P5-14-21: Predicting the Status of the Axillary Lymph Nodes in 1300 Consecutive Patients with Early Breast Cancer Treated in a Single Institution.

Abstract

Abstract Purpose. Lymph node involvement is the most important prognostic factor in breast cancer. It is a multifactorial event determined by tumour and patient characteristics. Several predictors of axillary lymph node metastasis (ALNM) in breast cancer have been described. The purpose of this study was to determine clinical and pathological factors predictive for ALNM in patients with early breast cancer and to build a model to portend lymph node involvement. Methods. We evaluated 1300 consecutive patients surgically treated in our institution (2007-2009) for cT1-T2 invasive breast cancer. The patient and tumour characteristics evaluated included: age at diagnosis, number of foci, histologic grade, location, tumour size, histologic subtype, lymphovascular invasion (LVI), estrogen-receptor (ER), progesterone-receptor (PR) and Her2 status. Univariate and multivariate analyses were performed. Factors significantly associated with lymph node metastasis by univariate analysis and histologic subtype were included in the multivariate model. We validated our model with the same 1300 patients on the basis of a multivariable logistic regression model containing the variables, using a correction factor. Results. By univariate analysis, the incidence of ALNM was significantly higher in patients with a tumour with LVI (P < 0.0001), larger tumours (P < 0.0001), tumours with a higher histologic grade (P < 0.0001), tumours located retroareolar or lateral in the breast (P < 0.0001), tumours with multiple foci (P = 0.0002) and in patients who underwent an axillary lymph node dissection. We found no effect of ER/PR nor HER-2 status. By multivariate analysis, lymph node involvement was significantly associated with the presence of lymphovascular invasion (P < 0.0001), larger tumour size (P < 0.0001), axillary lymph node dissection (P = 0.0003), retroareolar and lateral tumour location in the breast (P = 0.0019) and the presence of multiple foci (P = 0.0155). The discrimination of the model, calculated by the area under the curve (AUC), was 0.778. Conclusions. LVI and tumour size emerged as most powerful independent predictors of ALNM. A model with an AUC of 0.50 is equivalent to the toss of a coin, a model with an AUC > 0.70 is considered as good. In conclusion, our result is not perfect in discrimination which raises further questions. It is clear that underlying mechanisms of ALNM remain to be elucidated. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-14-21.