The objective of this study was to determine locoregional recurrence (LRR) patterns after mastectomy and doxorubicin-based chemotherapy to define subgroups of patients who might benefit from adjuvant irradiation.

A total of 1,031 patients were treated with mastectomy and doxorubicin-based chemotherapy without irradiation on five prospective trials. Median follow-up time was 116 months. Rates of isolated and total LRR (+/- distant metastasis) were calculated by Kaplan-Meier analysis.

The 10-year actuarial rates of isolated LRR were 4%, 10%, 21%, and 22% for patients with zero, one to three, four to nine, or >/= 10 involved nodes, respectively (P <.0001). Chest wall (68%) and supraclavicular nodes (41%) were the most common sites of LRR. T stage (P <.001), tumor size (P <.001), and >/= 2-mm extranodal extension (P <.001) were also predictive of LRR. Separate analysis was performed for patients with T1 or T2 primary disease and one to three involved nodes (n = 404). Those with fewer than 10 nodes examined were at increased risk of LRR compared with those with >/= 10 nodes examined (24% v 11%; P =.02). Patients with tumor size greater than 4.0 cm or extranodal extension >/= 2 mm experienced rates of isolated LRR in excess of 20%. Each of these factors continued to significantly predict for LRR in multivariate analysis by Cox logistic regression.

Patients with tumors >/= 4 cm or at least four involved nodes experience LRR rates in excess of 20% and should be offered adjuvant irradiation. Additionally, patients with one to three involved nodes and large tumors, extranodal extension >/= 2 mm, or inadequate axillary dissections experience high rates of LRR and may benefit from postmastectomy irradiation.