Poster #25: Stephanie Mohammed PhD

GENETIC MEDIATION OF CORONARY ARTERY DISEASE RISK THROUGH ANDROGEN AND METABOLIC PATHWAYS RELEVANT TO POLYCYSTIC OVARY SYNDROME

Stephanie Mohammed (1), Aude Ikuzwe Sindikubwabo(1), Lindsay Guare(1), Lanna Caruth(1), Marijana Vujkovic(2), Shefali Setia-Verma(1),(3) (1) Department of Pathology and Laboratory Medicine, University of Pennsylvania. (2) Department of Medicine, University of Pennsylvania. (3) Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania.

Women with polyendocrine metabolic ovarian syndrome (PMOS) have an approximately 2-fold increased risk of coronary artery disease (CAD) and adverse cardiovascular events compared with women without PMOS. Androgen excess, obesity, insulin resistance, and dyslipidemia are the hallmark biochemical derangements defining PMOS across diagnostic criteria, captured by total testosterone, BMI, HbA1c, and LDL cholesterol, making them biologically plausible mediators of this excess genetic CAD risk. Yet traditional CAD polygenic risk scores (PRS), built from marginal SNP-CAD associations, do not capture these specific pathways. We aimed to quantify the proportion of established CAD genetic risk mediated through each pathway and test whether mediation-informed PRS improve identification of high genetic risk. METHODS: We analyzed 406 established CAD-associated SNPs in a European-ancestry female cohort (All of Us). For each SNP, total genetic effect on CAD was decomposed into direct and indirect (mediated) components via the product-of-coefficients method, across mediators BMI, LDL cholesterol, HbA1c, and total testosterone. Proportion mediated (PM) was estimated per SNP-mediator pair and top-mediated loci were identified among SNPs with significant total CAD effect. These mediation estimates were also used to construct exploratory mediator-informed polygenic risk scores (PRS), benchmarked against a traditional CAD PRS in an independent validation cohort. RESULTS: Total testosterone was the dominant mediator of CAD genetic risk (median PM ~59% across tested SNPs), with near-complete mediation (94-96%) at its most strongly mediated loci (CFDP1, BLTP3A, LINC00881). LDL cholesterol mediated a smaller, biologically coherent share (median PM ~12%, rising to 48-58% at top loci including APOE and CELSR2), recovering APOE, a canonical cholesterol-metabolism gene, as a positive control validating the approach. BMI and HbA1c showed minimal mediation genome-wide (median <3%), reaching at most ~12% (BMI) and <2% (HbA1c) even at their single most-mediated locus, suggesting adiposity and glycemic control play a secondary role relative to the androgen axis. CONCLUSIONS: Androgen excess, and to a lesser extent LDL cholesterol, routes a substantial share of inherited CAD risk in women, directly implicating core PMOS mechanisms rather than adiposity or glycemic dysregulation in genetic CAD susceptibility. Exploratory mediation-informed PRS incorporating testosterone and LDL showed early evidence of improved discrimination over a traditional CAD PRS (AUC 0.75 vs 0.72) and flagged partly distinct high-risk subgroups, warranting further validation as a route toward mechanism-specific CAD risk stratification in PMOS.