BACKGROUND: Pancreatic neuroendocrine tumors (PNETs) display wide heterogeneity with highly variable prognosis. This study aimed to identify variables related to survival after surgical resection of PNET.

METHODS: A total of 143 patients were identified from a prospectively maintained database. Patient characteristics were analyzed and prognostic factors for overall survival and progression-free survival were evaluated. The WHO, ENETS and AJCC scoring systems were applied to the cohort, and their ability to predict patient outcomes were compared.

RESULTS: Multivariate analysis found that female gender, lymph node metastases and increasing WHO 2010 grade to be independently associated with reduced overall survival (P?<?0.05). Patients requiring multi-visceral resection or debulking surgery found to be associated with shortest survival. ROC analysis found the ENETS and AJCC scoring systems to be similarly predictive of 5-year overall survival. Modified Ki67 significantly improved its accuracy in predicting 5-year overall survival (AUROC: 0.699 vs 0.605; P?<?0.01).

CONCLUSIONS: Multi-visceral or debulking surgery is associated with poor outcomes. There seems to be no significant difference between enucleation and anatomical segmental resection. Available scoring systems have reasonable accuracy in stratifying disease severity, with no system identified as being superior. Prognostic stratification with modified grading systems needs further validation before applied in clinical practice.