Antiangiogenic Therapy for Glioblastoma: Complex Biology and Complicated Results.

Angiogenesis is one of the classic and diagnostic features of glioblastoma. These newly formed blood vessels are typically tortuous, forming vascular glomeruli, and they are notable for the lack of the tight junctions and complete pericyte coverage. This leads to peritumoral edema and extravasation of intravenous contrast on computed tomography or magnetic resonance imaging. Because angiogenesis is such an important aspect of glioblastoma biology, targeting this process has been considered a therapeutic priority. Vascular endothelial growth factor A is the most prominent and common mediator of angiogenesis, although there are a variety of other factors that have been identified in glioblastoma. Bevacizumab, a humanized monoclonal antibody against vascular endothelial growth factor A, showed activity in a variety of solid tumors. However, brain tumors were excluded from early studies of bevacizumab because of concerns regarding the risk of intracranial hemorrhage. A presentation in 2005 of a small number of patients with recurrent glioblastoma who were safely treated with bevacizumab led to larger formal clinical trials. These initial studies combined bevacizumab with irinotecan, emulating the successful treatment regimen in colorectal cancer, and demonstrated imaging response rates in the 30% to 40% range. These studies confirmed that the rate of intracranial hemorrhage was low and that the systemic toxicities were on par with this treatment regimen in other cancers. Two subsequent trials solidified these early results. The BRAIN Trial was a multicenter, randomized, noncomparative phase II study that treated patients with recurrent glioblastoma with bevacizumab either as a single agent or in combinationwith irinotecan. A parallel study was done at the National Institutes of Health, where patients were treated with bevacizumab in a single arm but at progression, irinotecan could be added. The results of these two studies were quite consistent with objective response rates ranging from 28% to 38%. The median overall survival (OS), measured from entry onto these studies, was similar. However, in the absence of a treatment arm without bevacizumab, assessment of the impact on OS was difficult. On the basis of the response rates, bevacizumab was granted accelerated approval by the Food and Drug Administration for treatment of patients with recurrent glioblastoma. Bevacizumab was then tested in newly diagnosed glioblastoma in two international placebo-controlled, double-blinded randomized phase III trials, AVAGlio and RTOG 0825. These studies added bevacizumab to the standard first-line treatment of glioblastoma, which combines external beam radiationwith concurrent daily temozolomide followed by six to 12 cycles of single-agent temozolomide. Both studies demonstrated prolonged progression-free survival (PFS) with bevacizumab; although because of differences in statistical design, the PFS difference was statistically significant in AVAGlio but not in RTOG 0825. However, neither study showed a difference in OS. A subsequent three-arm phase II trial in patients with recurrent glioblastoma, BELOB, randomly assigned patients to single-agent lomustine, single-agent bevacizumab, or the combination of lomustine and bevacizumab. This study enrolled approximately 50 patients per arm and was designed to look for a survival signal, designating the 9-month survival rate as the metric of success. Only the combination arm of bevacizumab and lomustine surpassed the preset threshold of a 9-month survival rate of 55%. The EORTC 26101 phase III trial was designed as the confirmatory trial. Patients with recurrent glioblastoma were randomized to either lomustine or the combination of lomustine with bevacizumab. The results were recently presented at the annual meeting of the Society for NeuroOncology. No survival difference was noted between the two treatment arms, although there was a difference in PFS. The article accompanying this editorial has the results of the GLARIUS trial, an open-label, randomized study that compared the standard treatment for newly diagnosed glioblastoma (chemoradiation with temozolomide, followed by maintenance temozolomide) with radiation followed by the combination of bevacizumab and irinotecan. Eligibility was restricted to patients with centrally confirmed glioblastoma that on molecular testing demonstrated that the promoter region of the O6-methylguanine DNA methyltransferase (MGMT) gene was unmethylated. These patients have a worse prognosis and the benefit of temozolomide is modest at best, allowing temozolomide to be excluded from the experimental arm. The primary end point of the trial was the progression-free rate at 6 months, a metric more commonly used for clinical trials for recurrent glioblastoma where statistical parameters are well established. The results did demonstrate a statistically significant difference in this metric, but the secondary end point of OSwas not different between the two treatment arms. There were, however, a high percentage of crossovers, with patients on the control (temozolomide) arm receiving bevacizumab at progression and, conversely, patients on the experimental (bevacizumab and irinotecan) arm receiving temozolomide at recurrence. How do we interpret this large composite of clinical trials? The gold standard benchmark of efficacy in cancer clinical trials is