P3-07-40: Impact of the Sentinel Lymph Node Procedure on the Detection of Positive Lymph Nodes in Breast Cancer.

Abstract

Abstract Introduction: The objective of the sentinel lymph node (SLN) procedure in breast cancer is to perform an accurate axillary staging and provide good local control, while sparing the patients the morbidity of an axillary lymph node dissection (ALND). Since its routine clinical use, questions have been raised concerning the upstaging of a subgroup of node-negative patients and an increase in the overall percentage of node-positive patients. The goal of our study was to investigate the impact of the SLN procedure on the detection of positive lymph nodes. Patients and methods: We included 1119 consecutive breast cancer patients from one center (2007-2009) who underwent primary surgery for a breast cancer smaller than 5 cm diameter without clinical involvement of axillary lymph nodes. 31 patients had a bilateral breast cancer. Logistic regression models were used to analyze the effect of various predictors on the presence of positive lymph nodes. First, we fitted univariable models for a number of predictors of which the relationship with lymph node involvement was to be expected. (multifocality, tumor location in the breast, tumor size, LVI, ER, PR, Her-2, tumor grade, age and histologic subtype). In a second stage, we fitted a multivariable model including the procedure (SLN biopsy or ALND) and the predictors that have been shown to be related to lymph node involvement in univariable analysis or for which this relationship has been suggested in the literature. The probability of finding positive lymph nodes was thus calculated in both groups correcting for relevant predictors of lymph node involvement. Results: The SLN biopsy group includes 830 patients, the ALND group 320. Tumors in the ALND group are larger (mean 26.65 mm versus 18.43mm) and more frequently multifocal (24.44% versus 8.19%). 33% of patients in the ALND group were lymph node positive, 28% in the SLN procedure group. Univariable and multivariable analyses revealed that LVI, tumor size and multifocality are the most significant predictors of lymph node involvement (table 1). The AUC for the multivariate model is 0.754. By comparing ALND en SLN biopsy an odds ratio (OR) of 0.7635 (95% confidence interval: 0.5334−1.0930) is observed, suggesting a higher probability of finding positive lymph nodes with SLN procedure. However, the difference between the procedures is not significant (p= 0.1404). Conclusion: For comparable tumors, the chance of detecting positive lymph nodes is not lower when applying SLN procedure compared to ALND. The results even point in the direction of a higher detection rate with SLN. However, care should be taken in making this conclusion given a non-significant result. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P3-07-40.