Objective  To analyze the correlation among the clinicopathologic features, ultrasound imaging features, and axillary lymph node metastasis in breast cancer patients with negative clinical evaluation of axillary lymph nodes (cN0), and to establish a logistic regression model to predict axillary lymph node metastasis, so as to provide a reference for more accurate evaluation of axillary lymph node status in cN0 breast cancer patients. Methods  The data of 501 female patients with cN0 breast cancer who were hospitalized and operated in the Affiliated Hospital of Wuhan University of Science and Technology (Xiaogan Central Hospital) from December 2013 to October 2020 were collected. Among them, 376 patients from December 2013 to December 2019 were selected to establish a prediction model for axillary lymph node metastasis of cN0 breast cancer. In the modeling group, the basic information, clinical pathological characteristics, and ultrasound imaging features of patients were analyzed by single factor analysis. The factors with statistical significance were included in the multivariate logistic regression analysis, and the logistic regression prediction model was established. The model was evaluated by the correction curve and Hosmer-Lemeshow test goodness of fit. The model was validated in the validation group (125 patients from January to October 2020), and the receiver operation characteristic (ROC) curve was drawn. Results  The probability of positive axillary lymph nodes in 501 patients with cN0 breast cancer was 28.14% (141/501). The univariate analysis results of the modeling group showed that the histological grade, vascular invasion, progesterone receptor (PR), Ki-67, age, molecular typing, ultrasound breast imaging-reporting and data system (BI-RADS) grade were associated with axillary lymph node metastasis. Multivariate logistic regression analysis showed that the vascular infiltration, positive estrogen receptor (ER) , ultrasound BI-RADS grade 4C and Ki-67≥14% increased the probability of axillary lymph node metastasis (P<0.05). Using the above prediction factors to establish the prediction nomogram, the area under the ROC curve (AUC) of the modeling group was 0.72 [95%CI (0.66, 0.78)], the cut-off value was 0.30, the sensitivity was 61.00%, and the specificity was 71.20%. The newly established axillary lymph node transfer logistic regression model was applied to the validation group (n=125), and the AUC was 0.72 [95%CI (0.53, 0.76)]. The truncation value was 0.40, and the total coincidence rate was 69.60% (87/125), positive predictive value was 47.37% (18/38), and negative predictive value was 91.95% (80/87). Conclusions  Vascular invasion, positive ER , ultrasound BI-RADS grade 4C, and Ki-67≥14% are risk predictors of axillary lymph node metastasis in cN0 breast cancer patients. The negative predictive value of the model is 91.95%, which has a higher value in predicting axillary lymph node metastasis in early breast cancer patients, and can provide a reference for screening exempt sentinel lymph node biopsy population.

Citation： WANG Shu, SHEN Haoyuan. Establishment and validation of logistic regression model for risk factors of axillary lymph node metastasis in cN0 early breast cancer. CHINESE JOURNAL OF BASES AND CLINICS IN GENERAL SURGERY, 2022, 29(10): 1326-1333. doi: 10.7507/1007-9424.202202014 Copy