Retrospective validation of nomograms used in the prediction of non-sentinel lymph node metastasis in breast cancer patients with sentinel lymph node metastasis

The aim of this study was to validate five models used to predict non-sentinel lymph node metastasis (NSLNM) in patients positive at SLN biopsy in a single center retrospective database and to identify tumor characteristics effective in the prediction of NSLNM incidence in our own patient population.Twenty-six parameters related to 42 patients diagnosed with positive SLN?s out of 197 patients who have undergone SLN biopsy for breast cancer between January 2007 and March 2011 at the Gazi University School of Medicine, Department of General Surgery have been examined. All patients have received Completion Axillary Lymph Node Dissection (CALND). The predictive models developed by the MSKCC, Stanford University, Cambridge University, Tenon Hospital and Gür et. al. have been applied. The area under the Receiver Operating Characteristics (ROC) curve has been calculated for all the nomograms. In addition, parameters with the highest predictive effect in our population have been determined using the Pearson chi-squared and Fisher exact tests.Fourteen of 42 SLN positive patients had non-SLN metastasis. Among the MSKCC, Cambridge, Stanford, Tenon and Gür et. al models, the best predictive values were obtained from the Tenon model. In chi-squared analysis, pathological tumor size, lymphovaskuler invasion (LVI) status and the number of positive SLN?s were found to be statistically significant in the prediction of non-SLN positivity.Scores for the low risk group in the original series and our series for the Tenon model were 3.5 and 3.75 respectively but the false positive rates for the other models were either unacceptably high or their sensitivity and specificity values were unacceptably low. In our evaluation of tumor characteristics for our population, pathological tumor size, LVI status and the number of positive SLN?s detected were determined to be parameters which must be used in the prediction of non-SLN metastasis. Due to the limited size of the patient population used in this study, our findings should be verified in larger series.