yet been identified, refined mapping and phenotyping procedures have deter- mined that there may be different genetic influences involved in the craniofacial endophenotypes cleft lip and cleft palate (Diehl and Erickson, 1997).
There is a paucity of published work regarding genetic influences on the phar- macodynamic actions of ACDs relative to their therapeutic effects, although one interesting line of research involves the development of phenytoin-sensitive and phenytoin-resistant strains of rats (Loscher and Rundfeldt, 1991). These
340 T. N. Ferraro
experiments were extended to show that rats selectively bred for extreme sensitiv- ity or resistance to phenytoin suppression of amygdala-kindled seizures respond in a correlated fashion to valproate and phenobarbital (Loscher et al., 1993), suggest- ing that similar genetic influences underlie responsiveness to ACDs with presum- ably different mechanisms of anticonvulsant action. More recent work has investigated the responsiveness of a variety of inbred strains of rats to phenytoin (Cramer et al., 1998), which will facilitate gene mapping.
Carbamazepine
Apart from the hydantoin-derived ACDs, other commonly prescribed nonsedating drugs are less well studied regarding metabolic polymorphisms. Carbamazepine conversion to its primary metabolite, carbamazepine-10,11-epoxide, has been shown to be mediated primarily by the CYP3A4 isoform (Kerr et al., 1994; Levy, 1995) although some biotransformation has been documented by CYP2C8 as well (Kerr et al., 1994). Idiosyncratic hypersensitivity reactions to carbamazepine are relatively common (McNamara, 1996); however, no data are available for review regarding associations between polymorphisms in P-450 isoforms and ability to metabolize carbamazepine. Nonetheless, several polymorphisms are indeed found in CYP3A4. One occurs in the 5 promoter region (Rebbeck et al., 1998; Westlind et al., 1999) and its frequency exhibits a striking ethnic specificity as it is especially prevalent in individuals of African descent (Ball et al., 1999). This variation does not appear to play a major role in determining constitutive CYP3A4 expression, however (Ball et al., 1999), and thus it may not significantly influence metabolism of phenytoin. Two other CYP3A4 alleles have been reported but they are very rare. Both are missense mutations, giving rise to Ser222Pro and Met445Thr (Sata et al., 2000), and thus have the potential to be functionally relevant. It is currently unknown whether these polymorphisms affect carbamazepine metabolism. Genetic influences on clinically relevant adverse effects of carbamazepine are doc- umented in a report of an idiosyncratic hypersensitivity reaction in a pair of iden- tical twins with primary generalized epilepsy (Edwards et al., 1999). Aplastic anemia is a clincally significant adverse effect of carbamazepine therapy, which may be related to the action of myeloperoxidase and formation of the 9-acridine car- boxaldehyde metabolite (Uetrecht, 1990). To-date, however, no studies have exam- ined the role of this or other genes that affect carbamazepine metabolism and the ability of this drug to induce aplastic anemia, although some evidence suggests that such an approach might prove useful (Gerson et al., 1983).
The paradigm shift towards identifying genes involved in therapeutic response of ACDs as opposed to polymorphic metabolism is represented by a recent study in which a mutated gene for a nicotinic receptor subunit associated with a rare form of epilepsy (Steinlein et al., 1995) was expressed in Xenopus oocytes. It was shown
341 Genetic influences on responsiveness to anticonvulsant drugs
that mutant receptors had acetylcholine-induced responses similar to wild-type cells but that mutants were three times more sensitive to inhibition by carbamaze- pine (Picard et al., 1999). This study indicates the importance of elucidating under- lying genetic basis of disease when focusing on pharmacogenetic aspects of drug action.
Studies relevant to the pharmacogenetics of carbamazepine that involve experi- mental animals are somewhat scarce. In one study, testing of three different rat strains showed that anticonvulsant effects of carbamazepine were strain dependent and correlated with drug levels (Graumlich et al., 1999). These data suggest that strain-specific pharmacokinetic processes may account for observed differences in response and, as such, they may be less useful regarding application to human research. In another study, ability of carbamazepine to suppress cocaine-induced seizures was tested in three strains of mice; however, all strains exhibited similar responsiveness (Marley et al., 1993), offering no interesting leads for gene mapping.
Valproic acid
Studies related to genetic influences on the effects of valproic acid are not abun- dant, particularly in humans. As with other ACDs, considerable inconsistencies between dose and serum concentration of valproic acid have been documented (Tisdale et al., 1992) and it is likely that this phenomenon has a genetic component that impacts individual response to treatment and contributes to unexplained response failures and toxicities (McNamara, 1996). Valproic acid metabolism is complex and the bioconversion process results in formation of a number of oxid- ative metabolites (Prickett and Baille, 1984), one of which, 2-n-propyl-4-pentenoic acid (4-ene-VPA), is associated with significant clinical hepatotoxicity (Rettie et al., 1987). So far, several P-450 isoforms have been implicated in valproic acid biotrans- formation including CYP2B and CYP4B (Rogiers et al., 1995; Guan et al., 1998), with some evidence to suggest that the CYP2B1 isoform is responsible for forming a metabolite with hepatotoxic properties (Rogiers et al., 1995). Additionally, other CYP isoforms have been implicated in the formation of 4-ene-VPA including CYP2C9 and CYP2A6 (Sadeque et al., 1997). Most of the isoforms involved in val- proic acid metabolism are polymorphic, including the CYP2B family with nine alleles (Lang et al., 2001), CYP2A6 with a loss-of-function missense polymorphism (Yamano et al., 1990) and several deletions (Nunoya et al., 1998; Oscarson et al., 1999), and, as described above, CYP2C9 with four alleles, each resulting in amino acid substitutions and altered enzyme activity (Rettie et al., 1994; Sullivan-Klose et al., 1996; Imai et al., 2000). In spite of the substantial amount of baseline data in the literature on these CYP polymorphisms, no studies have yet attempted to cor- relate relevant CYP genotypes in patients receiving valproic acid with drug blood levels or clinical response.
342 T. N. Ferraro
One line of study on pharmacodynamic issues of valproic acid treatment involves genetic models of neural tube defects in mice. These studies address the mechanism of teratogenic effects observed in valproic acid-exposed human fetuses. Support for a genetic component to valproic acid-induced developmental defects in the nervous system is provided by studies of inbred mouse strains that differ sig- nificantly in their sensitivity to defects induced by this drug, and further, by evi- dence that a close correlation exists between sensitivity to valproic acid-induced and spontaneous neural tube defects (Hall et al., 1997). Studies of sensitive and resistant mouse strains have also identified several provocative candidate genes that might be causally related to the development of valproic acid-induced congenital malformations including the gene for 5,10-methylenetetrahydrofolate reductase (Finnell et al., 1997), Hox genes (Faiella et al., 2000), the gene for the r1 subunit of ribonucleotide reductase (Craig et al., 2000), and numerous genes for neurotrophic factors (Bennett et al., 2000). So far, however, no studies that examine the role of these genes have been reported in humans.
Benzodiazepines
A wide variety of benzodiazepine compounds are available to treat neurological and psychiatric disorders, with diazepam and clorazepam being the most common ones used for treating patients with epilepsy (McNamara, 1996). While all benzo- diazepines are believed to exert their pharmacological effects by facilitating func- tion of GABA-A receptors (MacDonald and Kelly, 1993), some of them have other actions in the brain (Palmada et al., 1999). Additionally, the pharmacokinetic prop- erties of benzodiazepines vary (Baldessarini, 1996). Therefore, genetic variation potentially can affect response to this class of drug in many ways. Most promin- ently, these include polymorphisms in genes for both cytochrome P-450 drug- metabolizing enzymes and subunits of GABA-A receptors (see Ch. 15).
Although early evidence suggested that pharmacokinetic parameters of diaze- pam are not under genetic control (Alda et al., 1987), several polymorphic P-450 isoforms that significantly affect the oxidative metabolism of benzodiazepines have been reported more recently. One study in a Chinese population showed that the G681A nucleotide polymorphism of CYP2C19 segregated with impaired metabo- lism of both diazepam and desmethyldiazepam (Qin et al., 1999). Another study, in which European-American and African-American men were given midazolam revealed a modest association between systemic clearance of the drug and a poly- morphism (A290G) in the 5 regulatory region of CYP3A4 (Wandel et al., 2000). The guanine nucleotide, found only in the African-American group, was associated with a 30% reduction in systemic clearance (Wandel et al., 2000). Studies to eval- uate the potential significance of CYP2C and CYP3A polymorphisms on functional parameters of benzodiazepine action are needed.
343 Genetic influences on responsiveness to anticonvulsant drugs
Molecular studies of GABA receptors have begun to define not only the specific subunits required for benzodiazepine sensitivity (Rudolf et al., 1999) but also the specific amino acid residues that mediate GABA and benzodiazepine binding (Benson et al., 1998; Carlson et al., 2000; Hartvig et al., 2000). At the same time, a number of naturally occurring variations of GABA receptor subunit genes are being identified (Cheng et al., 1997; Iwata et al., 2000) and associations with a variety of central nervous system diseases are being investigated (Sander et al., 1996, 1999; Gade-Adavolu et al., 1998; Serretti et al., 1998; Loh et al., 1999, 2000). So far, however, little information regarding the possible association between GABA receptor subunit gene variation and anticonvulsant responses to benzodiaz- epines is available. One study investigating the influence of a Pro385Ser (C1236T) variation on the sensitivity of oculomotor responses to diazepam showed that the variant serine residue was associated with significantly less diazepam-induced impairment of saccadic velocity (Iwata et al., 1999). Such results provide impetus to examine potential associations between GABA receptor subunit polymorphisms and therapeutic responses to benzodiazepines.
Contributions of animal studies to the pharmacogenetics of benzodiazepines rel- evant to their behavioral effects have come primarily from comparisons of inbred strains and from studies of other specialized strains such as knockouts and transgen- ics. Studies of single-gene mutants can evaluate the effects of a defined genetic poly- morphism. For example, deletion of the long splice variant of the gene for the 2-subunit of the GABA-A receptor yielded a strain of mice that showed enhanced sleep times with both midazolam and zolpidem (Quinlan et al., 2000), demonstrat- ing the importance of the action and processing of this gene in the regulation of ben- zodiazepine sensitivity. Studies of the effects of benzodiazepines on inbred strains have been much more common. Strain comparisons are useful because they provide a foundation for subsequent studies to identify the genes involved in mediating spe- cific responses with no preconceived notion as to which genes these might be. Behavioral responses to benzodiazepines are complex traits determined by a mix of environmental and genetic influences (Crabbe et al., 1998). If a specific trait or response to drug has a sufficiently large genetic component, mapping studies can be undertaken to determine the genomic location of the major genes involved; however, once genetic mapping is complete, the identification of specific causative genes remains a major task. The correlation of genotype with complex trait phenotype and the resolution of gene mapping is generally insufficient to reduce the chromosomal interval containing the gene of interest to a size small enough to allow positional cloning even for genes that exert very large effects. Consequently, genetic mapping of complex traits in experimental animals must be followed up by other studies that involve systematic reduction of critical intervals and, ultimately, functional analysis of single genes with rigorous quantification of their phenotypic influence.
344 T. N. Ferraro
The use of selectively bred strains of experimental animals has facilitated study of genetic influences on benzodiazepine-associated traits. A particularly useful resource are lines of mice bred specifically for sensitivity and resistance to the ataxic effects of diazepam (Gallagher et al., 1987; Yoong and Wong, 1988). Some of these lines have generated evidence suggesting that common genetic mechanisms are involved in determining responsiveness to benzodiazepines and to volatile anes- thetic compounds such as halothane and enflurane (McCrae et al., 1993). The fact that strain differences may be a result of functional differences in neuronal sensi- tivity (Quinlan et al., 1993) rather than pharmacokinetics (Gallagher et al., 1987) underscores the importance of identifying the genes that mediate these discordant phenotypes. Identification of such genes could significantly advance the therapeu- tic use of benzodiazepines. Similarly, genes relevant to the anticonvulsant action of benzodiazepines may be amenable to identification using specially bred “ethanol withdrawal seizure prone” and “ethanol withdrawal seizure resistant” mice, since these respective lines differ with regard to the ability of a wide variety of benzodi- azepines and barbiturates to increase electrical seizure threshold (Crabbe et al., 1986, 1993). Wistar rat lines selectively bred for high- and low-anxiety-related behavior also differ in their sensitivities to the pharmacological effects of diazepam (Liebsch et al., 1998) and could also be exploited to identify causative genes.
Common, commercially available strains of experimental animals have also pro- vided useful data regarding genetic control of benzodiazepine effects. Inbred strains of mice have been a particular focus and several reported surveys document significant strain differences in behavioral and physiological responses to a variety of benzodiazepines including diazepam (Crabbe et al., 1998; Garrett et al., 1998; Griebel et al., 2000), phenazepam (Seredenin et al., 1990), alprazolam (Weizman et al., 1999), and midazolam (Homanics et al., 1999). Strain differences in benzodi- azepine receptor expression may contribute to these opposite behavioral pheno- types (Hode et al., 2000), although the multifactorial nature of drug responsiveness assures that other mechanisms are also involved. Assuming that drug response traits can be shown to be sufficiently heritable, strain survey data can provide a strong foundation to begin to consider strategies for mapping associated genes.
Barbiturates
Like benzodiazepines, the variety of different barbiturate compounds available for clinical use is associated with a complex pharmacokinetic profile for the drug class as a whole. Regarding their use as anticonvulsants, phenobarbital and mepho- barbital have found the widest application (Porter et al., 1984; McNamara, 1996). GABA-A receptors are the main target of barbiturates and facilitation of chloride flux through these receptors is believed to be largely responsible for their pharmaco- logical and therapeutic effects (MacDonald and Kelly, 1993; Hobbs et al., 1996).
345 Genetic influences on responsiveness to anticonvulsant drugs
Therefore, polymorphism in genes that code for GABA-A receptor subunits is a potentially important source of variation in barbiturate treatment response. Nonetheless, other mechanisms and genes may also be relevant (Steinbach et al., 2000).
As is true for other pharmacological classes, large individual differences in response to the clinical effects of barbiturates have been documented previously (Peck et al., 1976; Nightgale, 1988). Studies in monozygotic and dizygotic twin pairs indicate that the rate of metabolism of amobarbital is under genetic control (Endrenyi et al., 1976), and more recent studies have focused on the role of poly- morphic cytochrome P-450 isoforms in barbiturate metabolism. Hydroxylation of hexobarbital was shown to be correlated with hydroxylation of mephenytoin (Kato et al., 1992) but not phenobarbital (Schellens et al., 1990), documenting important kinetic differences between compounds closely related pharmacologically and structurally. The CYP2C isoform family has an integral role in barbiturate metab- olism since it influences disposition of hexobarbital (Yasumori et al., 1990; Adedoyin et al., 1994) and phenobarbital (Mamiya et al., 2000). Mephobarbital N- demethylation is mediated by CYP2B6 (Kobayashi et al., 1999). The relationships between polymorphisms in CYP isoforms and patient responsiveness to barbit- urates remain to be studied.
Studies relevant to genes that influence the action of barbiturates have been carried out with common inbred strains and selectively bred lines of mice. Mice selectively bred for long sleep time and short sleep time in response to ethanol are resistant and sensitive, respectively, to the hypnotic (sleep-inducing) effects of thio- pental, phenobarbital, and chlordiazepoxide (McIntyre and Alpern, 1986). These results suggest that a subset of genes with alleles that segregated between the two selectively bred lines mediate responses to a range of central nervous system depressants. The concept that a subset of genes influences response to different types of depressant drug is also supported by studies of withdrawal seizures. Results of experiments involving DBA and C57 mice, inbred strains that are sensitive and resistant to experimental seizures, respectively (Engstrom and Woodbury, 1988; Ferraro et al., 1998, 1999), suggest that similar genes underlie sensitivity to seizures induced by withdrawal from ethanol, pentobarbital, and diazepam (Buck et al., 1999; Metten and Crabbe, 1999). Identification of these genes could provide important information relevant to the therapeutic use of barbiturates and benzod- iazepines in so far as understanding genetic susceptibility to adverse effects such as physiological dependence will help to refine treatment regimens for individual patients.
The ability of phenobarbital to induce hepatic microsomal enzyme activity is a well-known factor in studies of drug metabolism and can affect the biodisposition and efficacy of other drugs administered concomitantly (Anderson and Levy,
346 T. N. Ferraro
1995). Great interest exists in defining the properties of genes that have their regu- lation affected by phenobarbital; recent work indicates the importance of specific 5 promoter sequences (Ramsden et al., 1999). In vitro studies have examined the effect of phenobarbital on the induction of known genes (Runge et al., 2000; Schilter et al., 2000) whereas newer in vivo approaches enable detection of unknown as well as known genes (Garcia-Allen et al., 2000). Once identified, all of these genes can become candidates to be examined with regard to influences of their variation on barbiturate responsiveness.
Miscellaneous drugs
Among drugs with a longer history of use in the treatment of epilepsy, several studies involving the drug acetazolamide are of interest. These include the success- ful use of this drug in treating two siblings affected with familial hemiplegic migraine and ataxia resulting from a missense mutation in CACNA1A, the gene for the calcium channel (Battistini et al., 1999) and another study in which two strains of mice (one sensitive and one resistant) were used to map genetic loci involved in acetazolamide-induced ectrodactyly (Biddle, 1975). In vitro studies on metabolism of trimethadione suggest that N-demethylation of this compound is mediated pri- marily by CYP2E1 with some participation by CYP2C8 (Kurata et al., 1998). Functional polymorphisms within CYP2E1 (Hu et al., 1997) suggest one factor that could influence responsiveness to trimethadione.
Studies of pharmacogenetic relevance to newer ACDs are scarce although a strong foundation for future studies is being developed. For instance, recent work documenting that felbamate blockade of NMDA receptors depends on the presence of the NR2B subunit (Harty and Rogawski, 2000) combined with the recent dis- covery of a silent mutation in the human gene NR2B (Nishiguchi et al., 2000) paves the way for studies to examine the effect of this polymorphism in patients treated with felbamate. Similarly, the ACD gabapentin has been shown to bind an 2- subunit of the calcium channel subunit (Wang et al., 1999), and polymorphisms of genes for related calcium channel subunits, CACNA1A (Yue et al., 1997) and CACNL1A2 (Yamada et al., 1995), have been reported. Genes for specific potassium channels, KCNQ2 and KCNQ3, have been identified as a likely target for the novel ACD retigabine (Main et al., 2000; Wickendon et al., 2000). Since a rare inherited form of epilepsy called benign familial neonatal convulsions has been shown to be caused by missense mutations in these genes (Leppert et al., 1989; Lewis et al., 1993), it is possible that epilepsy patients harboring variants of these or related genes would respond in a unique fashion to drugs such as retigabine. Genes for potassium channel subunits have so far proven to be a rich source of functional polymorphisms (Lai et al., 1994; Sakura et al., 1995; Hansen et al., 1997; Derst et al., 1998; Mylona et al., 1998; Laitinen et al., 2000; Sesti et al., 2000; Vaughn et al.,
347 Genetic influences on responsiveness to anticonvulsant drugs
2000) and it is likely that drugs designed against potassium channel proteins will be differentially effective in their interaction. MK-801, an investigational com- pound with anticonvulsant properties, has been the subject of several mouse strain surveys regarding its behavioral effects. Strain-dependent effects of MK-801 for a number of phenotypes have been observed including inhibition of electrical seizure threshold (Deutsch et al., 1998), passive avoidance behavior (Cestari et al., 1999), and “popping” behavior (Deutsch et al., 1997).
Conclusions
Pharmacogenetic studies involving ACDs have focused strongly on polymorphic metabolism, and a number of important associations have been uncovered. Nonetheless, substantial unexplained variability in responsiveness of patients to conventional drug treatments remains and it is probable that additional, as yet undiscovered, genetic factors are involved. Future studies will continue to investig- ate associations between genes relevant to ACD pharmacokinetics and drug blood levels; however, we need to begin to correlate genotypes with more functional measures of ACD activity. We also need more emphasis on possible associations between variations in genes for drug targets and ACD responsiveness. As genetic polymorphism information on these genes becomes available, it will become feas- ible to carry out studies involving ACD-treated responsive and unresponsive patients. For those newer drugs with unclear modes of therapeutic action, genetic studies must await further elucidation of drug mechanisms before a candidate gene approach is feasible.
When whole genome scans using dense SNP maps become a reality, there will be less need for candidate gene strategies; however, the current state of the art in complex trait genetics involves association analysis requiring candidate genes. Presently, candidate genes for human association studies derive from a knowledge of drug action and kinetics, or may come from animal models. In either case, iden- tification of specific genetic variation that influences a complex trait requires a combined mathematical and biological approach that starts by determining the relationship between a set of genotypes and phenotypes (Olsen et al., 1999). Several different methods of correlating genotype with complex trait phenotypes that have been developed are used for complex trait analysis in humans (Clayton and Jones, 1999; Page and Amos, 1999; Blangero et al., 2000). Studies in animals can yield important information about complex trait genes and can provide gene candidates for human studies (Ferraro and Berrettini, 1996). On the horizon are new strate- gies involving experimental animals that promise to enhance further gene mapping and identification in complex trait models (Nadeau and Frankel, 2000; Nadeau et al., 2000; Grupe et al., 2001).

348 T. N. Ferraro
In conclusion, it is clear that the future development of pharmacological strat- egies to treat disorders of the central nervous system will be most successful when these strategies begin to incorporate knowledge of the effects of genetic variation on individual drug responsiveness. In the case of ACDs, it is likely that patients with seizure disorders will be the primary focus of research; however, it will be crucial to extend these studies to patients with affective disorders, particularly bipolar dis- order, since there is a continuing increase in the use of ACDs in this population.
Acknowledgements
Sincere thanks to Russell J. Buono, PhD (Department of Psychiatry, University of Pennsylvania, Philadelphia, PA) and James J. Kocsis, PhD (Department of Biochemistry and Molecular Pharmacology, Thomas Jefferson University, Philadelphia, PA) for suggestions, discussion, and critical reading of this manuscript.
REFERENCES
Adedoyin A, Prakash C, O’Shea D, Blair IA, Wilkinson GR (1994). Stereoselective disposition of hexobarbital and its metabolites: relationship to the S-mephenytoin polymorphism in Caucasian and Chinese subjects. Pharmacogenetics 4, 27–38.
Alda M, Dvorakova M, Posmurova M, Balikova M, Zvolsky P, Filip V (1987). Pharmacogenetic study with diazepam in twins. Neuropsychobiology 17, 4–8.
Anderson GD, Levy RH (1995). Phenobarbital: chemistry and biotransformation. In Levy RH, Mattson RH, Meldrum BS, eds. Antiepileptic Drugs, 4th edn. New York: Raven Press, pp. 371–377.
Aynacioglu AS, Brockmoller J, Bauer S et al. (1999). Frequency of cytochrome P-450 CYP2C9 variants in a Turkish population and functional relevance for phenytoin. Br J Clin Pharmacol 48, 409–415.
Baldessarini RJ (1996). Drugs and the treatment of psychiatric disorders. In Hardman JG, Limbird LE, eds.