Pharmacogenetics of Antipsychotic Drug Treatment: Update and Clinical Implications Free

Schizophrenia is a severe neurodevelopmental disorder with high heritability [1, 2]. Likewise, genetic factors have also been implicated in antipsychotic medication used in the treatment of schizophrenia [3, 4]. Antipsychotics are currently the main medication for treatment of schizophrenia; however, the clinical response and adverse effects vary substantially between individuals, which typically leads clinicians to “trial-and-error” procedures before optimal treatment medication is found [5].

Pharmacogenetics/pharmacogenomics (PGx) is the use of genomic data to understand drug metabolism and response, which may allow clinicians to select medications based on the genetic variability across patients [6]. Multiple studies have investigated PGx approaches in order to identify genotype-specific dosing and predict antipsychotic responses and/or adverse effects [6]. Over the past years, specific recommendations on how to apply such genetic information, including commercial tests, have been proposed for clinical practice.

This narrative review summarizes most recent publications and recommendations related to the genetics of antipsychotic treatment. Furthermore, this review aims to shed light on the clinical utility of PGx.

First, recently published articles were searched using PubMed. The following search terms were applied: (schizophreni* or schizoaffective or psychosis or psychoses or “severe mental illness”) and (antipsychotic* or neuroleptic*) and (pharmacogenetic* or genetic* or genomic* or gene or “single nucleotide polymorphism” or SNP or polymorphism). Studies were included if they (1) were peer-reviewed original articles or meta-analyses, (2) referred to PGx in the treatment of patients with schizophrenia or related psychoses (i.e., serum/plasma concentrations or dosages, treatment response, and major adverse effects including antipsychotic-induced weight gain [AIWG], movement disorders, hyperprolactinemia, and clozapine-associated neutropenia), and (3) were published in English between January 2015 and April 2018. Next, expert group recommendations and Food and Drug Administration (FDA) drug labeling were reviewed regarding PGx information for antipsychotics.

Ninety-two original articles and 6 systematic reviews and meta-analyses were included in this review. Among the original articles, 7 reported on serum/plasma concentrations or dosage (Table 1), 44 reported on responses to antipsychotics (Table 2), and 45 reported on antipsychotic-induced adverse effects (Table 3); some of the included articles reported on two or three topics (i.e., antipsychotic concentrations or dosage, response, and adverse effects).

Cytochrome P450 (CYP) enzymes mediate the metabolism of various drugs, and for certain CYPs, the genotype affects serum (plasma) drug levels [7]. Previous studies have explored the role of pharmacokinetic (PK) genes for antipsychotic serum/plasma concentrations, particularly of the CYP genotype [8]. Four classes of CYP metabolizer profiles have been established: poor (PM), intermediate, normal, and ultrarapid metabolizers [9]. These phenotypes are based on the multiallelic nature of the CYP enzyme’s genetic construct, which reflects several polymorphisms such as single nucleotide polymorphisms (SNPs) [8]. Notably, the FDA has included warnings for higher side effects at standard dosages for CYP2D6 PM; the details are described later. Nonetheless, the relationship between antipsychotic serum/plasma concentrations and the CYP genotypes continues to be investigated.

Three studies in this review reported a potential in­fluence of polymorphisms of CYP genotypes (i.e., the CYP1A2, 2C19, 2D6, and 3A4 genes) on PK profiles (i.e., the antipsychotic serum/plasma concentration) [7, 10, 11]. One study reported a significant association between olanzapine serum concentrations and the CYP1A2*1D and *1F polymorphisms [10], while 2 other studies also reported that polymorphisms in CYP2D6, CYP1A2, and DRD3 affected antipsychotic concentrations to some extent [7, 11]. The membrane transport protein P-glycoprotein, which is encoded by the ATP-binding cassette subfamily B member 1 (ABCB1) gene, helps clear antipsychotic medications across the blood-brain barrier [12]. One study investigated the influence of the ABCB1 rs1045642 polymorphism and the ATP-binding cassette subfamily C member 1 (ABCC1) rs212090 polymorphism on clozapine and norclozapine serum levels [13]. The combination of ABCB1 and ABCC1 homozygotes was associated with increased clozapine and norclozapine serum levels, although there were no significant associations between the ABCB1 and ABCC1 SNPs and serum concentrations.

In addition, antipsychotic dosage is affected by PK and pharmacodynamic (PD) factors. Three studies that investigated the influence of genetics on antipsychotic dosage were included in the review [14‒16]. Hettige et al. [14, 17] reported a significant association between higher antipsychotic dosage and the γ-aminobutyric acid type A receptor β1 subunit (GABRB1) gene polymorphisms rs16860087 and rs4627835, although the authors were unsuccessful in reproducing the results of their previous study. However, the other 2 studies [15, 16] (using a genome-wide association approach) reported negative findings. One of the 2 studies investigated the association between antipsychotic dosage and polygenic risk scores in schizophrenia derived from significant risk loci identified by the Psychiatric Genomics Consortium (PGC2) but found no association between them [15]. Given that antipsychotic dosage can be influenced by several factors such as symptom severity and duration of illness, further research on this topic will be needed.

Response to antipsychotics is a complex phenotype involving genetic and clinical factors such as symptom severity and adherence level. Despite this challenging issue, various genetic variants related to PK and PD have been investigated in association with the response to antipsychotics.

Most studies included in this review focused on genetic polymorphisms, associated with response to antipsychotics, and how these polymorphisms related to clinical symptom improvement in patients with schizophrenia (e.g., a reduction in scores in the Brief Psychiatric Rating Scale [18] or the Positive and Negative Syndrome Scale [PANSS] [19]).

Antipsychotic drugs are primarily metabolized by CYP2D6, CYP3A4, and CYP1A2, and approximately 40, 23, and 18% of antipsychotics comprise major substrates for CYP2D6, CYP3A4, and CYP1A2, respectively [20]. Numerous studies have investigated the association between CYP variants and response to antipsychotics [8]. However, most have resulted in negative findings [8]. In the current review, 2 studies reported significant associations between antipsychotic response and the CYP3A43 rs680055 and CYP2D6 rs3892097 polymorphisms [21, 22]. Although another study found that CYP2C19 rs4986893 and CYP2D6 rs1135840 were associated with antipsychotic response, the associations did not remain significant after correction for multiple testing [22, 23].

In addition to the CYP enzymes, common polymorphisms of the P-glycoprotein-coding gene (ABCB1) have been associated with antipsychotic response. Previous studies investigating the rs1045642 (C3435T), rs2032582 (G2677T/A), and rs1128503 (C1236T) polymorphisms and their association with antipsychotic response are summarized elsewhere; the findings in this area of research are still inconsistent [12]. The current review includes 3 studies that investigated the association between polymorphisms in ABCB1 and response to antipsychotics. Of these, 1 study reported positive findings for the rs2032582 polymorphism, although the association did not remain significant after correction for multiple testing [23]. The other studies reported negative findings regarding the rs1045642 and rs2032582 polymorphisms, although a combination of a loss-of-function CYP2D6 allele and the TT genotype of ABCB1 rs2032582 was associated with poor response to antipsychotic treatment in one of those 2 studies [24, 25]. The current review includes several studies suggesting that response to antipsychotics is influenced by CYP genetic variants and ABCB1. However, this conclusion is limited by several factors such as heterogeneous ethnicity, medication, and study duration. Therefore, the association between PK gene variants and antipsychotic responses remains controversial, and further investigation is required.

Several studies investigated PD-related genes and treatment responses, particularly dopamine receptor D2 (DRD2), which plays a critical role in antipsychotic treatment. DRD2 antagonism is a common mechanism of antipsychotic action and is considered necessary for antipsychotic efficacy [26]. Accordingly, both DRD2 and D2-like receptor genes, including DRD3 and DRD4, have been extensively investigated as candidate genes that influence antipsychotic responses [27, 28]. We identified 7 studies that investigated the association between polymorphisms of the DRD2 gene and response to antipsychotics [21, 29‒34], where three polymorphisms of DRD2 (rs180498, rs2514218, and rs1079597) were significantly associated with treatment response [21, 30‒32]. In particular rs2514218, which is located 47-kb upstream of the DRD2 gene, was previously reported as one of the genome-wide significant SNPs associated with risk of schizophrenia [35]. Further, it was associated with response to antipsychotics such as clozapine and risperidone in independent samples [31].

In addition, Blasi et al. [33] reported that a combination of the DRD2 rs1076560 and serotonin 2A receptor gene (HTR2A) rs6314 polymorphisms affected the response to antipsychotic treatment. Although a previous meta-analysis reported an association of DRD2 rs1799732 (–141C Ins/Del) with antipsychotic treatment response, no statistically significant associations were found in 2 studies included in this review [29, 34]. The association of DRD3 rs6280 (Ser9Gly) with antipsychotic responses has been extensively investigated [36]. One study included in this review reported a significant association between DRD3 rs6280 (Ser9Gly) and response to quetiapine [23]. A previous meta-analysis showed a nonsignificant trend toward lower response in Ser allele carriers [27]. Although DRD3 may theoretically affect antipsychotic responses via a high affinity of antipsychotics for the D3 receptor, empirical research findings remain inconsistent [36]. In contrast, 3 studies included in this review reported no significant associations of DRD1 rs5326, rs4867798, rs4532, and rs686, DRD3 rs6280, and DRD4 rs1800955 and rs4646984 with response to antipsychotics [25, 29, 37]. Similarly, a recent meta-analysis also showed no significant association between the DRD1 rs4532 polymorphism and response to antipsychotics, including cloza­pine [38].

In addition, catechol-O-methyltransferase (COMT), which is another dopamine pathway-related gene, has been investigated. The COMT gene is located within the 22q11 chromosomal locus, which is involved in schizophrenia susceptibility [39]. The COMT enzyme is involved in dopamine degradation [40], making the COMT gene a candidate gene for an association with antipsychotic response. In this review, 6 of 9 studies reported associations of polymorphisms of the COMT gene (including marker rs4680) with improvement in the response to treatment by antipsychotics, including clozapine [23, 41‒45]. Furthermore, a recent meta-analysis also showed that the COMT Val158Met (rs4680) polymorphism was significantly associated with response to antipsychotics [46]. These findings suggest that COMT gene variants, particularly COMT Val158Met (rs4680), are associated with antipsychotic response. However, the role of Val158Met polymorphism in antipsychotic treatment regulation is yet to be entirely elucidated [47].

Most second-generation antipsychotics display a high affinity for serotonin receptors, which is considered as contributing to antipsychotic response [48]. Accordingly, several serotonin pathway-associated genes have been studied. Three studies in this review reported that the serotonin 1A receptor (HTR1A) polymorphism rs6295 was significantly associated with improvement in clinical symptoms [45, 49, 50]. Takekita et al. [49] indicated that the HTR1A polymorphism rs1364043 and the rs10042486-rs6295-rs1364043 haplotype might influence negative symptoms on the PANSS. A recent meta-analysis by Takekita et al. [51] showed that the HTR1A rs6295 polymorphism might be associated with improvement of negative symptoms but not with overall or positive symptoms. However, given the relatively small number of studies included in the analysis (n = 10) and the heterogeneity of antipsychotics, further studies using larger sample populations and homogeneous treatments are required to confirm this effect. It is thought that HTR2A contributes to the pathophysiology of hallucinations, and the receptor has been a major target among many second-generation antipsychotics [52]. Two of the 3 studies identified in this review indicated the impact of the HTR2A polymorphism rs6314 on antipsychotic treatment response modulation [21, 33, 42].

Associations of the HTR2C polymorphisms rs1328685, rs643627, rs498177, rs3813929, and rs1414334 with response to antipsychotics were also found [23, 42]. In contrast, the serotonin 7 receptor (HTR7) polymorphisms rs12412496, rs7916403, and rs1935349 did not appear to improve antipsychotic treatments [53].

Several other candidate genes, including glutamate-related genes, are also identified in this review (Table 2) [10, 23, 25, 34, 43, 54‒66]. For example, Stevenson et al. [67] examined the association between glutamate gene polymorphisms and treatment response to risperidone in patients with first-episode psychosis (n = 86). They found that several SNPs were associated with antipsychotic treatment response; of these, the type-7 metabotropic glutamate receptor (GRM7) polymorphism rs2069062 showed the most robust finding [67].

Similarly, GRM7 rs2133450 was significantly associated with Emsley’s positive domain derived from the PANSS in a genome-wide association study (GWAS), using a sample of patients treated with risperidone [68]. As for other genes, synaptosomal-associated protein 25 kDa (SNAP25) rs8636 and ankyrin repeat and sterile alpha motif domain-containing protein 1B (ANKS1B) rs7968606 were significantly associated with response to amisulpride [56, 69]. The brain-derived neurotrophic factor (BDNF) Val66Met (rs6265) polymorphism was not associated with response to antipsychotics in a recent meta-analysis that included 9 studies with a total of 2,461 patients treated with antipsychotics [70].

An alternative to the candidate gene approach is to use GWAS, which is characterized as a hypothesis-free approach [71]. To our knowledge, the first GWAS investigating antipsychotic response was performed in the context of a phase III clinical trial of iloperidone, published in 2009 [72]. Subsequently, several GWAS on antipsychotic response have been conducted [73‒76]. In the current review, we found 9 studies that used this approach [67, 68, 77‒82]. The following gene polymorphisms were significantly associated with treatment response: rs72790443 in multiple EGF-like domains 10 (MEGF10); rs1471786 in solute carrier family 1 member 1 (SLC1A1); rs9291547 in protocadherin 7 (PCDH7); rs12711680 in contactin-associated protein-like 5 (CNTNAP5); rs6444970 in TRAF2 and NCK-interacting kinase (TNIK); rs2133450, rs2069062, and rs2014195 in GRM7; rs9307122 and rs1875705 in glutamate ionotropic receptor delta type subunit 2 (GRID2); rs3129996 in protein phosphatase 1 regulatory subunit 18 (PPP1R18); and rs6435681 in erb-b2 receptor tyrosine kinase 4 (ERBB4) [67, 68, 77, 81‒83]. Among these studies, the study by Yu et al. [77] has one of the largest sample sizes reported so far (n = 2,413 in the discovery cohort and n = 1,379 in the replication sample). They found five novel genome-wide significant loci associated with treatment response in samples of Han Chinese ancestry (i.e., rs72790443 in MEGF10, rs1471786 in SLC1A1, rs9291547 in PCDH7, rs12711680 in CNTNAP5, and rs6444970 in TNIK). Furthermore, three additional loci were associated with drug-specific treatment responses (rs2239063 in CANCA1C for olanzapine, rs16921385 in SLC1A1 for risperidone, and rs17022006 in CNTN4 for aripiprazole). Although several factors such as duration of illness, duration of treatment, and concomitant therapy should be considered as covariates and validation studies in other ethnic populations are needed, the genes identified are clinically relevant and the findings can provide novel insights into the underlying mechanisms of antipsychotic action [84].

In addition, Li et al. [78] reported that common genetic variants related to synaptic adhesion complexes, scaffolding, and the alternative splicing regulator were associated with treatment response to lurasidone, although their findings were not statistically significant after correction for multiple testing. Ovenden et al. [79] found tentative associations between response to treatment and the mannosidase beta (MANBA), collagen type IX alpha 2 chain (COL9A2), and nuclear factor kappa B subunit 1 (NFKB1) genes.

Recently, the rapid progress of next-generation sequencing, also called massively parallel sequencing, has led to revolutionary changes in medical genomics, including the field of psychiatry [85]. This method enables the sequencing of the whole genomes or exomes [85]. In the first study using exome sequencing to examine the pharmacogenomics of antipsychotic response, Drögemöller et al. [86] sequenced 11 South African patients with first-episode schizophrenia to identify genetic traits related to antipsychotic response. In the current review, we found a whole-exome sequencing study by the same group [83]. They identified two novel variants that were significantly associated with antipsychotic treatment response in two independent first-episode schizophrenia cohorts: rs13025959 in myosin VIIB (MYO7B) and rs10380 in 5-methyltetrahydrofolate-homocysteine methyltransferase reductase (MTRR). Although these new findings provide valuable information regarding future pharmacogenomic antipsychotic studies, further research will be needed.

Four main categories of antipsychotic-induced adverse effects were identified: (1) metabolic dysregulation, such as AIWG and metabolic syndrome; (2) movement disorders, including extrapyramidal symptoms (EPS), tardive dyskinesia (TD), and restless legs syndrome; (3) hormonal abnormalities, such as hyperprolactinemia; and (4) clozapine-induced agranulocytosis/granulocytopenia (CIAG).

AIWG is a common side effect in patients treated with antipsychotics, coinciding with metabolic dysregulation and heritability [87]. Although the mechanisms of AIWG are not fully understood, various genes appear to contribute to symptom onset (e.g., genes related to antipsychotic metabolism, neurotransmitter systems, and neuroendocrine systems) [88].

Twenty-eight of the studies in this review investigated the association between genetic factors and AIWG and metabolic dysregulation (Table 3). Regarding genes related to antipsychotic metabolism, Czerwensky et al. [10] found no significant association between AIWG and CYP1A2*1D and *1F polymorphisms. The ABCB1 rs1045642 and ABCC1 rs212090 polymorphisms were associated with AIWG in male patients treated with clozapine, although no significant associations were found in the total sample [13].

As for neurotransmitter systems, serotonin receptors influence central pathways affecting satiety and hunger [88]. Of the serotonin receptor genes, HTR2C has been extensively studied in association with AIWG, and the −759C/T polymorphism of HTR2C has been targeted in previous studies. One study included in this review replicated an association of the −759C/T polymorphism of HTR2C with AIWG [89]. Although another study produced negative findings, the authors found a significant association between insulinemia and T allele carriers [90]. A recent meta-analysis that included papers up until the end of 2014 reported 13 SNPs from nine genes being significantly associated with AIWG, including the −759C/T polymorphism of HTR2C [91]. In contrast, the rs1414334 polymorphism of HTR2C and the studied polymorphisms of HTR3A and HTR3B were not associated with AIWG and metabolic dysregulation [92‒94]. The dopamine system also appears to be associated with AIWG and metabolic dysregulation [88].

BDNF is a neurotrophic factor that plays an important role in neurogenesis, neuronal survival, and synaptic plasticity [95]. Evidence exists to indicate that BDNF contributes to food intake and body weight control [95]. Although researchers continue to examine the BDNF Val66Met (rs6265) polymorphism and its potential association with AIWG, the results are still inconsistent [96‒98]. Two studies included in this review reported signi­ficant associations between AIWG and the BDNF Met66Met polymorphism [99, 100]. Furthermore, the recent meta-analysis mentioned above [91] also showed a significant association between the BDNF polymorphism and AIWG.

Leptin is an adipocyte hormone that acts on the hypothalamus to regulate appetite and energy expenditure, which makes it a natural candidate for genetic AIWG studies. In particular, the −2548A/G (rs7799039) polymorphism of the LEP gene has been investigated in several studies [92]. However, the association between the −2548A/G polymorphism and AIWG remains inconclusive. In the current review, 1 study reported no significant association [92].

Antipsychotic-induced lipid biosynthesis may explain the metabolic side effects of antipsychotics [101]. Le Hellard et al. [102] investigated the association between AIWG and five major genes involved in the sterol regulatory element-binding protein (SREBP) activation of fatty acids and cholesterol production (SREBF1, SREBF2, SREBP cleavage-activation protein [SCAP], insulin-induced gene 1 [INSIG1], and INSIG2). They found a significant association between three markers localized within or near INSIG2 (rs17587100, rs17047764, and rs10490624) and AIWG. In addition, nominally significant associations with AIWG were found in INSIG1 (rs13223383), SREBF2 (rs4822064), and SCAP (rs12490383). The following gene polymorphisms were significantly associated with AIWG or the metabolic syndrome in this review: rs11654081 in SREBF1; rs2267443 and rs1052717 in SREBF2; and rs12151787, rs1049626, and rs17047733 in INSIG2 [103‒105]. Taken together, these genes might enhance the risk of antipsychotic- induced metabolic dysregulation, although these mixed findings warrant further investigation.

The mitochondria play a key role in energy metabolism, and mitochondrial dysfunction contributes to metabolic disorders [106]. Thus, the mitochondrial system has become an interesting target for AIWG studies, although there have not been many studies investigating mitochondrial genetic variance in AIWG; Gonçalves et al. [107] reported an association between the mitochondrial NADH dehydrogenase (ubiquinone) Fe-S protein 1 (NDUFS1) gene and AIWG. Translocator protein 18 kDa (TSPO) is also a mitochondrial gene that may regulate weight through its effects on mitochondrial metabolism [108]. One study included in this review reported a nominal association between AIWG and the TSPO rs6971 polymorphism in two independent samples [66]. More recently, Mittal et al. [109] examined associations of variants in nuclear-encoded mitochondrial genes and mitochondrial DNA with AIWG. They identified three nuclear-encoded mitochondrial genes conferring a risk for AIWG. These findings suggest that mitochondrial genes have a significant role in AIWG, although replication studies are needed with larger samples and/or other ethnic groups.

As for other genes, Brandl et al. [110] conducted a GWAS using the data set derived from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) sample, selecting a well-characterized and homogeneous subsample of individuals of European ancestry. Although none of the SNPs were significantly associated with AIWG, there were nominal associations for rs9346455 upstream of opioid growth factor receptor-like 1 (OGFRL1) and rs1059778 in iron-sulfur cluster assembly (IBA57). Notably, rs9346455 showed a significant association with AIWG in an independent sample of Europeans patients (n = 86) [110]. Another GWAS of 534 Han Chinese patients with schizophrenia found significant associations with rs10977144 and rs10977154 in protein tyrosine phosphatase, receptor type D (PTPRD), and these results were further validated in an independent replication cohort (n = 547) [111]. Notably, a recent meta-analysis that included studies through the end of 2014 found that 13 SNPs from nine genes were significantly associated with AIWG, with the SNPs in adrenoreceptor alpha 2A (ADRA2A), DRD2, HTR2C, and melanocortin 4 receptor (MC4R) having the largest effect sizes [91]. Additional details are summarized in Table 2. The previous studies from 2010 to 2015, focusing on gene variants associated with AIWG, are summarized in a recent comprehensive review [87].

In summary, several genes and polymorphisms have been studied in relation to AIWG with some findings showing independent replications. Nevertheless, considering that AIWG is a polygenic phenotype interacting with clinical and demographic factors, further studies are required to elaborate an algorithm for clinical practice [91].

Antipsychotic drugs are associated with various movement disorders. Among them, TD is a severe and often irreversible side effect, characterized by involuntary trunk, limb, and orofacial muscle movements. Several studies using genome-wide association and candidate gene approaches have been conducted to identify risk variants associated with TD development [112]. In the current review, 12 studies reported associations between variants of related genes and EPS, TD, and restless legs syndrome, with 6 of them reporting TD [113‒124]. A large number of studies have examined an association between CYP genes, including CYP2D6, and EPS/TD, and produced mixed findings [87]. One meta-analysis from 2005 suggested that loss-of-function alleles in CYP2D6 increased the risk of developing TD [125], whereas a recent study reported that CYP2D6 ultrarapid metabolizers were associated with an increased risk of TD [126]. Two studies included in this review found associations of polymorphic 1846G>A CYP2D6*4 with EPS and limb-trunk TD [116, 119]. Combined with the existing knowledge, it appears that CYP2D6 is a potential gene associated with the onset of TD [112]. In addition, CYP1A2 is involved in the metabolism of several antipsychotics, such as cloza­pine, olanzapine, and some first-generation antipsychotics [8]. Two studies included in this review reported an association between the CYP1A2*F polymorphic variant and TD [119, 122]. However, further studies using larger samples are required to confirm the association between CYP enzyme genes and TD, and to evaluate the clinical utility of these genes.

Neurotransmitter genes, including those within the dopamine, serotonin, γ-aminobutyric acid, and glutamate systems, may be linked to antipsychotic-induced movement disorders including TD [112]. For example, the Taq1A (rs1800497) polymorphism of the DRD2 gene [127], Ser9Gly (rs6280) of the DRD3 gene [128], and T102C (rs6313) of the HTR2A gene [129] have previously been investigated as risk variants for TD. Furthermore, vesicular monoamine transporter 2, which is coded by the solute carrier family 18 member A2 (SLC18A2) gene, has been implicated in TD [130, 131]. Zai et al. [131] found that the polymorphisms rs2015586, rs363390, rs363224, and rs14240 were associated with TD. Remarkably, these findings are consistent with clinical evidence demonstrating that a novel selective vesicular monoamine transporter 2 inhibitor (valbenazine) improves TD in patients with schizophrenia [132, 133]. In the current review, 1 study examined the association of 43 tag SNPs in five neurotransmitter genes with TD [118]. As a result, only rs1345423 in the glutamate receptor, ionotropic, N-methyl D-aspartate 2A (GRIN2A) gene showed a significant association with TD. A recent study focused on the influence of the neuregulin-1 (NGR1) and ERBB4 genes on TD [114], which help regulate N-methyl-D-aspartate and dopaminergic activity [134, 135]. They found a significant association between ERBB4 rs839523 and TD, although additional replication studies are needed to replicate this preliminary finding.

One study included in this review investigated an association between EPS and 202 SNPs in 31 neurotransmitter genes [117]. Four SNPs (rs9567733 [HTR2A], rs363341 [SLC18A2], rs1334802 [glutamate ionotropic receptor kainate type subunit 3 (GRIK3)], and rs1124491 [DRD2]) were significantly associated with EPS. Mas et al. [123] reported that the mTOR pathway was important for developing EPS. Using protein-protein interaction network analysis in 12 antipsychotic-naïve patients with first-episode psychosis, the same group showed that the NF-κB pathway (inflammatory response) and the mTOR pathway (lipid biosynthesis, insulin signaling, and autophagy) played a key role in EPS [121].

In summary, a large number of genes are potentially involved in antipsychotic-induced movement disorders. In the light of clinical evidence, SLC18A2 is a promising marker of TD risk. Nevertheless, further research and genetic validation studies are required to predict the risk for antipsychotic-induced movement disorders, including TD [112].

With regard to hormonal abnormality, hyperprolactinemia has been reported as a common adverse effect of antipsychotics [136]. Prolactin synthesis and secretion are inhibited by dopamine in the anterior pituitary gland through DRD2 activation [137]. Thus, an association between the DRD2 gene and hyperprolactinemia has been investigated [138]. A recent meta-analysis that included studies up until May 2015 showed that DRD2 Taq1A A1 carriers had significantly higher prolactin levels than A1 noncarriers among patients with schizophrenia, although further investigation will be needed due to several limitations, including a small sample size and heterogeneity of antipsychotics (n = 475; Hedges’ g 0.250; 95% CI: 0.068–0.433; p = 0.007) [138]. This review included 3 original studies that focused on the association between genetic variants and hyperprolactinemia (or an increase in prolactin levels). It appears that rs1341239 in the prolactin (PRL) gene, rs569959 and rs17326429 in the HTR2C gene, and rs4680 in the COMT gene are significantly associated with hyperprolactinemia [44, 139, 140]. Thus, investigating the effect of combinations of several gene variants will be required to predict the risk for hyperprolactinemia.

Clozapine is the only drug with proven efficacy in treatment-resistant schizophrenia. However, it has a risk of hematological side effects such as CIAG [141]. The human leukocyte antigen (HLA) genes have been implicated in clozapine-induced agranulocytosis (CIA) [142]. The first GWAS in relation to CIA, which was conducted by the Clozapine-Induced Agranulocytosis Consortium (CIAC), demonstrated that the HLA-DQB1 126Q and HLA-B 158T alleles were significantly associated with CIAG [143]. The current review includes 2 studies that examined the genetic factor for CIAG [144, 145]. One study reported a significant association of HLA-B*59: 01 – which was not compatible with the HLA-B 158T allele detected in the previous study – with CIAG in a Japanese sample [144]. The other study, using a genome-wide association approach, HLA allele imputation, exome array, and copy number variation, reported that clozapine- associated neutropenia was associated with rs149104283 located between SLCO1B3 and SLCO1B7 (solute carrier organic anion transporter family, member 1B3 and member 1B7) through a meta-analysis using the CIAC samples [145]. HLA-DQB1 (126Q) and HLA-B (158T), implicated in a previous study [143], could not be imputed with sufficient quality and thus were investigated in the study, while no imputed classic HLA or amino acid polymorphism was associated with clozapine-induced neutropenia. As for another finding, the association of HLA-DBQ1 6672G>C (rs113332494), which had been shown to be associated with the risk for CIA in another previous study [146], was replicated in this study.

In 2007, a first commercial test kit for CIA, PGxPredict:CLOZAPINE (Clinical Data, Inc., New Haven, CT, USA), which included the 6672G/C polymorphism in HLA-DQB1 for the detection of high-risk patients carrying the C allele, was launched, with a high specificity of 98.4% but a low sensitivity of 21.5% [47]. However, this test was discontinued due to lack of general interest at that time [147]. Further research focusing on multiple genes and polymorphisms will be needed to predict CIAG.

Currently, the US FDA provides information on pharmacogenomic biomarkers in their drug labeling for nine antipsychotics (i.e., aripiprazole, aripiprazole lauroxil, brexpiprazole, clozapine, iloperidone, perphenazine, pimozide, risperidone, and thioridazine) [148]. Seven out of nine refer to the CYP2D6 PM status, where dose adjustment recommendations are provided for the following antipsychotics: aripiprazole, aripiprazole lauroxil, brexpiprazole, clozapine, iloperidone, pimozide, and thioridazine. Similarly, the Pharmacogenomics Knowledgebase (PharmGKB) website lists ten antipsychotics where caution is advised for patients who are poor CYP2D6 metabolizers [149]. Drug labels with PGx information are provided for the following ten antipsychotics: aripiprazole, aripiprazole lauroxil, brexpiprazole, clozapine, iloperidone, olanzapine, perphenazine, pimozide, risperidone, and thioridazine.

The Clinical Pharmacogenetics Implementation Consortium (CPIC) [150], which was established in 2009 as a shared project of PharmGKB and the Pharmacogenomics Research Network (PGRN) [151], also provides useful gene-drug information for clinicians. This is available on the PharmGKB website, which assigns CPIC levels to genes and drugs according to the extent to which genetic information affects drug prescription. PharmGKB also provides levels of evidence for gene-drug associations. However, recommendations for antipsychotic medications have not yet been published. In contrast, the Royal Dutch Association for the Advancement of Pharmacy – Dutch Pharmacogenetics Working Group has also provided PGx drug dosing guidelines based on CYP2D6 genotypes for six antipsychotics: aripiprazole, clozapine, haloperidol, olanzapine, risperidone, and zuclopenthixol [152]. The guidelines are available on the PharmGKB website. An overview of the current recommendations is provided in Table 4.

In summary, recommendations exist on how to use PK variants (such as CYP2D6) for some antipsychotics; however, no PD variants for antipsychotics are available for clinical practice.

Since the FDA first approved a PGx testing platform in 2004 (i.e., that of Roche Molecular Systems Inc. and Affymetrix Inc. [Pleasanton, CA, USA] for CYP2D6 and CYP2C19), several pharmacogenomic tests have become commercially available [153]. Several randomized clinical trials have previously evaluated the effectiveness of PGx testing for antidepressants and compared it with that for standard treatments (i.e., treatment as usual), and promising results regarding the clinical utility of PGx testing have been observed [154‒156]. However, the precise clinical validity and utility of PGx tests regarding antipsychotics remain to be validated in randomized clinical trials. Therefore, randomized clinical trials investigating the effectiveness of pharmacogenomic tests for antipsychotics in clinical practice are required.

In this study, we reviewed recent developments regarding PGx in the treatment of schizophrenia. Numerous new studies have been published in the past years on antipsychotic serum/plasma concentrations, response to treatment, and adverse effects. While some studies could replicate previous findings, others could not, but many new associations have been reported which warrant further validation. Overall, the strongest associations are being noted of CYP2D6 metabolizer status and outcome with antipsychotic medications. The influence of CYP2D6 on antipsychotic dosage and the occurrence of side effects has also been integrated in recommendations provided by various expert groups and agencies such as the FDA. Not surprisingly, genetic tests are now being commercially offered across the world and have recently been reviewed for their clinical use [153]. In addition, some promising studies have detected associations between “older” candidate genes such as COMT and treatment response.

However, newer findings need to be considered as preliminary until further replication studies will have been conducted. Importantly, there is still a large heterogeneity across studies, possibly explaining some of the inconsistent findings. Sriretnakumar et al. [141] proposed five factors in addressing mixed findings: (1) heterogeneity of the study design, (2) adherence to treatment, (3) individual study-wide limitations, (4) population stratification, and (5) additional mechanisms beyond polymorphic DNA sequence effects. Although these authors particularly focused on the field of clozapine pharmacogenetics in patients with schizophrenia, their suggestions may be valid for the overall field of PGx regarding the response to antipsychotics and adverse effects in patients with schizophrenia.

In addition to the traditional candidate gene approach, new methodologies including GWAS, whole-exome sequencing, and gene-gene interaction have been adopted in PGx research [157]. These studies have produced novel findings and keep the promise to develop PGx research, although further replication studies are needed in other ethnic populations. Moreover, collaborations with pooled samples using robust biostatistical strategies are strongly encouraged, similar to the efforts made by the PGC in the broader field of psychiatric genetics or by the Pharmacogenomics of Antipsychot ic-Induced Weight Gain Consortium [158] and the CRESTAR group for clozapine response [159] in psychiatric pharmacogenetics.

In conclusion, PGx has the potential for optimizing antipsychotic treatment through the prediction of clinical outcomes without the need for conventional “trial-and-error” approaches. However, further well-designed investigations (such as trials to create algorithms for use in clinical practice and replication studies in large, well-characterized samples) are needed.

D.J.M. is supported by the Canadian Institutes of Health Research (CIHR Operating Grant MOP 142192), the National Institutes of Health (R01MH085801), and the Centre for Addiction and Mental Health Foundation (Joanne Murphy Professorship).

The authors declare no conflict of interest.

K.Y. wrote the first draft of the manuscript, and both authors conducted the literature search and contributed to and approved the final manuscript.