基于CT影像组学列线图预测肺腺癌程序性细胞死亡受体配体1表达状态  

CT-Derived Radiomics Nomogram for Predicting the Expression of Programmed Cell Death Ligand 1 in Patient with Lung Adenocarcinoma

作　　者：徐婷 刘晓雯 陈亚曦 潘玉蝶 龚静山 XU Ting;LIU Xiaowen;CHEN Yaxi;PAN Yudie;GONG Jingshan(Department of Radiology,Shenzhen People's Hospital(the Second Clinical Medical College,Jinan University,the First Affiliated Hospital,Southern University of Science and Technology),Shenzhen 518020,China;不详)

机构地区：[1]暨南大学第二临床医学院,广东深圳518020 [2]深圳市人民医院(暨南大学第二临床医学院,南方科技大学第一附属医院)放射科,广东深圳518020

出　　处：《中国医学影像学杂志》2025年第1期33-40,共8页Chinese Journal of Medical Imaging

基　　金：国家自然科学基金面上项目(82172026)。

摘　　要：目的探讨基于术前CT影像组学列线图预测肺腺癌程序性细胞死亡受体配体1(PD-L1)表达状态的预测价值。资料与方法回顾性纳入2021年1月—2022年7月深圳市人民医院158例肺腺癌患者,其中PD-L1阴性82例,PD-L1阳性76例,按照7∶3随机分为训练集119例及验证集39例。分析临床病理及影像学资料,通过单因素及多因素Logistic回归分析筛选出PD-L1阴性与PD-L1阳性间有意义的特征构建临床模型。基于术前CT图像提取组学特征,筛选特征并建立模型。最后结合临床特征及影像组学分数构建联合模型,并通过列线图使模型可视化。使用受试者工作特征曲线及曲线下面积(AUC)评估模型的诊断效能。结果由癌胚抗原及血管集束征构成的临床模型,在训练集及验证集的AUC分别为0.774(95%CI0.687~0.860)、0.808(95%CI0.670~0.947);通过特征筛选,由17个影像组学特征组成的组学模型,其训练集及验证集的AUC分别为0.837(95%CI 0.764~0.910)、0.778(95%CI 0.633~0.923);由癌胚抗原、血管集束征及组学分数构成的联合模型,其训练集及验证集的AUC分别为0.892(95%CI0.832~0.952)、0.853(95%CI0.737~0.968);在训练集中,联合模型的AUC均高于其他两个模型,差异有统计学意义(Z=-2.640、-2.855,P均<0.05)。结论基于术前CT影像组学列线图对肺腺癌PD-L1表达状态具有较高的预测效能,可以为肺腺癌患者临床治疗方案的选择提供决策支持。Purpose To investigate the predictive value of nomogram based on preoperative CT imaging for predicting programmed cell death receptor ligand 1(PD-L1)expression in patient with lung adenocarcinoma.Materials and Methods A total of 158 patients with lung adenocarcinoma were enrolled in Shenzhen people's Hospital from January 2021 to July 2022,of which 82 were negative for PD-L1 and 76 were positive for PD-L1.They were randomly divided into training set(n=119)and verification set(n=39)according to the proportion of 7:3.The significant characteristics between PD-L1 negative and PD-L1 positive were screened by univariate and multivariate Logistic regression to construct a clinical model.Radiomics features were extracted from preoperative CT images,and then features were screened and modeled.Finally,the combined model was established by clinical factors and radiomics features,which was visualized by nomogram.The diagnostic performance of the model was evaluated using receiver operating characteristic curves and area under the curve(AUC).Results The area under the curve(AUC)of the clinical model composed of carcinoembryonic antigen and vascular convergence sign was 0.774(95%CI 0.687-0.860)and 0.808(95%CI 0.670-0.947)in the training set and validation set,respectively.Through feature screening,the radiomics model was composed of 17 radiomics features,and the AUC of the training and validation sets was 0.837(95%CI 0.764-0.910)and 0.778(95%CI 0.633-0.923).The training set and validation set of the combined model composed of carcinoembryonic antigen,vascular convergence sign and radiomics score were AUC 0.892(95%CI 0.832-0.952)and 0.853(95%CI 0.737-0.968).In the training set,the AUC of the combined model was higher than that of the other two models(Z=-2.640,-2.855,P<0.05).Conclusion Based on preoperative CT radiomics nomogram,it had high predictive efficacy on the expression of PD-L1 in lung adenocarcinoma and could provide decision-making support for the selection of clinical treatment regimens for lung adenocarcinoma pati

关 键 词：肺腺癌 影像组学 列线图表 程序性细胞死亡受体配体1 体层摄影术 X线计算机 

分 类 号：R445.3[医药卫生—影像医学与核医学] R734.2[医药卫生—诊断学]