| Background With the development of modern medicine, the treatment of breast cancer has entered the era of precision medicine. Axillary lymph node status is the most important factor in staging, treatment and prognosis, which is indispensable in precise management of early-stage breast cancer. Currently Sentinel lymph node biopsy(SLNB)has become the gold standard of surgical axillary staging in early-stage breast cancer. Despite the fact that SLNB is associated to less morbidity of complications than axillary lymph node dissection(ALND), SLNB can still cause a series of complications and should be considered as over-treatment in 60 percent of early-stage breast cancer patients with no axillary lymph nodal metastasis(ALNM). These patients could gain more benefits from a noninvasive staging method instead of SLNB. Ultrasound has been increasingly used in the preoperative evaluation of axillary lymph node status due to its relatively high sensitivity and specificity, which has the potential to be a new tool of axillary staging. However, axillary ultrasound(AUS) is operator-dependent and has a certain range of false-negative rate. It is essential to identify the most significant sonographic features to accurately predict ALNM. And with the identification of lower axillary metastatic burden related to false-negative results, AUS can be validated as a reliable and safe tool of noninvasive axillary staging.Objective To investigate the clinical application of AUS to the identification of ALNM in early-stage breast cancer via the analysis of predictive factors of ALNM and establish multivariate logistic regression models.Methods We prospectively enrolled 282 consecutive patients with stage Tis-T2 breast cancer between December 2013 and September of 2015. All the patients underwent axillary ultrasound performed by two specified senior ultrasound doctors. Sonographic features of their axillary lymph nodes(longitudinal and transverse diameters, cortical and hilar thickness, blood flow form) were collected. These patients were divided into metastatic, suspicious and non-metastatic groups based on the ultrasound features by ultrasound doctors. Clinical and pathologic features(age,T-stage, pathologic type, menopausal status, estrogen-receptor, progesterone-receptor Her-2 status, histologic subtype) of these patients were collected. The diagnostic accuracy of axillary ultrasound was compared with results of pathology. Univariate and multivariate logistic regression analyses were used to evaluate the relationship between sonographic features, clinical and pathologic features and axillary nodal metastasis. The area under the ROC curve was used to assess the accuracy of the univariate and multivariate logistic regression models.Results(1) 47.5%(134/282) of all the patients had ALNM confirmed by pathologic examination. 110 patients were divided into metastatic group according to the ultrasound results. 95 of them had ALNM. 117 patients were divided into non-metastatic group and 101 of them were confirmed non-metastatic by pathology. Only 19.2%(55/282) of all the patients were included into the suspicious group.(2) The sensitivity, specificity, positive predictive value, negative predictive value and accuracy of AUS were respectively 85.6%, 87.1%, 86.4%, 86.3%, and 86.3% in the metastatic and non-metastatic groups. The Kappa value was 0.727(P<0.001).(3) The axillary node metastasis burden in the non-metastatic group was significantly lower than in the metastatic group(1.2 vs 6.9, P<0.001). The false-negative results were found only in 16 cases, fourteen of whom had only 1 metastasis axillary node, and two had 2 and 3 metastasis nodes respectively.(4) Univariate logistic regression analysis showed that maximum cortical thickness was the most significant predictive factor of ALNM(the area under the ROC curve was 0.872). T-stage and histologic subtype are predictive factors of ALNM among all the clinical and pathologic features.(5) Multivariate logistic regression analysis suggested that T-stage, cortical thickness and the ratio of hilar thickness to cortical thickness were predictive factors of ALNM(P<0.05). The area under the ROC curve of the multivariate logistic regression model of sonographic features was 0.879 and its sensitivity and specificity were 77.0% and 85.1% respectively. The area under the ROC curve of the integrated multivariate logistic regression model of sonographic features and T-stage was 0.910 and its sensitivity and specificity were 85.1% and 83.8% respectively.Conclusion AUS is a valuable tool for detecting ALNM. Patients with false-negative results of AUS have a lower axillary metastatic burden. Maximum cortical thickness is the most significant predictive factor of ALNM. AUS may be a potential alternative method for sentinel lymph node biopsy as axillary lymph node staging in early-stage breast cancer patients. |
PDF Full Text Request