Increasing research into polygenic risk scores (PRSs) for disease prediction highlights their clinical potential for informing screening, therapeutics, and lifestyle.1 While their use enables risk prediction in individuals of European ancestry, PRSs can have widely varying and much lower accuracy when applied to non-European populations.2, 3, 4 Although the nature of this bias is not well understood, it can be attributed to the vast overrepresentation of European ancestry individuals in genome-wide association studies (GWASs), which is 4.5-fold higher than their percentage of the world population; conversely, there is underrepresentation of diverse populations such as individuals of African ancestry in GWASs, which is one-fifth their percentage.3 Potential explanations for the limited portability of European-derived PRSs across populations includes differences in population allele frequencies and linkage disequilibrium (LD), the presence of population-specific causal variants or effects, or potential differences in gene-gene or gene-environment interactions.4 However, in traits such as BMI and type 2 diabetes, 70%–80% of European-based PRS accuracy loss in African ancestry has been attributed to differences in allele frequency and LD; therefore, most causal variants discovered in Europeans are likely
