How well do the HapMap SNPs tag functional variants in 217 Drug Metabolizing Enzyme genes? F.C.L. Hyland¹, K. Lazaruk¹, K.A. Haque², R.A. Welch², F.M. De La Vega¹. 1) Applied Biosystems, Foster City, CA; 2) Core Genotyping Facility, Division of Cancer Epidemiology and Genetics, SAIC Frederick, National Cancer Institute, Gaithersburg, MD.
   We examined patterns of linkage disequilibrium (LD) among putatively functional polymorphisms in drug-metabolizing enzyme (DME) genes, many of which have previously been shown to alter drug responses in individuals. We have developed and wet-lab validated 2,394 specific and robust assays in 217 DME genes. Functional, especially rare, alleles may be less likely to be in high LD with a set of common tagging SNPs, such as those selected using HapMap data. We thus investigated the degree of pairwise LD among these putative functional DME SNPs, and between these putative functional variants and the HapMap SNPs (release 19). In collaboration with National Cancer Institutes Core Genotyping Facility we genotyped these polymorphisms on the HapMap samples (n=270). Few putative functional SNPs are in high or perfect linkage disequilibrium with each other: among 1702 SNP pairs within 50 kb both polymorphic in CEU, only 10.6% of the pairs have pairwise r² > 0.85. Low frequency variants (MAF < 0.05) showed less LD with the others. About 40% of the DME putative functional SNPs were found in HapMap. However, of these about 34% failed the internal HapMap quality control metrics, probably reflecting the difficulty in developing robust assays for variants in challenging, highly homologous genes such as these. 23% of DME genes have none of their non-HapMap SNPs tagged by any HapMap SNPs. By determining how many of the non-HapMap DME SNPs are tagged by HapMap SNPs, a predictive model can be generated to assess the power of an association mapping study to capture functional variants in genes. Of the SNPs polymorphic in CEU and not found in HapMap, about 38% were not effectively tagged by any combination of HapMap SNPs (R² < 0.8). Our results suggest that investigators interested in DME putatively functional SNPs should exercise caution in relying solely on tagging SNPs for disease association, drug efficacy, toxicity, and metabolism studies, and should consider typing directly the relevant variants when feasible.