基于CT影像组学预测不可切除性胃癌姑息性化疗疗效  

A study on predicting the efficacy of palliative chemotherapy for inoperable gastric cancer based on CT imaging radiomics

作　　者：程震 尤亚茹 詹鹏超 郑月 范儒阳 吕培杰[1] 梁盼[1] 高剑波[1] CHENG Zhen;YOU Ya-ru;ZHAN Peng-chao(Department of Radiology,the First Affiliated Hospital of Zhengzhou University,Zhengzhou 450052,China)

机构地区：[1]郑州大学第一附属医院放射科,郑州450052

出　　处：《放射学实践》2025年第1期16-23,共8页Radiologic Practice

摘　　要：目的:构建基于治疗前增强CT的临床影像组学模型,评估其用于个体化预测不可切除性胃癌姑息性化疗的临床反应。方法:纳入256例经病理证实的不可切除性胃癌患者,并在姑息性化疗前行增强CT检查。根据实体肿瘤临床疗效评价标准确定治疗效果。对性别、年龄、肿瘤位置、实验室检查(CA19-9,CEA,CA724)、肿瘤最长径(横截面沿胃壁的最长轴)、肿瘤最厚径(垂直于横截面图像长轴的最大直径)、肿瘤临床T分期(cT)、临床N分期(cN)及临床M分期(cM)进行单、多因素逻辑回归分析,获得有统计学意义的临床独立预测因子以构建临床模型。使用3D Slicer软件在静脉期图像勾画感兴趣区(ROI),提取影像组学特征。选择最小绝对收缩及选择算子(LASSO)算法筛选并获得影像组学特征,构建影像组学模型。联合影像组学模型与临床模型构建影像组学列线图,使用诺莫图对模型进行可视化。采用ROC曲线的曲线下面积(AUC)等指标对构建模型的预测效果进行全方位评估。结果:95例患者对姑息性化疗有反应,161例无反应。肿瘤的临床分期是姑息性化疗疗效的临床独立预测因子(P<0.05),将临床分期作为独立预测因素以构建临床模型,该模型在训练集和验证集中预测姑息性化疗治疗反应的AUC分别为0.591(95%CI:0.525~0.657)、0.674(95%CI:0.574~0.776)。在筛选得到12个最优影像组学特征的基础上构建影像组学模型,该模型在训练集和验证集中的AUC分别为0.799(95%CI:0.733~0.865)、0.761(95%CI:0.656~0.865),均高于临床模型。影像组学列线图在训练集和验证集中的AUC分别为0.814(95%CI:0.748~0.879)、0.785(95%CI:0.687~0.882)。校正曲线分析结果表明,模型表现出较好的拟合性,预测结果与实际情况相符。决策曲线分析结果显示,在大部分阈值概率范畴当中,不管在训练集还是验证集中均表现出较高的净收益,具备很好的预测效能。结论:基于�Objective:To build and assess a pre-treatment clinical radiomics model based on enhanced CT imaging for individualized prediction of clinical response to palliative chemotherapy in inoperable Gastric Cancer.Methods:This retrospective study included a total of 256 histologically confirmed cases of unresectable gastric cancer,and enhanced CT scans were performed before palliative chemotherapy.The treatment response was determined according to the RECIST criteria.Univariate and multivariate logistic regression analyses were conducted to identify independent predictive factors,including gender,age,tumor location,clinical stage,laboratory tests(CA19-9,CEA,CA724),CT image features[tumor longest diameter along the gastric wall,tumor thickest diameter perpendicular to the longest axis on cross-sectional images,clinical T stage(cT),clinical N stage(cN),and clinical M stage(cM)].3D Slicer software was used to delineate regions of interest(ROIs)on the venous phase images and extract radiomic features.The features were selected using the least absolute shrinkage and selection operator(LASSO)method,and radiomic labels were constructed to build a radiomics model.Furthermore,radiomic nomograms were constructed,and the model was visualized using nomograms.The predictive performance of the constructed model was comprehensively evaluated using metrics such as the area under the receiver operating characteristic curve(AUC).Results:Among the included patients,95 patients showed a response to palliative chemotherapy.Clinical stage of the tumor was identified as a clinically independent predictive factor for the efficacy of palliative chemotherapy(P<0.05).A clinical model was constructed using clinical stage as an independent predictor,and the AUC of this model in the training and validation sets were 0.591(95%CI:0.525~0.657)and 0.674(95%CI:0.574~0.776),respectively.On the basis of selecting 12 optimal radiomic features,a radiomics model was constructed,and its AUC in the training and validation sets were 0.799(95%CI:0.733~0.865)and 0

关 键 词：胃肿瘤 不可切除性胃癌 姑息性化疗 影像组学 体层摄影术 X线计算机 

分 类 号：R735.2[医药卫生—肿瘤] R814.42[医药卫生—临床医学]