Modified Histopathological Grading Optimizes Prediction of Survival Outcomes in Small Intestinal Neuroendocrine Tumors

Abstract

Abstract Context One of the major prognostic indices in neuroendocrine tumors (NETs) is Ki67 proliferation index. Objective To identify optimal grading Ki67 cutoffs to delineate differences in prognosis of patients with small intestinal NETs (SI-NETs). Methods Multicenter retrospective cohort analysis of 551 SI-NET patients diagnosed from 1993 through 2021 at 5 European referral centers with a mean (±SD) follow-up time of 51.5 (±52.9) months, measuring rates of overall survival (OS) and event-free survival (EFS). Results Median age at baseline was 62.3 (range, 17-90) years; 252 (45.7%) patients were female. All SI-NETs were well-differentiated, with 326 being grade 1 (G1; 59.2%), 169 G2 (30.7%), and 8 G3 (1.5), while 48 tumors were unspecified grade (8.7%). The median Ki67 was 2% (range, 1%-70%). At baseline, 247 (44.8%) patients had distant metastases (stage IV), 217 locoregional disease (41.1%; stage III), while 29 (7.1%) and 25 (4.5%) presented at stages II and I, respectively. Median OS was 214.7 (95% CI, 152.7-276.6) months and median EFS was 79.8 (68.2-91.5) months. In multivariable Cox-regression OS analysis, the proposed modified histopathological Ki67 grading system (Ki67 5%-10% group: HR = 2.2 [95% CI, 1.15-4.31], P = .018 and Ki67 ≥ 10% group: HR = 5.11 [2.87-9.09], P < .001), age (HR = 1.07 [1.04-1.09], P < .001), Charlson Comorbidity Index (HR = 1.08 [1-1.16], P = .028), and TNM stage (HR = 1.79 [1.05-3.06], P = .034) were independent predictors for death. Pertinent EFS analysis confirmed the proposed modified histopathological Ki67 grading system (Ki67 ≥ 10% group: HR = 4.01 [2.6-6.37], P < .001) and age (HR = 1.04 [1.02-1.05], P < .001) as independent predictors for recurrence, progression, and/or death. Conclusion Ki67 proliferation index was a strong and independent predictor of OS and EFS. A modified histopathological grading system applying Ki67 cutoffs of 5% and 10% could be superior to predict differences in SI-NET patient survival outcomes.