Metabolic signatures of lipid dysregulation and inflammation characterize pancreatic cancer risk in type 2 diabetes mellitus and improve risk stratification: a prospective cohort study.

[BACKGROUND] Type 2 diabetes mellitus (T2DM) increases pancreatic cancer (PC) risk, but specific metabolic mechanisms and effective risk stratification tools remain lacking. We aimed to identify metabolic features statistically accounting for this risk and construct a metabolic risk score (MRS).

[METHODS] We analyzed 445,931 UK Biobank participants with NMR metabolomics, excluding early PC cases to minimize reverse causality. Cox models estimated associations and calculated the proportion of the T2DM-PC association statistically accounted for by metabolites. LASSO regression constructed the MRS within the T2DM sub-cohort.

[RESULTS] Over a 13.6-year follow-up, T2DM increased PC risk (HR 1.40, 95% CI 1.18-1.67). We identified 30 candidate metabolites statistically accounting for a significant proportion of this association, primarily reflecting lipid dysregulation (triglyceride-enriched VLDL) and inflammation (GlycA). The derived MRS stratified long-term PC risk (highest vs. lowest tertile HR 1.79). Integrating the MRS with clinical factors provided incremental improvement in 15-year prediction, increasing the AUC from 72.3% to 75.8% (P < 0.001).

[CONCLUSION] Lipid dysregulation and inflammation statistically account for a substantial proportion of the T2DM-PC association. The MRS captures these molecular alterations, serving as a preliminary tool to improve long-term PC risk stratification, which warrants rigorous external validation before clinical application.