Molecular Subtypes and Targeted Therapy Advances in Glioblastoma: A Comprehensive Research Report

Introduction

Glioblastoma (GBM) remains one of the most aggressive primary brain malignancies, with median overall survival of approximately 12–17 months from diagnosis and fewer than 10% five-year survival rates despite standard multimodal therapy 85. The past decade has witnessed a fundamental transformation in GBM classification—from purely histology-based to integrative molecular-pathological frameworks mandated by the 2021 WHO Classification of Tumors of the Central Nervous System. Yet despite unprecedented advances in molecular profiling, the clinical translation of targeted therapies has consistently lagged behind biological understanding, constrained by the blood–brain barrier (BBB), profound intratumoral heterogeneity, clonal plasticity, and an immunosuppressive tumor microenvironment 1012.

This report synthesizes findings from the most recent clinical trials (2025–2026), regulatory developments, and foundational translational research to provide a comprehensive overview of GBM molecular subtypes, actionable biomarkers, targeted therapy progress, and clinical translation pathways.

1. Molecular Subtype Framework in GBM

1.1 Historical and Current Research Subtypes

The Cancer Genome Atlas (TCGA)-era transcriptional classification of GBM into classical, proneural, mesenchymal, and neural subtypes established the first molecular framework for understanding GBM heterogeneity. Subsequent work converged on three more durable biological states—classical, proneural, and mesenchymal—while increasingly emphasizing dynamic cellular-state plasticity rather than rigid subtype boundaries 17.

Critically, longitudinal profiling of 106 paired initial and recurrent IDH-wildtype GBM specimens revealed that methylation class transitions from RTK-like (RTK2) at initial diagnosis to mesenchymal (MES) at recurrence are common events 1. This epigenomic plasticity represents a fundamental challenge for any therapy designed against a fixed subtype. The observation that GBM undergoes substantial genetic and epigenetic evolution between diagnosis and recurrence—with alterations in EGFR, PTEN, TP53, and NF1 frequently being private to either the initial or the recurrent tumor—further undermines static, single-timepoint biomarker approaches 1.

1.2 The IDH-Mutant versus IDH-Wildtype Distinction: The Most Clinically Consequential Divide

The 2021 WHO Classification established IDH mutation status as a key diagnostic criterion distinguishing major adult-type diffuse glioma entities; under this framework, glioblastoma is defined as IDH-wildtype, while IDH-mutant grade 4 tumors are classified as astrocytoma, IDH-mutant, CNS WHO grade 4.

A particularly important nuance is the molecular GBM (mGBM) subgroup—tumors lacking classical histopathologic GBM features but meeting molecular criteria (TERT promoter mutation, EGFR amplification, or +7/−10 genotype). In the Beijing Tiantan cohort (n=18 mGBM patients), these tumors had smaller sizes, minimal contrast enhancement (77.8% vs 4.0% in histologic GBM), and higher seizure incidence (55.6% vs 23.0%), yet similar median survival to histologic GBM (21.2 vs 20.0 months, p=0.454). These findings illustrate that molecularly defined IDH-wildtype glioblastoma can present with less classic imaging or histologic features while still showing survival outcomes comparable to histologic glioblastoma; however, the small cohort size means this observation should be interpreted cautiously 2.

A predictive model incorporating age, tumor margin, and cortical non-enhancing tumor infiltration (CnCE) achieved AUC 0.890 in training and 0.951 in external validation for identifying IDH-mutant astrocytoma grade 4 before surgery, suggesting that preoperative imaging and clinical features can guide WHO classification decisions 2.

1.3 Actionable Alterations in IDH-Wildtype GBM

EGFR is the most frequently amplified or overexpressed oncogene in IDH-wildtype GBM, present in approximately 50–60% of cases 1417. The EGFRvIII variant—a constitutively active truncated form resulting from in-frame deletion of exons 2–7—occurs in a subset of EGFR-amplified GBMs and provides a tumor-associated target; however, EGFRvIII expression is heterogeneous and may not be present in all EGFR-amplified tumor cells 1217.

Beyond EGFR, several other alterations define the IDH-wildtype GBM landscape:

Radiogenomic analysis of 357 IDH-wildtype GBMs demonstrated distinctive MRI signatures for specific driver mutations. EGFR-mutant tumors showed hypervascularity with increased relative cerebral blood volume (rCBV); TP53-mutant tumors displayed altered vascular permeability in enhancing tumor and elevated neovascularization in peritumoral edema; PTEN-mutant tumors had increased water content in the non-enhancing core; NF1-mutant tumors showed greater vascularization and lower diffusion 3. These radiogenomic correlations, while still investigational, offer noninvasive approaches to infer driver mutations at preoperative assessment.

1.4 Gliosarcoma Transformation at Recurrence

An important clinical observation from longitudinal profiling is that 17% (18/106) of patients in one cohort transformed to gliosarcoma at recurrence. These transforming tumors were enriched for NF1, TP53, and RB1 alterations and lacked EGFR amplification, with predominance of mesenchymal epigenetic class—suggesting specific therapeutic opportunities targeting mesenchymal or TGF-β signaling pathways 1.

2. Biomarker Rationale and Patient Selection

2.1 Biomarker Overview and Clinical Utility

2.2 MGMT Methylation: The Most Clinically Actionable Biomarker

MGMT methylation occupies a uniquely important position because it directly informs both prognosis and treatment selection. A retrospective study of 142 inoperable IDH-wildtype GBM patients stratified by MGMT methylation status demonstrated 6:

• Unmethylated (<5%): n=70, 49.3% of patients
• Borderline methylated (5–15%): n=29, 20.4%
• Strongly methylated (>15%): n=43, 30.2%

One-year survival probabilities were 0.18 (unmethylated), 0.21 (borderline), and 0.33 (methylated). In Cox proportional hazard modeling, MGMT methylation was an independent predictor of OS (coefficient −0.30, p=0.01), with each methylation category increase associated with a 26% lower risk of shorter OS (HR 0.74). The model achieved a concordance index of 0.72. Critically, this predictive value held even in biopsy-only patients, supporting use of MGMT status to guide treatment intensity even when resection is not feasible 6.

Additionally, four specific CpG sites in the MGMT promoter region were reported to predict temozolomide-induced hypermutation with 92% sensitivity and 87% specificity, suggesting a potential biomarker for identifying patients at risk of TMZ-associated hypermutation rather than a biomarker proving benefit from prolonged alkylating exposure 1.

2.3 ADC Histogram Analysis as an Imaging Biomarker

A retrospective study of 79 GBM patients evaluated ADC histogram metrics from diffusion-weighted MRI for predicting MGMT and TERT status 7:

• MGMT methylation prevalence: 53.2%; TERT mutation prevalence: 44.3%
• ADC 10th percentile (ADCp10) was significantly lower in MGMT-unmethylated vs methylated groups (p=0.005), suggesting higher cellularity in unmethylated tumors
• For TERT status: entropy (histogram heterogeneity) showed best diagnostic performance (AUC=0.722, p=0.001)
• Overall survival positively correlated with ADCmin; TERT mutation was the only independent prognostic factor in multivariate analysis

This noninvasive approach may complement tissue profiling when sampling is limited or risky, though external validation in larger multi-institutional cohorts is necessary before clinical implementation 7.

2.4 Regulatory Framework for Companion Diagnostics

The FDA companion diagnostic framework requires cleared or approved in vitro diagnostics for safe and effective use of corresponding therapeutics 4. Current GBM testing (IDH, MGMT, TERT, EGFR) predominantly relies on laboratory-developed tests (LDTs) or research-grade assays rather than FDA-cleared companion diagnostics—a significant gap in regulatory standardization that could complicate pivotal trial registration. This regulatory context suggests that any targeted therapy seeking GBM approval would require development of a corresponding companion diagnostic assay meeting FDA approval standards 4.

2.5 Biomarker-Defined Subgroups Most Likely to Benefit from Targeted Approaches

3. Targeted Therapy Landscape

3.1 IDH-Targeted Therapies

Vorasidenib: Phase 3 Landmark Achievement in IDH-Mutant Glioma

The INDIGO phase 3 trial represents the most significant precision medicine advance in molecularly defined glioma in the past decade 19. The study enrolled 331 patients aged ≥12 years with residual or recurrent IDH1/2-mutant low-grade glioma.

The trial also confirmed no negative effects on health-related quality of life (FACT-Br scores: 154.2 vs 153.2 at end of treatment) or neurocognitive function across multiple domains. The investigators concluded that these findings support vorasidenib use in grade 2 IDH1/2-mutant diffuse glioma patients who underwent surgery only and are not in immediate need of radiotherapy or chemotherapy 19.

Critically, as of the retrieved materials, vorasidenib efficacy data in IDH-mutant GBM (grade 4) are not yet mature, and no comprehensive phase 3 GBM efficacy readout was found in the retrieved materials 17.

3.2 EGFR-Directed Therapies

3.2.1 Antibody-Drug Conjugate: Depatuxizumab Mafodotin (Depatux-M, ABT-414/ABBV-221)

A systematic review and meta-analysis synthesizing evidence from 10 studies (6 RCTs, 4 cohort studies; n=1,431 patients) provided the most comprehensive assessment of depatuxizumab mafodotin 21:

Safety profile: Ocular toxicity affected approximately 65% of patients (often Grade 1–2, frequently reversible); Grade 3+ ALT elevation in 10%; Grade 3+ AST in 5%; seizures Grade 3+ in 4% 21.

Key interpretation: EGFR amplification alone appears insufficient for patient selection. Spatial heterogeneity of EGFR expression, EGFRvIII heterogeneity, and BBB penetration challenges persist as barriers. Absolute OS improvements remain modest (approximately 2–3 months) 21.

3.2.2 Small Molecule EGFR TKIs

The historical pattern for erlotinib, gefitinib, afatinib, dacomitinib, lapatinib, and neratinib is consistently poor in GBM. While afatinib showed some indication of longer PFS in tumors with EGFRvIII expression or EGFR amplification, no EGFR TKI has established a clear precision-therapy win in GBM 17. A critical distinction from lung cancer is that GBM-associated EGFR alterations often involve amplification, EGFRvIII, extracellular-domain alterations, marked spatial heterogeneity, and limited CNS drug delivery, all of which may contribute to the poor clinical performance of conventional EGFR kinase inhibitors in GBM 19.

3.2.3 Monoclonal Antibodies

Cetuximab failed to demonstrate meaningful benefit in recurrent GBM as monotherapy or with bevacizumab/irinotecan. The nimotuzumab phase 3 trial (newly diagnosed GBM; RT + TMZ backbone) showed no significant OS benefit (PFS 7.7 vs 5.8 months; OS 22.3 vs 19.6 months) 17.

3.2.4 CAR T Cell Therapy: CART-EGFR-IL13Rα2

A first-in-human phase 1 trial enrolled 18 adults with recurrent IDH-wildtype GBM with EGFR amplification for intracerebroventricular (ICV) delivery of bivalent CAR T cells targeting EGFR epitope 806 and IL-13Rα2 20:

• Rationale for dual targeting: EGFR epitope 806 (accessible with EGFR amplification and activation) is present in 50–60% of GBMs; IL-13Rα2 (a cancer-testis antigen) is expressed in ≥75% of GBMs. Dual targeting was designed to address intratumoral antigen heterogeneity
• MTD identified: 2.5 × 10^7 CAR T cells
• Safety: Grade 1–2 CRS in 100% of patients; Grade 3 neurotoxicity in 56% (10/18 patients)—all reversible by Day 28 except one patient with chronic neurotoxicity maintaining durable disease control; no Grade 4–5 neurotoxicity
• Efficacy: ORR 8% (1/13 patients with measurable disease); stable disease 61%; progressive disease 31%; median PFS 1.9 months; 2 long-term responders with PFS of 7.7 and 16.6 months (both ongoing at data cutoff)
• Pharmacokinetics: CSF CAR T cell expansion peaked Day 4–10 and declined by Day 14–28; dose-dependent CSF expansion; marked pro-inflammatory cytokine increase post-infusion resolving by Day 14

Retreatment resistance was identified as a significant barrier: less robust CSF expansion, shorter CAR T persistence, and limited PFS upon re-challenge 20.

3.3 Anti-Angiogenic Therapy

Bevacizumab (anti-VEGF monoclonal antibody) represents the most clinically validated targeted approach in GBM, with clear but compartmentalized activity:

The mechanistic interpretation is instructive: bevacizumab targets an environmental/vascular phenotype rather than a subclonal tumor genotype, making it more likely to improve imaging, edema control, steroid-sparing, and PFS across a broad population—even without OS extension 17.

A bevacizumab biosimilar (SCT-510; Sinocelltech) has been approved in China for brain neoplasm malignant, confirming continued anti-angiogenic development in this indication regionally 18.

3.4 PI3K/AKT/mTOR Pathway Inhibitors

Despite the strong biological rationale (EGFR activation and PTEN loss are common in GBM), pathway inhibitors have consistently underperformed:

The failure pattern is explained by pathway redundancy: compensatory RTKs (PDGFR, c-MET), feedback loops, and state plasticity allow tumors to maintain downstream signaling despite pathway blockade 17.

3.5 Immunotherapy and Cancer Vaccines

DCVax-L Dendritic Cell Vaccine

Phase 3 trial results (n=331 newly diagnosed and recurrent GBM patients across 94 sites in four countries; published in JAMA Oncology 2022) 22:

Important methodological caveats highlighted by the American Brain Tumor Association: (1) trial design changes were made mid-study; (2) external controls were used rather than a concurrent randomized control arm; and (3) FDA approval has not yet been granted, with clinical availability limited 22. These limitations constrain the evidentiary level for regulatory purposes.

Checkpoint Inhibitors and Other Immunotherapies

The comprehensive GBM burden review explicitly states that "immunotherapies and targeted therapies currently have only a limited role due to disappointing clinical trial results, including in recurrent glioblastoma" 5. The immunosuppressive tumor microenvironment—characterized by exhausted CD8+ and CD4+ T cells, regulatory T cells (Tregs), and M2-like macrophages—remains a central barrier 9.

3.6 Oncolytic Virotherapy

Oncolytic viruses offer the potential to modulate neuroinflammatory responses, induce localized immune reactions, and deliver immunomodulatory factors directly to the tumor site 11. Encouraging preclinical and early clinical signals exist, but challenges include: overcoming the BBB, managing host antiviral immunity, and potential risks to normal neurons. Combination with checkpoint inhibitors, NK cell therapy, CAR-T cells, or dendritic cell therapy is thought to offer the most effective near-term development path 11.

3.7 Novel Molecular Targets: Alpha-Synuclein and Temozolomide Resistance

A mechanistically distinctive study revealed that alpha-synuclein (α-syn), a Parkinson's disease-associated protein, functions as a tumor suppressor in GBM 13:

• α-Syn is a TP53 transcriptional target reduced in GBM biopsies compared to lower-grade gliomas
• Downstream effects: α-syn lowers cyclin D1 protein and mRNA, reducing GBM cell proliferation
• Overcomes TMZ resistance: In temozolomide-resistant U87 cells, α-syn reduces MGMT expression and rescues drug sensitivity through XBP1-mediated transcriptional regulation (unfolded protein response)
• In vivo validation: α-syn lowers MGMT and cyclin D1 expression and reduces tumor development in allografted mice

This finding suggests that inducing α-syn expression or the unfolded protein response could sensitize MGMT-expressing, TMZ-resistant GBMs to alkylating chemotherapy—representing a novel mechanistic approach to overcome resistance 13.

3.8 Real-World Evidence: Molecular Tumor Board Outcomes

A prospective real-world study from the Tübingen molecular tumor board (MTB@CPM, NCT03503149; n=1,088 patients through June 30, 2024) provided important context for biomarker-guided precision medicine in clinical practice 23:

Setting-specific outcomes are particularly informative:

The 52-day median turnaround from molecular profiling to MTB recommendation represents a significant practical barrier in rapidly progressing GBM, where disease can change substantially during this window 23.

4. Clinical Translation Pathways

4.1 Why Biology Has Not Consistently Become Clinical Benefit

The most frequently investigated targeted-therapy class in GBM has been EGFR (40 EGFR-focused trials; 85 RTK-focused trials overall), followed by anti-angiogenic (75 trials, including 53 bevacizumab studies) from a systematic review of 257 phase II–IV trials 17. The translation record is sobering.

4.2 Focused Ultrasound as a BBB Delivery Solution

A proof-of-concept clinical study suggested that focused ultrasound BBB opening before chemotherapy may be feasible and safe in newly diagnosed GBM patients, with preliminary survival signals compared with matched controls that require confirmation in larger randomized studies 24. Dr. Graeme Woodworth noted: "Using focused ultrasound to open the blood-brain barrier and deliver chemotherapy could significantly improve patient survival." Ongoing studies seek to confirm and expand these findings, with potential extension to targeted therapies (antibodies, small molecules, ADCs) 24.

4.3 GBM AGILE Platform Trial: A Paradigm Shift in Trial Design

The GBM AGILE (Adaptive Global Innovative Learning Environment) Phase II/III Bayesian randomized platform trial reported the regorafenib arm in a Journal of Clinical Oncology article posted in April 2026, showing that regorafenib did not improve overall survival compared with control therapy in newly diagnosed or recurrent glioblastoma 25.

The regorafenib arm was stopped for limited efficacy at a defined interim analysis, demonstrating the platform's ability to efficiently terminate non-productive development pathways. This design provides a model for simultaneously testing multiple biomarker-stratified approaches within the same trial infrastructure 25.

4.4 Epigenomic Evolution Signature as a Novel Biomarker

A 347-CpG site DNA methylation evolution signature was identified that correlates with clinical outcomes in IDH-wildtype GBM; patients stratified by mean Δβ (change in methylation) across these sites showed prognostic separation with external validation support 1. This dynamic epigenomic signature captures inter-treatment changes and may ultimately inform risk stratification and treatment intensity decisions in a longitudinal fashion.

4.5 Trial Design Principles Most Likely to Improve Success

Based on synthesized evidence, the following design principles offer the highest probability of translational success 1715:

1. Molecular enrichment over all-comer enrollment: The N2M2 (NOA-20) concept—molecularly matched therapy in newly diagnosed MGMT-unmethylated GBM—exemplifies this approach and is specifically highlighted as the most rational frontline precision-trial architecture 17

2. Composite biomarker models: Single markers such as EGFR amplification alone are inadequate; composite panels integrating IDH, MGMT, TERT, EGFR, and pathway context (PTEN loss) are more likely to identify true therapeutic dependency 1617

3. Longitudinal profiling: Obtaining tissue at recurrence when feasible to capture clonal evolution; radiogenomics as a complementary noninvasive approach 13

4. Mechanism-informed combinations: Rational combinations addressing primary driver pathway plus common adaptive escape routes, aligned to disease setting and MGMT status 15

5. BBB-aware development: Brain-penetrant drug design, intratumoral PK/PD confirmation, and delivery innovations as prerequisites rather than afterthoughts 1516

6. Mechanism-aligned endpoints: Pairing OS with edema control, steroid-sparing, neurocognitive function, and seizure frequency to detect clinically meaningful benefit even when OS is difficult to move 196

5. Disease-Setting Implications

5.1 Comparison of Newly Diagnosed versus Recurrent GBM

5.2 MGMT Methylation in Inoperable Newly Diagnosed GBM

Among 142 inoperable IDH-wildtype GBM patients receiving only biopsy, overall median survival was 184 days. Treatment group outcomes 6:

Treatment category was independently significant (coefficient −0.39, p<0.005), with MGMT methylation status remaining independently predictive even in this exclusively inoperable cohort. This supports using MGMT status to guide treatment intensity and therapy selection even when resection is not feasible 6.

5.3 IDH-Mutant GBM (Grade 4) as a Distinct Disease Setting

IDH-mutant astrocytoma grade 4 represents a distinct clinical population requiring separate treatment considerations:

• Superior survival vs IDH-wildtype GBM (HR 0.39 in multivariate analysis; p=0.03) 2
• Higher MGMT methylation rate (29.0% vs 17.1%)
• Younger age at presentation
• Theoretically amenable to IDH inhibitors, but clinical validation in grade 4 disease is still lacking in the retrieved materials 17

6. Competitive and Development Outlook

6.1 Leading Programs by Target and Modality (Global, US/EU/China Focus)

6.2 China-Specific Development Landscape

China represents a significant hub for brain-targeted biologics and cellular therapies, with several notable features from the retrieved dataset 18:

• EGFR/EGFRvIII CAR-T programs (A-1910 and DCTY-0801) with explicit brain malignancy labeling are in preclinical stage, reflecting regionally concentrated early-stage innovation
• Bevacizumab biosimilar (SCT-510; Sinocelltech) is approved for brain malignancy in China
• Multiple China-based EGFR TKIs at advanced stages (rezivertinib approved; sacibertinib Phase 3) provide platform infrastructure potentially adaptable to GBM
• Ridaforolimus (mTOR inhibitor; Merck/Takeda) is in Phase 3 for brain malignancy in China—the most advanced mTOR-specific program with brain labeling in the dataset

No comprehensive NMPA 2025 glioma drug approval or IND filing data were found in the retrieved materials 18.

6.3 EGFR Inhibitor Competitive Landscape (Pathway-Level)

The breadth of EGFR-directed programs across modalities illustrates the multiple angles of attack being pursued against EGFR-altered CNS disease, even as brain-specific clinical validation remains elusive 18:

6.4 Recent Material Clinical Readouts and Regulatory Developments (2025–2026)

As of April 2026, vorasidenib is an FDA-approved targeted therapy for patients aged 12 years and older with Grade 2 astrocytoma or oligodendroglioma harboring a susceptible IDH1 or IDH2 mutation following surgery, while its efficacy in IDH-mutant astrocytoma, CNS WHO grade 4 remains unproven 1922.

6.5 Near-Term Catalysts and Open Questions

7. Practical Implications for Trial Design, Biomarker Selection, and Commercial Development

7.1 Routine Clinical Testing Recommendations

Based on synthesized evidence, the following molecular profiling framework is supported for all GBM patients 567:

• Mandatory: IDH1/2 mutation status (diagnostic, prognostic, treatment-defining); MGMT promoter methylation (predictive for TMZ; mandatory before therapy decisions)
• Strongly recommended: TERT promoter mutation (prognostic refinement; composite classifier role); EGFR amplification and EGFRvIII status (trial enrollment; monitoring with understanding of subclonal limitations)
• Contextually recommended: PTEN status (pathway context; PI3K/mTOR trial selection); NF1/TP53/RB1 (gliosarcoma transformation risk; mesenchymal transition indicators)
• Emerging/investigational: Radiogenomic imaging biomarkers (ADC histogram, rCBV, PSR); CSF liquid biopsy for longitudinal monitoring; 347-CpG epigenomic evolution signature

7.2 Enrollment Strategy for Targeted Therapy Trials

• Prioritize biomarker-enriched trials over all-comer designs; the INTELLANCE 1 negative result in unselected newly diagnosed GBM and the N2M2 precision-matching model collectively demonstrate the failure of unselected and the promise of enriched approaches 1721
• For EGFR-directed programs: require not just EGFR amplification by FISH, but assessment of EGFRvIII status, spatial heterogeneity considerations, and co-expression of relevant antigens (e.g., IL-13Rα2 for CAR-T) 20
• For frontline targeted therapy: MGMT-unmethylated newly diagnosed GBM represents the highest unmet need subgroup for non-TMZ alternative regimens, where a new targeted approach would add the most to existing options 17

7.3 Commercial Development Considerations

• IDH-mutant disease: Vorasidenib regulatory success in low-grade glioma establishes a commercial and clinical precedent; combination strategies and grade 4 extension are logical next development steps
• EGFR-amplified GBM: Remains commercially attractive but clinically unvalidated; future success requires brain-penetrant agents, subclonal monitoring, combination therapy, and BBB delivery solutions 17
• Cellular therapies: China represents a significant innovation hub for EGFRvIII and EGFR CAR-T programs with brain labels; partnering opportunities exist for clinical-stage development 18
• Anti-angiogenic: Established market with biosimilar competition (bevacizumab); differentiation requires improved OS data or combination strategies
• Manufacturing complexity: Personalized approaches (DCVax-L, autologous CAR-T) face logistical barriers including 52-day+ turnaround time from profiling to treatment, manufacturing cost, and reimbursement complexity 2223
• Companion diagnostic alignment: Companies developing targeted GBM therapies should engage with FDA CDx framework early; current reliance on LDTs represents a regulatory gap that could impede approval 4

Conclusions

The molecular understanding of glioblastoma has undergone a fundamental transformation. The IDH-mutant/IDH-wildtype dichotomy is the most clinically consequential molecular division, with IDH-mutant tumors now having their first validated targeted therapy (vorasidenib) showing phase 3 benefit in low-grade disease 19. MGMT promoter methylation remains the most clinically actionable biomarker across all GBM settings, predicting alkylator benefit in both operable and inoperable patients and stratifying prognosis across treatment modalities 56.

Despite decades of investigation, IDH-wildtype GBM remains the most challenging setting. EGFR amplification is biologically central but therapeutically poorly validated: depatuxizumab mafodotin showed no benefit in newly diagnosed GBM and only modest benefit at recurrence 21; CAR-T cells targeting EGFR and IL-13Rα2 demonstrated feasibility but limited durability (median PFS 1.9 months) 20; small-molecule EGFR inhibitors have consistently underperformed 17. The barriers are biological—intratumoral heterogeneity, subclonal antigen expression, bypass signaling, epigenomic plasticity—compounded by the blood-brain barrier and immunosuppressive microenvironment 910.

Potential near-term translation pathways include studying IDH inhibitors in IDH-mutant astrocytoma, CNS WHO grade 4; refining EGFR-amplified GBM trials with composite biomarker enrichment and rational combinations; improving CAR-T persistence; evaluating focused ultrasound BBB opening in controlled studies; and using platform trials such as GBM AGILE to test biomarker-stratified approaches more efficiently 19202425. The field has fundamentally shifted from asking "what is the target?" to asking "in which molecular state, in which disease setting, with what combination, and with what delivery strategy?" — a shift that reflects hard-won lessons from over two decades of largely unsuccessful single-agent trials in this devastating disease 17.