Analysis of risk factors and prognostic factors for gastrointestinal stromal tumors with gastrointestinal hemorrhage: Based on propensity score matching method.

BACKGROUND: This study aimed to analyze the relationship between risk factors and prognosis of patients with gastrointestinal stromal tumor associated with gastrointestinal bleeding.

METHODS: According to whether there was gastrointestinal bleeding, 246 patients with gastrointestinal stromal tumors were divided into 2 groups. The clinicopathological baseline characteristics of the 2 groups of patients were balanced by propensity score matching, and the Kaplan-Meier method was used to draw the survival curve and analyze the overall survival of the 2 groups of patients. The receiver operating characteristic curve was drawn to evaluate the accuracy of Modified National Institutes of Health criteria and Armed Forces Institute of Pathology criteria in predicting the prognosis and postoperative recurrence of patients. Logistic regression analysis of risk factors affecting gastrointestinal stromal tumor with gastrointestinal bleeding before matching. Univariate and multivariate analyses of risk factors affecting the prognosis of patients with gastrointestinal stromal tumors after matching were performed using Cox regression models.

RESULTS: Before matching, the accuracy of Modified National Institutes of Health criteria in predicting postoperative survival status and recurrence was higher than that of Armed Forces Institute of Pathology criteria. Modified National Institutes of Health criteria and relapse were the risk factors for gastrointestinal stromal tumor with gastrointestinal bleeding independent risk factors (P < .05). After 1:1 matching, the general clinical data of patients with gastrointestinal bleeding group and nongastrointestinal bleeding group were balanced (P > .05). The results of matched survival analysis indicated that tumor location and gastrointestinal bleeding were independent risk factors for the prognosis of patients with gastrointestinal stromal tumors (P < .05). The results of subgroup analysis according to anatomical site showed that there was no significant difference between the gastrointestinal bleeding group and the nongastrointestinal bleeding group (P > .05). Survival analysis showed that patients with gastrointestinal stromal tumors with gastrointestinal bleeding had a worse prognosis, and the results were also applicable in different tumor anatomical locations and different Modified National Institutes of Health criteria.

CONCLUSION: Modified National Institutes of Health criteria and relapse are independent risk factors for gastrointestinal stromal tumors with gastrointestinal bleeding; gastrointestinal bleeding is associated with poor prognosis in patients with gastrointestinal stromal tumors, and patients with gastrointestinal stromal tumors with gastrointestinal bleeding have a worse prognosis.