Pharmacogenetics of Antipsychotics

Abstract

Objective:

During the past decades, increasing efforts have been invested in studies to unravel the influence of genetic factors on antipsychotic (AP) dosage, treatment response, and occurrence of adverse effects. These studies aimed to improve clinical care by predicting outcome of treatment with APs and thus allowing for individualized treatment strategies. We highlight most important findings obtained through both candidate gene and genome-wide association studies, including pharmacokinetic and pharmacodynamic factors.

Methods:

We reviewed studies on pharmacogenetics of AP response and adverse effects published on PubMed until early 2012. Owing to the high number of published studies, we focused our review on findings that have been replicated in independent studies or are supported by meta-analyses.

Results:

Most robust findings were reported for associations between polymorphisms of the cytochrome P450 system, the dopamine and the serotonin transmitter systems, and dosage, treatment response, and adverse effects, such as AP-induced weight gain or tardive dyskinesia. These associations were either detected for specific medications or for classes of APs.

Conclusion:

First promising and robust results show that pharmacogenetics bear promise for a widespread use in future clinical practice. This will likely be achieved by developing algorithms that will include many genetic variants. However, further investigation is warranted to replicate and validate previous findings, as well as to identify new genetic variants involved in AP response and for replication of existing findings.

Since the first observations in the 1950s suggested genetic influences on drug response,¹ pharmacogenetic studies have rapidly evolved over the past decades. Modern laboratory techniques have given insights toward the identification of genetic variants influencing drug efficacy, metabolism, and occurrence of adverse effects. This is giving pharmacogenetics an important role in contemporary psychiatric research, strengthening the notion that the widespread use of genetic tests will become available in future clinical practice.

Besides studies investigating the genetics of outcome to antidepressants and mood stabilizers² (see also the In Review paper by Dr Serretti and colleagues³ in this issue), large efforts have been undertaken in the area of pharmacogenetics of APs. The importance of these efforts is increasing because APs are routinely used in the treatment not only of schizophrenia and related spectrum disorders but also for mood disorders and various other conditions, such as obsessive–compulsive disorder, eating disorders, or behavioural disturbances, associated with dementia. Specific APs may differ regarding alleviating various symptoms; for instance, SGAs may be more effective for the treatment of negative symptoms. APs also differ in their propensity to induce specific adverse effects; for example, SGAs may be associated with a higher risk of significant metabolic disturbances but a lower risk of TD, compared with FGAs.⁴ More importantly, large interindividual differences exist among patients. These differences cannot be predicted clinically but they can have serious consequences leading to repeated medication switches owing to intolerability or lack of efficacy. This issue is of particular relevance in patients with schizophrenia who need to be treated over a long period of time and whose adherence is hampered by adverse effects.⁵ Also, treatment resistance (that is, a lack of adequate response to medication) occurs in up to 40% of these patients.⁶ The possibility of meaningful prediction of individual responses and risks of adverse effects would, therefore, represent a milestone in AP pharmacotherapy.

To date, only a limited number of factors have been identified that are described to be either positively or negatively correlated with outcome of AP treatment. Among those are demographic factors (for example, family history, ethnicity, and gender), clinical factors (for example, duration of untreated psychosis and early response to APs), and environmental factors (for example, smoking habits, concomitant treatment, and diet).⁷ The potential influence of genetic factors is underpinned by different studies demonstrating similar treatment response and occurrence of adverse effects in monozygotic twins or first-degree relatives.^8–12

Clinical Implications

• Accurate prediction of individual response to AP treatment or individual risk for the occurrence of adverse effects has not yet been achieved. Large efforts have been made to find genetic factors in the hope of improving future patient care.

• Some genetic factors have been identified. However, valid tests with substantial specificity and sensitivity measures have not yet been developed to predict drug metabolism, response to treatment, or most adverse effects.

• Nonetheless, some results are starting to impact patient care through the development of genetic tests.

Limitations

• Lack of independent replication in well-characterized samples limits the clinical applicability of most of these genetic findings.

• Approaches integrating genome-wide, DNA sequencing, epigenetic and environmental factors, methods for prediction of treatment response, and occurrence of adverse effects are required to allow for a more widespread clinical use of pharmacogenetic-based algorithms.

Most studies conducted to date have used a candidate gene approach, investigating SNPs or changes in larger DNA segments of a gene considered to be involved in the mechanisms of medication response or in a given adverse effect. With further development of laboratory techniques, GWASs and next-generation sequencing are being applied to pharmacogenetic research. Although there are some inconsistencies in the results of some studies, efforts are being made to transfer some promising findings into clinical practice. Our article reviews the most robust pharmacogenetic results, including pharmacokinetic and pharmacodynamic factors, relevant to APs.

Methods

We reviewed studies on pharmacogenetics of AP response and adverse effects published on PubMed until January 2012. Owing to the large amount of available literature, a detailed summary of all studies would go beyond the scope of our review. Also, several recent reviews have focused on AP response and adverse events.^7,13–15 Thus we focused our review on pharmacogenetics findings that have been replicated in independent studies or are supported by meta-analyses. We also included selected studies that highlight new and promising candidate genes in relevant pathways.

Genetics and Pharmacokinetic Factors

Cytochrome Enzymes

Most psychotropics, including APs, are metabolized by cytochrome enzymes, mainly members of the CYP family, which includes CYP1A2, CYP2D6, CYP3A4, and CYP2C19. Besides environmental factors, such as induction of the CYP1A2 enzyme by smoking, genetic variation contributes to the variability in enzyme activity and drug metabolism. The genes encoding CYP enzymes can be highly polymorphic,¹⁶ with more than 80 genetic variations known in CYP2D6. For this gene, a distinction is made among extensive metabolizers, with normal enzyme activity having 2 functional alleles; PMs, with low enzyme activity caused by 2 nonfunctional alleles; IMs, with intermediate activity carrying 2 partly defective alleles or 1 nonfunctional allele; and UMs, with more than 2 functional alleles. Similarly, an activity score system was recently proposed.¹⁷ The main polymorphisms causing defective alleles in the population of European descent are CYP2D6*3, CYP2D6*4, CYP2D6*5, and CYP2D6*6.¹⁸ Genotype distribution differs considerably between ethnicities: 1% to 2% of Europeans are UMs and 5% to 10% are PMs, whereas only 1% to 2% of Asians are PMs, while 30% to 40% of the population in North Africa are UMs but only 1% are PMs.¹⁹ Metabolizer status highly influences the required dosage of substances; for example, it has been demonstrated that carriers of CYP2D6 defective allele (that is, *3, *4, *5, and *6) can have up to 80% higher serum concentrations of risperidone, compared with homozygous carriers of the wildtype CYP2D6*1.²⁰ The *1F polymorphism in CYP1A2 seems to be related to higher inducibility of the enzyme by smoking, leading to reduced clozapine plasma levels.²¹ Polymorphisms in CYP2C19 and CYP3A4 have also been reported to affect clozapine plasma levels.^22,23

These data suggest that CYP polymorphisms may affect treatment response. However, there are only a few studies with positive findings, most likely because plasma levels of APs often do not correlate with treatment response. For CYP1A2, the *1F polymorphism has been associated with poorer treatment response to clozapine and olanzapine.^24–26 One other study²⁷ indicated a weak effect of CYP3A4 polymorphisms on risperidone response. A recent study by Bigos et al²⁸ reported a highly significant influence of the marker rs472660 in CYP3A43, a member of the CYP3A4 enzyme family encoding the CYP family 3, subfamily A, polypeptide 43, on olanzapine clearance and response. By contrast, several studies investigating the influence of CYP2D6 polymorphisms on olanzapine response have reported no associations.^29–32

There are more data available indicating an impact of CYP polymorphisms on side effects: CYP2D6*3, CYP2D6*4, and CYP2D6*10, but not CYP1A2*1F, have been associated with increased AIWG.^24,33,34

Today, most evidence exists for an influence of CYP polymorphisms on motor adverse effects of APs, such as TD. Although some studies reported negative findings,^35–38 there is growing evidence that CYP2D6 PMs are at higher risk of developing acute EPSs^39,40 or TD^41–43 (see reviews^13,44,45). Some studies have also associated CYP1A2*1F with TD,^42,46 but other studies and a meta-analysis have yielded negative findings.^47,48

In summary, genetic variation in CYP genes seems to exert a stronger influence on plasma levels and the occurrence of adverse effects than on medication response. As sample sizes of many studies are small and these studies have been performed using different medications and protocols, further investigation is warranted to fully assess clinical implications of these findings. Further, there is still a paucity of studies assessing the cost-effectiveness of CYP genotyping in daily practice,^19,45,49 and further pharmacoeconomic investigations are needed. Despite these limitations, first steps to bring pharmacogenetics advances into clinical practice have been undertaken by offering tests for CYP2D6 and CYP2D19 usable for medication monitoring.⁵⁰ For information regarding ongoing related investigations in Canada, please see the Pharmacogenetics Research Clinic website.^51,52

Blood–Brain Barrier Transporters

P-gp is a member of the ATP-binding cassette transporters and is involved in the transmembrane efflux of different substrates, including toxins, peptides, and medications. P-gps are located in different organs, including the blood–brain barrier, where they pump medications back into the plasma.⁵³ An altered transporter activity influences intracerebral drug concentration and may thereby impact on treatment response as well. Polymorphisms in the encoding gene MDR1 (synonymous with ABCB1, ATP-binding cassette, subfamily B [MDR/transporter associated with antigen processing], member 1 gene) have been investigated in several studies. Although there are negative findings,^54,55 some positive findings exist for the G2677T/A and C3435T polymorphisms. Contradictory results have been reported, with studies showing better treatment response of T allele carriers of both polymorphisms,⁵⁶ while others reported worse response in the T allele carriers of C3435T.^57,58 Other studies found no association in the above-mentioned polymorphisms but had positive results for C1236T.⁵⁹ Overall, despite ambiguous evidence for the role of MDR1 polymorphisms in treatment response, probably owing to yet undetected confounding variables (for example, ethnicity), a role of this gene is likely and awaits further clarification.

Regarding AP-induced adverse effects, there has been one study showing no association of MDR1 polymorphisms with TD,⁶⁰ one study demonstrating an association with a specific haplotype and TD,⁶¹ and one study indicating an influence of MDR1 genotypes on the occurrence of EPSs,³⁰ while another study could not find an association with EPSs.⁵⁵ There are also hints of an influence of MDR1 on weight gain⁶² and metabolic disturbances caused by APs.⁶³ Overall, more studies are needed to assess the role of MDR1 polymorphisms in the occurrence of adverse effects.

Genetics and Pharmacodynamic Factors of Treatment Response

The DA System

The antidopaminergic effect of FGA and SGA is thought to be their main mechanism of action.⁶⁴ Therefore, genetic variation in DA receptors influencing receptor density, expression, and activity of the receptor may be an essential factor in regulation of individual treatment response.

As all APs block D₂ receptors,⁶⁵ polymorphisms in the DRD2 gene have been investigated most intensively. The del allele of the −141C ins/del polymorphism in the promoter region is assumed to reduce D₂ density and activity. In several studies, the del allele was associated with poorer response to AP treatment, for example, see Lencz et al⁶⁶ and Wu et al.⁶⁷ This finding has been confirmed by a recent meta-analysis,⁶⁸ despite some studies that failed to find an association with treatment response (for example, see Xing et al⁶⁹). The A1 allele of TaqIA polymorphism, which is located 10 kilobases downstream DRD2, leads to reduced gene expression and was therefore also hypothesized to influence treatment response. However, previous studies have reported inconsistent findings: while some studies indicated that A1^70–72 or A2⁷³ are associated with better treatment response, the above-mentioned meta-analysis⁶⁸ did not detect any association. Also for other DRD2 polymorphisms, for example, TaqIB,⁷⁴ Ser311Cys,⁷⁵ or A-241G,⁶⁶ a few studies have reported associations, but contradictory results and lack of replication make further investigations necessary.

In DRD3, the Gly9 variant of the Ser9Gly polymorphism, changes D₃ receptor density. Thus this impact of this variant on AP treatment response has been studied extensively, with at least 18 published studies.¹³ While some studies have indicated better response of the Ser allele, a meta-analysis has reported a nonsignificant trend toward lower response in carriers of Ser allele.⁷⁶ Because clozapine also binds to D₄ receptors, a 48-bp variable number tandem repeat in the DRD4 gene has been investigated in several studies. However, findings have been contradictory,^77,78 and many studies have yielded negative results (for example, see Rietschel et al⁷⁹ and Ikeda et al⁸⁰); the role of the polymorphism in treatment response is not fully resolved, to date. Polymorphisms of the DRD1⁸¹ and DRD5⁸⁰ genes have not yet been investigated intensively and their clinical relevance needs to be demonstrated in future studies.

The 5-HT System

Because of the higher affinity of SGAs to 5-HT receptors, compared with FGAs, genes in the serotonergic transmitter system are interesting candidates for investigation of AP treatment response. Several studies have focused on polymorphisms in HTR2A, the gene encoding 5-HT_2A receptors. The functional −1438A/G SNP in the promoter region of the gene and the SNP 102T/C are in high linkage disequlibrium with each other⁸² and −1438A/G has been demonstrated to influence promoter activity.⁸³ Association studies have repeatedly indicated poorer treatment response in G allele carriers of the −1438A/G SNP,^84,85 although there are negative findings⁷¹ and a study showing better response in G allele carriers.⁸⁶ Regarding the 102T/C marker, results are less consistent with some studies (see Anttila et al⁸⁷) and an early meta-analysis focusing on clozapine demonstrating better response in T allele carriers,⁸⁸ while later studies investigating other APs found no association⁸⁰ or association of the C allele with better symptom improvement.⁸⁹ Another polymorphism repeatedly investigated in HTR2A is His452Tyr (histidine and Tyr at the 452nd amino acid): the Tyr allele seems to be associated with poorer treatment response,⁸⁸ but this has not been observed in all studies.⁹⁰

In HTR2C, encoding 5-HT_2C receptors, the C/C genotype of the −759C/T SNP, which has been shown to lead to reduced transcription,⁹¹ was described to be associated with improvement of negative symptoms in one study.⁹² Also, association of the Ser allele of Cys23Ser has been reported once,⁹³ but all other studies^90,94 have failed to replicate this finding, indicating no major role of the SNP in treatment response. In the gene encoding solute carrier family 6 (neurotransmitter transporter), member 4 (SLC6A4, synonymous with 5-HTT) 5-HT transporters, which terminate 5-HT action by transporting it back into the presynaptic neuron, an association of the short allele of the 44-bp ins/del polymorphism serotonin-transporter-linked polymorphic region (commonly referred to as 5-HTTLPR) with poorer AP response has consistently been demonstrated.^95–97

Further association with AP treatment response has been described for the G allele of −1019C/G SNP in 5-HT receptor 1A, G protein-coupled (HTR1A), encoding 5-HT_1A receptors,^98,99 for polymorphisms in the genes encoding subtypes 3A, B, and E of 5-HT₃ receptors^100,101 and for the 267C/T SNP in the gene encoding 5-HT₆ receptors.^102,103 However, this last finding has not been confirmed in all studies.⁸⁰

Other Systems

Other candidate genes that may be involved in regulation of individual treatment response to APs have not been studied as extensively as the above-mentioned genes. However, several studies have suggested an association between response and several genetic variants. For example, the COMT gene encodes catechol-O-methyltransferase, which is involved in the degradation of monoamines. The Val(108/158)Met polymorphism of this gene influences activity of the enzyme, which is less active in Met/Met carriers. The Met allele has been repeatedly associated with better improvement of cognitive function^104,105 during AP treatment, compared with the Val allele. Other studies¹⁰⁶ have found no association or reported an association of the Val allele with better response.¹⁰⁷ Therefore, the role of the Val(108/158)Met SNP in the regulation of treatment response is not yet fully understood and deserves further study.

Further, studies reported associations in GNB3, encoding the β-subunit of G-protein, with the T allele of the −825C/T polymorphism leading to worse treatment response,^108,109 and in the genes encoding brain-derived neurotrophic factor,^110,111 GFRA,¹¹² oxytocin,¹¹³ and TNF¹¹⁴; however, those genes need to be studies more extensively.

In summary, candidate gene studies have delivered some promising results, consisting mainly of associations between genes involved in the DA system and overall treatment response or between genes involved in the serotonergic system and improvement of negative symptoms (Table 1). Therefore, findings suggest a clinical impact of those genetic variants. However, given the relatively small sample sizes and inconsistent results, further studies with larger sample sizes and more homogenous studies are needed before these findings can be incorporated in clinical algorithms.

Table 1.

Polymorphisms associated with AP treatment response supported by a meta-analysis or by 3 or more studies reporting the same allele association

Gene	Polymorphism	Main association finding	Study
CYP1A2	*1F	*1F/*1F genotype associated with lower plasma concentrations and lower response	Laika et al,24 Ozdemir et al,25 Eap et al26
DRD2	−141C ins/del	Del allele is associated with poorer response	Lencz et al,66 Wu et al,67 Zhang et al68
	TaqIA	A1 allele associated with better response in some studies; but a meta-analysis showed negative results	Suzuki et al,70 Schafer et al,71 Yamanouchi et al,72 Zhang et al68
DRD3	Ser9Gly	Meta-analysis with trend for lower response in Ser-allele carriers	Hwang et al76
HTR2A	−1438G/A	G allele was associated with poorer response	Arranz et al,84 Ellingrod et al,85 Chen et al177
	102T/C	T allele was associated with better response in a meta-analysis	Arranz et al88
	His452Tyr	Tyr allele was reported to be associated with poor response	Arranz et al,88 Arranz et al,178 Masellis et al179
5-HTT	5-HTTLPR	Short-allele was demonstrated to be associated with poorer treatment response	Dolzan et al,95 Arranz et al,96 Vazquez-Bourgon et al97
COMT	Val(108/158)Met	Better improvement of cognitive function was associated with Met allele	Bertolino et al,104 Woodward et al,105 Bertolino et al180

Genome-Wide Association Studies

During the past few years, the development of technologies has facilitated the investigation of polymorphisms in the whole genome through GWASs. Studies investigating the CATIE sample⁵ have reported the association of an intergenetic SNP on chromosome 4p15 with response to APs. Also, borderline significant results have been reported for SNPs in the ankyrin repeat and sterile alpha motif domain containing 1B (ANKS1B) and contactin associated protein-like 5 (CNTNAP5) genes¹¹⁵ as well as association of 6 SNPs in or near to the ets homologous factor (EHF), solute carrier family 26 (anion exchanger), member 9 (SLC26A9), DRD2, G protein-coupled receptor 137B (GPR137B), carbohydrate (N-acetylgalactosamine 4-0) sulfotransferase 8 (CHST8), and interleukin 1, alpha (IL1A) genes.¹¹⁶ In other samples, ATPase, Ca++ transporting, plasma membrane 2 (ATP2B2), heparan sulfate (glucosamine) 3-O-sulfotransferase 2 (HS3ST2), unc-5 homolog C (C elegans) (UNC5C), BCL2-associated athanogene 3 (BAG3), phosphodiesterase 7B (PDE7B), phosphoribosylaminoimidazole carboxylase, phosphoribosylaminoimidazole succinocarboxamide synthetase (PAICS), prostaglandin F2 receptor inhibitor (PTGFRN), nuclear receptor subfamily 3, group C, member 2 (NR3C2), zinc finger and BTB domain containing 20 (ZBTB20), ST6 beta-galactosamide alpha-2,6-sialyltranferase 2 (ST6GAL2), phosphatidylinositol-4-phosphate 5-kinase, type I, beta (PIP5K1B), ephrin receptor type A6 (EPHA6), potassium voltage-gated channel, sub-family H (eag-related), member 5 (KCNH5), and adherens junctions associated protein 1 (AJAP1) have been reported to be associated with response to risperidone.¹¹⁷ Similarly, SNPs within neuronal PAS domain protein 3 (NPAS3), XK, Kell blood group complex subunit-related family, member 4 (XKR4), tenascin R (TNR), glutamate receptor, ionotropic, AMPA 4 (GRIA4), GFRA2, and the nudix (nucleoside diphosphate linked moiety X)-type motif 9 pseudogene 1 (NUDT9P1) located in the 5-HT receptor 7, adenylate cyclase-coupled (HTR7) gene have been reported to be associated with response to iloperidone.¹¹⁸ In summary, hypothesis-free GWASs have yielded numerous interesting findings that warrant further investigation.

Genetics and Pharmacodynamic Factors of Adverse Effects

Antipsychotic-Induced Weight Gain

AIWG is observed in up to 30% of patients treated with SGAs. Owing to the higher morbidity and mortality associated with obesity and metabolic syndrome, as well as the social stigmatization and noncompliance that arises from weight gain (extensively reviewed elsewhere^14,119), this serious adverse effect needs special consideration in clinical practice. However, there is a large variation in the propensity of various APs to cause AIWG,¹²⁰ with olanzapine and clozapine causing the most extensive amount of weight gain. Interindividual variability is also influenced by clinical factors, such as sex, baseline weight, and age. However, there are no reliable clinical predictors available in clinical practice. As results of twin studies indicate a genetic influence on AIWG,^11,12,121 numerous candidate studies have been performed during the past few years. Most consistent findings exist for an association between polymorphisms of the HTR2C gene and AIWG. The functional polymorphism −759C/T¹²² has repeatedly been shown to influence AIWG, with the C allele being overrepresented in patients with higher AIWG (for example, see Reynolds et al,¹²³ Miller et al,¹²⁴ and Opgen-Rhein et al,¹²⁵ including 2 meta-analyses, see De Luca et al¹²⁶ and Sicard et al¹²⁷). Although negative findings have been reported,^128,129 with one study indicating the opposite allele as a risk variant,¹³⁰ to date, the role of this polymorphism is supported by the most robust studies. Other HTR2C polymorphisms have been shown to be associated with AIWG as well, including the −995G/A, −1165A/G,¹²⁵ and Cys23Ser polymorphisms as part of a high risk haplotype.¹²⁷

Leptin plays a major role in energy homoeostasis. The −2548A/G polymorphism in the encoding gene leptin (LEP) has been shown to impact on AIWG, with the G allele being the risk allele in some but not in all studies.^131,132 Nonetheless, negative findings (for example, see Opgen-Rhein et al¹²⁵; for review, see Lett et al¹⁴ and Lee and Bishop¹³³) hamper the assessment of the clinical impact of this polymorphism. Other replicated associations include Ddell, Mnll, and Tail polymorphisms in synaptosomal-associated protein, 25kDa (SNAP25),^134,135 and in the gene encoding alpha-2-adrenergic receptor, adrenoceptor alpha 2A (ADRA2A), with the G/G genotype of −1291C/G being at higher risk for AIWG.^136,137 Similarly, the T allele carriers of the −825C/T polymorphism in the gene encoding β3-subunit of G-protein receptors, that is, GNB3, are at higher risk for AIWG.^129,138 Other reported associations, such as polymorphisms in insulin-induced gene 2 (INSIG2),¹³⁹ have failed to be replicated in other samples.^125,140 Recent studies indicate an influence of DRD2 polymorphisms,^141,142 cholecystokinin B receptors,¹⁴³ and adiponectin.¹⁴⁴ A systematic analysis of the DA D₁–D₅ receptors and the association with weight gain reported a significant finding for the 957C/T polymorphism and 2 weak signals for DRD3 polymorphisms.¹⁴² Most recent data indicate a strong effect in the region of MC4R, the gene encoding the melanocortin 4 receptor, on development of obesity.¹⁴⁵ A GWAS in a cohort of drug-naive adolescents showed significant association of an SNP near this gene with weight gain induced by several APs, which was replicated independently by our group.¹⁴⁶ In a different study, our group has also shown a significant association of another SNP in the promoter region of MC4R with AIWG.¹⁴⁷

A recent GWAS found significant associations of AIWG with SNPs in the Meis homeobox 2 (MEIS2), protein kinase, cyclic adenosine monophosphate-dependent, regulatory, type II, beta (PRKAR2B), G protein-coupled receptor 98 (GPR98), formin homology 2 domain containing 3 (FHOD3), ring finger protein 144A (RNF144A), astrotactin 2 (ASTN2), sex determining region Y-box 5 (SOX5), and activating transcription factor 7 interacting protein 2 (ATF7IP2) genes¹⁴⁸ but these findings have not yet been replicated in other studies. Thus further investigation remains necessary. Table 2 presents a summary of the most important genetic findings regarding AIWG. Overall, except for the polymorphism −759C/T of HTR2C, with most robust findings obtained in a large number of studies, genetic findings regarding AIWG are not yet applicable in clinical practice.

Table 2.

Polymorphisms associated with AIWS or EPSs supported by a meta-analysis or by 3 or more studies reporting the same allele association

Gene	Polymorphism	Main association finding	Study
CYP2D6	metabolizer status	PMs were at higher risk for EPSs in several studies	Kobylecki et al,41 Fu et al,42 Patsopoulos et al43
HTR2C	−759C/T	Two meta-analyzes reported C allele to be associated with larger AIWG	Sicard et al,127 De Luca et al181
LEP	−2548A/G	G allele was associated with higher AIWG in most studies; others showed A allele to be associated	Ellingrod et al,131 Zhang et al,132 Templeman et al182; Zhang et al183
GNB3	−825C/T	T allele was associated with higher AIWG in several studies, and with a nonsignificant trend in a meta-analysis	Ujike et al,129 Wang et al,138 Souza et al184
DRD2	TaqIA	A2 allele was associated with higher risk for TD in 2 meta-analyzes	Bakker et al,48 Zai et al154
DRD3	Ser9Gly	Gly allele was associated with higher risk for TD in several studies	de Leon et al,60 Steen et al,155 Al Hadithy et al,156 Lerer et al,157 Bakker et al158
HTR2A	T102C	C allele was associated with TD in a meta-analysis	Bakker et al,48 Lerer et al164
COMT	Val(108/158)Met	Met allele was protective against TD in a meta-analysis	Bakker et al48
MnSOD	Ala-9Val	Val allele was protective against TD in a meta-analysis	Bakker et al48

Ala = alanine

Agranulocytosis

Agranulocytosis is a serious adverse effect occurring with a wide range of psychotropic medications. While clozapine is the most effective medication for treatment-resistant schizophrenia,¹⁵ it is associated with an incidence of up to 2% of agranulocytosis. Hypothesized mechanisms of CIA include immune-mediated response against neutrophils, enhanced release or destruction of neutrophils, and direct toxicity against bone marrow stromal cells.¹⁴⁹ Association studies investigating genetic factors of CIA have mainly focused on genes building the HLA system. Positive findings exist for the HLA-B38 marker, which is overrepresented in patients affected by CIA as well as a haplotype consisting of HLA-B38, DR4, and DQw3.¹⁵⁰ Other HLA-antigens associated with CIA include HLA-DRB5*0201 and HLA-Cw-7 and haplotypes with HLA-Cw-B and HLA-DRB5-DRB4.¹⁵¹ In addition, other genes for non-HLA components of the MHC have been investigated. For both TNF and heat shock protein replicated positive findings have been reported.¹⁵ There are also associations for genes that are not part of the MHC, for example, myeloperoxidase (MPO) and nicotinamide adenine dinucleotide phosphate-oxidase gene polymorphisms.^152,153 First steps toward a clinical use have been made by the temporary release of a commercial test-kit in 2007, that is, PGxPredict:CLOZAPINE (Clinical Data, Inc, New Haven, CT), which included the 6672G/C polymorphism in MHC, class II, DQ beta 1 (HLA-DQB1) for detection of high-risk patients carrying the C allele, with high specificity of 98.4% but a low of sensitivity of 21.5%.

TD and EPSs

TD is a serious adverse effect that occurs in roughly 25% of all AP-treated patients. FGAs bear a much higher risk for the development of TD than SGAs. Clinical risk factors include advanced age, dose and duration of medication, and sex.⁴⁵ Familial occurrence points to a genetic influence on TD.¹⁰ Candidate genes for TD have been extensively investigated during the past 2 decades, and there are numerous studies supporting the hypothesis of a genetic implication. Besides positive findings for pharmacokinetic factors (see above), there is growing evidence for pharmacodynamic factors having an impact on TD.

In DRD2, an increased risk for the development of TD has been described for carriers of the A2 allele of the TaqIA polymorphism in 2 meta-analyzes,^48,154 although more studies have delivered negative findings than positive ones.¹³ Most robust findings exist for the Ser9Gly polymorphism in DRD3, with the Gly allele being associated with higher risk of TD in different ethnicities (for examples, see de Leon et al,⁶⁰ Steen et al,¹⁵⁵ Al Hadithy,¹⁵⁶ Lerer et al,¹⁵⁷ and Bakker et al¹⁵⁸). Nonetheless, results are not unambiguous as negative results also exist,^38,159,160 and one study indicated the opposite allele as risk variant.¹⁶¹ DRD4 has not been studied as extensively as DRD2 and DRD3. Nonetheless, there have been positive findings for DRD4 polymorphisms,^162,163 which need further replication. Several studies have reported associations for polymorphisms in serotonergic genes. A meta-analysis found an association of the C allele in the 102T/C polymorphism of HTR2A with TD.¹⁶⁴ Nonetheless, there are numerous negative findings for this SNP as well (for example, see Wilffert et al¹⁶⁵). Studies also reported positive results with the −1438G/A¹⁶⁶ in HTR2A and the Cys23Ser polymorphism in HTR2C (for example, see Al Hadithy et al¹⁵⁶). As there are many studies with negative findings,^38,167 further studies of 5-HT genes remain necessary to evaluate their role in the development of TD. Other significant findings in meta-analyses exist for the Val(108/158)Met polymorphism in COMT⁴⁸ (with the Met allele being protective against TD) and for manganese superoxide dismutase (MnSOD)⁴⁴ (with the Val allele of alanine (Ala-9Val) polymorphism being protective against TD⁴⁸). There are single reports of positive results for a few other genes, including different COMT polymorphisms,¹⁶⁸ glycogen synthase kinase 3 beta (GSK3B), linked to dopaminergic signalling,¹⁶⁹ and nitric oxide synthase 3 (endothelial cell) (NOS3).¹⁷⁰

There has also been reports of an influence of regulator of G-protein signalling 2, 24kDa (RGS2) polymorphisms on EPSs.^171,172 A GWAS analyzing the CATIE sample^5,173 reported associations in the gene encoding zinc finger protein 202, (ZNF202), in GLI family zinc finger 2 (GLI2), a gene encoding a transcription factor involved in embryonal development of the dopaminergic system,¹⁷⁴ and statistical trends for other genes.¹⁷⁵

In summary, several candidate genes have been found to be involved in TD, and the most robust findings were obtained in dopaminergic genes. Table 2 summarizes the most important genetic findings. As TD and EPSs are mainly mediated by the influence of APs on the DA system, those findings are hardly surprising. However, to date, there is no definite evidence on the genetics of TD, and further research is needed to develop predictive tests that can be used in clinical practice.

Summary and Outlook

During the past 20 years, pharmacogenetic studies have identified many genetic variants implicated in serum levels, response to APs, and occurrence of adverse effects. While some of these results have been replicated in independent samples, and therefore are likely to represent true associations, negative studies have also been published, albeit, often in smaller and likely underpowered studies. Currently, most promising findings involve associations between DA receptor polymorphisms and response, or HTR2C SNPs and AIWG. There are several possible explanations for these inconsistencies: differences in study design, small effect sizes of most SNPs, small sample sizes, incomplete coverage of most genes, lack of control of environmental and clinical confounders, or varying definition of outcome parameters. Conflicting results may also be due to an insufficient incorporation of gene–gene and gene–environment interactions. Therefore, new approaches, beyond candidate gene studies and GWASs are needed: DNA-sequencing, gene expression studies, novel bioinformatic approaches, animal studies, epigenetic approaches, and large and prospectively assessed samples will illuminate underlying genetic mechanisms.

Although widespread clinical use of pharmacogenetics will likely take several more years of investigation, some findings are starting to be clinically applied. These include the assessment of drug metabolizer status and inclusion of other pharmacodynamics genes with promising findings showing superiority to treatment as usual.¹⁷⁶

Those tests are still relatively expensive and need to be improved by adding more gene variants. They also need to be implemented in ways that allow their use in clinical routine (that is, rapid processing and convenient costs), and physicians need to learn how to use their results in their practice. However, the clinical availability of these few tests suggests that pharmacogenetics is becoming a reality in patient care and will help to individualize and substantially improve treatment in the near future.

Abbreviations

5-HT

5-hydroxytryptamine (serotonin)

AIWG

AP-induced weight gain

AP

antipsychotic

ATP

adenosine triphosphate

bp

base pair

CATIE

Clinical Antipsychotic Trials of Intervention Effectiveness

CIA

clozapine-induced agranulocytosis

COMT

catechol-O-methyltransferase

CYP

cytochrome P450

CYP1A2

CYP, family 1, subfamily A, polypeptide 2

CYP2C19

CYP, family 2, subfamily C, polypeptide 19

CYP2D6

CYP, family 2, subfamily D, polypeptide 6

CYP3A4

CYP, family 3, subfamily A, polypeptide 4

Cys

cysteine

DA

dopamine

del

deletion

DNA

deoxyribonucleic acid

DRB5

MHC, class II, DR beta 5

DRD

DA receptor D

EPS

extra pyramidal symptom

FGA

first-generation AP

GFRA

glial cell derived neurotrophic factor family receptor alpha

Gly

glycine

GNB3

guanine nucleotide binding protein (G protein), beta polypeptide 3

GWAS

genome-wide association study

HLA

human leukocyte antigen

HTR2A

5-HT receptor 2A, G protein-coupled

HTR2C

5-HT receptor 2C, G protein-coupled

IM

intermediate metabolizer

ins

insertion

MC4R

melanocortin 4 receptor

MDR1

multi-drug resistance gene 1

Met

methionine

MHC

major histocompatibility complex

P-gp

permeability glycoprotein

PM

poor metabolizer

rs

reference SNP

SLC

solute carrier

Ser

serine

SGA

second-generation AP

SNP

single-nucleotide polymorphism

TD

tardive dyskinesia

TNF

tumour necrosis factor

Tyr

tyrosine

UM

ultra-rapid metabolizer

Val

valine