Kantar Health Blog

Will conflicting data for Avastin in glioblastoma bring on brain fog amongst physicians?

by Stephanie Hawthorne | Jun 2, 2013
Stephanie Hawthorne
Co-authored by Arnold DuBell, PhD, MBA, Associate Consultant

Glioblastoma has an angiogenic component, as the disease manifests itself as a focal lesion with central necrosis surrounded by an angiogenic tumor rim1. Genentech and Roche took advantage of this with the data from the phase II BRAIN study (AVF3708g), which suggested that single-agent Avastin® (bevacizumab) was active, with a six-month PFS rate of 42.6% and was slightly higher (50.3%) in combination with irinotecan2. Based on this data, the U.S. FDA conditionally approved the use of Avastin for glioblastoma patients with progressive disease.

As a condition to convert this approval to a full approval, the companies agreed to perform a confirmatory Phase III trial; moreover, the European Medicines Agency (EMA) chose not to approve Avastin based solely on the BRAIN study results, and so any European approval for Avastin in glioblastoma also relies on the results of Phase III studies. Initial results of the Roche-sponsored AVAglio (BO21990) study were presented in 2012 at the Annual Society for Neuro-Oncology meeting, which indicated that the addition of Avastin to standard therapy (temozolomide plus radiotherapy followed by temozolomide maintenance) significantly improved progression-free survival (mPFS: 10.6 months versus 6.2 months, HR 0.64, p<0.0001). However, this was coupled with news that the co-primary endpoint of overall survival (OS) had not reached significance (HR 0.89, p=0.2135)3. Anticipation was high at the 2013 American Society of Clinical Oncology (ASCO) annual meeting amongst neuro-oncologists for several reasons: final OS data were presented for AVAglio, and the initial data were presented for a second phase III trial sponsored by the Radiation Therapy Oncology Group (RTOG 0825) which also compared Avastin to standard therapy in newly-diagnosed glioblastoma patients. Both data may have muddied the picture more for Avastin in glioblastoma than it helped.

Dr. Wick presented the final AVAglio data (Abstract 2002), which confirmed that the addition of Avastin did not improve OS (mOS: 16.8 months versus 16.7 months, HR 0.88, p=0.0987). Moreover, Dr. Henricksson (Abstract 2005) presented further data that showed that patients treated with Avastin in AVAglio experienced improved quality of life as measured by five scales of the QLQ-C30 and BN20 survey instruments: global health status (HR 0.64, p<0.0001), physical functioning (HR 0.70, p<0.0001), social functioning (HR 0.63, p<0.0001), motor dysfunction (HR 0.67, p<0.0001), and communication deficit (HR 0.67, p<0.0001).

Confusing matters was RTOG 0825, which was presented at the plenary session (Gilbert, Abstract 1). The trial was designed similarly to AVAglio, in that newly-diagnosed patients were randomized to receive temozolomide-based radiotherapy with or without Avastin followed by maintenance therapy with either temozolomide or temozolomide plus Avastin. The discussant (Dr. Fine) noted that the two trials had similar distributions of patient characteristics such as performance status, resection type and MGMT methylation status. Unfortunately, the similarities end there. Although the addition of Avastin numerically improved PFS (mPFS: 10.7 months versus 7.3 months, HR 0.79, p=0.007), it did not meet the pre-specified threshold of a 30% reduction in the hazard of failure. Similar to AVAglio, Avastin did not meet its co-primary endpoint of OS (mOS: 15.7 months versus 16.1 months, HR 1.13, p=0.21)

Avastin administration was associated with an increase in several select Grade 3 or 4 adverse events: hypertension (4.6% versus 1.0%), deep-vein thrombosis/pulmonary embolism (9.9% versus 7.7%), wound issues (2.3% versus 1.0%), GI perforations (1.3% versus 0.7%), hemorrhage (1.3% versus 1.0%) and neutropenia (15.1% versus 7.3%).

Health-related quality of life (HRQoL) analyses were a secondary endpoint of RTOG 0825. Furthering the dissimilarity to AVAglio, Dr. Armstrong reported that Avastin was associated with declines in cognitive function (Interaction effect = 0.009), motor dysfunction (Interaction effect = 0.014), and communication deficits (Interaction effect = 0.003; Abstract 2003). Moreover, Dr. Wefel reported that Avastin was associated with greater rates of neurocognitive decline (CTB Composite, p=0.02; Abstract 2004). There are a variety of reasons for these differences as noted by the plenary session discussant, Dr. Fine, and included that these were subset analyses – neither trial had a full set of HRQoL data – and unspecified differences in how these data were analyzed.

Based on the combined data set from AVAglio and RTOG 0825, it is clear that Avastin has only a limited role for newly-diagnosed patients. Dr. Gilbert noted in a post-plenary discussion that there is “anecdotal” experience that Avastin may be effective in the small population of patients with high edema.

The question remains what might happen with the current U.S. FDA conditional approval for Avastin’s use in relapsed or refractory patients. At the post-plenary discussion, Drs. Gilbert and Fine noted that they use Avastin for their relapsed patients, and would probably continue to do so. Their stated rationale was that there are no approved agents other than Avastin for these patients for whom prospects were so dire. There are only two agents who have active Phase III trials, rindopepimut (Celldex Therapeutics) and DC-Vax-L® (Northwest Biotherapeutics). Avastin differs from other recently failed trials such as cilengitide’s (Merck KGaA) CENTRIC trial (Stupp, Abstract LBA2009) in that a PFS benefit has been seen.

The discussant to the RTOG 0825 data, Dr. Fine, provided a biological rationale for continuing to use Avastin in relapsed patients. As alluded above, there are two predominant growth patterns within glioblastoma: VEGF-driven and VEGF-independent parts of the tumor. These have histological components, as the VEGF-driven tissues are central masses while the VEGF-independent tissues appear as peripheral flares. Dr. Fine provided data that in newly-diagnosed GBM, the tumor lesions appear to be much less vascularized, which may explain the lack of efficacy of anti-angiogenic agents in this setting. Additionally, in patients treated with front-line Avastin, the VEGF-independent peripheral flares predominate, which would render any recurrences insensitive to angiogenic inhibition as well. In contrast, VEGF-driven central masses predominate more in patients who relapse after cytotoxic therapy, which may explain the sensitivity of these tumors to Avastin.

There is sure to be a high level of debate about the fate of Avastin in GBM, and Kantar Health believes that although the FDA will not entertain a new application for Avastin in newly-diagnosed patients, the agency might be willing to convert the conditional approval in relapsed patients to a full one, or at least maintain the accelerated approval. The PFS benefit seen in these two separate Phase III trials (albeit a non-significant trend in RTOG 0825), coupled with the long-term responders seen in the Phase II trials in recurrent patients may be sufficient evidence of activity for Avastin in the relapsed setting. In spite of the failure to reach the OS endpoint, the data do highlight that Avastin is still the most promising agent since the introduction of temozolomide in 2005. As Dr. Fine noted in his title of his discussion, “You knew it wasn’t going to be that easy.”


  1. Verhoeff et al., BMC Cancer, 9: 444, 2009
  2. Friedman et al., J Clin Oncol, 27: 4733, 2009
  3. Chinot et al., Neuro-Oncology, 14: Abstract OT-03, 2013



Leave a comment