Limited Association of the 2988g>a Single Nucleotide Polymorphism with CYP2D6*41 in Black Subjects

To the Editor In the August 2004 issue of the Journal, Raimundo et al1 described a novel intronic single nucleotide polymorphism (SNP), 2988G>A, that is highly predictive for intermediate metabolizer (IM) status as assessed with the probe drug sparteine. In their white population, 2988G>A was present in 52 of 56 CYP2D6*41 alleles. Because all 2988A/*null subjects were IMs and this SNP was not found on any other allele, they recommended identifying CYP2D6*41 as [−1584C, 2988A] and using 2988A as a single assay marker. Currently, CYP2D6*2 and *41 are defined by identical coding regions but discriminated by −1584C>G (*41 = C and *2 = G). Investigating phenotype-genotype-discordant cases, we have independently discovered this SNP in North American white subjects. Genotyping of 203 subjects revealed that all 31 CYP2D6*41 alleles carried 2988A (frequency = 0.076).2,3 Within this cohort, the only CYP2D6*41×2 allele observed was negative for the 2988G>A SNP (confirmed by pedigree analysis), similar to that for subject Tu 140 described by Raimundo et al1 (Fig 1). Whereas the 2988G>A genotyping results are in agreement with their findings, the functional consequences with the use of dextromethorphan as a phenotyping probe are less conclusive. Of 7 subjects in our data set with a *41[2988A]/*null genotype, 3 were extensive metabolizers (EMs) (Fig 2). Hence, although 2988A appears to be associated with an IM sparteine phenotype, this does not necessarily seem to be the case for assessments with dextromethorphan. Simultaneous polymerase chain reaction (PCR)–restriction fragment length polymorphism (RFLP)–based genotype analysis of 2850C>T and 2988G>A. A 225–base pair (bp) fragment was generated from a 6.6-kilobase CYP2D6-specific amplicon5 with primers 5′-TGC TCT CGG CCC TGC TC and 5′-TTC ATG GGC CCC CGC Cga aAC CCT T (partial Xmn I restriction site [bold type]). PCR fragments were cut with Fsp I (2850C) and Xmn I (2988G). Depending on the absence or presence of 2850C>T and 2988G>A, the following patterns were generated: *1, 62 + 142 + 21 bp; *2 and *41[2988G], 204 + 21 bp; and *41[2988A], 225 bp. Presence of 2850C>T and 2988G>A and genotypes are as indicated for each subject. F, Father; M, mother; D, daughter (segregation of CYP2D6*41×2[2988G] in a 2-generation family). *41/*41 with diamond symbol denotes a black subject. Correlation between CYP2D6*41[2988A] alleles and CYP2D6 activity. Metabolic urinary metabolic ratios of dextromethorphan (DM) to dextrorphan (DX) lower than 0.3, greater than 0.3, and greater than 0.03 indicate poor metabolizer (PM), intermediate metabolizer (IM), and extensive metabolizer (EM) phenotypes, respectively; *null, *red, and *fct denote nonfunctional, reduced, and fully functional alleles, respectively. The sequenced CYP2D6*5/*41[2988A] subject is indicated by asterisk within triangle. The number of subjects within each genotype group is as shown. In contrast to the strong linkage of 2988A with CYP2D6*41 in white subjects, only 10 of 48 CYP2D6*41 alleles among 251 black subjects were positive for this SNP (phenotype available for 8) (Fig 2).3 Allele frequencies were 0.076 and 0.021 for CYP2D6*41[2988G] and *41[2988A], respectively. As a sole marker, 2988G>A would correctly exclude a poor metabolizer (PM) phenotype but would not detect about 80% of CYP2D6*41 alleles in this population sample. Among other explanations, Raimundo et al1 speculated that 2988G>A may cause low expression as a result of erroneously spliced messenger ribonucleic acid. Information theory-based splice site analysis, however, did not reveal any cryptic splice sites that may be activated by 2988A but did reveal the potential loss of an SF2/ASF splice-enhancer binding site.4 Furthermore, resequencing 9 kilobases of a CYP2D6*41[2988A] allele did not reveal any other unique SNP that may explain lower activity when compared with CYP2D6*1 (AY545216) and CYP2D6*2 (NG_003180) reference sequences. In summary, cytochrome P450 (CYP) 2D6 phenotype prediction from genotype data remains a challenging task, especially in nonwhite and ethnically diverse or admixed populations. Our previously reported approach using −1584C>G as a PM exclusion criterion5 may be compromised by the occurrence, albeit infrequently, of nonfunctional −1584G-positive alleles (1 CYP2D6*4 and *8, respectively, reported by Raimundo et al1) and may rather be used as a screening tool than a definite assessment. Using 2988G>A as a marker may be helpful to identify a small proportion of African Americans and other subjects of black African descent as non-PMs, but it is not as powerful a predictor in these populations as it may be in white subjects. Furthermore, our genotype-to-phenotype correlation data imply that the presence of 2988A does not necessarily predict intermediate metabolism for all CYP2D6 substrates alike. To further assess the usefulness and associations of these SNPs, −1584C>G and 2988G>A, both are included in ongoing investigations. The authors do not have any conflicts of interest