An exploratory survival analysis of anti-angiogenic therapy for recurrent malignant glioma
Recent clinical trial results suggest that anti-angiogenic therapy may be effective against recurrent malignant glioma. Though these treatments prolong progression-free survival, the extent to which they prolong overall survival is unknown. We pooled data from 34 patients treated at a single institution on phase II clinical trials of bevacizumab and cediranib, and we compared these data to 18 patients treated on clinical trials of cytotoxic chemotherapies. In univariate and multivariate analyses, treatment group was a significant predictor of progression-free but not overall survival. Median progression-free survival was 8 vs. 22 weeks in patients treated with cytotoxic as compared to anti-angiogenic therapy (P = 0.01). Median overall survival was nearly identical in the two groups (39 vs. 37 weeks). The results of this exploratory analysis suggest that anti-angiogenic therapy may fail to prolong overall survival in patients with recurrent malignant glioma. If this conclusion proves correct, progression-free survival may be an inappropriate endpoint for phase II trials of anti-angiogenic therapies.
KeywordsMalignant glioma Anti-angiogenic therapy Progression-free survival
Despite advances in surgery, radiation therapy, and chemotherapy, the prognosis for patients with recurrent malignant gliomas (MG) remains poor. Until recently, radiographic responses in this population have been observed infrequently. Only 5–6% of recurrent glioblastoma (GBM) patients achieve complete or partial responses [1, 2], while 14–35% of recurrent anaplastic glioma (AG) patients respond to treatment, depending on histology and other factors [1, 3]. Survival outcomes are equally disappointing. The six-month progression-free survival (PFS6) for recurrent GBM is 15% and for AG 31% . Overall survival (OS) at 1 year for recurrent GBM is 21% and for AG 47% .
Emerging data indicate that anti-angiogenic therapy is an effective treatment strategy for recurrent MG. Among 32 patients treated in a phase II clinical trial with irinotecan (Camptosar®) and bevacizumab (Avastin®), the humanized monoclonal antibody against vascular endothelial growth factor (VEGF), PFS6 for recurrent GBM was 30% and for AG, 56% . A second phase II study randomized 167 recurrent GBM patients to treatment with bevacizumab alone or in combination with irinotecan. Patients treated with bevacizumab monotherapy had a PFS6 of 43% and radiographic response rate of 28%; patients treated with combination therapy had a PFS6 of 50% and radiographic response rate of 38%. Median OS was approximately 9 months in both groups . The small molecule pan-VEGF receptor (VEGFR) inhibitor cediranib (AZD2171) also has activity against recurrent GBM, with PFS6 of 26% in a phase II trial .
The majority of phase II clinical trials in neuro-oncology utilize PFS6 as a primary endpoint. This approach avoids the imprecision inherent in the use of radiographic response or symptom assessment as measures of efficacy. Furthermore, there is increasing evidence that PFS6 is a valid predictor of OS among recurrent MG patients treated with cytotoxic chemotherapy or targeted molecular agents . In the current era in which anti-angiogenic therapy is rapidly becoming the first-line therapy for recurrent MG, the utility of PFS as an outcome measure is uncertain. Standard Macdonald response and progression criteria rely primarily on measurements of contrast-enhancing tumor on MRI scans . Patients treated with anti-angiogenic therapies often experience worsening of infiltrative tumor that is apparent on T2-weighted MRI scans rather than contrast-enhanced T1 sequences [9, 10, 11, 12]. Additionally, the biological effect of anti-angiogenic therapy in humans with recurrent MG is unknown. Some have proposed that the primary effects of anti-angiogenic therapies are to reduce vascular permeability and vasogenic edema without directly or indirectly killing tumor cells. Though these treatments prolong PFS, the extent to which they have true anti-tumor activity and prolong OS is unknown.
In an exploratory analysis, we therefore reviewed data for patients treated at our institution on phase II clinical trials of anti-angiogenic therapies that are thought to be active against recurrent MG. We compared this to data for patients treated on clinical trials with cytotoxic chemotherapy in an effort to describe the impact of anti-angiogenic treatments on PFS and OS.
Methods and materials
After obtaining written approval from the local institutional review board, we identified patients treated at our institution on clinical trials with bevacizumab or cediranib. Bevacizumab patients had recurrent GBM in first or second relapse. They were treated on a phase II clinical trial protocol (Dana-Farber Cancer Institute Protocol 06-142) in which they were randomized to bevacizumab 10 mg/kg every 2 weeks with or without irinotecan every 2 weeks. The irinotecan dose was 340 mg/m2 for patients who were taking cytochrome P450 enzyme-inducing anti-epileptic drugs or 125 mg/m2 for patients not taking these drugs. Cediranib patients had recurrent GBM and had received no more than two prior chemotherapy treatments. They were treated on a phase II clinical trial protocol (Dana-Farber Cancer Institute Protocol 05-254) in which they received cediranib 45 mg daily.
We also identified cytotoxic chemotherapy patients who were treated on protocols using gimatecan and edotecarin. These protocols were selected because complete data were readily available. In addition, although patients on the cytotoxic chemotherapy studies were generally enrolled earlier than patients treated with the anti-angiogenic therapies, there was significant overlap between the two groups in terms of the time of enrollment. Gimatecan patients had recurrent MG in first relapse and were treated on a phase I clinical trial protocol (Dana-Farber Cancer Institute protocol 02-257) in which they received gimatecan at a dose of at least 0.32 mg/m2/d for 5 consecutive days every 4 weeks. Edotecarin patients had recurrent GBM in first relapse and were treated on a phase II clinical trial protocol (Dana-Farber Cancer Institute protocol 03-205) in which they received edotecarin 13–15 mg/m2 every 3 weeks.
All clinical trials were approved by the local institutional review board, and informed consent was obtained from each patient or health care proxy before enrollment. Patients were evaluated with serial neurologic examinations and gadolinium-enhanced MRI scans at 6–8 weekly intervals or more frequently. Treatment continued until progression or unacceptable toxicity. Response and progression were determined using standard Macdonald criteria . End-points of interest were PFS and OS. Both were measured from the date of trial enrollment.
T-tests, Wilcoxon tests, and Chi square tests were used to identify significant differences between patients in the cytotoxic and anti-angiogenic therapy groups. The impact of treatment group on PFS and OS was evaluated using the Kaplan–Meier method and tested for statistical significance with the log-rank test. Cox multivariate proportional hazards models were constructed to adjust for known predictors of survival including age, sex, surgical resection, histology, performance status, and number of previous chemotherapy regimens. Model assumptions were verified by visual assessment of log-minus-log plots.
Univariate P value
Median age (range), y
Male sex, N (%)
Glioblastoma, N (%)
Median KPS (range)
Median number of previous chemotherapies (range)
Resection, N (%)
Univariate survival analysis
Median OS (95% CI) for subjects who received cytotoxic chemotherapy was 39 weeks (23, 74). Overall survival at 3, 6, and 9 months for subjects who received cytotoxic chemotherapy was 89%, 65%, and 54%, respectively. Median OS (95% CI) for subjects who received anti-angiogenic therapy was 37 weeks (30, 42). Overall survival at 3, 6, and 9 months for subjects who received anti-angiogenic therapy was 91%, 74%, and 53%, respectively. Two-sided p value for the log-rank test comparing OS in patients treated with cytotoxic and anti-angiogenic therapy was 0.60 (Fig. 2b). When non-GBM patients were excluded from the analysis, survival estimates did not appreciably change, and the P value increased to 0.89.
Univariate predictors of survival
Median PFS (95% CI)
P value for PFS
Median OS (95% CI)
P value for OS
18.6 (10.1, 22.9)
37.6 (30.5, 44.9)
6.6 (4.0, –)
39.4 (10.6, –)
18.7 (11.0, 23.1)
37.1 (30.4, 42.3)
8.0 (4.0, 68.7)
50.6 (11.3, 164.6)
18.7 (12.0, 23.7)
37.6 (30.6, 46.3)
8.0 (5.7, 23.1)
38.1 (11.6, 48.3)
Number of previous chemotherapies
7.0 (4.1, 34.0)
18.7 (10.6, 57.3)
18.6 (10.1, 22.9)
37.9 (33.9, 46.3)
17.0 (9.0, 22.9)
34.9 (28.6, 48.3)
17.4 (8.0, 32.1)
38.6 (33.9, 46.3)
17.0 (8.3, 23.7)
37.9 (33.9, 46.3)
14.4 (8.0, 22.9)
29.5 (19.4, 48.3)
8.0 (7.9, 16.4)
39.4 (23.4, 74.1)
21.9 (16.0, 28.0)
37.4 (30.4, 42.3)
17.0 (9.0, 22.9)
37.6 (30.6, 44.9)
Multivariate survival analysis
Multivariate predictors of survival
HR (95% CI)
HR (95% CI)
Age (≥40 vs.<40 years)
1.26 (0.27, 5.93)
0.93 (0.20, 4.32)
Histology (GBM vs. AG)
1.42 (0.45, 4.48)
1.67 (0.50, 5.56)
KPS (≥80 vs.<80)
0.82 (0.38, 1.81)
0.84 (0.36, 1.98)
Number of previous chemotherapies (≥2 vs.<2)
1.84 (0.73, 4.62)
1.75 (0.66, 4.63)
Surgery (resection vs. biopsy)
1.08 (0.50, 2.30)
Treatment (Anti-angiogenic vs. cytotoxic)
0.38 (0.17, 0.84)
1.01 (0.45, 2.29)
These data indicate that recurrent MG patients treated with anti-angiogenic therapies experience longer PFS than patients treated with conventional cytotoxic chemotherapy. This finding is consistent with the results of recent, prospective phase II clinical trials of bevacizumab [4, 13, 14] and cediranib  in addition to several retrospective analyses [11, 15, 16, 17]. Despite the observation that median PFS increased from 8 to 22 weeks, we found that OS was nearly identical in the two cohorts (39 weeks in patients treated with cytotoxic chemotherapy versus 37 weeks in patients treated with anti-angiogenic therapy). These relationships were observed on both univariate analysis and multivariate analysis that accounted for other known prognostic factors. A multivariate Cox proportional hazards regression model found a hazard ratio of 0.38 for the effect of anti-angiogenic therapy on PFS, indicating that patients treated with bevacizumab or cediranib had only 38% the hazard of progression or death compared to patients treated with cytotoxic chemotherapy.
These findings are important because of their potential implications regarding the biology of anti-angiogenic therapy for MG. We and others who treat MG patients with bevacizumab, cediranib, and other VEGF/VEGFR inhibitors have observed rapid radiographic progression and clinical decline following the cessation of anti-angiogenic therapy. In addition, salvage therapy for patients who progress on their initial VEGF/VEGFR therapy is generally ineffective [9, 18]. In a retrospective review of 54 recurrent MG patients who progressed on a bevacizumab-containing regimen and were then treated with an alternate bevacizumab-containing regimen, no patients achieved radiographic responses. The median PFS on the second bevacizumab-containing regimen was only 37.5 days, and PFS6 was 2% . These observations led us to hypothesize that the primary effect of anti-angiogenic agents is to delay tumor progression without necessarily providing an OS benefit. The question of why withdrawal of anti-angiogenic agents results in rapid failure is unanswered. Potential explanations include upregulation of edema-producing mechanisms, activation of alternative angiogenesis signaling pathways such as bFGF , stromal-derived factor-1α (SDF1α) , tie2, or placental growth factor (PlGF) , and stimulation of tumor invasion machinery .
A similar analysis with a larger dataset performed in 1999 found no difference in outcomes between patients treated with cytotoxic and cytostatic chemotherapy agents . Ours is the first study to pool data from patients treated with anti-angiogenic agents and show an improvement in PFS compared to patients treated with cytotoxic chemotherapy. These pooled data are likely generalizable in that the inclusion and exclusion criteria for the clinical trials from which patients were selected are similar to those used by most recurrent MG trials.
A potential implication of our findings is that PFS may not be an optimal endpoint in phase II studies of anti-angiogenic drug therapies. In patients treated with anti-angiogenic therapy, determination of the precise time at which tumor progression occurs is challenging. These agents frequently provoke robust, sustained radiographic responses that render subsequent MRI scans difficult to interpret . Patients have been observed to progress clinically in the absence of radiographic progression as defined by Macdonald criteria. Since clinical worsening may be due to a number of factors aside from tumor progression (e.g., seizures, toxic-metabolic disturbance, infection, other medical illness), the use of clinical features to determine progression is subjective. Another issue concerns the observation that T2 or FLAIR hyperintensity may worsen on MRI despite stable gadolinium enhancement, which is the only MRI parameter of interest for assessing tumor progression by Macdonald criteria. In addition to infiltrating tumor, T2 or FLAIR hyperintensity may reflect edema, gliosis, or seizure activity, thus rendering the use of T2/FLAIR MRI sequences for assessment of progression also subjective.
These results warrant confirmation with a larger, prospective dataset, such as may be obtained in a phase III clinical trial. We did not find that other prognostic factors such as age, performance status, or histology were significant predictors of survival, likely indicating insufficient statistical power. However, since the difference in PFS was highly significant on univariate and multivariate analysis, it is unlikely that there exists an OS difference of similar magnitude that this study failed to detect. Another flaw in this exploratory analysis involves the differences between the two treatment groups. Patients in the anti-angiogenic therapy group were more likely to have GBM histology and to have received more chemotherapy regimens. Because these are adverse prognostic factors, they would be expected to bias the results toward the null hypothesis that survival is the same in the two groups. The finding of a significant PFS difference on univariate analysis indicates that these discrepancies were not highly problematic. Furthermore, in an effort to verify that our results were not biased by the inclusion of some AG patients in the cytotoxic chemotherapy cohort, survival analyses were repeated with these patients excluded. This had no significant impact on the results.
In conclusion, this preliminary study suggests that anti-VEGF/VEGFR therapies increase PFS but produce little or no gain in OS. While the prolongation of PFS, together with concomitant reduction in corticosteroid use, represents an important clinical benefit, much work is still needed to enhance the effectiveness of this class of agents and increase OS in patients with recurrent MG.
- 2.Yung W, Albright R, Olson J, Fredericks R, Fink K, Prados M, Brada M, Spence A, Hohl R, Shapiro W, Glantz M, Greenberg H, Selker R, Vick N, Rampling R, Friedman H, Phillips P, Bruner J, Yue N, Osoba D, Zaknoen S, Levin V (2000) A phase II study of temozolomide vs. procarbazine in patients with glioblastoma multiforme at first relapse. Br J Cancer 83:588–593. doi:10.1054/bjoc.2000.1316 PubMedCrossRefGoogle Scholar
- 3.Yung WK, Prados MD, Yaya-Tur R, Rosenfeld SS, Brada M, Friedman HS, Albright R, Olson J, Chang SM, O’Neill AM, Friedman AH, Bruner J, Yue N, Dugan M, Zaknoen S, Levin VA (1999) Multicenter phase II trial of temozolomide in patients with anaplastic astrocytoma or anaplastic oligoastrocytoma at first relapse. Temodal Brain Tumor Group. J Clin Oncol 17:2762–2771PubMedGoogle Scholar
- 4.Vredenburgh JJ, Desjardins A, Herndon JEII, Dowell JM, Reardon DA, Quinn JA, Rich JN, Sathornsumetee S, Gururangan S, Wagner M, Bigner DD, Friedman AH, Friedman HS (2007) Phase II trial of bevacizumab and irinotecan in recurrent malignant glioma. Clin Cancer Res 13:1253–1259. doi:10.1158/1078-0432.CCR-06-2309 PubMedCrossRefGoogle Scholar
- 5.Cloughesy T, Prados M, Wen P, Mikkelson T, Abrey L, Schiff D, Yung WK, Paleologos N, Nicholas MK, Dorr A, Zheng M, Dimery I, Friedman H (2007) A phase II, randomized, non-comparative clinical trial of bevacizumab alone or in combination with CPT-11 prolongs 6-month PFS in recurrent, treatment-refractory glioblastoma. Society for Neuro-Oncology 12th Annual MeetingGoogle Scholar
- 6.Batchelor TT, Sorensen AG, di Tomaso E, Zhang WT, Duda DG, Cohen KS, Kozak KR, Cahill DP, Chen PJ, Zhu M, Ancukiewicz M, Mrugala MM, Plotkin S, Drappatz J, Louis DN, Ivy P, Scadden DT, Benner T, Loeffler JS, Wen PY, Jain RK (2007) AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, normalizes tumor vasculature and alleviates edema in glioblastoma patients. Cancer Cell 11:83–95. doi:10.1016/j.ccr.2006.11.021 PubMedCrossRefGoogle Scholar
- 7.Lamborn KR, Yung WK, Chang SM, Wen PY, Cloughesy TF, DeAngelis LM, Robins HI, Lieberman FS, Fine HA, Fink KL, Junck L, Abrey L, Gilbert MR, Mehta M, Kuhn JG, Aldape KD, Hibberts J, Peterson PM, Prados MD (2008) Progression-free survival: an important end point in evaluating therapy for recurrent high-grade gliomas. Neuro Oncol 10:162–170. doi:10.1215/15228517-2007-062 PubMedCrossRefGoogle Scholar
- 9.Lassman AB, Iwamoto FM, Gutin PH, Abrey LE (2008) Patterns of relapse and prognosis after bevacizumab (BEV) failure in recurrent glioblastoma (GBM). J Clin Oncol 26:2028Google Scholar
- 10.Narayana A, Raza S, Golfinos JG, Johnson G, Knopp EA, Zagzag D, Fischer I, Medabalmi P, Eagan P, Gruber ML (2008) Bevacizumab therapy in recurrent high grade glioma: impact on local control and survival. J Clin Oncol 26:13000Google Scholar
- 11.Norden AD, Young GS, Setayesh K, Muzikansky A, Klufas R, Ross GL, Ciampa AS, Ebbeling LG, Levy B, Drappatz J, Kesari S, Wen PY (2008) Bevacizumab for recurrent malignant gliomas: efficacy, toxicity, and patterns of recurrence. Neurology 70:779–787. doi:10.1212/01.wnl.0000304121.57857.38 PubMedCrossRefGoogle Scholar
- 12.Zuniga RM, Torcuator R, Doyle T, Anderson J, Jain R, Orley J, Rosenblum M, Mikkelsen T (2008) Retrospective analysis of patterns of recurrence seen on MRI in patients with recurrent glioblastoma multiforme treated with bevacizumab plus irinotecan. J Clin Oncol 26:13013Google Scholar
- 13.Cloughesy TF, Prados MD, Mikkelsen T, Abrey LE, Schiff D, Yung WK, Maoxia Z, Dimery I, Friedman HS (2008) A phase II, randomized, non-comparative clinical trial of the effect of bevacizumab (BV) alone or in combination with irinotecan (CPT) on 6-month progression free survival (PFS6) in recurrent, treatment-refractory glioblastoma (GBM). J Clin Oncol 26:2010bGoogle Scholar
- 14.Vredenburgh JJ, Desjardins A, Herndon JEII, Marcello J, Reardon DA, Quinn JA, Rich JN, Sathornsumetee S, Gururangan S, Sampson J, Wagner M, Bailey L, Bigner DD, Friedman AH, Friedman HS (2007) Bevacizumab plus irinotecan in recurrent glioblastoma multiforme. J Clin Oncol 25:4722–4729. doi:10.1200/JCO.2007.12.2440 PubMedCrossRefGoogle Scholar
- 15.Guiu S, Taillibert S, Chinot O, Taillandier L, Honnorat J, Dietrich PY, Maire JP, Guillamo JS, Guiu B, Catry-Thomas I, Capelle F, Thiebaut A, Cartalat-Carel S, Deville C, Fumoleau P, Desjardins A, Xuan KH, Chauffert B (2008) Bevacizumab/Irinotecan. An active treatment for recurrent high grade gliomas: Preliminary results of an ANOCEF Multicenter Study. Rev Neurol (Paris) 164:588–594 doi:10.1016/j.neurol.2008.04.003 Google Scholar
- 17.Stark Vance V (2005) Bevacizumab (Avastin®) and CPT-11 (Camptosar®) in the Treatment of Relapsed Malignant Glioma. Neuro Oncol 7:369Google Scholar
- 18.Quant E, Norden AD, Drappatz J, Ciampa A, Doherty L, LaFrankie D, Kesari S, Wen PY (2008) Role of a second chemotherapy in recurrent malignant glioma patients who progress on a bevacizumab-containing regimen. J Clin Oncol 26:2008Google Scholar
- 19.Willett CG, Boucher Y, Duda DG, di Tomaso E, Munn LL, Tong RT, Kozin SV, Petit L, Jain RK, Chung DC, Sahani DV, Kalva SP, Cohen KS, Scadden DT, Fischman AJ, Clark JW, Ryan DP, Zhu AX, Blaszkowsky LS, Shellito PC, Mino-Kenudson M, Lauwers GY (2005) Surrogate markers for antiangiogenic therapy and dose-limiting toxicities for bevacizumab with radiation and chemotherapy: continued experience of a phase I trial in rectal cancer patients. J Clin Oncol 23:8136–8139. doi:10.1200/JCO.2005.02.5635 PubMedCrossRefGoogle Scholar
- 20.Sorensen AG, Batchelor TT, Wen PY, Zhang WT, Jain RK Response criteria for glioma (2008) Nat Clin Pract Oncol 5:634–644Google Scholar