In-Depth Survey of the Transferability of Genetic Associations of Lipid Traits to African Populations

Abstract

The transferability of genetic associations detected in European and US based populations has been shown to be lowest in African populations with high variations among traits as well as different African regions. We report an assessment of the transferability of genetic associations from the Global Lipids Genetics Consortium (GLGC) study, the largest meta-analysis for lipid traits (LDL, HDL, and Triglyceride), to seven African ancestry cohorts comprising participants from eastern, western, and southern Africa. Independent genome-wide significant signals (p- value < 5x10-8) from the GLGC were identified using the LD clumping function in PLINK 2.0. The transferability of these signals were assessed in the seven cohorts, at the standard genome- wide significance threshold (p-value <5x10-8) as well as a less stringent replication threshold (p-value <5x10-6). For LDL-C and HDL-C, the GLGC-African-specific ancestry GWAS signals were considerably more transferable compared to the GLGC-multi-ancestry GWAS. Although, transferability estimates generally improved with the size of the replication cohort, there were several exceptions. Moreover, we noted major differences in replication among the African regions with LDL-C signals showing higher transferability in Southern Africa, HDL-C signals showing higher transferability in East and TG signals showing higher transferability to West African cohorts. Among the three lipid traits, LDL-C signals showed highest concordance in beta values with the GLGC beta-values. Our results suggest that transferability of signals depends on several factors including the genetic architecture of the trait being investigated. Therefore, the application of polygenic risk scores and other estimates for precision medicine requires independent in-depth evaluations for each trait in each African region.