Development of nomograms to predict axillary lymph node status in breast cancer patients

Background: Prediction of axillary lymph node (ALN) status preoperatively is critical in the management of breast cancer patients. This study aims to develop a new set of nomograms to accurately predict ALN status. Methods: We searched the National Cancer Database to identify eligible female breast cancer patients with profiles containing critical information. Patients diagnosed in 2010-2011 and 2012-2013 were designated the training (n=99,618) and validation (n=101,834) cohorts, respectively. We used binary logistic regression to investigate risk factors for ALN status and to develop a new set of nomograms to determine the probability of having any positive ALNs and N2-3 disease. We used ROC analysis and calibration plots to assess the discriminative ability and accuracy of the nomograms, respectively. Results: In the training cohort, we identified age, quadrant of the tumor, tumor size, histology, ER, PR, HER2, tumor grade and lymphovascular invasion as significant predictors of ALNs status. Nomogram-A was developed to predict the probability of having any positive ALNs (P_any) in the full population with a C-index of 0.788 and 0.786 in the training and validation cohorts, respectively. In patients with positive ALNs, Nomogram-B was developed to predict the conditional probability of having N2-3 disease (P_con) with a C-index of 0.680 and 0.677 in the training and validation cohorts, respectively. The absolute probability of having N2-3 disease can be estimated by P_any*P_con. Both of the nomograms were well-calibrated. Conclusions: We developed a set of nomograms to predict the ALN status in breast cancer patients.