It may also be challenging to interpret the clinical significance of the findings \[[@B1]\]. To address this concern, we have suggested the use of new clinical end points to describe the response of tumors and patients to antiangiogenic therapy, including the time to progression, the total time during treatment, and the probability of disease progression or death \[[@B1], [@B3]\]. We have now added a further category of end point, the probability of overall survival and disease-free survival for those patients whose tumor did not progress or died before the end of the treatment. These can be combined with other data, such as progression-free survival and objective response rates, to allow the clinical interpretation of efficacy. 2.2. Clinical Benefits {#sec2.2} ---------------------- ### 2.2.1. Phase I Studies {#sec2.2.1} Data from phase I studies of first-generation antiangiogenic agents in advanced disease are sparse. However, in a study of a TKI, aflibercept, in patients with advanced melanoma, the drug appeared to be well tolerated, although this was not a dose-escalation study \[[@B6]\]. Phase II data were available for bevacizumab. In a phase II study of first-line bevacizumab in combination with chemotherapy, median survival for the control arm (carboplatin/paclitaxel) was 12.7 months compared with 10.2 months with bevacizumab, while the progression-free survival rates were 5.5 months and 7.6 months, respectively \[[@B7]\]. In a second-line study with bevacizumab, patients had a median survival of 10.6 months compared with 6.9 months for those in the control arm \[[@B8]\]. Results of a phase II study with bevacizumab in patients with recurrent glioblastoma were recently reported. Median survival in patients treated with bevacizumab plus irinotecan was 13 months, compared with 9.9 months for those treated with irinotecan alone \[[@B9]\]. These data indicated that antiangiogenic therapy may lead to improvements in clinical outcomes. ### 2.2.2. Phase II Studies {#sec2.2.2} There were very few phase II data available in the peer-reviewed literature, and this is also the case for the phase III data to date. However, as discussed above, phase II results can help us in our understanding of the optimal dose and schedules for further development in phase III studies. There have been two reported phase II studies of bevacizumab in patients with recurrent glioblastoma. The phase II study by Hainsworth et al. \[[@B10]\] demonstrated a median survival of 20 weeks, with 2/17 patients surviving 6 months or more. Three studies have evaluated the antiangiogenic agent sunitinib in phase II studies in patients with recurrent glioblastoma \[[@B11]--[@B13]\]. Two of these demonstrated median survival of 6.4--9 months with sunitinib \[[@B11], [@B13]\], while the third study reported the median survival for sunitinib was 7.4 months \[[@B12]\]. Median progression-free survival rates were 4.6 months and 5.3 months, respectively. 2.3. Implications of Biomarker Analysis {#sec2.3} --------------------------------------- Biomarker studies have shown that changes in vascular endothelial growth factor levels following antiangiogenic therapy do correlate with outcomes, suggesting that such measurements may be useful in clinical practice for evaluating responses \[[@B14]--[@B16]\]. However, the majority of samples in such studies have been plasma, and it is uncertain whether these will be valid in the brain, since bevacizumab may have an effect in the brain \[[@B17]\]. While it may be of clinical benefit to measure vascular endothelial growth factor levels in patients with glioblastoma, a better understanding of the tumor environment would be useful in predicting response to therapy. 3. The Effect of Bevacizumab on Endothelial Growth Factor {#sec3} ========================================================= 3.1. Evidence from Experimental Models {#sec3.1} -------------------------------------- The effect of antiangiogenic therapy in glioblastoma may be at least partly explained by inhibition of vascular endothelial growth factor and its receptor \[[@B18]\]. In animal models, bevacizumab appears to induce responses in a manner which correlates with its level of vascular endothelial growth factor activity \[[@B19]\]. 3.2. Phase I Studies {#sec3.2} -------------------- Phase I studies of antiangiogenic agents have shown that there are both immediate and late responses in clinical outcome. This supports the findings of experimental models, where treatment of tumors with antiangiogenic agents induces a period of transient regression in tumor size \[[@B20]\]. The initial phase of this tumor "awakening" appears to involve an acute increase in vascular permeability and edema, followed by a period of tumor stasis or regression. This then leads to an increase in tumor perfusion and re-initiation of tumor growth. This model of antiangiogenic therapy is consistent with the use of bevacizumab to induce transient tumor response, rather than complete eradication of all of the tumor. The same principles can be applied to the use of bevacizumab in recurrent disease, where the tumor may be reduced in size in the short term, but subsequently resumes growth at the same rate as baseline. 3.3. Phase II Studies {#sec3.3} --------------------- There were no phase II data available in the peer-reviewed literature, but data from phase I studies were available, and these supported the results of clinical studies. There were no phase III studies available. A study of sunitinib in phase II trials was terminated for slow recruitment, and it is uncertain whether there is sufficient evidence to determine whether antiangiogenic agents have a role in the treatment of patients with recurrent glioblastoma. 3.4. End Points for Future Studies {#sec3.4} ---------------------------------- Antiangiogenic therapy may be of value as a component of treatment of patients with malignant gliomas \[[@B21]\], particularly in combination with other agents \[[@B22], [@B23]\]. For the phase III studies of bevacizumab in recurrent glioblastoma, the median progression-free survival rates were approximately 5 months, and this was improved in some patients (see below). Based on these results, the phase III trials used median progression-free survival rates as the primary end point. Future trials should include alternative end points, such as the total duration of disease control, median duration of progression-free survival, and the number of patients with complete remission. 4. Potential for Alternative Agents {#sec4} =================================== Data regarding the use of bevacizumab as a single agent have been disappointing, with a clear need for alternative agents in the treatment of patients with glioblastoma. Several agents are now undergoing clinical testing, although the benefit of combining these agents with bevacizumab has not been established. We discuss some of these below. 4.1. Therapeutic Implications {#sec4.1} ----------------------------- Data on alternative treatments for recurrent disease have been derived from phase I trials in patients with refractory solid tumors \[[@B24]\]. Patients were divided into two groups, and the dose for each group was increased if there were no limiting toxicities. Doses were then escalated in three dose increments, and the maximum tolerated dose was then determined. However, the authors suggested a new design for phase I studies in advanced solid tumors \[[@B24]\]. In their approach, each patient receives at least two different doses of the agent being studied in combination with bevacizumab. If no toxicities are observed, then at least a total of 12 doses will be administered. With further dose escalation, if no patient develops a grade 3 or 4 toxicity, then the maximum tolerated dose will be that dose associated with the least amount of toxicity. Following this, phase II studies may be conducted to confirm the activity of the agent in patients with solid tumors \[[@B25]\]. The design of phase II studies should also include a preliminary analysis of progression-free survival, although this is less important for evaluating single-agent therapy. 4.2. Bevacizumab Plus Bevacizumab {#sec4.2} --------------------------------- The optimal dose of bevacizumab in combination with other agents in recurrent glioblastoma has not yet been determined. A phase I/II study evaluated the addition of bevacizumab to irinotecan and temozolomide (with or without dexamethasone) in patients with recurrent glioblastoma \[[@B26]\]. There were 12 patients, and the median overall survival was 8.6 months, with an overall survival rate at 6 months of 66.7%. The median progression-free survival was 2.8 months. Median progression-free survival and overall survival were slightly better with bevacizumab and temozolomide alone (progression-free survival = 1.7 months, overall survival = 8.7 months). 4.3. Combination with Liposomal Irinotecan {#sec4.3} ------------------------------------------ There is currently insufficient evidence from clinical studies to recommend use of the liposomal irinotecan, SN-38, in the treatment of patients with glioblastoma \[[@B27]\]. Nevertheless, there may be a role for the use of liposomal irinotecan in patients who are resistant to other treatments. It has been suggested that combination therapy of liposomal irinotecan and bevacizumab should be given for at least 3 months in patients who have been treated with bevacizumab and have progression of the tumor \[[@B28]\]. 4.4. Bevacizumab Plus Erlotinib {#sec4.4} ------------------------------- Erlotinib