肝细胞癌患者TACE术后感染的危险因素分析及预测模型构建  

Analysis and prediction model construction of infection after transcatheter arterial chemoembolization in hepatocellular carcinoma patients

作　　者：杜姣 答秀维 张庆 延琰 刘丹 高楠 高敏 张洪新 王苗苗 DU Jiao;DA Xiuwei;ZHANG Qing;YAN Yan;LIU Dan;GAO Nan;GAO Min;ZHANG Hongxin;WANG Miaomiao(Department of Interventional Pain,The Second Affiliated Hospital of the Air Force Medical University,Xi'an 710038,China;Department of Burn and Plastic Surgery,The Second Affiliated Hospital of the Air Force Medical University,Xi'an 710038,China;Research Guarantee Center Information Technology Office,The Second Affiliated Hospital of the Air Force Medical University,Xi'an 710038,China)

机构地区：[1]空军军医大学第二附属医院介入疼痛科,西安710038 [2]空军军医大学第二附属医院烧伤整形科,西安710038 [3]空军军医大学第二附属医院教研保障中心信息技术室,西安710038

出　　处：《医学新知》2024年第12期1345-1356,共12页New Medicine

基　　金：陕西省自然科学基础研究计划(2020JM-327)。

摘　　要：目的探究肝细胞癌(hepatocellular carcinoma,HCC)患者经导管动脉化疗栓塞(transcatheter arterial chemoembolization,TACE)术后感染病原菌的危险因素,并构建预测模型。方法回顾性分析2020年1月1日至2023年12月31日在空军军医大学第二附属医院介入科接受TACE治疗的HCC患者的临床资料。根据TACE术后30 d内是否发生感染,分为感染组和非感染组。根据患者就诊时间分为训练队列和验证队列。使用多因素Logistic回归分析确定独立危险因素,并构建预测模型。采用受试者工作特征(ROC)曲线及其曲线下面积(AUC)、校准曲线分析(CCA)、决策曲线分析(DCA)评估模型效能。结果共纳入592例行TACE的HCC患者,TACE术后感染率为10.30%,以革兰氏阴性菌(55.41%)最为常见。合并糖尿病[OR=12.694,95%CI(4.415,36.497)]、最大病灶直径>5 cm[OR=7.620,95%CI(1.994,29.111)]、腹腔积液[OR=5.106,95%CI(2.226,11.711)]、术中出血量≥500 mL[OR=20.588,95%CI(7.269,58.311)]和手术时间≥120 min[OR=1.284,95%CI(1.093,1.872)]是TACE术后感染的独立危险因素。TCAE术后感染预测模型的AUC值在训练集和验证集中分别为0.907和0.931。CCA显示预测模型“预测概率”和“实际概率”的一致性较好,Brier评分在训练集和验证集分别为0.084和0.075。DCA提示预测模型具备较好的临床净获益。结论合并糖尿病、最大病灶直径>5 cm、腹腔积液、术中出血量≥500 mL和手术时间≥120 min是TACE术后感染发生的独立危险因素。TACE术后感染预测模型能较好的识别HCC患者TACE感染的发生情况,是一种早期识别TACE术后感染的有效工具。Objective To analyze the risk factors of infectious in hepatocellular carcinoma(HCC)patients after transcatheter arterial chemoembolization(TACE),and construct a prediction model.Methods The complete clinical data of HCC patients who received TACE treatment in the department of Interventional Pain of the Second Affiliated Hospital of the Air Force Medical University from January 1,2020 to December 31,2023 were retrospectively analyzed.According to whether infection occurred within 30 days after TACE,the patients were divided into the infection group and the non-infection group.According to the visiting time of HCC patients undergoing TACE,they were divided into the training and validation cohorts.Multivariate Logistic regression analysis was used to determine independent risk factors and construct a prediction model.The area under the curve(AUC)of the receiver operating characteristic curve,calibration curve analysis(CCA)and the Brier score,decision curves analysis(DCA)were used to evaluate the performence of the prediction model.Results A total of 592 HCC patients with TACE were included,the infection rate after TACE was 10.30%,with Gram-negative bacteria was most common infection(55.41%).Comorbid diabetes[OR=12.694,95%CI(4.415,36.497)],maximum lesion diameter>5 cm[OR=7.620,95%CI(1.994,29.111)],ascites[OR=5.106,95%CI(2.226,11.711)],intraoperative blood loss≥500 mL[OR=20.588,95%CI(7.269,58.311)],and operation time≥120 minutes[OR=1.284,95%CI(1.093,1.872)]were independent risk factors for post-TACE infection.The AUC of the prediction model in the training set and the validation set were 0.907 and 0.931.CCA showed good consistency of the"predicted probability"and the"actual probability"of the post-TACE infection prediction model,and the Brier scores in the training set and the validation set were 0.084 and 0.075.The DCA curve suggested that the prediction model provided good clinical net benefit.Conclusion Comorbid diabetes,maximum lesion diameter>5 cm,ascites,intraoperative blood loss≥500 mL,and operation tim

关 键 词：肝细胞癌 经导管动脉化疗栓塞 病原菌 感染 预测模型 

分 类 号：R735.7[医药卫生—肿瘤] R63[医药卫生—临床医学]