Next-generation therapeutics for glioblastoma: Challenges and future directions

Abstract

Glioblastoma (GBM) is the most common and aggressive primary malignant brain tumor in adults. Although recent World Health Organization classifications integrating molecular features have improved diagnostic precision, therapeutic outcomes for IDH-wild-type glioblastoma remain dismal. Standard treatment with radiotherapy and temozolomide has changed little over the past two decades, and most investigational therapies have failed to produce durable clinical benefit. Tumor heterogeneity, adaptive resistance mechanisms, a profoundly immunosuppressive tumor microenvironment, and limited drug delivery across the blood–brain barrier have collectively contributed to these failures. This review summarizes the evolving molecular landscape of glioblastoma and examines emerging therapeutic approaches designed to overcome these barriers. We discuss advances in molecularly targeted therapies, including strategies directed at BRAF and FGFR alterations, as well as continued efforts to address EGFR-driven disease. We also review immunotherapeutic approaches such as immune checkpoint inhibition, chimeric antigen receptor T-cell therapies, and oncolytic viruses, highlighting key clinical trial results and biological challenges. In parallel, we explore metabolic targeting strategies and novel technologies aimed at improving central nervous system drug delivery, including focused ultrasound and convection-enhanced delivery. Collectively, current evidence indicates that meaningful progress in glioblastoma will require biomarker-driven patient selection, rational combination strategies, and improved methods for intracranial drug delivery. While substantial challenges remain, ongoing translational and clinical efforts provide a framework for the development of more effective and personalized therapeutic strategies.

Introduction

The classification of central nervous system (CNS) tumors has been fundamentally reshaped by the integration of molecular biology into routine pathology. Adult-type diffuse gliomas, the most common malignant primary brain tumors in adults, are now divided into three subtypes based on distinct molecular alterations. According to the most recent WHO classification of CNS tumors, glioblastoma (GBM) is defined by the absence of isocitrate dehydrogenase 1/2 (IDH1/2) mutations, together with the classical histological features of microvascular proliferation and/or necrosis, or by additional molecular criteria such as TERT promoter mutations, EGFR amplification, or combined whole chromosome +7/−10 copy-number changes [1].

Despite these advances in classification, IDH-wildtype GBM remains a highly lethal disease, with a median overall survival of approximately 14 months and a 5-year survival rate of only 5–7% [2,3]. This dismal prognosis reflects the lack of meaningful therapeutic advances over the past two decades. Since the pivotal phase 3 study published in 2005, the combination of radiotherapy with the alkylating agent temozolomide has remained the standard of care for newly diagnosed glioblastoma [4]. Attempts to improve outcomes through alternative strategies have largely failed [[5], [6], [7], [8], [9]], with the exception of tumor treating fields (TTFs), which demonstrated a statistically significant survival benefit [10]. However, TTFs have not been widely adopted, in part due to the practical challenges of long-term device use, but also ongoing variability around access and ease of use.

In the recurrent setting, the absence of a definitive superior treatment has led to individualized decision-making. Therapeutic choices are typically based on prior response, toxicity profiles, clinical status, availability of clinical trials, targetable alterations, and patient preferences.

Several factors contribute to the therapeutic resistance of GBM. Clinical trials have traditionally adopted a “one-size-fits-all” approach, overlooking the biological heterogeneity of the disease and leading to generalized conclusions across diverse molecular subtypes. In addition, a profoundly immunosuppressive tumor microenvironment (TME) [11], together with challenges in drug delivery across the blood–brain barrier (BBB), particularly for larger molecules or agents that are substrates of ATP-binding cassette transporters such as P-glycoprotein, which actively efflux drugs back into the systemic circulation [12], has further limited therapeutic efficacy [13].

Nevertheless, decades of negative clinical trials, advances in other tumor types, and ongoing basic and translational research have created new opportunities. Emerging strategies, some of which have already shown promising preliminary results, aim to overcome these barriers. In this review, we highlight novel and upcoming therapeutic approaches for GBM, ranging from systemic therapies to innovative modalities designed to enhance drug delivery across the BBB.

Molecular landscape and therapeutic approaches

Glioblastoma is widely recognized as a highly heterogeneous tumor. Traditionally, this heterogeneity refers to the broad spectrum of morphological features observed by pathologists, ranging from cells with astrocytic characteristics to poorly differentiated, highly anaplastic forms. These include fusiform and pleomorphic cells, as well as mixed populations containing gemistocytic elements [1]. More recently, advances in genomic, transcriptomic, and epigenetic profiling have demonstrated that malignant glioblastoma cells differentiate into four distinct cellular states, with the proportion of cells in each state varying both within and between tumors [14]. These findings indicate that the marked morphological diversity of glioblastoma is paralleled by profound molecular heterogeneity across key signaling pathways, as well as by substantial heterogeneity of the tumor microenvironment. Some alterations represent potentially targetable drivers, including BRAF V600E mutations (1–2%) [15], FGFR3-TACC3 fusions (∼3%) [16], and NTRK fusions (∼1%) [17]. Others, such as TP53, PTEN, PDGFRA, and TERT promoter mutations, are more frequent but have not yet been successfully translated into effective targeted therapies. A particularly distinctive feature of the glioblastoma molecular landscape is the frequent coexistence of multiple alterations within the same gene. The most notable example is EGFR, which is altered in approximately 50–60% of glioblastoma. Interestingly, unlike many other tumor types in which EGFR alterations are typically truncal and clonal, EGFR alterations in glioblastoma often occur at subclonal levels. This results in marked intratumoral heterogeneity, with alterations such as EGFR amplification being present in distinct spatial regions of the same tumor [18]. EGFR amplification is the most common genetic alteration in glioblastoma, observed in approximately 40–50% of cases, and is characterized by marked intratumoral heterogeneity, with amplified and non-amplified subclones often coexisting within the same tumor [19]. Up to 70% of EGFR-amplified tumors harbor an additional EGFR alteration [18,[20], [21], [22]]. Among them, the best characterized is EGFRvIII, an in-frame deletion of exons 2–7 that produces a truncated receptor with constitutive, ligand-independent activation [19], contributing to enhanced oncogenic signaling, tumor heterogeneity and therapy resistance.

Initial efforts to better subclassify glioblastoma based on genomic profiling identified four different subtypes -classical, mesenchymal, neural and proneural- defined by the expression of signature genes. These subtypes correlate with differences in cellular linage, prognosis and response to therapy [20]. However, despite being proposed over 15 years ago, this classification, has not yet been relevant to routine clinical practice. A more recent computational approach using core biological traits of single cells and bulk tumors RNA-sequencing identified four GBM subtypes distributed along neurodevelopmental and metabolic axes: proliferative/progenitor, neuronal, mitochondrial and glycolytic/plurimetabolic. Interestingly, the mitochondrial subtype was associated with a better outcome and increased sensitivity to oxidative phosphorylation inhibitors [23], revealing a potential new therapeutically targetable pathway.

Targeted Therapies

BRAF alterations are one of the most frequently found driver mutations and have been successfully targeted in a variety of tumors, including glioblastoma. Recent analyses using the CODEai platform (Caris Life Sciences, Irving, Texas) demonstrated that BRAF V600E mutations are not prognostic of overall survival but represent a predictive biomarker of therapeutic benefit from MAPK inhibitors in patients with glioblastoma [24]. In the phase II basket ROAR trial, patients with recurrent BRAF V600E mutant low- and high-grade gliomas were treated with the dabrafenib–trametinib combination. In the glioblastoma subgroup (n = 31), the objective response rate (ORR) was 35% (95% CI, 16.7–51.4), with a modest median progression-free survival (PFS) of 2.8 months (95%CI, 1.8–13.7) [25]. Next-generation BRAF inhibitors, including PAN-RAF inhibitors and dimer breakers, capable of targeting both BRAF V600 and non-V600 alterations (including BRAF fusions) without inducing paradoxical MAPK pathway activation, have shown promising activity in early-phase clinical trials. Although data are scattered and not particularly promising in adults with GBM [26], tovorafenib demonstrated impressive ORR in pediatric low-grade gliomas, with 69% in patients harboring BRAF:KIAA1549 fusions and 50% in those with BRAF V600E mutations. With respect to safety, the most frequent grade ≥3 treatment-related adverse events were elevated creatine phosphokinase (12%), anemia (11%), and rash (8%). Other reported adverse events included decreased growth velocity (12%) and intratumoral hemorrhage, which occurred in 15 patients (10.9%) [27]. Although BRAF:KIAA1549 fusions are exceedingly rare in glioblastomas [28], a sub-study of the FIRELIGHT-1 trial evaluated the efficacy of tovorafenib in patients with relapsed solid tumors harboring BRAF fusions or RAF amplifications, including a cohort of eight patients with CNS tumors. The ORR for the entire cohort was 38% [29]; however, despite these encouraging results, the study was ultimately terminated by the sponsor. Similarly, plixorafenib an oral BRAF inhibitor that targets V600 and non-V600 alterations without paradoxical activation of the MAPK pathway [30] has demonstrated an ORR of 66.7% in patients with V600-mutant primary CNS tumors (n = 9), with a highly favorable safety profile, mainly including grade 1–2 liver function alterations, fatigue and nausea [31]. The phase II FORTE trial is recruiting for patients with recurrent BRAF V600 mutant or BRAF fusions positive CNS tumors (NCT05503797). A summary of the ongoing clinical trials discussed in this article is provided in Table 1.

Table 1.

Ongoing clinical trials.

Trial name NCT/EUCT number	Treatment	Patient population	Phase
FORTE NCT05503797	Plixorafenib (FORE8394)	≥10 years of age recurrent BRAF V600E gliomas	2
ETCTN:10559 NCT05859334	Erdafitinib	≥18 years of age with progressive FGFR-TACC fusion positive IDH-WT gliomas.	2
NCT06934889	ABBV-637 or ABBV-155 with ERAS-801	≥18 years of age with newly diagnosed or recurrent GBM with EGFR amplification	1
A072201 NCT06325683	Relatlimab/Nivolumab	≥18 years of age with recurrent GBM	2
GIANT trial NCT06816927	Relatlimab/Nivolumab plus RT and TMZ	≥18 years of age with newly diagnosed GBM	2
NCT06504381	DB107-FC (Toca FC), DB107-RRV (Toca 511) plus RT with or without TMZ	≥18 years of age with newly diagnosed GBM	1/2
NCT06264388	DB107-FC (Toca FC) with DB107-RRV (Toca 511)	≥18 years of age with recurrent AA or GBM	2
NCT05664464	Gabapentin, sulfasalazine and memantine plus RT and TMZ	≥18 years of age with newly diagnosed GBM	1/2
SENIPERA trial NCT07284069	Senicapoc and perampanel plus RT and TMZ	≥18 years of age with newly diagnosed GBM	1
PerSurge trial EUCT: 2023-503938-52-00	Perampanel (presurgical)	≥18 years of age with recurrent GBM	2 (Europe only)

NCT = National Clinical Trial; EUCT = European Union Clinical Trial; GBM = glioblastoma; AA = anaplastic astrocytoma; RT = radiotherapy; TMZ = temozolomide.

As we await results from the new BRAF inhibitors, both for canonical BRAF V600E mutations and for newly targetable BRAF fusions, important questions regarding optimal treatment strategies continue to arise, particularly given the different tolerability profiles of these new drugs. This emerging evidence suggests that these agents may demonstrate sufficient activity to warrant evaluation in first-line clinical trials, with the potential to complement or redefine the current standard of concurrent radiotherapy and temozolomide. In addition, with the rapid development of targeted therapies that act through different mechanisms, designing sequential treatment approaches will be difficult without a clearer understanding of the resistance pathways associated with these agents.

Current investigations

Alterations in the fibroblast growth factor receptor (FGFR) genes result in oncogenic mutant proteins with increased kinase activity and have been identified across multiple tumor types, culminating in a major clinical milestone with the approval of the FGFR inhibitor erdafitinib for FGFR2–3-mutant urothelial carcinomas. Approximately 3% of GBM harbor an FGFR oncogenic alteration [32]. The single arm phase 2 RAGNAR basket trial evaluated erdafitinib in multiple cancer types harboring FGFR1–4 alterations; among 30 heavily pretreated high-grade glioma patients, the ORR was 10% [33]. To further characterize the activity of FGFR inhibitors in gliomas, a multicenter phase II trial of infigratinib—a potent FGFR1–3 inhibitor—enrolled 26 patients with recurrent FGFR-altered gliomas and demonstrated a 3.8% ORR with a median PFS of 1.7 months [34]. Similarly, the FIGHT-209 study evaluated pemigatinib, another FGFR1–3 inhibitor, in patients with recurrent FGFR-altered gliomas. Cohort A included 79 patients with GBM, and cohort B included 9 patients with glioneuronal, neuronal, or other glioma subtypes. The ORRs were 8% and 22% for cohorts A and B, respectively [35]. Although all 3 trials failed to show an overall significant efficacy, interestingly in 2 of the trials, long lasting responses were reported in patients with FGFR3-TACC3 fusions [33,34]. FGFR3–TACC3 fusions are found in approximately 3% of glioblastomas and represent a highly oncogenic and recurrent gene fusion described in GBM. Preclinical models have demonstrated that tumors harboring FGFR3–TACC3 fusions are highly sensitive to FGFR inhibition [16,36]. Moreover, clinical data from patients with FGFR3–TACC3 positive glioblastomas show that these tumors exhibit distinct histologic features and are associated with more favorable outcomes compared with other GBM subtypes [37]. Consequently, in an effort to achieve greater molecular selectivity, the ETCTN 10559 trial (NCT05859334) is recruiting patients with recurrent IDH–wild-type gliomas harboring FGFR–TACC3 fusions to receive erdafitinib.

As previously noted, epidermal growth factor receptor (EGFR) alterations are among the most common molecular alterations in glioblastoma. Given the well-established efficacy of EGFR inhibitors in lung adenocarcinoma, it is not uncommon for patients, and even physicians, to inquire about the potential role of EGFR-directed therapies in glioblastoma. However, despite early optimism, clinical trials evaluating first- and second-generation EGFR tyrosine kinase inhibitors, EGFR-targeted vaccines, and EGFR-directed antibodies have consistently failed to demonstrate a survival benefit in either the recurrent or newly diagnosed GBM setting [[38], [39], [40], [41], [42], [43], [44]]. The rationale for the striking difference in therapeutic response between GBM and lung adenocarcinoma lies not only in the known clonal heterogeneity of GBM cells and the limitations of drug delivery across the BBB, but mainly in the distinct EGFR mutations observed in these tumors. In lung adenocarcinoma, canonical mutations such as L858R (exon 21) or exon 19 deletion affect the intracellular tyrosine kinase domain and confer strong sensitivity to EGFR inhibitors. In contrast, the most common EGFR mutations in glioblastoma, EGFR vIII, involves deletions in extracellular domain and does not confer sensitivity to EGFR inhibitors, which are designed to bind into the altered ATP-binding pocket in the kinase domain [45]. EGFR amplification is one of the definitive criteria for GBM classification [1] and is present in approximately 40% of cases [20]. Nonetheless, gene amplifications do not necessarily imply oncogene addiction, that is, sustained dependence of cancer cell survival on a single activated oncogenic driver [46]. Consequently, targeting EGFR amplification with small molecules or with directed monoclonal antibodies has not demonstrated clinical benefit [44,[47], [48], [49]]. Although there is only a moderate correlation between gene amplification and protein overexpression, increased surface receptor density has served as the rationale for developing antibody-drug-conjugates (ADC) capable of delivering cytotoxic payload to selected antigen expressing cells.

Depatuxizumab mafodotin (Depatux-M, formerly ABT-414) is an ADC that couples Depatuxizumab (formerly ABT-806) a monoclonal antibody against EGFR that binds to EGFRvIII or amplified wild-type EGFR [50], to the microtubule toxin mafodotin. A phase 2 trial enrolled 260 patients with recurrent EGFR amplified glioblastomas and randomized them 1:1:1 to Depatux-M monotherapy, Depatux-M plus temozolomide, or a control arm of temozolomide or lomustine. Although the primary endpoint of overall survival (OS) for the combination vs. control arm was not met (HR 0.71, 95%CI 0.50–1.02) p = 0.06, long term follow up of prespecified subgroup analyses demonstrated a statistical significant benefit of Depatux-M over temozolomide in patients whose tumors recurred in more than 16 weeks after prior temozolomide treatment [51]. The larger INTELLANCE-1 phase 3 trial evaluated Depatux-M combined with radiotherapy and temozolomide versus radiotherapy, temozolomide and placebo in newly diagnosed EGFR amplified glioblastomas. A prespecified interim analysis showed no OS benefit (median OS 18.9 months vs. 18.7 months, HR 1.02, 95% CI 0.82–1.26, p = 0.63) and the trial was terminated early [7].

Despite these setbacks, efforts to target EGFR alterations in gliomas have continued. An ongoing phase 1b trial is recruiting patients with newly diagnosed or recurrent EGFR amplified glioblastomas to receive either ABBV-637, an ADC linking an anti-EGFR antibody to a BCL-XL inhibitor or ABBV-155, another first in class ADC composed of an antiB7H3 antibody conjugated to a BCL-XL inhibitor, in combination with ERAS-801 an oral potent inhibitor against EGFR amplifications, point mutations, and EGFRvIII. Notably, ABBV-155 does not target EGFR but rather B7H3, a transmembrane glycoprotein overexpressed in many solid tumors (NCT06934889).

Immunotherapy approaches in glioblastoma

Immunotherapy has thus far shown limited efficacy in glioblastoma, largely due to pronounced tumor-intrinsic heterogeneity, a highly immunosuppressive tumor microenvironment, and low baseline T-cell infiltration [52]. Randomized trials of single-agent immune checkpoint inhibitors targeting PD-1 have failed to demonstrate a survival benefit in unselected glioblastoma populations [8,52], prompting investigation of rational combination strategies aimed at overcoming adaptive immune resistance. LAG-3 is a coinhibitory receptor expressed on activated CD4⁺ and CD8⁺ T cells, that together with PD-1 contributes to T-cell exhaustion [53]. In the phase 2/3 RELATIVITY-047 trial, the addition of relatlimab, an anti-LAG3 monoclonal antibody, to nivolumab (anti PD-1) resulted in a significant survival benefit in patients with metastatic melanoma as compared to nivolumab alone [54,55]. Based on these results, the relatlimab-nivolumab combination is currently being investigated in patients with recurrent glioblastoma (NCT06325683), as well as in the perioperative setting (NCT06816927).

CAR-T cells in GBM

Based on a similar biological rationale of exploiting tumor specific antigen expression in glioblastoma, chimeric antigen receptor T-cell therapy (CART) has emerged as a potential strategy. CART cells are genetically engineered autologous T cells modified to express a synthetic receptor capable of recognizing antigens expressed on cancer cells [56]. Interleukin 13 receptor subunit ⍺ 2 (IL-13R⍺2), expressed in approximately 75% of GBM [46], served as a target in the first-in-human pilot study to evaluate the safety and feasibility of CART cell therapy in recurrent GBM. In this study, engineered CD8⁺ cells were delivered via a catheter directly into the resection cavity in 3 patients, with no significant adverse effects, and some evidence of biological activity, including reductions in IL-13R⍺2 expression on tumor cells [57]. A larger phase I trial conducted at the same institution treated 65 patients with recurrent gliomas (72% GBM) using IL-13Rα2-directed CAR-T cells delivered intratumorally, intraventricularly, or through combined routes. Median OS for patients with recurrent GBM was 7.7 months, confirming that this approach remained feasible and safe [58]. As expected, EGFRvIII has also been targeted using CAR-T therapy in recurrent glioblastoma. In a report of 10 patients treated with intravenously delivered EGFRvIII-CAR-T cells, 7 underwent tumor resection after infusion. CAR-T cells were identified in the resected tumors of 5 patients, demonstrating that peripherally infused CAR-T cells can traffic to the brain. Median OS was 8 months, and no radiographic responses were observed [59]. Building on these findings, a “second-generation” EGFRvIII-targeted CAR was developed to additionally secrete T-cell–engaging antibody molecules (TEAMs) directed against wild-type EGFR. These CARv4-TEAM-E cells were tested in the first-in-human phase 1 INCIPIENT trial via intraventricular administration in three patients. Short-interval MRI studies (<5 days after infusion) reported marked reductions in enhancing tumor volume; however, responses were transient in two patients (≤1 month), while one patient exhibited a sustained response lasting at least 150 days (5 months). Cerebrospinal fluid liquid biopsy revealed decreases in EGFR and EGFRvIII copy number in all three cases [60]. Bivalent CAR-T cells targeting both EGFR and IL-13Rα2 have also been evaluated. In a phase 1 study of intracerebroventricular delivery in 18 patients with EGFR amplified recurrent GBM, 56% experienced grade 3 neurotoxicity. By modified RANO criteria, the ORR was 8%, with a median PFS of 1.9 months. At a median follow-up of 8.1 months, OS data were still immature [61].

Overall, CAR-T cell therapy for GBM remains an evolving field. Although studies consistently demonstrate that this approach is technically feasible and generally safe, clinically meaningful and durable efficacy has yet to be achieved.

Oncolytic viruses

The use of viruses as therapeutic agents in cancer is an active and increasingly promising area of investigation. Oncolytic viruses, engineered from platforms such as herpes simplex virus, adenovirus, and poliovirus, exert antitumor effects through two principal mechanisms. First, they selectively replicate within malignant cells, leading to direct oncolysis and release of tumor antigens. Second, they enhance the immunogenicity of the TME by inducing the release of pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), which in turn activate innate and adaptive immune responses against the tumor [62]. Teserpaturev (formerly G47Δ) is a third-generation oncolytic herpes simplex virus type 1 and the first oncolytic therapy approved for glioblastoma, receiving regulatory approval in Japan in 2021. Approval was based on a phase II study in which 19 patients with residual or recurrent glioblastoma were treated with stereotactically guided intratumoral administration of the virus. One year after initiation of G47Δ therapy, 84.2% of patients were alive (the primary endpoint of the study). Median OS was 20.2 months from the start of viral therapy and 28.8 months from initial surgery. One patient achieved a partial response, while the remaining patients experienced stable disease as their best response. Fever was the most common adverse event, occurring in 17 patients [63].

DNX-2401 also known as tasadenoturev, is an oncolytic adenovirus, that was tested in a phase I trial in 25 patients with recurrent gliomas via intratumoral injection, resulting in 20% of patients surviving more than 3 years from treatment and 12% of patients having a >95% tumor reduction. Analysis from samples from 12 patients from another cohort that had tumor resection after viral delivery, showed tumor infiltration by T-cells [64]. These findings led to the CAPTIVE/KEYNOTE-192 phase 1/2 study evaluating intratumoral DNX-2401 in combination with systemic pembrolizumab in 49 patients with recurrent glioblastoma. ORR was 10.4%, and median OS was 12.5 months (95% CI: 10.8–14.6). Notably, five patients achieved objective responses, including two with durable complete responses lasting more than 45 months. All patients who responded had tumors characterized by a moderately inflamed microenvironment, suggesting that highly inflamed tumors, despite their abundance of immune cells, may not provide the most favorable substrate for this type of therapy [65].

Retroviruses interact with host cells by integrating fragments of their DNA into the host-cell genome. Leveraging this mechanism, retroviral replicating vectors (RRVs) are engineered to retain the elements necessary for viral replication, enabling in situ amplification as they spread through the tumor while simultaneously delivering therapeutic genes [66]. DB-107 (formerly TOCA-511, vocimagene amiretrorepvec) is an RRV that encodes a yeast cytosine deaminase transgene, which converts 5-fluorocytosine (an antifungal agent) into 5-fluorouracil (5-FU), a well-established antimetabolite chemotherapeutic. When DB-107 is administered intratumorally or into the resection cavity, it enables localized conversion of orally administered 5-fluorocytosine into cytotoxic 5-FU. In a phase 1 clinical trial, 56 patients with recurrent high-grade gliomas, including IDH-mutant astrocytoma and glioblastoma, received DB-107 injected into the resection cavity, followed by tumor resection and weekly oral 5-fluorocytosine. Investigators identified a subgroup of patients who derived the greatest benefit, achieving an ORR of 21.7%, interestingly all complete responses [67]. These patients were subsequently selected for enrollment into a phase III trial that included individuals with recurrent high-grade gliomas, who were randomized to receive DB-107 plus 5-fluorocytosine versus standard of care. A total of 403 patients were randomized in a 1:1 fashion between the two treatment arms. Median OS was 11.10 months in the investigational arm and 12.22 months in the control group (HR 1.06; 95% CI: 0.83–1.35; p = 0.62) [68]. In a retrospective analysis, patients harboring a DGM7 germline mutation demonstrated a significant overall survival benefit compared with DGM7-negative patients (18.3 months vs. 12.0 months in the standard-of-care group; HR = 0.50; 95% CI: 0.28–0.89; p = 0.0167). These findings suggest that DGM7 may represent a potential biomarker for improved response to DB-107–based therapy [69]. These findings have led to the ongoing trials testing DB107 plus 5-fluorocytosine in patients harboring germline DGM7 mutation in high grade gliomas in the newly diagnosed (NCT06504381) and recurrent (NCT06264388) settings.

Targeting Metabolism

Efforts to develop drugs targeting cellular metabolism in glioblastoma have thus far failed to demonstrate meaningful clinical benefit. Preclinical studies have shown that inhibition of oxidative phosphorylation with biguanides reduces mitochondrial-dependent ATP production and oxygen consumption, leading to decreased cellular proliferation and increased apoptosis [70,71]. Nonetheless, clinical trials with novel agents have been terminated due to toxicity [72,73], or in the case of metformin, consistently failed to demonstrate an improvement in survival [74]. Beyond the narrow therapeutic index of these novel agents, the lack of clinical benefit may also arise from the fact that current data have largely been generated in unselected patient populations. As mentioned above, a recent classifier identified a glioblastoma subtype (mitochondrial GBM) that rely exclusively on oxidative phosphorylation for energy production. This subgroup demonstrated sensitivity to mitochondrial complex I inhibition, including agents such as metformin [23]. These findings suggest that metabolic-targeted therapies may require a more defined patient selection to reveal meaningful clinical efficacy.

Next generation oxidative phosphorylation (OX-PHOS) inhibitors may address some of the limitations in the therapeutic use of these inhibitors. Lixumistat (formerly IM156), a novel biguanide with higher brain penetration and proven antitumoral activity in GBM preclinical models [71], has reached the clinical development stage and is being tested in patients with pancreatic adenocarcinoma (NCT05497778). A window-of-opportunity study using lixumistat in newly diagnosed GBM is under development.

Enhancing delivery across the BBB

Low intensity focused ultrasound (FUS)

Major limitations in the treatment of glioblastoma include the difficulty of identifying effective therapeutic targets, selecting the appropriate patient populations and achieving adequate drug delivery across the BBB. Low intensity focused ultrasound (FUS) has emerged as an image-guided technology capable of inducing a transient, reversible opening of the BBB, thereby enhancing drug delivery and immune effectors to access the tumor and TME. Preclinical studies have shown that FUS increases the delivery of otherwise non-brain-penetrant agents and immune cells, such as doxorubicin [75] and natural killer immune cells [76], as well as enhancing delivery of brain-penetrant drugs [77,78]. Initial human evidence for enhanced brain penetration using this approach comes from studies employing MRI-guided FUS (MRgFUS), a noninvasive modality that improved delivery of radiolabeled trastuzumab, a large molecule with limited intracranial activity, in patients with HER2-positive breast cancer brain metastases [79]. The BT008NA trial enrolled 34 patients with newly diagnosed glioblastoma (53% MGMT-methylated) who had completed concurrent chemoradiation with temozolomide and then received FUS during each temozolomide cycle for up to six cycles. This approach was feasible and generally safe, with only one grade 3 adverse event attributed to FUS. Median PFS was 13.5 months, and median OS was 31.3 months [80]. Several clinical trials using a FUS non-invasive approach have been completed and results are much awaited (NCT04446416, NCT01473485, NCT03712293). Based on the same biotechnological principles, a phase 1/2 study evaluated an implantable sonication device placed after standard tumor resection in 34 patients with recurrent glioblastoma. Patients received carboplatin at a standard AUC of 4–6 mg/ml∗min with sonication during cycles 1–3, resulting in MRI-evident BBB opening in 90% of cases; no grade 4 events were observed [81].

The potential benefits of FUS may extend beyond drug delivery to modulation of the TME. Glioblastomas arise within a profoundly immunosuppressive TME, in which glioma-associated microglia and macrophages (GAMs) constitute the dominant immune cell population. These cells exist along a spectrum of functional states, including a pro-inflammatory, tumor-suppressive (M1) state triggered by DAMPs and PAMPs (the mechanisms by which the previously mentioned oncolytic viruses effect their action); and an anti-inflammatory and immunosuppressive (M2) state [11]. In a preclinical study using mice with injected glioma cells, MRgFUS increased the abundance of genetically color-coded microglia and macrophages within tumors, and FUS-treated tumors exhibited a higher proportion of pro-inflammatory cells [82].

Convection-enhanced delivery (CED)

By delivering drugs at a slow, controlled rate directly into the brain through an implanted catheter, CED generates positive hydrostatic pressure that distributes the agent through the interstitial space. This approach enables higher local drug concentrations while minimizing systemic toxicity [83]. A small phase 1b study successfully delivered topotecan via CED in five patients with recurrent GBM, with no grade ≥3 adverse events and no significant systemic toxicities. Notably, analysis of pre- and post-treatment biopsies demonstrated a reduction in the Ki-67 proliferation index and a decrease in SOX2 labeling, along with an increase in pro-inflammatory cytokines [83]. Delta24-RGD, the previously mentioned DNX-2401 oncolytic adenovirus, is engineered to replicate selectively in tumor cells with disruptions in the Rb pathway and was tested in 19 patients with recurrent GBM via CED. Two patients developed increased intracranial pressure, presumed to be related to treatment-induced intracranial inflammation, although only one grade 4 treatment-related adverse event (confusion) was reported. Median OS was 129 days, and notably, the single patient who achieved a complete response remained alive for more than seven years; this patient had an IDH-mutant tumor. Furthermore, analyses of tumor tissue and cerebrospinal fluid demonstrated evidence of prolonged immune activation [84].

Targeting the Tumor Microenvironment

As noted above, a hallmark of the glioblastoma tumor microenvironment is the predominance of glioma-associated macrophages that adopt an immunosuppressive (M2) phenotype. Three distinct, yet mechanistically convergent, therapeutic strategies targeting the myeloid compartment of the glioblastoma TME have advanced into clinical development.

Activation of CSF1R by its ligands, CSF-1 and IL-34, promotes macrophage polarization toward an M2 phenotype which has been associated to a poorer prognosis in patients with high-grade gliomas. M2 differentiation suppresses T-cell activity, contributing to an immunosuppressive microenvironment and supporting angiogenesis [85]. In preclinical glioblastomas models, CSF1R inhibition has been shown to prevent M2 polarization and improve cell survival [86]. Pexidartinib (formerly PLX3397) is a small molecular inhibitor of CSF1R that has gained FDA approval for the treatment of tenosynovial giant cell tumors. However, a phase II trial evaluating pexidartinib monotherapy in recurrent glioblastoma failed to demonstrate a survival benefit [87], with a 6-month progression-free survival rate of only 8.6%. Similarly, a study combining pexidartinib with chemoradiotherapy in newly diagnosed glioblastoma did not improve survival outcomes [88].

BLZ945, an experimental CSF1R inhibitor, has demonstrated promising preclinical activity in glioma models by preventing M2 macrophage polarization and improving survival outcomes [89]. Nevertheless, no active clinical trials are currently evaluating BLZ945 in patients with glioma.

CD40 is a member of the TNFR superfamily that is expressed on antigen-presenting cells. Activation of the CD40/CD40L pathway enhances antigen presentation and promotes T-cell priming, leading to the generation of effective adaptive immune responses [90]. In addition to its roles in T-cell priming and B-cell activation, CD40 signaling promotes macrophage polarization toward a pro-inflammatory M1 phenotype, thereby enhancing antitumor immune responses [91]. Early-phase clinical studies evaluating CD40 agonists have demonstrated encouraging clinical activity in advanced solid tumors [92], particularly pancreatic adenocarcinoma, which is known for its profoundly immunosuppressive tumor microenvironment [93]. The PBTC-051 trial evaluated sotigalimab, a CD40 agonist monoclonal antibody, in pediatric and young adult patients with recurrent gliomas or newly diagnosed diffuse intrinsic pontine glioma (DIPG). A total of 31 patients were enrolled. Although the treatment was well tolerated, no objective responses were observed [94]. An ongoing clinical trial is evaluating 2141-V11, a CD40-targeting monoclonal antibody, in combination with D2C7-IT, an EGFR-directed immunotoxin, in patients with recurrent WHO grade III or IV gliomas (NCT04547777).

CCAAT/enhancer-binding protein beta (C/EBPβ) is a basic leucine zipper transcription factor that plays a central role in myeloid cell differentiation, inflammatory signaling, and immune regulation. In the context of the tumor microenvironment TME, C/EBPβ has emerged as a master regulator of mesenchymal transformation in gliomas [95]. It drives the differentiation and functional programming of myeloid-derived suppressor cells (MDSCs) and TAMs toward immunosuppressive phenotypes that promote tumor immune evasion, angiogenesis, and tissue remodeling [96].

Lucicebtide is a first-in-class C/EBPβ antagonist that binds to the leucine zipper domain of C/EBPβ, preventing its dimerization and promoting its degradation. This mechanism has demonstrated antitumor activity in vitro across multiple tumor cell lines, including glioblastoma models [97]. Preliminary results from a window-of-opportunity trial in patients with newly diagnosed glioblastoma demonstrated that lucicebtide penetrates the blood–brain barrier and is associated with a reduction in the mesenchymal transcriptional signature of tumor cells. Notably, most patients who received lucicebtide in combination with standard-of-care chemoradiation achieved prolonged progression-free survival, exceeding 21 months in several cases [98]. Final results from the study are awaited.

Blocking signaling interactions between tumor cells and the surrounding neurons and astrocytes has emerged as an alternative and complementary strategy to modulate the tumor microenvironment. Glutamate is the principal excitatory neurotransmitter in the central nervous system and has been shown to play a key role in glutamatergic synaptic signaling that promotes glioma growth, invasion, and network integration [99]. GLUGLIO is an ongoing repurposing clinical trial evaluating the efficacy of adding the glutamate modulating agents gabapentin, sulfasalazine, and memantine to standard-of-care chemoradiotherapy in patients with newly diagnosed glioblastoma (NCT05664464). Similarly, two trials are investigating the antiepileptic agent perampanel, a selective AMPA-type glutamate receptor antagonist. The PerSurge (NOA-30) trial is ongoing in Germany and is evaluating the perioperative activity of perampanel in recurrent glioblastoma (EU Clinical Trials Register identifier: 2023-503938-52-00), whereas the SENIPERA trial is assessing perampanel in combination with senicapoc, an inhibitor of calcium-activated potassium channels, in patients with newly diagnosed glioblastoma (NCT07284069).

Challenges

Despite the growing number of investigational therapies, several obstacles continue to limit progress in glioblastoma. Tumor heterogeneity and cellular plasticity remain major challenges, as glioblastoma contains multiple coexisting subpopulations that respond differently to therapy and can rapidly adopt resistant phenotypes [14]. This complexity makes it difficult to identify consistently actionable targets and explains why most targeted agents have shown limited or short-lived benefit. In addition, although molecular profiling can identify small patient subsets who might benefit from specific treatments, many clinical trials continue to enroll unselected populations. More rigorous biomarker-driven trial design will be essential to accurately evaluate targeted therapies.

Another persistent barrier is effective delivery of therapeutic agents to the central nervous system. The BBB restricts penetration of many promising treatments, including antibodies, cell-based therapies, and large-molecule inhibitors. While emerging technologies such as focused ultrasound, implantable sonication devices, and convection-enhanced delivery show early signs of improving drug access to the tumor, their clinical value still needs to be established. Additional challenges include the highly immunosuppressive tumor microenvironment, which limits the activity of immune-based treatments, and the lack of preclinical models that fully replicate the complexity of human glioblastoma. Addressing these issues will be essential for translating ongoing scientific advances into meaningful therapeutic improvements.

Conclusion

Glioblastoma remains an extremely difficult disease to treat, driven by marked molecular heterogeneity, rapid adaptation, and resistance to both standard and investigational therapies. Although recent advances in molecular classification have greatly improved diagnostic precision, these developments have not yet translated into major improvements in survival. Standard therapy with radiotherapy and temozolomide has remained essentially unchanged for nearly twenty years, and treatment options for recurrent disease continue to offer limited benefit.

Despite this, several areas of research are beginning to show progress. Molecular profiling has identified small but clinically relevant subgroups of patients with targetable alterations such as BRAF V600E mutations. Early experiences with next-generation targeted inhibitors suggest that selected patients may benefit from more precise treatment strategies, but larger studies and better understanding of resistance mechanisms are still needed. Evaluating these targeted treatments to the upfront setting, where potential better outcomes are expected is encouraged.

Immunotherapy approaches, including CAR-T cells, oncolytic viruses, and retroviral replicating vectors, have demonstrated biological activity and acceptable safety profiles, although consistent and durable clinical responses remain uncommon. These studies underline the challenges of activating an effective immune response within the highly immunosuppressive glioblastoma microenvironment.

Metabolic targeting has produced encouraging preclinical observations, particularly in defined subgroups that depend on oxidative phosphorylation. However, clinical trials in unselected populations have not been successful, reinforcing the need for better patient selection.

New technologies designed to improve drug delivery across the blood–brain barrier, such as focused ultrasound, implantable sonication devices, and convection-enhanced delivery, have shown feasibility and early signs of improved drug distribution. These methods may expand the range of agents that can be tested in the CNS, including large molecules and cell-based therapies.

Overall, current research suggests that future progress will require combining accurate molecular classification, rational patient selection, and improved methods for CNS drug delivery. Although significant barriers remain, ongoing clinical and translational efforts offer a more practical foundation for developing therapies that may provide meaningful benefit for patients with glioblastoma..

Author contributions

All authors contributed to the drafting, writing, and critical revision of the manuscript. Juan B Blaquier was responsible for the design and preparation of the figure. All authors approved the final version of the manuscript.

Funding

Juan B. Blaquier and Macarena de la Fuente were supported in part by The Sylvester Comprehensive Cancer Center Support Grant 5P30CA240139-04.

Declaration of Interest

JBB: None to declare.

MJLF: has received honoraria for consultation from Servier, Novocure and Ipsen.

MDLF: reports a relationship with Servier Laboratories Suresnes that includes: board membership; Medscape Education that includes: consulting or advisory and speaking and lecture fees; Fore Therapeutics that includes: consulting or advisory; Nuvation Bio, Inc that includes: consulting or advisory; with Caris Life Sciences Inc that includes: non-financial support; Rigel Pharmaceuticals Inc that includes: consulting or advisory.