Association Studies of Genomic Variants with Treatment Response to Risperidone, Clozapine, Quetiapine and Chlorpromazine in the Chinese Han Population

Home , Clozapine

ORIGINAL ARTICLE Association studies of genomic variants with treatment response to risperidone, clozapine, quetiapine and chlorpromazine in the Chinese Han population

QXu1,XWu1,MLi1, H Huang2,3, C Minica4,ZYi5, G Wang6, L Shen1, Q Xing7, Y Shi1,LHe1,7,8 and S Qin1,2,3,8

Schizophrenia is a widespread mental disease with a prevalence of about 1% in the world population. Continuous long-term treatment is required to maintain social functioning and prevent symptom relapse of schizophrenia patients. However, there are considerable individual differences in response to the antipsychotic drugs. There is a pressing need to identify more drug-response- related markers. But most pharmacogenomics of schizophrenia have typically focused on a few candidate genes in small sample size. In this study, 995 subjects were selected for discovering the drug-response-related markers. A total of 77 single-nucleotide polymorphisms of 25 genes have been investigated for four commonly used antipsychotic drugs in China: risperidone, clozapine, quetiapine, and chlorpromazine. Significant associations with treatment response for several genes, such as CYP2D6, CYP2C19, COMT, ABCB1, DRD3 and HTR2C have been verified in our study. Also, we found several new candidate genes (TNIK, RELN, NOTCH4 and SLC6A2) and combinations (haplotype rs1544325–rs5993883–rs6269–rs4818 in COMT) that are associated with treatment response to the four drugs. Also, multivariate interactions analysis demonstrated the combination of rs6269 in COMT and rs3813929 in HTR2C may work as a predictor to improve the clinical antipsychotic response. So our study is of great significance to improve current knowledge on the pharmacogenomics of schizophrenia, thus promoting the implementation of personalized medicine in schizophrenia.

The Pharmacogenomics Journal (2016) 16, 357–365; doi:10.1038/tpj.2015.61; published online 18 August 2015

INTRODUCTION poor. There is a pressing need to identify more drug-response- Schizophrenia is a chronic and debilitating mental disorder related markers based on which response to treatment can be affecting 1% of the general population, which is often described predicted. And on this basis, a number of pharmacogenetic in terms of both positive and negative symptoms, such as studies have been performed. To date, most studies have typically hallucinations, delusions, thought disorders, avolition and social focused on candidate genes, mainly selected for drug- withdraw, as well as cognitive and functional impairment.1–3 metabolizing enzyme genes, transport genes and neurotransmit- 5 Continuous long-term treatment is required to maintain social ter receptors genes. functioning and prevent symptom relapse, which causes signifi- Several studies have shown positive associations between 6,7 cant public health and economic burden.4 genetic variation and different antipsychotic drug response. Antipsychotic drugs are the mainstay of treatment for schizo- However, these studies mainly focus on a few genes through phrenia. The vast majority of antipsychotic drugs in current use can candidate gene association study method and most were be classified into two categories: first-generation antipsychotic performed in small samples, which make these results could not medications and second-generation antipsychotic medications. be confirmed by diffident groups and be used in clinical practices.4 Currently available antipsychotic drugs achieve a certain degree At present, there is only a few variant that is well replicated and of clinical effective schizophrenia treatment in about 50% of holds promise as clinical biomarkers, such as CYP2D6 and ABCB1.8 patients.2 And treatment failure for schizophrenia might cause In recent years, genome-wide association study, which has the weeks of unremitted illness and potential adverse drug reactions, as potential to discover new molecular markers of drug actions, had well as following larger clinical and economic cost. Risperidone, been applied to antipsychotic response and drug-induced side clozapine, quetiapine and chlorpromazine are the most commonly effect studies. However, these studies were also mainly performed used antipsychotic drugs in China. As with all antipsychotics, there in relatively small samples, which has the low power to uncover are considerable individual differences in response to the four the significant variants, and sometimes genome-wide association – drugs, in terms of both therapeutic effects and adverse effects. study results are not replicated across studies or population.9 11 Identifying the optimal treatment for any patient is often a trial There has been an increasing awareness that larger sample size is and error process that can span a long time, and response is often important for future studies to discover the reliable genomic

1Bio-X Institutes, Key Laboratory for the Genetics of Developmental and Neuropsychiatric Disorders (Ministry of Education), Shanghai Jiao Tong University, Shanghai, China; 2Analytic and Translational Genetics Unit, Department of Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; 3Broad Institute of Harvard and MIT, Cambridge, MA, USA; 4Department of Biological Psychology, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands; 5Shanghai Institute of Mental Health, Shanghai, China; 6Nanjing Medical University Affiliated Wuxi Mental Health Center, Wuxi, China; 7Institutes of Biomedical Sciences, Fudan University, Shanghai, China and 8Shanghai Key Laboratory of Psychotic Disorders, Shanghai, China. Correspondence: Dr S Qin, Bio-X Institutes, Key Laboratory for the Genetics of Developmental and Neuropsychiatric Disorders (Ministry of Education), Shanghai Jiao Tong University, Little White House Building 1954 Hua Shan Road, Shanghai 200030, China. E-mail: [email protected] Received 13 April 2015; revised 8 June 2015; accepted 14 July 2015; published online 18 August 2015 Pharmacogenomics of schizophrenia QXuet al 358 variants with treatment response.8 Furthermore, the drug response the end of week 2. After that, the dosage was adjusted according to of antipsychotic drugs involves the process of absorption, individual tolerance. For all the participants, medication compliance was distribution, metabolism, target response and excretion. Also, closely monitored and confirmed by nursing staff, and no other medication genes do not work in isolation, instead, complex molecular was given except trihexylphenidy for extrapyramidal side effects, networks and pathways are often involved in drug response. The clonazepam or lorazepam for insomnia and sennoside for constipation common variants associated with antipsychotics may have smaller during the 2-month study period. Clinical effect was assessed on the Positive and Negative Syndrome Scale effect sizes and combining genetic information may improve their (PANSS), including the positive, negative and general psychopathology global predictive value. A combination of variants in the genes subscales. For all of the recruited patients, clinical assessments were coding for 5-HT2A, 5-HT2C, H2 receptors (histamine receptor type conducted on the day of admission, as well as in the treatment process. All 2) and for the serotonin transporter (serotonin-transporter-linked PANSS ratings were conducted independently by two qualified psychiatrists polymorphic region: HTTLPR) have been reported to result in the who were blind to the genotype of patients. And the inter-rater reliability correct prediction of clozapine response in 76% of cases.12 between the two psychiatrists is good. Treatment efficacy of the four Recently, a great deal of attention has been focused on antipsychotics was measured in terms of the reduction in PANSS scores. pathway-based association analysis for multivariate studies, which Patients were classified as good responders (reduction of PNASS score ⩾ o can combine variants in genes in a known molecular pathway to 50%) and poor responders (reduction of PNASS score 50%) for analysis. test whether the pathway is associated with the phenotype.13,14 Unlike single variant analyses or genome-wide association study, Genotyping which are often underpowered to uncover variants with small Genomic DNA was extracted from the blood samples using the QiaAmpH effect sizes, this approach could provide a more powerful analysis isolation system (Qiagen, Basel, Switzerland). All of the SNPs were for pharmacogenomic analysis. genotyped on the ViiA 7 System (Life Technologies, Carlsbad, CA, USA) In the present study, in order to comprehensively discover the using TaqMan technology. The genotyping probes were designed by the μ predictor of these four antipsychotic drug response, based on Applied Biosystems service. The standard 5 l PCR reaction was carried out pathway approach, genetic variants previously associated with or using TaqMan Universal PCR Master Mix Reagent Kits (Applied Biosystems, Foster City, CA, USA) under the guidelines provided. Genotype accuracy potentially involved in antipsychotic drug response were selected was assessed by re-genotyping within a subsample, and reproducibility for genotyping with Taqman genotyping technology in a relatively was routinely 499%. Genotyping completion rate was 490%. larger schizophrenia subjects. Also, we performed combined analysis of different markers, which lay the foundation for discovering higher predictive value genomic variants with anti- Statistical analysis psychotic treatment response in Chinese population. Power calculations for the association analysis were performed using the case–control model for discrete traits.15 Hardy–Weinberg equilibrium, pairwise linkage disequilibrium estimation and haplotype reconstruction MATERIALS AND METHODS were performed using SHEsis (http://analysis.bio-x.cn).16 The allele, genotype and haplotype frequencies of each polymorphism were compared between Candidate gene and single-nucleotide polymorphism (SNP) the good-responders and poor-responder groups using χ2 test. Odds ratios selections and their 95% confidence intervals were also calculated. Correction for All of the candidate genes were drawn from previous reports in which multiple testing was conducted using the Bonferroni correction method. genes were previously associated with or potentially involved in P ⩽ 0.05 was regarded as statistically significant. The gene–gene interactions antipsychotic drug response based on the five criteria: (1) genes and was analyzed on multifactor dimensionality reduction (MDR) software, which SNPs associated with disease pathogenesis; (2) genes and SNPs associated is a model-free and non-parametrical approach method that can identify with the mechanism of action of drugs; (3) genes associated with drug high dimensional gene–gene or gene–environment interactions in a small metabolism (phase I and II reactions); (4) genes and SNPs associated with population as previously described.17–19 During the MDR process, the data drug transporters; and (5) pleiotropic genes and SNPs involved in are randomly divided into 10 equal parts, and the MDR is developed on 9/10 multifaceted cascades and metabolic reactions. Through systematically of the data (training set) and then tested on 1/10 of the remaining data searching literatures and databases (dbSNP, HapMap, 1000 Genome), the (testing set). Two-locus interactions through four-locus interactions were functional and tag polymorphisms across every gene, of which the minor analyzed in the present study. Training-balanced accuracy and testing- allele frequency is at least 1% in Chinese population, were finally selected. balanced accuracy are obtained for each selected models. This procedure is The selected genes and variants are listed in Supplementary Table S1. A repeated 10 times with the data split into 10 different training and testing total of 77 SNPs of 25 genes have been selected for association analysis, sets. The number of times a particular set of factors is identified is defined as including pathogenic genes and genes encoding drug metabolism cross-validation consistency. The multifactor model with the best testing- enzymes, drug transporters, neurotransmitter receptors and other genes. balanced accuracy and cross-validation consistency is selected. Statistical significance of testing-balanced accuracy of each selected multifactor model Subjects is determined by comparing the average prediction error from the observed data with the distribution of average prediction errors under the null The unrelated Chinese Han inpatients with schizophrenia were enrolled in hypothesis of no association derived empirically from 1000 permutations. – the present study. All of the patients (men and women, aged 18 60 years) The null hypothesis was rejected when the upper-tail Monte Carlo P-value were recruited from Shanghai Mental Health Center and have been derived from the permutation test was ⩽ 0.05. prescribed risperidone, clozapine, quetiapine and chlorpromazine separately. According to the following criteria, 995 subjects were recruited in this study: (1) they satisfied the Diagnostic and Statistical Manual of Mental RESULTS Disorders-IV criteria for schizophrenia; (2) they had no physical complica- tion or other substance abuse; (3) they had no history suggesting Patient-related variables and clinical parameters resistance to antipsychotic treatment; and (4) they had not previously Of the total of 995 subjects included in this study, the number of received other antipsychotics. The study protocol was reviewed and good responders and poor responders for each drug are listed in proved by the Shanghai Ethical Committee of Human Genetic Resources. Supplementary Table S2. Descriptive statistics for patient-related Statement of informed consent was obtained from all subjects after full variables such as age and daily dosage with regard to sex in four explanation of the procedure. antipsychotic drugs are summarized in Supplementary Table S2. No significant difference in drug dosage between the sexes was noted. Clinical assessment For the risperidone, clozapine, quetiapine and chlorpromazine subjects, the initial dosages were 1, 50, 100–200, 50–100 mg day − 1, respectively, Statistical power and then gradually increased to 4–6, 200–400, 600–800, 400–600 - Given the parameters from this study, a power calculation mg day − 1 within the first week, respectively, which was maintained until indicated that the study was sufficiently powered (at least

The Pharmacogenomics Journal (2016), 357 – 365 © 2016 Macmillan Publishers Limited Pharmacogenomics of schizophrenia QXuet al 359 82.86%) to detect many of the SNPs. With the present sample size, DISCUSSION we assigned the risk allele frequency as 0.2, as most SNPs were It is well known that the sample recruitment is more difficult for common with risk allele frequency 40.2 in our study. Assuming a antipsychotic drug-efficient pharmacogenomics studies. So, many dominant 1 d.f. test (A* versus aa) model, we have 82.86% power studies are mainly performed in a small sample set, which account to detect a genotypic relative risk (GRR) of 1.5 at a significance for the inconsistent results in our and other group’s studies. And level of 0.05. A GRR of 1.5 is even small for drug response of this promotes us to further expand our sample size. In the present schizophrenia as the GRRs of most SNPs were 41.5 in this study. study, nearly 1000 subjects were selected for discovering the effect marker. In order to unravel the complicating genetic factors Effects of individual polymorphisms on drug response implicated in response to antipsychotic medication, 77 SNPs of 25 genes involved in drug-metabolizing enzymes, transporters and Schizophrenia inpatients were genotyped for genetic variants drug target pathways have been investigated for four commonly previously associated with or potentially involved in the four used antipsychotic drugs, which provide the potential for multi- antipsychotics' response. We have excluded several variations that variate interaction analysis and comparison between different had too low a degree of heterozygosity in our subjects or SNP that fi fi – o drugs. To the best of our knowledge, this is the rst comprehen- signi cantly deviated from Hardy Weinberg equilibrium (P 0.05); sive analysis of genomic variants with treatment response to we did not analyze their effects on antipsychotic treatment risperidone, clozapine, quetiapine and chlorpromazine in the response. The comparison of the genotype and allele frequencies Chinese Han population. in the good- and poor-responder groups revealed several We compared the significant SNPs in this study with previous significant allelic and genotypic associations with schizophrenia reports in Table 5. Despite existing biological evidence, genetic (Table 1). Genotype and allele frequencies of other SNPs associated studies have not consistently shown significant associations with resperidone, clozapine, quetiapine and chlorpromazine treat- between polymorphisms of suggested genes and antipsychotic ment are listed in Supplementary Table S3. However, most of the response. Our study can validate most of previous associations associations remain no longer significant after Bonferroni correc- (such as CYP2D6, CYP2C19, COMT, ABCB1, DRD3 and HTR2C). In tion. But there are three SNPs, rs6269, rs5993883 and rs4818, particular, the polymorphisms of COMT, CYP2C19, DRD3, HTR2C located in the COMT gene that remain significant even after and AKT1 have been observed to be associated with more than Bonferroni correction associated with quetiapine response one antipsychotic drug in our study. And this study can also (P =0.008, P = 0.007 and P = 0.007, respectively). The rs6269 A, provide some new findings to the current relationships between rs5993883 T and rs4818 C alleles in the COMT gene were very genetic polymorphisms and antipsychotic response (such as TNIK, frequently observed in the poor responders; odds ratios were 1.938 RELN, NOTCH4 and SLC6A2). So our study is of great significance to (95% confidence interval, 1.307–2.873), 1.877 (95% confidence improve current knowledge on the pharmacogenomics of interval, 1.289–2.732) and 0.504 (95% confidence interval, 0.336– schizophrenia, thus promoting the implementation of persona- 0.755) respectively (Table 1). We also compared our results with lized medicine in schizophrenia. previous reports in Table 5. Most antipsychotic drugs undergo extensive metabolism by CYP enzymes. Our study found that rs1135840 in CYP2D6 was associated with quetiapine response. The result may be attributed to the Association of haplotypes with drug response current findings that polymorphism of CYP2D6 would influence the We carried out linkage disequilibrium analysis between each pair clinical response of quetiapine. The rs1135840 allele in the CYP2D6 of all the SNPs in every gene for the four antipsychotic drugs. Then gene (G4C) is a putative nonsynonymous SNP known to influence fi we de ne the blocks to evaluate haplotype association with the enzyme activity both in vitro experiments and in some genetic critera of minor allele frequency 45% and D’40.68. We have association studies for other drugs.20 Another study suggests that listed haplotypes with significant different frequencies associated this SNP might be associated with the metabolic rate of with the four antipsychotics' response in Table 2. There are two antipsychotics.21 In the present study, C allele was observed to be haplotypes consisting of four SNPs in COMT gene, rs1544325, overrepresented in good responders to quetiapine treatment. Our rs5993883, rs6269 and rs4818, that showed significant association results also showed that rs4986893 in CYP2C19 was associated with with quetiapine treatment response (Table 2). Even after resperidone treatment response, and this result was replicated in permutation, the P-values indicate significant association (GGGG, chlorpromazine response. In vitro study indicated that rs4986893 in χ2 = 9.219 permuted Po0.001; ATAC, χ2 = 6.745 permuted CYP2C19 involves a G4A mutation and creates a premature stop P = 0.016). Haplotype GGGG was more prevalent in good codon and a truncated protein that would cause poor responders than in poor responders, while haplotype ATAC was metabolism.22 Whereas in vivo study indicated that patients more prevalent in poor responders than in good responders. harboring A allele in rs4986893 did not show significant differences in serum drug levels when compared with those harboring G 23 Multivariate interaction analysis of drug response allele. And our result found that the rs4986893 A allele were observed more frequently in the good responders (Table 2). But the MDR analyses were used to investigate probable multivariate associations were marginal association, which remained no interactions (including all SNPs, sex, age, weight and so on) significant associations after multiple correction (P = 0.038 and associated with each drug treatment response. Two-, three- and 0.044 for clozapine and chlorpromazine, respectively). The data of four-order interaction models were considered. The results are in vivo and in vitro studies was not contradictory. Patients who summarized in Table 3. The two-locus model (including the rs6269 harbor poor metabolizer phenotype do not mean that they would polymorphisms in the COMT gene, the rs3813929 polymorphism be poor responders to drug treatment. Our findings that CYP2D6 in the HTR2C gene), which had a testing-balanced accuracy of and CYP2C19 were associated with antipsychotic treatment 60.08% and a cross validation consistency of 6/10, was regarded response could provide new insight into interactions between the as the best. The permutation testing showed that the two-locus cytochrome P450 system and antipsychotics, thus it can lay the model was significant at the 0.044 level. Table 4 summarizes the foundation for personalized medicine. two-locus–genotype combinations associated with high and low Taking a comprehensive overview of our results, we found that sensitivity, along with the corresponding distribution of good the most significant SNPs associated with antipsychotic treatment responders and of poor responders, for each multilocus–genotype response were located in the COMT gene. Because of the combination. importance of the dopamine system in antipsychotic activity,

Table 1. Genotype and allele frequencies of signiﬁcant SNPs associated with resperidone, clozapine, quetiapine and chlorpromazine treatment

Associated Gene SNPs Allele OR (95% CI) Pa(d.f. = 1) χ2 Genotype Pa(d.f. = 2) χ2 antipsychotics

Resperidone COMT rs737865 A G AA AG GG Good responders 241 (0.730) 89 (0.270) 0.525 (0.323–0.853) 0.009 6.894 80 (0.485) 81 (0.491) 4 (0.024) 0.011 8.961 Poor responders 134 (0.838) 26 (0.163) 55 (0.688) 24 (0.300) 1 (0.013) AKT1 rs2494738 A G AA AG GG Good responders 163 (0.420) 225 (0.580) 0.599 (0.422–0.85) 0.004 8.311 34 (0.175) 95 (0.490) 65 (0.335) 0.015 8.463 Poor responders 104 (0.547) 86 (0.453) 27 (0.284) 50 (0.526) 18 (0.189) CYP2C19 rs4986893 A G AA AG GG Good responders 34 (0.086) 360 (0.914) 2.267 (1.029–4.994) 0.038 4.327 2 (0.010) 30 (0.152) 165 (0.838) 0.120 4.245 Poor responders 8 (0.040) 192 (0.960) 0 (0.000) 8 (0.080) 92 (0.920) Clozapine ABCB1 rs2032582 C T CC CT TT Good responders 225 (0.745) 77 (0.255) 1.586 (1.049–2.399) 0.028 4.819 85 (0.563) 55 (0.364) 11 (0.073) 0.009 9.525 Poor responders 105 (0.648) 57 (0.352) 41 (0.506) 23 (0.284) 17 (0.210) HTR2C rs498177 A G AA AG GG Good responders 1 (0.003) 301 (0.997) 0.138 (0.015–1.244) 0.039 4.242 0 (0.000) 1 (0.007) 150 (0.993) 0.218 3.045 Poor responders 4 (0.024) 166 (0.976) 1 (0.012) 2 (0.024) 82 (0.965) DRD3 rs167771 A G AA AG GG Good responders 220 (0.743) 76 (0.257) 0.588 (0.36–0.958) 0.032 4.600 78 (0.527) 64 (0.432) 6 (0.041) 0.074 5.207 Poor responders 133 (0.831) 27 (0.169) 54 (0.675) 25 (0.312) 1 (0.013) TNIK rs2088885 A C AA AC CC Good responders 103 (0.337) 203 (0.663) 1.57 (1.031–2.393) 0.035 4.450 19 (0.124) 65 (0.425) 69 (0.451) 0.104 4.533 Poor responders 42 (0.244) 130 (0.756) 7 (0.081) 28 (0.326) 51 (0.593) HTR2C rs3813929 C T CC CT TT Good responders 266 (0.917) 24 (0.083) 2.001 (1.109–3.613) 0.020 5.449 123 (0.848) 20 (0.138) 2 (0.014) 0.081 5.036 Poor responders 144 (0.847) 26 (0.153) 64 (0.753) 16 (0.188) 5 (0.059) COMT rs4646316 C T CC CT TT Good responders 145 (0.558) 115 (0.442) 0.597 (0.388–0.919) 0.019 5.540 35 (0.269) 75 (0.577) 20 (0.154) 0.025 7.375 Poor responders 95 (0.679) 45 (0.321) 28 (0.400) 39 (0.557) 3 (0.043) COMT rs4680 A G AA AG GG Good responders 58 (0.196) 238 (0.804) 0.684 (0.433–1.08) 0.102 2.671 5 (0.034) 48 (0.324) 95 (0.642) 0.024 7.495 Poor responders 41 (0.263) 115 (0.737) 10 (0.128) 21 (0.269) 47 (0.603) TCF4 rs9960767 A C AA AC CC Good responders 257 (0.857) 43 (0.143) 1.895 (1.168–3.074) 0.009 6.837 108 (0.720) 41 (0.273) 1 (0.007) 0.016 8.317 Poor responders 123 (0.759) 39 (0.241) 43 (0.531) 37 (0.457) 1 (0.012) SLC6A2 rs2242480 C T CC CT TT Good responders 221 (0.762) 69 (0.238) 1.556 (1.012–2.392) 0.043 4.085 78 (0.538) 65 (0.448) 2 (0.014) 0.008 9.709 Poor responders 105 (0.673) 51 (0.327) 35 (0.449) 35 (0.449) 8 (0.103) SLC6A2 rs5569 A G AA AG GG Good responders 80 (0.284) 202 (0.716) 0.631 (0.421–0.947) 0.026 4.970 15 (0.106) 50 (0.355) 76 (0.539) 0.057 5.736 Poor responders 64 (0.386) 102 (0.614) 12 (0.145) 40 (0.482) 31 (0.373) AKT1 rs3001371 C T CC CT TT Good responders 107 (0.374) 179 (0.626) 0.658 (0.447–0.97) 0.034 4.493 20 (0.140) 67 (0.469) 56 (0.392) 0.084 4.963 Poor responders 79 (0.476) 87 (0.524) 21 (0.253) 37 (0.446) 25 (0.301) Quetiapine NOTCH4 rs3131296 C T CC CT TT Good responders 391 (0.931) 29 (0.069) 1.851 (1.024–3.345) 0.039 4.256 182 (0.867) 27 (0.129) 1 (0.005) 0.110 4.406 Poor responders 153 (0.879) 21 (0.121) 68 (0.782) 17 (0.195) 2 (0.023) DRD3 rs6280 C T CC CT TT Good responders 111 (0.275) 293 (0.725) 0.999 (0.663–1.505) 0.995 0.000 21 (0.104) 69 (0.342) 112 (0.554) 0.046 6.142 Poor responders 44 (0.275) 116 (0.725) 3 (0.037) 38 (0.475) 39 (0.487) CYP2D6 rs1135840 C G CC CG GG Good responders 117 (0.284) 295 (0.716) 1.598 (1.042–2.45) 0.031 4.661 21 (0.102) 75 (0.364) 110 (0.534) 0.095 4.715 Poor responders 35 (0.199) 141 (0.801) 6 (0.068) 23 (0.261) 59 (0.670) COMT rs6269 G A GG GA AA Good responders 161 (0.389) 253 (0.611) 1.938 (1.307–2.873) 8.9 × 10 − 4 11.041 31 (0.150) 99 (0.478) 77 (0.372) 0.004 11.176 Poor responders 44 (0.247) 134 (0.753) (corrected 6 (0.067) 32 (0.360) 51 (0.573) Pb = 0.008) COMT rs5993883 G T GG GT TT Good responders 186 (0.447) 230 (0.553) 1.877 (1.289–2.732) 9.4 × 10 − 4 10.948 38 (0.183) 110 (0.529) 60 (0.288) 0.003 11.626 Poor responders 53 (0.301) 123 (0.699) (corrected 7 (0.080) 39 (0.443) 42 (0.477) Pb = 0.007) COMT rs4818 C G CC CG GG Good responders 245 (0.609) 157 (0.391) 0.504 (0.336–0.755) 8.1 × 10 − 4 11.217 76 (0.378) 93 (0.463) 32 (0.159) 0.004 11.011 Poor responders 127 (0.756) 41 (0.244) (corrected 47 (0.560) 33 (0.393) 4 (0.048) Pb = 0.007) COMT rs1544325 A G AA AG GG Good responders 102 (0.255) 298 (0.745) 0.629 (0.422–0.936) 0.022 5.272 12 (0.060) 78 (0.390) 110 (0.550) 0.061 5.602 Poor responders 55 (0.353) 101 (0.647) 8 (0.103) 39 (0.500) 31 (0.397) HTR2C rs1414334 C G CC CG GG Good responders 22 (0.056) 368 (0.944) 0.002 9.982 2 (0.010) 18 (0.092) 175 (0.897) 0.009 9.389 Poor responders 0 (0.000) 170 (1.000) 0 (0.000) 0 (0.000) 85 (1.000) Chlorpromazine RELN rs7341475 A G AA AG GG Good responders 26 (0.160) 136 (0.840) 3.174 (1.173–8.586) 0.018 5.643 3 (0.037) 20 (0.247) 58 (0.716) 0.073 5.228 Poor responders 5 (0.057) 83 (0.943) 0 (0.000) 5 (0.114) 39 (0.886) DRD3 rs6280 C T CC CT TT Good responders 33 (0.201) 131 (0.799) 0.532 (0.293–0.965) 0.036 4.377 4 (0.049) 25 (0.305) 53 (0.646) 0.067 5.403 Poor responders 27 (0.321) 57 (0.679) 3 (0.071) 21 (0.500) 18 (0.429) AKT1 rs2494732 C T CC CT TT Good responders 12 (0.162) 62 (0.838) 0.813 (0.256–2.579) 0.725 0.124 0 (0.000) 12 (0.324) 25 (0.676) 0.018 8.077 Poor responders 5 (0.192) 21 (0.808) 2 (0.154) 1 (0.077) 10 (0.769) COMT rs6269 G A GG GA AA Good responders 71 (0.382) 115 (0.618) 2.021 (1.152–3.543) 0.013 6.139 9 (0.097) 53 (0.570) 31 (0.333) 0.012 8.832 Poor responders 22 (0.234) 72 (0.766) 3 (0.064) 16 (0.340) 28 (0.596) COMT rs174696 C T CC CT TT Good responders 78 (0.443) 98 (0.557) 0.593 (0.359–0.98) 0.041 4.184 20 (0.227) 38 (0.432) 30 (0.341) 0.134 4.024 Poor responders 55 (0.573) 41 (0.427) 16 (0.333) 23 (0.479) 9 (0.188) CYP2C19 rs4986893 A G AA AG GG Good responders 11 (0.069) 149 (0.931) 6.423 (0.815–50.604) 0.044 4.060 1 (0.013) 9 (0.113) 70 (0.875) 0.156 3.713 Poor responders 1 (0.011) 87 (0.989) 0 (0.000) 1 (0.023) 43 (0.977) Abbreviations: CI, conﬁdence interval; OR, odds ratio; SNP, single-nucleotide polymorphism. aUncorrected P-values. bPermuted P-values, number of permutations: 1000.

Table 2. Haplotypes with signiﬁcant different frequencies associated with antipsychotics

Associated Gene Haplotype Good Poor χ2 P-valuea P-valueb antipsychotics responder (%) responder (%)

Clozapine COMT rs4646312 rs6269 rs4818 rs4680 rs4646316 C G G G T 34.9 25.8 3.95 0.0469 0.400 Quetiapine COMT rs1544325 rs5993883 rs6269 rs4818 G G G G 31.4 19.3 9.219 0.0024 0.010 A T A C 23.0 33.2 6.745 0.0094 0.039 CYP2D6 rs1135840 rs16947 C G 27.9 19.8 4.257 0.0391 0.107 CYP2D6 rs1135840 rs1065852 C G 27.6 19.8 3.946 0.047 0.118 CYP2D6 rs16947 rs1065852 G G 30.7 22.2 4.394 0.0361 0.141 Chloropromazine CYP2C19 rs4986893 rs4244285 rs2104161 G G T 15.0 25.2 4.53 0.0333 0.056 COMT rs6269 rs4818 G G 38.2 26.5 4.076 0.0435 0.226 A C 16.7 28.8 5.842 0.0156 0.051 COMT rs4680 rs174697 rs174699 rs165599 A A T A 0.6 4.3 4.658 0.0309 0.144 G G C G 0.3 3.2 4.467 0.0346 0.163 DRD3 rs963468 rs6280 G T 40.7 28.7 3.842 0.05 0.071 aUncorrected P-value. bPermuted P-value, number of permutations: 1000. P-values in bold indicate signiﬁcant for the association.

Table 3. Summary of gene–gene interaction models detected by MDR

Antipsychotics Model Training Testing Cross- P-valuea Bal. Acc. Bal. Acc. validation consistency

Resperidone rs737865, rs208885 0.6572 0.5025 3/10 0.449 rs737865, rs208885, rs2032582 0.7458 0.5125 3/10 0.375 rs1076562, rs4646316, rs16947, rs2032582 0.8581 0.48 2/10 0.656 Clozapine rs4646316, rs2242480 0.6663 0.4423 2/10 0.865 rs1079727, rs4646316, rs1065852 0.7621 0.4362 2/10 0.905 rs1800497, rs174696, rs324036, rs2032582C/A 0.8772 0.5008 2/10 0.453 Chlorpromazine rs1125394, rs174696 0.7145 0.5573 8/10 0.218 rs1544325, rs174696, rs1135840 0.8243 0.5149 3/10 0.387 rs4244285, rs6275, rs174696, rs2494746 0.9317 0.5687 2/10 0.302 Quetiapine rs3813929, rs6269 0.6739 0.6008 6/10 0.044 rs6275, rs1065852, rs174699 0.7452 0.5086 4/10 0.370 rs6265, rs2075652, rs2032582, rs174699 0.8581 0.5213 3/10 0.291 Abbreviations: Acc, accuracy; Bal, balanced; MDR, multifactor dimensionality reduction. aP-value based on 1000 permutations. P-value in bold indicates signiﬁcant for the association.

proteins regulating dopamine availability or functionality may Table 4. inﬂuence response to treatment. The COMT enzyme that catalyzes Distribution of high- and low-sensitivity genotypes in the the degradation of dopamine and limits its availability falls within best two-locus model 24,25 this group. Among the investigated genes, COMT polymorph- Multilocus– Good Poor responders' Association with isms were found to be strikingly associated with treatment genotype responders' number drug response response in the present study. A total of eight SNPs were combinations number associated with antipsychotic treatment response, and the associations were observed in all the four antipsychotics. These rs6269 rs3813929 results provide more evidence that COMT has an important role in the pharmacokinetics and the therapeutic efﬁcacy of many GG CC 19 3 High-sensitivity antipsychotics. A famous common functional variant rs4680 GG CT 5 2 High-sensitivity (G1947A) in the COMT gene has been shown to reduce the GG TT 5 0 High-sensitivity activity of the soluble isoform of COMT by approximately fourfold, GA CC 77 16 High-sensitivity which is hypothesized to result in relatively greater dopamine GA CT 7 7 Low-sensitivity activity.26 This is one of the most extensively studied SNPs and the GA TT 6 7 Low-sensitivity AA CC 61 42 Low-sensitivity high activity Val allele has been linked to disorders, such as 27–30 AA CT 8 2 High-sensitivity schizophrenia and diminished frontal cognitive function. AA TT 4 5 Low-sensitivity Many studies have evaluated the possible association of this

Table 5. Summary of the signiﬁcant SNPs in this study and in previous reported signiﬁcant association with response to antipsychotics

Gene Association in this study Previous studies

SNPs Associated antipsychotics

CYP2D6 rs1135840 Quetiapine A57,58;I59;N60–62 CYP2C19 rs4986893 Resperidone, chlorpromazine A61,63;I59;N21 COMT rs737865, rs4646316, rs4680, rs6269, rs5993883, Resperidone, clozapine, quetiapine, A6,7,27,31,32,55,64–67 rs4818, rs1544325, rs174696 chlorpromazine ABCB1 rs2032582 Clozapine A21,61,68–71 SLC6A2 rs2242480, rs5569 Clozapine N72 DRD3 rs167771, rs6280 Clozapine, quetiapine, chlorpromazine A73–75;I76;N77–79 HTR2C rs498177, rs3813929, rs1414334 Clozapine, quetiapine A12,80–82;N74,83,84 AKT1 rs2494738, rs3001371 Resperidone, clozapine A43;N37 TNIK rs2088885 Clozapine — NOTCH4 rs313296 Quetiapine — RELN rs7341475 Chlorpromazine — Abbreviation: SNP, single-nucleotide polymorphism. Note: A, indicates associated signiﬁcantly (Po0.05); I, implicated in vitro experiments; N, not associated (P40.05); —, not found.

common polymorphism with antipsychotic response.6,27,31–33 In and mouse model studies have reported the significant associa- this study, the AA genotype of rs4680 was observed to be tions of these three genes with schizophrenia among populations overrepresented in poor responders. Nevertheless, rs4680 is not around the world.44–52 However, no study has aimed to the only variation in the gene, and several other variations have investigate the associations of the three high-risk schizophrenia also been observed to be associated with antipsychotic response susceptibility genes with antipsychotic response. Also, there have in our study (Table 5). Notably, rs6269, rs5993883 and rs4818 were been many studies reporting the association of norepinephrine found to be striking significantly associated with quetiapine transporter (SLC6A2) polymorphism with drug response in response, and they remained significant associations even after attention-deficit hyperactivity disorder or major depressive Bonferroni correction. There are few studies to preform associa- disorder, but no one has reported the association with anti- tion analysis of the three SNPs with antipsychotic treatment psychotic response. We are the first to study the associations, and response. This is the first study to report the associations of the observed significant associations of TNIK, RELN, NOTCH4 and three SNPs with antipsychotic response. So our results have SLC6A2 polymorphisms with antipsychotic response. enriched the current data on association of genetic polymorphism Current data suggest that haplotype-based association methods with antipsychotic treatment response. are more powerful than analogous allele-based methods when RAC-alpha serine/threonine-protein kinase (AKT1) is known to variants occur within the same gene. The study of haplotypes, influence the biology of dopamine signaling.34–36 As AKT1 gene which characterize the linkage disequilibrium structure of several has a crucial functional role in intracellular signaling pathways, markers, may show stronger association with phenotype than many studies reported genetic association of the AKT1 gene with individual markers.53 In support of this idea, Drysdale et al.54 schizophrenia to prove that AKT1 is important for the pathophy- showed that the interaction of several SNPs in a haplotype can siology of schizophrenia.37–40 A recent study reported that a affect the physiological reaction and response to treatment and number of the top canonical pathways dysregulated in treatment- that a single SNP is not sufficient to predict the genetic basis of naive patients signal through AKT1 that was upregulated. After drug response. In this study, haplotype analysis showed significant treatment, AKT1 returned to control levels and less dysregulation associations of COMT, CYP2D6, CYP2C19 and DRD3 haplotypes of these canonical pathways was observed. This study supports with antipsychotic response (Table 2). What is striking about the immune dysfunction and pathways involving AKT1 in the data presented here is that the association of COMT haplotypes etiopathophysiology of schizophrenia and their response to with antipsychotic response was observed in three different drugs, antipsychotic medication.41 A study reported that the DRD2 and especially two haplotypes consisting of four markers in COMT AKT1 polymorphisms altered dose–response effects of antipsy- gene rs1544325, rs5993883, rs6269 and rs4818 (GGGG and ATAC) chotic drugs on cognition in schizophrenia, suggesting that showed highly significant association with quetiapine treatment genetic polymorphism of AKT1 may, at least in part, be relevant to response. To our knowledge, only one study reported an individual variation in the cognitive outcome of antidopaminergic association of two-marker COMT haplotype with drug response treatment.36 The study of Blasi et al.42 demonstrated that DRD2– before.55 The haplotypes consisting of rs1544325, rs5993883, AKT1 signaling pathway is involved in modulating the effect of rs6269 and rs4818 may serve as better markers to guiding clinical antipsychotic treatment in humans. However, to the best of our individual medication in the future than a single SNP marker. As knowledge, there is only one study reporting the direct this is a new finding, our results should be treated with caution association of AKT1 genetic polymorphisms with antipsychotic and these findings need to be replicated in more studies. treatment in a small sample size.43 Our study validates the There is a growing awareness that drug response involves association in large sample size. And moreover, the associations complex molecular networks and pathways. The common variants were not sporadic, two SNPs: rs2494738 and rs3001371 were associated with drug response traits may have small effects so that observed to be associated with resperidone and clozapine they are able to predict a response only when combined. A response, respectively. It is the new direct evidence to associate previous study observed this combination effect in drug AKT1 genetic polymorphisms with antipsychotic treatment. response.56 In our study, we conducted a combined analysis As TRAF2 and NCK interacting kinase (TNIK), reelin gene (RELN) between all the selected genes encoding drug metabolism and notch homolog 4 (NOTCH4) have important roles in central enzyme, drug transporters, neurotransmitter receptors and so on nervous system, they are the most extensively examined using MDR analysis. The two-locus model (including the rs6269 schizophrenia candidate genes. Many genetic association studies polymorphisms in the COMT gene, the rs3813929 polymorphism

The Pharmacogenomics Journal (2016), 357 – 365 © 2016 Macmillan Publishers Limited Pharmacogenomics of schizophrenia QXuet al 363 in the HTR2C gene) between drug metabolism enzyme and 10 McClay JL, Adkins DE, Aberg K, Stroup S, Perkins DO, Vladimirov VI et al. Genome- neurotransmitter receptors was selected as the best combined wide pharmacogenomic analysis of response to treatment with antipsychotics. variants, which means that the interactions among the two SNPs Mol Psychiatry 2011; 16:76–85. might be associated with antipsychotic response. 11 Clark SL, Souza RP, Adkins DE, Aberg K, Bukszar J, McClay JL et al. Genome-wide Our results, concerning the safety profile of antipsychotics, are association study of patient-rated and clinician-rated global impression of severity 23 – in agreement with other studies. Because of the cross-sectional during antipsychotic treatment. Pharmacogenetics Genomics 2013; :69 77. 12 Arranz MJ, Munro J, Birkett J, Bolonna A, Mancama D, Sodhi M et al. Pharmaco- study design and limited sample size, as well as problems inherent genetic prediction of clozapine response. Lancet 2000; 355:1615–1616. in multiple testing, the results need to be interpreted with caution 13 Sun YV. Integration of biological networks and pathways with genetic association and require replication and functional validation. studies. Hum Genet 2012; 131: 1677–1686. 14 Atias N, Istrail S, Sharan R. Pathway-based analysis of genomic variation data. Curr Opin Genet Dev 2013; 23:622–626. CONCLUSION 15 Purcell S, Cherny SS, Sham PC. Genetic Power Calculator: design of linkage In conclusion, based on pathway approach, the present study first and association genetic mapping studies of complex traits. Bioinformatics 2003; performed an integrative study by investigating most of the 19:149–150. genetic variants previously associated with or potentially involved 16 Shi YY, He L. SHEsis, a powerful software platform for analyses of linkage disequilibrium, haplotype construction, and genetic association at polymorphism in four commonly used antipsychotic drug response in about 1000 loci (vol 15, pg 97, 2005). Cell Res 2006; 16:851–851. Chinese schizophrenia cases. Associations with treatment response 17 Hahn LW, Ritchie MD, Moore JH. Multifactor dimensionality reduction software for for several genes, such as CYP2D6, CYP2C19, COMT, ABCB1, DRD3 detecting gene-gene and gene-environment interactions. Bioinformatics 2003; 19: and HTR2C have been verified in our study. Also, we found several 376–382. new candidate gene (TNIK, RELN, NOTCH4 and SLC6A2), and 18 Qin S, Zhao X, Pan Y, Liu J, Feng G, Fu J et al. An association study of the combinations (haplotype rs1544325–rs5993883–rs6269–rs4818 in N-methyl-D-aspartate receptor NR1 subunit gene (GRIN1) and NR2B subunit gene COMT), which are associated with treatment response to the four (GRIN2B) in schizophrenia with universal DNA microarray. Eur J Hum Genet 2005; drugs. Also, multivariate interaction analysis show the combination 13:807–814. of rs6269 in COMT and rs3813929 in HTR2C may work as a 19 Su MW, Tung KY, Liang PH, Tsai CH, Kuo NW, Lee YL. Gene-gene and gene- environmental interactions of childhood asthma: a multifactor dimension predictor to improve the clinical antipsychotics response. Although reduction approach. PLoS One 2012; 7: e30694. we expanded the sample size compared with previous studies, 20 de Groot MJ, Wakenhut F, Whitlock G, Hyland R. Understanding CYP2D6 inter- these findings also need to be verified by subsequent studies in actions. Drug Discov Today 2009; 14:964–972. larger cohorts of patients with longer follow-up design and collect 21 Lee ST, Ryu S, Kim SR, Kim MJ, Kim S, Kim JW et al. Association study of 27 data including more relevant clinical factors as well as further annotated genes for clozapine pharmacogenetics: validation of preexisting molecular basis studies, which has the potential to firmly establish studies and identification of a new candidate gene, ABCB1, for treatment the potential of these polymorphisms as pharmacogenetic markers response. J Clin Psychopharmacol 2012; 32: 441–448. of antipsychotics drugs in Chinese population. 22 Tsai M-H, Lin K-M, Hsiao M-C, Shen WW, Lu M-L, Tang H-S et al. Genetic polymorphisms of cytochrome P450 enzymes influence metabolism of the antidepressant escitalopram and treatment response. Pharmacogenomics 2010; CONFLICT OF INTEREST 11:537–546. 23 Lee S-T, Ryu S, Kim S-R, Kim M-J, Kim S, Kim J-W et al. Association study of 27 fl The authors declare no con ict of interest. annotated genes for clozapine pharmacogenetics: validation of preexisting studies and identification of a new candidate gene, ABCB1, for treatment 32 – ACKNOWLEDGMENTS response. J Clin Psychopharmacol 2012; : 441 448. 24 Akil M, Kolachana BS, Rothmond DA, Hyde TM, Weinberger DR, Kleinman JE. This work was supported by grants from the 863 Program (2012AA02A515, Catechol-O-methyltransferase genotype and dopamine regulation in the 2012AA021802), the 973 Program (2010CB529600), the National Key Technology human brain. J Neurosci 2003; 23: 2008–2013. R&D Program (2012BAI01B09), the National Nature Science Foundation of China 25 Chen JS, Lipska BK, Halim N, Ma QD, Matsumoto M, Melhem S et al. Functional (81121001, 81273596, J1210047, 30900799, 30972823, 81171272), Public Science and analysis of genetic variation in catechol-O-methyltransferase (COMT): effects on Technology Research Funds (201210056), the Shanghai Jiaotong University mRNA, protein, and enzyme activity in postmortem human brain. Am J Hum Genet Interdisciplinary Research fund and the Shanghai Leading Academic Discipline 2004; 75:807–821. Project (B205). 26 Shield AJ, Thomae BA, Eckloff BW, Wieben ED, Weinshilboum RM. Human catechol O-methyltransferase genetic variation: gene resequencing and functional characterization of variant allozymes. Mol Psychiatry 2004; 9:151–160. REFERENCES 27 Woodward ND, Jayathilake K, Meltzer HY. COMT val108/158met genotype, 1 Wyatt RJ. Neuroleptics and the natural course of schizophrenia. Schizophr Bull cognitive function, and cognitive improvement with clozapine in schizophrenia. 1991; 17: 325–351. Schizophr Res 2007; 90:86–96. 2 Arranz MJ, de Leon J. Pharmacogenetics and pharmacogenomics of schizo- 28 Chen X, Wang X, O'Neill AF, Walsh D, Kendler KS. Variants in the catechol-O- phrenia: a review of last decade of research. Mol Psychiatry 2007; 12:707–747. methyltransferase (COMT) gene are associated with schizophrenia in Irish 3 van Os J, Kapur S. Schizophrenia. Lancet 2009; 374:635–645. high-density families. Mol Psychiatry 2004; 9: 962–967. 4 Zhang JP, Malhotra AK. Pharmacogenetics and antipsychotics: therapeutic effi- 29 Goghari VM, Sponheim SR. Differential association of the COMT Val15Met cacy and side effects prediction. Expert Opin Drug Metab Toxicol 2011; 7:9–37. polymorphism with clinical phenotypes in schizophrenia and bipolar disorder. 5 Xu Q, Wu X, Xiong Y, Xing Q, He L, Qin S. Pharmacogenomics can improve Schizophr Res 2008; 103:186–191. antipsychotic treatment in schizophrenia. Front Med 2013; 7: 180–190. 30 Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE et al. 6 Yamanouchi Y, Iwata N, Suzuki T, Kitajima T, Ikeda M, Ozaki N. Effect of DRD2, 5- Effect of COMT Val(108/158) Met genotype on frontal lobe function and risk for HT2A, and COMT genes on antipsychotic response to risperidone. Pharmacoge- schizophrenia. Proc Natl Acad Sci USA 2001; 98:6917–6922. nomics J 2003; 3: 356–361. 31 Molero P, Ortuno F, Zalacain M, Patino-Garcia A. Clinical involvement of 7 Gao S, Hu Z, Cheng J, Zhou W, Xu Y, Xie S et al. Impact of catechol-O- catechol-O-methyltransferase polymorphisms in schizophrenia spectrum dis- methyltransferase polymorphisms on risperidone treatment for schizophrenia orders: influence on the severity of psychotic symptoms and on the response to and its potential clinical significance. Clin Biochem 2012; 45: 787–792. neuroleptic treatment. Pharmacogenomics J 2007; 7:418–426. 8 Brennan MD. Pharmacogenetics of second-generation antipsychotics. Pharma- 32 Illi A, Kampman O, Hanninen K, Anttila S, Mattila KM, Katila H et al. Catechol-O- cogenomics 2014; 15: 869–884. methyltransferase val108/158met genotype and response to antipsychotic 9 McClay JL, Adkins DE, Aberg K, Bukszar J, Khachane AN, Keefe RS et al. Genome- medication in schizophrenia. Hum Psychopharm Clin 2007; 22: 211–215. wide pharmacogenomic study of neurocognition as an indicator of antipsychotic 33 Weickert TW, Goldberg TE, Mishara A, Apud JA, Kolachana LS, Egan MF et al. treatment response in schizophrenia. Neuropsychopharmacology 2011; 36: Catechol-O-methyltransferase Val(108/158)Met genotype predicts working 616–626. memory response to antipsychotic medications. Biol Psychiatry 2004; 56:677–682.

© 2016 Macmillan Publishers Limited The Pharmacogenomics Journal (2016), 357 – 365 Pharmacogenomics of schizophrenia QXuet al 364 34 Emamian ES, Hall D, Birnbaum MJ, Karayiorgou M, Gogos JA. Convergent evi- 57 Spina E, Avenoso A, Salemi M, Facciola G, Scordo MG, Ancione M et al. Plasma dence for impaired AKT1-GSK3beta signaling in schizophrenia. Nat Genet 2004; concentrations of clozapine and its major metabolites during combined treat- 36: 131–137. ment with paroxetine or sertraline. Pharmacopsychiatry 2000; 33: 213–217. 35 Harris SL, Gil G, Robins H, Hu WW, Hirshfield K, Bond E et al. Detection of func- 58 Centorrino F, Baldessarini RJ, Frankenburg FR, Kando J, Volpicelli SA, Flood JG. tional single-nucleotide polymorphisms that affect apoptosis. Proc Natl Acad Sci Serum levels of clozapine and norclozapine in patients treated with selective USA 2005; 102: 16297–16302. serotonin reuptake inhibitors. Am J Psychiatry 1996; 153: 820–822. 36 Tan HY, Chen AG, Kolachana B, Apud JA, Mattay VS, Callicott JH et al. 59 Olesen OV, Linnet K. Contributions of five human cytochrome P450 isoforms to Effective connectivity of AKT1-mediated dopaminergic working memory net- the N-demethylation of clozapine in vitro at low and high concentrations. works and pharmacogenetics of anti-dopaminergic treatment. Brain 2012; 135: J Clin Pharmacol 2001; 41:823–832. 1436–1445. 60 Melkersson KI, Scordo MG, Gunes A, Dahl ML. Impact of CYP1A2 and CYP2D6 37 Xu MQ, Xing QH, Zheng YL, Li S, Gao JJ, He G et al. Association of AKT1 gene polymorphisms on drug metabolism and on insulin and lipid elevations and polymorphisms with risk of schizophrenia and with response to antipsychotics in insulin resistance in clozapine-treated patients. J Clin Psychiatry 2007; 68: the Chinese population. J Clin Psychiatry 2007; 68:1358–1367. 697–704. 38 Thiselton DL, Vladimirov VI, Kuo PH, McClay J, Wormley B, Fanous A et al. AKT1 is 61 Jaquenoud Sirot E, Knezevic B, Morena GP, Harenberg S, Oneda B, Crettol S et al. associated with schizophrenia across multiple symptom dimensions in the Irish ABCB1 and cytochrome P450 polymorphisms: clinical pharmacogenetics of clo- study of high density schizophrenia families. Biol Psychiatry 2008; 63:449–457. zapine. J Clin Psychopharmacol 2009; 29: 319–326. 39 Schwab SG, Hoefgen B, Hanses C, Hassenbach MB, Albus M, Lerer B et al. Further 62 Kakihara S, Yoshimura R, Shinkai K, Matsumoto C, Goto M, Kaji K et al. Prediction evidence for association of variants in the AKT1 gene with schizophrenia in a of response to risperidone treatment with respect to plasma concencentrations of sample of European sib-pair families. Biol Psychiatry 2005; 58:446–450. risperidone, catecholamine metabolites, and polymorphism of cytochrome 40 Norton N, Williams HJ, Dwyer S, Carroll L, Peirce T, Moskvina V et al. Association P450 2D6. Int Clin Psychopharmacol 2005; 20:71–78. analysis of AKT1 and schizophrenia in a UK case control sample. Schizophr Res 63 Altar CA, Hornberger J, Shewade A, Cruz V, Garrison J, Mrazek D. Clinical validity of 2007; 93:58–65. cytochrome P450 metabolism and serotonin gene variants in psychiatric 41 Kumarasinghe N, Beveridge NJ, Gardiner E, Scott RJ, Yasawardene S, Perera A et al. pharmacotherapy. Int Rev Psychiatry 2013; 25: 509–533. Gene expression profiling in treatment-naive schizophrenia patients identifies 64 Bertolino A, Caforio G, Blasi G, De Candia M, Latorre V, Petruzzella V et al. abnormalities in biological pathways involving AKT1 that are corrected by anti- Interaction of COMT Val(108/158) met genotype and olanzapine treatment on psychotic medication. Int J Neuropsychopharmacol 2013; 16: 1483–1503. prefrontal cortical function in patients with schizophrenia. Am J Psychiatry 2004; 42 Blasi G, Napolitano F, Ursini G, Taurisano P, Romano R, Caforio G et al. DRD2/AKT1 161:1798–1805. interaction on D2 c-AMP independent signaling, attentional processing, and 65 Zhao QZ, Liu BC, Zhang J, Wang L, Li XW, Wang Y et al. Association between a response to olanzapine treatment in schizophrenia. Proc Natl Acad Sci USA 2011; COMT polymorphism and clinical response to risperidone treatment: a 108: 1158–1163. pharmacogenetic study. Psychiatr Genet 2012; 22:298–299. 43 Ikeda M, Yamanouchi Y, Kinoshita Y, Kitajima T, Yoshimura R, Hashimoto S et al. 66 Fijal BA, Kinon BJ, Kapur S, Stauffer VL, Conley RR, Jamal HH et al. Candidate-gene Variants of dopamine and serotonin candidate genes as predictors of response to association analysis of response to risperidone in African-American and white risperidone treatment in first-episode schizophrenia. Pharmacogenomics 2008; 9: patients with schizophrenia. Pharmacogenomics J 2009; 9: 311–318. 1437–1443. 67 Illi A, Kampman O, Anttila S, Roivas M, Mattila KM, Lehtimaki T et al. Interaction 44 Ayalew M, Le-Niculescu H, Levey DF, Jain N, Changala B, Patel SD et al. Con- between angiotensin-converting enzyme and catechol-O-methyltransferase vergent functional genomics of schizophrenia: from comprehensive under- genotypes in schizophrenics with poor response to conventional neuroleptics. standing to genetic risk prediction. Mol Psychiatry 2012; 17: 887–905. Eur Neuropsychopharmacol 2003; 13: 147–151. 45 Potkin SG, Turner JA, Guffanti G, Lakatos A, Fallon JH, Nguyen DD et al. A genome- 68 Consoli G, Lastella M, Ciapparelli A, Catena Dell'Osso M, Ciofi L, Guidotti E et al. wide association study of schizophrenia using brain activation as a quantitative ABCB1 polymorphisms are associated with clozapine plasma levels in psychotic phenotype. Schizophr Bull 2009; 35:96–108. patients. Pharmacogenomics 2009; 10: 1267–1276. 46 Wang Q, Charych EI, Pulito VL, Lee JB, Graziane NM, Crozier RA et al. The psy- 69 Bozina N, Kuzman MR, Medved V, Jovanovic N, Sertic J, Hotujac L. Associations chiatric disease risk factors DISC1 and TNIK interact to regulate synapse com- between MDR1 gene polymorphisms and schizophrenia and therapeutic position and function. Mol Psychiatry 2011; 16: 1006–1023. response to olanzapine in female schizophrenic patients. J Psychiatr Res 2008; 42: 47 Andreasen NC, Wilcox MA, Ho BC, Epping E, Ziebell S, Zeien E et al. Statistical 89–97. epistasis and progressive brain change in schizophrenia: an approach for exam- 70 Vijayan NN, Mathew A, Balan S, Natarajan C, Nair CM, Allencherry PM et al. ining the relationships between multiple genes. Mol Psychiatry 2012; 17: Antipsychotic drug dosage and therapeutic response in schizophrenia is influ- 1093–1102. enced by ABCB1 genotypes: a study from a south Indian perspective. Pharma- 48 Tost H, Weinberger DR. RELN rs7341475 and schizophrenia risk: confusing, yet cogenomics 2012; 13: 1119–1127. somehow intriguing. Biol Psychiatry 2011; 69:e19. 71 Xing Q, Gao R, Li H, Feng G, Xu M, Duan S et al. Polymorphisms of the ABCB1 gene 49 Wedenoja J, Tuulio-Henriksson A, Suvisaari J, Loukola A, Paunio T, Partonen T et al. are associated with the therapeutic response to risperidone in Chinese schizo- Replication of association between working memory and Reelin, a potential phrenia patients. Pharmacogenomics 2006; 7: 987–993. modifier gene in schizophrenia. Biol Psychiatry 2010; 67: 983–991. 72 Meary A, Brousse G, Jamain S, Schmitt A, Szoke A, Schurhoff F et al. Pharmaco- 50 Yang XB, Kang C, Liu H, Yang J. Association study of the reelin (RELN) gene with genetic study of atypical antipsychotic drug response: involvement of the nor- Chinese Va schizophrenia. Psychiatr Genet 2013; 23: 138. epinephrine transporter gene. Am J Med Genet B Neuropsychiatr Genet 2008; 51 Aberg KA, Liu Y, Bukszar J, McClay JL, Khachane AN, Andreassen OA et al. A 147B:491–494. comprehensive family-based replication study of schizophrenia genes. JAMA 73 Scharfetter J, Chaudhry HR, Hornik K, Fuchs K, Sieghart W, Kasper S et al. Dopa- Psychiatry 2013; 70:573–581. mine D3 receptor gene polymorphism and response to clozapine in schizo- 52 Ikeda M, Aleksic B, Yamada K, Iwayama-Shigeno Y, Matsuo K, Numata S et al. phrenic Pakastani patients. Eur Neuropsychopharmacol 1999; 10:17–20. Genetic evidence for association between NOTCH4 and schizophrenia supported 74 Vehof J, Burger H, Wilffert B, Al Hadithy A, Alizadeh BZ, Snieder H. Clinical by a GWAS follow-up study in a Japanese population. Mol Psychiatry 2013; 18: response to antipsychotic drug treatment: association study of polymorphisms in 636–638. six candidate genes. Eur Neuropsychopharmacol 2012; 22:625–631. 53 Shifman S, Bronstein M, Sternfeld M, Pisante-Shalom A, Lev-Lehman E, Weizman A 75 Adams DH, Close S, Farmen M, Downing AM, Breier A, Houston JP. Dopamine et al. A highly significant association between a COMT haplotype and schizo- receptor D3 genotype association with greater acute positive symptom remission phrenia. Am J Hum Genet 2002; 71: 1296–1302. with olanzapine therapy in predominately caucasian patients with chronic schi- 54 Drysdale CM, McGraw DW, Stack CB, Stephens JC, Judson RS, Nandabalan K et al. zophrenia or schizoaffective disorder. Hum Psychopharmacol 2008; 23:267–274. Complex promoter and coding region beta(2)-adrenergic receptor haplotypes 76 Daskalakis ZJ, George TP. Clozapine, GABA(B), and the treatment of resistant alter receptor expression and predict in vivo responsiveness. Proc Natl Acad Sci schizophrenia. Clin Pharmacol Ther 2009; 86:442–446. USA 2000; 97: 10483–10488. 77 Hwang R, Zai C, Tiwari A, Muller DJ, Arranz MJ, Morris AG et al. Effect of dopamine 55 Gupta M, Bhatnagar P, Grover S, Kaur H, Baghel R, Bhasin Y et al. Association D3 receptor gene polymorphisms and clozapine treatment response: exploratory studies of catechol-O-methyltransferase (COMT) gene with schizophrenia and analysis of nine polymorphisms and meta-analysis of the Ser9Gly variant. Phar- response to antipsychotic treatment. Pharmacogenomics 2009; 10:385–397. macogenomics J 2010; 10:200–218. 56 Ritchie MD, Haas DW, Motsinger AA, Donahue JP, Erdem H, Raffanti S et al. 78 Xuan J, Zhao X, He G, Yu L, Wang L, Tang W et al. Effects of the dopamine D3 Drug transporter and metabolizing enzyme gene variants and nonnucleoside receptor (DRD3) gene polymorphisms on risperidone response: a reverse-transcriptase inhibitor hepatotoxicity. Clin Infect Dis 2006; 43: 779–782. pharmacogenetic study. Neuropsychopharmacology 2008; 33: 305–311.

The Pharmacogenomics Journal (2016), 357 – 365 © 2016 Macmillan Publishers Limited Pharmacogenomics of schizophrenia QXuet al 365 79 Barlas IO, Cetin M, Erdal ME, Semiz UB, Basoglu C, Ay ME et al. Lack of association 82 Liu BC, Zhang J, Wang L, Li XW, Wang Y, Wei ZY et al. HTR2C promoter poly- between DRD3 gene polymorphism and response to clozapine in Turkish schi- morphisms are associated with risperidone efficacy in Chinese female patients. zoprenia patients. Am J Med Genet B Neuropsychiatr Genet 2009; 150B:56–60. Pharmacogenomics 2010; 11:685–692. 80 Sodhi MS, Arranz MJ, Curtis D, Ball DM, Sham P, Roberts GW et al. Association 83 Rietschel M, Naber D, Fimmers R, Moller HJ, Propping P, Nothen MM. Efficacy and between clozapine response and allelic variation in the 5-HT2C receptor gene. side-effects of clozapine not associated with variation in the 5-HT2C receptor. Neuroreport 1995; 7:169–172. Neuroreport 1997; 8:1999–2003. 81 Reynolds GP, Yao Z, Zhang X, Sun J, Zhang Z. Pharmacogenetics of treatment in 84 Klemettila JP, Kampman O, Seppala N, Viikki M, Hamalainen M, Moilanen E et al. first-episode schizophrenia: D3 and 5-HT2C receptor polymorphisms separately Association study of the HTR2C, leptin and adiponectin genes and serum marker associate with positive and negative symptom response. Eur Neuropsycho- analyses in clozapine treated long-term patients with schizophrenia. Eur Psy- pharmacol 2005; 15: 143–151. chiatry 2014; 30:296–302.

Supplementary Information accompanies the paper on the The Pharmacogenomics Journal website (http://www.nature.com/tpj)