A nomogram to predict nonsentinel lymph node involvement in breast cancer patients with sentinel lymph node metastases.

14 Background: Several nomograms have been described as predictors of non-sentinel axially lymph node (non-SN) metastases in breast cancer with positive sentinel nodes (SN). However, all these predicting models were based on data from western countries. The purpose of this study was to examine predictive factors of non-SN status among SN metastatic patients, in order to develop a nomogram based on Japanese large data set. Methods: This research was analyzed by using a clinical database of 11,228 Japanese breast cancer patients who registerd to cohort study as SN biopsy between March 2008 and Octover 2009 in Japan. We reviewed data retrospectively to extract patients with SN metastases who underwent complementary axillary lymph node dissection. In this cohort, we examined predictive factors of non-SN metastases. All clinical and pathologic features were analyzed to predict the non-SN status, by using univariate and multivariate logistic regression model. A receiver operating characteristic curve was constructed and the area under the curve (AUC) was calculated. Results: Among the database, SN metastases were found in 1,029 patients, and 345 (33.5%) were non-SN positive. Univariate analysis showed a significant association between non-SN involvement and primary tumor size (p<0.001), histologic grade (p=0.011), lymphatic invasion (p<0.001), venous invasion (p=0.005) and the number of involved SNs among all identified SNs (p<0.001). Tumor size (p<0.001), lymphatic invasion (p<0.001), and the size of SN metastasis (p<0.001) were associated with non-SN metastasis in multivariate analysis. Based on the multivariate analysis, we developed a scoring system to predict the likelihood of non-SN metastases in breast cancer patients with SN involvement. The discriminatory ability of our nomogram, as measured by the AUC, was 0.752. Conclusions: In patients with invasive breast cancer and a positive SN, primary tumor size, lymphatic invasion, and the size of SN metastases among all identified SNs were independently predictive of non-SN involvement, and used for a nomogram. Validation study will be performed in the future investigation.