51-OR: Prognostic Proteomics Modeling in Type 2 Diabetes Reveals Shared and Distinct Mechanisms of Macro- and Microvascular Complications Free

Introduction and Objective: T2D complications manifest across various organs, but are fundamentally rooted in vascular dysfunction, including fibrosis, inflammation, calcification and endothelial damage. Much of complication heterogeneity arises from variation in angiostructures (e.g., larger vessels vs. smaller capillaries) and how they are impacted by elevated glucose levels. This study aims to identify mechanisms underlying macro- and microvascular complications in T2D using a proteomics approach, and to build models that predict future vascular complications.

Methods: We analysed 2,923 plasma proteins in UK Biobank European-ancestry participants with prevalent T2D but no prior vascular diseases at baseline measurement. The primary outcomes were time to first macro- or microvascular complications, identified by ICD-10 codes. We used adaptive LASSO to select proteins in the prediction model, which included key clinical variables. Robustness and internal validation were ensured via stratified bootstrapping.

Results: Among 917 T2D patients, 243 developed macrovascular, 297 microvascular, and 157 both complications over a mean follow-up of 10.69 years. Shared clinical variable predictors included waist-to-hip ratio, glucose, eGFR, and phosphate-calcium homeostasis. Incorporating proteomics significantly improved predictive accuracy compared to only using clinical variables (Harrell’s C: macrovascular 0.747 vs. 0.612; microvascular 0.710 vs. 0.642) by adding 62 proteins for macrovascular (42 in the extracellular region) and 30 for microvascular (11 in the secretory granule lumen). Six proteins (SEPTIN8, IL15, CLEC3B, IDS, FN1, ROBO4) were shared across the micro- and macrovascular prediction models.

Conclusion: Proteomics significantly enhances prediction precision for T2D vascular complications and reveals distinct subcellular compartment localisation for macro- and microvascular prognostic proteins.