Genetic Risk Scores for Cardiometabolic Traits in Sub-Saharan African Populations

Abstract

There is growing support for the use of genetic risk scores (GRS) in routine clinical settings. Due to the limited diversity of current genomic discovery samples, there are concerns that the predictive power of GRS will be limited in non-European ancestry populations. Here, we evaluated the predictive utility of GRS for 12 cardiometabolic traits in sub-Saharan Africans (AF; n=5200), African Americans (AA; n=9139), and European Americans (EA; n=9594). GRS were constructed as weighted sums of the number of risk alleles. Predictive utility was assessed using the additional phenotypic variance explained and increase in discriminatory ability over traditional risk factors (age, sex and BMI), with adjustment for ancestry-derived principal components. Across all traits, GRS showed upto a 5-fold and 20-fold greater predictive utility in EA relative to AA and AF, respectively. Predictive utility was most consistent for lipid traits, with percent increase in explained variation attributable to GRS ranging from 10.6% to 127.1% among EA, 26.6% to 65.8% among AA, and 2.4% to 37.5% among AF. These differences were recapitulated in the discriminatory power, whereby the predictive utility of GRS was 4-fold greater in EA relative to AA and up to 44-fold greater in EA relative to AF. Obesity and blood pressure traits showed a similar pattern of greater predictive utility among EA. This work demonstrates the poorer performance of GRS in AF and highlights the need to improve representation of multiethnic populations in genomic studies to ensure equitable clinical translation of GRS. Key MessagesO_LIGenetic Risk Score (GRS) prediction is markedly poorer in sub-Saharan Africans compared with African Americans and European Americans C_LIO_LITo ensure equitable clinical translation of GRS, there is need need to improve representation of multiethnic populations in genomic studies C_LI