Abstract

Background The addition of bevacizumab to chemotherapy improves progression-free survival in metastatic breast cancer and pathological complete response rates in the neoadjuvant setting. Micrometastases are dependent on angiogenesis, suggesting that patients might benefit from anti-angiogenic strategies in the adjuvant setting. We therefore assessed the addition of bevacizumab to chemotherapy in the adjuvant setting for women with triple-negative breast cancer. Methods For this open-label, randomised phase 3 trial we recruited patients with centrally confirmed triple-negative operable primary invasive breast cancer from 360 sites in 37 countries. We randomly allocated patients aged 18 years or older (1:1 with block randomisation; stratified by nodal status, chemotherapy [with an anthracycline, taxane, or both], hormone receptor status [negative vs low], and type of surgery) to receive a minimum of four cycles of chemotherapy either alone or with bevacizumab (equivalent of 5 mg/kg every week for 1 year). The primary endpoint was invasive disease-free survival (IDFS). Efficacy analyses were based on the intention-to-treat population, safety analyses were done on all patients who received at least one dose of study drug, and plasma biomarker analyses were done on all treated patients consenting to biomarker analyses and providing a measurable baseline plasma sample. This trial is registered with ClinicalTrials.gov, number NCT00528567. Findings Between Dec 3, 2007, and March 8, 2010, we randomly assigned 1290 patients to receive chemotherapy alone and 1301 to receive bevacizumab plus chemotherapy. Most patients received anthracycline-containing therapy; 1638 (63%) of the 2591 patients had node-negative disease. At the time of analysis of IDFS, median follow-up was 31·5 months (IQR 25·6–36·8) in the chemotherapy-alone group and 32·0 months (27·5–36·9) in the bevacizumab group. At the time of the primary analysis, IDFS events had been reported in 205 patients (16%) in the chemotherapy-alone group and in 188 patients (14%) in the bevacizumab group (hazard ratio [HR] in stratified log-rank analysis 0·87, 95% CI 0·72–1·07; p=0·18). 3-year IDFS was 82·7% (95% CI 80·5–85·0) with chemotherapy alone and 83·7% (81·4–86·0) with bevacizumab and chemotherapy. After 200 deaths, no difference in overall survival was noted between the groups (HR 0·84, 95% CI 0·64–1·12; p=0·23). Exploratory biomarker assessment suggests that patients with high pre-treatment plasma VEGFR-2 might benefit from the addition of bevacizumab (Cox interaction test p=0·029). Use of bevacizumab versus chemotherapy alone was associated with increased incidences of grade 3 or worse hypertension (154 patients [12%] vs eight patients [1%]), severe cardiac events occurring at any point during the 18-month safety reporting period (19 [1%] vs two [<0·5%]), and treatment discontinuation (bevacizumab, chemotherapy, or both; 256 [20%] vs 30 [2%]); we recorded no increase in fatal adverse events with bevacizumab (four [<0·5%] vs three [<0·5%]). Interpretation Bevacizumab cannot be recommended as adjuvant treatment in unselected patients with triple-negative breast cancer. Further follow-up is needed to assess the potential effect of bevacizumab on overall survival. Funding F Hoffmann-La Roche.