Prediagnostic plasma proteomic signatures associated with incident chronic kidney disease in type 2 diabetes

Chronic kidney disease (CKD) is a major complication of type 2 diabetes (T2D), but the circulating protein state that precedes clinically recognised CKD remains incompletely defined. We aimed to determine whether baseline plasma proteomics identifies a biologically interpretable prediagnostic protein signature associated with incident CKD in adults with T2D who were CKD-free at baseline, and to explore its incremental predictive information beyond clinical variables. We analysed 1715 adults with type 2 diabetes who were free of chronic kidney disease at baseline and had plasma proteomic profiling in the UK Biobank Pharma Proteomics Project. During a median 14.6-year follow-up, Cox models evaluated 2920 baseline plasma proteins for incident chronic kidney disease. Baseline protein patterns were compared across future chronic kidney disease timing groups. Functional enrichment and protein–protein interaction network analyses characterised biological pathways. For prediction modelling, training-set Cox screening and least absolute shrinkage and selection operator regression selected candidate proteins for Cox models as selected proteins or a protein risk score. Exploratory Mendelian randomisation, colocalisation and translational annotation prioritised candidate proteins. During follow-up, 262 participants developed incident chronic kidney disease, and 449 proteins were associated with chronic kidney disease risk after false discovery rate correction. Chronic kidney disease-associated proteins showed graded baseline abundance patterns by future chronic kidney disease timing and were enriched in immune activation, receptor signalling, cytokine-related and extracellular matrix pathways, with network hubs including TNF, EGFR and IL10. Adding selected proteins or a protein risk score to clinical variables yielded numerically higher C-index from 0.715 to 0.732 and 0.741, respectively, compared with 0.715 for the clinical model, although incremental differences were statistically uncertain. Exploratory prioritisation integrated association, prediction, genetic and translational evidence to identify proteins for further investigation. Baseline plasma proteomics revealed a biologically organised prediagnostic signature associated with future chronic kidney disease among adults with type 2 diabetes who were free of CKD at baseline. Integrated biological, predictive, genetic and translational evidence further prioritised candidate proteins, supporting continued investigation of plasma proteomics as a complementary source of information for earlier CKD risk assessment.