Genome-wide analysis and polygenic prediction of clinical obesity and comparison with body mass index

A new definition of clinical obesity was introduced by the 2025 Lancet Diabetes & Endocrinology Commission to characterize excess body fat and dysfunctional adiposity beyond body mass index (BMI). We performed the first genome-wide analysis (GWAS) of 151,642 cases of clinical obesity and 128,874 controls without obesity from the All of Us Research Program and UK Biobank spanning five ancestry groups and compared these results to GWAS for BMI. We identified 127 independent loci associated with clinical obesity, 63 of which did not share any significant association with BMI. We highlight rs15285 as the most discordant variant, with larger effect size and significance for clinical obesity compared to BMI (ΔZ: +6.35, Δ -log10 P: 17.67). This variant is located in the LPL gene, colocalized to expression quantitative loci for lipoprotein lipase, and associated with elevated triglyceride levels and proteomic markers of insulin resistance and inflammation. Next, we constructed a clinical obesity polygenic score, which had improved association with cardiovascular risk factors and proteomic markers of inflammation (interleukin-6, fibroblast growth factor 21, hepatic growth factor) and insulin resistance (adiponectin, resistin, leptin, and leptin receptor) over a BMI polygenic score. Stratifying individuals into low, intermediate, and high inherited obesity risk groups, we found that clinical obesity polygenic risk reclassified 35% of individuals compared to BMI polygenic risk. Clinical obesity polygenic risk improved discrimination for myocardial infarction, heart failure and stroke over BMI polygenic risk. We replicated the significant improvement in cardiovascular risk prediction of clinical obesity polygenic scores in Atherosclerosis Risk in Communities, Multi-Ethnic Study of Atherosclerosis, Framingham Heart Study, and Women’s Health Initiative. These findings demonstrate that clinical obesity captures genetic loci distinct from BMI that are biologically and clinically relevant to cardiovascular health and can improve cardiovascular genetic risk prediction.