Cross-Ancestry Comparison of Aptamer and Antibody Proteomics Measures

About This Grant

PROJECT SUMMARY Cardiovascular disease (CVD) is the leading cause of death in the United States, yet drivers of disease and risk markers remain poorly characterized. Although CVD is more prevalent among non-Hispanic Blacks and Hispanics compared to non-Hispanic Whites, these populations are underrepresented in existing genetic and epidemiological studies of disease. New affinity-based high-throughput proteomics platforms offer unprecedented opportunities to study disease mechanisms across diverse populations, and to search for new clinical biomarkers of disease. However, initial comparisons of the leading proteomics platforms suggest that at least one-third of protein measures correlate poorly between platforms, resulting in platform-specific or platform-discordant genetic and epidemiological associations, which may confound biological interpretation of results across studies. Genetic variants associated with circulating protein abundance measures may highlight assays with on-target affinity-probe binding, as well as assays which may be non-specific or prone to interference from protein-altering variants, which may affect affinity-probe binding and thereby alter protein measure. As frequencies of these protein-altering variants may vary between populations, they may drive differences in assay performance across populations. Here, we propose the first examination of protein measure agreement in diverse populations, using newly generated affinity-based proteomics data across four large studies of participants within the Trans-Omics for Precision Medicine (TOPMed) consortium. After quantifying the correlation of protein measures for each platform within each study, and the heterogeneity of these correlations between ancestral populations, I will identify technical and biological predictors of platform correlation. Next, I will analyze whole genome sequences for contributing individuals to identify genetic predictors of circulating protein abundances, as measured by each platform. I will characterize signals across platforms to identify protein-altering genetic variants which may drive assay interference, and synthesize these genetic associations with platform correlation coefficients to predict assay efficacy. Finally, I will model protein measures from each platform against relevant cardiometabolic phenotypes to evaluate agreement of epidemiological associations across platforms, and explore whether adjustment for protein-altering genetic variants in epidemiological models may improve the concordance of associations. The results of this analysis will disentangle reproducible protein measures and association results from those likely reflecting genetically- driven assay interference, within and across populations, and present new strategies to harmonize downstream associations for the impacted proteins across platforms.