Comparison of two models for the prediction of nonsentinel node metastases in breast cancer

Background It is debated whether all patients with a positive sentinel node dissection (SLND) should be submitted to axillary lymph node dissection (ALND). Models have been developed to estimate the likelihood of nonsentinel node (non-SLN) metastases. Methods The accuracy of the Memorial Sloan-Kettering Cancer Center (MSKCC) nomogram and MD Anderson scoring system for the prediction of non-SLN status was tested in a consecutive series of 186 SLN-positive breast cancer patients. A multivariate analysis was performed to assess which parameters independently predicted the presence of non-SLN metastases. Results The predictive accuracy of the MSKCC nomogram measured by the receiver operating characteristic curve was 0.71, and it was best in patients with <10% risk of non-SLN metastases (sensitivity 100% and specificity 96%). The MD Anderson score predicted non-SLN involvement with low accuracy because it classified 85% of the patients in the intermediate-risk groups. Only SLN macrometastases and tumor multifocality independently predicted non-SLNs involvement. Conclusions The MSKCC nomogram can help individualize the surgical treatment of SLN-positive breast cancer when the likelihood of further axillary involvement is low or surgical risks are higher.