关键词: 转移性结直肠癌, 消化道穿孔, 多因素Logistic分析, 预测模型

Abstract: Objective To explore the independent risk factors of metastatic colorectal cancer(mCRC) combined with digestive tract perforation(DTP),and to construct the prediction model.Methods A total of 171 patients with mCRC treated in the hospital were enrolled between April 2018 and October 2021.According to presence or absence of DTP,they were divided into perforation group and non-perforation group.The data of patients were collected,including gender,age,types of primary tumors,history of smoking and drinking,diabetes mellitus,RAS/RAF status,colonoscopy,intestinal obstruction,anemia,diverticulum and usage of vascular endothelial growth factor(VEGF) inhibitors.The risk factors of mCRC combined with DTP were analyzed.The risk prediction model was constructed,and its predictive efficiency was analyzed. Results In the 171 patients with mCRC,there were 33 cases with DTP and 138 cases without,and the incidence of perforation was 19.30%.Colonoscopy,intestinal obstruction and usage of VEGF inhibitors were independent risk factors of mCRC combined with DTP(P＜0.05).The expression of risk prediction model was as follow:P=1/[1+e(-3.374+1.521×(colonoscopy)+1.418×(intestinal obstruction)+1.872×(usage of VEGF inhibitors)].The results of Hosmer-Lemeshow test were as follows:χ2=2.267,P=0.894.AUC and 95%CI of the model for predicting DTP were 0.734 and 0.724-0.859,showing good fit and predictive efficiency.Conclusion Multivariate Logistic regression model can better predict the occurrence of mCRC combined with DTP.Clinically,close attentions can be given for mCRC patients with aged ≥ 60 years,colonoscopy,intestinal obstruction and usage of VEGF inhibitors.

Key words: metastatic colorectal cancer, digestive tract perforation, multivariate Logistic analysis, prediction model