Gene-based polygenic risk scores analysis of alcohol use disorder in African Americans

Abstract

ABSTRACTGenome-wide association studies (GWAS) in admixed populations such as African American (AA) have limited sample sizes, resulting poor performance of polygenic risk scores (PRS). Based on the observations that many disease-causing genes are shared between AA and European ancestry (EA) populations, and some disease-causing variants are located within boundaries of these genes, we proposed a novel gene-based PRS method (PRSgene) by using variants located in these shared disease-causing genes. Using AA GWAS of alcohol use disorder (AUD) from the Million Veteran Program and EA GWAS of problematic alcohol use as the discovery datasets, we identified 858 variants from 410 genes that were AUD-related in both AA and EA. PRSgene calculated using these variants were significantly associated with AUD in three AA cohorts (P-values: 7.61E-05-6.27E-03; Betas: 0.15-0.21) and outperformed PRS calculated using all variants (P-values: 7.28E-03-0.16; Betas: 0.06-0.18). PRSgene was also associated with AUD in an EA cohort (P-value=0.02, Beta=0.11). In AA, individuals in the highest PRSgene decile had an Odds Ratio of 1.76 (95% CI: 1.32-2.34) to develop AUD compared to those in the lowest decile. The 410 genes were enriched in 54 Gene Ontology biological processes, including ethanol oxidation and processes involving synaptic system, which are known to be AUD-related. Additionally, 26 genes were targets of drugs to treat AUD or other diseases, but may be repurposed to treat AUD. Our study demonstrated that our gene-based PRS had improved performance in evaluating AUD risk in AA and provided new insight into identification of AUD genes.