Improving risk stratification in breast cancer through a novel mathematical model of axillary lymph-node stage migration

Background: Surgical axillary staging is an important evidence in breast cancer staging, prognosis and subsequent therapies. This study aims to stratify breast cancer with >3 positive axillary lymph nodes (PLN) using a novel mathematical model, and to evaluate the impact of axillary stage migration on breast cancer staging. Methods: This study included female patients diagnosed with breast invasive ductal carcinoma with >3 PLNs from the Surveillance, Epidemiology, and End Results database (n = 9642) and Sun Yat-sen University Cancer Center database (n = 416), which were used as the training and validation cohorts, respectively. A mathematical model of removed axillary lymph nodes (RLNs) number was built based on the hypergeometric distribution and Bayes' theorem. A multivariate Cox regression model and a nomogram were established to evaluate the correlation of RLNs number with stage migration and overall survival. Results: RLNs number was an independent prognostic factor in breast cancer with >3 PLNs (P < 0.001). A mathematical model calculated the probability of stage migration in every situation of RLNs number and showed that patients with stage migration had worse survivals than those with no stage migration (training: P < 0.001; testing: P = 0.015). A nomogram based on stage migration and other factors was built. The AUC for 3-/5-year survival rate was 0.765/0.730 and 0.890/0.834 in training and testing group, respectively. Conclusion: Axillary stage migration is associated with overall survival and could differentiate higher risk populations from breast cancer with >3 PLNs. Our established model may help the surgeon better assess the accurate stage, predict prognosis and select strengthened or extended therapies in those higher-risk populations.