The significance of extracapsular extension of axillary lymph node metastases in early-stage breast cancer

PURPOSE: To investigate if extracapsular extension (ECE) of axillary lymph node metastases predicts for a decreased rate of disease-free survival or an increased rate of regional recurrence of breast carcinoma. METHODS: The study population consisted of 368 patients with T1 or T2 breast cancer and pathologically-positive lymph nodes treated with breast-conserving therapy between 1968 and 1986. The median number of sampled lymph nodes was 10. Median follow-up time for the surviving patients was 139 months (range 70-244). Twenty percent of the patients were treated with supraclavicular RT, and 64% received both axillary and supraclavicular RT, with a median dose to the nodes of 45 Gy. The following factors were evaluated: presence of ECE, number of sampled lymph nodes (LN), number of involved LN, size of primary tumor, histologic grade of tumor, presence of lymphatic vessel invasion (LVI), presence of an extensive intraductal component (EIC), radiation dose, use of adjuvant chemotherapy, and age of patient. Recurrences were reported as the 5-year crude sites of first failure, and were divided into breast recurrences (LR), regional nodal failure (RNF, defined as isolated axillary, supraclavicular, or internal mammary recurrence), and distant metastases (DM). RESULTS: One hundred twenty-two patients (33%) had ECE and 246 patients did not. The median number of LN with ECE was 1 (range 1-10) and 20% of patients had ECE in > or =4 LN. Patients with ECE tended to be older (median age 51 vs. 47, p = 0.01), and had a higher number of involved LN (median 3 vs. 2, p = 0.005) than patients without ECE. Forty-three percent of patients with ECE had > or =4 involved LN compared to 15% of patients without ECE (p<0.0001). Models of ECE and the above factors revealed no significant correlation between ECE and either disease-free or overall survival. There was no statistically significant increase in local, regional nodal, or distant failures in patients with ECE as compared to patients without ECE. CONCLUSION: In this population of patients with nodal involvement, the presence of ECE correlates with the number of involved LN but does not appear to add predictive power to models of local, regional, or distant recurrence when the number of positive LN is included.