摘要
目的探讨CT影像组学预测非小细胞肺癌免疫检查点抑制剂治疗疗效的价值。方法分别回顾性连续纳入本院及外院经二线PD-1/PD-L1单药抑制剂免疫治疗患者114例和52例作为训练集和外验证集。根据首次免疫治疗后疗效分为有反应组及无反应组。基于治疗前增强CT提取影像组学特征,使用Logisti c回归分别构建临床、影像组学及临床-组学预测模型,并评价模型的区分度、校准度及临床有效性。结果在训练集中,反应组和无反应组分别为45及69例,外部验证集中,反应组和无反应组分别22及30例。共12个影像组学特征纳入模型,临床特征中的肿瘤最大径(OR=2.547[1.092-5.941],P=0.031)及中性粒细胞计数(OR=3.642[1.303-10.179],P=0.014)对免疫治疗疗效有预测价值。临床模型、组学模型及临床-组学模型的AUC值依次为0.653(95%Cl:0.559-0.746)、0.856(95%Cl:0.784-0.926)、0.894(95%CI:0.838-0.950),在外验证集中,临床模型、组学模型及临床-组学模型的AUC值依次为0.624(95%Cl:0.535-0.712)、0.794(95%Cl:0.715-0.871)、0.817(95%CI:0.838-0.950)。结论CT影像组学联合临床特征对预测非小细胞肺癌免疫检查点抑制剂治疗疗效有较高的价值。
Objective To explore the value of CT radiomics in predicting the efficacy of immune checkpoint inhibitor therapy for non-small cell lung cancer.Methods A total of 114 and 52 patients who received second-line PD-1/PD-L1 monotherapy immunotherapy in our hospital and external hospitals were retrospectively included as the training set and external validation set,respectively.According to the efficacy of the first immunotherapy,patients were divided into a responsive group and a non-responsive group.Based on pre-treatment enhanced CT to extract radiomics features,logistic regression was used to construct clinical,radiomics,and clinical omics prediction models,and the discrimination,calibration,and clinical effectiveness of the models were evaluated.Results In the training set,there were 45 and 69 cases in the response group and 69 cases in the non-response group,respectively.In the external validation set,there were 22 and 30 cases in the response group and 30 cases in the non-response group,respectively.A total of 12 radiomics features were included in the model,and clinical features such as tumor maximum diameter(OR=2.547[1.092-5.941],P=0.031)and neutrophil count(OR=3.642[1.303-10.179],P=0.014)have predictive value for the efficacy of immunotherapy.The AUC values of the clinical model,omics model,and clinical omics model were 0.653(95%CI:0.559-0.746),0.856(95%CI:0.784-0.926),and 0.894(95%CI:0.838-0.950),respectively.In the external validation group,the AUC values of the clinical model,omics model,and clinical omics model were 0.624(95%CI:0.535-0.712),0.794(95%CI:0.715-0.871),and 0.817(95%CI:0.838-0.950),respectively.Conclusion The combination of CT imaging omics and clinical features has high value in predicting the efficacy of immune checkpoint inhibitor therapy for non-small cell lung cancer.
作者
孙泽恒
张旭胤
陈瑞
张硕
梁挺
杜永浩
牛刚
SUN Ze-heng;ZHANG Xu-yin;CHEN Rui;ZHANG Shuo;LIANG Ting;DU Yong-hao;NIU Gang(Department of Radiology,The First Affiliated Hospital of Xi'an Jiaotong University,Xi'an 710000,Shaanxi Province,China;Department of Radiology,Xi'an Yanliang District People's Hospital,Xi'an 710089,Shaanxi Province,China)
出处
《中国CT和MRI杂志》
2025年第8期36-39,共4页
Chinese Journal of CT and MRI
基金
《多模态MRI评估非小细胞肺癌的肿瘤微环境预测免疫相关性肺炎分级的临床研究》(2022JQ-884)。
关键词
非小细胞肺癌
影像组学
CT
疗效
Non-small Cell Lung Cancer
Radiomics
CT
Curative Effect
