摘要
目的利用生物标志物与CT征象建立肺癌风险模型,以期为伴有胸腔积液的肺部良恶性病变的诊断提供依据。方法纳入1664例胸腔积液患者中筛选出训练集460例,验证集134例,根据临床-放射学及实验室参数纳入20个危险因素。通过多因素Logistic回归分析建立两个列线图模型(模型1、模型2),并对模型进行外部验证。用曲线下面积(AUC)来量化模型预测区分能力,以Delong检验进行AUC的比较,并通过校准曲线和决策曲线分析评价模型及其临床应用价值。结果通过多因素Logistic回归分析,肿瘤病史,胸腔积液癌胚抗原(CEA)、细胞角蛋白19片段(CYFRA21-1)、清蛋白/球蛋白比值(A/G),以及肿块最大直径、空气支气管征、分叶征和血管集束征为肺癌的危险因素。模型1与模型2在训练集中AUC分别为0.979(95%CI:0.968~0.991)和0.932(95%CI:0.909~0.955),两组AUC比较,差异有统计学意义(P<0.001);在验证集中AUC分别为0.911(95%CI:0.861~0.961)和0.846(95%CI:0.776~0.917),两组AUC比较,差异有统计学意义(P=0.009)。在诊断分类表中两个模型的正确分类比例分别为86.6%和82.1%。决策曲线分析表明在50%的阈值概率下模型1的净效益为45.0%,模型2为39.8%,并且在校准曲线中前者的预测值与实际值显示了更好的一致性。结论通过验证,胸腔积液生物标志物和CT征象结合构建的模型可用于评估伴有胸腔积液的肺部病变的癌症风险。
Objective To establish a risk model of lung cancer by using biomarkers and CT signs,so as to provide a basis for the diagnosis of benign and malignant pulmonary lesions with pleural effusion.Methods Among 1664 patients with pleural effusion,460 patients were selected from the training set and 134 patients in the validation set.20 important clinical risk factors were included according to clinical-radiological and laboratory parameters.Two nomogram models(model 1,model 2)were established by multivariate logistic regression analysis and were externally verified.The area under the curves(AUC)were used to quantify the predictive differentiation ability of the models,the AUC between the models was compared with Delong test,and the clinical application value of the models were evaluated by decision curve analysis and calibration curve.Results Through multivariate,logistic regression analysis,the history of extrathoracic malignancy,tumor diameter,air bronchogram,lobulation,vessel convergence,hydrothorax carcinoembryonic antigen(CEA),cytokeratin 19 fragment(CYFRA21-1)and the ratio of albumin to globulin(A/G)were risk factors for lung cancer.The AUC of model 1 and Model 2 in the training set were 0.979(95%CI:0.968—0.991)and 0.932(95%CI:0.909—0.955),the difference in AUC between the two groups was statistically significant(P<0.001).In the validation cohort,the AUC were 0.911(95%CI:0.861—0.961)and 0.846(95%CI:0.776—0.917),the difference in AUC between the two groups was statistically significant(P=0.009).The correct classification ratios of the two models in the diagnostic classification table were 86.6%and 82.1%,respectively.Decision curve analysis showed that the net benefit of model 1 was 45.0%and model 2 was 39.8%under the threshold probability of 50%.The predicted value of model 1 showed better consistency with the actual value in the calibration curve.Conclusion It has been demonstrated that the model constructed by combining pleural effusion biomarkers with CT signs can be used to assess the risk of cancer in
作者
涂宇琴
吴燕
陆云峰
陈特
毕小云
TU Yuqin;WU Yan;LU Yunfeng;CHEN Te;BI Xiaoyun(Department of Medical Laboratory;Department of Blood Transfusion;Department of Radiology,the First Affiliated Hospital,Chongqing Medical University,Chongqing 400016,China)
出处
《国际检验医学杂志》
CAS
2021年第18期2228-2233,共6页
International Journal of Laboratory Medicine
