Prediction of CYP2D6 phenotype from genotype across world populations

Abstract

Purpose: Owing to its highly polymorphic nature and major contribution to the metabolism and bioactivation of numerous clinically used drugs, CYP2D6 is one of the most extensively studied drug-metabolizing enzymes and pharmacogenes. CYP2D6 alleles confer no, decreased, normal, or increased activity and cause a wide range of activity among individuals and between populations. However, there is no standard approach to translate diplotypes into predicted phenotype.

Methods: We exploited CYP2D6 allele-frequency data that have been compiled for Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines (>60,000 subjects, 173 reports) in order to estimate genotype-predicted phenotype status across major world populations based on activity score (AS) assignments.

Results: Allele frequencies vary considerably across the major ethnic groups predicting poor metabolizer status (AS = 0) between 0.4 and 5.4% across world populations. The prevalence of genotypic intermediate (AS = 0.5) and normal (AS = 1, 1.5, or 2) metabolizers ranges between 0.4 and 11% and between 67 and 90%, respectively. Finally, 1 to 21% of subjects (AS >2) are predicted to have ultrarapid metabolizer status.

Conclusions: This comprehensive study summarizes allele frequencies, diplotypes, and predicted phenotype across major populations, providing a rich data resource for clinicians and researchers. Challenges of phenotype prediction from genotype data are highlighted and discussed.Genet Med 19 1, 69-76.

Figure 1

Comparison of selected allele frequencies across world populations. Allele frequencies differ considerably between populations, as demonstrated on selected allelic variants. Frequencies were calculated from all studies (Supplementary Table S2 online; CYP2D6*2xN (column L), CYP2D6*4 (column Y); CYP2D6*5 (column AA); CYP2D6*10 (column R), and CYP2D6*17 (column S)).

Figure 2

CYP2D6 allele frequencies across world populations. The graph depicts frequencies of no (no)-, decreased (↓)-, normal (↔)-, and increased (↑)-function alleles that were calculated from studies reporting a minimum of allelic variants (Supplementary Table S2 online columns BF–BL). Allele frequencies represent the average of an allele in respective populations and therefore do not add up to 100%. An accompanying graph generated from all studies listed in Supplementary Table S2 online (columns AL–AR) is provided as Supplementary Figure S1 online.

Figure 3

CYP2D6 phenotype prediction from genotype data. An activity score (AS) was assigned to each genotype (no, ↓, ↔, and ↑ indicate genotypes with no-, decreased-, normal-, and/or increased-function allele combinations). Panel a shows average frequencies for the different allele combinations and their respective phenotype classifications into poor (gPM), intermediate (gIM), normal-slow (gNM-S), normal-fast (gNM-F), and ultrarapid (gUM) metabolizer groups. The prefix “g” indicates that the phenotype is predicted from genotype. For panel b, genotypes giving rise to AS = 1 or AS = 2 were grouped as indicated. Panel c depicts the translation of genotype or AS into phenotype according to the classification used in CPIC guidelines. Note that genotypes falling into the AS = 1 group are inconsistently classified as gIM or gNM throughout the literature.