Abstract
BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) has a high fatality rate, with surgery as the only curative treatment. Identification of new biomarkers related to survival may help guide discovery of new pathophysiologic pathways and potential therapeutic targets. As long-chain ceramides have been linked to tumor proliferation, we sought to determine if ceramide levels were prognostic in PDAC. METHODS: Patients from two phase I studies of PDAC were followed for all-cause mortality. Ceramide levels (C24:0, C22:0, and C16:0) were quantified before treatment and at study intervals. Multivariable Cox regression models assessed the association of ceramide levels and mortality after adjusting for other univariable predictors, including time-dependent tumor resection. The ability of repeated ceramide measures to discriminate patients at risk for mortality was also assessed using multivariable modeling and the c-statistic. RESULTS: Higher plasma C16:0 concentration was associated with higher all-cause mortality in univariable and multivariable analysis (adjusted hazard ratio [aHR] 1.41, 95% confidence interval [CI] 1.09-1.82; p < 0.01). In contrast, a higher plasma C24:0/C16:0 ratio was associated with lower all-cause mortality in multivariable analysis (aHR 0.69, 95% CI 0.49-0.97; p = 0.032). Discrimination of mortality was significantly improved with the addition of either plasma C16:0 or C24:0/C16:0 levels, with optimal discrimination occurring using repeated measures of the C24:0/C16:0 ratio (c-statistic 0.73 vs. c-statistic 0.66; p < 0.001). CONCLUSIONS: Higher plasma C16:0 and lower C24:0/C16:0 ratios are independently associated with mortality in PDAC and show an ability to improve discrimination of mortality in this deadly disease. Further studies are needed to confirm this association and evaluate this novel pathway for potential therapeutic targets.