Cross-biobank comparison of ASCVD heritability and genetic correlation

Estimates of SNP-based heritability for atherosclerotic cardiovascular disease (ASCVD) vary widely across populations, complicating interpretation of genetic architecture. We compared SNP-based heritability and cross-cohort genetic correlation for ASCVD and its nested subphenotypes, coronary events (CE) and myocardial infarction (MI), across Geisinger's MyCode, the UK Biobank, and the NIH's All of Us using GWAS summary statistics and LD score regression. Heritability estimates differed significantly across cohorts, with consistently higher values in the UK Biobank and higher heritability for more narrowly defined phenotypes (MI) relative to broader ASCVD definitions. In contrast, genetic correlations between cohorts were uniformly high for all phenotypes, with confidence intervals spanning 1.0. These results suggest substantial sharing of common variant genetic effects across biobanks and indicate that cross-cohort differences in SNP-based heritability may primarily reflect differences in ASCVD ascertainment and phenotype capture rather than underlying genetic architecture.