Abstract 13873: Deep Learning-Based GWAS of Aortic Valve Function Links 81 Loci to Aortic Stenosis Risk

Background & Aims: The genetic influences on normal aortic valve (AV) function and their impact on aortic stenosis (AS) risk are of significant interest. We sought to identify common genetic determinants of normal variation in AV function, and to understand their relationship with clinical disease. Methods: We used deep learning to measure peak velocity, mean gradient, and aortic valve area from MRI in UK Biobank participants. We conducted genome-wide association studies (GWAS) in a subset of 44,780 participants without cardiovascular disease. We performed multi-trait analysis of GWAS (MTAG) to incorporate data from AV measurements and AS diagnoses in UK Biobank and FinnGen, and applied PRScs to construct polygenic scores (PGS). Results: In the unadjusted analysis, 26 loci were associated with at least one AV measurement. Incorporating the AS GWAS using MTAG, we identified 81 distinct loci (62 with AV traits, 54 with AS, 35 overlapping), including PCSK9 (β=0.032, P=5.8E-09) and LDLR (β=0.018, P=2.3E-10). Additional loci were associated with lipid metabolism (LPA, FADS2, SORT1), atherosclerosis (CDKN2B, ARHGEF26, OTUD7B, PRDM16), inflammation (IL6, KLF2, TRAF1), and cardiac development and aortic root function (SMAD3, GATA4, ELN, TBX20, TEX41). The PGS for the unadjusted mean gradient was predictive of AS in FinnGen (HR 1.36 for the top 5% of participants vs all others, P=1.8E-14). All of Us participants in the top 5% of the MTAG-adjusted peak velocity PGS had an HR of 2.5 for incident AS (P=4.4E-08). Mendelian randomization showed association between Lp(a) and LDL—but not ApoA—on greater peak velocity (P=2.7E-13, P=1.6E-16, and P=0.60, respectively). Conclusion: We identified 81 genetic loci linked to AV function or AS. AV measurement PGS were strong predictors of AS risk. ApoB and Lp(a) showed causal association with higher peak velocity. These findings have implications for the early pathogenesis of AS and suggest modifiable pathways as targets for prevention.