Abstract
PURPOSE: To identify hidden risk factors for recurrence in patients with very low- and low-risk jejunoileal gastrointestinal stromal tumors (GISTs) in order to improve patient risk stratification. METHODS: This retrospective study included 123 patients with very low- or low-risk jejunoileal GISTs who underwent R0 resection at Zhongshan Hospital of Fudan University between 2011 and 2021. Clinical and pathological variables, including tumor location, size, mitotic index, Ki-67 index, and mutation status, were analyzed. Recurrence-free survival (RFS) and overall survival (OS) were assessed using Kaplan-Meier and Cox regression analyses. Optimal cut-off values were determined using receiver-operating characteristic (ROC) curves, based on which a prognostic nomogram was constructed. RESULTS: The median age of the 123 patients with very low- or low-risk jejunoileal GISTs was 57 years, with a moderate male predominance (60.2%). Tumors were located in the jejunum in 61.8% of the patients, while the remaining 38.2% of the cases were in the ileum. The median tumor size was 3.5 cm, and the median Ki-67 index was 4.89%. During the median 70-month follow-up period, 19 patients developed recurrence or metastasis, and 11 patients died due to disease progression. Jejunal GISTs were more prevalent (61.8%) and showed a significantly higher recurrence rate compared to ileal tumors (21.1% vs. 6.4%). ROC curve analysis determined optimal cut-off values of 3.9 cm for tumor size and 3.5% for Ki-67 index for predicting recurrence. Multivariate Cox regression analysis revealed that tumor size > 3.9 cm, mitotic index > 2/50 HPF, Ki-67 > 3.5%, and jejunal location independently predicted worse RFS, while female sex, tumor size > 3.9 cm, and Ki-67 > 3.5% were associated with worse OS. CONCLUSION: Jejunoileal GISTs have a significant risk of recurrence despite their low-risk NIH classification. Tumor size, mitotic count, Ki-67 index, and anatomical location should be incorporated into refined risk models to guide follow-up and treatment decisions.