Development and validation of a postoperative prognostic nomogram for early-stage pancreatic cancer

Our study aims to develop and validate individualized risk prediction models for all-cause mortality (ACM) and cancer-specific mortality (CSM) following surgery for early-stage pancreatic cancer (ESPC). Patients with pathologically confirmed stage I–II pancreatic cancer (PC) who underwent curative surgery between 2010 and 2019 were identified from the Surveillance, Epidemiology, and End Results (SEER) database. To identify the optimal predictor, the integrated discrimination improvement (IDI) and net reclassification improvement (NRI) indices were calculated to compare the predictive performance of the log odds of positive lymph nodes (LODDS) with that of other lymph node (LN)-related measures. LODDS demonstrated superior incremental discriminative ability for predicting ACM and exhibited a nonlinear association with ACM. The Cox regression-based nomogram for predicting ACM yielded area under the curve (AUC) values of 0.751 (95% CI 0.732–0.769), 0.799 (95% CI 0.784–0.813), and 0.824 (95% CI 0.808–0.840) for 1-, 3-, and 5-year survival, respectively, in the internal validation cohort. The competing risk model-based nomogram for predicting CSM achieved AUC values of 0.746 (95% CI 0.728–0.765), 0.808 (95% CI 0.794–0.823), and 0.832 (95% CI 0.816–0.847) for 1-, 3-, and 5-year CSM, respectively, in the validation cohort. The nomograms developed using Cox regression and competing risk models demonstrated favorable predictive performance based on internal validation. However, these findings should be interpreted with caution, as the nomograms remain preliminary prediction tools that require further validation. In the future, we will continue to incorporate additional external validation data to further validate and update the model.