Frequency and predictors of axillary lymph node metastases in invasive breast cancer.

BACKGROUND: The objectives of the present study were to evaluate the incidence and predictors of axillary lymph node metastases (ALNM) in patients with breast cancer, and to identify if axillary surgery could be safely omitted in selected patients.
METHODS: Between January 1996 and May 2000, 492 patients underwent 501 axillary lymph node dissections (ALND). The incidence of ALNM was correlated with clinical and pathological characteristics by univariate and multivariate analyses.
RESULTS: Axillary lymph node metastases were found in 41% (207/501) of cases. Univariate analysis showed that palpability of primary and axillary lymph node (ALN), pathological tumour size, grade, lymphovascular invasion (LVI) and multifocality or multicentricity were significant predictors of ALNM. By multivariate analysis, palpability of ALN, pathological tumour size, LVI and multifocality or multicentricity remained as independent predictors. Among the 431 cases without palpable ALN, no ALNM were found if the tumour was < or = 5 mm, non-multifocal or multicentric, and without LVI, or the tumour was a tubular or mucinous carcinoma < or = 15 mm (n = 21). The frequency of ALNM in the absence of the other risk factors was 11% (7/64) if the tumour size was > 5-10 mm, and 17% (19/113) if the tumour was > 10-20 mm. However, the incidence of ALNM was 72% for the 32 clinically node-negative cases with multifocal or multicentric tumour > or = 10 mm and LVI. Those patients with palpable ALN (n = 66) had a greater than 50% risk of ALNM.
CONCLUSIONS: Routine ALND could be omitted in clinically node-negative patients with either a < or = 5-mm, LVI-negative tumour, or a < or = 15-mm tubular or mucinous carcinoma. Axillary lymph node dissection is still useful for determining pathological nodal status in all other cases, and in most cases with palpable ALN, as a therapeutic manoeuvre.