Résumé
Purpose Predictive factors of non-sentinel lymph node (NSN) involvement at axillary lymph node dissection (ALND) have been studied in the case of sentinel node (SN) involvement, with validation of a nomogram. This nomogram is not accurate for SN micrometastasis. The purpose of our study was to determine a nomogram for predicting the likelihood of NSN involvement in breast cancer patients with a SN micrometastasis. Methods We collated 909 observations of SN micrometastases with additional ALND. Characteristics of the patients, tumours and SN were analysed. Results Involvement of SN was diagnosed 490 times (53.9%) with standard staining (HES) and 419 times solely on immunohistochemical analysis (IHC) (46.1%). NSN invasion was observed in 114 patients (12.5%), whereas 62.3% (71) had only one NSN involved and 37.7% (43) two or more NSN involved. In multivariate analysis, significant predictive factors were: tumour size (pT stage ≤10 mm or >11 and ≤20 or >20 mm [odds ratio (OR) 2.1 and 3.43], micrometastases detected by HES or IHC [OR 1.64], presence or absence of lymphovascular invasion (LVI) [OR 1.76], tumour histological type mixed or not [OR 2.64]. The rate and probability of NSN involvement with the model are given for 24 groups, with a representation by a nomogram. Conclusion One group, corresponding to 10.1% of the patients, was associated with a risk of NSN involvement of less than 5%, and five groups, corresponding to 29.8% of the patients, were associated with a risk ≤10%. Omission of ALND could be proposed with minimal risk for a low probability of NSN involvement.