Pharmacogenetics of Cytochrome P450 Enzymes in American Indian and Caucasian Children Admitted to a Psychiatric Hospital

Introduction

Pharmacogenetics (PGx) refers to interactions between an individual's genetics and that individual's response to medication, often on the basis of a single gene polymorphism (Vogel 1959). The utilization of PGx testing facilitates more individualized drug therapy and may improve patient and public health outcomes. The cytochrome P450 (CYP450) enzyme system is composed of 57 genes from 18 families (Spina and de Leon 2015). CYP450 enzymes are utilized for phase I metabolism, with CYP2D6 responsible for ∼25% of medication metabolism with >70 alleles described (Ingelman-Sundberg 2005; Human Cytochrome P450 [CYP] Allele Nomenclature Database website 2015). CYP2C9 is responsible for ∼15% of medication metabolism with 60 alleles described (Zhou et al. 2010; Human Cytochrome P450 Allele Nomenclature Database website 2015) and CYP2C19 is responsible for ∼10% of medication metabolism with 35 alleles described (Ingelman-Sundberg et al. 2007; Human Cytochrome P450 Allele Nomenclature Database website 2015).

Genetic polymorphisms can occur and lead to altered rates of drug metabolism, with a large degree of CYP450 expression variability existing across racial groups (Wilkinson 2005). The degree of metabolic activity of a specific CYP450 enzyme can correspond with drug plasma concentrations for medications that are respective substrates (Baumann et al. 2004). Accordingly, patients who are poor metabolizers (PMs) may be at higher risk for adverse drug events, whereas ultrarapid metabolizers (UMs) may have increased risk for therapeutic failure while taking Food and Drug Administration (FDA) approved maximum doses (Baumann et al. 2004). A large proportion of psychotropic medications are metabolized through the CYP450 system, and clinical guidelines exist for the use of CYP2D6 and CYP2C19 phenotyping in a psychiatric setting (de Leon et al. 2006; Hicks et al. 2013, 2015). de Leon and colleagues emphasize that in order to use CYP450 PGx testing properly, the clinician must acknowledge that drug response is not only controlled by genetic factors, but also by environmental factors (de Leon et al. 2006). Hicks and colleagues similarly acknowledge that PGx testing is one of many important pieces of clinical information that can be applied to medication management, with application of PGx testing being most appropriate during therapy initiation (Hicks et al. 2013, 2015). Therefore, good judgement and knowledge of pharmacologic principles must be utilized when making treatment decisions for patients who have CYP450 PGx test results available (de Leon et al. 2006).

Shodair Children's Hospital is an 88 bed acute and residential psychiatric hospital serving children and adolescents in the Northwestern United States. Our patient population consists primarily of American Indians (AIs) and Caucasians with mood disorders, aggressive behavior, suicidal ideation, or a combination of these. The Shodair Children's Hospital Clinical Genetics Laboratory performs genotyping for CYP2D6, CYP2C9, and CYP2C19 when ordered by providers. Our genetics laboratory does not genotype the highly important CYP3A family. This is largely because of the unpredictability of CYP3A4 phenotypes, as a substantial number of both environmental and genetic factors outside of the CYP3A4 locus lead to variability in metabolic activity (Klein and Zanger 2013). Few reports of CYP450 PGx phenotype and genotype frequencies have been described within the context of a psychiatric setting for Caucasians, and none have been described for AIs. AIs are a largely underinvestigated racial group in regard to PGx in general (Fohner et al. 2013). However, with an estimated 5,200,000 AI/Alaska Natives (AN) residing in the United States (Census Brief 2010), AIs are an important racial group in whom the application of PGx may improve drug therapy outcomes. Because of the paucity of PGx reports for AIs and Caucasians in a psychiatric setting, we report and compare allele and phenotype frequencies for CYP2D6, CYP2C9, and CYP2C19.

Methods

A retrospective chart review was conducted in order to evaluate CYP450 PGx data from the Shodair Children's Hospital Clinical Genetics Laboratory between 2006 and 2014. The factors determining which patients were selected to have CYP450 testing performed were clinical judgment (multiple failed responses to medications or a history of unexpected adverse drug reactions), and an active health insurance coverage benefit for PGx testing. Included patients self-identified as either AI or Caucasian and were between the ages of 5 and 17 at hospital discharge. All patients gave written informed consent for CYP450 PGx testing and for use of this data for research purposes. Patient DNA was isolated from whole blood using the Puregene system (QIAGEN, Valencia, CA). CYP2D6 genotyping was performed using the xTAG® CYP2D6 Kit (Luminex, Austin, TX). CYP2C9 and CYP2C19 genotyping was performed by laboratory-developed assays using allele-specific quantitative polymerase chain reaction (qPCR) and/or melt-curve analysis. Phenotype calls for CYP2D6 were conducted as per the xTAG® CYP2D6 Kit (Luminex, Austin, TX). CYP2D6 phenotypes classified as “unknown” had a duplication detected for any combination of tested alleles, with the exception of *1/*1, *2/*2, *35/*35, *1/*2, *1/*35, or *2/*35. Phenotype calls for CYP2C9 are as follows: Extensive metabolizers (EMs) have two active alleles (*1); intermediate metabolizers (IMs) have one active allele and one inactive allele (*2-*6); and PMs have two inactive alleles. Phenotype calls for CYP2C19 are as follows: EMs have two active alleles (*1); EM/IMs have one inactive allele (*2-*8) and one rapid allele (*17); IMs have one active allele and one inactive allele; PMs have two inactive alleles; rapid metabolizers (RMs) have one active and one rapid allele; and UMs have two rapid alleles. We employed descriptive statistics using 95% confidence intervals (CIs) in order to estimate significance. If the CIs of two frequencies do not overlap, there is a significant difference between them (p < 0.05). These findings were confirmed using the χ² test. Patient genetic data was de-identified prior to analysis through removal of unique identifiers. This project was designated as institutional review board (IRB) exempt by our hospital's appointed research review committee because of its retrospective design, patient de-identification, and the Basic Health and Human Services (HHS) policy for protection of human research subjects (56 FR 28012,28022 46.101[b]4).

Results

A total of 123 AIs and 688 Caucasians met criteria for inclusion. Patient's sex was 50% and 42% female for AIs and Caucasians, respectively. Patient's mean age in years was 12 (SD = 3.3) and 12 (SD = 3.1) for AIs and Caucasians respectively. Phenotype and allele frequency data can be found in Table 1. All alleles were in Hardy–Weinberg equilibrium. The overall prevalence of CYP2D6 PMs was 8.3% (95% CI 6.1%, 10.4%), 9.3% in Caucasians (95% CI 7.1%, 11.5%), and 2.4% in AIs (95% CI 0%, 5.2%). The overall prevalence of CYP2D6 UMs was 1.6% (95% CI 0.7%, 2.5%), 1.6% in Caucasians (95% CI 0.7%, 2.5%), and 1.6% in AIs (95% CI 0%, 3.9%). The overall prevalence of CYP2C9 PMs was 3% (95% CI 1.7%, 4.2%), 3.2% in Caucasians (95% CI 1.8%, 4.6%), and 1.8% in AIs (95% CI 0%, 4.2%). The overall prevalence of CYP2C19 PMs was 2.5% (95% CI 1.3%, 3.6%), 2.9% in Caucasians (95% CI 1.6%, 4.2%), and 0% in AIs. The overall prevalence of CYP2C19 UMs was 1.5% (95% CI 0.6%, 2.4%), 1.6% in Caucasians (95% CI 0.6%, 2.6%), and 0.9% in AIs (95% CI 0%, 2.6%).

Discussion

We found clinically important phenotypes (PMs or UMs) in 4% of AIs and 11% of Caucasians for CYP2D6; 2% of AIs and 3% of Caucasians for CYP2C9; and 1% of AIs and 5% of Caucasians for CYP2C19. The AI CYP2D6 and CYP2C9 allele frequencies in this study are similar to those previously described, whereas no comparator is currently available for CYP2C19 (Fohner et al. 2013). Our study found that the prevalence of clinically important phenotypes for Caucasians was similar to previous reports (de Leon et al. 2009; Jürgens et al. 2012). As discussed previously, patients with these clinically important phenotypes may experience either increased risk of medication adverse events or reduced therapeutic effect secondary to altered serum drug concentrations. Psychiatric medications that are substrates of CYP2D6 primarily include, but are not limited to, antidepressants (tricyclic antidepressants, fluoxetine, paroxetine, citalopram, escitalopram, venlafaxine, duloxetine, and mirtazapine), antipsychotics (haloperidol, chlorpromazine, fluphenazine, perphenazine, clozapine, olanzapine, risperidone, aripiprazole, and quetiapine), and atomoxetine for attention-deficit/hyperactivity disorder (Spina and de Leon 2015). Medications that are substrates of CYP2C9 primarily include, but are not limited to, antidepressants (fluoxetine), hypnotics (zolpidem and zopiclone) and the antiepileptic valproic acid (Spina and de Leon 2015). Medications that are substrates of CYP2C19 primarily include, but are not limited to, antidepressants (sertraline, citalopram, and escitalopram) and the anxiolytic diazepam (Spina and de Leon 2015). Although there were subtle differences in CYP450 phenotypes between our two studied racial groups, these results should not dictate choice of medication in the absence of CYP450 PGx testing or the decision to perform CYP450 PGx testing.

Conclusions and Clinical Significance

PGx outcomes research in a psychiatric setting is in its infancy. Although PGx testing has been demonstrated to affect clinicians' treatment decisions, it is uncertain if these treatment decisions lead to clinically important outcomes (Brennan et al. 2015). de Leon and colleagues performed CYP450 testing on 4532 patients at the Kentucky state hospitals utilizing a commercially available PGx test. The CYP2D6 overall prevalence was 7.6% (CI 7%, 8.3%) PMs and 1.5% (CI 1.2%, 1.9%) UMs. The CYP2C19 overall prevalence was 2.2% (CI 1.8%, 2.7%) PMs and 97.8% (CI 97.3%, 98.2%) EMs (de Leon et al. 2009). Although genotype frequencies were not presented, Jürgens and colleagues found that 12% of 967 psychiatric inpatients were either PMs or UMs for either CYP2D6 or CYP2C19. Of these patients, 86% were taking medications metabolized by CYP2D6 or CYP2C19 (Jürgens et al. 2012). Although most CYP2D6 phenotype studies have not found differences between the general population and patients with psychiatric disorders, primarily schizophrenia (Dawson et al. 1994; Daniels et al. 1995; Dandara et al 2001; Rasmussen et al. 2006), it is unknown if this relationship is true for our pediatric population who primarily have affective disorders. Therefore, until further research becomes available, our results should not be extrapolated to the general AI and Caucasian populations of the Northwestern United States. CYP450 PGx testing offers additional information to those making drug treatment decisions. A percentage of AI and Caucasian patients tested in the Northwestern United States were found to have clinically important phenotypes. Further outcomes research will shape the role of CYP450 PGx testing in the psychiatric setting. When PGx test results are available, the use of good clinical judgment as well as PGx practice guidelines should be utilized. This study is the first to identify differences in genetic polymorphism frequencies of the CYP450 system in AIs and Caucasian youth admitted to a psychiatric hospital. Our findings are important, as they found that clinically important CYP450 phenotype abnormalities were uncommon, but not rare, in both AI and Caucasian youth suspected of having abnormal drug metabolism. Clinical judgment and insurance coverage determines who receives PGx testing at our facility, with the majority of patients receiving testing after multiple failed medication trials or adverse drug events. It is unknown what genotypes and phenotypes would present if our sample were randomized or if CYP2D6 testing had been fully genotyped without use of a commercially available kit. Additionally, patient's race was self-identified, and verification of blood quanta was not conducted. Our findings warrant further study to determine if allele and phenotype frequencies are generalizable to the larger population of Caucasian and AI/AN youth in the Northwestern United States. Finally, our results assist clinicians in determining circumstances in which PGx testing may have the most impact in a psychiatric population. Our found frequencies of clinically important phenotypes should be considered, along with other clinical information, when deciding if PGx testing is cost effective.