RESUMO
BACKGROUND: Tumor-positive sentinel node biopsy (SNB) suggests a risk of nonsentinel node metastases in breast cancer. This risk is lower after micrometastasis or isolated tumor cells (ITC) in the sentinel node (SN), and recent studies suggest that completion axillary lymph node dissection (ALND) might not improve outcome in these patients. We aim to validate existing predictive models and to develop a new model for micrometastatic and ITC patients. METHODS: A series of 484 patients with micrometastases or ITC in SN followed by ALND was used to evaluate factors affecting nonsentinel node involvement. Logistic regression analysis was performed to construct a predictive model, which was validated by a separate series of 51 patients. RESULTS: Only 7.2% of patients had additional metastases on completion ALND. Tumor diameter and multifocality associated with nonsentinel status on multivariate analysis. A predictive model was constructed showing good [area under the curve (AUC) 0.791] discrimination in the validation series. Previously published models performed poorly in our patient population. CONCLUSIONS: Nonsentinel node metastases are rare with micrometastasis or ITC in SN. Most published predictive models for nonsentinel node involvement perform poorly in the present patient population. We developed a new predictive model which seems to perform well in discriminating patients with more than 10% risk of additional metastases. However, the presented nomogram needs to be validated with an independent patient series to evaluate its accuracy, especially for high-risk patients.