Olink proteomics identifies serum protein biomarkers and an age-stratified diagnostic model for diabetes in Chinese patients
Background Aging is a critical risk factor for the progression and complications of type 2 diabetes mellitus (T2DM). However, routine clinical indicators fail to accurately quantify biological senescence burden in T2DM patients. This study aimed to screen plasma protein signatures associated with metabolic senescence in T2DM using Olink targeted proteomics and to construct a novel biological aging evaluation model for diabetic populations. Methods A total of 21 healthy controls and 66 T2DM patients were enrolled. Plasma protein profiles were detected via Olink proteomics. Differentially expressed proteins (DEPs) were identified and subjected to GO and KEGG functional enrichment analyses. Random forest and LASSO regression analyses were applied to screen core T2DM-related protein molecules. T2DM patients were further stratified by a 55-year cutoff, a well-recognized critical threshold for metabolic senescence. After adjustment for sex, BMI, glycated hemoglobin (HbA1C), and hypoglycemic medication use, age-independent senescence-related proteins were identified to establish a multi-protein predictive model. A 1000-time Bootstrap resampling procedure was performed for internal validation to evaluate model discrimination and stability. Results A total of 84 DEPs (P < 0.05) were identified between T2DM patients and healthy controls, mainly enriched in cytokine–cytokine receptor interaction, lipid metabolism, and atherosclerosis pathways. Machine learning screened six core proteins with high discriminatory value for T2DM, including MERTK, BOC, TNFRSF10A, CCL3, AGRP, and PD-L2. Within the T2DM cohort, nine age-dependent plasma proteins were independently identified (FDR < 0.05), among which VSIG2, LPL, S100A11, and CST5 exhibited the most robust correlations (FDR < 0.01). A three-protein model comprising VSIG2, S100A11, and S100A5 achieved favorable discriminative performance (AUC = 0.874), which was superior to conventional clinical indicators including BMI (AUC = 0.572) and HbA1c (AUC = 0.593). Bootstrap validation confirmed reliable model stability with a correction optimism of −0.024, a bias-corrected AUC of 0.900, and an overfitting degree of −2.65%. Functional analyses indicated that candidate proteins were primarily involved in immune homeostasis, lipid remodeling, inflammatory responses, and calcium signaling. Conclusion This study identified novel T2DM-associated plasma biomarkers and established a robust three-protein signature (VSIG2, S100A11, S100A5) for age stratification and biological senescence evaluation in T2DM. The proposed model compensates for the limitations of routine clinical indices, providing a promising non-invasive serological tool for precise risk stratification and individualized intervention in elderly diabetic patients.