Introduction

Glioblastoma (GBM) is the most aggressive type of primary malignant tumor of the brain in adults [1]. Despite the new combination of Stupp protocol, including radiation and chemotherapy with maximal surgical resection and tumor-treating field (TTF), the prognosis remains unsatisfactory as most tumors recur in situ [1, 2]. Several interventions, including targeted therapies, have been attempted to improve the prognosis of GBM. As GBM is a hyperemic tumor involving the upregulation and activation of VEGFA and HIF [3], VEGFA is a reasonable target molecule in the treatment of GBM. Bevacizumab (BEV), a humanized monoclonal antibody inhibiting VEGFA, was considered a promising candidate for treating GBM, given its clinical benefits in other cancers such as colorectal cancer [4], renal cell carcinoma [5], non-squamous non-small cell lung cancer [6], and cervical cancer [7]. Success in the treatment of other tumors persuaded researchers to conduct phase III AVAglio and RTOG 0825 clinical trials in patients with newly diagnosed GBM. However, both clinical trials didn’t improve the overall survival (OS) in the BEV treatment arm. Further, a randomized phase II TAVAREC clinical study demonstrated that BEV treatment had no significant improvement on progression-free survival (PFS) and OS in Grade 2 and Grade 3 gliomas [8]. A phase III trial by Wick et al. did not find any OS benefits with combined therapy of BEV plus lomustine, compared with lomustine alone [9]. Based on these several clinical trials, BEV is considered ineffective in prolonging OS for recurrent GBM (rGBM) by the European Association Neuro-Oncology (EANO) [10, 11]. Nevertheless, clinical benefits other than the prolongation of survival were possibly observed. The EORTC protocol demonstrated that BEV decreased steroid dependence and relieved para-tumor edema in patients with GBM [8]. Despite a lack of evidence supporting its ability to prolong OS, BEV was approved by the FDA (U.S. Food and Drug Administration) in 2009 as a treatment for rGBM and was included in the 2021 EANO guidelines due to its demonstrated improvement in quality of life and safety [11].

BEV might not be suitable for the treatment of all rGBM patients in general based on the outcome of these randomized trials. In the AVAglio trial, subgroup analysis revealed that the TCGA-proneural GBM subtype had an OS benefit from the administration of BEV. Further, epigenetic mechanisms could also influence the sensitivity of BEV, as demonstrated by Cloughesy et al.‘s finding that methylguanine-DNA methyltransferase (MGMT) methylation may be predictive for onartuzumab (ONA) + BEV outcomes in GBM. It is necessary to perform subgroup analyses to specifically identify the survival benefits of the treatment of BEV. However, no consensus has been reached regarding the subset of rGBM patients who are sensitive to BEV. Furthermore, the optimal combination therapy, dosage efficacy, and correct indication for BEV therapy are still controversial.

Given the considerable uncertainty surrounding BEV treatment strategies, we aimed to systematically review the current evidence associated with BEV therapy by mapping evidence. We aimed to answer the following five questions: (1) Could BEV-containing regimens bring survival benefits to patients with rGBM, compared with non-BEV treatment regimens? (2) Could BEV combined with other therapies prolong the OS of patients with rGBM, compared with BEV monotherapy? (3) Could BEV treatment improve quality of life (QoL) and reduce the adverse events (AEs) in rGBM? (4) Could some subgroups harboring specific clinical or molecular characteristics gain survival benefits from BEV treatment? (5) What are the optimal dosages and indications for the BEV treatment in rGBM?

Methods

Search strategy and study selection

The scoping review and mapping evidence were conducted following the PRISMA extension for scoping reviews [12]. A comprehensive literature search was performed in electronic databases including PubMed, Embase, and the Cochrane Library, on March 27th, 2022.

Inclusion criteria

(1) Patients with recurrent high-grade glioma (WHO grades 3–4) or GBM (WHO grade 4), regardless of age, gender, or pathological type; (2) Patients who were treated with BEV alone or in combination. Treatment types were focused on but were not limited to BEV alone, or BEV plus radiotherapy, chemotherapy (including carmustine implants), chemoradiotherapy, surgery, immunotherapy, and TTFs; (3) The outcomes of interest included OS, PFS, QoL, and AEs (cerebral edema and cognitive deficits) incidence; (4) Study types included randomized controlled trials (RCT), case-control trials (CCT), observational trials, pre- and post-control studies, and systematic reviews.

Exclusion criteria

(1) Case reports and conference abstracts; (2) Protocols but not reports of the study result; (3) Studies that were not reported in English.

Two reviewers independently screened the titles and abstracts of the retrieved records. Following the initial screening, full texts of the trials that passed title/abstract screening were scrutinized to confirm eligibility for the analyses. Disagreements were resolved by discussion with a third person if necessary. A PRISMA flow diagram was constructed to show the full article-selection process.

Data extraction

Two authors (Minjie Fu and Xiao Huang) independently examined the studies and extracted data using a standardized spreadsheet with the following characteristics: trial type, number of participants, type and administration of interventions, the definition of outcomes, measurement variables, and key findings. In situations of discrepancies, the third author (Zhirui Zhou) was consulted for final decision-making.

Data coding and definition

Selected studies were coded according to the type and administration of interventions (BEV monotherapy or BEV combined therapy). Classification criteria were discussed by the professional group. The term “Beneficial” was defined as a finding that had one or more of the following results: prolonged OS or improved QoL or PFS. The term “Harm” was defined as a finding that had one or more of the following results: decreased OS or PFS or worse QoL. The term “No difference” was defined as no significant difference or no difference reported between the groups for OS, PFS, or QoL. “Inconclusive” was defined as a finding that demonstrated both beneficial and harmful results in the studies.

Presentation of evidence mapping

We provided a scoping review and mapping evidence through a descriptive table that consisted of the characteristics of selected studies. The narrative description was presented.

Results

Study selection

A primary search yielded a total of 405 studies. After the removal of 15 duplicated publications, 390 studies were subsequently screened. Subsequently, full texts of 132 studies were scrutinized for eligibility. Ultimately, 90 studies met the eligibility criteria for inclusion in the scoping review and mapping evidence, comprising 2 phase I trials, 22 phase II trials, 2 phase III studies, 5 prospective studies, 36 retrospective studies, and 23 reviews (see Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Could BEV-containing treatment regimens bring survival benefits to patients with rGBM, compared with non-BEV treatment regimens?

Table 1 The therapeutic efficacy of BEV-containing treatment regimens and non-BEV treatment regimens
Full size table

In total, 31 studies (2 phase III studies, 5 phase II studies, 2 prospective studies, 9 retrospective studies, and 13 reviews, see Table 1) compared the therapeutic efficacy of BEV-containing treatment regimens with non-BEV treatment regimens. Of these, 17 studies investigated the benefits of adding BEV to chemotherapy (1 phase III trial, 1 phase II trial, 4 retrospective studies, and 11 reviews), while 5 studies investigated the efficacy of BEV plus lomustine (1 phase III trial, 3 phase II trial, and 1 retrospective study). Although a phase II study by Taal et al. showed the OS benefits of BEV plus lomustine versus the lomustine monotherapy group (mOS: LOM vs. BEV/LOM 110 vs. BEV/LOM 90, 8 months vs. 16 months vs. 11 months) [13], the other four randomized studies (including a phase III trial by Wick et al.) didn’t support this finding. Other phase II/III trials did not identify the OS benefits of BEV with a range of other chemotherapy partners (temozolomide (TMZ), trebananib, irinotecan, and nivolumab) compared with the non-BEV regimen.

Four studies (3 retrospective studies and 1 review) reported that the combination of BEV and radiotherapy improved OS, compared with radiotherapy alone. Meanwhile, 1 retrospective study and 1 prospective study on BEV plus re-surgery regimen showed that rGBM patients benefitted from BEV after receiving re-surgery.

Although some randomized clinical trials showed positive effects of the BEV-containing regimen on PFS, other palliative effects, and neurological improvement as meaningful benefits, gain on OS was not observed among the entire patient population in the majority of the trials.

Could BEV combined therapy prolong the OS of patients with rGBM compared with BEV monotherapy?

Because BEV treatment alone lacked evidence to prolong OS of patients with rGBM, 41 studies were further conducted to identify the optimal combination therapies. These studies included 14 phase II trials, 1 phase I trial, 15 retrospective studies, 1 prospective study, and 10 reviews (Table 2).

Table 2 The therapeutic efficacies of BEV monotherapy and combined therapy
Full size table

A range of chemotherapy candidates was studied, including lomustine, ONA, celecoxib, vorinostat, dasatinib, valganciclovir, and trebananib. The phase II trials by Taal et al. and Weathers et al. did not find the OS benefits of the addition of lomustine to BEV [13, 14], while the results varied in the two retrospective studies [15, 16]. Although several studies found that BEV plus lomustine could prolong PFS, compared with BEV monotherapy, its benefits on OS warranted further validation [17, 18]. In addition to lomustine, the OS benefits of Irinotecan (IRI), osimertinib, and valganciclovir were reported in some retrospective studies [19,20,21]. But currently, no high-quality evidence from RCT was found to further verify their positive effect on OS.

Five studies investigated the efficacy of BEV plus radiotherapy versus BEV monotherapy (1 prospective study and 4 retrospective studies) on OS. Although the prospective study found no significant difference in OS between the combination group and monotherapy group [22], the other four retrospective studies stated that radiotherapy plus BEV improved the rGBM prognosis by enhancing OS [23,24,25,26].

A retrospective study by Yamaguchi et al. in 2021 showed that the BEV plus re-surgery improved OS (mOS, Cytoreductive surgery + BEV vs. BEV, 16.3 months vs. 7.4 months, p = 0.0008) [27] while another retrospective study in 2017 did not find any difference between BEV combination and single regimen groups [28].

As TTF has emerged as a promising technique for tumor therapy, the efficacy of TTF plus BEV was also elucidated. A post-analysis of the EF-14 trial demonstrated that the combination of BEV and TTF brought more OS benefits, compared with BEV alone (mOS, TTF + BEV vs. BEV: 11.8 months vs. 9.0 months, p = 0.043) [29].

The combinatory partners of BEV were widely studied, and some BEV combined therapies (especially with lomustine and radiotherapy) were proved to have superior efficacy to BEV monotherapy. But additional research is required to determine the optimal combination of treatment modalities.

Median OS reported in the studies included in the analyses is summarized in Fig. 2. Although it was difficult to prove the OS benefits of BEV treatment through a single study, there was a trend to suggest that rGBM patients treated with BEV combined therapy may experience longer median OS.

Fig. 2
figure 2

Median OS of patients with rGBM reported in studies

Full size image

Could BEV treatment improve the quality of life and reduce the adverse events in rGBM?

In total, 19 studies (1 phase I trial, 4 phase II trials, 1 phase III trial, 4 retrospective studies, and 9 reviews) investigated the BEV effect on QoL and AEs (edema and cognitive dysfunction) (Table 3). While the effect of BEV monotherapy and combined therapy on OS prolongation remains unclear and controversial, three studies have verified BEV’s potential to reduce steroid use [30,31,32]. Additionally, three studies have reported that BEV could reduce the AEs induced by radiotherapy [33,34,35]. BEV also effectively controlled the tumor mass. However, only two retrospective studies found that the health-related QoL improved after receiving BEV containing therapy [30, 36], while other studies, including a phase II trial, did not find associations between BEV treatment and QoL [37, 38]. A review suggested that BEV combined therapy increased the incidence of side effects compared to BEV monotherapy [39]. Therefore, the potential for BEV to improve QoL remains uncertain and requires further validation.

Table 3 The effect of BEV on improving quality of life and reducing adverse events
Full size table

Could subpopulations harboring some clinical or molecular characteristics gain survival benefits from BEV treatment?

A total of 17 studies (6 phase II trials, 2 prospective studies, and 9 retrospective studies) analyzed the types of rGBM that may favorably benefit from BEV-containing therapies. These studies analyzed the association between different genetic alterations, such as MGMT methylation, IDH mutation, and EGFR alteration and clinical features such as age groups, laboratory examinations, and radiological characteristics (Table 4).

Table 4 Responses to BEV in different rGBM subpopulations
Full size table

MGMT methylation status

MGMT methylation status was assessed in six studies (4 phase II trials and 2 retrospective studies) to determine its association with responses to BEV [38, 40, 41]. A phase II trial found that BEV plus ONA improved OS in patients with rGBM having unmethylated MGMT (mOS, ONA + BEV vs. PLA + BEV, 10.9 vs. 7.5 months, p = 0.0836), compared with BEV plus placebo while BEV monotherapy favored outcome in patients with rGBM harboring methylated MGMT (mOS, ONA + BEV vs. PLA + BEV, 7.7 months vs. NR, p = 0.0150) [40]. A retrospective study on BEV plus osimertinib treatment was marginally effective in most GB patients with simultaneous EGFR amplification plus EGFRvIII mutation [20]. Another retrospective study compared the post-recurrence survival between patients with MGMT methylation and unmethylation, treated with BEV plus alkylating agents and found no difference between the two groups [41]. Nevertheless, another phase II trial did not find differences in QoL between the groups with GBM having MGMT methylation and unmethylation to BEV plus TMZ [38].

IDH mutation status

The association between IDH mutation status and response to BEV has been investigated in one phase II trial and two retrospective studies. Subgroup analysis of the BELOB trial revealed that patients with IDH mutation had higher OS and PFS compared to the control (mOS, IDH mutant vs. IDH wildtype: 20 vs. 9 months, p = 0.021) [13]. Dono et al. revealed an association between the genetic alterations and response to stereotactic radiosurgery (SRS) and BEV-containing chemotherapy in patients with rGBM carrying IDH-wildtype. Moreover, PTEN mutant subgroup in IDH WT group was found to have longer PFS and OS after combination therapy (mOS, PTEN mutant vs. PTEN wildtype: 22.5 vs. 13.6 months, p = 0.07; mPFS, PTEN mutant vs. PTEN wildtype: 17.5 vs. 8.1 months, p = 0.04) [42]. A retrospective study conducted by Lv et al. revealed that rGBM carrying IDH mutation had a better prognosis (OS and PFS) after receiving a BEV-containing regimen, compared with rGBM without IDH mutation (BEV monotherapy, mOS, IDH mutant vs. IDH wildtype: 10.16 vs. 4.9 months; mPFS, IDH mutant vs. IDH wildtype: 3.23 vs. 1.37 months, p = 0.04; BEV plus sunitinib, mOS, IDH mutant vs. IDH wildtype: 7.53 vs. 4.83 months; mPFS, IDH mutant vs. IDH wildtype: 2.07 vs. 1.10 months, p = 0.06), while no difference was found between IDH wildtype and mutated rGBM receiving non-BEV regimens (cetuximab and sunitinib) [43].

EGFR alteration status

A phase II trial found that EGFR vIII positive rGBM had PFS and OS benefits from BEV plus rindopepimut therapy (HR for BEV plus rindopepimut, 0.58, p = 0.01).

Radiological characteristics

Apart from genetic alterations, the association between radiological examination outcome and response to BEV was elucidated. Cox regression analysis in a phase II trial showed that BEV improved survival in patients with large enhancing tumors with low apparent diffusion coefficient (ADCL). It also revealed that the pretreatment tumor volume was an independent risk factor for the BEV-treated group [44]. A prospective study revealed that patients with hyperintense lesions in T1 and diffusion-weighted restriction (double-positive) benefited more than others from BEV treatment [34, 45]. A retrospective study demonstrated that rGBM with a large tumor burden might be benefitted most favorably from BEV-containing regimens [15].

Laboratory examinations

A prospective trial in 2019 stated that low neutrophil counts (below 3.9 G/L) and high Treg counts (above 0.011 G/L) predicted prolonged OS [46].

Age groups

No consensus was found regarding the association between BEV efficacy and age groups. Two retrospective studies found that there was a better improvement in non-elderly patients with rGBM/recurrent high-grade glioma (rHGG) patients compared with elderly patients treated with BEV-containing regimens [47, 48]. However, another retrospective study concluded controversially that elderly patients had more prognostic benefits compared with younger patients [49].

What are the optimal dosages and indications for BEV administration?

The optimal dosages and indications for BEV administration are still under investigation. In the US, the recommended dosage of BEV in the US is a 10 mg/kg intravenous infusion administered every 2 weeks. However, different studies (2 retrospective studies and 2 reviews, Table 5) have adopted varying dosages, and recent research has elucidated the optimal dosage. Two retrospective studies stated that lower doses were at least equal or even superior to the recommended doses [50, 51]. Two reviews had similar conclusions [37, 49]. Although BEV at the recommended dose and lower dose exhibits equal efficacy on survival, influence on other outcomes such as QoL and side effects reduction needs further investigation.

Table 5 The optimal dosage and indication for the BEV treatment
Full size table

The window of opportunity for BEV treatment is also still under debate. Matsuoka et al. argued that the initiation of a treatment regimen containing BEV at first recurrence may improve prognosis. However, they also noted that BEV administration could lead to chemotherapy resistance and rapid progression in some cases [34]. Similar conclusions were made in other studies. A retrospective study found that BEV treatment before surgery might be beneficial for young and high-performance patients [52]. No significant difference in OS was identified between patients receiving BEV-containing regimens after the first relapse and the second relapse [53]. However, some studies concluded contrastingly. Funakoshi et al. found that BEV administration after recurrence (post-BEV) improved PFS and deterioration-free survival (DFS) than pre-recurrence BEV administration (pre-BEV) (mPFS, post-BEV vs. pre-BEV: 9.9 vs. 7.5 months, p = 0.0153; mDFS, post-BEV vs. pre-BEV: 13.8 vs. 8.5 months, p = 0.0046) [54]. Therefore, the optimal opportunity window of BEV treatment warrants further validation through future large-scale clinical trials. Table 5 summarizes the different findings across studies.

Discussion

BEV has shown improved PFS in clinical studies, but OS benefits have not been consistently observed. Despite this, BEV has been proposed as a promising drug in GBM due to its ability to reduce side effects from steroid use and radiotherapy. To further maximize benefits from BEV treatment, investigations could be summarized in two ways. One was to combine BEV with other treatment modalities to enhance synergistic anti-tumor effects. The other one was to identify the BEV-response groups which could gain more prognostic benefits from the treatment of BEV. Additionally, we investigated the optimal dosage and treatment opportunity window to maximize the BEV treatment benefits. To the best of our knowledge, BEV-containing multimodality treatment was associated with clinical benefit and is worthy of administration. The outcome depends on the unique clinical and molecular features linked to varied BEV responses.

Despite many efforts in the past, the efficacy of BEV remains to be optimized and needs further investigations focusing on the two mechanisms mentioned above. First, newly emerging therapies for rGBM bring further opportunities for BEV-containing multimodality treatment. TTF was the landmark therapy in the treatment of GBM [55]. Post-hoc analysis of EF-14 in a phase III trial on newly diagnosed GBM revealed that the addition of TTF to BEV could further prolong the median OS by 2 months beyond the period that patients with rGBM achieved with second-line treatment alone [56]. Studies of higher evidence are warranted to investigate the efficacy of BEV plus TTF combination therapies. Besides TTF, an increasing number of combination therapies are currently explored via several clinical trials (e.g., NCT02511405, VB-111 plus BEV; NCT01308684, RO5323441 plus BEV; NCT01349660, BKM120 plus BEV).

Second, biomarker-enrichment strategies are warranted to direct the clinical administration of BEV. While BEV administration has been shown to improve OS in the TCGA-proneural newly diagnosed GBM subtype, characterizing rGBM according to TCGA transcriptome classification in a realistic manner requires further exploration. Moreover, high-quality evidence is lacking regarding the associations between molecular and clinical features with BEV response. Therefore, RCTs focusing on specific subpopulations of rGBM are warranted.

In summary, current RCTs are not sufficient to make a definitive statement that BEV could improve OS and QoL in patients with rGBM although some clinical benefits (including PFS, decreased steroid use, and cognitive ability protection) are observed. Combing BEV with TTF and administration at first recurrence may improve prognosis. In the meantime, rGBM with low ADCL, large tumor burden, or IDH mutation is more likely to benefit from BEV treatment. Of note, observational studies have yielded conflicting results due to heterogeneity. High-quality clinical trials are needed to gain new insights into BEV treatment, and breakthroughs may emerge from the use of BEV-containing multimodality treatment on unique subpopulations of rGBM.