Glycemic status-dependent proteomic signatures of biological aging for health risk prediction.

Abstract

Existing proteomic aging clocks have been derived from the overall population, with little consideration of extended models tailored to individuals with different glycemic status. We aimed to quantify glycemic status-dependent proteomic signatures of aging and developed proteomic aging scores (ProAS) for health risk prediction. A total of 2923 plasma proteins were measured using Olink in 46,047 UK Biobank participants, including 37,353 with normoglycemia, 5977 with prediabetes, and 2717 with diabetes. Using a three-step screening approach, we identified 11, 23, and 21 representative protein biomarkers associated with all-cause mortality among individuals with normoglycemia, prediabetes, and diabetes, respectively. Three proteins (GDF15, EDA2R, and WFDC2) were shared across all groups, with GDF15 emerging as the top-ranked important protein in normoglycemia and prediabetes and WFDC2 in diabetes. The protein-based ProAS according to glycemic status showed significant associations with diverse health outcomes. Adding the ProAS in the models improved the predictive accuracy of mortality and incident diseases beyond conventional risk factors, but the performance progressively diminished as glycemic status deteriorated. In addition, 72, 51, and 36 out of 102 modifiable factors spanning seven categories were identified as determinants for ProRS in normoglycemia, prediabetes, and diabetes, respectively. Our findings extend the current proteomic clocks by revealing glycemic status-specific aging patterns and their ability to predict age-related outcomes, potentially refining risk stratification and targeted interventions for healthy aging.