Taming glioblastoma: targeting angiogenesis.

More than 30 years ago, Judah Folkman hypothesized that tumor angiogenesis could be a target of anticancer therapy. Some of the initial studies were actually performed in patients with glioblastomas, which had the greatest extent of tumor angiogenesis among a variety of human malignancies. The identification of vascular endothelial growth factor (VEGF) also provided a firm foundation for the development of antiangiogenic drugs. Fast forward to 2007: We now have multiple agents that specifically block tumor angiogenesis in various malignancies. In this issue of the Journal of Clinical Oncology, two groups independently describe phase II clinical trials of bevacizumab plus irinotecan for recurrent glioblastomas. Vredenburgh et al noted an objective response rate of 57% and a 6-month progression-free survival of 46% with this regimen. These results are remarkable indeed when compared with a benchmark response rate of 6% and 6-month progression-free survival of 15% from salvage cytotoxic chemotherapies. Although the radiographic responses were impressive, resolution of gadolinium enhancement could also be caused by bevacizumab-induced changes in vascular permeability, impairing our ability to visualize these tumors on magnetic resonance imaging (MRI). To demonstrate that bevacizumab plus irinotecan had an impact on the underlying tumor, Chen et al went one step further by using F-fluorothymidine positron emission tomography (FLT-PET) as an adjunctive measure of tumor response. They reported a 47% response rate by FLT-PET and a 65% 6-month survival, whereas metabolic responders lived three times longer than did nonresponders. These results are consistent with earlier findings that response to cytotoxic chemotherapies was associated with a significantly lower treatment failure rate and a hazard ratio of 0.5. Taken together, both studies establish a role for combining antiangiogenic drug and cytotoxic chemotherapy to treat recurrent glioblastomas. However, these results should be interpreted within the context of glioblastoma behaviors and our ability to measure treatment efficacy. The rationale for combining an antiangiogenic agent with cytotoxic chemotherapy is based on the phenomenon of vascular normalization, which leads to increased tissue oxygenation, decreased interstitial hypertension, and improved drug delivery to tumors. This was demonstrated experimentally with DTC101, a monoclonal antibody specific for VEGF receptor 2. As hyperpermeable vasculatures were pruned away by DTC101, increased tissue oxygenation acted synergistically with external-beam radiation in controlling U87 glioblastomas implanted in mouse brains. Interestingly, this normalization window peaked at day 5 of DTC101 treatment. For combination bevacizumab and cytotoxic chemotherapy, precise scheduling does not appear to be important, possibly because of immediate permeability changes and the prolonged half-life of bevacizumab. In patients treated with AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, Batchelor et al observed improvement in glioblastoma perfusion and permeability as measured by dynamic susceptibility contrast MRI within the first 24 hours. This is in contradiction to vascular damaging agents, which shut down well-developed tumor blood vessels and are highly schedule dependent when combined with cytotoxic chemotherapies. For example, ZD6126 induced hypoxia in U87 glioblastomas and interfered with radiation efficacy when administered 1hour before radiation, but it was synergistic with radiation when administered 24 hours after radiation. Similarly, 5,6-dimethyl-xanthenone-4 acetic acid and combretastatin A-4 disodium phosphate had the greatest tumor-cell kill when administered 1 to 3 hours after cisplatin. The current response criteria, or Macdonald’s criteria, may not be adequate for assessing glioblastoma response to antiangiogenic drugs. This is because tumor size is “estimated” on the basis of gadolinium leakage from hyperpermeable vasculatures. When antiangiogenic drugs change vascular permeability, they may alter the apparent size of glioblastomas on contrast-enhanced MRI or computed tomography, without affecting the underlying tumor mass. Therefore, it is critically important to incorporate adjunctive measures of tumor response, such as dynamic susceptibility contrast MRI for vascular perfusion and permeability and FLT or C-methylmethionine PET for tumor metabolism. Chen et al used FLT-PET and noted that metabolic response correlated with MRI response and patient survival. Unlike [F]fluorodeoxyglucose PET, FLT-PET has higher signal-to-noise ratio and is not confounded by a high rate of glucose utilization in the brain. But in order for FLT and C-methylmethionine PET to become accepted methods of assessing response in glioblastomas, more data are needed to evaluate their positive and negative predictive values. It is equally notable that not all patients responded to bevacizumab plus irinotecan. Heterogeneity in vascular response to bevacizumab may explain some of these failures. In the AZD2171 trial, some patients had worsening or unchanged perfusion and permeability, whereas others responded favorably. There may be secondary pathways for tumor angiogenesis to occur in nonresponders when either VEGF or the VEGF receptor is blocked. In Chen et al’s cohort, metabolic heterogeneity was also noted, as seen JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 25 NUMBER 30 OCTOBER 2