Abstract 4653
Background
Pancreatic cancer (PaC) remains extremely lethal worldwide even after resection. This large international population-based study aimed at exploring factors associated with survival in resected PaC, and at developing and globally validating a survival-predicting nomogram.
Methods
Data of PaC patients resected in 2003-2014 were obtained from multiple European national cancer registries and the US SEER-18 Program. Multivariable Cox proportional hazards models were constructed to investigate the associations of patient and tumor characteristics with overall survival. Prognostic factors remaining after backward selection in SEER-18 were used to build a nomogram, which was subjected to bootstrap internal validation and external validation using the European databases. Predictive accuracy was assessed using the concordance-index.
Results
Totally 24,863 resected PaC patients were included, with median survival of 12-19 months and 3-year survival rates of 14%-28%. In main analysis, patient age, tumor T, N, and M stages, histology, and differentiation were significantly associated with survival, with country-specific association patterns and strengths. Additionally, hospital type, tumor size, harvested lymph node number, performance status, and certain comorbidities were associated with survival in countries with available information. A nomogram incorporating the backward-selected variables in the main analysis was established. Calibration curves showed good agreement between nomogram-prediction and actual observation. The concordance-index of the nomogram was significantly higher than that of the TNM staging for predicting survival.
Conclusions
In these international population-based cohorts, resected PaC patients have distinct characteristics independently associated with survival. A personalized postoperative survival-predicting nomogram is established and internationally validated, which would be practical and helpful clinically and aid to patient stratification in international studies.
Clinical trial identification
Legal entity responsible for the study
Division of Clinical Epidemiology and Aging Research, German Cancer Research Center (DKFZ).
Funding
German Cancer Aid (Deutsches Krebshilfe).
Editorial Acknowledgement
NA
Disclosure
All authors have declared no conflicts of interest.