Plasma Proteomics Identifies Insulin Resistance Signatures Predictive of Cardiometabolic Disease Risk

Insulin resistance drives cardiometabolic disease, yet its molecular signatures and tissue origins remain incompletely characterized, and scalable assessment methods are lacking. Here, we apply Multi-Workflow Proteomics on plasma from 161 individuals spanning the metabolic spectrum defined by hyperinsulinemic–euglycemic clamp–derived insulin sensitivity. We identify 488 proteins associated with insulin sensitivity, revealing contributions from liver, adipose tissue, and immune cells alongside underappreciated roles for brain and heart. An exercise intervention demonstrated these signatures are modifiable. We developed a model combining 13 proteins, including IGFBP1, LEP, GDF15, PON3, and LDLR, with clinical variables (sex, HbA1c, TG/HDL ratio) that estimates hyperinsulinemic–euglycemic clamp-derived insulin sensitivity (R² = 0.73). Applied to ~20,000 UK Biobank participants, estimated insulin sensitivity outperformed TG/HDL in predicting type 2 diabetes (c-index 0.86 vs. 0.71) and other cardiometabolic outcomes, including obesity, cardiovascular disease, and chronic kidney disease. This proteomic atlas enables scalable insulin resistance assessment and precision risk stratification.