Comparison of three mathematical models for predicting the risk of additional axillary nodal metastases after positive sentinel lymph node biopsy in early breast cancer

Abstract

Abstract Background Women with breast cancer and a positive axillary sentinel lymph node (SLN) are recommended to undergo complete axillary lymph node dissection; however, further nodal disease is not always present. Mathematical models have been constructed to determine the risk of metastatic disease; three of these were evaluated independently. Methods Data from 108 women with breast cancer who had a positive SLN biopsy and completion axillary lymph node dissection were used. Measurements of additional parameters over those usually determined (such as size of SLN metastasis) were assessed under the supervision of two pathologists. These data were used to determine the predicted risk of non-SLN metastases using three mathematical models (from Memorial Sloan-Kettering Cancer Center (MSKCC), Cambridge University and Stanford University) and a comparison made with the observed findings. Analyses were made using the area under the receiver operating characteristic (ROC) curve (AUC). Results Some 53 (49·1 per cent) of 108 patients had a positive non-sentinel axillary lymph node metastasis. The AUC values were 0·63, 0·72 and 0·67 for the MSKCC, Cambridge and Stanford nomograms respectively. Conclusion This independent comparison found no significant difference between the models, although the Cambridge model had the advantage of requiring fewer measurements with a more accurate predictive performance.