基于TM&M分级系统肺癌胸腔镜肺主要手术并发症列线图预测模型的建立和验证

A Nomogram for Prediction of Complications Based on TM&M System of VATS Major Lung Surgery for Lung Cancer

背景与目的并发症是肺切除术后患者死亡的重要原因,目前我国肺癌胸腔镜手术普及率逐年增高,但肺癌胸腔镜手术术后并发症的预测模型尚缺乏基于大样本数据库的支持。本研究采用TM&M(Thoracic Mortality and Morbidity)分级系统全面描绘我中心胸腔镜肺癌手术术后并发症,并建立和验证并发症的预测模型。该模型可为此类患者术后并发症的预防和干预提供依据,从而加速患者康复。方法回顾性收集我中心2007年1月-2018年12月胸腔镜肺癌手术患者临床资料,仅纳入Ⅰ期-Ⅲ期肺癌行胸腔镜肺主要手术的肺癌患者,术后并发症登记严格采用TM&M分级系统。将患者按照手术时期分为两组:前期组(2007年-2012年)和后期组(2013年-2018年),以倾向评分匹配法对两组基线数据进行匹配;匹配后数据采用二元Logistic回归分析建立并发症的预测模型,Bootstrap法内抽样进行内部验证。结果研究共纳入2,881例肺癌患者,平均年龄(61.0±10.1)岁,其中发生主要并发症180例(6.2%)。匹配后的1,268例患者进行二元Logistic回归分析显示:年龄(OR=1.04,95%CI:1.02-1.06,P<0.001)、手术时期(OR=0.62,95%CI:0.49-0.79,P<0.001)、病理类型(OR=1.73,95%CI:1.24-2.41,P=0.001)、术中出血量(OR=1.001,95%CI:1.000-1.003,P=0.03)、清扫淋巴结数目(OR=1.022,95%CI:1.00-1.04,P=0.005)为术后并发症发生的独立危险因素;将其纳入列线图模型,受试者工作特征曲线(receiver operating characteristic curve,ROC)提示该模型区分度较好(C-指数为0.699),1,000次重复Bootstrap内部抽样验证C-指数为0.680。校准曲线显示预测模型的校准度良好。结论TM&M并发症分级系统可全面准确地报告胸腔镜肺癌外科的术后并发症。年龄、手术时期、病理类型、术中出血量、清扫淋巴结数目是胸腔镜肺癌手术后主要并发症的独立危险因素,以此建立的并发症预测模型具有较好的区分度和校准度。

Background and objective Postoperative complications are an important cause of death after lung resection.At present,the adoption of video assisted thoracoscopic surgery(VATS)for lung cancer in China is increasing every year,but the prediction model of postoperative complications of VATS for lung cancer is still lack of evidence based on large sample database.In this study,Thoracic Mortality and Morbidity(TM&M)classification system was used to comprehensively describe the postoperative complications of VATS major lung resection in our center,and the prediction model of complications was established and verified.The model can provide basis for the prevention and intervention of postoperative complications in such patients,and accelerate the recovery of patients.Methods The clinical data of patients underwent VATS major lung resection in our center from January 2007 to December 2018 were collected retrospectively.Only patients with stageⅠ-Ⅲlung cancer were included.The postoperative complications were registered strictly by TM&M classification system.The patients were divided into two groups according to the operation period:the early phase group(From 2007 to 2012)and the late phase group(From 2013 to 2018).The baseline data of the two groups were matched by propensity score matching.After matching,binary logistic regression analysis was used to establish the prediction model of complications,and bootstrap internal sampling was used for internal verification.Results A total of 2,881 patients with lung cancer were included in the study,with an average age of(61.0±10.1)years,including 180 major complications(6.2%).Binary Logistic regression analysis of 1,268 matched patients showed:age(OR=1.04,95%CI:1.02-1.06,P<0.001),other period(OR=0.62,95%CI:0.49-0.79,P<0.001),pathological type(OR=1.73,95%CI:1.24-2.41,P=0.001),blood loss(OR=1.001,95%CI:1.000-1.003,P=0.03),dissected lymph nodes(OR=1.022,95%CI:1.00-1.04,P=0.005)were independent risk factors for postoperative complications.The ROC curve indicates that the model has good discrimination(C-index=0.699),and the C-index is 0.680 verified by bootstrap internal sampling for 1,000 times.The calibration curve shows a good calibration of the prediction model.Conclusion TM&M system can comprehensively and accurately report the postoperative complications of thoracoscopic lung cancer surgery.Age,operative period,pathological type,intraoperative bleeding and dissected lymph nodes were independent risk factors for postoperative complications of VATS major lung resection for lung cancer.The established complication prediction model has good discrimination and calibration.