Abstract P256: Generalization and Fine Mapping of Genetic Associations with Hematocrit in Multi-ethnic Populations: The Population Architecture Using Genomics and Epidemiology (PAGE) Study
Introduction: Variability within the normal population range of hematocrit is associated with stroke and myocardial infarction. Published GWAS of hematocrit have identified multiple loci, yet few studies have included populations of Hispanic or African descent, thereby limiting opportunities to identify population-specific variants or narrow associated regions for functional analysis. We present a fine-mapping analysis of six previously identified hematocrit loci in African American and Hispanic/Latino participants of the PAGE study.
Methods: Approximately 200,000 genotyped or imputed Metabochip variants were examined for association with hematocrit (proportion of whole blood comprising red blood cells) in 19,822 Hispanic/Latino and 19,973 African American participants. SNPs were excluded on a population-specific basis if effective heterozygosity was < 30. Primary and conditional analyses were performed in Plink, ProbABEL, or SuGen; fixed-effects meta-analyses were performed in Metal. Trans-ethnic and ancestry-specific meta-analyses were performed in MANTRA to generate 99% credible intervals for previously published variants that generalized to our populations.
Results: We first examined whether 8,261 variants in five previously identified hematocrit loci (HFE, ABO, HK1, SH2B3/ATXN2, and TMPRSS6) were associated with hematocrit in our study populations. Three loci generalized (p<1.7x10-4) to Hispanic/Latino participants (ABO, HK1, and TMPRSS6) and three generalized to African Americans (HFE, ABO, and SH2B3/ATXN2). Among generalized loci, conditional analyses adjusting for published variants in European-ancestry or East Asian populations did not identify any independently associated SNPs in Hispanic Latinos or African Americans (p<1.3x10-5). Trans-ethnic meta-analysis for the ABO locus resulted in a 5 SNP, 13kb 99% credible interval, shorter than both the Hispanic/Latino (17kb) and African American (360kb) credible intervals. In discovery analysis, we identified one variant associated with hematocrit in Hispanic/Latinos at array-wide significance levels (PROX1 locus, p<3.0x10-7). No novel loci were identified in African Americans.
Conclusion: Our findings provide evidence that the same genomic loci influence normal variation in hematocrit values across diverse ancestral populations. Trans-ethnic fine mapping of the gene-rich ABO locus—which has been associated with ischemic stroke, thrombosis, and myocardial infarction in addition to hematocrit GWA studies—suggests that a functional variant may reside in the first intron of the ABO coding region. Additionally, identification of previously unidentified associations in Hispanic/Latinos emphasizes the importance of including diverse populations in association studies as well as the potential to identify population-specific functional variants within known or discovery loci.
Author Disclosures: C.J. Hodonsky: None. U. Schick: None. J. Kocarnik: None. C. Schurmann: None. S. Buyske: None. M. Fornage: None. L.A. Hindorff: None. D. Lin: None. B.M. Psaty: None. A.P. Reiner: None. R. Tao: None. K.E. North: None. R.J.F. Loos: None. C. Kooperberg: None. C.L. Avery: None.
- © 2016 by American Heart Association, Inc.