Plasma proteomic signatures of incident cardiovascular disease according to serum urate levels: a prospective cohort study

Hyperuricemia is associated with increased cardiovascular disease (CVD) risk. Whether plasma protein–CVD associations vary with serum urate and whether proteomic information adds predictive value in hyperuricemia remain unclear. This prospective UK Biobank study included 45823 participants without baseline CVD and Olink measurements of 2920 plasma proteins. Hyperuricemia ( n = 7336) was defined as serum urate >400 μmol/L in men or >360 μmol/L in women. CVD comprised coronary heart disease, stroke and heart failure. Multivariable Cox models tested protein interactions with continuous serum urate and, in complementary analyses, binary hyperuricemia status, with false discovery rate (FDR) correction. Among participants with hyperuricemia, weighted gene co-expression network analysis identified protein modules for functional enrichment analysis. In an exploratory prediction analysis, a proteomic risk score (ProRS) was developed using penalized Cox regression in the training set ( n = 5868) and evaluated in the internal validation set ( n = 1468). During a median follow-up of 13.4 years, 5636 participants developed CVD. At FDR < 0.05, 130 proteins showed interactions with continuous serum urate and five with hyperuricemia status; four (FAM3B, CELSR2, FAM3C, and CST6) were significant in both analyses with concordant directions. Module M02 contained 76 of the 130 continuous-urate interaction proteins and remained enriched after adjustment for Olink assay panel. M02 had vascular, extracellular matrix, and immune-related functional annotations. Each 1-SD higher M02 eigengene was associated with higher CVD risk (hazard ratio, 1.31; 95% CI, 1.22–1.41). Adding the 47-protein ProRS to the enhanced clinical model increased the validation C-index from 0.703 to 0.738 (difference, 0.0358; 95% CI, 0.0111–0.0591). The corresponding gain over this clinical model plus N-terminal pro-B-type natriuretic peptide (NT-proBNP) and growth differentiation factor 15 (GDF15) was 0.0133 (95% CI, −0.0001–0.0261). Plasma protein–CVD associations varied according to serum urate levels, with interaction proteins enriched in a module with vascular, extracellular matrix, and immune-related annotations. The 47-protein ProRS improved discrimination beyond the enhanced clinical model in the internal validation set. Independent replication and external validation are needed to assess reproducibility and clinical value.