Redefining the Ki-67 Index Stratification for Low-Grade Pancreatic Neuroendocrine Tumors: Improving Its Prognostic Value for Recurrence of Disease

The Ki-67 index is an established prognostic marker for recurrence after resection of pancreatic neuroendocrine tumors (PanNETs) that groups tumors into three categories: low grade (< 3%), intermediate grade (3-20%), and high grade (> 20%). Given that the majority of resected PanNETs have a Ki-67 less than 3%, this study aimed to stratify this group further to predict disease recurrence more accurately. The Ki-67 index was pathologically re-reviewed and scored by a pathologist blinded to all other clinicopathologic variables using tissue microarray blocks made in triplicate. All patients who underwent curative-intent resection of non-metastatic PanNETs at a single institution from 2000 to 2013 were included in the study. The primary outcome was recurrence-free survival (RFS). Of 113 patients with well-differentiated PanNETs resected, 83 had tissue available for pathologic re-review. The Ki-67 index was lower than 3% for 72 tumors (87%) and between 3 and 20% for 11 tumors (13%). Considering only Ki-67 less than 3%, the tumors were further stratified by Ki-67 into three groups: group A (< 1%, n = 43), group B (1-1.99%, n = 23), and group C (2-2.99%, n = 6). Compared with group A, groups B and C more frequently had advanced T stage (T3: 44% and 67% vs 12%; p = 0.003) and lymphovascular invasion (50% and 83% vs 23%; p = 0.007). Groups B and C had similar 1- and 3-year RFS, both less than group A. After combining groups B and C, a Ki-67 of 1-2.99% was associated with decreased RFS compared with group A (< 1%). This persisted in the multivariable analysis (hazard ratio [HR] 8.6; 95% confidence interval [CI] 1.0-70.7; p = 0.045), with control used for tumor size, margin-positivity, lymph node involvement, and advanced T stage. PanNETs with a Ki-67 of 1-2.99% exhibit distinct biologic behavior and earlier disease recurrence than those with a Ki-67 lower than 1%. This new stratification scheme, if externally validated, should be incorporated into future grading systems to guide both surveillance protocols and treatment strategies.