基于血清肿瘤标志物和影像学的肺腺癌患者EGFR突变概率临床预测模型的构建和验证

目的:利用肿瘤标志物、凝血功能指标与影像学相关指标建立预测肺腺癌患者EGFR突变状态的临床模型并进行预测价值验证。方法:随机选取2018年至2019年沧州市中心医院肿瘤科收治的NSCLC患者225例为建模组研究对象。通过多因素Logistic 回归分析筛选出NSCLC患者EGFR突变的独立影响因素，并建立回归模型，绘制回归模型预测NSCLC患者EGFR突变概率的受试者工作特征（ROC）曲线。选取2020年至2021年沧州市中心医院肿瘤科收治的肺腺癌患者75例为验证组，对预测模型进行内部验证。结果:多因素Logistic 回归分析结果显示，磨玻璃密度影、充气支气管征、血管集束征、CEA是预测NSCLC患者发生EGFR突变的独立影响因素（P＜0.05）。预测模型表达式为:P=eX /（1+eX），其中X =-2.196+（0.036×CEA）+（0.984×肺部磨玻璃阴影）+（1.923×支气管充气征）+（1.206×血管集束征）。预测模型曲线下面积为0.824［95%CI（0.767，0.881）］，敏感度和特异性分别为78.6%和77.5%。将验证组75例患者临床资料带入预测模型，ROC曲线下面积为0.762，H-L拟合优度检验显示，χ2 =3.140，P=0.872。结论:影响肺腺癌患者EGFR突变的危险因素较多，临床要进行及时有效评估。本研究建立的预测模型具有较高的评估效能，值得临床推广。

Establishment and evaluation of EGFR mutation prediction model based on serum tumor markers and medical imaging in patients with lung adenocarcinoma

Objective:To establish a clinical model for predicting EGFR mutation status in patients with lung adenocarcinoma using tumor markers,coagulation function indicators and imaging related indicators,and to verify the predictive ability.Methods:A total of 225 non-small cell lung cancer (NSCLC) patients admitted to the oncology department of Cangzhou Central Hospital from 2018 to 2019 were randomly selected as the research object of the modeling group.The independent factors that affected EGFR mutation in NSCLC patients were screened by multivariate Logistic regression analysis,and a regression model was established to draw the receiver operating characteristic (ROC) curve for predicting EGFR mutation probability in NSCLC patients.A total of 75 patients with lung adenocarcinoma admitted to the oncology department of Cangzhou Central Hospital from 2020 to 2021 were randomly selected as the validation group to conduct internal verification of the predictive model.Results:Multivariate Logistic regression analysis showed that density shadow of ground glass,aerated bronchial sign,vascular cluster sign and CEA were independent influencing factors for predicting EGFR mutation in NSCLC patients (P<0.05).The multi-factor Logistic regression modelexpression was P=eX/(1+eX),X=-2.196+（0.036×CEA）+（0.984×ground glass opacity）+（1.923×air bronchogram）+（1.206×vessel convergence）.The prediction model ROC curve was 0.824［95%CI（0.767，0.881）］.The sensitivity and the specificity were 78.6% and 77.5%.The 75 patients factors in the validation group were brought into the prediction model,and the results showed that the area under the ROC curve was 0.762.The H-L goodness-of-fit test showed that χ2 =3.140，P=0.872．Conclusion:There are many risk factors affecting EGFR gene mutations in lung adenocarcinoma patients and clinical evaluation should be carried out in a timely and effective manner．A clinic prediction model constructedin this study has a good evaluation performance,which is worthy of clinical promotion.