Efficacy and Safety of Bevacizumab for Treating Glioblastoma: A Systematic Review and Meta-Analysis of Phase II and III Randomized Controlled Trials

Abstract Objective To fully investigate the efficacy and safety of bevacizumab for glioblastoma. Methods Databases were searched for phase II/III randomized controlled trials treated with bevacizumab. Results Bevacizumab significantly improved the PFS in glioblastoma patients, but did not prolong OS. PFS was significantly prolonged in both first-line and second-line treatment. Bevacizumab plus temozolomide was correlated with improved PFS for patients with different MGMT methylation status. Bevacizumab could increase the risk of hypertension, proteinuria, thromboembolic, and infection. Hypertension should be well concerned. Conclusions Bevacizumab-containing regimen can significantly improve PFS, but did not prolong OS.


Introduction
Glioblastoma (GBM) is the most common primary malignant brain tumor in adults. Recent clinical trials have reported a median survival of only 14-16 months with a 2-year survival rate of 26-33% (1). Standard care for newly diagnosed GBM is surgical resection as extensively as is safely possible, followed by radiotherapy with concurrent temozolomide and six monthly cycles of adjuvant temozolomide (2). Nearly all patients experience recurrence following standard-of-care surgical resection, radiotherapy, and temozolomide (1). Survival rate of recurrent GBM is extremely low, and the prognosis after recurrence is dismal with a short progression-free survival (PFS) period and overall survival (OS).
Glioblastomas are highly vascularized tumors in which the vascular endothelial growth factor (VEGF) signaling pathway is up-regulated. Therefore, the inhibition of VEGF is assumed to slow down tumor growth and enhance the effects of radiotherapy and chemotherapy (3). Bevacizumab is a humanized monoclonal antibody that targets VEGF-A and prevents VEGF from binding to receptors on the surface of endothelial cells, thereby inhibiting angiogenesis. It was the first drug targeting tumor angiogenesis approved by the FDA and was applied for the treatment of glioblastomas since 2009 (4).
Clinical trials showed that bevacizumab has a marked response rate from 30 to 50% in patients with recurrent GBM, prolonging PFS period and improving patients' quality of life (5). Moreover, clinical trials have also demonstrated that bevacizumab was efficacious in treating newly diagnosed GBM. However, there has been a substantial variation in the efficacy and safety among the available clinical trials. We therefore conducted this systematic review and meta-analysis of randomized controlled trials (RCTs) to fully investigate the efficacy and safety of bevacizumab in the treatment of GBM.

Search strategy and study selection
This meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (6). Relevant studies published in English were selected by searching EMBASE, PubMed, and Medline (data from January 1, 2000 to September 1, 2022). We searched the databases using the combinations of the following keywords "bevacizumab" AND "Avastin" AND "glioblastoma" AND "randomized controlled phase trials" AND "RCT", and the relevant Medical Subject Heading (MeSH) terms were searched.
Studies that met the following criteria were included in the meta-analysis: (1) patients receiving bevacizumab for the treatment of GBM; (2) subjects are diagnosed with GBM, irrespective whether primary or recurrent type; (3) the studies were designed as RCTs; (4) data were reported on progression-free survival (PFS), overall survival (OS), and adverse events (AEs).
The studies that did not meet the above inclusion criteria were excluded. Bevacizumab in both treatment and control arms was excluded. If multiple publications of the same trial were retrieved, the articles providing the most detailed data were included.

Data extraction and quality assessment
Two investigators independently performed data extraction, and any discrepancies were resolved by consensus. For each study, the following information was collected: the first author's name, year of publication, study design, number of treatment arms, median age, therapeutic regimen, median durations, drug doses, and outcomes. Clinical outcomes collected from the trials included hazard ratios (HR) for PFS and OS and their 95% confidence intervals (CIs), and numbers or rate of each type of AEs. In this study, we separated AEs into all-grade (grade 1-5) and high-grade (grade 3-5) for our analysis.
Study quality was assessed according to the Jadad ranking system recommended by The Cochrane Handbook for Systematic Reviews of Interventions (7). The Jadad five-point scale was introduced to evaluate the overall methodological quality of RCTs. Each study was independently assessed for quality and potential bias based on randomization sequence generation, intention-totreat (ITT) analyses, blinding of participants and personnel, allocation concealment, incomplete outcome data, study withdrawals, and other biases. Each criterion was categorized as "low", "unclear", or "high" risk of bias.

Statistical analysis
All statistical analyses were performed using Review Manager Version 5.4. HRs with 95% confidence interval (CI) were used to assess OS and PFS, and the dichotomous outcomes (AEs) were analyzed as the risk ratios (RRs) with 95% CI. The Inverse Variance method was used to calculate the 95% CIs and HRs, and the Mantel-Haenszel method was used to calculate the 95% CIs and RRs. The HR and associated variances were extracted directly from the trial publications when possible. If these data were not reported, the HR was obtained indirectly from the data extracted from published Kaplan-Meier curves. An HR greater than one favors the control arm, whereas an HR less than one favors the research arm. Heterogeneity was tested with I 2 statistics. Heterogeneity was considered statistically significant when I 2 >50%. If heterogeneity existed, data were analyzed using a random-effects model. In the absence of heterogeneity, a fixed-effects model was used. Subgroup analysis was conducted to examine whether the PFS and OS varied by the treatment regimens (bevacizumab þ temozolomide vs. temozolomide; bevacizumab þ lomustine vs. lomustine; bevacizumab (þPD-1) vs. PD-1), and GBM type (newly diagnosed GBM vs. recurrent/progressive GBM). Subgroup difference was considered statistically significant when I 2 >50%.

Results
A total of 206 potentially relevant articles were initially identified by searching databases.

Characteristics of included studies and assessment of methodological quality
A total of 3015 patients from 11 trials were available for the present meta-analysis ( Table 1). The median age of the enrolled patients ranged from 52 to 70 years, and the sample size in the included trials varied from 56 to 911. Six trials were conducted in the first-line setting, and five were in the ! second line setting. Bevacizumab was administrated 10 mg every 2 weeks in all the included trials. One trial (11) was a four-arm study, and all the others were two-arm study.
The baseline characteristics of the included trials are shown in Table 1.
All the included studies were assessed for study quality using the Jadad scoring system. The methods of randomization were reported in all the included studies, and ITT analysis was conducted in 10 studies. Four of these studies were double-blind study, five were open-label study, and two were unclear. All of the included RCTs were judged to be of adequate quality (Jadad score ! 3/5). The quality assessments of included trials are presented in Supplementary Table 1.

Progression-free survival (PFS)
All of the included trials reported PFS, and there was a statistically significant difference in PFS in favor of the bevacizumab therapy ( Figure 2(A)), with an HR of 0.59 (95% CI: 0.52-0.66; p < 0.00001). Statistically significant heterogeneities were found among these studies (I 2 ¼79%), and the random-effects model was used. The results showed a decrease in heterogeneity (I 2 ¼69%) when the study of Gilbert2014 (9) was removed and still had a statistical significance (p < 0.00001, Supplementary Figure 2). This data indicated that the results were robust. Among the eleven included studies, nine studies (8)(9)(10)(11)(13)(14)(15)(16)18) reported that the bevacizumab-containing remedy were associated with a significantly prolonged the median PFS than without bevacizumab (Table 1), which also shows that PFS could be improved in bevacizumab-containing remedy.  The heterogeneity in the meta-analysis might be due to different GBM types and treatment regimens. Thus, subgroup analyses were performed in accordance with these differences. Subgroup analysis (Table 2) suggested that PFS was significantly prolonged in all subgroups of treatment regimens, and a significant difference was observed in PFS according to treatment regimens (I 2 ¼84.2%). Our analysis implied that bevacizumab plus lomustine (HR ¼ 0.53) was superior to bevacizumab plus temozolomide (HR ¼ 0.71) in enhancing PFS, and bevacizumab (bevacizumab þ PD-1) was superior to PD-1 inhibitor in improving PFS. We further analyzed the HRs of PFS in patients with newly diagnosed GBM and recurrent/progressive GBM, and a significant difference was observed in PFS according to GBM types (I 2 ¼56.8%). It seems to be that recurrent patients tend to have a better PFS than patients with newly diagnosed GBM. Among most of our included studies, however, newly diagnosed patients were treated with bevacizumab plus temozolomide and recurrent patients received bevacizumab plus lomustine. Therefore, we performed separate analyses on these differences in the following section (Section 3.4) to confirm the above results. Besides, it should be noted that a four-arm study (11) was included in our meta-analysis, and thus there were more numbers of comparisons than the number of included studies in the subgroup analysis.

Overall survival (OS)
All of the included trials reported OS, and there was no statistically significant difference in OS, with an HR of 0.96 (95% CI: 0.88-1.05; p ¼ 0. 36). No statistically significant heterogeneities was found among these studies, and the fixed-effects model was used (Figure 2(B)).
Subgroup analysis (Table 2) suggested that OS was not significantly prolonged in any of the subgroups by treatment regimens, and no significant difference was observed in OS according to treatment regimens (I 2 ¼0%). Patients were further stratified according to GBM type, and no significant difference was observed in OS according to GBM type (I 2 ¼0%).

Separate analyses of bevacizumab containing treatment in first-or second-line treatment
Among most of our included studies, (Table 1), newly diagnosed patients were treated with bevacizumab plus temozolomide and recurrent patients received bevacizumab plus lomustine. In order to indicate whether future studies should test bevacizumab containing treatment in first-or second-line treatment, we performed separate analyses on patients who were treated with bevacizumab plus temozolomide in the first-line treatment and bevacizumab plus lomustine in the recurrent GBM. As shown in Figure 3(A), PFS was significantly prolonged in newly diagnosed patients who were treated with bevacizumab plus temozolomide in the first-line treatment (HR: 0.71; 95% CI: 0.61-0.82; p<0.0001), but the OS was not. Similarly, PFS was significantly prolonged in recurrent patients treated with bevacizumab and lomustine in the second-line treatment (HR: 0.52; 95% CI: 0.45-0.60; p<0.0001), but the OS was not (Figure 3(B)). A significant difference was observed in PFS between the two groups (p ¼ 0.004), which implied that recurrent patients who were treated with bevacizumab and lomustine (HR, 0.52) were associated with a higher PFS than that of bevacizumab plus temozolomide in the first-line treatment (HR, 0.71).
The O 6 -methylguanine-DNA methyltransferase (MGMT) has been identified as a major mechanism by which cells can exhibit resistance to temozolomide. It was reported that temozolomide is more effective in causing cell death when tumors cells exhibit lower levels of MGMT. Our subgroup analysis evaluated the effect of bevacizumab plus temozolomide for newly diagnosed GBM with different MGMT methylation status (unmethyleted and methyleted). As shown in Supplementary Figure 1(A), PFS was significantly prolonged in all subgroups of MGMT methylation status, and there was no significant difference in PFS according to the MGMT methylation status between bevacizumab and controls (I 2 ¼0%). However, OS was not significantly prolonged in neither MGMT unmethyleted nor methylated groups (Figure 1(B)).

Adverse events (AEs)
The common reported AEs associated with bevacizumab including cardiovascular toxicities and gastrointestinal events. Compared with controls, bevacizumab was associated with an increased risk of all-grade hypertension (RR: 3.15; 95% CI: 1.98-5.01; p<0.00001) and proteinuria (RR: 1.84;  Table 3.

Publication bias
Funnel plots of the studies used in the meta-analysis to evaluate the HRs of PFS are shown in Figure 4. No significant publication bias was detected in the meta-analysis of PFS.

Discussion
Bevacizumab, a humanized monoclonal antibody was approved by the FDA for the treatment of GBM. Our meta-analysis evaluated the efficacy and safety of bevacizumab in patients with GBM. Our analysis provided evidence that bevacizumab could significantly improve the PFS in patients with GBM, but could not prolong OS in these studies. The median PFS could be prolonged by bevacizumab-containing remedy in most of the included studies. Temozolomide is an oral alkylating agent with a broad spectrum of antineoplastic activity (19), which has shown some efficacy in treating newly diagnosed GBM (20). Lomustine is an oral alkylating nitrosourea chemotherapy used in recurrent disease in GBM, and clinical data suggested benefits in combination with bevacizumab for recurrent GBM (10). The combination therapy of bevacizumab with other chemotherapies such as temozolomide and lomustine was supposed to be more effective than that of the monotherapy (21,22). Hence, clinical trials have been initiated to explore the efficacy of combination therapy (23)(24)(25). The available clinical trials showed inconsistent results of the combination therapy in terms of the safety and efficacy for the treatment of GBM. In order to explore the potential factors that might influence the efficacy of bevacizumab combination therapy, we calculated HRs according to treatment regimen. Subgroup analysis suggested that PFS was significantly prolonged in both bevacizumab plus temozolomide group and bevacizumab plus lomustine group, whereas OS was not significantly prolonged in any of these subgroups. We found that the combination of bevacizumab and lomustine was associated with a better PFS than bevacizumab plus temozolomide, thus we speculated that bevacizumab plus lomustine may be superior to bevacizumab plus temozolomide in enhancing PFS. Moreover, we found that PFS varies significantly according to GBM type, and patients with recurrent GBM tend to have a better PFS than patients with newly diagnosed GBM. Our results implied that bevacizumab-containing regimens may be relatively more effective in the second-line treatment of recurrent GBM. Among the included studies, however, temozolomide was used in the first-line treatment for treating newly diagnosed GBM, and lomustine was used in the second-line treatment for treating recurrent GBM. Therefore, we performed separate analyses on patients treated with bevacizumab plus temozolomide in the first-line treatment and bevacizumab plus lomustine in the recurrent GBM. Our separate analyses suggested that patients with both newly diagnosed and recurrent GBM could benefit from bevacizumab containing treatment in PFS, but not in OS. Our analysis implied that recurrent patients who were treated with bevacizumab and lomustine may associate with a relatively higher PFS in the  second-line treatment than that of bevacizumab plus temozolomide in the first-line treatment. However, PFS is influenced by pseudo progression (PsP) and -response in varying degree in first-and second-line treatment (PsP is more frequent in first-line treatment), which could bring some uncertainty to our results. The differentiation between PsP and tumor progression has not been reported in the included studies, therefore, future researches should focus on this factor to confirm our results. In addition, developing advanced techniques and approaches to distinguish between PsP and true progression is also important for the treatment of GBM. As shown in Figure 3, bevacizumab-containing regimens tend to be associated with a prolonged OS in patients with GBM, but the improvement is not significant. Some of the included studies (9,10,13,18) in the analysis allowed cross-over, and the frequency of cross-over to bev-containing treatment was ranged from 8 to 81.8%. This potentially influenced OS. Future researches should include more bevacizumab-containing trials or consider using more stringent inclusion criteria to indicate whether bevacizumab helps improve OS efficacy. In this analysis, all the included studies of bevacizumab plus temozolomide were combined with radiotherapy, therefore, the comparison between temozolomide single treatment regimens and bevacizumab plus temozolomide could not be made in our analysis. Moreover, our subgroup analysis evaluated the effects of bevacizumab plus temozolomide for newly diagnosed GBM with different MGMT methylation status. We found that both PFS and OS was not affected by MGMT methylation status. PFS was significantly prolonged in all the subgroups of MGMT methylation status, but OS was not. Our results were consisted with a pervious meta-analysis (26), which suggested that bevacizumab combined with temozolomideradiotherapy was correlated with improved PFS for the treatment of patients with different MGMT methylation status of newly diagnosed GBM. Similar to the previous meta-analysis, our study did not find evidence for synergistic effects of bevacizumab-containing regimens on improving OS in GBM with different MGMT methylation status. Currently, there are no biomarkers that have been validated for clinical use to predict response to bevacizumab. Tissue markers such as VEGF, CD68þ and CD11þ tumor-associated macrophages, myoinositol, angiotensinogen, and HLA class II gene expression have been studied and suggest possible correlation with bevacizumab treatment response. However, the included studies did not report such data for us to conduct a subgroup analysis.
Recurrent GBM is different from newly diagnosed GBM, it is a severer disease characterized by poor physical condition, reoperation, multiple radiotherapy, and chemotherapy (27). Nearly all patients experience recurrence following standard-of-care surgical resection, radiotherapy, and temozolomide (2). Our study implied that the combination of temozolomide and lomustine is a potential treatment option for patient with recurrent GBM. When patients are treated with the combination of temozolomide and lomustine, the mechanism behind the observed increase in PFS is still not completely clear. It has been proposed that the normalization of vessels around the tumor, improvement of regional cerebral blood flow, augmentation of the antitumor effects of chemotherapy and radiotherapy are key components of the antiangiogenic activity (3,28,29). However, only 4 studies on lomustine combined with bevacizumab were included in our analysis, and more studies should be included to confirm our conclusions in the future study.
Although bevacizumab is widely used for treatment of GBM, it is well known that AEs due to bevacizumab can cause early discontinuation of treatment. It was reported that the development of AEs to bevacizumab or bevacizumab-containing regimens is associated with unfavorable glioma-related survival outcomes in patients with GBM (30). Therefore, we evaluated the AEs of bevacizumab in patients with GBM. We found that bevacizumab was associated with an increased risk of all-grade hypertension and proteinuria, as well as high-grade thromboembolic, hypertension and infection. A previous meta-analysis suggested that bevacizumab treatment significantly increases the risk of all-grade and highgrade infectious events developing in cancer patients, which was partly consist with our results that bevacizumab only increases the risk of high-grade infectious in GBM patients (31). The pathogenesis of bevacizumab-related infections is currently unknown, and neutropenia could be a possible mechanism that links VEGF-targeted agents and the risk of infections (32). It was noteworthy that the incidences of all-grades and high-grade hypertension were 3.15 and 4.16 times more likely to occur in patients receiving bevacizumab than those of controls, respectively. Hypertension is one of the most common cardiovascular toxicities caused by VEGF-targeted therapy, which may result in a decreased functional capacity and quality of life. Serious hypertension can be the reason for discontinuation of VEGFtargeted therapy. Therefore, hypertension should be well concerned and cancer patients receiving bevacizumab require a close monitoring.
A previous meta-analysis (27) evaluated the value of bevacizumab combined with chemotherapy and single-agent therapy in recurrent GBM treatment. This meta-analysis suggested that bevacizumab combined with chemotherapy can significantly improve PFS, but did not prolong OS, and can even lead to higher odds of AEs. However, this meta-analysis mainly focused on the patients with recurrent GBM, and bevacizumab in both treatment and control arms were included. Another previous meta-analysis (33) evaluated effectiveness of lomustine combined with bevacizumab in GBM. This study suggested that bevacizumab plus lomustine can effectively increase OS, PFS, and 6-month PFS in patients with GBM. Another meta-analysis (34) also showed that bevacizumab therapy was associated with a longer PFS in adult patients with newly diagnosed GBM, but had an inconsistent effect on OS in this patient population. However, this meta-analysis mainly focused on the first-line setting of bevacizumab, and all the included RCTs were conducted before 2018. Moreover, the individual adverse events have not been evaluated in these previous meta-analyses. A previous study (35) assessed the efficacy of bevacizumab in recurrent GBM patients, and they found that bevacizumab can effectively increase the rate of objective response and longer median PFS in patients with recurrent GBM. However, this meta-analysis mainly focused on the recurrent GBM, and they did not explore the potential factors that might influence the efficacy of bevacizumab therapy such as treatment regimens or MGMT status. A previous study (36) even revealed the OS benefits. However, this previous study mainly studied the lomustine and bevacizumab combination therapy versus the single therapy of lomustine or bevacizumab. They found that lomustine and bevacizumab can effectively increase OS and PFS than the single use of lomustine or bevacizumab in patients with GBM. Our study focused on bevacizumab. We studied the lomustine and bevacizumab combination therapy versus the single therapy of lomustine, and bevacizumab in both treatment and control arms were excluded. Therefore, we did not find the OS benefits of the combination therapy. In addition, there are some other previous metaanalysis studied bevacizumab in the treatment of GBM, but these studies were conducted a few years ago, and the newly published RCTs were not included in these analysis (37)(38)(39)(40). The present meta-analysis included 11 phase II and phase III RCTs of bevacizumab in the treatment of both newly diagnosed and recurrent GBM, and the effects of treatment regimen on curative effect were also analyzed. To the best of our knowledge, this is the largest and most up-to-date meta-analysis evaluating the efficacy and safety of bevacizumab in treating GBM. According to our analysis, we recommend that bevacizumab could be used in combination with temozolomide or lomustine in treating newly diagnosed or recurrent GBM, respectively. The adverse events associated with bevacizumab combination therapy were generally manageable. Hpertension should be close monitored and certain measures should be taken if necessary.
Our study has some limitations. First, this study was a study-level meta-analysis, and there is publication bias leading to heterogeneity among the included studies. Moreover, the incorporated studies were conducted by different researchers from various institutions, and the PFS response assessments differ between the included studies. Therefore, the outcomes may have not been reported consistently across studies because of subjectivity and disparities in investigator interpretation. Second, only 11 RCTs were included in the present study. If even a minor clinical difference existed, a great heterogeneity would be found. Therefore, more large-scale, multi-center RCTs of bevacizumab-containing regimens are needed for the future study. Third, the present meta-analysis suggested that bevacizumab-containing regimens can significantly improve PFS, but did not prolong OS, which may due to the limited number of included studies. Future researches are needed to analyze whether bevacizumab helps improve OS efficacy.
In summary, our meta-analysis provided evidence that bevacizumab significantly improved the PFS in patients with GBM, but did not prolong OS in these studies. PFS was significantly prolonged in both newly diagnosed GBM patients who were treated with bevacizumab plus temozolomide and recurrent GBM who were treated with bevacizumab plus lomustine, but OS was not. Recurrent patients who were treated with bevacizumab and lomustine were associated with a relatively higher PFS than that of bevacizumab plus temozolomide in the first-line treatment. Bevacizumab combined with temozolomide was correlated with improved PFS for treatment of patients with different MGMT methylation status for newly diagnosed GBM. Bevacizumab was associated with an increased risk of hypertension, proteinuria, thromboembolic, and infection. Hypertension was the most common AEs that should be well concerned. These conclusions provide concrete evidence for further researches on bevacizumab-containing regimens for GBM.

Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors. For this type of study, formal consent is not required.

Author contributions
XiaoHong Xie and Shangyi Bao conceived and designed the study. Liuying Li and XiaoJun Fu wrote the main manuscript text. Hong Zhao and XiaoHong Xie extracted and reviewed the data. Liuying Li and XiaoJun Fu analyzed the data.

Declaration of interest
The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.