Chronic kidney disease onset, progression, and cardiovascular outcomes: proteomics informs biology and risk stratification

BACKGROUND: Large-scale proteomics provides an opportunity to understand chronic kidney disease (CKD) and cardiovascular disease, yet research in this field is limited. This study utilized proteomics to inform biology and risk stratification for these diseases. METHODS: This cohort study included 44,779 participants free of prevalent CKD, and 3,749-4,272 participants with prevalent CKD from the UK Biobank. The Olink Explore 3072 platform quantified 2,923 plasma proteins. Cox proportional hazards models were used to assess associations of proteins with kidney diseases including CKD and end stage kidney disease, and cardiovascular diseases including coronary heart disease (CHD), stroke, and heart failure (HF). Mendelian randomization examined genetic associations, pathway analyses identified biological pathways, and predictive models were developed for incident diseases. RESULTS: Median follow-up periods were 12.2-12.6 years. We identified 598 (20.5%) proteins shared across ≥ 2 diseases, with 595 (20.4%) showing consistent directions of associations, and 471 (16.1%) unique to a single disease. CKD and HF specifically shared the largest number of 279 (9.6%) proteins. POLR2F, TNFRSF10B, and IGFBP2 were positively associated with all five diseases, with Mendelian randomization supporting genetic associations of POLR2F with CHD and IGFBP2 with hypertensive renal disease. Pathway analyses highlighted cell adhesion, signal transduction, and cytokine-cytokine receptor interaction for disease-associated proteins. Incorporating predictive proteins into clinical models improved risk prediction for CKD, CHD, stroke, and HF, yielding Harrell's C indices of 0.750-0.818 (corresponding increases of 0.027-0.090). CONCLUSIONS: This study deepens insights into disease biology and provides a foundation for early detection and integrated risk stratification in CKD and cardiovascular disease.