转移性结直肠癌并消化道穿孔的多因素Logistic分析及预测模型的构建与验证

Multivariate Logisticanalysis of metastatic colorectal cancer combined with digestive tract perforation,construction and validation of the prediction model

目的探讨转移性结直肠癌(mCRC)并消化道穿孔(DTP)的独立危险因素并构建预测模型。方法2018年4月~2021年10月本院就诊的mCRC病人171例,根据是否发生DTP,将病人划分为穿孔组与未穿孔组。收集入院病人的临床资料,包含性别、年龄、原发性肿瘤类型、吸烟饮酒史、糖尿病、RAS/RAF状态、肠镜检查、合并肠梗阻、合并贫血、合并憩室、使用血管内皮生长因子(VEGF)抑制剂。Logistic回归分析影响mCRC并DTP的危险因素,建立风险预测模型,并分析预测效能。结果171例mCRC病人中,33例发生DTP,138例未发生DTP,穿孔率为19.30%;接受肠镜检查、合并肠梗阻、使用VEGF抑制剂是影响mCRC并DTP的独立危险因素(P<0.05);建立风险预测模型的模型表达式为P=1/[1+e(-3.374+1.521×(接受肠镜检查)+1.418×(合并肠梗阻)+1.872×(使用VEGF抑制剂)],采用Hosmer-Lemeshow检验χ^(2)=2.267,P=0.894,模型预测DTP发生的AUC为0.734,95%CI为0.724~0.859,具有良好的拟合度和预测效能。结论多因素Logistic回归模型可以较好的预测mCRC并DTP的发生,临床可重点关注年龄≥60岁、接受肠镜检查、合并肠梗阻、使用VEGF抑制剂的mCRC病人。

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×(intestinalobstruction)+1.872×(usage of VEGF inhibitors)].The results of Hosmer-Lemeshowtest wereas follows:χ^(2)=2.267,P=0.894.AUC and 95%CI of the model for predicting DTP were0.734and0.724-0.859,showing good fitandpredictive 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.