Plasma proteomics enhances heart failure risk prediction among individuals with type 2 diabetes: a prospective cohort study

BACKGROUND: Heart failure (HF) is a severe complication of type 2 diabetes (T2D), yet the association between plasma proteins and HF in individuals with T2D remains underexplored. OBJECTIVES: This study aimed to investigate the association between plasma proteomic profiles and HF, and further evaluate whether proteomic data could enhance HF prediction beyond clinical variables, polygenic risk, and N-terminal prohormone of brain natriuretic peptide (NT-proBNP). METHODS: This cohort study included 2198 participants with T2D from the United Kingdom Biobank. Cox proportional hazards models were employed to examine associations between 2920 plasma proteins and incident HF. Clinical and protein predictors were selected using the least absolute shrinkage and selection operator method based on 10-fold cross-validation. The predictive performance of models was evaluated using Harrell's C-index, calibration slope, net reclassification improvement, integrated discrimination improvement, decision curve analysis, and calibration plots. RESULTS: During a median follow-up of 13.1 y, 298 individuals developed incident HF. A total of 455 proteins (447 positively and 8 inversely) were associated with HF, primarily involved in cell adhesion, extracellular space, signaling receptor activity, and cytokine-cytokine receptor interaction pathways. The top protein associated with increased risk of HF was whey acidic protein (WAP) 4-disulfide core domain protein 2, with a per-SD increment hazard ratio (HR) of 1.90 95% confidence interval (CI): 1.65, 2.19). Conversely, the top protein inversely associated with HF risk was apolipoprotein C-I, with a per-SD increment HR of 0.75 (95% CI: 0.66, 0.85). Seventeen proteins were subsequently selected as proteomic predictors, and the resulting 17-protein risk score improved HF prediction beyond clinical variables, polygenic risk, and NT-proBNP, yielding a maximum C-index of 0.833 with an increment of 0.091. CONCLUSIONS: This study identified proteomic biomarkers for HF, highlighted potential pathways that inform biological mechanisms, and demonstrated that proteomic data enhanced HF risk prediction in individuals with T2D.