小乳腺癌超声直接和间接征象与腋窝淋巴结转移关系的Logistic回归分析

The correlation between direct or indirect signs of small breast cancer and axillary lymph node metastasis with the binary Logistic regression model by unltrasound

目的探讨小乳腺癌(最大径均≤2.0cm)超声直接征象和间接征象与腋窝淋巴结转移的关系。方法对65例(65个肿块)乳腺癌患者的(23例伴腋窝淋巴结转移)术前超声声像图表现进行分析,以病理表现为基础将超声图像分为直接征象和间接征象。直接征象包括:(1)肿块形态不规则;(2)小分叶;(3)边缘毛刺;(4)内部回声不均匀;(5)内部微钙化;(6)后方回声衰减;(7)纵横比≥0.71;(8)血流分级Ⅱ～Ⅲ级;(9)内部穿支血管;(10)最大血流速度Vmax≥15cm/s;(11)阻力指数RI≥0.7;(12)速升速降型血流频谱。间接征象包括:(1)肿块周边高回声晕;(2)乳腺间质内纤维结构(浅筋膜、深筋膜及Cooper韧带)的模糊或中断;(3)腋窝淋巴结肿大(长短径比<2.0、最大皮质厚度≥3.0mm、皮髓质比≥1.0以及血流类型)。以病理诊断淋巴结转移为因变量,小乳腺癌各超声直接和间接征象为自变量,建立Logistic回归模型,用ROC曲线评价Logistic模型的预报能力。结果小乳腺癌直接征象中内部微钙化、血流分级Ⅱ～Ⅲ级和间接征象中浅筋膜的改变、淋巴结最大皮质厚度≥3.0mm进入方程,血流分级Ⅱ～Ⅲ级和淋巴结最大皮质厚度≥3.0mm对预测腋窝淋巴结转移有统计学意义(OR=33.246、99.579,P=0.005、0.001),ROC曲线下面积分别为0.909±0.043和0.923±0.044,P=0.000、0.000。微钙化和浅筋膜的改变对预测腋窝淋巴结转移无统计学意义(OR=12.036、19.406,P=0.098、0.066)。结论小乳腺癌直接征象中血流分级Ⅱ～Ⅲ级和间接征象中淋巴结最大皮质厚度≥3.0mm与腋窝淋巴结转移关系密切,对术前预测小乳腺癌淋巴结是否转移具有重要价值。

Objective To explore the correlation between signs of small breast cancer （ ≤ 2.0 cm） and axillary lymph node metastasis by ultrasound. Methods The sonographic features of 65 cases with 65 breast malignant lesions （23 with axillary lymph metastasis） were analyzed by classifying into direct and indirect signs according to pathological assessment. The direct ultrasonographic signs included irregular shape, microlobulation, speculated margin, centric eehogenity, microcalcification, posterior attenuation, length/ width ratio ≥0.71, the Ⅱ-Ⅲ grade flow of tumor, afferent artery, Vmax≥ 15 cm/s, RI≥ 0.7 and spectral pattern. The indirect uhrasonographic signs included halo, changes of fibred structures（including superficial fasia, deep fascia, cooper＇ s ligament） , fetures of axillary lymph nodes （including longtiudina-transverse axis ratio 〈 2.0, maximum cortex thickness ≥3.0 mm, cortex-hilus thickness ratio ≥ 1.0 and the vascular pat- tern）. A Logistic regression model for predicting axillary lymph node metastasis on the basis of pathology and ultrasonographic signs was obtained. A receiver operating characteristic （ROC） curve was used to assess the performance of the Logistic regression model. Results The direct sonographie features of micro calcifications, the Ⅱ -Ⅲ grade flow of tumor and the indirect signs of changes of superficial fascia, maximum cortex thickness≥3.0 mm of axillary lymph node were applied into the model. There was statistical significance of the Ⅱ -Ⅲ grade flow and maximum cortex thickness ≥3.0 mm （ OR =33. 246,99. 579 ,P =0. 005,0. 001 ）. Their areas under the ROC curve were 0. 909 ±0. 043 and O. 923 ±0.044, P = 0. 000,0.000 . No statistical significance of micro calcifications and changes of superficial fascia were observed （ OR = 12. 036,19. 406, P = 0. 098 ,0. 066 ） . Conclusion The Ⅱ -Ⅲ grade of tumor flow and maximum cortex thickness ≥3.0 mm were positively correlated with metastasis of axillary lymph node, which was useful for predicting lymph node metastasis of small breast cancer before operation.