Conditional models were tested to evaluate evidence in EAs and AAs for contributions from multiple independent risk loci in the region containing ADH1B, LOC100507053, ADH1C, PDLIM5 and METAP1. These models incorporated pairwise combinations of the most significant SNPs in each of the genes. They showed that in EAs the signal at rs1229984 in ADH1B remained highly significant after adjustment for rs116203444 in LOC100507053 (Prs1229984 = 4.41 × 10−14) or rs11724023 in PDLIM5 (Prs1229984 = 4.65 × 10−11), but rs116203444 was not significant after adjustment for the ADH1B*rs1229984 (P = 0.06). PDLIM5*rs11724023 remained significant after adjusting for either of the other SNPs, with similar adjusted P-values (Prs11724023 = 7.73 × 10−7, Prs11724023 = 7.73 × 10−7). In AAs, the SNPs in ADH1B and LOC100507053 are highly correlated, and mutual adjustment rendered both P-values non-significant. The association in METAP1 was no longer significant after adjustment for either ADH1B or LOC100507053. Although attenuated, ADH1C remained significant after adjustment for any of the other three loci tested. Conditioning on ADH1B caused the greatest attenuation (Prs12639833 = 0.01). Thus, our results support the presence in both populations of multiple independent risk loci in this region.
