Phenome-wide Mendelian randomization identifying circulating proteins for cardiovascular traits in populations of African ancestry

BackgroundCirculating proteins represent robust drug targets with therapeutic potential. Many discoveries have focused on European-ancestry populations, disregarding minuscule yet substantial proteomic differences that may contribute to disease and/or alter drug generalizability in other ancestry groups. MethodsUsing two-sample Mendelian randomization and colocalization, we analyzed the effects of 1,556 circulating proteins on 145 cardiometabolic centric outcomes to identify robust protein-phenotype associations in African-ancestry populations and reveal African-ancestry associations with heterogenous effects. We further replicated these findings using the proteomic data available from the UK Biobank Pharma Proteomics Project (UKB-PPP), and tested the effect of protein quantity in association with select phenotypes. Population branch statistics (PBS) were also constructed to examine whether protein-genetic instruments under natural selection could lead to significant protein-outcome associations specific to the African ancestry. ResultsWe identified 115 robust protein-phenotype associations in African-ancestry populations. Among these, 52 demonstrated heterogenous effects between African- and European-ancestry populations. We further replicated four cross-platform African-ancestry associations in the UKB-PPP and also revealed four significant, direct associations between protein levels and phenotypes. Ultimately, based on our prioritization criteria, we found that CD36, APOC1, GSTA1, and FOLH1, were shown to influence lipids and heart diseases and were uniquely represented in African-ancestry populations. In addition, using PBS, we showed that 47.5% of the 115 significant protein-outcome associations were possibly driven by cis-acting-protein quantitative trait loci under natural selection. ConclusionsMultiple lines of evidence were used to interrogate proteomic determinants of cardiometabolic diseases and traits in African-ancestry populations. We highlighted actionable circulating protein targets that could represent potential drug targets for cardiovascular diseases specific to populations with African ancestry.