Clinical relevance of different WHO grade 3 pancreatic neuroendocrine neoplasms based on morphology

PURPOSE: Emerging evidence suggests G3 pancreatic neuroendocrine neoplasms (pNENs) present heterogeneous morphology and biology. The 2017 WHO classification has introduced a new category of well-differentiated pancreatic neuroendocrine tumors (WD-pNETs) G3, compared with poorly differentiated pancreatic neuroendocrine carcinomas (PD-pNECs) G3. We aim to analysis the demographics and outcomes of patients with resectable 2017 WHO G3 pNENs to facilitate the distinction between two entities. METHODS: The multi-institutional retrospective cohort involving 57 surgically treated patients affected by 2017 WHO G3 pNENs were morphologically identified and clinically analyzed. Patients having WD-pNETs G3 and those having PD-pNECs G3 were compared. RESULTS: Thirty patients had WD-pNETs and 27 patients had PD-pNECs. The distributions of Ki-67 and mitotic count in patients with PD-pNECs or WD-pNETs showed remarkable disparities. ROC indicated cut-off value of Ki-67 was 45. PD-pNECs were more common in patients with elevated Ki-67 and mitotic count, advanced AJCC TNM stage, vascular invasion, regional lymph-node metastases, elevated NSE and decreased CgA levels compared with WD-pNETs (P < 0.05). The association between 2017 WHO G3 grade and TTR was statistically significant (P < 0.05). Univariate analysis indicated OS rates were associated with morphologic differentiation (WD-pNETs vs PD-pNECs), Ki-67, TNM staging, synchronous distant metastases, initial treatments, vascular invasion, regional lymph nodes metastases, mitotic count and age (P < 0.05). Multivariate analyses illustrated Ki-67, differentiation, TNM staging and vascular invasion were independent predictors (P < 0.05). CONCLUSIONS: PD-pNECs G3 presented malignant biological behavior and dismal outcome compared with WD-pNETs G3. These findings challenge 2010 WHO classification and suggest the categorization can be improved by refined tumor grading.