基于自动乳腺超声的列线图模型早期预测HER-2阳性乳腺癌新辅助化疗病理完全缓解的临床价值  

Clinical value of a nomogram model based on automated breast ultrasound in early prediction of pathological complete response to neoadjuvant chemotherapy in HER-2 positive breast cancer

作　　者：赵阳 肖迎聪 巨艳 党晓智 蔡林利 薛文欣 李洋 肖瑶 郭妤绮 宋宏萍 Zhao Yang;Xiao Yingcong;Ju Yan;Dang Xiaozhi;Cai Linli;Xue Wenxin;Li Yang;Xiao Yao;Guo Yuqi;Song Hongping(Department of Ultrasound Medicine,The First Affiliated Hospital of Air Force Military Medical University(Xijing Hospital),Xi'an 710032,China;Shaanxi University of Traditional Chinese Medicine,Xianyang 712046,China)

机构地区：[1]空军军医大学第一附属医院(西京医院)超声医学科,西安710032 [2]陕西中医药大学医学技术学院,陕西咸阳712046

出　　处：《中华临床医师杂志（电子版）》2024年第4期355-362,共8页Chinese Journal of Clinicians(Electronic Edition)

基　　金：国家自然科学基金面上项目(82071934);陕西省科技计划项目国合重点项目(2020KWZ-022);陕西省高等教育教学改革研究重点项目(21JZ009);空军军医大学临床研究项目(2021LC2210)。

摘　　要：目的探讨自动乳腺超声(ABUS)早期预测HER-2阳性乳腺癌患者新辅助化疗(NAC)后获得病理完全缓解(pCR)的临床价值。方法回顾性分析2019年3月至2023年5月于空军军医大学附属西京医院乳腺外科收治的248例HER-2阳性女性乳腺癌患者,比较pCR组和非病理完全缓解(npCR)组NAC前各项参数的差异,并行多因素二元Logistic回归分析确定HER-2阳性乳腺癌pCR的独立预测因素,构建基于ABUS特征的列线图预测模型。应用Bootstrap方法(1000次重抽样)对模型进行内部验证;采用受试者工作特征(ROC)曲线评估模型的区分度,校准曲线评估模型的准确性,临床决策曲线(DCA)评价模型的临床获益。结果HER-2阳性乳腺癌NAC前2组肿瘤的ER状态、PR状态、分子亚型、皮肤侵犯、后方回声、冠状面汇聚征、冠状面白墙征比较差异有统计学意义(均P<0.05)。行二元Logistic回归分析显示分子亚型、皮肤侵犯、冠状面汇聚征、冠状面白墙征是HER-2阳性乳腺癌pCR的独立预测因素(均P<0.05)。基于这些变量构建列线图模型,其ROC曲线下面积为0.805(95%CI:0.751~0.859)。Hosmer-Lemeshow检验表明模型良好的拟合度(χ^(2)=6.597,P=0.360)。采用Bootstrap法迭代1000次进行内部验证,平均AUC为0.806(95%CI:0.742~0.855),表明模型稳定性良好,校准曲线表明列线图模型的预测概率与实际概率一致性好,决策曲线(DCA)线显示模型的临床获益及应用价值较高。结论基于NAC前肿瘤ABUS特征的列线图模型可以在一定程度上早期准确预测HER-2阳性乳腺癌NAC后pCR,可为乳腺癌患者临床治疗方案的制定提供超声影像依据。Objective To assess the clinical usefulness of automated breast ultrasound(ABUS)in predicting pathological complete response(pCR)to neoadjuvant chemotherapy(NAC)in HER-2 positive breast cancer patients.Methods A retrospective analysis was performed on 248 HER-2 positive female breast cancer patients admitted to the Department of Breast Surgery,Xijing Hospital Affiliated to Air Force Military Medical University from March 2019 to May 2023.The differences in parameters before NAC were compared between patients with pCR and non-pCR(npCR)patients.Multivariate binary Logistic regression analysis was performed to identify the independent predictors of pCR in HER-2 positive breast cancer.Additionally,a nomogram prediction model was developed based on features obtained from ABUS.The Bootstrap method(1000 times of sampling)was used to verify the model internally.Receiver operating characteristic(ROC)curve analysis was employed for assessing the discriminative performance of the model,and decision curve analysis(DCA)was used to evaluate the clinical benefit and application value of the model.Results There were significant differences in ER status,PR status,molecular subtype,skin invasion,posterior echo,retraction phenomenon in the coronal plane,and white wall sign in the coronal plane between the the pCR and npCR groups before NAC(P<0.05 for all).Binary Logistic regression analysis showed that molecular subtype,skin invasion,retraction phenomenon in the coronal plane,and white wall sign in the coronal plane were independent predictors of pCR in HER-2 positive breast cancer(P<0.05 for all).A nomogram model was constructed based on these variables,and the area under the ROC curve of this model for predicting pCR in HER-2 positive breast cancer was 0.805(95%confidence interval[CI]:0.751~0.859).The Hosmer-Lemeshow test showed that the model had a good fit(χ^(2)=6.597,P=0.360).The calibration curve demonstrated an excellent correspondence between the predicted and observed probabilities of the joint model,thereby indicating it

关 键 词：HER-2阳性乳腺癌 自动乳腺超声 病理完全缓解 新辅助化疗 

分 类 号：R737.9[医药卫生—肿瘤]