Patient Population

Patients with platinum-resistant EOC or PSC were enrolled. Platinum-resistant disease was defined as progression during or within 6 months after completing either first- or second-line platinum-based chemotherapy. Patients were also required to have subsequently progressed during, or within 3 months after discontinuing, treatment with topotecan or liposomal doxorubicin. Patients must have received no more than three prior treatment regimens, which could include chemotherapy, biologics, or other targeted therapies. They were also required to have measurable disease by Response Evaluation Criteria in Solid Tumors (RECIST)¹⁶ and an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1. Patients were excluded if they had brain metastases or a history of other malignancies within 5 years before study entry; uncontrolled blood pressure (> 150/100 mmHg); unstable angina; New York Heart Association grade 2 or greater congestive heart failure (CHF); prior liposomal doxorubicin therapy with an ejection fraction less than 50%; myocardial infarction, stroke, or transient ischemic attack within 6 months; clinically significant peripheral vascular disease; major surgical procedure within 28 days; history of abdominal fistula, GI perforation (GIP), or intra-abdominal abscess; or urine protein/creatinine ratio more than 1.0 at screening. Patients were not excluded from participation if they had signs or symptoms of intestinal obstruction or required parenteral hydration and/or nutrition. Eligibility criteria also included an absolute neutrophil count more than 1,500/μL; platelet count more than 75,000/μL; hemoglobin more than 8.5 g/dL; serum bilirubin less than 2.0× the upper limit of normal (ULN); alkaline phosphatase, AST, and ALT less than 2.5× ULN; serum creatinine less than 2.0 U/mL; international normalized ratio less than 1.5; and activated partial thromboplastin time less than 1.5× ULN (except for patients receiving anticoagulation therapy).

Study Design

This multicenter trial was to be conducted at 40 investigative sites in the United States and include approximately 120 patients. The trial was sponsored and monitored by Genentech Inc (South San Francisco, CA). A two-stage design was employed. If two or more of 35 enrolled patients achieved a confirmed CR or partial response (PR; stage 1), an additional 85 patients were to be enrolled (stage 2). Each patient provided written informed consent, and the trial was conducted in accordance with the Helsinki Declaration.

Bevacizumab 15 mg/kg was administered by intravenous infusion on the first day of each 3-week cycle until unacceptable toxicity or radiographic progression. The bevacizumab dosage was based on the patient's baseline weight at study entry and remained unchanged throughout the study. The initial dosage was administered over 90 ± 10 minutes; in subsequent cycles, infusions could be shortened to 60 minutes, and then to 30 minutes, as tolerated.

The primary objective was to evaluate the activity of bevacizumab in patients with platinum-resistant EOC or PSC, as measured by objective RR. Secondary objectives included estimating durations of PFS and OS, in addition to characterizing the safety of single-agent bevacizumab in this patient population.

Toxicity and Efficacy

At baseline, patients were evaluated by treatment history (including prior cancer and concomitant medications) and physical evaluation (vital signs, height, and weight), ECOG PS, CBC, chemistry panel, CA-125, serum pregnancy test, urine protein creatinine ratio (UPCR), and by computed tomography (CT) scan to define RECIST-measurable disease.¹⁶

Every 3 weeks and before each bevacizumab infusion, patients were reassessed for toxicity and evaluated by physical examination (including vital signs), CBC, and serum chemistry testing. Every 6 weeks, patients on study underwent UPCR evaluation, tumor assessment with CT scan (up to week 24), and CA-125 testing (until week 24). After week 24, patients were assessed by CT scan and CA-125 every 12 weeks until disease progression. Toxicity was reported using Common Terminology Criteria for Adverse Events (CTCAE) version 3 (http://ctep.cancer.gov/forms/CTCAEv3.pdf).

Statistical Analysis

The primary and secondary efficacy end points were analyzed using the intent-to-treat (ITT) patient population, based on patients who received a subject identification number from the Interactive Voice Response System. The primary efficacy end point was the objective RR as determined by an independent review facility (IRF). A two-stage design was used to test Ho: RR = 0.05 versus Ha: RR more than 0.05 at the .05 significance level. The null hypothesis was rejected if 12 or more of 120 patients achieved an objective response by the IRF. The protocol was modified between stages 1 and 2 to allow 67 additional patients into stage 2. This rule is adjusted to control the type I error resulting from protocol modification occurring with knowledge of the stage 1 RR; 95% Blyth-Still-Casella exact CI for objective RR was to be calculated.

Patients

Between January 26 and August 30, 2005, 44 patients were enrolled at 14 centers. Patients received at least one infusion of bevacizumab. As of August 2005, there were two objective responses in stage 1, and enrollment continued into stage 2. The sponsor closed the study early, after enrollment of 44 patients, because of a higher than expected incidence of GIP (11.4%).

Demographic and baseline characteristics are presented in Table 1. Patients had a mean age of 59.5 years, more than half had a baseline ECOG PS of 0, and 81.8% had serous tumors. All patients had platinum-resistant disease, and close to half (47.7%) of the patients had received three prior chemotherapy regimens.

Patients received a median of five (range, two to 16) bevacizumab doses. Twenty-four patients had one dose omission; one patient had a dose interruption because of grade 1 oral hypoaesthesia, and one had an infusion prematurely discontinued because of grade 3 hypersensitivity and urticaria.

Efficacy

Seven patients (15.9%; 95% CI, 7.2% to 29.0%) experienced a confirmed PR based on RECIST (Table 2). Four of these patients had received two prior chemotherapy regimens, and three patients had received three prior chemotherapy regimens. All seven patients were platinum resistant on the basis of progressing within 6 months of receiving first-line platinum-based chemotherapy. No patient experienced a CR; 27 (61.4%) had stable disease as their best response, 11 (25.0%) had stable disease for at least 12 weeks, and five (11.4%) had progressive disease. For the seven patients with an objective response, the median response duration was 4.2 months (range, 1.7 to 9.2 months).

Thirty-six patients died or experienced disease progression during the 15-month study period. Eight patients were excluded for being removed from the study before radiographic progression or death on study: four for a clinical impression of disease progression (usually based on a rising CA-125 level), three for adverse events, and one for early study closure. The median PFS was 4.4 months (95% CI, 3.1 to 5.5 months), with 27.8% of patients remaining progression free at 6 months. Of interest, we obtained an equivalent median PFS of 4.4 months by assessing progression on the basis of previously published CA-125 criteria (doubling of the CA-125 nadir value; 95% CI, 3.2 to 5.6 months; data not shown).¹⁷ The median survival duration was 10.7 months, with the longest survival time excluded at study termination at 14.6 months (Fig 1). At data cutoff (May 26, 2006), 25 patients had died. The median follow-up duration was 10.2 months. Nineteen survival dates were excluded because of study termination.

Toxicity

The sponsor halted the study because of a higher than expected incidence of GIP after 44 patients had received at least one bevacizumab dose. The most common reason for discontinuation was disease progression in 25 patients (65.9%). Eight patients (18.2%) discontinued because of an adverse event. One patient discontinued because of the early closure of the trial and had an ongoing PR. Overall, 17 patients (38.6%) had a grade 3 event, five (11.4%) a grade 4 event, and three (6.8%) a grade 5 event (Table 3). Serious adverse events occurred in 18 of 44 patients (40.9%; Table 4). Perforation and obstruction of the GI tract and arterial thromboembolic events (ATEs) were the most commonly reported serious adverse events.

Twenty-five patients died during the study, including 21 (47.7%) because of disease progression. The remaining four (9.1%) were attributed to adverse events: one from bowel obstruction related to metastatic disease, one from cerebrovascular ischemia, one from intestinal perforation, and one from sepsis due to an abdominal fistula. Three of these deaths were suspected to be related to bevacizumab treatment: one patient died of myocardial infarction and cerebrovascular ischemia, one of intestinal perforation, and one of convulsion and hypertensive encephalopathy.

Five patients (11%) developed an event attributed to GIP, one of which was grade 5. Perforation occurred 51 to 178 days after starting bevacizumab. All five patients had radiographic evidence of bowel involvement at study entry, and all had received three prior chemotherapy regimens. Each of these five patients had stable disease at the time of perforation. To assess potential radiographic risk factors for GIP, an IRF radiologist, blinded to patients' tumor responses and adverse events (including GIP), reviewed baseline radiographs from all study patients (Table 5). Radiographic variables for assessment of GIP risk included tumor involvement of the stomach, small bowel, or colon. In cases where the anatomic site of bowel involvement could not be accurately identified, a separate category of GI tumor involvement was used. Other radiographic variables included the presence of bowel obstruction, distention, or wall thickening. None of these variables seemed related to GIP. However, a trend toward increased frequency of GIP was observed for bowel wall thickening and bowel wall obstruction (Table 5). These associations were not statistically significant after accounting for multiple possible risk factors.¹⁸ There appeared to be an association between GIP and increased number of prior chemotherapy regimens. Specifically, five (23.8%) of 21 patients who received three prior chemotherapy regimens developed GIP compared with 0 of 23 patients who had received two prior chemotherapy regimens (P < .01; Table 5).

Five patients (11.3%) experienced bowel obstruction, in each case associated with disease progression. Four patients underwent surgical correction of the obstruction; two developed postoperative wound complications (one developed a grade 3 incision site hematoma requiring evacuation, and one developed grade 3 wound dehiscence). One patient developed bowel obstruction after being withdrawn from study because of hypertensive encephalopathy, and she subsequently died. This death was attributed to disease progression and bowel obstruction, although a contribution of the hypertensive event could not be ruled out.

Four ATE events were reported in three patients (6.8%), all of whom were younger than 65 years and had not received aspirin before their ATE event. One patient experienced a grade 3 silent myocardial infarction and grade 2 CHF after surgical correction of a bowel obstruction. A second patient with a history of deep venous thrombosis had an indwelling inferior vena cava filter and migratory thrombophlebitis at study entry. During the study, this patient developed a grade 3 venous thromboembolism, grade 5 cerebral ischemia, and myocardial infarction. The third patient experienced a grade 4 myocardial ischemia event, which was attributed to bevacizumab and to her underlying hypertension.

Fourteen patients (31.8%) experienced hypertension of any grade, most of which were grade 1 to 2 events; 4 (9.1%) experienced grade 3 hypertension, all of which resolved; no patient developed grade 4 hypertension. One patient experienced grade 5 hypertensive encephalopathy that presented with seizure, somnolence, blood pressure of 203/105 mmHg, epileptiform activity on EEG in the right occipital/parietal region, and magnetic resonance imaging changes in the right occipital lobe, most likely representing an area associated with hypertensive encephalopathy. The patient's death was attributed to disease progression and bowel obstruction, although a contribution of hypertensive encephalopathy could not be excluded.