【摘要】 目的 探讨多模态超声对浸润性乳腺癌患者新辅助化疗（NAC）后转移性腋窝淋巴结病理完全缓解（pCR）的预测价值。方法 选取行腋窝淋巴结常规超声、经皮超声造影（pCEUS）和剪切波弹性成像（SWE）检查的NAC前临床N₁期乳腺癌患者148例，其中训练集112例，测试集36例。以腋窝淋巴结清扫术后病理结果为“金标准”,分为腋窝淋巴结pCR组和腋窝淋巴结病理未完全缓解（npCR）组，比较2组乳腺癌患者原发灶T分期、分子分型、腋窝淋巴结短径、长径/短径比、淋巴门结构、实质厚度、实质回声、边界、彩色多普勒超声（CDFI）血流类型、淋巴结造影增强模式、淋巴管连续性、淋巴结平均弹性模量(E_mean)、最大弹性模量(E_max)、最小弹性模量(E_min)、标准差（SD）和弹性模量比值比(E_ratio)等相关因素的差异，筛选出与腋窝淋巴结pCR相关的独立影响因素，构建logistic回归模型。在测试集中采用受试者工作特征（ROC）曲线、Hosmer-Lemeshow拟合优度检验、校准曲线和决策曲线综合评价模型的预测性能。结果 训练集112例患者中，腋窝淋巴结pCR组49例，npCR组63例。单因素分析结果提示，腋窝淋巴结pCR组与npCR组的乳腺癌分子分型、淋巴结门结构、实质厚度、E_mean、E_max、E_min、SD、E_ratio、造影增强类型和淋巴管连续性比较，差别有统计学意义（P<0.05）。多因素分析结果提示，乳腺癌HER2⁺型、淋巴结实质厚度≤0.38 cm、E_max<10.0 kPa、造影呈均匀/环形增强为腋窝淋巴结pCR的保护因素（P<0.05）。以多模态超声指标构建腋窝淋巴结pCR回归模型，预测模型的曲线下面积为0.831（95%CI:0.749～0.923,P<0.05）,灵敏度为87.8%,特异度为70.9%。Hosmer-Lemeshow拟合优度检验、校准曲线和决策曲线结果表明，该预测模型在准确性、区分度、校准度和临床获益各方面的性能较好。结论 多模态超声可有效识别NAC后腋窝淋巴结pCR病例，为临床N₁期乳腺癌患者NAC后腋窝淋巴结手术方式的选择提供影像学依据。更多还原

【Abstract】 Objective To explore the predictive value of multimodal ultrasonography in lymph node pathologically complete response（pCR） after neoadjuvant chemotherapy（NAC） in invasive breast cancer. Methods 148 patients, including 112 in the training set and 36 in the testing set, with breast cancer before NAC, who underwent routine ultrasound, percutaneous contrast-enhanced ultrasound（pCEUS） and shear wave elastography（SWE） of axillary lymph nodes, were collected. T staging and molecular typing of breast cancer and the short diameter, long short diameter ratio, hilum structure, parenchymal thickness, parenchymal echo, boundary, color doppler flow imaging（CDFI） grade, contrast-enhanced mode of lymph nodes, lymphatic continuity, lymph nodes average elasticity(E_mean), maximum elastic modulus(E_max), minimum elastic modulus(E_min), standard deviation（SD） and elastic modulus ratio(E_ratio) were collected. According to the pathological results after axillary lymph node dissection, axillary lymph nodes were divided into pCR group and non-pathologically complete response（npCR） group. The influencing factors related to axillary lymph node pCR were analyzed and a logistic regression model was established. In the testing set, the predictive performance of the model was comprehensively evaluated using receiver operating characteristic（ROC） curves, Hosmer-Lemeshow goodness of fit tests, calibration curves, and decision curves. Results Among the 112 patients in the training set, 49 axillary lymph nodes were pCR and 63 were still positive. Univariate analysis showed that there were statistically significant differences in molecular typing of breast cancer, hilar structure, parenchymal thickness, E_mean, E_max, E_min, SD, E_ratio, contrast-enhanced type, and lymphatic continuity between pCR group and npCR group（P<0.05）. The results of multivariate analysis showed that HER2⁺, parenchymal thickness ≤ 0.38 cm, E_max <10.0 kPa and homogeneous/circular contrast enhancement were the protective factors of axillary lymph node pCR（P<0.05）. The regression model was constructed for axillary lymph nodes pCR, and area under curve was 0.831（95%CI:0.749-0.923, P<0.05）, with a sensitivity of 87.8% and a specificity of 70.9%. The results of the Hosmer-Lemeshow goodness-of-fit test, calibration curves, and decision curves all indicated that the predictive model performed well in all aspects of accuracy, discrimination, calibration, and clinical benefit. Conclusion Multimodal ultrasound of axillary lymph nodes can effectively recognize the cases of lymph nodes pCR after NAC. It can offer the reliable evidence in the choice of axillary lymph nodes surgical methods in patients with clinical N₁ breast cancer after NAC.更多还原