Background and purpose: The increasing use of systemic adjuvant therapy even in lymph node-negative breast cancer patients and breast cancer screening programs detecting smaller tumors with less probability of metastatic lymph nodes questions the need for routine axillary lymph node dissection. Since morbidity of breast cancer surgery is predominantly related to axillary lymph node dissection, predictive models for lymph node involvement may provide a way to avoid lymph node surgery and its side effects in subgroups of patients.
Patients and methods: Using a multivariate logistic regression model, tumorbiological parameters such as expression of estrogen and progesterone receptors, Ki-67, p53, cathepsin D, HER2, S-phase fraction, and ploidy were analyzed regarding their ability to predict axillary lymph node involvement in 655 breast cancer patients.
Results: The model correctly predicted axillary lymph node metastases in 58% of the patients by including expression of progesterone receptor, HER2, and Ki-67. In a subgroup of 200 patients predicted to be at extremely high or extremely low risk for axillary lymph node metastases, the accuracy of the prediction was 70%.
Conclusion: With a model just based on tumorbiological parameters obtained in the primary tumor it is possible to predict axillary lymph node status. By including additional parameters it appears to be feasible to further improve the model in order to avoid axillary lymph node surgery in low-risk women.