非小细胞肺癌纵隔淋巴结转移数学预测模型的建立与验证  

Development and validation of a mathematical prediction model for mediastinal lymph node metastasis in non-small cell lung cancer

作　　者：赵方超 李书军[1] ZHAO Fangchao;LI Shujun(Department of Thoracic Surgery,the Second Hospital of Hebei Medical University,Hebei Shijiazhuang 050051,China.)

机构地区：[1]河北医科大学第二医院胸外科,河北石家庄050051

出　　处：《现代肿瘤医学》2021年第16期2833-2837,共5页Journal of Modern Oncology

摘　　要：目的:分析影响非小细胞肺癌(NSCLC)纵隔淋巴结转移的相关危险因素,构建风险预测模型并验证其预测能力。方法:收集2017年10月-2018年8月唐山市人民医院行手术治疗的159例NSCLC患者病例为建模组,通过单因素和多因素Logistic回归分析筛选出纵隔淋巴结转移的独立危险因素。根据各危险因素权重,构建风险预测模型。通过受试者工作特征(ROC)曲线下面积评估模型的准确性。收集2018年9月-2019年3月的84例NSCLC患者病例为验证组,验证模型的预测效能。结果:多因素Logistic回归分析显示肿瘤部位、分化程度、胸膜侵犯、脉管侵犯是影响NSCLC纵隔淋巴结转移的独立危险因素(P<0.05)。根据Logistic回归分析,可得回归方程:P预测=-1.29+0.141X 1+0.763X 2+0.7442X 3+0.1325X 4,X 1:肿瘤部位(1=中央型,2=周围型)、X 2:分化程度(1=低分化,2=中分化,3=高分化)、X 3:胸膜侵犯(1=有,0=无)、X 4:脉管侵犯(1=有,0=无)。ROC曲线的Youden指数最大值为0.523,截断值为0.678,敏感性为71.3%,特异性为29.1%,曲线下面积为0.80。将验证组患者各因素带入预测模型,检验该模型预测效能,结果发现,ROC曲线下面积为0.86,Hosmer-Lemeshow拟合优度检验显示,χ2=2.45,P=0.96。模型拟合优度好,预测价值高。结论:影响NSCLC纵隔淋巴结转移的高危因素多,临床应进行及时有效评估。本研究构建的预测模型有较好的评估效能,有一定的临床应用价值。Objective:To analyze the risk factors affecting mediastinal lymph node metastasis in non-small cell lung cancer,construct a risk prediction model and verify its predictive ability.Methods:The 159 cases of NSCLC patients who underwent surgical treatment in Tangshan People's Hospital from October 2017 to August 2018 were collected as the modeling group.The independent risk factors of mediastinal lymph node metastasis were screened out by single factor and multivariate logistic regression analysis.According to the weight of risk factors,a risk prediction model is constructed.The accuracy of the model is evaluated by the receiver operating characteristic(ROC)area under the curve.Collect 84 cases of NSCLC patients from September 2018 to March 2019 as the verification group to verify the predictive ability of the model.Results:Multivariate logistic regression analysis showed that tumor location,degree of differentiation,pleural invasion,and vessel invasion were independent risk factors affecting mediastinal lymph node metastasis in NSCLC(P<0.05).According to Logistic regression analysis,the regression equation can be obtained:P prediction=-1.29+0.141X 1+0.763X 2+0.7442X 3+0.1325X 4,X 1:Tumor location(1=central type,2=peripheral type),X 2:Degree of differentiation(1=low differentiation,2=moderately differentiated,3=well differentiated),X 3:Pleural invasion(1=yes,0=no),X 4:Vascular invasion(1=yes,0=no).The maximum Youden index of the ROC curve was 0.523.The cutoff value was 0.678.The sensitivity was 71.3%.The specificity was 29.1%,and the area under the curve was 0.80.The factors in the validation group were brought into the prediction model to test the predictive power of the model.The results showed that the area under the ROC curve was 0.86.The Hosmer-Lemeshow goodness-of-fit test showed thatχ2=2.45,P=0.96.The model had good fit validity and high predictive value.Conclusion:There are many high-risk factors affecting mediastinal lymph node metastasis in non-small cell lung cancer,and clinical evaluation should be carr

关 键 词：非小细胞肺癌 纵隔淋巴结 风险模型 受试者工作曲线 

分 类 号：R734.2[医药卫生—肿瘤]