A phase II study of preoperative (preop) bevacizumab (bev) followed by dose-dense (dd) doxorubicin (A)/cyclophosphamide (C)/paclitaxel (T) in combination with bev in HER2-negative operable breast cancer (BC).

Abstract

1026 Background: Two recent preop studies evaluating bev showed conflicting results, particularly in hormone receptor (HR)+ BC. Identification of predictive markers and their relationship to the pharmacodynamic effects of bev would facilitate the identification of BCs most likely to benefit from bev. To accomplish these goals, we conducted a unique preop trial with a run-in of single agent bev followed by ddACT with bev in two cohorts, one with HR+HER2– BC, and a smaller triple negative (TN) cohort. Methods: Pts with HR+, HER2– or TN BC were eligible if their tumor (T) was ≥1.5 cm and high grade, or had axillary LN involvement; or if T≥2.5cm and was low/intermediate grade. Treatment consisted of a single dose of bev 10 mg/kg, followed two wks later by A 60 mg/m2 and C 600 mg/m2 with bev 10 mg/kg q2 wks x 4, followed by T 175 mg/m2 with bev 10 mg/kg q2 wks x 3, followed by T 175 mg/m2 x1. Research core biopsies and interstitial fluid pressure (IFP) were assessed pre- and post- bev alone. Pathologic response was confirmed centrally and Miller-Payne (MP) was assessed. Results: 84 pts with HR+ and 20 pts with TN breast cancer were enrolled. Amongst HR+ pts, 74 had surgical tissue centrally reviewed, and 6 (8%) had a pCR. Amongst TN pts, 18 pts had tissue centrally reviewed and 8 (44%) had a pCR. Grade was found to predict MP response in both HR+ and TN pts (p=0.001). Several biomarkers were evaluated as predictors of response to bev. Baseline sVEGFR1 correlated with MP response to treatment among TN pts (p=0.015). Single-agent bev reduced the mean vascular density by 18.5% (p=0.049) in HR+ patients and the mean IFP in the overall cohort and HR+ patients by 20 (p=0.020) and 24.5% (p=0.001), respectively. The reductions in IFP correlated with higher levels of sVEGFR2 (p=0.003). The IFP decreased > 50% in 24/65 pts and did not change in others. Gene expression profiling by PAM50 is underway. Conclusions: The addition of bev to preop chemotherapy is well tolerated. Tumor grade appears to predict MP response in HR+ and TN tumors, and sVEGFR1 may be a predictor of MP response to bev in TN tumors. Further work for biomarker predictors of response to bev is ongoing.