Abstract 051: Trans-ethnic Metabochip Genotyping of Established Lipid Loci Identifies Low Frequency Susceptibility Variants and Additional Independent Signals in Known Loci

Abstract

Genome-wide association studies (GWAS) have identified ∼100 loci for triglycerides (TG), high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C), but much of the trait heritability remains unexplained. We conducted a trans-ethnic study in 5,884 African Americans (AA) in the Population Architecture using Genomics and Epidemiology (PAGE) consortium [the Atherosclerosis Risk in Communities Study (ARIC), the Multiethnic Cohort Study (MEC) and the Women's Health Initiative (WHI)], and 1,719 East Asians (ASN) from the Cebu Longitudinal Health and Nutrition Survey (CLHNS) to analyze high-density genotyped Metabochip SNPs for association with TG, HDL-C and LDL-C. We aimed to identify population-specific and additional independent signals at established lipid loci and to assess the relative evidence of association with SNPs previously reported to have a functional effect. Among the 58 lipid loci tested, evidence of association (P<10–4) was observed in at least one of the populations for 21 loci, including 19 loci in AA and 7 loci in ASN. Using sequential conditional analysis, we detected in AA that the LDL-C locus PCSK9 harbored five independent nonsense or non-synonymous variants (C679X, A443T, N425S, Y142X and L253F). All had low frequency (MAF: 0.004–0.095), were unique to AA (MAF=0 in non-African ancestry populations) and have been reported previously to affect protein function. These five SNPs collectively explained 2.9% of the LDL-C variability compared to 1.1% explained by C679X alone. Similarly, the APOE-C1 cluster revealed two AA-specific TG signals represented by rs12721054 (APOC1 3'UTR, MAF=0.12) and rs769455 (APOE R163C, MAF=0.002, prior evidence of biological function) that together accounted for 1.7% of the total TG variation compared to the 1.0% explained by the strongest APOE-C1 SNP alone. At a third locus, TG-associated APOA5, two reportedly functional variants, S19W (MAF=0.059) and −3A>G (MAF=0.26), exhibited the strongest association in AA and ASN, respectively. In AA, three independent signals at APOA5 explained 1.7% of the TG variation; in ASN, two independent signals explained 3.3% of total variation. The reportedly functional variants were included or well-represented on Metabochip at ten loci. At eight of the ten loci, the reported variants (at 6 loci) or proxies (r2>.95, at 2 loci) showed the strongest association in at least one population. This study highlights the value of the trans-ethnic high-density genotyping to identify a wider spectrum of susceptibility variants at reported loci, both additional independent signals and population-specific SNP effects, that more than double the trait variation explained. In addition, the large fraction of reportedly functional SNPs that showed the strongest evidence of association suggests that lead SNPs at novel loci frequently may be good candidates for functional studies.