肺癌脑转移患者早期死亡预测模型的构建与验证  

Development and validation of a predictive model for early death in lung cancer patients with brain metastases

作　　者：王朝[1,2,3] 潘宴青 孙云刚[1,2,3] 邵丰 WANG Zhao;PAN Yanqing;SUN Yungang;SHAO Feng(Department of Thoracic Surgery,Nanjing Chest Hospital,Nanjing,Jiangsu210029,China;Department of Thoracic Surgery,Affiliated Nanjing Brain Hospital,Nanjing Medical University,Nanjing,Jiangsu210029,China;Pulmonary Nodule Diagnosis and Treatment Research Center,Nanjing Medical University,Nanjing,Jiangsu210029,China)

机构地区：[1]南京市胸科医院胸外科,江苏南京210029 [2]南京医科大学附属脑科医院胸外科,江苏南京210029 [3]南京医科大学肺部结节诊疗研究中心,江苏南京210029

出　　处：《临床肺科杂志》2024年第11期1698-1705,共8页Journal of Clinical Pulmonary Medicine

摘　　要：目的构建并验证一个模型以预测肺癌脑转移(lung cancer with brain metastases,LCBM)患者确诊后三个月内死亡的风险。方法本研究纳入监测,流行病学和最终结果(Surveillance,Epidemiology and End Results,SEER)数据库内2010年1月至2015年12月期间确诊为LCBM的患者。以是否发生早期死亡为研究终点将患者分为早期死亡组和非早期死亡组。以8∶2为比例将数据分为训练集和验证集。在训练集上采用最小绝对值收缩和筛选算子(least absolute shrinkage and selection operator,LASSO)回归法筛选预测因子,并使用多因素Logistic回归构建预测模型并创建列线图。使用受试者工作特征(receiver operating characteristic,ROC)曲线、校准曲线和临床决策曲线(decision curve analysis,DCA)分别在训练集和验证集上评估模型性能。结果共纳入5035例患者,早期死亡发生率28.3%。LASSO回归筛选出13个变量,Logistic回归最终保留了13个与LCBM患者早期死亡相关的危险因素,包括年龄、从诊断到开始治疗时间、肿瘤大小、肿瘤部位、肿瘤分化程度和组织学类型、T分期、N分期、手术、放疗、化疗、肝转移和骨转移。验证集的一致性指数(concordance index,C-index)为0.84,校准曲线和DCA显示模型具有较好的预测效能和临床净效益。结论基于多因素Logistic回归构建的LCBM患者发生早期死亡的预测模型的区分度较好,能够为临床决策提供一定的帮助。Objective To develop and validate a model to predict the risk of early death(death within 3 months)in patients with lung cancer and brain metastases(LCBM).Methods This study included patients with LCBM from January 2010 to December 2015 in the Surveillance,Epidemiology,and End Results(SEER)database.The patients were divided into an early-death group and a non-early death group according to whether early death occurred.The data is divided into a training set and a validation set in a ratio of 8:2.Least absolute shrinkage and selection operator(LASSO)regression was used to screen predictors on the training set,and multivariate logistic regression was used to construct a prediction model and create a nomogram.The performance of the model was analyzed and evaluated on both the training set and validation set using the concordance index(C-index),the calibration curve,and the decision curve analysis(DCA),respectively.Results A total of 5035 patients were included,and the incidence of early death was 28.3%.LASSO regression screened 13 variables.Logistic regression finally retained 13 risk factors related to the early death of LCBM patients,including age,time from diagnosis to treatment,tumor size,tumor differentiation,histological type,T stage,N stage,surgery,radiotherapy,chemotherapy,liver metastasis,and bone metastasis.The C-index of the validation set was 0.84.The calibration curve and DCA showed that the model had a good predictive performance and clinical net benefit.Conclusion The predictive model of early death in LCBM patients based on multivariate Logistic regression has good discrimination and can provide some help for clinical decision-making.

关 键 词：肺肿瘤 脑转移 早期死亡 预后模型 

分 类 号：R734.2[医药卫生—肿瘤]