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

Abstract

640 Background: Breast cancer patients with a positive sentinel lymph node invariably have a 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. We have independently compared three of these models. Methods: Data from 108 breast cancer patients who underwent sentinel lymph node biopsy followed by complete axillary dissection were compiled. Measures additional to those usually determined (e.g., size of sentinel lymph node) were assessed on stored slides. All assessments were made under the supervision of one pathologist (MF). Data were used to determine the predicted risk of non-sentinel lymph node metastases using three mathematical models (MSKCC model, Ann Surg Oncol 2003, 10[10]1140-1151; Cambridge model, Br J Surg 2008, 95[3]302-309; Stanford model, BMC Cancer 2008, 8:66) and a comparison made with the observed status. Analyses were made of the areas under the receiver operating characteristic (ROC) curves using SAS/STAT software v9.2 (SAS Institute Inc., Cary, NC). Results: The areas under the ROC curves are shown in the Table. None of the models was significantly better than the others, although the Cambridge model required fewer measurements (grade, metastasis size, and proportion of positive nodes). Conclusions: This independent comparison of three mathematical models for predicting the risk of additional nodal metastases after positive sentinel lymph node biopsy in early breast cancer found no significant difference between the models, although the Cambridge model has the advantage of requiring fewer measurements. Comparison of ROC curves of the three models (n=108) Area Standard error 95% Wald confidence limits MSKCC 0.629 0.0584 0.515–0.743 Cambridge 0.724 0.0530 0.620–0.828 Stanford 0.667 0.0574 0.555–0.780 No significant financial relationships to disclose.