Pharmacogenetics and Antipsychotics: Therapeutic Efficacy and Side Effects Prediction

2.1 DRD2

From a candidate gene perspective, DRD2 is the ideal gene to study in relation to antipsychotic drug response. Antipsychotic clinical potency is highly correlated with the binding affinity to the dopamine D2 receptor^{4, 9}, D2 receptor occupancy by antipsychotic agents has been demonstrated to occur with all antipsychotic agents, and drugs targeting other receptor sites without dopamine D2 blockade have not yet been successfully developed as antipsychotics. DRD2 is located on chromosome 11q22, and consists of eight exons separated by seven introns. It contains a number of SNPs with differing frequencies amongst populations, and several of them have been studied in association with antipsychotic drug response. Among these, −141C Ins/Del (rs1799732), Taq1A (rs1800497), A-241G (rs1799978), Ser311Cys (rs1801028), and Taq1B (rs1079597) have been extensively studied.

2.2 DRD3

The dopamine D3 receptor is preferentially expressed in limbic and basal ganglia regions associated with cognitive, emotional, and motor functions. The D3 receptor inhibits spontaneous secretion of the neurotransmitter, hence may play an important role in the regulation of neurotransmission. Because many antipsychotic drugs exhibit a high affinity for the D3 receptor, it is reasonable to suspect that DRD3 genetic variants may affect the clinical efficacy of antipsychotic drugs. DRD3 gene is located at chromosome 3q13.3 and contains five exons, and has a missense polymorphism in the exon 1 leading to a serine to glycine substitution at amino acid position 9 (Ser9Gly, rs6280) in the N-terminal extracellular domainof the receptor protein. Previous research suggests this SNP is associated with altered dopamine binding affinity and the Gly9 variant may increase DRD3 densities in some brain areas²⁴.

At least 18 studies have examined the association between the Ser9Gly SNP and antipsychotic drug response, with earlier studies focusing on clozapine and recent studies using risperidone. An early meta-analysis²⁵ found intriguing results; the Ser allele was associated with better response to FGAs, but it was associated with non-response to clozapine treatment. In a recent meta-analysis specifically targeting clozapine response, Hwang et al²⁶ showed a non-significant trend that the Ser allele and Ser/Ser genotype were more frequent in non-responders than in responders in 8 cohorts with 758 patients. Examining the DRD3 studies in Table 1 reveals that most recent studies of SGAs including risperidone and aripiprazole did not produce significant associations with the Ser9Gly SNP. Although it makes theoretical sense that DRD3 may affect antipsychotic drug response, empirical research does not yield consistent data on the hypothesis.

2.3 DRD4

Due to clozapine's superior antipsychotic efficacy and the fact that it potently binds to the dopamine D4 receptor, it has been hypothesized that that the D4 receptor genotype plays a role in mediating clozapine and other SGA's effects. The DRD4 gene codes for the D4 receptor protein and is located at chromosome 11p15.5. The coding DNA sequence of the DRD4 is highly polymorphic, resulting in functionally different receptor variants. A well-studied 48-bp variable number tandem repeat (48-bp VNTR) in the third exon, with 2–10 repeats, results in a different length of the third cytoplasmatic loop. The 48-bp VNTR may be functionally important because this region of the D4 receptor is involved in G-protein coupling, and the longer repeat alleles are associated with reduced clozapine binding. Moreover, the potency of dopamine to inhibit cAMP formation was decreased by twofold in the D4,7 variant (i.e., repeating 7 times), when compared with both the D4.4 and the D4.2. Despite several negative studies of clozapine response in early 1990s, Hwu et al²⁷ found that longer repeat alleles of the 48-bp VNTR were associated with higher response rate using a variety of antipsychotic drugs in 80 Chinese patients. However, this was not replicated in two later studies, one Chinese sample²⁸ and one Caucasian sample²⁹. The inconsistency is yet to be reconciled, but it is likely contributed by different study design, different treatment duration, and various medications used.

2.4 HTR2A

Serotonin system has long been suspected to play a major role in mediating antipsychotic drug action. All SGAs tightly bind to serotonin receptor 2A relative to dopamine D2 receptor, and this was once thought to be one of the defining characteristics of “atypicality” of SGAs. As such, genetic variations in different serotonin receptors have been extensively studied to examine their potential associations with drug response. HTR2A, HTR2C, HTR6, and 5HTT are reviewed here.

HTR2A is the gene coding for the 5-HT 2A receptor and is located at chromosome 13q14-q21. The 2A receptor is widely distributed in the cortex, and may be associated with negative symptoms of schizophrenia. Neuroimaging studies have suggested that high occupancy of 5-HT2 receptors by SGAs is associated with improvement in negative symptoms and cognition. HTR2A is highly polymorphic. One polymorphism that has been investigated in many studies is a synonymous SNP at codon 102 (T102C, rs6313). Although this SNP does not result in an amino acid change, it is in nearly complete linkage disequilibrium (LD) with another functional SNP (A-1438G, rs6311) in the promoter region in Caucasian populations³⁰. A recent study suggests the C allele of the T102C SNP and the G allele of the A-1438G SNP may cause lower promoter activities and thus decreased 2A receptor densities in some brain areas, which may lead to a less flexible serotonin system and lower dopaminergic modulation³¹. Conversely, specific methylation of the C allele of the T102C SNP could increase HTR2A expression in human temporal cortex³². These findings suggest that A-1438G/T102C polymorphisms may influence HTR2A densities in the brain.

At least 15 studies have been published on the association between the T102C SNP and antipsychotic drug response. Early studies focused on clozapine and later studies also examined other SGAs including risperidone, olanzapine, and aripiprazole. A meta-analysis summarized the first six studies of clozapine response in 733 patients revealed that the C allele of T102C was more prevalent among non-responders³³. However, after excluding the first published study³⁴, the pooled odds ratio became non-significant. After the meta-analysis was published in 1998, two more studies^{35, 36} examined clozapine response, and neither of them found significant association with the T102C SNP. More recently, four studies focused on risperidone response, two of which found a significant association between the C/C genotype and better response, especially in negative symptoms^{37, 38}. This is the converse finding of earlier studies of clozapine. It should be noted, however, that earlier clozapine studies were mostly in Caucasians with some African American patients, but more recent studies were mostly from Asian countries. Another recent study of Chinese patients added further complexity by showing that the C/C genotype responded better to aripiprazole treatment with a larger improvement in negative symptoms, compared to other genotypes³⁹.

Compared to the conflicting findings of the T102C SNP, the A-1438G SNP was more consistently found to be associated with antipsychotic drug response. Three studies^{30, 40, 41} showed that the G/G genotype was less likely to respond to clozapine, olanzapine, and aripiprazole, especially in negative symptoms, than other genotypes. However, a recent study of 100 Algerian patients treated with haloperidol reported that the G allele was actually associated with better treatment response⁴². Haloperidol has only minimal effects at the 5-HT 2A receptor, so it is not clear how a genetic variant in HTR2A would mediate clinical response to haloperidol. Although in complete LD with the T102C SNP, the A-1438G SNP has produced more positive findings in fewer studies. Many studies reported one but the other SNP, hence the concordance of the findings between the two SNPs is not clear.

A third SNP in the HTR2A gene, His452Tyr, was also examined in several studies. This nonsynonymous SNP is located in exon 3 of the gene and the change from C to T results in a change from histidine to tyrosine at the 452th amino acid. The Tyr variant is associated with reduced calcium release and reduced ability to activate phospholipases. In vitro data indicated that the Tyr variant showed lowered antipsychotic binding affinity and decreased drug potency. Three clinical studies^{30, 35, 43} found that the Tyr allele was significantly associated with poor response to clozapine treatment, compared to the His allele. Although other studies failed to replicate the findings^{44, 45}, an early meta-analysis of five samples showed a clear association of the Tyr/Tyr genotype with poor response to clozapine (OR = 5.55, p = .04)³³. It should be noted, however, that only 10 out of 676 patients included in the meta-analysis were Tyr/Tyr homozygotes. Intriguingly, all of the studies of this SNP were published before 2002, and publication bias towards significant findings may play an important role.

2.5 HTR2C

Most SGAs except quetiapine tightly bind to serotonin 2C receptors, in addition to the 2A receptors⁵. The 5-HT2C receptors are widely distributed in many areas of the human brain including striatum, prefrontal cortex, and the limbic system, indicating a role in executive functioning, memory, emotional processing, feeding behavior and motor functions. Like the 2A receptors, these postsynaptic receptors are excitatory and positively couple with G-protein. The HTR2C gene is located at chromosome Xq24. Several SNPs in the gene have been linked to antipsychotic drug response, but the most studied is the Cys23Ser SNP (rs6318). This non-synonymous SNP is in the coding region with a change from G to C resulting in an amino acid substitution of Cysteine with Serine. Despite a change in receptor protein structure, there has been no evidence of alteration in receptor function. Only the first clinical study⁴⁶ found that patients who carry the Ser allele were more likely to respond to clozapine treatment compared to patients who are Cys/Cys homozygotes, but five later studies failed to confirm the finding.

2.6 5HT6 and 5HTT

Genetic variants of other components of the serotonin system have also been linked to antipsychotic drug efficacy, notably the 5HT6 and 5HTT genes, which code for the serotonin 6 receptor and serotonin transporter. In animal studies, 5-HT6 receptors are associated with the endogenous 5-HT-mediated facilitation of dopamine release, and specific 5-HT6 receptor antagonists produced a favorable outcome for reducing positive symptoms of schizophrenia. HTR6 is located at chromosome 1p36. One SNP, C267T (rs1805054), was examined in four pharmacogenetic studies of clozapine and risperidone. Two studies^{47, 48} found a significant association between the T/T genotype and better treatment response, both of which happen to be in Han Chinese patients. Two other studies in Caucasians and Japanese did not yield significant results^{20, 49}. It is not clear whether this is a specific biomarker for antipsychotic response in the Chinese population, and it is not clear what the functional consequences of this variant are, but it is clear that the SNP deserves more research.

5HTT is the gene coding for serotonin transporter (SLC6A4; solute carrier family 6 member 4). It is located at chromosome 17q11.2. The serotonin transporter is an integral membrane protein that transports serotonin from synaptic spaces into presynaptic neurons, thus terminates the action of serotonin and recycles it in a sodium-dependent manner. A repeat length polymorphism, 5-HTTLPR, involves insertion/deletion of a 44-bp segment located upstream of the transcription start site in the promoter region. It has been shown to affect the rate of serotonin uptake and is the most studied genetic variant in psychiatry⁵⁰. Recent meta-analyses of the relationship between the polymorphism and antidepressant response have shown that patients carrying the long allele are about twice as likely to respond to treatment at 4 weeks and reach remission, and less likely to suffer from side effects, than patients with the short/short genotype⁵¹. Although only a few studies have focused on this polymorphism and response to antipsychotic drugs, three studies showed that the short allele is associated with poor response to clozapine and risperidone treatment^52-54. Future research should attempt to replicate this finding with other antipsychotic drugs.

2.7 COMT

The catechol-O-methyltransferase (COMT) enzyme is one of the main pathways of dopamine clearance and metabolically terminates dopamine activity, especially in the frontal cortex. It may moderate antipsychotic drug action because all antipsychotics exert their effects on the dopamine system. The COMT gene is located at chromosome 22q11.21. A common polymorphism, Val108Met, can causes substantial variations in enzymatic activity. This is due to a G to A transition at codon 158 of the membrane-bound form of COMT, which corresponds to codon 108 of the soluble form of COMT, resulting in a valine to methionine substitution. The met/met genotype results in 3- to 4-fold lower enzyme activity compared with the val/val allele pair, while the met/val heterozygote results in intermediate enzyme activity. In other words, the val allele results in reduced dopamine in synapse due to more rapid degradation. An early case-control study⁵⁵ found that patients with the met/met genotype were less likely to respond to treatment with various FGAs. However, a later study of 59 Caucasian first episode schizophrenic patients⁵⁶ showed the opposite, that is, patients with the val/val genotype were less likely to respond to 8 weeks of olanzapine treatment, especially in negative symptoms, compared to other patients. This may be due to the fact that schizophrenia is characterized by dopamine hypoactivity in prefrontal cortex and reduced metabolism of dopamine associated with the met allele helps to restore dopamine level. Consistently, another study⁵⁷ of clozapine indicated that the met allele carriers were more likely to respond, especially improvement in cognitive functions. This is consistent with the data indicating that the met/met genotype has been associated with higher IQ scores in a meta-analysis⁵⁸.

2.8 CYP2D6

The cytochrome P450 enzyme family in the liver is responsible for the metabolism of many psychotropic drugs. Among its subtypes, 2D6 is the main metabolic pathway for several antipsychotics including risperidone, aripiprazole, haloperidole, perphenazine, and chlorpromazine, and a secondary pathway for clozapine, olanzapine, and quetiapine. CYP2D6 is also most relevant to pharmacogenetics because it has more than 100 genetic variants (as catalogued by the website: http://www.cypalleles.ki.se, as of July 20, 2010) and many of them yield non-functional or low-functional enzymes. The CYP2D6 gene is located at chromosome 22q13.1. The polymorphisms in this gene involve various single nucleotide substitutions and insertion/deletion of certain DNA segments. In Caucasians, four polymorphisms (*3, *4, *5, and *6) are responsible for most inactive alleles (98%). There are four phenotypes of CYP2D6 produced by combinations of various alleles with different degrees of enzymatic activities: poor metabolizer (PM), intermediate Metabolizer (IM), extensive metabolizer (EM), and ultrarapid metabolizer (UM). EMs have normal CYP2D6 enzyme activity, whereas PMs and IMs have no or reduced activity, respectively. Ums have duplicate or multiple copies of the gene which result in increased enzyme activity. Approximately 7-10% of Caucasians and 1-2% of Asians are PMs⁸, who tend to accumulate higher drug levels in blood, and theoretically, require lower doses to achieve therapeutic effects. UMs, in contrast, who are rare and consist of only 1% of the population, may require higher doses of an antipsychotic due to faster elimination of the drug. Therefore, CYP2D6 metabolic status could play an important role in determining antipsychotic efficacy for a particular patient.

However, there are few empirical data to support the above hypothesis. None of the six studies investigating the association between CYP2D6 genotypes and antipsychotic response have reported significant findings. Two risperidone studies^{59, 60} demonstrated that PMs had higher ratio of blood levels of risperidone to 9-hydroxyrisperidone than other patients, but neither genotypes nor blood levels predicted clinical improvement. Other studies have found significant relationships between PMs and higher rate of drug-induced side effects, which are reviewed in later sections of this article. Although theoretically appealing and clinically meaningful, studies with positive findings are needed to prove the utility of CYP2D6 genotyping in predicting antipsychotic drug response.

In summary, pharmacogenetic research of antipsychotic response has examined a number of genetic variants, from both pharmacodynamic and pharmacokinetic perspectives. Past studies showed promising results for a few polymorphisms including the −141C Ins/Del in DRD2, Ser9Gly in DRD3, −1438G/A in HTR2A, 5-HTTLPR, and Val108Met in COMT. Studies with larger samples and better designs are needed to validate these findings.

3.1 Tardive Dyskinesia

Tardive dyskinesia (TD) is a chronic involuntary body movement caused by exposure to neuroleptics. It is often irreversible and debilitating. A recent review of 12 clinical trials reported the both FGAs and SGAs can cause TD with a one-year risk of 5.5% and 3.9%, and a prevalence of 32.4% and 13.1%, respectively⁶¹. Several demographic and clinical factors are known to increase the risk of TD including older age, female gender, African American descents, higher antipsychotic dosage, and early EPS⁶². Smoking and alcohol abuse may also increase the risk of TD. The etiology of TD is unknown, but the nigrostriatal dopaminergic tract, which is closely involved in the regulation of motor behavior, may play a key role. Dopamine antagonism of antipsychotic drugs results in up-regulation of D2 receptors post-synaptically, which contributes to nigrostriatal dopaminergic hyperactivity. Genetic factors including variants in dopamine and other neurotransmitters genes may therefore play an important role.

TD is the most studied antipsychotic-induced ADR in pharmacogenetics. Most studies are case-control design in nature which affords larger sample sizes than studies of drug efficacy. Multiple genes of various neurotransmitter systems and many polymorphisms have been examined in association with TD. Among these, Taq1A, −141C Ins/Del, and Ser311Cys in DRD2, Ser9Gly in DRD3, T102C and −1438G/A in HTR2A, Cys23Ser in HTR2C, Val108Met in COMT, and CYP2D6 have accumulated sufficient data to warrant discussion below. It is not surprising that many of these genes are the ones that have been studied in association with antipsychotic drug efficacy because they are primary drug targets or metabolic pathways.

Although our recent meta-analysis did not reveal a significant relationship between the DRD2 Taq1A SNP and antipsychotic drug response¹⁷, it has been associated with TD in two meta-analyses. Despite the fact that only 2 out of 8 studies listed in Table 2 reported significant findings that the A2 allele and the A2/A2 genotype display increased risk of TD, a cumulative sample of 1,256 patients (507 with TD and 749 without TD) from 6 cohorts demonstrated an odds ratio of 1.30 for the risk of TD in the A2 allele⁶³. This means that each copy of the A2 allele confers a 30% more risk of developing TD, relative to the A1 allele. Compared to A1/A1 homozygote or A1/A2 heterozygote, patients with the A2/A2 genotype have a 50% increased risk of TD (odds ratio = 1.50). Another meta-analysis⁶⁴ of 764 patients (297 with TD and 467 without TD) from 4 studies, which represent a sub-sample of the first meta-analysis, confirmed the previous findings. One mechanistic explanation is that the A1 allele is associated with reduced density of D2 receptors in the striatum, which results in less dopamine antagonism by antipsychotic drugs. Therefore, the A1 allele is protective of TD development. However, a recent report of 710 patients from the CATIE trial (207 with TD and 503 without TD), which was not included in either meta-analyses, did not find any association between the Taq1A SNP and TD, casting some doubts on previous findings. Other SNPs in DRD2, including −141C Ins/Del and Ser311Cys, have not been found to affect TD development^{63, 64}, despite their promising roles in predicting clinical response to antipsychotic treatment.

The dopamine D3 receptor was initially implicated in TD development because D3 blockade in the basal ganglia produced hyperactivity in animal models⁶⁵. Antipsychotic drugs that have minimal D3 affinity, such as clozapine and quetiapine, tend to have lower liability of causing TD. One of the SNPs in DRD3, Ser9 Gly, has been examined for association with TD in least 20 studies. Interestingly, the Gly allele, previously associated with clinical response to antipsychotics drugs, is also associated with higher risk of TD in at least 8 studies, and this was confirmed in two early meta-analyses with overlapping samples^{66, 67}. However, the most recent meta-analysis, with 2,026 patients (928 with TD and 1098 without TD) from 13 cohorts, found only a non-significant trend that the Gly allele carriers may confer a slightly higher risk of TD compared to non-carriers⁶⁸. The odds ratio of 1.16 is modest and there was significant evidence of publication bias and sample heterogeneity. Data from the CATIE trial also did not support any link between the Ser9Gly SNP and antipsychotic-induced TD⁶⁹.

The facts that SGAs are less likely to cause TD and that all SGAs bind to serotonin receptors make it plausible that the serotonin system may play an important role in preventing antipsychotic-induced TD⁵. The same SNPs in HTR2A and HTR2C that have been examined for their association with antipsychotic drug efficacy are also related to TD in a number of studies. As reviewed earlier, the C allele of the T102C SNP in HTR2A was associated with poor response to antipsychotic drug treatment in several studies³³. Interestingly, the C allele was also associated with risk of TD in at least three studies^70-72. A meta-analysis⁷³ summarized 6 cohorts with 635 patients (256 with TD and 379 without TD) and found that the C allele carriers have a 64% higher risk of developing TD than the non-carriers (T/T homozygotes) (odds ratio = 1.64, p = 0.004), especially in older patients and in patients with limb-truncal TD. However, three later studies in three different populations (Indians, African-Caribbeans, and Americans of mixed ethnicities) were not able to replicate this finding^{69, 74, 75}. Another SNP, −1438G/A, that is in complete LD with T102C, was also found to be significantly associated with TD in a couple of studies^{70, 76}. The G allele was associated with reduced expression of 5-HT2A receptor. Thus, it is plausible that reduced availability of the receptor in certain brain regions such as basal ganglia may be a risk factor for developing TD. Nevertheless, more data are needed to support this hypothesis.

In addition to dopamine and serotonin receptors that are major targets of antipsychotic drugs, genetic variation in metabolic pathways of the drugs have also been studied in association with TD. Theoretically, if a drug is not cleared fast enough, prolonged stimulation of dopamine receptors may put patients at risk of developing TD. Therefore, the Val108Met SNP of COMT is a candidate polymorphism of interest because the Met allele results in lower enzyme activity and slower clearance of synaptic dopamine. A meta-analysis⁶⁴ of 5 studies with 1,089 patients (382 with TD and 707 without TD) found that the Met allele was actually protective and the Met allele carriers were less likely to develop TD, with an odds ratio of 0.66. In other words, patients with the Val/Val genotype had a 51% higher risk of TD than others.

CYP2D6 is the other metabolic pathway candidate gene that has been studied extensively. Poor metabolizers of CYP2D6 may have higher blood levels of antipsychotic drugs such as risperidone and haloperidol, which put these patients at risk of developing TD. Interestingly, not only is it found in the liver, CYP2D6 is also expressed in some brain areas, particularly in those areas rich in the dopamine transporter, and may play an important role in protecting certain susceptible brain regions from toxicants that gain access to the central nervous system⁷⁷. Several studies have demonstrated that patients who are poor or intermediate metabolizers of 2D6 are more likely to have TD^78-80, although there are a few negative studies as well, as shown in Table 2. A meta-analysis⁸¹ of 8 studies with 569 patients (220 with TD and 349 without TD) showed an odds ratio of 1.43 for the PMs compared to EMs. In other words, PMs have 43% higher risk of developing TD, compared to EMs. One issue with this literature is that the PMs account for only 7-10% of Caucasian population and even rarer in some other ethnic groups. Hence, it is difficult to find a large sample of PM patients, of whom TD can be ascertained. Potentially, sample sizes in a range of thousands are needed to robustly test the hypothesis regarding the association between CYP2D6 and TD.

CYP1A2 is another member of the cytochrome P450 enzyme family, and metabolizes many antipsychotics drugs including both FGAs and SGAs such as chlorpromazine, fluphenazine, perphenazine, clozapine, and olanzapine. The CYP1A2 gene is located on chromosome 15q24.1 and contains a number of non-functional variants, several of which have been associated with the development of TD. CYP1A2 enzyme can be induced by smoking and two SNPs, *1F (−163C>A) and *1C (−3860G>A), affect the inducibility of the enzyme^{82, 83}. An early study found that chronically treated patients with the *1F C/C genotype were more likely to have TD symptoms than those *1F A allele carriers, especially among smokers⁸⁴. However, a meta-analysis of seven studies did not find significant association between the *1F SNP and TD frequency⁶⁴. It should be noted that many studies did not report what antipsychotic drugs patients were taking, and it appears that some patients on SGAs were included⁸⁵. The two SGAs metabolized by CYP1A2, clozapine and olanzapine, have low TD liability. Other SGAs that have relatively high TD liability, such as risperidone, were not metabolized through CYP1A2. As such, inclusion of patients on SGAs might have reduced signal-to-noise ratio.

In addition to the association between TD and genetic variants on the pharmacodynamic and pharmacokinetic pathways of antipsychotic drugs, antipsychotic-inducing oxidative stress and free radicals may cause neuronal injury and contribute to the development of TD. Several studies have examined whether TD is associated with variants of MnSOD, the gene encoding manganese superoxide dismutase, a mitochondrial enzyme involved in oxidative metabolism. One SNP, Ala9Val, results in a substitution of alanine with valine and less efficient MnSOD transporter in mitochondria. A meta-analysis of four studies suggested that the Val carriers are less likely to develop TD compared to the Ala/Ala homozygotes⁶⁴, although the result of a more recent meta-analysis with ten samples was not significant⁸⁶. Other oxidative enzymes have also been studied, but findings were also mixed⁸⁶.

Several studies have focused on EPS instead of TD in association with CYP2D6. Six out of nine studies found that CYP2D6 PMs and IMs were more likely to experience antipsychotic drug-induced EPS^{80, 87-91}. In general, pharmacogenetic research of EPS has been less extensive than that of TD, although EPS are much more commonly encountered in clinical practice.

3.2 Weight Gain and Metabolic Syndrome

Weight gain is the most prominent side effect associated with the SGAs, especially clozapine, olanzapine, and quetiapine⁹². Although weight gain is commonly associated with SGAs, patients on FGAs can also gain large amount of weight. In the European Union First Episode Schizophrenia Trial (EUFEST), 53% of patients on haloperidol gained more than 7% of baseline body weight at 1-year follow-up⁷. Due to large inter-individual variation in weight gain, no clear clinical predictors have been identified, and the mechanism remains poorly understood⁹². Food intake and body weight are regulated by complex interactions between multiple neurotransmitter systems in multiple brain regions. Several studies have examined the possible role genetic variation in the dopamine system may play in drug-induced weight gain^93-95, but pharmacogenetic research has primarily focused on the serotonin system, especially the 5-HT2C receptor.

The 5-HT2C receptor is involved in the regulation of food intake in rodents. HTR2C knockout mice display chronic hyperphagia leading to obesity and hyperinsulinemia. In humans, C-759T (rs3813929), a SNP in the promoter region of the HTR2C gene has been related to late-onset diabetes and obesity in a normal population. At least 17 studies have reported on the association between the C-759T SNP in HTR2C and antipsychotic drug-induced weight gain. 10 out of 17 studies listed in Table 3 reported significant findings that the C allele was associated with more weight gain than was the T allele after antipsychotic drug treatment, especially clozapine and olanzapine, both of which have high affinity to 5-HT2C. A meta-analysis⁹⁶ of 8 studies with 588 patients found that the T allele was significantly protective against antipsychotic drug-induced weight gain. The C allele was associated with more than two fold increase of risk for clinically significant weight gain, i.e., gaining 7-10% or more of baseline body weight. The C allele is the common allele and the T allele is the rare allele with a frequency ranging from 3.3% in African descent to 33.3% in Asians. Of the 9 studies published after the meta-analysis, 5 have also reported positive findings. Although one recent study did not find a significant association between this SNP and iloperidone-induced weight gain⁹⁷, iloperidone binds to 5-HT2C only minimally. Overall, the evidence so far suggests that the C-759T SNP in HTR2C may play an important role in antipsychotic drug-induced weight gain.

Another gene that has attracted much attention in pharmacogenetics of drug-induced weight gain is GNB3, which codes for G-protein β3 subunit. G-proteins are ubiquitous in many intracellular signaling pathways, and relay signals from receptors to effector proteins. The C825T polymorphism of the GNB3 gene is associated with a GB3 splice variant that results in a deletion of 41 amino acids, although it does not seem to alter protein function. This SNP has been linked to hypertension and obesity, two of the major components of metabolic syndrome, in the general population. A meta-analysis of 3 published and 2 unpublished studies with 402 patients found that the T allele was marginally associated with increased weight gain, but the finding was not statistically significant due to considerable heterogeneity among studies and relatively small sample size⁹⁸. Two recent studies, both from East Asia and with olanzapine, did not provide consistent findings^{99, 100}.

In summary, a number of antipsychotic drug-induced side effects have been examined in relation to genetic variants. Many genes and polymorphisms were studied, but so far very few have gained consistent support. For TD, the Taq1A in DRD2, the Ser9Gly in DRD3, the T102C SNP in HTR2A, and the loss of functional variants in CYP2D6 may warrant further research. For weight gain, the only promising variant that has accumulated substantial data is the C759T SNP in HTR2C.