Value of computed tomography evaluation in pathologic classification and prognosis prediction of gastric neuroendocrine tumors

Background: The study aims to investigate the correlation of CT characteristics with pathological classifications and the prognostic value of CT features in patients with gastric neuroendocrine neoplasms (g-NENs). Methods: Ninety-one cases of pathologically diagnosed g-NENs, including 15 cases of well-differentiated neuroendocrine tumors (NETs) (G1 and G2) and 76 cases of poor-differentiated neuroendocrine carcinomas (NECs) (G3 and MANEC) were retrospectively studied. All cases were included in correlation analysis of CT characteristics with pathologic grades. Among them, 76 patients who had fulfilled follow-up data were included for overall survival (OS) and disease-free survival (DFS) analysis. Results: CT characteristics that favor poor differentiation include tumor location (fundus and cardia), larger tumor size (>3.0 cm), infiltrative growth, unclear tumor margin, serosa involvement, ulceration and lymph node metastasis (P<0.05). Most variables had sensitivities >80% and specificities >60% to distinguish NECs from NETs. Through log-rank analysis, it was revealed that serosa involvement, cystic degeneration, necrosis, heterogeneous enhancement and lymph node metastasis led to worse 1)FS and OS for patients with g-NENs (P<0.05). COX regression analysis showed that serosa involvement and lymph node metastasis were independent risk factor for DFS and OS, respectively, despite of grading, staging and therapeutic choices (P<0.05). Moreover, high Ki-67 index (>55%) in G3 g-NENs is in correlation with serosa involvement and lymph node metastasis; accordingly, patients with higher Ki-67 index had worse 1-year DFS (61.7% vs. 92.3%; P<0.05). Conclusions: CT characteristics can be useful discriminators and prognostic factors for g-NENs and may help identify G3 g-NEC from G3 g-NEN by revealing its poor differentiation and high invasive potential.