A Pathology-Trained Lipid Aging Clock Reveals Accelerated Aging and Prognostic Sphingolipid Signatures in Pancreatic Disease

Altered lipid metabolism is increasingly recognized as a feature of aging and age-related disease, including cancer. Here, we present a lipid-based biological aging clock developed from serum lipidomics data of 443 patients with pancreatic ductal adenocarcinoma (PDAC). Using penalized Cox proportional hazards models, we derived a risk-equivalent Lipid Age and Lipid Age Acceleration (LAA) metric and evaluated its relationship to survival, healthy aging, and pancreatitis. Despite being trained exclusively on a pathological PDAC cohort, the lipid clock predicted chronological age in healthy individuals with high accuracy (Pearson r=0.86; median absolute error = 3.46 years), indicating that pathological cohorts can retain biologically meaningful aging signals. Both PDAC patients and, in exploratory analyses, pancreatitis patients exhibited accelerated lipid aging relative to healthy individuals, with mean LAA values of 5.57 and 2.71 years, respectively. Increased continuous LAA was significantly associated with worse overall and progression-free survival and remained predictive of overall mortality even after adjusting for standard clinical cancer measures, staging, and the conventional tumor marker CA19-9 (HR=1.09 per 1-year increase, p<0.005). Specific individual lipid species, particularly sphingolipids, carried independent prognostic value. Lipid age acceleration was associated with pancreatic disease and with lipidomic patterns consistent with systemic aging-associated and inflammatory remodeling independent of traditional oncological risk parameters. Because the model was developed and evaluated within a single parent dataset and analytical platform, external validation in independent cohorts will be needed to establish broader generalizability.