基于多模态超声成像特征构建的乳腺癌雄激素受体表达水平预测模型的临床价值  

Clinical Value of the Prediction Model of Androgen Receptor Expression in Breast Cancer Based on Multimodal Ultrasound Imaging Features

作　　者：高卓然 赵诚[1] 李晶晶[2] 刘清华[2] 李丽[2] Gao Zhuoran;Zhao Cheng;Li Jingjing;Liu Qinghua;Li Li(Department of Ultrasound,The Affiliated Hospital of Qingdao University,Qingdao,Shandong 266000,China;Department of Ultrasound,People's Hospital of Rizhao,Rizhao,Shandong 276800,China)

机构地区：[1]青岛大学附属医院超声科,山东省青岛市266000 [2]日照市人民医院超声科,山东省日照市276800

出　　处：《中国超声医学杂志》2024年第6期649-653,共5页Chinese Journal of Ultrasound in Medicine

摘　　要：目的 探讨基于多模态超声成像特征构建的乳腺癌雄激素受体(AR)表达水平预测模型的临床价值。方法 收集70例乳腺癌患者,包含73个病灶,分析常规超声特征、剪切波弹性成像(SWE)参数、血流定量分析(FQ)参数与AR表达的关系,多因素Logistic回归分析筛选AR表达独立预测因素,构建预测模型,绘制受试者工作特征(ROC)曲线分析模型的预测效能,应用校准曲线和临床决策曲线评价模型的校准度和临床有效性。结果 多因素Logistic回归分析显示:病灶杨氏模量最大值(Emax)、毛刺征、血流最丰富切面的面积(area)是AR表达的独立预测因素,联合上述3个指标构建的预测模型曲线下面积(AUC)为0.884 (95%CI:0.803~0.965),灵敏度为78.8%,特异度为81.0%,校准曲线和临床决策曲线显示预测模型校准度和临床有效性较好。结论 基于多模态超声成像特征构建的乳腺癌雄激素受体表达水平的预测模型有一定的临床价值,能对AR表达进行术前无创评估。Objective To investigate the clinical value of a prediction model for breast cancer androgen receptor(AR)expression level constructed based on multimodal ultrasound imaging features.Methods We collected 70 breast cancer patients,including 73 lesions,analyzed the relationship of conventional ultrasound features,shear wave elas-tography(SWE)parameters,blood flow quantitative(FQ)analysis parameters and AR expression,and screened in-dependent predictors of AR expression by multifactor Logistic regression analysis.The predictive model was construc-ted and the predictive efficacy was evaluated by the receiver operating characteristic(ROC)curve.The calibration and clinical effectiveness of the model was evaluated using calibration curve and clinical decision curve.Results Multifac-tor Logistic regression analysis showed that the maximum Young's modulus(Emax),burr sign,and the most abun-dant section area of lesion flow(area)were independent predictors of AR expression.The above three indexes were combined to construct the prediction model.The area under the ROC curve(AUC)was 0.884(95%CI:0.803-0.965),the sensitivity 78.8%,and the specificity 81.0%.The calibration curve and the clinical decision curve showed good calibration and clinical validity of the prediction model.Conclusions The predictive model for AR ex-pression level in breast cancer based on multimodal ultrasound imaging features has certain clinical value for preopera-tivenon-invasive evaluation of AR expression.

关 键 词：乳腺癌 多模态超声 雄激素受体 

分 类 号：R445.1[医药卫生—影像医学与核医学] R737.9[医药卫生—诊断学]