Autism spectrum disorder (ASD) is a neurodevelopmental disorder typically diagnosed in early childhood, and is characterized by deficits in social and linguistic communication skills, along with repetitive and stereotyped behaviors. In addition, autism patients show symptoms like irritability, aggressive attitude toward other or property, anxiety, self-injurious behavior, hyperactivity, and severe temper tantrums (Ghosh et al., 2013; Bowers et al., 2015). Risperidone, an atypical antipsychotic, is approved by the Food and Drug Administration (FDA) to treat disruptive behaviors associated with autism in children and adolescents(Cohen et al., 2013) demonstrating antagonism at dopamine D₂, serotonin 5HT_2A, and adrenergic α₂ receptors (Thyssen et al., 2010).

Plasma concentrations of risperidone and 9-hydroxyrisperidone vary among individuals, which in turn influences clinical responses and adverse-effects justifying the regular need to monitor plasma concentrations of risperidone, 9-hydroxyrisperidone, and the total active-moiety (Riedel et al., 2005; Locatelli et al., 2010). Genetic heterogeneity is commonly regarded as the main cause of this inter-individual variability. Pharmacogenomics explores the genetic causes of variability in drug responses, and aids in tailored therapy according to the individual’s genetic make-up (Collins et al., 2016).

Risperidone undergoes 9-hydroxylation producing 9-hydroxyrisperidone as its active metabolite. CYP2D6 is the key enzyme for the hydroxylation of risperidone, but CYP3A4 and CYP3A5 enzymes have also been reported to play minor roles in the metabolism of risperidone (Fang et al., 1999). 9-Hydroxyrisperidone and risperidone, which have similar pharmacological activities, are collectively termed as the total active-moiety. Nevertheless, studies have shown that risperidone and 9-hydroxyrisperidone differ from each other in terms of regulation of receptor-mediated signaling pathways (Clarke et al., 2013) and 5-HT_2A/D₂ affinity ratios with differences in their therapeutic efficacies or adverse effect profiles (Kozielska et al., 2012). Risperidone and 9-hydroxyrisperidone are substrates of P-glycoprotein, an adenosine triphosphate–binding cassette (ABC) superfamily transporter, and their dispositions and absorptions might be affected by the functional status of P-glycoprotein (Nakagami et al., 2005; Corena-McLeod, 2015).

Studies have reported the inter-individual variability in pharmacokinetic and pharmacodynamics of risperidone leading to unusual exposures, an occurrence of toxic effects, and therapeutic failure. Presence of single nucleotide polymorphisms (SNPs) in the genes encoding phase I and phase II drug-metabolizing enzymes (DME) alters the activity of enzymes resulting in four possible phenotypes: poor metabolizer (PM), intermediate metabolizer (IM), extensive metabolizer (EM), and ultrarapid metabolizer (UM; Zhou et al., 2009). Drug transporters are classified as influx solute carrier (SLC) family transporters and efflux ABC family transporters (Yiannakopoulou, 2013). A study analyzing the associations between drug metabolizing enzyme and transporter (DMET) polymorphisms and pharmacokinetic variability of risperidone found associations of polymorphisms in CYP2D6, TPMT, ADRB1, VKORC1, and COMT with pharmacokinetic parameters (Cabaleiro et al., 2014). ABCB1 3435C>T genotypes had significant effect on maximum concentration observed (C_max) of risperidone and the total active-moiety, but when ABCB1 3435C>T genotypes were combined with CYP2D6^∗10/^∗10 genotype carrying subjects, significant differences were observed in C_max, the area under the curve (AUC) of risperidone, and the total active-moiety as well (Yoo et al., 2011). Several polymorphisms have been studied emphasizing the influence of allelic variations among metabolic enzymes and transporters in the disposition of risperidone and its phenotypes. An association study of risperidone and 9-hydroxyrisperidone plasma levels with CYP2D6 polymorphisms in ASD patients found a significant influence of CYP2D6 variants on plasma risperidone concentrations which provides evidence of genetic variations of drug exposure and safety (Vanwong et al., 2016a). Previous studies also reported the influence of polymorphisms in DMET genes, including CYP2D6, CYP3A4, CYP3A5, ABCB1, COMT, and VKORC1 on the plasma concentrations of risperidone, 9-hydroxyrisperidone, and the total active-moiety (Jovanovic et al., 2010; Xiang et al., 2010; Suzuki et al., 2012; Cabaleiro et al., 2014). The present authors previously reported the prevalence of clinically important DMET SNPs in Thai ASD patients (Medhasi et al., 2016).

The influence of CYP2D6, CYP3A5, and ABCB1 gene variants on plasma levels of risperidone, 9-hydroxyrisperidone, and the total active-moiety clearly argues for the need of pharmacogenetic testing and therapeutic drug monitoring (TDM) during risperidone therapy (Jovanovic et al., 2010). The association between genetic variants of DMET genes and steady-state plasma concentrations of risperidone, 9-hydroxyrisperidone, total active-moiety, and metabolic ratio remains unclear. The objective of the present study was to present the results of the association between genetic variants of the DMET genes and steady-state plasma concentrations of risperidone and its metabolite using the Affymetrix DMET Plus genotyping microarrays.

In this study, the associations of DMET genetic markers with steady-state plasma concentrations of risperidone and 9-hydroxyrisperidone among Thai ASD risperidone-treated patients using Affymetrix DMET Plus microarray platform were evaluated. The major findings from this exploratory pharmacogenomic investigation were the abilities of the genetic markers involved in drug metabolism and transport to predict inter-individual variability in steady-state plasma concentrations of risperidone and its active metabolite, 9-hydroxyrisperidone. It showed that six polymorphisms (4 ABCB11 variants and 2 ADH7 variants) were strongly associated with plasma concentrations of risperidone; six SLC polymorphisms (3 SLCO1B3 variants, 2 SLCO1B1 variants, and 1 SLC7A5 variant) were strongly associated with plasma concentrations of 9-hydroxyrisperidone and the total active-moiety; and four polymorphisms (3 CYP2D6 variants and 1 UGT2B4 variant) were strongly associated with metabolic ratio.

With the increased recognition of the role of transporters in drug disposition with pharmacologically important allelic variability, both ABC and SLC transporters have been the scope of analysis in the present study. These results showed that rs497692, rs496550, rs495714, and rs473351 in ABCB11 were associated with plasma concentrations of risperidone. ABCB11 is a member of the ABC transporter group, subfamily B, member 11, and is also known as sPgp (sister of P-glycoprotein)¹. Although ABCB11 has been studied extensively in other conditions, the report herein found new SNPs of ABCB11 that are associated with plasma concentrations of risperidone. In the study, genes and SNPs that do not have direct interactions with risperidone and its metabolite have been discovered and showed associations with plasma concentrations of risperidone and its metabolite. SNPs related to ADH7 (rs971074 and rs1442477) showed an association with plasma concentrations of risperidone. ADH7 is a class IV alcohol dehydrogenase which metabolizes ethanol to acetaldehyde (Jairam and Edenberg, 2014). It can then be concluded that there is a possible role for multiple genetic variants which previously lacked evidence of pharmacogenomic associations and known biological pathways and that now lead to phenotypic outcomes similar to the established ones. This explains the advantage of microarray technology where an a priori assumption is avoided and an established physiological network among several genes in the body is recognized (Nigam, 2015; Arbitrio et al., 2016). The mechanistic functional effects of previously unrelated genes, however, should be further investigated in an independent sample set to reveal the linking molecular pathways.

Other top SNPs associated with plasma concentrations of 9-hydroxyrisperidone and the total active-moiety were SLC transporter variants. A previous study reported the influence of SLC6A4 in sexual dysfunction and weight gain among risperidone treated patients (Almoguera et al., 2013). The present study, however, found SNPs among SLCO1B1, SLCO1B3, and SLC7A5 were associated with plasma concentrations of 9-hydroxyrisperidone and the total active-moiety. CYP2D6 polymorphisms influenced the plasma risperidone to 9-hydroxyrisperidone ratio in this study which suggests the functional role of CYP2D6 on a genetic basis. Among the Phase II enzymes, SNP rs1131878 in UGT2B4 was shown to affect the risperidone metabolic ratio. UDP-glucuronyltransferase (UGT) enzymes catalyze the glucuronidation reactions and are an important aspect for variability in pharmacokinetics and clinical responses of antipsychotic agents (Kiang et al., 2005).

This pharmacogenomic study identifies newer variants of DMET gene polymorphisms with greater susceptibility to alter plasma concentrations of risperidone, 9-hydroxyrisperidone, total active-moiety, and metabolic ratio. Notably, this study did not find any significant associations among the variants previously reported to influence risperidone and 9-hydroxyrisperidone plasma concentrations. Polymorphisms among the main metabolic polymorphic enzyme CYP2D6 and the major transporter protein ABCB1 did not correlate with the plasma concentrations of risperidone or 9-hydroxyrisperidone as has been reported in earlier studies among the schizophrenic patients (Roh et al., 2001; Jovanovic et al., 2010). This finding implicates the limited importance of CYP2D6 and ABCB1 polymorphisms for the plasma concentrations of risperidone and 9-hydroxyrisperidone. ABCB1 3435C > T polymorphisms, however, were significantly associated with pharmacokinetic parameters (C_max, AUC, and MR) of risperidone among individuals who were carriers of CYP2D6^∗10/^∗10 (Yoo et al., 2011). The reduced function variant CYP2D6^∗10 which is highly prevalent among Asian populations and associated with plasma concentrations of risperidone (Roh et al., 2001; Vanwong et al., 2016a) failed to pass the QC in the microarray assay. To make these results more conclusive, CYP2D6 variants (rs3892097, rs1065852, and rs28371725) that showed significant influence in pharmacokinetics of risperidone in previous studies were genotyped using the TaqMan assay. The assay found the involvement of CYP2D6 variant alleles in the metabolism of risperidone. The presence of a strong relationship between CYP2D6 variants and metabolism of risperidone in both microarray and TaqMan assays suggests the important role of CYP2D6 genotyping during risperidone treatment.

A cautious approach for the implementation of the information from this study is recommended as many SNPs did not pass the QC step and these SNPs could not be analyzed for associations with plasma concentrations of risperidone and 9-hydroxyrisperidone. A majority of the SNPs were discarded due to the QC filtering criteria of MAF < 0.05. It is likely that rare variants affecting the plasma concentration of risperidone and 9-hydroxyrisperidone might have been missed. Due to the lack of sample size, only the SNPs with a MAF > 0.05 were included and the possibility of false-positive findings was avoided.

The median risperidone dose was relatively lower (0.5 mg/day) along with plasma levels of the total active-moiety (8.48 ng/ml) which are lower than the values reported earlier causing extrapyramidal side effects (EPS; Spina et al., 2001). Linear regression analysis showed a significant influence of the risperidone dose in the plasma concentration of risperidone, 9-hydroxyrisperidone, and the total active-moiety but no effect on the metabolic ratio. The duration of risperidone therapy did not affect the plasma levels of risperidone, 9-hydroxyrisperidone, active-moiety, and metabolic ratio. Age affected the plasma level of 9-hydroxyrisperidone, the total active-moiety, and metabolic ratio. These findings suggest the possible variables responsible for interindividual variations in plasma concentrations of risperidone and its metabolite. Similar findings were reported in an earlier study with a similar effect of age on the plasma concentrations of the total active-moiety (Aichhorn et al., 2005).

This study has limitations. The lack of validation of the most significant (P < 0.01) DMET variants and association of steady-state plasma concentrations with drug response limits the significance of the study. These results should be verified and clinically interpreted in an independent cohort to avoid the risk of false-positive findings. Also, none of the variants met Bonferroni cutoff value. The exploratory nature of this study, however, suggests the preliminary results to be taken into consideration for further correlation studies. Another limitation is the absence of genomic control which could have false-negative conclusions. The lack of independent validation creates a space for other causative factors, such as age, dosage, and environmental interactions to be implicated for the influence on plasma concentrations of risperidone and its metabolite.

Despite limitations, therapeutic implications from these findings may aid in avoiding toxic plasma levels of risperidone and its metabolite in long term treatment. This pharmacogenomic study provides a mechanistic understanding for individual differences in pharmacokinetics of drugs and may have an outstanding application in the improved personalized treatment of autistic patients in risperidone dosing, efficacy, and side effects. This study opens up the prospect for association of DMET genetic variants with treatment response and adverse effects in future studies. All the patients involved in this study were unrelated which potentially avoided the bias of genotype within families being over-represented.

In summary, this study provides a pharmacogenomic approach to further investigate the DMET genetic variants which influence plasma concentrations of risperidone and its metabolite. The treatment of ASD should be based on genetic factors making the challenge of psychopharmacological treatment more efficacious with fewer adverse events. A caution must be exercised, however, in transferring the pharmacogenomic data into clinical settings unless the genotype-phenotype link is clearly validated.

SM, NV, and CS wrote the manuscript; CS, DP, PL, SM, and AP designed the research study; SM, NV, MC, YH, and JP performed the microarray, genotyping and laboratory part; NN and PL diagnosed and recruited the subjects; SM, CS, and EP analyzed the data.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.