The CYP D inferred metabolizer phenotype
The CYP2D6-inferred metabolizer phenotype describes only one phase of the tramadol (T) ADME (absorption, distribution, metabolism, and excretion) and response process and does not explain all genotypic contribution of an individual’s phenotypic expression . Numerous polymorphisms in the downstream metabolic enzymes uridine diphosphate glucuronosyltransferase, family 1, polypeptide B7 (UGT2B7), K-7174 binding cassette, subfamily B, number 1 (ABCB1), opioid receptor mu 1 (OPRM1), and catechol-O-methyltransferase (COMT) also have been implicated in idiosyncratic response to drugs. These ADME proteins are less well characterized and typically are interrogated in single-gene studies that associate relatively few SNPs/INDELs to rate of drug metabolism and/or enzyme activity , , , , , . It has been demonstrated that combinatorial pharmacogenetic profiles (i.e., data from multiple genes) improved patient outcomes in response to antidepressants ,  and opiates . Therefore, a higher confidence in predicting a metabolizer phenotype may be realized if information from multiple enzymes in an ADME pathway, such as CYP2D6, UGT2B7, ABCB1, OPRM1, and COMT, are included in the analysis. For example, a CYP2D6*4/CYP2D6*4 homozygote is considered a PM and may be prescribed a higher dose of pro-drug (e.g., T) to reach the therapeutic window. However, that same individual may harbor an ABCB1 diplotype which confers decreased efflux of O-desmethyltramadol (M1, the primary active metabolite of T) across the blood brain barrier, enabling a relatively large concentration of M1 to reach OPRM1 and stimulate analgesia propagation. Ultimately, a patient with this pair of diplotypes at CYP2D6 and ABCB1 should experience the desired, and safe, degree of pain relief, but relying solely on CYP2D6 information for this patient would support increasing the tramadol dose which potentially could cause hyperalgesia.
While combinatorial studies have been performed, they rely on targeted genotyping approaches to interrogate a priori SNPs and/or INDELs , , , , , , . Consequently, novel polymorphism(s) cannot be identified that refine estimates of enzyme activity . Massively parallel sequencing (MPS) of the full gene region increases the potential to discover polymorphisms that are currently excluded from phenotype predictions .
Herein, the SNP and INDEL variant effect prediction data presented by Wendt et al.  are expanded upon using the phased data of the 1000 Genomes Project . Full-gene haplotypes of UGT2B7, ABCB1, OPRM1, and COMT were characterized in self-reported healthy individuals. When compared to CYP2D6-predicted metabolizer phenotype for the same individuals , it was demonstrated that NMs by CYP2D6 genotyping may possess poorly active downstream metabolic enzymes. Logistic regression suggests that phenotype predictions using CYP2D6-inferences alone do not explain all phenotypic variability as there may be contribution from polymorphisms in UGT2B7, ABCB1, OPRM1, and COMT.
Materials and methods Polymorphisms in the UGT2B7, ABCB1, OPRM1, and COMT gene regions, including introns, exons, 5′ and 3′ untranslated regions (UTRs), and promoters, were downloaded from Phase 3 of the 1000 Genomes Project and analyzed individually in 5 super- and 26 sub-populations (Table S1) according to Wendt et al. . Haplotypes for each gene were produced according to Table 1 and individual haplotypes are listed in Table S2. Certain polymorphisms characterized were removed from haplotype formation to simplify downstream analyses but capture meaningful levels of variation within each gene. Those excluded variants differ for each gene based on gene size, number of polymorphic sites within each gene, and the consensus variant effect prediction of each polymorphism. In general, polymorphisms that were not scored by Sort Intolerant From Tolerant (SIFT) , , , , , , Polymorphisms Phenotyping v2 (PolyPhen-v2) , , , Protein Variant Effect Analyzer (PROVEAN) , , , or Human Splicing Finder (HSF) , were removed. Private mutations (SNPs or INDELs observed once in the 1000 Genomes Project) were included/excluded on a gene-by-gene basis. ABCB1 was divided into four haplotype blocks based on Sai et al. , . Herein, haplotype block ABCB1-Block-1 has been extended to include untranslated exon 1 (Fig. 1).