Genetics of antipsychotic drug outcome and implications for the clinician: into the limelight

Abstract

Background and purpose

Antipsychotics (APs) are the primary method of treatment for schizophrenia and other psychotic disorders. Unfortunately, lengthy trial-and-error approaches are typically required to find the optimal medication and dosage due to a large interindividual variability with outcome to AP treatment. The literature has shown abundant evidence for a genetic component in individuals’ responses to APs. Pharmacogenetic studies analyze specific genetic markers and their association with symptom improvement and occurrence of side effects with APs. This research aims to optimize AP drug treatment by usage of predictive testing and to personalize medicine.

Recent findings

This review will highlight the most consistent findings in pharmacogenetics of APs and will update the reader on the clinical implications. This will include how genetic variants modulate AP drug levels, side effects, and therapeutic symptom improvement (i.e. response) to AP treatment.

Summary

Several promising findings were obtained implicating gene variants of the dopamine receptor genes in addition to gene variants of serotonin receptors for response and common side effects. Notably, effect sizes appear to be particularly high in the genetics of side effects compared to response. One example is antipsychotic-induced weight gain where the leptin, HTR2C and in particular the melanocortin-4-receptor (MC4R) genes have been implicated in weight gain in children and adolescents. Consistent findings were also obtained for genes implicated in tardive dyskinesia and agranulocytosis. However, the most clinically relevant findings pertain to genes involved in drug metabolism such as the CYP2D6 and CYP2C19 genes which have been included in the first genetic test kits such as the Amplichip^® CYP450 Test and more recently the DMET™ Plus Panel, the Genecept™ Assay, the Genomas HILOmet PhyzioType™ System, and the GeneSight^® Test.

• pharmacogenetics
• antipsychotics
• response
• side effects
• CYP450

View correction statement:

Corrigendum: Genetics of antipsychotic drug outcome and implications for the clinician: into the limelight

A Corrigendum has been published for this paper. Please see http://www.translationaldevelopmentalpsychiatry.net/index.php/tdp/article/view/25715

Due to a large interindividual variability with response to psychotropic drugs, a lengthy trial-and-error approach is typically required to find the right medication and dosage. This variability results from various factors such as age, gender, ethnicity, disease symptoms, drug–drug interactions, and most importantly genetics.

Friedrich Vogel coined the term pharmacogenetics in 1959, referring to interactions between the response to drugs and genetics in an individual (1–(4)). In particular, pharmacogenetic research aims to identify specific genetic markers and their association with treatment outcome to drugs (5, 6). Pharmacogenetics bears the promise to become a tool in psychiatry, used to tailor treatment to an individual's genetic make-up.

Several areas of research have now embraced pharmacogenetics as a way to improve treatment with medication. There are many tests emerging in all fields of medicine such as cardiology, pain management, cancer treatment, and psychiatry (7, 8). Many efforts are currently being made to integrate pharmacogenetic testing into the clinical setting to improve patient care as well as work toward personalized medicine.

Pharmacology usually differentiates between pharmacokinetics and pharmacodynamics. Pharmacokinetics is the study of how drugs are metabolized (including degradation and excretion) in various body compartments. Pharmacokinetic effects include genetic variations that lead to differences in the activity of drug-metabolizing enzymes. This can influence the availability and plasma concentrations of antipsychotic drugs in the brain, and affect the synthesis of active metabolites (9). In contrast, pharmacodynamics is the course of action a drug exerts on various organs. This includes how gene variants may affect the interindividual variability seen in outcome to medications (9).

Antipsychotics (APs) are the most used class of medications for schizophrenia and related psychotic disorders. However, complete therapeutic benefit may not always be achieved in patients. AP use can frequently result in adverse reactions such as weight gain, leading to metabolic syndrome. Such side effects increase discontinuation and relapse rates (1, 10). Response to APs depends partially on demographic and clinical factors (1). However, the literature has shown substantial evidence for a genetic component in individuals’ outcomes to APs which include variation in drug metabolism and side effects (11).

A large number of genes are involved in AP treatment outcome, and gene variants (or polymorphisms) can be studied for their potential impact on this outcome. Most genes will typically carry a variable amount of polymorphisms that can be investigated and analyzed for their impact on drug treatment outcome. Such polymorphisms may result in a lack of gene expression, altered levels and functions of gene expression, which also have an impact on drug metabolism in the organism (1).

This review will highlight the most important findings in pharmacogenetics of APs and will also update the reader on the clinical implications. This will include examples of how genetic variants modulate AP drug levels, side effects, and therapeutic symptom improvement (i.e. response) to AP treatment.

Methods

We reviewed published review articles and meta-analyses on pharmacogenetic studies on APs published using Medline database (http://www.ncbi.nlm.nih.gov/pubmed/) and OVID (psycINFO) from 2008 to 2013 with an emphasis on studies published between 2012 and 2013. Selected key words were ‘pharmacogenetics’, ‘pharmacogenetic testing’, ‘response’, ‘side effects’, ‘pharmacokinetics’, ‘pharmacodynamics’, ‘APs’, ‘CYP alleles’, ‘CYP450’, ‘weight gain’, ‘tardive dyskinesia (TD)’, ‘agranulocytosis’, ‘sexual dysfunction’. In addition, we included highly relevant original articles recently published. We excluded letters to the editor and editorials as well as conference abstracts. In total, we reviewed 41 studies. Our review article emphasizes findings that have been consistently replicated in independent samples, or were observed using larger and well-characterized samples (e.g. CATIE sample), or were found through GWAS or any combinations of the above.

Results

Pharmacokinetics

The liver cytochrome P450 (CYP450) enzymes are involved in the metabolism of various psychotropic drugs (12) and are involved in metabolic phase I oxidative reactions (13). The enzymes which are most critically involved in the metabolism of psychotropic medications are CYP2D6 and CYP2C19 and to a lesser extent CYP1A2, CYP2C9, CYP3A4, and CYP3A5 (13). Due to their role in the metabolism of psychotropic medications, for which there are interindividual differences, the CYP450 enzymes are extensively studied.

As for CYP2D6 and CYP2C19, individuals are considered to be normal or extensive metabolizers (EM) if they have two functional alleles. Those who carry one normal and one non-functional allele are categorized as intermediate metabolizers (IM); those with two non-functional alleles are known as poor metabolizers (PM). Finally, individuals with duplicate copies of alleles are categorized as ultra-rapid metabolizers (UM) (11).

As for AP drugs metabolized by CYP2D6 and CYP2C19 enzymes, individuals with the PM phenotype may suffer from dose-dependent complications due to increased, and potentially toxic, plasma concentration (1). Those with IM phenotypes are also likely to have increased plasma levels compared to the EMs. The UM phenotypes can result in sub-therapeutic drug levels when standard doses of medications are given, requiring increased dosage in affected individuals (1).

Knowledge of an individual's metabolizer status of these genes may have potential benefits in order to optimize drug treatment by reducing the likelihood of side effects or therapeutic failure. By knowing the metabolizer status, dose adjustments can be made or alternative medications can be selected (11).

Of note, frequencies for PM, IM, EM, and UM vary considerably between ethnic groups. For example, frequencies for CYP2D6 UMs are much higher in Africans (11, 14) (see Table 1).

Table 1 Variability in the distribution of genes (allele frequencies) in different ethnicities^† (15, 16)

	Africans	Asians	European	South American Indians
CYP2D6*1 (EM)	55% (Ethiopian) 52% (Sub-Saharan African)	48% (Japanese)	37% (Northern European)	73%
CYP2D6*3 (PM)	<0.1% (Ethiopians) <0.1% (Sub-Saharan African)	<0.1% (Japanese)	2% (Northern European)	<0.1%
CYP2D6*4 (PM)	3% (Ethiopian) 2% (Sub-Saharan African)	<0.1%	19% (Northern European)	3%
CYP2D6*10 (IM)	6% (African Americans; Ethiopian) 4% (Sub-Saharan African)	39–51%	1–2%	1%
CYP2D6*17 (Altered affinity for substrates of 2D6)	20–35% (African Americans)	<0.1% (Japanese)	<0.1% (Northern European)	<0.1%
CYP2D6*M X N ^†† (Depending on the allele, it represents an increased function) (17)	1.0–2.4% (African Americans) 0.9–16.0% (Black Africans)	0.9–1.3% (East Asians)	0.7–5.6% (Caucasians)	Not determined (Native Americans)

For determination of metabolizer status using *allele combinations, the reader is referred to comprehensive review articles such as that by Mrazek (15).

^†The precise frequencies can differ in small communities due to their geographic locations (15, 18).

^††XN refers to the number of gene copies; for example, CYP2D*1X2 refers to two copies of CYP2D6*1 (17).

Furthermore, both enzymes can be inhibited or induced by drugs and diet and create phenocopies (11). An example of a phenocopy would be the conversion of a CYP2D6 EM to a CYP2D6 PM after exposure to a potent CYP2D6 inhibitor such as paroxetine (19).

AP drugs such as aripiprazole, haloperidol, perphenazine, risperidone, thioridazine, and zuclopenthixol are metabolized to a various degree by CYP2D6. Other most commonly prescribed APs such as olanzapine and clozapine are metabolized by CYP1A2 (20). For additional information on AP and antidepressant medications metabolized by the CYP450 enzymes, please refer to the Pharmacogenomics Knowledgebase (https://www.pharmgkb.org/) and the CYP450 Drug Interaction Table by the Indiana University Department of Clinical Pharmacology (http://medicine.iupui.edu/clinpharm/ddis/main-table/). More details on this topic can also be found in two reviews published by Ravyn et al. (1) and Altar et al. (5).

CYP2D6

This is a highly polymorphic enzyme, with more than 100 described allelic variants (3, 15). Variation occurs through single nucleotide polymorphisms (SNPs); a single nucleotide variation in the DNA sequence varies between person to person; for example, one individual can have the ‘A’ nucleotide while another person can have the ‘G’ nucleotide on the same locus (21), as well as from copy number variations (CNVs; i.e. duplications of the gene) (22). These variations can lead to loss-of-function, decreased activity, or increased enzymatic activity through gene duplications (3).

Genotyping and defining metabolizer status for CYP2D6 can be challenging because of the vast amounts of allelic variations and their interactions. A common limitation is that rare and population-specific variations are typically not tested which could impact the results of published articles. Another limitation with respect to gene duplication is that they are not usually determined quantitatively but rather qualitatively (3).

There are a few examples which show how pharmacogenetic testing could have prevented critical situations: A 9-year-old boy, diagnosed with encephalopathy secondary to fetal alcohol syndrome, attention-deficit hyperactivity disorder, and Tourette's disorder, died due to accidental toxic plasma levels of fluoxetine which also likely resulted from his previously unknown CYP2D6 PM status. Adopted when he was 5 years old, he was diagnosed with physical aggressiveness, negativity, and impulsivity. He also had a history of absence seizures. Clonidine 0.6 mg/day (in divided doses) was initially started to treat his tics, and the dose was increased over a year to 0.9 mg/day. At age 6, he was initially started on treatment with fluoxetine at 5 mg/day, and the dose was gradually increased to a high dose of 100 mg by the time he was 8 years old. In addition, he was taking clonidine 0.9 mg/day and methylphenidate 60 mg/day concurrently. This particular constellation of type and dose of drugs and his PM status had likely exacerbated the fluoxetine toxicity which was revealed at the autopsy where toxic fluoxetine levels were noted. Fluoxetine and norfluoxetine blood plasma levels were 21 µg/mL, substantially higher concentrations than what is normally seen in adults who are taking fluoxetine (with levels typically below 5 µg/mL) (23).

An infant's death was due to opioid toxicity from morphine formation because his mother was taking the prodrug codeine. CYP2D6 is responsible for the conversion of codeine to morphine. The mother's UM status caused an increase in morphine formation leading to the neonate's death (24).

CYP2C19

Although CYP2C19 is not primarily involved in the metabolism of APs, it is an important enzyme involved in the metabolism of antidepressants and the antiplatelet drug, clopidogrel (16, 25) (26). Given that it is included in various commercially available genetic test kits, its genetic architecture is briefly discussed. CYP2C19 is a fairly polymorphic enzyme, with more than 30 known allelic variants (www.cypalleles.ki.se/cyp2c19.htm). When a drugs’ primary metabolism pathway is non-functional (i.e. CYP2D6 PM), CYP2C19 plays a secondary role in its metabolism (15). Similar to CYP2D6, phenotypes of CYP2C19 can be categorized into the UM, IM, EM, and PM criteria (25). UMs have been found to be non-responders to standard doses of medication and PMs endorse more side effects. IMs can be given a lower dose of the CYP2C19 substrate medication (15).

CYP2C19*2 is the dominant defective allele (27). The distribution of common variant alleles of CYP2C19 has been found to vary among different ethnic groups. The PM phenotype is more prevalent in the Asian population and less in Caucasian and African populations (please see below) (17). The PM phenotype of CYP2C19 is inherited as an autosomal recessive trait (28, 29) (Table 2).

Table 2 Distribution of common variant alleles of CYP2C19 (allele frequencies) (21, 30–32))

	Africans	Asian	Europeans (Caucasians)	Native American
CYP2C19*2 (PM)	17%	18–23% (Japanese)	2–15%	12.0–19.1
CYP2C19*3 (PM)	0.4%	5%	0.04%	Not determined
Total PM status:	2–5%	~25%	2–5%	3.3–11.65%
CYP2C19*17 (UM)	18% (Ethiopians)	1.3% (Japanese) 0.64% (Chinese)	18% (Swedes)	Not determined

CYP1A2

The CYP1A2 allele has 41 described variants (www.cypalleles.ki.se/cyp1a2.htm) and this enzyme can be induced by various other drugs and substances. This can include smoking (the most common CYP1A2 inducer), insulin, modafinil, nafcillin, omeprazole, cruciferous vegetables (15), and oral contraceptives (32). The polycyclic aromatic hydrocarbons found in tobacco smoke induces CYP1A2 activity (33–(36)), which results in decreased plasma concentration of the drugs metabolized by CYP1A2 (34). Individuals who quit smoking need to be informed that their CYP1A2 activity decreases which then increases the risk for potential side effects (34, 37).

Two well-studied CYP1A2 alleles are CYP1A2*1A and CYP1A2*1F. The *1A allele is not easily induced whereas the *1F allele is. With the absence of an inducer, both the alleles are similar. When an inducer is present, the *1F activity (and *1C activity in people of European ancestry) increases. Individuals who begin to smoke and take a CYP1A2 substrate medication while having the *1F/*1F genotype are at an increased risk for a loss of efficacy of that medication (15). There is a marked ethnic difference in this CYP enzyme activity; Swedes had a 1.54-fold higher CYP1A2 activity than Koreans (27, 38). When compared to the Caucasian population, both Asian and African populations have lower CYP1A2 activities (39). Fourteen percent of Japanese people and 5% of Chinese people were found to be PMs (17, 27).

Few studies have investigated the role of CYP1A2 and response. A study by Laika et al. found that CYP1A2*1F/*1F genotype was associated with plasma levels for olanzapine (40); however, an association with therapeutic response has not yet been established (1). Smoking status was considered in the study; out of a total of 124 participants, 44 were smokers. For the olanzapine serum cohort, there were 73 participants out of whom 30 were smokers (40).

CYP2C9

CYP2C9 is also fairly polymorphic with more than 50 allelic variants (www.cypalleles.ki.se/cyp2c9.htm). CYP2C9 plays an important role in drug clearance if the primary metabolic pathway is non-functional. It also has a role to code an enzyme allowing for the oxidation of approximately 100 medications (15). It is estimated that for drugs requiring phase I metabolism, CYP2C9 is responsible for approximately 15–20% of metabolic clearance (41–(43)). CYP2C9 only plays a minor role for APs while it is involved in the metabolism of various antidepressants and S-warfarin. Genetic variants which cause changes in metabolic activity can lead to pathogenesis due to adverse drug reactions. This can be seen with patients who have low enzyme activity for CYP2C9 substrates (41, 44). Reduced CYP2C9 activity is caused by the CYP2C9*2 and CYP2C9*3 polymorphisms (45), These allelic variations are not common in East Asian, African-American, and Ethiopian populations (45, 46); CYP2C9*2 is absent in the East Asian population and is present in 3.2% of the African American and Ethiopian populations. CYP2C9*2 is present in 8–19% of the Caucasian population (46). The presence of CYP2C9*3 in East Asian populations (Koreans, Japanese, and Chinese) range from 1.1 to 3.3%. In Africans, it is present in 1.3% (for both African Americans and Ethiopians combined). In contrast, CYP2C9*3 is present between 3.3 and 16.2% in the Caucasian population (46). Carrying two non-functional CYP2C9 alleles can result in a risk for serious side effects such as life threatening bleeding episodes with warfarin and toxicity with phenytoin (47). For those who are heterozygous for the CYP2C9*2 or CYP2C9*3 allele (i.e. *1/*2 or *1/*3), a 25% dose reduction is suggested for the administration of phenytoin. Those who are homozygous for the two alleles (i.e. *2/*2 or *3/*3) require a 50% dose reduction for phenytoin (45, 48) Likewise, a 50% reduction is also suggested for those who have the *2/*3 allelic variation (48).

CYP3A4 and CYP3A5

CYP3A4 comprises more than 30% of hepatic enzymes and for marketed drugs which rely on metabolic elimination, CYP3A4 is involved in the metabolism for over half of them (49–(53)). In people with reduced CYP1A2 activity, CYP3A4 is involved in clozapine metabolism while CYP3A5 does not influence clozapine pharmacokinetics (54). Based on in-vitro affinity constants, it has been suggested that the role of CYP3A4 is relevant with higher doses of clozapine (54, 55).

Structurally, CYP3A5 is similar to CYP3A4 but the substrate selectivity of these proteins differ (51, 56–58)). CYP3A5*1 is present in approximately 10–20% of White-European individuals (50) and is present in approximately 55% of African American individuals (51, 56) (59). Apart from the liver, CYP3A5 is also distributed in the intestine and kidneys (50, 60–62)). There are over 26 allelic variants (www.cypalleles.ki.se/cyp3a5.htm) for CYP3A5. Protein expression and activity are affected by genetic polymorphisms of CYP3A5 which may also affect drug response (51, 56–58)). Those who are homozygous or heterozygous for the CYP3A5*1 allele express CYP3A5 to normal levels while having the CYP3A5*3, *6, and *7 alleles result in diminished expression CYP3A5 (www.cypalleles.ki.se/cyp3a5.htm.) (51, 56–58, 63).

Linkage studies conducted in a Japanese population demonstrated that three SNPs in CYP3A5 showed some linkage with a SNP in CYP3A4 suggesting that CYP3A4 and CYP3A5 are within the same gene block (64).

CYP3A4 plays a major role in the metabolism of a number of drugs, chemicals, and compounds. This is mainly due to the relatively high abundance of CYP3A4 and the minor impact of genetic variants for this enzyme (65).

Notably, for CYP3A43, a member of the same family as CYP3A4, Bigos et al. found that marker rs472660 was associated with olanzapine clearance in African Americans and where clearance was associated with therapeutic response (66).

Pharmacodynamics

Response.

There are some variants that have been shown repeatedly to be associated with overall response to treatment.

One of the most intensely studied genes is the DRD2 receptor gene due to its known role in the pathway of many AP drugs, and in particular, for positive symptoms (9). In addition, the dopamine D1, D3, D4, and D5 receptor polymorphisms were also investigated. Interesting findings were more recently reported for the DRD1 rs4532 polymorphism, which showed an increased risk for treatment resistance (67). As for the DRD3 gene, a possible role was suggested for the Ser allele of the Ser9Gly polymorphism (rs6280) to be associated with poorer response to clozapine in a meta-analysis by Hwang et al. (68). The DRD4 variants have been found to have a minor contribution to clozapine response using 18-item BPRS scale, while DRD5 showed none (69). A study by Miura et al., investigated the Taq1A polymorphism (rs1800497), in linkage disequilibrium and related to the DRD2 receptor, in relation to response to aripiprazole treatment, using the PANSS and CGI scales. They found a non-significant trend that carriers of the A1 allele responded better to aripiprazole compared to non-carriers. They also found a significant effect for response and genotype/response interaction on the change of plasma levels of homovanillic acid (pHVA) but not for the change of plasma levels of 3-methoxy-4hydroxyphenylglycol (pMHPG). This indicates a possible, partial role for the Taq1A polymorphism in response to aripiprazole (70). Two other previous studies in 2008 both also had positive findings in relation to Taq1A and clinical response to aripiprazole using the PANSS scale (71). In 2008, Ikeda et al. found that two SNPs in DRD2 (−241A>G [rs1799978] and Taq1A [rs1800497]) and two SNPs in AKT1 (AKT1-SNP1 [rs3803300] and AKT1-SNP5 [rs2494732]) were significantly associated with response to risperidone, using the PANSS scale (72).

The HTR2A gene is a target for atypical APs, making it another candidate for pharmacogenetic testing (9). Earlier studies have reported some interesting results implicating serotonergic gene variants to AP response. HTR2A gene polymorphisms have been shown to be associated with hallucinations, overall treatment response (73, 74) and in particular with negative symptom response (75). Earlier studies also suggested the HTR2C receptor was involved in response to AP treatment, given its role in regulating the effect of the drug on cognitive functioning (76). Consistent with this notion, a highly studied polymorphism of HTR2C, Cys23Ser (rs6318), has been found to be associated with clozapine response, although most studies were negative (77). In 2009, Wei et al. investigated the HTR7 gene and response using the BPRS scale, but found no significant association (78). Also in 2009, Lavedan et al. reported results of a GWAS study showing an association between iloperidone response (using the PANSS scale) and the nudix (nucleoside diphosphate linked moiety X)-type motif 9 pseudogene (NUDT9P1), which is located approximately 200kb away from the HTR7 gene. The NUDT9P1 gene could act as a remote regulatory unit (79) on the HTR7 gene, where the encoding receptor has shown high affinity for iloperidone (80). Overall, dopamine and serotonin receptors are strongly involved in AP mechanism of action and there is moderate evidence that gene variants of these receptors play a major role in response to APs.

More recent studies, including genome-wide approaches, have suggested other genes to be possibly involved in therapeutic response to APs. The zinc finger gene ZNF804A potentially plays a role in neurodevelopment, similar to other zinc-finger domain–containing proteins, and has been suggested to be involved in susceptibility to schizophrenia and psychotic bipolar disorder, particularly the functional marker rs1344706 (81–(83)). Furthermore, replication of this finding has been successful (81, 84–87)). Zhang et al. reported that Chinese T allele carriers showed significantly less symptom improvement, using the PANSS scale, while being treated with various SGAs (82). In contrast, Mössner et al. reported that those individuals of European descent who were homozygous for the A allele showed significantly less symptom improvement in positive symptoms, based on the PANSS scale (81).

Also of interest, the brain derived neurotrophic factor gene (BDNF) has been reported to be associated with treatment resistance in schizophrenic patients. Zai et al. found that the rs11030104 marker and the Val66Met marker (rs6265) were associated with AP response; furthermore, the rs11030104 and Val66Met C-A haplotypes were underrepresented in responders (assessed with the BPRS scale) compared to non-responders (88). Furthermore, Zhang et al. found the rs11030104 marker and the rs6265 marker along with the rs10501087 marker to be significantly associated with treatment resistance (89). Xu et al. reported that two haplotypes (230-bp/C-270/rs6265G and 234-bp/C-270/rs6265A) were significantly associated with response to risperidone based on BPRS scores. Patients with the 230-bp allele of the (GT)_n polymorphism or the 230-bp/C-270/rs6265G haplotype responded better to risperidone than those with other alleles or haplotypes (90).

The ATP-Binding Cassette (ABCB1) gene has been investigated in relation to response. ABCB1 functions as an efflux transporter for some APs in the blood–brain barrier thus decreasing concentrations of APs in the brain tissue (91). In their association study, Lee et al. found that the markers rs7787082 and rs10248420 of ABCB1 were significantly associated with response (using the CGI scale) in a sample of Korean patients treated with clozapine. Markers rs7787082 G and rs10248420 A alleles in ABCB1 were more frequently observed in non-responders (92). A naturalistic study by Vijayan et al. found that individuals with 3435CC (rs1045642) and 1236CC (rs1128503) genotypes responded better to a variety of APs, using the BPRS scale (93). Notably, earlier monotherapy studies in other populations found that individuals with the 3435TT genotype showed better response to olanzapine (94), and those with the 1236TT polymorphism showed better response to risperidone (95). On the contrary, Crisafulli et al. did not observe an association between ABCB1 and response in a Korean population treated with a variety of APs using the PANSS scale (96).

Souza et al. investigated several variants of the neurexin 1 (NRXN1) gene with clozapine response. They found a non-significant trend with the rs12467557 marker and association with clozapine response, using the BPRS scale (97). In 2011, Lett et al. reported a significant association between the rs1045881 marker of NRXN1 and BPRS negative symptoms score (98).

Spellmann et al. found a significant association between a number of variants of the Homer homolog 1 gene (HOMER-1) and response to APs. The dendritic HOMER-1 protein regulates group 1 metabotropic glutamate receptor function and is therefore an excellent candidate for pharmacogenetic studies. There were several variants with a strong association, particularly with the rs2290639 marker showing significant associations between both PANSS baseline scores as well as at the 4 weeks of treatment mark (99).

Side effects.

There are potentially serious adverse reactions to treatment with APs. Typical side effects include transient sedation, extrapyramidal symptoms, weight gain, sexual side effects, and in rare cases agranulocytosis depending on the type of AP drug (100).

Based on their side effect profile, APs were historically classified into typical (or first generation) and atypical (or second generation) APs. While all APs have antagonistic properties at the level of the dopamine D2 receptor, typical APs are much more likely to induce motoric side effects such as TD.

TD is a movement disorder which typically develops after months or years of AP treatment which often persists even after discontinuation of APs (101). Given the prominent role of APs on dopaminergic receptors and the role of the dopaminergic system in movement control, dopamine receptor genes have been extensively investigated (102). Most promising findings include the Taq1A polymorphism (rs6280) related to DRD2 and the Ser9Gly polymorphism (rs1800497) of DRD3 (103). The Taq1A polymorphism (which was associated with a lower density of D2 receptors in the brain in vitro) (104, 105) is likely associated with TD as reported by three different studies (106). The Gly variant of the Ser9Gly polymorphism has been shown several times to be linked to increased risk for TD (107). However, there have also been negative findings in this area and thus more investigation is required.

There are numerous findings in regards to CYP2D6 metabolizer status and TD where the majority suggested that increased or decreased CYP2D6 activity are associated with the presence of TD (108–(111)), although some negative findings, most likely due to heterogeneous study designs, were also reported (112).

Meta-analyses have found three other genes implicated in TD: manganese superoxide dismutase (MnSOD), cytochrome P450 1A2 (CYP1A2), and serotonin 2A receptor (HTR2A) (113). It has also been proposed that BDNF levels contribute to TD and that certain polymorphisms of BDNF may increase response to treatment for TD using ginkgo biloba, a potent antioxidant with neuroprotective effects mediated through enhancing BDNF levels (114). BDNF is a neuronal growth and survival peptide that modulates a variety of processes including regulation of the dopamine D3 receptor (DRD3) and could thus be involved in TD through this mechanism (75). The heparan sulfate proteoglycan 2 (HSPG2) gene emerged as the top hit in a GWAS of TD screening by Syu et al. (115) and was later replicated in a prospective sample by Greenbaum et al. (116) but not in a study by Bakker et al. (113).

Second-generation AP medications often induce a large amount of weight gain (117). This increased weight gain often leads to metabolic syndrome, which predisposes the patient to further conditions, such as diabetes and cardiovascular disease.

Since atypical APs frequently act as antagonists at the HTR2A receptor, this gene has been extensively investigated for response and side effects (118). Ujike et al. reported four gene variants to be associated with olanzapine-induced weight gain, two of which were the 102T allele (rs6313) of the HTR2A receptor, and the 23Cys allele (rs6318) of the HTR2C receptor (119). However, there have also been negative findings in this area. Mou et al., in 2005, reported that the −1438G/A polymorphism (rs6311) of the HTR2A gene was not associated with antipsychotic-induced weight gain (AIWG) in Chinese Han patients (120).

The HTR2C receptor is one of the most highly studied in relation to AIWG stemming from a study by Tecott et al. in 1995 that depicted HTR2C knockout mice overeating and developing obesity (121, 122). Furthermore, a study by Bonhaus et al. showed that HTR2C antagonists increase food intake (123). In their review, Altar et al. reported 22 studies showing significant pharmacodynamic association between HTR2C and weight gain (5). Recently, promising findings surrounding obesity-related genes have come in to play, such as the leptin and melanocortin receptor 4 (MC4R) genes. The protein products of these genes are influenced by APs, making them potential key factors in AIWG (124). A GWAS has recently identified marker rs489693, approx. 200 kb downstream of the MC4R gene to be associated with AIWG (125). Given that this finding was replicated across various samples in different populations using different APs, a major role for rs489693 is likely in AIWG (125–(127)). The findings by Nurmi et al. also indicated that the same gene variant associated with AIWG in adults is implicated in AIWG in children and adolescents as well (127). A study by Zai et al. found that patients with a Val66Met G-A haplotype (rs6265) of the BDNF gene (frequency 57%) were associated with more pronounced weight gain compared with patients taking clozapine or olanzapine (88). Given the role of dopamine in motivational aspects related to rewarding effects such as food intake, the dopamine receptors DRD1-DRD5 were investigated in one study suggesting an association among a DRD2 promoter region variant, −141C Ins/Del (rs1799732), and AIWG (128). Another study on the DRD1-DRD5 receptors by Müller et al. found three DRD2 SNPs (rs6277 (C957T), rs1079598, and rs1800497 (Taq1A)) to be associated with weight gain after stratifying their sample by ethnicity and high-risk APs (129). Another gene that has recently been suggested to be linked to AP-induced metabolic side effects is the methylenetetrahydrofolate reductase (MTHFR) gene. Deficiency of MTHFR will affect the central nervous system and cardiovascular system and in severe cases can lead to homocystinuria (130). Three studies focused on the association between APs, MTHFR polymorphisms, and metabolic disturbances. All of these studies reported on two variants: the A1298C variants (rs1801131) and the C677T variant (rs1801133). Two of them reported that the −1298C variant and the C677/1298C haplotype were associated with increased weight and metabolic syndrome, whereas the third reported that the −677T, not the −1298C, variant was associated with AP-induced metabolic syndrome (130–(133)).

Clozapine is particularly used in treatment-resistant schizophrenia. Unfortunately, clozapine has the potential to cause agranulocytosis (potentially fatal drop of white blood cells) in up to 1% of patients. For this reason, recurrent white blood cell counts are mandatory for clozapine which limits its use (134). A large-scale, retrospective study in 2012 by Lahdelma et al. analyzed 163 Finnish patients with clozapine-induced agranulocytosis and reported that the highest risk is during the first year of treatment (135). There are a couple of theories surrounding clozapine-induced agranulocytosis; the first being that clozapine activates electrophilic nitrenium ions, which is the first step in agranulocytosis, and the second theory suggests an immune-mediated mechanism (136). A few gene variants have been clearly implicated in clozapine-induced agranulocytosis with the most promising being variants of the HLA-DQB1 region of the major histocompatibility complex (136). The 6672G>C marker of HLA-DQB1 was recently implicated in agranulocytosis and can help identify a small subset of individuals with extremely high-risk for agranulocytosis (137).

Another common and distressing side effect of treatment with many AP drugs is hyperprolactinemia potentially contributing to sexual dysfunction. Hyperprolactinemia causes a variety of reversible symptoms such as menstrual dysfunction, lactation, gynecomastia, infertility and loss of libido. In addition, APs may cause other sexual side effects such as erectile dysfunction, delayed orgasm and impaired lubrication (138–(140)). Sexual side effects are most prevalent in those APs that have the most pronounced D2 antagonism, mainly typical APs and risperidone (141). One study suggests that erectile dysfunction and other sexual dysfunctions in men, as a side effect of AP treatment, is linked to a polymorphism of the dopamine D2 receptor (−141C Ins/Del [rs179973]) (142). A case study by Mesa et al. suggested the addition of a D2 agonist could eliminate the sexual side effects caused by AP use (143). Overall, genetic causes of AP-induced sexual side effects have not been studied extensively and there is a paucity of literature on this subject.

All of the aforementioned side effects may affect patient compliance, self-esteem, and quality of life. Note that for research purposes, side effects can be advantageous as some side effects can be measured with high reliability (e.g. weight gain or WBC counts) when compared to response which is often assessed using different scales. In addition, identifying possible genetic candidates may be easier in cases where AP effects are known to interfere with other pathways. For example, and as mentioned above, the HTR2C receptor is known to be involved in appetite regulation and also antagonized by most APs which cause AIWG.

Available pharmacogenetic tests.

A number of pharmacogenetic tests are available either directly to the consumer or through their healthcare providers. Since proof of clinical and economic benefits of using genetic tests through large randomized controlled trials on a global level are still pending, insurance providers are disinclined to provide coverage, causing a financial barrier to the dissemination of the tests (144).

The following are commercially available pharmacogenetic tests relating to AP treatment:

Amplichip^® CYP450 test

The Amplichip^® CYP450 Test from Roche Molecular Diagnostics was the first FDA-approved pharmacogenetic test. It is intended as an aid to treatment with APs and personalizing AP choice. The test identifies a patient's CYP2D6 and CYP2C19 genotype, which is an important factor in the patient's metabolizing rate for different drugs that are substrates for the CYP enzymes, and can provide information on which medications to prescribe (145, 146). In a study by Dunbar et al., sponsored by Roche Molecular Diagnostics, the authors report that upon surveying 33 clinicians who were using the Amplichip^® test there were some logistic difficulties with the test (i.e. ordering and receiving results) however, the clinicians widely reported advantages to the test, such as support of their dosing decisions (147). The reported disadvantages varied from over-reliance on the test to the actual implementation of a new clinical procedure, breaking away from routine practice. Their findings indicate that overall, psychiatric clinicians are open to pharmacogenetic testing in clinical practice if its implementation is done with economic consideration (147). On the patient side, Swen et al. performed a study using the Amplichip^® to investigate the feasibility of pharmacy-initiated pharmacogenetic testing of CYP metabolizer status by inviting 93 patients to participate in the testing. They found that approximately 60% of patients consented to the testing; a high percentage considering the screening was not directly related to their clinical treatment. Their study provided evidence for pharmacy-initiated pharmacogenetic screening to be feasible in a primary care setting (148). A retrospective pilot study by Müller et al. examined 35 patients with schizophrenia genotyped for CYP2D6 and CYP2C19 metabolizer status using the Amplichip^® test. They reported no significant association of CYP2D6 metabolizer status with treatment response. However, case reports of PM and UM metabolizers showed generally poor response or high rates of side effects to APs suggesting a potential clinical benefit for preemptive testing (12).

DMET™ plus panel

The DMET™ Plus Panel from Affymetrix, Inc. is a large-scale test that covers a number of genetic variants including frequent and uncommon SNPs (145). It includes 1,936 SNPs, copy number, and index markers across 231 genes and 100% coverage of ‘Core ADME (Absorption, Distribution, Metabolism and Excretion) Genes’ (total 32 genes) and 95% coverage of ‘Core ADME Markers’ (185 variants) selected by PharmaADME; a multidisciplinary group of representatives from an academic center and the pharmaceutical industry (149, 150). The test also allows for the translation of genotypes into metabolizer status groups to help indicate which drugs will be the best ‘fit’ for the patient (149).

PGxPredict:Clozapine^® test

The PGxPredict:Clozapine^® test from PGxHealth, Division of Clinical Data, Inc. was developed in 2007 and based on two variants of the HLA-DQB1 gene, aiming to predict if an individual has a lower versus higher risk of developing agranulocytosis (151). However, this test was discontinued due to lack of general interest at that time since individuals still would require regular blood tests.

GeneSight^® test

The GeneSight^® test from Assurex Health Inc.^® analyzes four cytochrome genes (CYP2D6, CYP2C19, CYP2C9 and CYP1A2) as well as the serotonin transporter gene (SLC6A4) and serotonin 2A receptor (HTR2A). The genotype results are then converted into a ‘bin’ system; each medication is placed into either the red, yellow, or green bin based on level of recommended caution when prescribing a given drug (152). Winner et al. of Assurex Health Inc.^® performed a 1-year blinded and retrospective study evaluating the GeneSight test^® and eight direct or indirect health care utilization measures of 96 patients with a diagnosis of depression or anxiety. Their results demonstrated that patients carrying gene variants with higher risk for non-response or side effects had significantly higher health care visits, more general medical visits, more medical absence days, and more disability claims (152). Hall-Flavin et al. also of Assurex Health Inc.^® conducted a study to evaluate the potential benefit of using the GeneSight Test^® for the management of psychotropic medication used to treat major depression and compared one group where genetic testing was considered versus one ‘treatment-as-usual’ group. Their findings demonstrated that use of the test would lead to improved outcomes with antidepressant medication. The guided group experienced greater improvement in three depression scores from baseline on the HAMD, QIDS, and PHQ, when compared to the unguided group (153).

Genecept™ assay

The Genecept™ Assay from Genomind, L.L.C. investigates 10 genes including three cytochromes (CYP2D6, CYP2C19, and CYP3A4) as well as the serotonin 2C receptor (HTR2C) and the serotonin transporter gene (SLC6A4). These genes were chosen because they have each been associated with multiple psychiatric disorders, an increased likelihood of side effects to AP treatment and/or poor response to treatment, or the metabolism of APs. The test's purpose is to aid clinicians in finding a fast and suitable treatment for the patient (154).

HILOmet PhyzioType™ system test

The Genomas, Inc. Laboratory of Personalized Health (LPH) offers various tests such as the High-Low Metabolic Risk for Neuro-Psychiatric and Cardio-Metabolic Drugs (HILOmet) PhyzioType™ System test. This test aims to guide drug choices for patients with treatment resistance or drug intolerance and investigates a total of 37 variants in the genes coding for CYP2D6 (20 alleles), CYP2C9 (7 alleles), and CYP2C19 (10 alleles), assessing metabolizer status in a patient (155, 156). Further information on additional pharmacogenetic tests can be found at http://www.genomas.com/.

Resources

There are several resources clinicians may use to gather information on the strength of evidence related to pharmacogenetic-based dosing and treatment decisions and specific dosing guidelines. These include resources from:

1. The Clinical Pharmacogenetics Implementation Consortium (CPIC) of the Pharmacogenomics Research Network, established in 2009 and offer peer-reviewed guidelines along with providing supplemental information and data (http://www.pharmgkb.org/page/cpic).

2. The Pharmacogenomic Knowledgebase (PharmGKB) provides regular updates on pharmacogenomics (http://www.pharmgkb.org/index.jsp).

3. The Pharmacogenetic Research Network – funded by the National Institutes of Health; and involving distinct research initiatives (http://www.nigms.nih.gov/Research/SpecificAreas/PGRN/Pages/default.aspx).

4. The Royal Dutch Association for the Advancement of Pharmacy established the Pharmacogenetics Working Group to provide dosing guidelines and recommendations (48).

5. The Evaluation of Genomic Applications in Practice and Prevention (EGAPP) initiative, established in 2004 by the Centers for Disease Control and Prevention (CDC), aims to establish and evaluate a process to assess evidence-based applications of genetic tests (http://www.egappreviews.org/about.htm).

Discussion

Pharmacogenetic testing has made progress in recent years to build toward personalized medicine and to reduce the trial-and-error process of AP prescription. There are many studies suggesting benefits of pharmacogenetic testing in psychiatry and the most consistent findings relate to the CYP450 enzymes. In this context, our Pharmacogenetics Research Clinic at CAMH in Toronto is conducting a study where testing of CYP2D6 and CYP2C19 was found to be well accepted by physicians and patients and considered useful to guide future treatment decisions (11). We have now expanded this study and have included several other genes (for more details, see www.im-pact.ca and (157)).

Given that genotyping costs are constantly dropping, more attention is given to the clinical applications of the CYP genetic test results (11). Initially, genotyping for two drug-metabolizing genes often exceeded US $1,000 whereas now a much larger amount of informative gene panels can be genotyped for an equivalent cost. Also, more public and private insurance companies have consistently been willing to cover the cost of pharmacogenetic testing for psychiatric patients, in the case of appropriate indications (15).

With respect to pharmacogenetic studies reviewed above, several limitations should be kept in mind such as relatively small sample sizes, low statistical power, retrospective designs, heterogeneity in study design, and inconsistencies in genotyping or phenotyping predictions (1). There is also a paucity of psychiatric pharmacogenetic research on patient populations such as elderly, incapable patients and children (see also section above on CYP2D6). Given their limited capability to report symptoms and side effects, tools such as genetic markers may even be of greater benefit in these populations. Another general limitation is that large randomized control trials (RCT) have not been conducted. RCTs could evaluate whether improved response to treatment, including the reduced likelihood of adverse effects, is associated with genetic testing (including CYP genes) (1). Although RCTs are costly and labor intensive, such studies are needed to evaluate in more depth the clinical and health-cost benefits of routine CYP testing (1, 11) (158).

General ethical concerns regarding pharmacogenetic testing were raised in the studies by Haga et al. who assessed attitudes toward pharmacogenetic testing in the general public and in various focus groups (healthcare professionals, primary care professionals and geneticists) (159, 160). In the focus groups, they found that the overall attitude was generally positive, but a few concerns were raised in both groups. These included clinical utility, insurance coverage, treatment delay, and the communication of ancillary disease information as principal concerns (161). Those studies concluded that additional educational resources, access to genetic counseling, improved clinical guidelines and cost-benefit analyses are warranted prior to large-scale implementation (159, 160).

The FDA provides a table of approved drugs containing pharmacogenomic information in their labeling (http://www.fda.gov/drugs/scienceresearch/researchareas/pharmacogenetics/ucm083378.htm; last updated January 22, 2014). Some labeling sections include which actions should be taken depending on the biomarker information provided. The table also provides which CYP enzyme metabolizes the drug. Included in the list are APs such as aripiprazole, clozapine, iloperidone, perphenazine, pimozide, and risperidone. Other drugs such as atomoxetine, clomipramine, fluvoxamine, fluoxetine, paroxetine, thioridazine, and valproic acid used for children are also included.

Altogether, outcomes of most medications in psychiatry have been linked to specific gene variants and as suggested by the FDA labeling, information of drug-metabolizing genes is currently recommended to be considered if metabolizer status is known. This clearly suggests that pharmacogenetics is becoming an increasingly important source of information for use in clinical practice in addition to other gene–drug pairs discussed in this review article that are likely to become implemented in the near future.

Future studies should include larger samples, populations with a wider age range (i.e. children and geriatric populations); address potential gender effects; and consider variations due to ethnicity. In addition, further studies comparing pharmacogenetic-guided treatment versus treatment as usual (TAU) are required, such as the study on antidepressants published by Hall-Flavin et al. (153).

Once these obstacles are overtaken, pharmacogenetic testing will be a great asset to clinical practice. As it was insightfully pointed out elsewhere, pharmacogenetic testing was quickly adopted by other fields of medicine once its importance became clear (15). We are confident that clinical implementation in the field of psychiatry is impending.

In conclusion, there is substantial evidence for the benefit of pharmacogenetic testing in clinical psychiatry. Promising first steps have been taken and will likely lead to a wider acceptance and implementation in clinical practice.

Conflict of interest and funding

Brain & Behaviour Research Foundation (NARSAD Independent Investigator) Award to DJM, CIHR Michael Smith New Investigator Salary Prize for Research in Schizophrenia to DJM, and an Early Researcher Award by the Ministry of Research and Innovation of Ontario to DJM.

Notes

A Corrigendum has been published for this paper. Please see http://www.translationaldevelopmentalpsychiatry.net/index.php/tdp/article/view/25715