Genomic atlas of cardiac and adiposity imaging phenotypes

Abstract

Cardiac and adiposity imaging-derived phenotypes (IDPs) provide quantitative measures of organ structure, function and body composition that may resolve genetic contributions to cardiometabolic disease more directly than heterogeneous clinical endpoints. Here, we constructed a whole-genome sequencing-based atlas of 42 IDPs in UK Biobank, comprising 11 cardiac MRI traits in up to 75,562 participants and 31 DXA-derived adiposity traits in up to 66,194 participants. Across these traits, we identified 1,305 independent association signals and prioritized candidate effector genes by integrating fine-mapping, FLAMES, cS2G, Open Targets L2G and rare-variant burden testing. Genetic architecture was structured predominantly within cardiac and adiposity domains, with selective cross-domain sharing concentrated in lean-mass, body-size, growth and hemodynamic signals. In 257,716 All of Us participants, IDP polygenic scores were associated with distinct disease spectra: the left ventricular end-diastolic volume (LVEDV) score with cardiomyopathy and heart failure; the BMI- and height-adjusted gynoid fat mass score inversely with type 2 diabetes. An LV composite polygenic score encoding higher LVEDV and lower left ventricular ejection fraction (LVEF) was associated with increased dilated cardiomyopathy (DCM) risk, with discrimination comparable to that of DCM polygenic score. Combining the two scores improved DCM risk prediction beyond either alone, indicating that the LV composite score captures information not represented by the disease-based score. Finally, multi-trait clustering of LVEDV signals identified clusters with distinct disease associations and epigenomic signatures. One cluster showed the strongest associations with heart failure and DCM, with epigenomic enrichment in cardiac tissues including left ventricle, and nominated candidate effector genes including TBX2, whose expression in independent single-nucleus RNA sequencing data localized to vascular muscle cells. These results highlight cardiac and adiposity IDPs as complementary genetic endophenotypes for dissecting cardiometabolic disease, with all summary statistics available as an open source at https://imaging-pheweb.cerc-genomic-medicine.ca/.