Genetics and Psychopharmacology: Prospects for Individualized Treatment

Abstract

This article provides a clear and succinct description of the components of inheritance, such as trait transmission, genetic variability, and gene interaction. Genetic sequences constitute the prime focus of pharmacogenetic studies. Variations in drug-metabolizing enzyme systems tend to be monogenic, whereas the pharmacologic effects of medications appear to be polygenic, i.e., complex phenotypes shaped by the interaction of genes and environment. Translated into clinical terms, a history of a good response to a drug in a close relative of a patient is presumed to indicate a good response to the same medication by the patient. This seems to hold for antidepressants, antipsychotics, and lithium, but the evidential studies generally have meaningful limitations. Bit by bit, information about the relationship between particular genetic formations and the effectiveness of these medications as well as their side effects, is appearing. The authors cite a number of examples, one such being an association between impaired antidepressant activity and the short allele of SLC6A4. This research promises to strengthen the accuracy, effectiveness, safety, and cost of our psychopharmacological practices.

Introduction

Considerable heterogeneity exists in the way individuals respond to psychotropic drugs.^1–3 This is a major source of concern for the skillful clinician, especially in a managed care environment, because the consequences are potentially devastating in patients who are seriously mentally ill.⁴ Interindividual variability in psychotropic drug response might result in lengthy drug trials, more hospital days, drug side effects, unnecessary drug exposure, and unremitting symptoms that could result in suicide.¹ Thus, ready availability of accurate methods of predicting the therapeutic response to psychiatric medications is important—more so nowadays than before, given the wide range of psychotropic medications that have been introduced since the early 1950s.⁵

Currently, the initial choice of psychopharmacotherapy is based on clinical variables such as the individual’s age, sex, premorbid functioning, illness severity, comorbidity, and liver and renal function.^6,7 Clinical parameters are also used to declare the treatment a failure, sometimes with inevitable misclassification errors.⁸ Not surprisingly, this practice has led to psychiatric morbidity, mortality, and cost-effectiveness that are less than desirable in the 21st century.^9,10

Pharmacogenetics can be used to individualize drug therapy, using molecular genetics to predict the likelihood of a response and the risk for toxicity before the medication is dispensed.¹¹ Worldwide interest in pharmacogenetics is intense, and the field is poised to take full advantage of recent advances in genomics and psychopharmacology.^12,13 The major strength of the molecular genetics approach lies in the accuracy with which investigators can document genotype using such techniques and the observation that treatment with psychotropic drugs does not affect the genotype of individuals.⁶ However, there are important barriers to translating the potential of pharmacogenetics into reality. For instance, the field of psychiatry is still refining how it defines and ascertains some of its key phenotypes, including diagnosis, drug response, and side effects. Some of these conditions also entail a complex relationship between genotype and phenotype. These constraints, which have not allowed centers world-wide to carry out investigations in a consistent manner across all pharmacogenetic studies, have resulted in the failure to replicate some of the results in follow-up studies. In addition, the overall success of pharmacogenetics will depend on technological advances because of the need for highly automated systems that can minimize the cost of genotyping each sample examined and maximize the flexibility of studying candidate genes.

In this article, we provide a review of the current state of research in the pharmacogenetics of psychotropic drug response. We conducted a Medline database search on psychiatric pharmacogenetics for the period of 1966 to 2004. The existing data clearly indicate that the field of pharmacogenetics is still evolving and ultimately should provide clinicians with a simple prospective tool to identify patients who will respond to psychopharmacotherapy with specific benefits or adverse events.

The Genetic Basis of Inheritance: An Overview

To fully appreciate the complex premise and clinical applications of pharmacogenetics in psychiatry, an understanding of the components of inheritance—such as trait transmission, genetic variability, and gene interactions—is essential.

Trait Transmission

Broadly speaking, traits might pass from parents to offspring by means of monogenic (Mendelian), environmental, or polygenic inheritance. However, genotype–phenotype associations might occur with no direct cause–effect relationship. For example, 2 or more genes that are sufficiently close on a chromosome may be inherited together following meiosis.

Meiosis is the process of cell division that gives rise to the gametes. A key step in meiosis is a process known as meiotic recombination, which involves identical pairs of chromosomes. In the process, homologous chromosomes pair up, crossover, and exchange bits of genetic material at the points of crossover events without the gain or loss of any genes. Thus, genes that are close enough on a chromosome can show a statistical association termed linkage disequilibrium that is based on a lack of recombination events in between. This lack of historical (ancestral) recombination has some biological implications. For example, traits that are inherited as a group are often related to genes that are near one another on the same chromosome. Similarly, a gene that has no known causal effect may be associated with a biological trait only because of its proximity to the true culprit. Such data can be used in gene-hunting studies.

Genetic Sequences

Three regions of the gene—namely, the regulatory, the coding (exon) and the noncoding (intron) sequences—constitute the prime focus of pharmacogenetic studies. These regions are important because they form the physical basis of the human genotype. The regulatory portions of genes usually contain a promoter sequence that initiates transcription with the subsequent production of proteins. During transcription, the exons and introns are initially copied onto the pre-messenger RNA (pre-mRNA). Posttranscriptional processing of the pre-mRNA involves the removal of the introns (the so-called junk DNA) and exon splicing to make a “mature” mRNA. Thereafter, the information in the mature mRNA is translated within the ribosomes to synthesize the protein.

The process of protein production follows the genetic code faithfully. The reading of genotypic information and the use of the information to produce proteins that underlie cellular activities are essential for development and reproduction. In this regard, a variation in the promoter region can influence the quantity of proteins that are synthesized, whereas a variation in the exon might result in the production of unsuitable proteins. However, it is important to note that the genes, especially the active genes that code for proteins, have a proofreading mechanism that maintains the fidelity of the information they carry. Errors that are missed by the “maintenance crew” can lead to the production of functionally defective proteins.

Interestingly, certain junk DNA sequences (introns) are transcribed and retained in the mature mRNA sequence in an alternative splicing process termed exonization. The exonization of intron elements into a coding sequence is thought to play a role in protein diversity in primates that is yet to be clarified.^14,15

Genetic Variations

Genomic instability is well recognized in the fields of evolutionary biology and cancer genetics.¹⁶ Some recent studies suggest that current investigative methods underestimate the frequency with which various kinds of genomic changes occur.¹⁷ Nevertheless, allelic variations with stable population frequencies of 1% or more are known as polymorphisms. Single nucleotide polymorphisms (SNPs), insertion/deletion sequence polymorphisms, and repeat polymorphisms are common examples. Two types of repeat polymorphisms have been described: variable number of tandem repeats (VNTR) and simple tandem repeats.

Polymorphisms might result in functional or nonfunctional gene products (i.e., proteins). Pharmacogenetics is primarily concerned with functional polymorphisms. Inasmuch as the identity of an individual can be tracked from parents to offspring using unique polymorphisms, regulatory bodies scrutinize human genetics research carefully to address important ethical concerns.

Gene-by-Gene and Gene-by-Environment Interactions

Family, twin, and adoption studies are important aids in the field of behavioral genetics. Family studies are used to establish familial traits. Certain characteristics might run in families owing to a shared environment, however. The possibility of a genetic contribution to a familial trait might be excluded in twin studies based on identical twins raised in the same environment for whom different trait frequencies are reported. Conversely, environmental influences on a familial trait may be ruled out in adoption studies carried out in identical twins who were raised in different environments but have similar frequencies for the same familial characteristic.

Genetic variations in drug-metabolizing enzyme systems tend to be monogenic, but the pharmacologic effects of medications appear to be polygenic.¹⁸ Polygenic inheritance results in complex phenotypes that are shaped by the interaction of genes and environment. Indeed, genes appear to have only a partial role in complex context-dependent phenotypes such as drug pharmacodynamics (what the drug does to the body) and pharmacokinetics (what the body does to the drug). Also, for genes that have multiple SNPs, a combination of SNPs (haplotype) might determine the phenotype rather than one SNP alone. Furthermore, the influence of multiple genes on a phenotype might be through additive or nonadditive (epistatic) mechanisms. For common diseases such as schizophrenia and cancer, there is a nonlinear relationship between genotype and phenotype. This may be due to 1 or more genes masking or enhancing the effects of some other gene(s) in the biological trait. This phenomenon is called epistasis.

Genes contributing to polygenic or complex traits are also called quantitative trait loci (QTLs). Typically, each QTL (e.g., for height) has a normal distribution in the population. There are multiple constraints on QTLs, and this has led to a resurgence of interest in gene–environment interactions.^18,19

Heritability of Pharmacogenetic Phenotypes

The Predictive Power of Drug Response Polymorphisms

The ability to detect genes depends on the allele frequency, trait heritability, and quality of the phenotypes. Investigators carrying out pharmacogenetic studies have hypothesized that variations in drug responses and adverse reactions are under genetic control, suggesting that culprit genes are easier to identify when their phenotypic effects are strong.

As described for familial traits in the preceding section, inherited polygenic traits may arise as a result of genetic factors or being in a shared or unique environment, the latter consisting of environmental factors that are not shared by genetically identical individuals.

Psychotropic Drug Response Heritability

The findings of epidemiologic studies support the use of psychopharmacological criteria to select drug responses that are genuinely genetic in nature. In general clinical practice, a history of a good drug response in a close relative of a patient is presumed to be an indicator of a good response to the same medication by the patient.

Antidepressant pharmacoepidemiologic studies conducted in the 1960s provided limited support for the heritability of the therapeutic effects of psychotropic drugs. These early studies reported a statistically significant response concordance between patients and first-degree relatives treated with antidepressant drugs.^20–23 However, these studies have been criticized for the failure of the investigators to control for illness course, case variability, drug heterogeneity, and other nongenetic factors that might influence the treatment response.

Only sparse data on the heritability of neuroleptic drug responses are available. However, a role for genetics in the response to antipsychotic drugs is suggested by the differences in response to antipsychotic drugs that have been observed among various ethnic groups.^24–26 Frackiewitz and colleagues provided some important data regarding ethnicity, but failed to clarify the role of disparate antipsychotic treatments.²⁴ Ruiz and coworkers suggested that the antipsychotic drug dose needed to control schizophrenia symptoms was lower in Hispanic patients than in African-American or white patients.²⁵ Emsley and colleagues reported that the acute response to antipsychotic treatment in schizophrenia was greater in African-American patients than in white patients of European descent.²⁶ However, their findings might have been altered if they had controlled for dietary intake, nutritional status, body mass, and substance use, each of which differentially affects the pharmacokinetics of antipsychotic drugs.²⁶ DeLisi and coworkers found no significant concordance for the heritability of antipsychotic response in sibling pairs.²⁷ Consequently, these studies provided a rudimentary indicator of the effects of genetics on the antipsychotic medication responder profile.

Although only scant data exist concerning the heritability of mood stabilizer response, family, twin, and adoption studies have shown that the drug response in bipolar disorder may be heritable. Grof and associates²⁸ evaluated the response to lithium in 24 individuals with bipolar disorder who were related to 106 lithium-responsive patients with bipolar disorder (N = 106). A series of 40 consecutive lithium-treated patients was used as a comparison group. The proportion of lithium responders was higher among the relatives of lithium-responsive patients than among patients in the comparison group, which suggests that the response to lithium might be a heritable trait. Sampling bias was a key issue with this study.²⁸ Other studies suggest that the distribution of lithium ions in the body might be genetically determined.^29,30 Taken together, the pharmacoepidemiologic studies of antipsychotic, antidepressant, and antibipolar medications provide strong preliminary evidence that the response to psychotropic drugs is heritable.

Methods Used in Pharmacogenetic Studies

There are, generally speaking, 2 ways in which investigators have approached research questions in pharmacogenetics. One approach to pharmacogenetic investigations uses the case-control association design to compare the frequencies of a candidate gene among unrelated patients who are treatment responders (“cases”) and nonresponders (“controls”). Studies of adverse reactions might examine treatment dropouts (“cases”) versus completers (“controls”).

Case-control association studies are hypothesis driven. The candidate gene with a known or putative effect on the mechanism of action of the drug is generally preferred for these studies. Because disease pathways and drug target sites are not always the same, genes that are linked to disease mechanisms are not always good candidates for pharmacogenetic studies.

The candidate gene approach may lead to spurious associations if there is unsuspected population stratification (subdivisions in the population owing to nonrandom mating) or admixture (the result of mating between subdivisions in the population). Thus, ethnic groups can have different allele frequencies at randomly chosen loci, and the early studies were criticized for ethnic mismatch between the cases and the controls.

To address this issue, another approach to pharmacogenetic studies—family-based genome-wide scans—was proposed. Genome-wide scans are used to compare the DNA in the patients with that of their parents, based on treatment response profiles. Unlike case-control studies, genome-wide studies are not based on the hypothesis concerning any particular gene. However, the genotype information from the parents has not always been readily available for genome-wide scans, which has added to their expense.

Genotyping procedures using leukocyte DNA are available. The laboratory methods are beyond the scope of this review.

Evidence-based Prediction of Response to Psychotropic Drugs

Use of Genotype to Predict Response to Antipsychotic Medications

The key phenotypes that have been observed in psychiatric pharmacogenetic studies are antipsychotic drug efficacy, antidepressant drug efficacy, and adverse reactions to psychotropic drugs.

Most pharmacogenetic studies of antipsychotic drugs have been based on the atypical antipsychotic clozapine because of its unique efficacy in treatment-resistant schizophrenia. The early studies were carried out in patients enrolled in clinical trials or were based on retrospective data on patients who were treated with clozapine. These studies suffered from the effects of selection bias based on age, ethnicity, and diagnostic heterogeneity.

The prime targets of clozapine studies were candidate polymorphisms of serotonin and dopamine receptor genes.³¹ Arranz and colleagues³² examined the T102C polymorphism in the coding region of the serotonin 5-HT2A receptor on chromosome 13p (p for petit). These investigators found that a significant proportion of clozapine nonresponders were homozygous for the 102C allele.³² Other research groups were unable to replicate these findings in small clozapine studies.³³ However, when Arranz and colleagues re-examined the T102C alleles in 2 clozapine studies,^34,35 they concluded that their association with the clozapine response was probably not spurious. One of the studies showed a significant association between clozapine response and a functional promoter polymorphism— −1438 guanin-adenine—on the 5HT2A gene.³⁴ The other study was a meta-analysis of the 5HT2A gene in 373 clozapine responders and 360 nonresponders and showed a clear association between the response and the presence of a T102C polymorphism.³⁵

Most studies, but not all,³⁶ have reported no significant relationship between clozapine response and each of the common polymorphic markers on the gene for any other serotonin receptor subtypes (5-HT2C, 5-HT3A, 5-HT3B, 5-HT6, 5-HT7) or for the serotonin transporter (SLC6A4).^37–39 Interestingly, 5HT2C binds preferentially with clozapine, but no linkage to schizophrenia has been reported for this receptor.⁴⁰

Mixed results have been reported for dopamine receptor polymorphisms and the therapeutic response to antipsychotic drugs. For example, despite the ability of clozapine to prevent the high-affinity binding of dopamine with its D4 receptor (DRD4), clozapine studies have failed to demonstrate a role for the repeat polymorphism on exon 3 of the DRD4 in treatment response. In one preliminary report, a novel repeat polymorphism in the first intron of DRD4 did predict the clozapine efficacy; however, the reliability of this finding is limited by the small sample size used in this study.⁴¹

Relatively few pharmacogenetic studies have focused on the D2 receptor (DRD2), even though several antipsychotic drugs bind with this receptor and neuroimaging reports suggest that DRD2 affinity might be related to the efficacy of these drugs. Schafer and colleagues⁴² have studied the DRD2 Taq I polymorphism and short-term response to haloperidol (Haldol), a strong DRD2 antagonist. They found that homozygosity for the A2 alleles predicted a poorer response.⁴² The association of poorer response with high-affinity DRD2 antagonists and the A2/A2 genotype was strengthened in a risperidone (Risperdol) study that showed similar results. In addition, a 141C-insertion/deletion polymorphism in the DRD2 promoter region was linked to response to treatment with bromperidol (Impromen, Bromidol) or nemonapride (Sepan),⁴³ but not clozapine.⁴⁴ These data need to be verified by studies using larger sample sizes.

Use of Genotype to Predict Response

Pharmacokinetic Effects:Antidepressants

Most pharmacogenetic studies of the efficacy and side effects of antidepressants have focused on pharmacokinetic effects (representing hepatic drug-metabolizing enzyme system activity) and pharmacodynamic effects (representing the effects of target receptor proteins, membrane transporters, and signal transduction systems).

Drug-metabolizing enzymes are classified as phase I or phase II enzymes. Phase I enzymes are sometimes referred to as mixed-function oxidases, because they oxidize 2 substrates at the same time. Phase II enzymes are responsible for the conjugation of xenobiotic substrates.

The cytochrome P-450 enzyme system catalyzes a huge proportion of the oxidative metabolic reactions in the liver. Among the cytochrome P-450 enzymes, the 2D6 (CYP2D6) isozyme is very important in the pharmacogenetics of antidepressant drugs.⁴⁵ The CYP2D6 isozyme—also known as desbrisoquine hydroxylase—is a highly polymorphic enzyme with more than 80 allele variants.^46–49 For current human cytochrome P450 allele nomenclature visit, http://www.imm.ki.se/CYPalleles/.

The variability in CYP2D6 expression and function has been used to classify individuals as ultrarapid, extensive, intermediate, or poor metabolizers. For example, the presence of 2 null alleles confers undetectable CYP2D6 activity, or a poor metabolizer phenotype.⁴⁹ Heterozygosity for the allele with null function or homozygosity for the alleles with intermediate function results in an intermediate metabolizer phenotype, which is characterized by severely impaired metabolic activity compared with that of the normal “extensive” metabolizers. Ultrametabolizers carry multiple copies of the gene for desbrisoquine hydroxylase.^50,51 Duplication-negative ultrarapid metabolizers have also been described.⁵²

Studies of the influence of CYP2D6 alleles on plasma concentrations of tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) have suggested the possibility of predicting side effects with high plasma drug levels on the basis of genotype.⁵³ Some studies, but not all,⁵⁴ have reported a significant correlation among CYP2D6 alleles, plasma levels of antidepressant medications, and side effects. Indeed, the poor metabolizer and the ultrametabolizer phenotypes appear to carry a greater burden in the cost of treatment because of adverse events, lack of drug benefit, or both.^55,56 The general safety of the SSRIs appears to offset the risk for adverse events associated with extreme metabolic phenotypes.¹

Inhibitors of the CYP2D6 enzyme might convert individuals who are extensive and intermediate metabolizers into poor metabolizers. The SSRIs fluoxetine (Prozac) and paroxetine (Paxil) are more potent inhibitors of CYP2D6 than setraline (Zoloft). Concomitant therapy with drugs that are CYP2D6 substrates is prevalent among patients receiving SSRIs.⁵⁷ Thus, it is judicious to carefully monitor all patients receiving medications, including those taking SSRIs alone or in combination with a CYP2D6 substrate such as a TCA or an antihypertensive or antiarrhythmic drug.^58–60 Murphy and coworkers⁵⁴ reported that concomitant use of medications that are metabolized by CYP2D6 did not result in significant adverse events in elderly patients with depression being treated with paroxetine for 8 weeks.⁵⁴ Zourkova and coworkers⁶¹ studied 30 patients receiving long-term paroxetine therapy and found that those who converted to the poor metabolizer phenotype reported more symptoms of sexual dysfunction than those who did not convert.⁶¹

Another important cytochrome P-450 enzyme, CYP3A4, catalyzes more drug substrates than all the other isoforms put together⁶² and has more than 40 polymorphisms. The CYP3A4 isozyme is involved in the metabolism of methadone (Dolmed), diazepam (Valium), and some antidepressant drugs (e.g., amitriptyline [Elavil]).⁶³ CYP3A4 polymorphisms have not been studied adequately in pharmacogenetic laboratories. Reports of potent inhibitors of CYP3A4 have included ketoconazole (Nizoral), protease inhibitors, and nefazadone (Serzone), but not setraline, fluoxetine, or venlafaxine (Effexor).⁶⁴ No pharmacokinetic interaction was demonstrated during coadministration of ritonavir (a protease inhibitor) and the SSRI escitalopram (Lexapro).⁶⁵

Pharmacodynamic Effects: Antidepressants

Pharmacodynamic pharmacogenetic studies have been conducted to identify genetic predictors of the antidepressant drug response. Smeraldi and others showed that the “short” form of the serotonin transporter gene promoter insertion/deleltion polymorphism (SLC6A4) predicts a poor response to the SSRI fluvoxamine (Luvox).⁶⁶ Subsequent studies in geriatric⁶⁷ and nongeriatric patients with depression⁶⁸ produced data indicating that the short allele impairs the response to another SSRI, paroxetine. Serretti and coworkers⁶⁹ used a sample of 221 patients with depression (128 with major depression; 93 with bipolar depression) to demonstrate a relationship between impaired antidepressant activity and the short allele of SLC6A4. However, this association was not observed in a study of Asian patients treated with paroxetine or fluoxetine.⁷⁰ Interestingly, the short allele of SLC6A4 also appears to induce vulnerability to affective disorders.^19,71

Some reports have suggested that the response to antidepressant drugs might be associated with common polymorphic genes for tryptophan hydroxylase, the 5HT2A receptor, and G protein systems.^{69, 72,73} Conversely, research has not provided evidence of a significant effect of DRD2 and DRD4 polymorphisms on fluvoxamine and paroxetine activities.⁷⁴

Predicting Response to Mood Stabilizers

Well-established candidates for pharmacogenetic studies of mood stabilizers are just becoming available. Although lithium has been the mainstay of treatment for bipolar disorder for nearly half a century, its therapeutic mechanism of action is unclear. The findings of pre-clinical and clinical mood stabilizer studies have included the suggestion of putative targets in the second messenger systems (e.g., G proteins, the inositol pathway, and phospholipase systems). These signal transduction pathways have been implicated in the neurobiology of bipolar disorder, although sites involved in the therapeutic response to psychotropic drugs might not necessarily involve disease-related neurobiological pathways.

In an important study of the C973A polymorphism—a silent coding SNP region (cSNP) in the gene for myoinositol polyphosphate monophosphatase—in bipolar disorder, Steen and coworkers⁷⁵ reported mixed results with lithium prophylaxis. They noted a significant association between lithium prophylaxis and the C973A polymorphism in one group of bipolar patients (N = 23) and no association in another group of bipolar patients (N = 54).⁷⁵ In 2 studies, an excellent response to lithium was linked with the intron dinucleotide repeat allele in the phospholipase C gamma-1 (PLCG1) gene.^76,77 In these studies, the investigators found an overrepresentation of the 5-repeat allele in patients with bipolar disorder compared with a control group, which suggests a trait marker.

Conversely, no association was demonstrated between the lithium response and any of a number of candidate polymorphisms in the gene for catecholamine-O-methyl transferase (COMT), monoamine oxidase (MAO) A, dopamine receptors (D2, D3, and D4), serotonin receptors (5-HT2A and 5-HT2C), the serotonin transporter SLC6A4 promoter, or the b3 subunit of G protein.¹

The Use of Genotype to Predict Adverse Reactions to Psychotropic Medications

Antidepressant-induced Mania

In general, the mechanisms underlying serious adverse reactions to drugs used in psychiatry are poorly understood. Antidepressant-induced mania is no exception. Antidepressant-induced mania is inherent in 25% to 33% of patients diagnosed with bipolar illness. A higher risk is associated with a family history of bipolar disorder, a history of antidepressant-induced mania, and exposure to multiple antidepressants.⁷⁸

Serretti and coworkers⁷⁹ conducted one of the few pharmacogenetic studies of antidepressant-induced mania, which was based on a retrospective analysis of patients diagnosed with sudden mania. From a pool of 169 patients who had had a sudden manic episode while taking an antidepressant, they selected 65 patients who “switched” to mania while taking antidepressant drugs without mood stabilizer therapy. They then selected 117 patients who “never switched” to mania from a sample of 247 patients who were diagnosed with bipolar disorder but had never been diagnosed with antidepressant-induced mania. To create a comparison group, the Serretti team randomly recruited 133 patients who had never had manic symptoms from a large pool of patients diagnosed with major depressive disorder. The data analysis took into account multiple comparisons and revealed no significant differences among the 3 treatment groups in terms of the prevalence of a polymorphism in the serotonin transporter in the upstream regulatory region or in the gene for tryptophan hydroxylase, the G-protein b3 subunit, MAO-A, COMT, 5-HT2A, or variants in the D2 and D4 receptors.⁷⁹

Mundo and others⁸⁰ used a similar study design in a subanalysis of 29 patients selected from 300 individuals with mania induced by serotonergic agents and 27 patients selected from a large pool of individuals who did not demonstrate symptoms of mania for at least 10 weeks following antidepressant therapy. Two polymorphisms of SLC6A4 were examined, namely, the intron repeat and the promoter region repeat. In this study, antidepressant-induced mania was associated with the short allele of the promoter polymorphism.⁸⁰

Neuroleptic-induced Tardive Dyskinesia

Neuroleptic-induced extrapyramidal symptoms (EPS) comprise a major drawback in the pharmacologic treatment of psychosis. Tardive dyskinesia (TD), a potentially irreversible side effect of neuroleptic medications, could compromise compliance and outcomes.^6,81 Clozapine has a superior efficacy and a lower proclivity for EPS than traditional antipsychotic drugs. This suggests that pharmacogenetic strategies in TD might be revealed through the study of antipsychotic drugs.

Perhaps the most well-studied gene in TD is the one for the DRD2 receptor.⁸¹ TD has also been inconsistently associated with common polymorphisms of this gene, including TaqI A and B and ser311cys, and several lines of evidence have linked it with the gene for the DRD3 receptor (e.g., the serine-to-glycine polymorphism on exon 1 of the DRD3 gene).¹ Additional support for the role of the glycine allele comes from meta-analysis and positron emission tomography studies.^82,83 Also, an additive interaction may occur between DRD3 and CYP1A2 through the combination of glycine-glycine (DRD3) and cytosine-cytosine (CYP1A2) alleles, resulting in the highest risk for TD, and the presence of 1 or 2 copies of the alleles, which carries an intermediate risk for TD.^84,85

Antipsychotic-induced Weight Gain

Multiple factors appear to contribute to antipsychotic-induced weight gain, which, in turn, might result in psychological distress, treatment nonadherence, and serious medical consequences, including diabetes, hyperlipidemia, cardiovascular disease, and malignancy.^86,87

The genetic and environmental factors influencing body weight and the pathways to obesity remain to be fully elucidated.⁸⁸ Among several genes currently being screened for a possible role in weight gain during antipsychotic therapy are the genes for 5-HT2C, pro-opiomelanocortin, leptin, ghrelin, tumor necrosis factor alpha, adiponectin, resistin, DRD2, the histamine (H1) receptor, and several adrenergic receptors (a1, b1, and b3).⁸⁷

A number of research centers are finding clear evidence implicating 5HT2C in weight gain induced by psychotropic medications. Reynolds and colleagues⁸⁹ have reported that the −759C/T promoter SNP in the 5-HT2C receptor gene may be linked with antipsychotic-induced weight gain. Of the 30 clozapine-treated patients who participated in this study, less weight gain (indicated by a change in body mass index) was noted in those with −759T alleles and accounted for 18% of the observed variance.⁸⁹ The −759C/T promoter polymorphism has been associated with alterations in 5-HT2C gene expression.

Neuroleptic Malignant Syndrome (NMS)

Kishida and coworkers⁹⁰ have examined the association of NMS with 3 functional polymorphisms in the DRD2 gene—namely TaqIA −141C insertion/deletion, and ser311cys—in Japanese patients. They recruited 32 patients diagnosed with NMS and used 132 patients diagnosed with schizophrenia without a documented diagnosis of NMS as controls. Allele frequencies of TaqI A, and ser311cys were similar between the 2 groups. However, the genotype and allele frequencies of the −141C ins/del polymorphism were significantly higher in patients with NMS than in controls.⁹⁰

Future Directions: The Right Drug for the Right Person

There is a considerable amount of evidence suggesting that genetic heterogeneity is an important factor in interindividual differences in drug response and toxicity. The current status of pharmacogenetic studies supports the clinical use of molecular genetic approaches to predict the therapeutic response to psychotropic medications. This would allow the clinician to use an individual patient’s unique genotype markers individually or in combination to make more informed initial treatment decisions and improve confidence in the selection of psychotropic medications based on the possibility of maximum benefit and minimum risk for adverse events. Genotyping is a simple test that can be carried out in the laboratory using samples of whole blood or buccal scrapings.

Pharmacogenetic studies of antipsychotic medications have primarily examined the use of genes encoding polymorphic target receptor sites, transporters, and second messenger systems to predict the response to clozapine. These include polymorphic genes for the D2, D3, and D4 receptors; the 5HT2A and 5HT2C receptors, and serotonin transporter proteins. These systems are fertile for research into drug response genes and appear to be inextricably interwoven with the phenomenon of psychosis. Thus far, the polymorphic 5HT2A genes appear to be strong candidates for participation in the therapeutic response to antipsychotic drugs. Conversely, the Taq1A allele of DRD2 was shown to predict a poor response to strong D2 antagonists.

The metabolism of antidepressant medications clearly depends on the cytochrome P-450 system, because CYP2D6 gene variations influence plasma levels of these drugs. Individuals who do not carry copies of the CYP2D6 gene are poor metabolizers of psychotropic drugs that are substrates for CYP2D6 and, thus, are at risk for high plasma drug concentrations. Recent randomized prospective studies examined the predictive value of CYP 2D6 genotypes for antidepressant-related adverse events and found no more events among poor metabolizers than among extensive metabolizers.^54,91 The reason for this is unclear. However, some medications may be metabolized through more than one enzyme system and some target sites that are responsible for medication sensitivity may be controlled by polymorphic genes.⁹²

Few pharmacogenetic studies of mood stabilizers have been conducted, and they have yielded less impressive results, most probably because the mechanism of action for mood stabilizers is yet to be understood. Also, it is still unclear whether bipolar disorder is a single disease entity or represents a group of disorders. Such information would be pertinent, because case uniformity might have a critical influence on the outcome of drug response studies. Preliminary results appear encouraging, with pharmacogenetic studies of the side effects of psychotropic drugs pointing to some informative candidate genes, although no firm results have been seen as yet.

It has been suggested that case definition in pharmacogenetic studies will improve if some hidden traits or endophenotypes that are more informative than the current descriptive classifications were used to identify bipolar disorder and the various forms of schizophrenia. This, in turn, suggests that a complete understanding of the neurobiology of psychiatric disorders and the mechanism of action of psychotropic drugs would increase the chance of identifying susceptibility genes in pharmacogenetic studies.⁹³

Along the same lines, it is important to note that the genetic determinants of heterogeneity in drug-metabolizing enzymes, target receptor proteins, membrane transporters, and signal transduction systems are yet to be fully characterized.^54,94 In addition, ethnic heterogeneity in the study sample may lead to spurious results. Although this pitfall may be avoided by using family-based genome-wide scans, prohibitive costs and issues of case ascertainment seriously impact the usefulness of such scans in pharmacogenetic studies.⁹⁵ For this reason, Pritchard and Rosenberg⁹⁶ recommended that genomic control be used to address ethnic stratification in case-control association studies. Genomic control involves a statistical test for stratification using alleles that have no known link to the phenotype under study in each group. If no prior reason existed to suspect stratification, a test showing a statistically significant association between randomly chosen alleles and the phenotype of interest suggests an ethnic mismatch. For samples that contain admixed groups, there appears to be a higher power to detect stratification if the markers that show allele frequency differences across populations are used for the test. No more than 40 unlinked alleles are needed for this test.⁹⁶ Additionally, novel methodologies are being used in pharmacogenetic studies to examine the collective roles of multiple common genes.^10,97 In a recent study, for example, the investigators demonstrated the contribution of multiple rare alleles to low plasma levels of high-density lipoprotein cholesterol.⁹⁸

Before long, it is likely that medicolegal pharmacogenetics might emerge as a field that depends on the strength of the evidence for genotype-based monitoring of medication benefits and adverse events. In short, important advances in pharmacogenetic studies are under way and may ultimately provide clinicians with a simple tool that can be used to predict the response to psychotropic medications and the risk for adverse events.

Table 1.

Antipsychotics: Candidate Gene Studies

Candidate Gene	Candidate Variation*	Genotype Implicated in Preliminary Reports*
Serotonin receptor 5-HT2A gene (Location: chromosome 13p or short arm)	T → C change at position 102 (coding region) G → A change at position −1438 (promoter region)	C/C genotype was linked to a poorer response (clozapine) G/G genotype was linked to a poorer response (clozapine)
Dopamine receptor D2 gene (Location: chromosome 11q or long arm)	TaqI A fragments (A1 and A2) C insert/deletion at position −141 (promoter region)	A2/A2 genotype linked to a poorer response (haloperidol) −141 C insertion allele linked to a poorer response (bromperidol or nemonapride)

^*The letters represent the DNA bases: adenine (A), guanine (G), thymine (T) and cytosine (C). The arrow → indicates single-base substitution at the identified position or locus.

Table 2.

Antidepressants: Candidate Genes

Candidate Gene	Candidate Variation	Preliminary Results
Cytochrome P-450 2D6 (Location: chromosome 22q)	0, 1, 2, or 3 copies of the allele	Poor metabolizers (higher levels of antidepressants) Ultrarapid metabolizers (lower levels of antidepressants)
Serotonin transporter gene (Location: chromosome 17q)	2 forms of the 44-base pair repeats allele; an insertion (“long”) or deletion (“short”) form	The “short” allele predicted poorer response to antidepressants

Table 3.

Mood Stabilizers: Candidate Genes

Candidate Gene	Candidate Variation	Preliminary Results
Inositol polyphosphate monophosphatase gene (Location: chromosome 2q)	C → A variation at position 973	Mixed results in studies of Li+ prophylaxis
Phospholipase C gamma-1 gene (Location: chromosome 20q)	Dinucleotide repeat in the intron; 5-repeat form	5-repeat allele overrepresented in Li+ responders

Li⁺ = Lithium

Table 4.

Vocabulary

Term	Definition
Allele	Alternative forms of a gene at a given locus. The alleles at a given locus may differ among individuals.
Dominance	When one of a pair of heterozygous alleles (see below) suppresses the expression of the other allele. If each allele produces half of the effect observed in the phenotype, this interaction is said to be additive.
Epistasis	The interaction of alleles at different loci, such that one suppresses the expression of the other. The outcome is nonadditive.
Exon	The segment of the gene that contains the genetic code for the basic unit of a protein (i.e., amino acid).
Gene	The unit of heredity (i.e., a segment of a chromosome) that contains genetic code for an entire polypeptide or protein.
Genotype	The genetic constitution of an individual based on the alleles that the parents transmitted to the individual.
Haplotype	The combination of alleles at 2 or more closely linked gene loci on the same chromosome.
Heterozygous	A genotype at a given locus that consists of 2 alleles that are not the same (e.g., the genotypes AB,AO and BO in the ABO blood group system).
Homozygous	A genotype at a given locus that consists of 2 alleles that are same (e.g., the genotypes AA, OO or BB in the ABO blood group system).
Intron	Intervening sequences that separate the exons. Introns do not code for proteins and are sometimes called noncoding sequences.
Linkage disequilibrium	The statistical association of 2 alleles at a rate that is more than would be expected by chance.
Locus	A specific region along a chromosome.
Promoter	The region within a gene that controls the initiation of protein synthesis.
Messenger RNA (mRNA)	A ribonucleic acid that shuttles genetic information from the nucleus of the cell to the ribosomes, where the information is used to produce proteins.
Monogenic	A phenotype that is based on alleles on the activity of a single locus (also referred to as Mendelian inheritance).
Nonsynonymous	Genetic variation that produces changes in protein function (also referred to as functional polymorphism).
Phenotype	A genetic effect that can be is observed (or measured). The phenotype varies with the genotype.
Polygenic	A phenotype that is based on the activity of alleles located at multiple loci.
Polymorphism	A genetic variation that occurs with a frequency of 1% or more in a random sample of the population.
Ribosome	A ribonucleic acid-rich cell structure where the proteins are made.
Synonymous	Genetic variation that produces no change in protein function.
Quantitative trait	A characteristic that is controlled by a known number of genes.
Quantitative trait loci	(QTL) The loci on a chromosome that contain the genes that control a quantitative trait.