Patterns and predictors of pancreatic neuroendocrine tumor prognosis: Are no two leaves alike?

Pancreatic neuroendocrine tumors (PanNETs) are heterogeneous; thus, individual prognostic prediction is important. Clinicopathological features, like TNM stage, grade, and differentiation, are independent clinical predictors. However, single predictors are insufficient, as patients sharing similar clinicopathological features usually show distinct prognoses. Accordingly, novel nomograms and risk stratifications have been developed for more accurate PanNET prognostic prediction. Moreover, the exploration of molecular mechanisms has identified novel prognostic predictors for PanNET. Multi-analyte assays of molecular biomarkers provide a deeper understanding of PanNET features; however, the priority, and the optimal combination of classic and novel predictors for PanNET prognosis prediction remain unclear. In this review, we summarized the patterns and predictors of PanNET prognosis and discussed their clinical utility; we emphasized that PanNET at different stages have different superior predictor, and that multi-analyte assays are more sensitive than mono-analyte biomarkers. Therefore, combined biomarkers improve the accuracy of surveillance and optimize decision-making in clinical practice.