Phase II Trial of Bevacizumab in Combination with Weekly Docetaxel in Metastatic Breast Cancer Patients

Bevacizumab (Avastin; Genentech, San Francisco, CA) is a recombinant humanized anti–vascular endothelial growth factor (VEGF) antibody that specifically inhibits VEGF, has activity in multiple cancer cell lines, and is synergistic with several chemotherapeutic agents (1, 2). Phase I trials of bevacizumab alone or with chemotherapy show the drug has acceptable toxicities at doses up to 10 mg/kg every 2 weeks or 15 mg/kg every 3 weeks (3, 4). Randomized trials in metastatic colorectal cancer (5, 6) and non–small cell lung cancer patients (7) show that bevacizumab added to chemotherapy improves the response rate, progression-free survival (PFS), and overall survival relative to chemotherapy alone.

With respect to breast cancer, bevacizumab alone has very modest antitumor activity in pretreated metastatic breast cancer (MBC; ref. 8). In a phase III trial in MBC patients who had received prior chemotherapy for metastatic disease, the addition of bevacizumab to capecitabine did not improve PFS or overall survival, but was associated with a doubling of the response rate (9). More recently, a phase III trial in MBC patients without prior treatment for metastases showed statistically significant improvements in response rate, PFS, and overall survival for the combination of weekly paclitaxel and bevacizumab versus paclitaxel alone (10).

Docetaxel is among the most active drugs in breast cancer. Docetaxel also has antiangiogenic activity and potentiates the activity of antiangiogenic drugs (11–13). In vitro and in vivo preclinical studies show that VEGF protects endothelial cells against the antiangiogenic effects of docetaxel and this effect is inhibited by anti-VEGF monoclonal antibody (14). These observations provided the rationale for initiating a phase II trial with docetaxel and bevacizumab as either first- or second-line treatment in MBC patients. In an exploratory analysis, the relationship between clinical outcomes and plasma endothelial and adhesion cell markers and was evaluated.

The trial was done at The Ohio State University and University of Colorado. MBC patients over the age of 18 years with histologically proven breast cancer with evidence of MBC and the following characteristics were eligible: The presence of at least one metastatic site measurable by Response Evaluation Criteria in Solid Tumors (15), Eastern Cooperative Oncology Group performance status ≤2, leukocytes ≥3,000/μL, absolute neutrophil count ≥1,500/μL, platelets ≥ 100,000/μL, total bilirubin ≤ institutional upper limit of normal, transaminases ≤2.5 times upper limit of normal, and creatinine ≤ upper limit of normal or creatinine clearance ≥60 mL/min for patients with creatinine above upper limit of normal. No more than one prior chemotherapy regimen for MBC was permitted. In the event that prior therapy contained a taxane, at least 6 months must have elapsed since completing prior taxane treatment. Written informed consent was required and the institutional review boards of the participating centers approved the study.

Patients were excluded for having a known allergy to polysorbate 80 or docetaxel, history of uncontrolled seizures, brain metastases, stroke, bleeding diathesis, coagulopathy, proteinuria >500 mg/24-hour urine collection, or a recent history (within 10 days before study treatment) of aspirin or anticoagulant therapy. Other exclusions were uncontrolled infection, symptomatic cardiac failure, unstable angina, cardiac arrhythmia, and uncontrolled hypertension. In addition, patients were not eligible if they had a major surgical procedure, open biopsy, nonhealing wound, or significant traumatic injury within 28 days, or placement of a vascular access device within 7 days before study treatment.

Study treatment. Patients received 10 mg/kg bevacizumab i.v. on days 1 and 15 in combination with 35 mg/m² i.v. of docetaxel on days 1, 8, and 15 repeated every 28 days or one treatment cycle. Patients having clinical benefit after six cycles received bevacizumab alone until disease progression. The first dose of bevacizumab was administered over 90 minutes with careful monitoring of vital signs every 15 minutes during the infusion, and every 30 minutes for 3 hours after the infusion. If the first infusion was without incident, the second dose was administered over 60 minutes and subsequent doses over 30 minutes. A 1-hour docetaxel infusion was administered after the bevacizumab infusion. Oral dexamethasone 8 mg was given 12 hours before, immediately before, and 12 hours after the docetaxel infusion. If the patient developed febrile neutropenia in any treatment cycle, granulocyte colony-stimulating factor (Neupogen) was allowed in the subsequent cycles at the discretion of the treating physician.

Docetaxel dose reductions of 15% were permitted for febrile neutropenia or grade 3 or 4 nonhematologic toxicity. Continued study treatment was contingent upon all toxicities resolving to grade ≤1 within 2 weeks. If toxicity failed to resolve within 2 weeks, the patient was removed from study. There was no modification of the bevacizumab dose. A urine dipstick for proteinuria was done before each treatment cycle. Any patient developing an increase in proteinuria had a 24-hour urine collection for total protein. Patients who developed >2 g proteinuria/24 hours did not receive further bevacizumab until the proteinuria was <2 g/24 hour. Bevacizumab was discontinued in patients who developed grade 4 hypertension, or uncontrolled hypertension despite oral antihypertensives. Bevacizumab was held for ≥grade 3 liver function tests until the abnormalities recovered to ≤grade 1. If there was a recurrence of ≥grade 3 liver function tests, bevacizumab was discontinued.

Study evaluations. The baseline evaluation included complete history and physical examination, assessment of performance status, complete blood count and differential, metabolic profile, coagulation studies, electrocardiogram, urine dipstick, and serum pregnancy test in women of childbearing age. Computed tomography scans of head, chest, abdomen, and bone scan comprised the baseline assessment. Response to treatment was evaluated every two treatment cycles using Response Evaluation Criteria in Solid Tumors (15). All tumor responses were confirmed by repeat imaging 4 to 6 weeks after the initial documentation. Toxicities were graded using the Common Toxicity Criteria version 2.0.

Biological correlates. Plasma endothelial and adhesion cell markers, including E-selectin, P-selectin, immune cell adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1, platelet derived growth factor, fibroblast-derived growth factor, matrix metalloproteinase-2, and matrix metalloproteinase-9 were evaluated at baseline (within 14 days before treatment) and after one cycle (pretreatment cycle 2, day 1). The assays were done in the Translational Research Laboratories of the University of Colorado Comprehensive Cancer Center under the direction of Dr. S. Gail Eckhardt. Each marker was assayed in triplicate wells using commercially available ELISA kits according to the instructions of the manufacturer. The interday and intraday precision and the coefficient of variation did not exceed 20%.

Statistical considerations. The primary objectives were to assess toxicity, response rate, and PFS. A secondary end point was to explore the relationship between plasma endothelial and adhesion cell markers and response rate and PFS. PFS was estimated using Kaplan-Meier curves. Logistic regression was used to evaluate models relating plasma cell markers to response. Cox proportional hazard models investigated relationships between plasma markers and PFS.

The trial used Simon's (16) optimal two-stage design for phase II trials. A response rate of ≥60%, which is an ∼50% improvement over patients receiving weekly docetaxel alone (17, 18), was considered sufficient to proceed with further investigation of the combination, whereas a response rate of ≤35% would be of little clinical interest. If seven or more responses were observed in the first 16 patients, 11 additional patients were enrolled (p₀ = 0.35, p₁ = 0.6, α = 0.10, β = 0.10). Further evaluation of the drug combination would be justified if a minimum of 13 patients responded to treatment.

Patient characteristics. Between August 2002 and February 2004, 27 patients enrolled with the following characteristics (Table 1): median age 51 years (range 39-68 years), median performance status 1 (range 0-2), 23 (85%) postmenopausal, and 16 (59%) estrogen receptor positive. Twenty-one (78%) patients received adjuvant chemotherapy. Nineteen (70%) received prior anthracyclines in either the adjuvant (n = 16) or metastatic (n = 3) setting; 11 (41%) patients received prior taxanes in either the adjuvant (n = 8) or metastatic (n = 3) setting. Twenty-one (78%) patients had no prior chemotherapy for MBC.

Treatment. One hundred fifty-eight treatment cycles were administered with a median of six cycles (range 1-15 cycles) per patient. Thirteen (48%) patients completed all six planned cycles, and 11 (41%) of them went on to receive bevacizumab alone for a median of two cycles (range 1-9 cycles).

Toxicity. Most of the toxicities were ≤grade 2 (Table 2). One (4%) patient had grade 3 hypertension, one (4%) patient had febrile neutropenia, and two (7%) patients had docetaxel dose reductions. Nine (33%) patients discontinued treatment for the following reasons: two (7%)—pulmonary embolism, one (4%)—serious bacterial infection, one (4%)—pleural effusion that required pleurodesis, one (4%)—grade 3 hypertension, one (4%)—persistent dyspnea of undetermined etiology, one (4%)—nonresolving neuropathy, one (4%)—required hip arthroplasty, and one (4%)—required radiation for bone metastases. There were no treatment-related deaths.

Efficacy. All patients were evaluable for efficacy analysis based on intent-to-treat. Fourteen (52%) patients had a partial response, nine (33%) had stable disease, and two (7%) had progressive disease (Table 3). The overall response rate was 52% [95% confidence interval (95% CI), 32-71%]. The median duration of response was 6.0 months (95% CI, 4.6-6.5 months), and the median PFS was 7.5 months (95% CI, 6.2-8.3 months; Fig. 1). The median duration of stable disease in nine patients was 5.3 months (range 2.9-8.0 months).

Endothelial and adhesion cell markers. Plasma samples were available in 21 (78%) patients pretreatment and 18 (67%) patients after cycle 1. Pretreatment and the decrease from pretreatment to after cycle 1 of E-selectin and ICAM-1 were the only markers that were associated with response in univariate analyses (Tables 4 and 5). Due to the small sample size, multivariate logistic regression models were limited to no more than two predictors at once. Pretreatment E-selectin maintained significance after controlling for performance status [odds ratio (OR), 1.6; 95% CI, 1.0-2.5; P = 0.05], age (OR, 1.6; 95% CI, 1.0-2.6; P = 0.05), estrogen receptor negativity (OR, 1.8; 95% CI, 1.0-3.0; P = 0.04), and disease-free interval (OR, 1.6; 95% CI, 1.0-2.5; P = 0.05). Likewise, the decrease in E-selectin after cycle 1 persisted after controlling for performance status (OR, 0.1; 95% CI, 0.0-0.9; P = 0.04), age (OR, 0.1; 95% CI, 0.0-0.9; P = 0.04), estrogen receptor negativity (OR, 0.1; 95% CI, 0.0-0.8; P = 0.04), visceral disease (OR, 1.0; 95% CI, 0.0-1.0; P = 0.04), and disease-free interval (OR, 0.1; 95% CI, 0.1-0.9; P = 0.03). In contrast, neither pretreatment nor the change in ICAM-1 after cycle 1 was significant when adjusting for these factors.

Weekly docetaxel and bevacizumab is active in MBC with the majority of side effects consistent with those expected from weekly docetaxel, including fatigue, nail changes, lacrimation, and mild to moderate neutropenia. Most of the adverse effects attributable to bevacizumab were minor and similar to those observed in other clinical trials. Notable was the absence of serious hypertension with only one (4%) patient developing grade 3 hypertension. This contrasts with the randomized trials in MBC (9, 10) and other solid tumors (5, 19), where grade 3 hypertension was observed in 11% to 18% of patients. However, the trial sample size was relatively small with resultant wide confidence intervals surrounding estimates of grade 3 hypertension.

Table 6 describes other trials of bevacizumab with chemotherapy in MBC. Eastern Cooperative Oncology Group 2100 is the largest randomized trial and included only patients without prior chemotherapy (10). The observed increase in PFS (hazard ratio, 0.49; 95% CI, 0.41-0.61; P < 0.001) and overall survival (hazard ratio, 0.67; 95% CI, 0.49-0.92; P = 0.01) is consistent with randomized trials in metastatic colon cancer (5, 6) and advanced non–small cell lung cancer (7) in patients without prior chemotherapy. However, when bevacizumab was combined with capecitabine, there was a doubling of the response rate but no improvement in PFS or overall survival in patients, most of whom received prior chemotherapy for metastases (9).

Currently, there is no defined marker that predicts for clinical benefit to bevacizumab, akin to HER-2 neu overexpression predicting response to trastuzumab (20, 21). In a retrospective subset analysis in a randomized trial (9), VEGF expression measured by in situ hybridization in primary breast cancers did not correlate with response to bevacizumab (22). E-selectin is an endothelial cell–specific membrane glycoprotein implicated in angiogenesis, the activation of mitogen-activated protein kinase pathways, and tumor metastases (23–27). In MBC patients, higher circulating E-selectin is associated with liver metastases, higher-grade tumors, and estrogen receptor–negative tumors (28–32). Likewise, higher circulating E-selectin may be an independent adverse prognostic factor in node-negative breast cancer patients (33, 34). Taken together, these studies suggest that higher pretreatment E-selectin may be a marker of angiogenesis as it relates to increasing tumor burden, more aggressive tumor biology, or both.

E-selectin is required for the antiangiogenic activity of endostatin, a potent angiogenesis inhibitor, and may predict or modulate endostatin efficacy in antiangiogenic therapy (35). Few studies have examined the role of E-selectin to predict response to therapy and none has included bevacizumab or antiangiogenic drugs (36, 37). In a preliminary hypothesis-generating exploratory analysis, we sought to identify whether plasma endothelial and adhesion cell markers might predict response to bevacizumab and weekly docetaxel. Of the markers that were measured, higher-pretreatment E-selectin and ICAM-1, and a greater decline in these markers after one treatment cycle, were significantly associated with response in univariate analyses. However, none of the markers correlated with PFS (data not shown).

Because of the relatively small sample size, multivariate models were limited to no more than two variables. E-selectin, but not ICAM-1, retained significance after controlling for other variables thought to be associated with response to chemotherapy in MBC patients (38). The ORs corresponding to E-selectin were consistent in each set of multivariate models. This suggests that E-selectin may be an independent predictive factor for response. However, because of the trial design, it was not possible to determine whether E-selectin predicts response to bevacizumab or docetaxel alone.

The analysis of E-selectin has several limitations, including the small sample size, missing data points, the multiple markers evaluated, and the lack of correlation to PFS. Nonetheless, the preclinical data supporting the role of E-selectin in angiogenesis, and the preliminary results of the current trial, justify larger prospective trials evaluating E-selectin as a marker of response bevacizumab-containing chemotherapy.

Our results support the Eastern Cooperative Oncology Group 2100 trial (10) and add important information about the safety and efficacy of bevacizumab in combination with weekly docetaxel. Trials are planned to evaluate bevacizumab in adjuvant setting in which the magnitude of the benefits may be greatest.