Glioblastoma in the post-2021 era: Diagnostic recalibration, therapeutic innovations, and future horizons

Abstract

Glioblastoma (GBM) remains a devastating diagnosis, and since 2021, the field has witnessed both incremental advances and renewed paradigms in diagnosis, therapeutic development, and trial design. The 2021 WHO Central Nervous System 5 classification redefined GBM as isocitrate dehydrogenase (IDH)-wildtype, sharpening the molecular underpinnings of clinical trial cohorts. Targeted approaches in selected molecular subsets (e.g., IDH inhibitors) have matured, while tumor-treating fields remains a debated but utilised adjunct. Immunotherapy via checkpoint inhibitors has been disappointing in unselected populations, but newer vaccines, oncolytic viral, and combination strategies are under active development. The most hopeful advances in the 2023-2025 window come from engineered cell therapies, especially chimeric antigen receptor T-cell delivered locoregionally or with multivalent antigen specificity. Advances in liquid biopsy, spatial transcriptomics, imaging biomarkers, and adaptive trial platforms are converging to accelerate translation. This narrative review synthesises developments since 2021, integrating translational insight and practical guidance, and outlines a roadmap for the next decade of GBM research.

Key Words: Glioblastoma; Chimeric antigen receptor T-cell therapy; Tumor treating fields; Immunotherapy; Radiotherapy, molecular diagnostic techniques

Core Tip: The management of glioblastoma (GBM) has entered a post-2021 era characterized more by convergent advances in molecular classification, trial design, and immunoengineering than by a single therapeutic breakthrough. The 2021 WHO Central Nervous System 5 redefinition of GBM as an isocitrate dehydrogenase-wildtype entity has altered clinical trial interpretation, diagnosis, and prognosis. Molecularly selected targeted therapies, improved tumor treating fields, and a deliberate shift away from ineffective single-agent immunotherapy toward combinatorial and neoadjuvant approaches are driving incremental progress, even though conventional chemoradiation continues to be the cornerstone of care. Locoregionally administered, multivalent engineered cell therapies, along with developments in liquid biopsy, spatial profiling, and adaptive platform trials, provide the most convincing indications of future impact. When taken as a whole, these advancements highlight a move in GBM treatment paradigms toward precision-guided, biologically informed, and adaptive approaches.

INTRODUCTION

Glioblastoma (GBM) continues to pose one of the greatest therapeutic challenges in oncology. Current treatment option include maximal safe resection followed by standard radiotherapy (RT) (or hypo fractionated RT for elderly or poor performance status) with concurrent and adjuvant temozolomide (TMZ) followed by alternating electrical field therapy as per availability and patient compliance[1]. Despite incremental improvements in supportive care and imaging, median survival from diagnosis remains under two years[1,2]. The past half-decade has brought changes not via a single “magic bullet” but by a mosaic of advances: Molecular redefinition, targeted therapies, immunoengineering, and more sophisticated trial designs. This narrative review is intended for oncologists, neuro-oncologists, and translational researchers, summarizing key developments post-2021 contextualizing them in a historically difficult disease space. We review the diagnostic changes, immunotherapy developments, translational advances [oncolytic viruses, chimeric antigen receptor T-cell (CAR-T)], liquid biopsy and trial/biomarker themes.

LITERATURE REVIEW

A PubMed search was conducted for English-language articles of randomised clinical trials, meta-analyses, systematic reviews, and observational studies of GBM published from January 1, 2021, to September 30, 2025. Central Nervous System (CNS) 5 was published in 2021, which clearly defined GBM not just by histopathological features but added molecular definitions, that’s why the cutoff for the search was set at 2021. We reviewed various guidelines databases and prioritised recent randomized clinical trials (RCTs) based on rigor of the study design, sample size, and length of follow-up. Articles published before 2021 that define standard practices were also included. Of the 4327 articles reviewed, we included 99, comprising of 37 trials (15 phase 3 RCTs or trials, 15 phase 2 RCTs or trials, 4 phase I trials, 2 basket trials, 1 platform trial), 33 reviews, 3 systematic reviews and meta-analysis, 14 translational research, 9 observational studies, and 3 guideline recommendations. These various landmarks post 2021 in the management of GBM have been summarized in Table 1. This table is illustrative rather than exhaustive. Many ongoing trials and publications will further enrich the field.

Table 1 Selected post-2021 clinical and translational landmarks.

Year	Milestone/trial	Context/population	Key result/insight	Ref.
2021	Official WHO CNS5 adoption	All gliomas	Molecular reclassification; IDH-wt glioblastoma defined	[3,4]
2021-2025	TTFields reanalysis and adoption debates	Newly diagnosed GBM	Reaffirmed modest benefit, raised compliance/cost issues	[25-28]
2021	Proton vs photon radiotherapy (randomized phase II)	Newly diagnosed glioblastoma	Proton RT reduced grade ≥ 3 radiation-induced lymphopenia (14% vs 39%) with comparable tumor control; no OS/PFS superiority (NCT01854554)	[94]
2024	Multicenter CAR-T trial launches	Recurrent GBM	Locoregional dosing in selected patients	[68]
2024	First public bivalent CAR-T interim reports	Refractory GBM	Intrathecal dual-antigen responses	[70]
2022-2024	G47Δ (teserpaturev) oncolytic virus approval (Japan)	Residual/recurrent GBM	Safety and select durable responses	[61,62]
2021-2025	Liquid biopsy advances	Glioma/GBM	Better ctDNA/EV detection in CSF	[77-79]

CNS5: Central nervous system 5; IDH: Isocitrate dehydrogenase; IDH-wt: Isocitrate dehydrogenase wild type; GBM: Glioblastoma; TTFields: Tumor treating fileds; CAR-T cell: Chimeric antigen receptor T cell; ctDNA: Circulating tumor DNA; EV: Extracellular vesicles; CSF: Cerebrospinal fluid; OS: Overall survival; PFS: Progression free survival.

MOLECULAR REDEFINITION AND DIAGNOSTIC EVOLUTION

WHO CNS5 and its impact

The 2021 fifth edition of WHO Classification of CNS5 shifted the diagnostic paradigm by elevating molecular markers [e.g., isocitrate dehydrogenase (IDH) status, 1p/19q codeletion, ATRX, TERT promoter, H3 K27 alterations] to central roles in classification[3]. Under CNS5, “GBM” is defined as a diffuse astrocytic tumor that is IDH-wildtype and demonstrates either classic histologic features (necrosis, microvascular proliferation) or one of the specific molecular features (TERT promoter mutation, EGFR amplification, or combined +7/-10 chromosomal changes). While grade 4 astrocytoma with these features and is IDH-mutant has been reclassified as “astrocytoma, IDH-mutant, CNS WHO grade 4”. These changes effectively exclude many previously classed “secondary GBMs” (i.e., IDH-mutants) from the classical GBM category, refining the eligible populations for most GBM trials[3,4]. This reclassified GBM has poor overall survival (OS), distinguishing it sharply from IDH-mutant astrocytoma which has better outcomes[5]. Recent studies using magnetic resonance imaging (MRI) radiomics models aligned with CNS5 have been able to noninvasively predict prognosis and biological features like DNA repair pathways and immune infiltration relevant to GBM aggressiveness[6].

The diagnostic redefinition has downstream consequences: Trials must now stratify by molecular subtypes, historical controls need re-evaluation, and clinicians must ensure comprehensive molecular panels are available for all glioma diagnoses. Operationally, this means that MGMT promoter methylation, TERT, ATRX, IDH1/2, and copy number profiling become standard, not optional.

Implications for trial design and interpretation

The new classification impacts clinical trial design, patient stratification, and development of targeted therapies aligned with molecular-defined subtypes of GBM[7]. With molecularly homogeneous cohorts, trial power may improve and results become more interpretable. However, eliminating IDH-mutant gliomas reduces patient numbers in classic GBM cohorts, posing enrollment challenges. Moreover, historical data that included molecularly mixed cohorts must be reinterpreted cautiously. Many progressions and responses in the pre-CNS5 era may not generalize to strictly defined GBM (IDH-wildtype). This shift mandates that ongoing and future trials clearly disaggregate results by molecular subtype or conform to CNS5 definitions.

TARGETED THERAPIES: MATURATION IN MOLECULAR SUBSETS

IDH inhibitors in gliomas

As per CNS5 IDH-mutant GBM are essentially a grade 4 astrocytoma, and such broader class of IDH-mutant diffuse gliomas (grade 2 to 4) has become targetable with IDH inhibitor. In particular, vorasidenib and ivosidenib have emerged as promising agents[8-10]. These results have triggered regulatory interest and expanded access in certain jurisdictions.

While these therapies are important but largely apply to IDH-mutant, often lower-grade gliomas (or IDH-mutant advanced gliomas), not the canonical IDH-wildtype GBM defined by CNS5. They exemplify how precise molecular targeting can shift the management paradigm in diffuse gliomas but at present not applicable to IDH-wildtype GBM. Moreover, they bolster the concept that precision glioma therapy is feasible, encouraging investigators to seek analogous drivers in IDH-wildtype GBM.

Other molecularly targeted ventures

Beyond IDH, occasional GBMs harbor actionable alterations: BRAF V600E, NTRK fusions, FGFR alterations, or EGFR mutations beyond amplification. Systematic screening for actionable molecular alterations resulted in lower rates (< 10%) with European Society for Medical Oncology Scale for Clinical Actionability of Molecular Targets tier I being 3.4% to 4.1%, and tier II up to 7%[11-13]. For some of these alterations, few case reports and phase 2 studies have documented durable responses, for example BRAF and MEK inhibitors (Dabrafenib plus trametinib) in BRAF^V600E-mutant high-grade gliomas[14-16]. The phase 2 basket trials (VE-BASKET and ROAR trial), with limited number of BRAF^V600E mutant recurrent GBM patients showed an objective response rates of 25%-33% with these inhibitors[15,16]. NTRK-fusion GBMs have responded to agents like entrectinib or larotrectinib[17,18]. Treatment of children with these alterations have shown better outcomes compared to adults. Nonetheless, the general applicability of targeted therapy in GBM remains limited by intratumoral heterogeneity, subclonal driver architecture, and blood-brain barrier (BBB) penetrance[19].

A particularly persistent challenge is that many targetable alterations in GBM are subclonal or spatially restricted, leading to early escape under selective pressure[20]. As such, combination targeted approaches, adaptive targeting (switching therapies midcourse based on liquid biopsy), or targeting downstream common pathways (e.g., PI3K/AKT/mTOR, CDK, DNA damage repair) are being explored[21-24].

TUMOR-TREATING FIELDS: CONTINUED DEBATE AND REFINEMENT

Evolving evidence and adoption

Tumor-treating fields (TTFields) delivered via scalp arrays in GBM as a non-invasive modality consisting of low intensity (1-3 V/cm), intermediate-frequency (100-300 kHz), alternating electric fields remains an approved adjunct in many regions. The foundational randomized trial, EF-14 (n = 695), predates 2020, reported increased progression-free survival (PFS) (6.7 months vs 4 months) and OS (20.9 months vs 16.0 months) advantage with TTField plus TMZ vs TMZ alone, respectively. with benefit most marked in patients with high device compliance[1]. But debates concerning its real-world uptake, cost-effectiveness, compliance, quality-of-life trade-offs, and mechanistic basis have continued vigorously[25-28]. Some meta-analyses and tertiary center reviews have reaffirmed an OS and PFS benefit when added to TMZ maintenance, in de-novo and recurrent cases, particularly in patients with high compliance (average usage > 75%)[27,28]. Yet critics point to selection bias, lack of blinding, and real-world tolerability challenges.

Integration with newer modalities

A recent conceptual advance is combining TTFields with immunotherapy or other biological therapies to potentiate immune responses by inducing tumor stress or immunogenic cell death[29]. Preclinical data suggest that TTFields may upregulate neoantigen presentation, DNA damage responses, or immunomodulatory signals, which could synergize with checkpoint inhibitors or vaccines. A phase 2 study by Chen et al[30] assessed efficacy and safety of TTFields plus pembrolizumab and TMZ in newly diagnosed GBM. The study showed improved PFS (12.0 months vs 5.8months) and OS (24.8 months vs 14.6 months) with this triple regimen compared to TMZ alone.

Overall, TTFields retains a role in selected patients who can adhere to usage requirements (> 75% average usage approximately 18 hours/day) and desirous of any incremental benefit, though it is no longer regarded as a transformational modality[31].

IMMUNOTHERAPY: SOBERING LESSONS AND EVOLVING STRATEGIES

Checkpoint inhibitors in GBM

One of the most disappointing outcomes in GBM research has been the consistent failure of immune checkpoint inhibitors (ICIs) to deliver survival benefit in unselected GBM populations in both recurrent and adjuvant setting (Table 2). A phase II trial (n = 80) in recurrent GBM, a combination of pembrolizumab and bevacizumab vs pembrolizumab alone, showed modest improvements in PFS6 (26% vs 6.7%) but failed to shift OS meaningfully (8.8 months vs 10.3 months)[32]. Worsened OS was correlated with baseline dexamethasone use and MGMT status. Iwamoto et al[33] conducted a phase 2 study (n = 60) of pembrolizumab with re-irradiation alone (in bevacizumab-naïve) or bevacizumab continuation in recurrent GBM. Neither primary endpoint of ORR or OS at 12 months was reached. For each cohort, median PFS (4.0 months each), median OS (mOS) (11.0 months vs 7.0 months), PFS6 (13.3% vs 10.6%) rate of survival at 12 months (OS12) was (40% vs 16.6%) did not differ much (approximately 90% in each cohort were IDH-wild type). For selected PD-L1 positive recurrent GBM patients multicohort phase 1b KEYNOTE-028 study demonstrated durable antitumor activity with manageable toxicity, while nivolumab monotherapy did not improve OS compared to single agent bevacizumab in recurrent GBM in Check Mate 143[34,35]. Further studies in such selected population of recurrent GBM patients is warranted.

Table 2 Summary of “randomized disappointments” with immunotherapy in unselected glioblastoma.

Ref.	n	Trial number	Phase	Patient population	Design	Results
Reardon et al[35]	369	NCT02017717 Checkmate 143	Phase 3 RCT	Recurrent GBM	Nivolumab vs bevacizumab	mPFS 1.5 months vs 3.5 months, mOS 9.8 months vs 10 months, ORR 7.8% vs 23.1%
Nayak et al[32]	80	NCT02337491	Phase 2 RCT	Recurrent GBM	Pembrolizumab + bevacizumab vs pembrolizumab	PFS6 rate 26% vs 6.7%; mOS 8.8 months vs 10.3 months; ORR 20% vs 0%
Iwamoto et al[33]	60	NCT03661723	Phase 2 RCT	Recurrent GBM	Pembrolizumab + ReRT (bevacizumab naïve) vs bevacizumab continuation	mPFS 4 months each; mOS 11 months vs 7 months
Lim et al[36]	716	NCT02667587 CheckMate 548	Phase 3 RCT	Newly diagnosed GBM, MGMT methylated	Nivolumab + RT + TMZ vs placebo + RT + TMZ	mPFS 10.6 months vs 10.3 months; mOS 28.9 months vs 32.1 months
Omuro et al[37]	560	NCT02617589 CheckMate 498	Phase 3 RCT	Newly diagnosed GBM, MGMT unmethylated	Nivolumab + RT vs TMZ + RT	mPFS 6.0 months vs 6.2 months; mOS 13.4 months vs 14.9 months
Lassman et al[38]	159	NCT04396860 NRG Oncology BN007	Phase 2 RCT	Newly diagnosed GBM, MGMT unmethylated	Nivolumab + ipilimumab + RT vs TMZ + RT	mPFS 7.7 months vs 8.5 months; OS immature (approximately 13 months each)
Brown et al[39]	119	ISRCTN84434175 Ipi-Glio trial	Phase 2 RCT	Newly diagnosed GBM	Ipilimumab + TMZ vs TMZ	mPFS 10.8 months vs 12.5 months; mOS months 18.0 vs 23.0 months

RCT: Randomized controlled trial; mOS: Median overall survival; mPFS: Median progression free survival; GBM: Glioblastoma; PFS6: 6 months progression free survival; ORR: Objective response rate; RT: Radiotherapy; TMZ: Temozolomide; Ipi: Ipilimumab.

In newly diagnosed GBM with either unmethylated or methylated MGMT promotors, no survival advantage was seen with addition of immunotherapy in various randomized controlled trials (Table 2)[32,33,35-39]. Similarly, systematic reviews in 2024 note that PD-1/PD-L1 inhibitors, whether as monotherapy or combined with radiation or chemotherapy, have largely yielded negative or marginal results[40-42].

Why the failure? GBM is immunologically “cold”, with low tumor mutational burden, T cell exclusion, dominance of immunosuppressive myeloid populations, and adaptive immune suppression[43-45]. Many tumors lack PD-L1 expression or have spatial heterogeneity in immune infiltration, making biomarker-driven selection essential but still nascent[46,47].

Biologic rationale for neoadjuvant/combinatorial approaches: Neoadjuvant and combinatorial immunotherapy strategies for GBM are biologically attractive because administering immune checkpoint blockade or other immune modifiers while the tumor remains in situ can prime a more robust, tumor-antigen-directed T-cell response, increase tumor infiltration by activated lymphocytes, and induce interferon-γ–driven transcriptional programs that remodel the otherwise profoundly immunosuppressive glioma microenvironment[48]. Early clinical and translational studies, most notably neoadjuvant anti-PD-1 work, have shown on-treatment increases in T-cell and IFN-γ-related gene signatures and reductions in proliferation signatures in resected specimens, suggesting a pharmacodynamic impact that may translate into clinical benefit when combined with surgery, RT, or strategies that target myeloid-cell mediated suppression[48,49]. Rational immunotherapy combinations could induce strong immune response, and can be aimed to overcome barriers such as low TMB, T-cell exclusion, and dominant myeloid/Treg immunosuppression that limit single-agent efficacy in GBM[50].

Vaccines, neoantigen approaches, and oncolytic viruses

Given the limitations of ICIs, vaccination and viral immunotherapy approaches have gained renewed attention. Cancer vaccines use tumor antigens to activate the adaptive immune system of GBM patients, which can be delivered in the form of peptides, dendritic cell (DCs), DNA/RNA, and viral vectors. Neoantigen vaccines (autologous peptide or RNA-based) are now entering mid-phase trials; early-phase readouts demonstrate immunogenicity and durable responses in small subsets, though survival translation remains elusive. The common peptide vaccines used in GBM include rindopepimut (CDX-110, targeting EGFRvIII mutation), surviving vaccine (SurVaxM, targeting survivin), multipeptide vaccine (IMA950), HSPPC-96-specific vaccine, and personalised neoantigen vaccine. The lessons from earlier EGFRvIII vaccines (e.g., rindopepimut), which failed in randomized trials, caution against overoptimism but provide mechanistic insights in immune escape[51,52]. SurVaxM, IMA950, HSPPC-96, personalised neoantigen vaccines have showing mixed results in phase I and II trials and further validation of their efficacy is warranted[53-57]. The various DC-based vaccines used in management of GBM includes DCVax-L, ICT-107, and CMV-DCs. Most promising results were seen with tumor lysate-loaded DC vaccines, DCVax-L which has shown modest survival gains (mOS of 19.3 months for newly diagnosed GBM and 13.2 months in recurrent GBM) in phase III prospective externally controlled settings[58].

Oncolytic viruses, such as engineered HSV derivatives, nonpathogenic poliovirus/rhinovirus chimeric virus (PVSRIPO) and adenoviruses (DNX-2401), provide both oncolysis, immune activation and are under active research. For example, in Japan, the G47Δ (Teserpaturev) 3^rd generation oncolytic HSV virus achieved conditional time-limited regulatory approval for residual or recurrent GBM, an important milestone[59-61]. Phase II data from intratumoral G47Δ suggest safety and 1-year survival rate of 84.2%, underscoring the viability of viral platforms in CNS tumors[61]. DNA (as plasmids) and RNA (as mRNA) vaccines are currently being investigated at preclinical and phase I/II trial level for newly diagnosed GBM patients[62,63].

Combination and microenvironment modulation

Moving beyond monotherapy, current strategies aim at synergistic combinations: Vaccines or oncolytic viruses plus ICIs, epigenetic modulators, radiation, or modulators of myeloid cells. Preclinical and early clinical studies suggest that priming the tumor microenvironment (e.g., with cerebrospinal fluid (CSF)-1R inhibitors, PI3Kγ inhibition, CCL5 or TLR agonists) may sensitize tumors to immunotherapy[64-66]. However, optimal scheduling, toxicity management, and patient selection remain unsettled.

CAR AND ENGINEERED CELL THERAPIES: THE HOPEFUL FRONTIER

Adoptive T-cell therapies are patient’s T-cells isolated and expanded ex vivo and then reinfused into patient targeting tumor like a personalized immunotherapy. The adoptive T-cell therapies include CAR-T-cell therapy, CMV-specific T-cell therapy, tumor-infiltrating lymphocyte therapy in GBM.

Locoregional delivery, bypassing systemic barriers

Unlike hematologic malignancies, GBM poses unique barriers to CAR-T therapy such as the BBB, antigen heterogeneity, and neurotoxicity risk[67]. A pivotal innovation has been the use of locoregional delivery, intratumoral, intracavitary, intraventricular, or intrathecal administration, which circumvents systemic trafficking limitations and concentrates effector cells within the tumor milieu. Also, by modifying T-cells to express receptors that selectively target tumor associated antigens such as EGFRvIII, interleukin-13 receptor alpha 2 (IL-13Ra2), and human epidermal growth factor receptor (EGFR) 2, CAR-T therapy seeks to overcome these obstacles. Brown et al[68] in their phase I trial, evaluated three routes of locoregional delivery [viz. intratumoral (ICT), intraventricular (ICV), and dual (ICT/ICV)) of IL-13Rα2-targeted CAR-T cell in 65 patients with recurrent GBM. They showed such locoregional delivery is feasible, well tolerated with no dose limiting toxicities (one grade 3 encephalopathy and one grade 3 ataxia) with 50% having stable or better response.

Multivalent and armored CAR constructs

Given antigen escape is a major failure mode, contemporary CAR designs incorporate multivalent antigen targeting (e.g., dual or tri-antigen constructs), switchable binding domains, or “armored” features (e.g., co-expression of cytokines, checkpoint blockade molecules, or immune stimulants)[69].

Bagley et al[70] reported interim data on six patients treated intrathecally with bivalent CAR-T cell targeting EGFR and IL13Rα2, which was associated with early-onset neurotoxicity, most consistent with immune effector cell-associated neurotoxicity syndrome (ICANS). While the early MRI timepoints showed reduction in enhancement and tumor size in all six patients supporting primary safety and bioactivity of the dual blockade.

These enhancements aim to mitigate antigen-negative escape, improve T cell persistence, and modulate the immunosuppressive microenvironment[69]. Preclinical modelling using glioma organoids, PDX models, and mathematical simulations informs optimal construct design and dosing regimens[71,72].

Challenges and safety

Key hurdles include potential neurotoxicity (e.g., cytokine release syndrome, ICANS, cerebral edema, neuroinflammation, tumor lysis syndrome), off-target effects, T cell exhaustion, and variable infiltration[73]. Manufacturing complexity, cost, and delivery logistics remain significant barriers to broad adoption[74,75]. Some investigators are exploring repeated dosing or maintenance infusions, but sustaining efficacy while controlling toxicity is an active area of research.

Early outcomes and proof-of-concept

Although still early, some treated patients have achieved meaningful survival beyond expected benchmarks (mOS: 10.2-11.8 months)[68,76]. However, these are small numbers, heavily selected recurrent GBM, previously treated with other regimens, and with short follow-up[68]. The key determinant will be whether spectacular individual responses can be generalized and made reproducible in larger cohorts.

BIOMARKERS, LIQUID BIOPSY, IMAGING, AND SPATIAL PROFILING

Liquid biopsies: Circulating tumor DNA, CSF, extracellular vesicles

Liquid biopsy has emerged as one of the most promising translational developments in neuro-oncology because it offers the possibility of non-invasive molecular monitoring of GBM evolution over time. In systemic cancers, circulating tumor DNA (ctDNA) analysis has already transformed biomarker discovery and treatment monitoring. However, translation to GBM has been slower due to the unique biological barriers of CNS tumors, particularly the BBB and low levels of tumor DNA shedding into peripheral circulation. Studies in glioma and GBM have shown that ctDNA in plasma is often low, but CSF-derived DNA or extracellular vesicle (EV) cargo can more reliably reflect tumor mutational status and clonal evolution[77-79]. These assays enable noninvasive (plasma)/invasive (CSF) molecular monitoring over time, detection of emerging resistance clones, and dynamic biomarker–guided adjustments (e.g., switching targeted therapy). However, sensitivity remains limited, particularly for small-volume disease or compartmentalized relapse. Another limitation is short half-lives of ctDNA (2.5 hours), EV (30 minutes), circulating tumor cells (1 hour to 2.4 hours).

Despite strong research interest, liquid biopsy has not yet entered routine clinical practice for GBM. Current National Comprehensive Cancer Network and European Association of Neuro-Oncology guidelines do not recommend ctDNA or EV assays for diagnosis, treatment selection, or surveillance in GBM. Instead, molecular profiling continues to rely primarily on tumor tissue obtained through surgical resection or biopsy, with imaging remaining the principal modality for disease monitoring.

Consequently, liquid biopsy in GBM should currently be regarded as a promising research tool rather than a guideline-endorsed clinical test. Most applications remain confined to translational studies and clinical trials, where serial molecular monitoring can be incorporated into investigational therapeutic strategies. Large prospective studies will be required to establish standardized methodologies, validate clinical utility, and determine whether liquid biopsy-guided management can improve outcomes.

Advanced imaging and response assessment

Differentiating pseudoprogression, treatment effect, and true tumor recurrence is a persistent challenge, especially when immunotherapy or TTFields are in play. Innovations in MRI (advanced perfusion, diffusion, spectroscopy), hyperpolarized MRI, and PET imaging (e.g., amino acid tracers such as FET-PET, MET-PET) are now integrated into therapeutic trials to improve specificity[80,81]. Some adaptive trial designs incorporate imaging biomarkers in real time to modulate therapy allocation.

Spatial transcriptomics and tumor heterogeneity

Single-cell RNA sequencing and emerging spatial transcriptomics have unveiled the intricate intratumoral heterogeneity of GBM: Coexisting cellular states (e.g., neuronal-progenitor-like, astrocytic-like, mesenchymal-like), myeloid niches, vascular niches, and metabolic gradients[82-84]. These high-resolution maps inform antigen selection, cell therapy targeting, and combinatorial strategies. They also help explain why a single-target therapy may fail in a dynamically heterogeneous ecosystem.

CLINICAL TRIALS, ADAPTIVE DESIGNS, AND REGULATORY ENVIRONMENT

Platform and basket trials

Post-2021, the GBM trial ecosystem has evolved toward master protocols, phase 0, basket trials, and adaptive designs, allowing simultaneous testing of multiple experimental arms in molecularly stratified cohorts[15,85,86]. This approach parallels trends in other tumor types, accelerating evaluation timelines and reducing control arm burden. For rare molecular subsets (e.g., NTRK or BRAF in gliomas), such designs ensure efficient accrual. These trials can remove ineffective therapies earlier during drug development and improve trial efficacy. The INdividualized Screening trial of Innovative GBM Therapy (INSIGhT) a phase II platform trial, evaluated three experimental arms (abemaciclib, neratinib, CC-115) in MGMT unmethylated newly diagnosed GBM cases with chemoradiotherapy as control arm[87]. Though there was just PFS benefit (with abemaciclib and neratinib, not with CC-115), and no OS benefit, the trial showed such trial design may promote improved and more efficient therapeutic discovery in GBM. GBM Adaptive Global Innovative Learning Environment (GBM AGILE), a platform trial evaluates investigational therapies and associated biomarker signatures to support new drug approvals and registrations (NCT03970447). The trials adaptive nature and its role in evaluating multiple regimens simultaneously provides real infrastructure solutions to accrual and heterogeneity problems.

Endpoints beyond survival

Although OS remains the gold standard, new trials increasingly incorporate PFS, immune-related response criteria, patient-reported outcomes, neurocognitive metrics, and quality-of-life endpoints. Surrogate endpoints may expedite early decision-making, though their validity remains under scrutiny. Recent trends in efficacy endpoints for trials conducted in fiscal years (FY) 2020-2022 indicates PFS (22%), OS (20%) as most common primary endpoints. ORR as primary endpoint was significantly lower compared to FY 2017-2019 (8% vs 20%)[88].

Table 3 summarizes the evolving therapeutic strategies and their corresponding opportunities and challenges in GBM management. Emerging modalities such as cell-based immunotherapies, checkpoint inhibitors, and tumor-targeted vaccines highlight the diversification of approaches beyond conventional chemoradiation. However, each strategy faces unique translational barriers, ranging from tumor heterogeneity and immune evasion to logistical hurdles in delivery and trial design. Parallel advances in liquid biopsy and spatial or single-cell profiling offer real-time molecular insights that can guide adaptive and platform trial structures. Collectively, these integrative frameworks underscore a shift toward precision and dynamic treatment paradigms in GBM research.

Table 3 Strategies, opportunities, and reframing challenges.

Strategy/modality	Therapeutic rationale	Key challenges and barriers	Research directions
CAR-T/cell immunotherapy	Engineered T cells targeting tumor antigens	Antigen heterogeneity, neurotoxicity, manufacturing, infiltration	Multivalent/armored constructs, locoregional dosing, repeat infusions
Checkpoint inhibitors/immunotherapy	Release immune suppression	Cold tumor microenvironment, low TMB	Combination therapies, microenvironment modulation, biomarker selection
Vaccines and oncolytic viruses	Prime systemic/Locoregional immune responses	Delivery, immune suppression, limited by scale	Combinatorial adjuvants, intratumoral delivery, optimized scheduling
TTFields	Disrupt mitosis and sensitize cells	Compliance, cost, modest benefit	Synergy with immunotherapy, biomarker-guided use
Liquid biopsy	Dynamic molecular monitoring	Low sensitivity, compartmentalization	Improved assay sensitivity, integration into adaptive trials
Spatial profiling/single-cell	Map heterogeneity and niche interactions	Data complexity, translation to therapy	Use to guide trial designs and antigen selection
Adaptive and platform trials	Efficient evaluation across arms	Statistical complexity, regulatory harmonization	Master protocols, biomarker stratification, real-time adaptive arms

TMB: Tumor mutational burden; TTFields: Tumor treating fileds; CAR-T cell: Chimeric antigen receptor T cell.

TRANSLATIONAL INSIGHT AND BIOLOGICAL CONSTRAINTS

Tumor microenvironment and immunosuppression

GBM is dominated by immunosuppressive myeloid cells (microglia, macrophages), M2 phenotypes, TGF-β, adenosine pathways, and metabolic constraints (hypoxia, nutrient depletion) that suppress effector T cell function[43,44]. T cell exhaustion signatures are pervasive and severity varies by tumor region[44]. Efforts to reprogram myeloid populations (via CSF-1R inhibition, PI3Kγ targeting, CD47-SIRPα blockade) are underway, both preclinically and in early clinical trials[89,90]. The goal is to convert a “cold” microenvironment into an “inflamed” state receptive to immunotherapy.

Intratumoral heterogeneity and resistance

Single-cell and spatial transcriptomics post-2021 have mapped multiple coexisting cellular states (e.g., neuronal-like, astrocyte-like, mesenchymal-like) which may shift under therapy, contributing to adaptive resistance[82-84]. The usual linear clonal model is insufficient; plasticity allows phenotypic switching under therapy pressure. Thus, monotherapies targeting a single pathway or antigen are frequently undermined by clonal escape. Effective therapy must anticipate and counter heterogeneity and plasticity—through combinatorial, sequential, or adaptive strategies.

Modeling, simulation, and predictive tools

To manage complexity, mathematical and computational models are increasingly used. For example, recent preprints describe dynamic models combining CAR-T and chemotherapy protocols in glioma[71,72]. These models help predict optimal dosing, timing, and antigen targeting strategies. Further, PDX models, glioma organoids, and humanized mouse models provide translational platforms to test combinatorial regimens before clinical application[91,92].

INNOVATIONS IN GBM RT

RT remains a foundational component of GBM management, yet its role in improving long-term outcomes has remained largely static despite decades of investigation. Since 2021, clinical and translational work in GBM RT has focused on: (1) Altered fractionation and dose-escalation (including accelerated/hypofractionated regimens); (2) Particle therapy (protons) to reduce normal-tissue toxicity (including lymphopenia); (3) Stereotactic/hypofractionated re-irradiation for recurrences; (4) Integration of RT with immunotherapy and other radiosensitizers (including nanoparticle strategies and radio-immunotherapy); and (5) Personalization using imaging/biomarkers and adaptive RT. Importantly, while several approaches have demonstrated promising biological or surrogate endpoints, no phase III trial has yet shown a definitive OS advantage over standard fractionated photon RT in molecularly defined IDH-wildtype GBM[93-95].

Key themes and evidence

Proton therapy-less immune suppression, same (or comparable) tumour control: Particle therapy, particularly proton beam therapy, has gained attention in GBM primarily as a strategy to reduce treatment-related toxicity rather than to improve tumor control. A randomized phase II study found that proton RT reduced high-grade radiation-induced lymphopenia (14%) vs photons (39%) in GBM patients, an important biological endpoint given the immunosuppressive impact of RT in brain tumours and the interest in combining RT with immunotherapy[94]. Proton RT is being actively evaluated in trials comparing clinical outcomes and toxicity profiles vs photon IMRT. A phase II randomised trial compared proton vs photon, showed proton RT was not associated with delay in time to cognitive failure but reduced toxicities with no difference in PFS or OS[93].

The clinical relevance of lymphocyte preservation has become increasingly important in the era of immunotherapy, as radiation-induced immune depletion may blunt the efficacy of ICIs, vaccines, and adoptive cell therapies. Ongoing trials are therefore evaluating proton therapy as a platform for immunotherapy integration, rather than as a stand-alone intensification strategy. However, at present, proton therapy cannot be considered superior to photon RT in terms of survival, and its role remains selective, influenced by availability, cost, and patient-specific considerations.

Altered fractionation and accelerated/hypofractionated approaches: Multiple prospective and pooled analyses have tested higher-dose hypofractionated or accelerated regimens (e.g., 52.5 Gy in 15 fractions in elderly or frail patients; 3 weeks 40 Gy in 15 fractions vs 4 weeks 50 Gy in 20 fractions regimen; dose-escalated accelerated hypofractionation 60 Gy in 20 fractions)[95-97]. Results suggest comparable survival in selected patients, acceptable toxicity in many series, and potential practical benefits (shorter treatment time), though definitive phase III evidence for superiority remains limited and it is not a universal replacement for standard chemoradiation.

Re-irradiation and stereotactic hypofractionation for recurrence: Randomized and prospective phase II work has evaluated hypofractionated stereotactic RT (HSRT) in recurrent GBM comparing fractionation/dose schedules; these trials focus on balancing local control and toxicity in the re-treatment setting. Early data support HSRT schedules (e.g., 35 Gy/5 fx) as feasible with numerically higher toxicity for selected relapses[98].

RT combined with immunotherapy/radiosensitizers-biologic rationale, limited clinical success so far: A major conceptual advance in recent years has been the recognition that RT exerts profound immunomodulatory effects. Ionizing radiation can enhance antigen release, increase major histocompatibility complex expression, promote immunogenic cell death, and transiently disrupt the BBB, providing a strong biological rationale for combining RT with immunotherapy.

However, clinical translation has been challenging. Randomized trials combining RT with ICIs in both newly diagnosed and recurrent GBM have largely failed to improve survival (Table 2). Several mechanistic barriers have been implicated: (1) Radiation-induced lymphopenia, particularly with large-field cranial irradiation, may counteract immune activation; (2) Field size and dose distribution influence immune cell trafficking and systemic immune competence; (3) Timing and sequencing of immunotherapy relative to RT remain poorly optimized; and (4) The profoundly myeloid-dominated, immunosuppressive tumor microenvironment of GBM limits effector T-cell activity even when immune priming occurs.

Current research efforts therefore emphasize lymphocyte-sparing techniques (including proton therapy), reduced elective margins, hypofractionation, and optimized sequencing strategies rather than simple concurrent administration.

Personalization, advanced imaging and adaptive RT: Work toward individualized RT planning uses advanced MRI/PET imaging to define high-risk regions, MR-LINAC adaptive workflows, and molecular stratification (e.g., MGMT methylation, IDH status) to guide RT intensity and integration with systemic agents. Reviews emphasize the need for imaging-based target definition and biologically adapted dose painting trials[99].

Limitations in the current literature

Most positive signals are biological (e.g., reduced lymphopenia with protons) or from phase II/nonrandomized cohorts; phase III survival evidence is still scarce.

Heterogeneous patient selection, dose/fractionation schedules, and concomitant systemic regimens make cross-study comparisons difficult.

CLINICAL GUIDANCE AND REAL-WORLD INTEGRATION

Clinicians must ensure comprehensive molecular profiling, including IDH1/2, 1p/19q, ATRX, TERT promoter, MGMT methylation, EGFR amplification, FGFR/NTRK fusions, on initial glioma diagnosis. This enables accurate CNS5 classification, trial eligibility, and tailored therapy selection. Given the slow progress in unselected GBM, patients (especially at recurrence) should be prioritized for molecularly guided or early-phase trials. Referral to centers capable of advanced immunotherapy, CAR-T manufacturing, or locoregional delivery is increasingly important.

Transparency about the modest odds of benefit, risks, and logistical burdens of experimental therapies (e.g., CAR-T, vaccine, TTFields) is essential. Discussions should include quality-of-life trade-offs, potential for neurotoxicity, and the experimental nature of newer modalities.

When possible, integrate liquid biopsy, imaging biomarkers, and early response assessments to modulate therapy mid-course. For example, rising ctDNA or EV mutations may prompt transition to alternate therapies or trial arms.

GAPS, CHALLENGES, AND STRATEGIC PRIORITIES

One of the greatest challenges is transforming anecdotal “hero responses” in CAR-T or viral therapy into reproducible, durable outcomes across broader populations. This requires rigorous patient selection, biomarker development, and rational combinations. Improving trafficking, persistence, and resistance avoidance of engineered cells remains a central priority. Locoregional delivery is promising but requires optimization of dosing, routes, and safety margins.

Robust biomarkers predictive of response to immunotherapy (beyond PD-L1 or TMB) are lacking in GBM. Integrative models of neoantigen burden, spatial immune infiltration, and transcriptomic signatures must be prospectively validated. GBM trials struggle with slow accrual and heterogeneous endpoints. Expanding master protocols, international consortia, and common data platforms is essential. Regulatory alignment and centralized biobanks will facilitate cross-trial comparisons and meta-analyses.

Emergent therapies (especially CAR-T) are resource-intensive, and real-world access may lag in lower-resource settings. Strategies to reduce cost, streamline manufacturing, and democratize access are essential for equitable impact.

CONCLUSION

Since 2021, GBM research has not (yet) delivered a paradigm-shifting cure, but it has laid increasingly robust foundations: Molecularly defined cohorts, viable precision approaches in subsets, emerging immunoengineering, and refined translational frameworks. The clearest promise rests in engineered cell therapies delivered locoregionally, combinatorial immunotherapy, and responsive biomarker-guided designs. Over the next decade, success will depend not on silver bullets but on integrative, adaptive, patient-centered strategies, rigorous biomarker validation, and global cooperation.