A risk score model predictive of the presence of additional disease in the axilla in early-breast cancer patients with one or two metastatic sentinel lymph nodes

Abstract Background Axillary lymph node dissection (ALND) in early-breast cancer patients with positive sentinel node (SLN+) may not always be necessary. Aims To predict the finding of ≥1 metastatic axillary node in addition to SLN+(s); to discriminate between patients who would or not benefit from ALND. Methods Records of 397 consecutive patients with 1-2 SLN+s receiving ALND were reviewed. Clinico-pathological features were used in univariate and multivariate analyses to develop a logistic regression model predictive of the risk of ≥1 additional axillary node involved. The discrimination power of the model was quantified by the area under the receiver operating characteristic curve (AUC) and validated using an independent set of 83 patients. Results In univariate analyses, the risk of ≥1 additional node involved was correlated with tumor size, grade, HER-2 and Ki-67 over-expression, number of SLN+s. All factors, but Ki-67, retained in multivariate regressions were used to generate a predictive model with good discriminating power on both the training and the validation sets (AUC 0.73 and 0.75, respectively). Three patient groups were defined based on their risk to present additional axillary burden. Conclusions The model identifies SLN+-patients at low risk (≤15%) who could reasonably be spared ALND and those at high risk (>75%) who should receive ALND. For patients at intermediate risk, ALND appropriateness could be individually evaluated based on other clinico-pathological parameters.