Abstract
Interpreting proteomic associations with chronic kidney disease (CKD) is challenging due to the disease's clinical heterogeneity and complex overlap with systemic conditions. We present a framework that identifies causal circulating protein drivers of CKD and delineates their subtype-specific and systemic effects using electronic health record (EHR) data at biobank scale. Using proteome-wide Mendelian randomization, we instrumented cis-acting protein quantitative trait loci for 2,807 circulating proteins and tested them against detailed, EHR-based kidney function outcomes in 464,631 Million Veteran Program participants. Proteins were mapped to nine kidney disease subtypes defined by genome-wide association meta-analyses from the Million Veteran Program, UK Biobank, and FinnGen. Phenome-wide association studies across 1,020 traits distinguished renal versus extra-renal associations. This integrative strategy prioritizes 93 proteins with proteome-wide significance for kidney outcomes, demonstrates subtype-specific relevance, exposes systemic associations, and maps therapeutic targets to nominate candidates for drug development and repurposing.