Single‐institution retrospective review of patients with recurrent glioblastoma treated with bevacizumab in clinical practice

Abstract Background and aims This retrospective review of patients with recurrent glioblastoma treated at the Preston Robert Tisch Brain Tumor Center investigated treatment patterns, survival, and safety with bevacizumab in a real‐world setting. Methods Adult patients with glioblastoma who initiated bevacizumab at disease progression between January 1, 2009, and May 14, 2012, were included. A Kaplan‐Meier estimator was used to describe overall survival (OS), progression‐free survival (PFS), and time to greater than or equal to 20% reduction in Karnofsky Performance Status (KPS). The effect of baseline demographic and clinical factors on survival was examined using a Cox proportional hazards model. Adverse event (AE) data were collected. Results Seventy‐four patients, with a median age of 59 years, were included in this cohort. Between bevacizumab initiation and first failure, defined as the first disease progression after bevacizumab initiation, biweekly bevacizumab and bevacizumab/irinotecan were the most frequently prescribed regimens. Median duration of bevacizumab treatment until failure was 6.4 months (range, 0.5‐58.7). Median OS and PFS from bevacizumab initiation were 11.1 months (95% confidence interval [CI], 7.3‐13.4) and 6.4 months (95% CI, 3.9‐8.5), respectively. Median time to greater than or equal to 20% reduction in KPS was 29.3 months (95% CI, 13.8‐∞). Lack of corticosteroid usage at the start of bevacizumab therapy was associated with both longer OS and PFS, with a median OS of 13.2 months (95% CI, 8.6‐16.6) in patients who did not initially require corticosteroids versus 7.2 months (95% CI, 4.8‐12.5) in those who did (P = 0.0382, log‐rank), while median PFS values were 8.6 months (95% CI, 4.6‐9.7) and 3.7 months (95% CI, 2.7‐6.6), respectively (P = 0.0243, log‐rank). Treatment failure occurred in 70 patients; 47 of whom received salvage therapy, and most frequently bevacizumab/carboplatin (7/47; 14.9%). Thirteen patients (18%) experienced a grade 3 AE of special interest for bevacizumab. Conclusions Treatment patterns and outcomes for patients with recurrent glioblastoma receiving bevacizumab in a real‐world setting were comparable with those reported in prospective clinical trials.

Conclusions: Treatment patterns and outcomes for patients with recurrent glioblastoma receiving bevacizumab in a real-world setting were comparable with those reported in prospective clinical trials. KEYWORDS bevacizumab, real-world setting, recurrent glioblastoma, survival, treatment patterns Glioblastomas are the most common and most aggressive form of primary brain tumors in adults. 1 Patients with glioblastoma have a very poor prognosis because of the high propensity for relapse, 2 with reported median survival times for patients with recurrent disease of just 3 to 9 months. 3,4 The angiogenic factor vascular endothelial growth factor (VEGF) is expressed at high levels in glioblastoma relative to other cancer types 4 ; thus, inhibitors of VEGF have been investigated for the treatment of glioblastoma. 5 In 2009, the anti-VEGF monoclonal antibody bevacizumab received accelerated United States Food and Drug Administration (FDA) approval as a single agent for the treatment of recurrent glioblastoma. 6 Approval of bevacizumab was based on durable objective response rates (ORRs) and 6-month progression-free survival (PFS) data obtained from two single-arm phase 2 studies.
The first study evaluated the activity of bevacizumab monotherapy followed by bevacizumab plus irinotecan at disease progression (PD). The 6-month PFS rate was 29.0%, median overall survival (OS) was 31 weeks, and radiographic response was recorded in 71% and 35% of patients, based on Levin and Macdonald criteria, respectively. 7 A companion study assessed the efficacy of bevacizumab alone and in combination with irinotecan following PD. In the bevacizumab monotherapy arm, the 6-month PFS rate was 42.6%, median OS was 9.2 months, and the ORR was 28.2%. 8 Grade 3/4 adverse events (AEs) across the two studies were mostly nonhematologic and included hypertension and thromboembolic events. 7,8 Patients enrolled in randomized clinical trials do not always reflect real-world populations. Real-world data can provide valuable insights into treatment patterns as well as therapeutic benefits for patients in routine clinical practice. However, there are few realworld studies evaluating survival outcomes among patients with recurrent glioblastoma. A recent retrospective, online chartabstraction study examined patterns of treatment, outcomes, and use of cancer-related health-care resource for 503 patients with glioblastoma treated in real-world clinical practices in the United States, but only 11 patients (2.2%) had recurrent disease. 9 We retrospectively reviewed treatment patterns, survival, and safety outcomes for patients with recurrent glioblastoma receiving bevacizumab-containing regimens outside of a clinical protocol at a large specialist brain cancer center in the United States. The aims of this retrospective review were to describe patterns of treatment, survival outcomes, and toxic effects in patients who received bevacizumab at the time of recurrence and were previously bevacizumab-naïve. We also explored the associations of baseline demographic and disease characteristics on survival outcomes in these patients.

| Survival
The median duration of follow-up at the time of data cut off was

| Performance status
Of the 61 patients with a baseline KPS available within 2 months prior to starting bevacizumab and at least one follow-up KPS assessment, 17 patients (27.9%) experienced a greater than or equal to 20% reduction in KPS from bevacizumab initiation ( Figure 2). The median time to greater than or equal to 20% reduction in KPS from bevacizumab initiation was 29.3 months (95% CI, 13.8-∞), and the proportion of patients without a KPS reduction of greater than or equal to 20% from bevacizumab initiation was 77.4%, 51.8%, and 41.4% at 12, 24, and 36 months, respectively.

| Corticosteroid use during bevacizumab treatment
Overall, 31/74 patients (41.9%) were receiving corticosteroids at the time of bevacizumab initiation. Of these, 17 patients (54.8%) discontinued corticosteroids during bevacizumab therapy and one patient (3.2%) discontinued corticosteroids while not receiving

| Safety
AEs reported at a cumulative incidence of greater than or equal to 10% across all grades during bevacizumab therapy (including salvage bevacizumab treatment) are displayed in Table 3. However, assessment of attribution to bevacizumab or other therapy was not com-

| Patterns of treatment until bevacizumab failure
Treatment failure after initial bevacizumab occurred in 70 patients.
Sixty of these patients (85.7%) received bevacizumab-based therapy until failure, including 55 patients (78.6%) who had PD and five patients (7.1%) who died ( Figure 3 and Table S1). Nine of the 70 patients (12.9%) were off all therapy at the time of failure. Six of these patients stopped the bevacizumab-based regimen because of AEs, two patients had intercurrent illnesses, and one patient refused further treatment. One of the 70 patients (1.4%) received bevacizumab-based therapy followed by nonbevacizumab therapy prior to failure.
After initial bevacizumab treatment failure, 47 patients received salvage therapy (Table S2)    In contrast, the addition of bevacizumab to lomustine did not extend   Our study is limited by its retrospective nature, the inclusion of patients from only a single center, and the lack of data on the molecular genetics of the tumors. The wide variety of bevacizumabcontaining regimens received by the patients, and the fact that patient-specific treatment was not mandated by a study protocol, also limit the findings. However, in terms of survival, corticosteroid usage, and baseline KPS, our results compare favorably with those published in prospective clinical trials, which may not always accurately reflect real-world populations.
Taken together, these data indicate an important role for bevacizumab as part of the treatment modality for recurrent glioblastoma. Bevacizumab-based regimens were the most commonly used regimens both prior to, and following, first bevacizumab failure. Patient outcomes in this real-world setting were comparable with those reported in prospective clinical trials, and bevacizumab was generally well tolerated.

FUNDING
The study was funded by Genentech, Inc. Genentech, Inc was involved in the design, writing of the report, and in the decision to submit the report for publication.

CONFLICTS OF INTEREST
Annick Desjardins has received grants or research support from Genentech, PTC Therapeutics, Celldex, Triphase Research and Development Corp, Eli Lilly and Co, Eisai, Symphogen A/S, Pfizer, and Orbus Therapeutic and is an advisory board member for Genentech; Arliene Ravelo is an employee of Genentech and owns stock options in Roche; Nicolas Sommer is an employee of Genentech and owns stock options in Roche; and the remaining authors have no conflicts of interest to declare.