PROJECT SUMMARY/ ABSTRACT Despite mixed evidence of efficacy, drugs like aspirin, and vitamin supplements, are widely used in the US population to prevent chronic diseases. The failure of randomized trials of these over-the-counter (OTC) drugs to consistently demonstrate efficacy, may in part be attributed to a complex interacting heterogeneity of genetic and other effects. Identifying pharmacogenomically defined subpopulations for benefit or harm from drugs and supplements, could enhance precision prevention of cardiovascular disease (CVD), one of the leading causes of death worldwide. Thus, pharmacogenomics, the study of how genomic variation modifies effects of pharmacologically active compounds, has the potential to guide development of precision CVD prevention strategies. Our group has identified candidate genes from multiple pathways related to cardiovascular function that modify the effects of aspirin, and vitamin E in randomized CVD prevention trials. Preliminary data from the Women’ Health Study (WHS), suggests that SCARB1, the gene encoding scavenger B1, a major HDL receptor, is a novel candidate gene for both aspirin and vitamin E effect modification in CVD prevention. Translation of pharmacogenomic markers requires identification of the full set of actionable pharmacogenomic targets, elucidation of their underlying mechanism, and replication of their effects in other cohorts. In this proposal, we examine SCARB1 pharmacogenomic effects on CVD in the WHS and four validation cohorts, ASPirin in Reducing Events in the Elderly (ASPREE), The Physician’s Health Study (PHS), The Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study (ATBC), and The Partner’s Biobank (PBB), allowing broad generalizability of these findings. Given the complexity of CVD and the number of pathways highlighted by pharmacogenomic candidate genes, – estrogen and catecholamine metabolism (COMT), nitric oxide signaling (GUCY1A3), and cholesterol transport (LPA and SCARB1), – there are likely many more pharmacogenomic loci that give rise to a heterogeneity of effects. Identification of the full complement of treatment modifying genes is required to define effects of these overlapping pathways on efficacy and safety of CVD preventive therapy. Currently, discovery of pharmacogenomic loci is limited to candidate gene analysis because genome-wide association study (GWAS) datasets from clinical trials lack sufficient disease events to power classical statistical models. Machine learning approaches, like random forests (RF) obviate the need for statistical power offering an alternative path to discovery. Here, we propose to identify and validate combinatorial pharmacogenomic effects that influence aspirin and vitamin E in CVD prevention in The WHS and PBB by using the RF machine learning methodology. Thus, this proposal seeks to accelerate translation of precision cardiovascular disease prevention by identifying and validating individual and combinatorial effects of novel...