Objective To investigate the impact of prior non-pancreatic cancer on the survival outcomes of patients with localized pancreatic neuroendocrine tumors (PanNETs).
Methods We reviewed the Surveillance, Epidemiology, and End Results database and selected patients with localized PanNETs diagnosed between 1973 and 2015. We divided the patients into two groups according to the presence or absence of prior non-pancreatic malignancy. Before and after propensity score matching, we compared the clinicopathological characteristics and studied the overall survival and cancer-specific survival.
Results A total of 357 (12.9%) of 2778 patients with localized PanNETs had prior cancer. A total of 1211 cases with only a localized PanNET and 133 cases with a localized PanNET and prior cancer had complete data and met the inclusion criteria of the current study. Patients with prior cancer were associated with advanced age (>65 years, 57.9% prior cancer vs. 31.0% no prior cancer, P<0.001), later year of diagnosis (87.2% vs. 80.2%, P=0.049), a higher proportion of poorly differentiated/undifferentiated grade tumors (4.5% vs. 1.5%, P=0.025), and a higher proportion of no primary site surgery (19.5% vs. 10.4%, P=0.003). Prostate (29.32%), breast (18.05%), other genitourinary and retroperitoneal (16.54%), and gastrointestinal (12.78%) cancers were the most common prior cancer types. Most of the prior cancers (95.49%) were localized and regional, and only 4.51% of the prior cancers were distant. Patients with interval periods between the prior cancer and PanNET of ≤36 months, 36-60 months, 60-120 months, and >120 months accounted for 33.08%, 13.53%, 24.06%, and 29.32% of all cases with prior cancers, respectively. Univariate and multivariate Cox proportional hazards analyses were performed. The presence/absence of prior cancers did not impact survival outcomes of patients with localized PanNETs before and after propensity score matching (PSM). Further subgroups analysis showed that, patients with localized PanNETs and prior distant cancer had worse cancer-specific survival than patients with prior local/regional cancer or patients without prior cancer (P<0.001). No significant differences in cancer-specific survival were observed in terms of the different sites of the prior cancers and the different interval periods of prior cancers and PanNETs (P<0.05).
Conclusions Patients with localized PanNETs and a history of prior cancer had survival outcomes that were comparable to those of patients with no history of prior cancer. Patients with localized PanNETs and prior cancer could be candidates for clinical trials if they satisfy all other conditions; aggressive and potentially curative therapies should be offered to these patients.

Key words: pancreatic neuroendocrine tumor, prior cancer, propensity score matching