Grading Using Ki-67 Index and Mitotic Rate Increases the Prognostic Accuracy of Pancreatic Neuroendocrine Tumors

OBJECTIVES: To measure the usefulness of Ki-67 proliferative index (Ki-67 index) as a prognostic variable for grading pancreatic neuroendocrine tumors. METHODS: A multi-institutional prospective database comprising 350 patients. Grading based on mitotic activity (<2 mitoses/10 high-power fields, 2-20 and >20) and Ki-67 index (<3% per 10 high-power fields, 3%-20% and >20%). Final grade selected based on higher grade of either variable. RESULTS: Most patients were in the less than 3% (n = 158) and 3% to 20% Ki-67 category (n = 107), with a minority being high-grade (Ki-67 > 20%, n = 27). Discordance between Ki-67 and mitotic rate was noted in 58 patients. On multivariate analysis, final-grade (grade 2: P = 0.010, hazard ratio [HR], 1.2; grade 3: P = 0.002; HR, 2.8), Ki-67, mitotic rate, and lymph node status were significant prognostic markers for overall survival (OS). For disease-free survival (DFS), only final-grade (grade 2: P = 0.05; HR, 1.4; grade 3: P = 0.009; HR, 2.3), Ki-67, mitotic rate, and margin status significantly predicted DFS. Ki-67 was a better model for OS and mitotic rate for DFS. Overall combined final grade was the best model based on HR. CONCLUSION: Ki-67 is a strong prognostic factor for OS and DFS and should be included in all pancreatic neuroendocrine tumor pathology.