Current Status and Evolution of Immunotherapy in Glioma Management Authors: Jiaying Liu, Guangzhao Yang, Zhengcong Cao, Haozhe Qin, Maorong Zhu, Cheng Zou, Xiao Liu, Yalong He, Jintao Gu Published: 2026 ABSTRACT AbstractGliomas, especially glioblastoma (GBM), are the most aggressive primary brain tumors, and their treatment faces multiple challenges. These challenges include incomplete surgical resection, the blood-brain barrier (BBB) that limits drug delivery, and an immunosuppressive tumor microenvironment (TME). Despite the "cold tumor" nature of gliomas, significant progress has been made in immunotherapy strategies. In this review, we focus on the following four strategies: 1) Cancer vaccines, especially personalized vaccines based on neoantigens, can activate specific immune responses; 2) Immune checkpoint inhibitors (ICIs) to reverse T cell exhaustion (e.g., anti-PD-1/PD-L1 antibodies); 3) Adoptive immune cell therapies, particularly chimeric antigen receptor T-cell (CAR-T) therapy, have shown encouraging results in preclinical trials; 4) Oncolytic viruses, which have dual roles of directly lysing tumor cells and stimulating immune responses. Immunotherapy offers a new paradigm to overcome the limitations of conventional glioma therapy. However, tumor heterogeneity, BBB limitation, target selection, and immunosuppressive microenvironment remain major obstacles. Future research should focus on developing combination therapies such as immunotherapy combined with radiotherapy, targeted therapy, exploring novel targets, and using biomarkers for patient stratification, which may help improve the survival outcomes of glioma patients. This review comprehensively evaluates the latest clinical progress of glioma immunotherapy, assesses the current limitations, and explores future directions.Keywords: glioma, immunotherapy, immune checkpoint inhibitors, CAR-T cell therapy, cancer vaccine, oncolytic virus Introduction Glioma is a central nervous system tumor arising from glial cells. It is the most common and most aggressive primary brain tumor in adults 1. The pathogenesis of glioma is complex, involving various genetic and molecular alterations 1. The 2021 WHO Classification (5th edition) has shifted the classification of adult-type diffuse gliomas from a purely histologic grading system to an integrated paradigm that combines histologic features with molecular biomarkers 2. These tumors are primarily categorized into three subtypes: 1) IDH-mutant astrocytoma: cases with CDKN2A/B homozygous deletion are directly assigned WHO grade 4, irrespective of classic histologic features such as microvascular proliferation or necrosis 3; 2) IDH-mutant oligodendroglioma with 1p/19q codeletion 3; 3) IDH-wildtype glioblastoma, which is strictly defined as IDH-wildtype grade 4 astrocytoma and represents the most malignant and aggressive subtype 2. Furthermore, distinct genetic alterations are increasingly recognized to shape the tumor immune microenvironment and consequently modulate the response to immunotherapeutic interventions 4. Although its incidence is relatively low, the median survival for GBM patients is only approximately 15 months, indicating a very poor prognosis 5. Standard treatment includes surgical resection, radiotherapy, and temozolomide (TMZ) chemotherapy 6. Additional therapeutic options include DNA damage response (DDR) inhibitors and targeted molecular therapies 7. Despite these measures, patient outcomes remain poor, with high recurrence rates 7. The poor prognosis stems from multiple factors. First, the highly invasive nature of glioma cells enables diffuse infiltration into normal brain tissue, precluding complete surgical resection 8. Second, the BBB severely hinders effective drug delivery, limiting systemic therapy efficacy 5. Third, the glioma microenvironment is highly immunosuppressive. It features abundant M2-polarized tumor-associated macrophages (TAMs), increased infiltration of regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), exhausted cytotoxic T lymphocytes, and upregulation of multiple immune checkpoints 1. These factors, together with metabolic suppression (e.g., tryptophan catabolism via the IDO pathway) and hypoxia-driven immunosuppression, collectively shape the "cold tumor" phenotype of gliomas 9. Furthermore, glioma has a relatively low tumor mutational burden and high intra- and inter-tumoral heterogeneity. This further weakens the immune system's ability to recognize and eliminate the tumor 1. Together, these factors present unique challenges, prompting the exploration of immunotherapy as a new strategy. Despite inherent immunosuppression, immunotherapy for glioma is feasible 10. Recent re-evaluation of central nervous system (CNS) immune privilege and the discovery of skull-meningeal channels and intracranial lymphatic vessels provide new perspectives on brain tumor immunology 10. Studies show that inducing immunogenic cell death (ICD) can transform the glioma microenvironment from "cold" to "hot", promote Th17 cell migration to the tumor, and enhance immunotherapy efficacy 11. Additionally, immune cell infiltration patterns in the TME correlate with patient prognosis. Stronger immune cell infiltration often predicts better survival 11. These findings provide a theoretical basis for developing immunotherapies for glioma. Current glioma immunotherapy mainly includes four classic strategies. Cancer vaccines introduce tumor-specific dendritic cells (DCs) or tumor-associated antigens (TAAs) to activate anti-tumor immune responses 12. Personalized vaccines, especially those targeting neoantigens, show great potential 1. ICIs like PD-1, PD-L1, and CTLA-4 inhibitors aim to reverse T cell exhaustion and reactivate anti-tumor immunity 1. Based on the available literature, adoptive immune cell therapy has not yet demonstrated satisfactory clinical efficacy in glioma studies, and its safety profile requires further investigation 13-15. However, breakthroughs in new therapies such as universal CAR-T, fourth-generation CAR-T, and novel-target CAR-T in preclinical research have brought hope for improving the prognosis of glioma patients. Oncolytic viruses have a dual mechanism: directly killing tumor cells and activating immune responses. They have made breakthrough progress in clinical trials 1. This review comprehensively evaluates recent clinical progress in these four main immunotherapy strategies for glioma. It critically analyzes limitations and provides insights into future directions. 1. Cancer Vaccines Cancer vaccines aim to induce the immune system to recognize and attack TAAs or tumor-specific antigens (TSAs) 12 (Figure 1). In early exploration (around the 1990s), the main type of tumor vaccine was the whole tumor cell vaccine 16. They had manageable safety but limited clinical efficacy 16; In the early 2000s, DC vaccines emerged. They are made by isolating monocytes from patients' blood, differentiating them into DCs in vitro, loading them with tumor antigens (e.g., cell lysates, synthetic peptides), and reinfusing them to activate specific T cells 17. In a large Phase 3 trial of a dendritic cell vaccine, the median overall survival (OS) among 331 patients was 23.1 months. Among the 131 patients with MGMT promoter methylation, the 3-year survival rate reached 46.4%. These findings provide strong evidence supporting the survival benefit of the DC vaccine 16; Around the 2010s, molecularly targeted peptide vaccines appeared. They use defined immunodominant epitope peptides to induce precise immune responses 18. IMA950 is the first multi-peptide vaccine targeting glioblastoma to enter clinical trials 19. In its first-in-human Phase I study, it successfully met the safety and immunogenicity endpoints, marking a transition in glioma vaccine development from cell-based approaches to precision molecular targeting 19. Since 2015, personalized/neoantigen vaccines have developed rapidly. These are peptide-based personalized cancer vaccines composed of immunogenic mutant epitopes derived from the patient's specific glioma 20. The figure illustrates the mechanisms of action of four types of vaccines: dendritic cell vaccines, peptide vaccines, DNA vaccines, and RNA vaccines. These vaccines promote the activation and maturation of immune cells by presenting tumor-associated antigens. Subsequently, the activated immune cells induce immunogenic cell death of tumor cells, which in turn leads to the release of a broader repertoire of tumor-associated antigens and neoantigens. The newly released antigens further recruit and activate additional immune cells and also promote the reversion of suppressive immune cells, thereby effectively reversing the immunosuppressive state within the TME. 1.1 Dendritic Cell Vaccines Based on literature from 2015-2025, 14 DC vaccine studies entered clinical trials. Comparative analysis preliminarily confirmed the safety and tolerability of DC vaccines 21-25, demonstrating good prospects in the field of glioma immunotherapy. Representative studies include 1) Wilms Tumor 1 (WT1) mRNA vaccine 21; 2) autologous tumor antigen (ATA) vaccine 23, 24; 3) ICT-107 25; 4) DC vaccines loaded with glioma stem cell (GSC) (A2B5+) antigen 26; 5) DC vaccines targeting CMVpp65 27-29. First, in a single-arm Phase I/II study of advanced solid tumors, WT1 mRNA-electroporated autologous DCs (WT1-mRNA/DC) demonstrated promising therapeutic activity. The induced type 1 T-cell responses were associated with improved OS 21. Second, the ATA vaccine's antigens come from self-renewing irradiated autologous tumor cells in culture. It contains personal neoantigens and any GBM TAAs, but not normal cell antigens, offering potential advantages over autologous tumor lysate 23. However, its production requires sufficient tumor tissue, limiting use in advanced patients 23. Finally, the efficacy and safety of ICT-107 and GSC antigen-loaded DC vaccines need further validation in larger trials 25, 26. Specific pretreatment of DC vaccines (e.g., using tetanus-diphtheria (Td) toxoid 28, 29) combined with adjuvants (e.g., keyhole limpet hemocyanin, KLH) 21 or cytokines (e.g., GM-CSF) 25, or using activators (e.g., IFNγ or lipopolysaccharide 25), can enhance vaccine potency and patient immune responses to TSAs. This may improve patient outcomes 21, 25. Many studies indicate that patients with O-6-methylguanine-DNA methyltransferase (MGMT) promoter methylation, combined Telomerase Reverse Transcriptase (TERT) promoter and Isocitrate Dehydrogenase 1/2 (IDH1/2) mutations, low B7-H4 expression, Human Leukocyte Antigen A2 (HLA-A2) positivity, and high immune responses are more sensitive to DC vaccines. This may be associated with better prognosis 25, 26, 30. 1.2 Molecularly Targeted Peptide Vaccines Representative studies from 2015 to 2025 include 1) the Epidermal Growth Factor Receptor variant III (EGFRvIII) targeted vaccine (Rindopepimut/CDX-110) 31, 32; 2) IDH mutation-targeted vaccine (IDH1-vac) 33; 3) Survivin-targeting peptide vaccine (SurVaxM) 34, 35; 4) and WT1 peptide vaccine 36 as single-target vaccines. Multi-target peptide vaccines include 1) IMA950 19, 37; 2) WT1 peptide cocktail vaccine 38; 3) VEGFR1-VEGFR2 peptide vaccine 39; 4) glioma-associated antigen (GAA) synthetic peptide vaccine 40. 1.2.1 Single-Target Vaccines First, Rindopepimut is a single-target peptide vaccine against EGFRvIII. It consists of an EGFRvIII-specific peptide conjugated to KLH. In the ACT III trial, Rindopepimut successfully induced specific immune responses, achieving a 5.5-month PFS rate of 66%, significantly exceeding the preset null hypothesis of 53% (P=0.0168) 32. In terms of safety, vaccine-related adverse events were common but generally tolerable 32. However, the subsequent Phase III ACT IV trial showed that Rindopepimut did not improve survival in patients with newly diagnosed glioblastoma (nGBM) 31. The ACT III trial demonstrated promising efficacy; however, the subsequent Phase III ACT IV trial failed to confirm these findings. Several factors may account for this discrepancy: 1) Patient selection bias. ACT III enrolled a highly selected population with favorable responses to initial therapy 32. Comparing their survival outcomes with unselected historical controls likely overestimated the therapeutic benefit of the vaccine 32. 2) Improvements in standard of care. In ACT IV, the median OS in the Rindopepimut arm (20.1 months) was consistent with that observed in ACT III, indicating stable vaccine activity 31. However, advances in standard therapy likely prolonged survival in the control group relative to historical data, thereby diluting the relative benefit of Rindopepimut and rendering the difference statistically nonsignificant 31. 3) Instability of target expression. In a small subset of ACT IV patients, EGFRvIII expression was lost at recurrence in approximately 60% of cases, irrespective of treatment received 31. This finding suggests that single-target strategies against EGFRvIII may be limited and supports the exploration of multi-target approaches in future studies 31. Second, the IDH1(R132H) peptide vaccine (IDH1-vac) showed good safety in its first trial, but the pseudoprogression rate was high, correlating with increased peripheral T cell responses 33. Third, SurVaxM has good safety and tolerability 34, 41. Clinically, benefit was observed in both methylated and unmethylated patients, with longer survival in the former. Notably, this clinical benefit was associated with upregulation of tumor-associated B cells, which predicted better outcomes 35. However, it is essential to interpret such early-phase findings with caution. A well-documented challenge in GBM research is the frequent discrepancy between encouraging Phase I/II results and the outcomes of subsequent Phase III trials. Initial efficacy signals often fail to be reproduced in larger, more rigorous, and controlled settings, as exemplified by the Rindopepimut trials and other candidate immunotherapies in neuro-oncology 31, 32, 42, 43. A large randomized trial of SurVaxM for nGBM (NCT02455557) was completed in March 2026, but the results have not yet been reported. Finally, the WT1 peptide (WT2725) vaccine is safe and tolerable but its efficacy needs confirmation 36. 1.2.2 Multi-Target Vaccines First, IMA950 is a novel multi-target vaccine containing 11 tumor-associated peptides identified from HLA surface receptors in GBM tissue. It was used to induce antigen-specific peripheral CD8+ T cell responses in HLA-A*02-positive patients. In terms of safety, drug-related toxicity was relatively benign 19. Second, combining TMZ with VEGFR1-VEGFR2 peptide vaccine showed synergistic activity 39. Third, the GAA synthetic peptide vaccine targets IL-13Rα2, EphA2, WT1, and Survivin. The adjuvant poly-ICLC promotes type 1 polarization of T cell responses against these antigens, enabling effective T cell infiltration into brain tumors and therapeutic benefit 40. Furthermore, patients with WHO grade II astrocytoma or oligoastrocytoma without prior non-surgical treatment may be particularly suitable for this vaccine 40. Finally, the WT1 HLA class I/II peptide cocktail vaccine is safe in patients 38. 1.3 Personalized/Neoantigen Vaccines Representative studies from 2015 to 2025 include 1) IDH1 mutation individualized peptide vaccine 20; 2) personalized peptide vaccine (PPV) 44; 3) Glioma Active Personalized Vaccine Consortium (GAPVAC) 45; 4) autologous formalin-fixed tumor vaccine (AFTV) 46. First, a clinical trial of a personalized neoantigen-targeting vaccine for IDH1-mutant glioma used whole-exome sequencing to customize treatment, which can cover tumor heterogeneity 20. Results showed that the vaccine was well-tolerated and induced persistent T cell responses 20. Moreover, patients with T cell responses to multiple neoantigens had a trend toward longer OS compared to those with no/few responses 20. Additionally, detection of CDR3 sequences may help track CD4+ and CD8+ T cell clones in blood and tumor tissue 20. However, this study was retrospective and non-randomized, with variability in the number of peptides administered and the timing of vaccinations. Therefore, the findings warrant further validation in prospective randomized controlled trials. Second, a PPV trial for HLA-A24-positive recurrent GBM (rGBM) found that SART2-93 positivity was a major factor limiting clinical benefit for all patients 44. In contrast, biomarkers associated with better OS included lower proportions of CD11b+CD14+HLA-DRlow immunosuppressive monocytes, higher proportions of CD4+CD45RA- activated T cells, and medium levels of CCL2, VEGF, IL-6, IL-17, and haptoglobin 44. Third, AFTV is a personalized vaccine derived from formalin-fixed autologous glioblastoma specimens 46. Theoretically, it preserves all tumor antigens, including neoantigens, obviating the need for HLA-based patient selection 46. However, this strategy has two major limitations: the inability to monitor therapeutic response via specific antibody titers, and the dependence of vaccine yield on the amount of resected tumor tissue, precluding flexible dose escalation or prolonged administration. A Phase IIb double-blind randomized trial showed no significant difference in efficacy between the vaccine and placebo groups in the overall population, likely due to the small sample size and the high quality of surgery and TMZ therapy received by the placebo group (median OS 31.5 months) 46. Subgroup analyses suggested that patients with gross total resection or p53-negative tumors may be more sensitive to AFTV. This hypothesis is being evaluated in a completed Phase III trial (UMIN000010602), the results of which have not yet been reported 46. Finally, the GAPVAC vaccine consists of APVAC1 (a pre-manufactured unmutated antigen repertoire) and APVAC2 (targeting neoepitopes) for HLA-A*02:01+ or HLA-A*24:02+ nGBM patients 45. Results showed that APVAC1 induced sustained central memory CD8+ T cell responses, while APVAC2 induced Th1-type CD4+ T cell responses against predicted neoepitopes 45. Furthermore, both vaccines, administered with poly-ICLC and GM-CSF adjuvants, were considered safe and highly immunogenic 45. 1.4 New Preclinical Advances Recent advances focus on new delivery systems/pathways, optimized antigen selection, immune microenvironment modulation, and personalized vaccine design. 1.4.1 New Delivery Systems Nanotechnology offers promising solutions. Optimizing particle size, surface modification, and targeting ligands significantly improves brain enrichment and immune activation 47-50. Representative platforms include lipid nanoparticles (LNPs), polymer nanoparticles, and inorganic nanomaterials. A key LNP-based preclinical study is the CVGBM vaccine. CVGBM is a novel multi-targeted mRNA-LNP vaccine encoding a fusion protein comprising eight tumor-associated antigen-derived epitopes 48. Its platform efficacy was validated in a surrogate melanoma mouse model, and antigen presentation specificity was confirmed in vitro 48. However, immunopeptidomics analysis detected the presentation of only four of the eight encoded epitopes, while the remaining four were not identified 48. Currently, a Phase I trial for HLA-A*02:01+ MGMT unmethylated GBM is ongoing (NCT05938387) 48. Although several mRNA-LNP vaccines have entered clinical trials for infectious diseases and cancer, their clinical translation remains limited by poor targeting precision, potential immunogenicity and toxicity, and the difficulty of developing universal delivery systems 51. Biomimetic nano-DC vaccines show improved biocompatibility, targeting, and immune activation 52. For example, conjugating biomimetic DC membranes with tumor-associated costimulatory molecules or paclitaxel nanoparticles, combined with ICIs, alleviated T cell exhaustion in mouse models, significantly extended survival, potentially overcoming GBM immunotherapy resistance 53, 54. However, the clinical translation of cell membrane-coated nanoparticle technology is hindered by challenges in scalable production and purification, safety assurance, quality control, and analytical method development 52. 1.4.2 Delivery Route Selection and Optimization Vaccine administration route, frequency, and timing significantly impact efficacy and require systematic optimization. 1) The intranasal route is an innovative strategy 49. It bypasses the BBB, delivering vaccine directly to the brain and cervical lymph nodes 49. It reduces systemic exposure and has good patient compliance 49. Sequential targeted sonodynamic nanovaccine (Stars NV) 55, Survivin peptide-CpG nanovaccines (SPOD-NV) 49, and enhanced nanovaccines using biomineralizing virus-like particles 50 are delivered through the nasal cavity, with drug enrichment observed in the brain and lymph nodes 49, 50, 55. 2) Intravenous delivery is an established technique, allows high dosing, and activates immune cells in systemic lymphoid organs 49. Serum IgG titers were higher after intravenous delivery (IV) than after intranasal delivery 49. However, challenges include poor brain/tumor penetration, rapid clearance by liver macrophages, potential cytokine storm, and systemic toxicity risk 49. A sequential dosing strategy for SPOD-NV vaccine used 5 full IV, 5 full intranasal, or 3 intranasal + 2 IV injections, all combined with anti-CTLA-4 antibody 49. Mice with full IV dosing had significantly longer median survival than full intranasal 49. The 3 intranasal + 2 IV group extended survival to 41 days, with a 43% complete response rate 49. This suggests combining intranasal and IV routes may synergize, producing local and systemic immunity, improving efficacy 49. 3) Subcutaneous delivery allows antigen uptake by subcutaneous DCs, which migrate to regional lymph nodes 49. In the SPOD-NV study, subcutaneous + IV sequential dosing was significantly less effective than intranasal + IV, highlighting the key role of intranasal delivery in GBM treatment 49. Intratumoral delivery directly delivers drugs to the tumor, bypassing the BBB 56. It enables local immune activation, remodels the TME, and reduces systemic exposure risk 56. For example, intratumoral pIpC-Dox LNPs significantly increased infiltration and activation of TAMs, DCs, and T cells. Drawbacks include invasive procedure, difficulty covering infiltrative cells, and increased risk with multiple injections 56. 1.4.3 Optimized Antigen Selection Neoantigens are antigens produced by specific mutations in tumor cells 20. They are specific and immunogenic, and do not exist in normal tissues, thus becoming ideal vaccine targets 20. Representative studies include the IDH1-R132H 20-mer peptide vaccine 20 and nanoshield vaccines 57. The (multi-cationic protein (MCP))-muIDH1 nanoshield vaccine (NaV) uses a MCP scaffold, genetically fuses the muIDH1 neoantigen to MCP, and assembles carboxylated PEG into 100 nm spherical nanoparticles 57. This enables lysosomal escape, enhanced cross-presentation, NLRP3 inflammasome activation, and prolonged in vivo retention 57. In tumor-bearing mice, combining NaV with anti-PD1 delayed postoperative recurrence and increased median survival by over 40% 57. TAAs are normal proteins overexpressed or over-presented on tumor cells 48, 49, 58, 59. While less immunogenic than neoantigens, they are shared, potentially benefiting more GBM patients 48, 49, 58, 59. Representative TAAs include 1) Survivin Baculoviral IAP Repeat Containing 5 49; 2) Ephrin type-A receptor 2 49; 3) Plasminogen activator urokinase receptor 48; 4) Transforming growth factor beta 2 (TGFB2) 58. Representative vaccines include 1) SurVaxM 49; 2) CVGBM mRNA vaccine 48; 3) and GVac 58. GVac is a multi-epitope vaccine targeting TGFB2, and its antigen selection and screening have been strongly assisted by computer-aided design 58. Allergenicity, antigenicity, toxicity, epitope selection, molecular docking, dynamics simulation, and immune simulation can be predicted computationally, theoretically enabling rapid screening of low-toxicity, highly immunogenic candidate vaccines with strong target affinity, though experimental validation is needed 58. 1.4.4 Immune Microenvironment Modulation GBM's immunosuppressive TME limits vaccine efficacy 7. New vaccine strategies remodel the TME by reprogramming TAMs, reducing Tregs, and enhancing T cell infiltration 47, 49, 55, 57, 60. However, the observed TAM reprogramming is mediated not by the vaccine itself, but by secondary components such as sonosensitizers 55 or ICIs 61 incorporated into these strategies. The vaccine Star NV contains a sonosensitizer, the Toll-like receptor agonist R848 (which promotes M1 polarization), and CM-β-CD (for macrophage targeting) 55. Mechanically, it uses ultrasound to activate sonosensitizers, generate reactive oxygen species (ROS) in tumor tissues to directly destroy GBM cells, and release R848 to achieve TAM transformation to M1 phenotype 55. In vivo experiments in the glioma mouse model confirmed that this method could kill tumor cells and help to reshape the immunosuppressive microenvironment 55. Clofazimine (CFZ), an anti-leprosy drug, has potent Wnt/β-catenin inhibitory activity 60. Combining DC vaccine with CFZ inhibited the Wnt/β-catenin pathway in cancer stem cells, blocking glioma phenotypic heterogeneity/plasticity and β-catenin-mediated immune escape, thereby reducing immunosuppression 60. Immune checkpoint synergy can also improve the TME. For example, nanoshield vaccine combined with anti-PD-1 improved postoperative relapse-free survival by over 40% in mouse models 57. 2. Immune Checkpoint Inhibitors Immune checkpoint molecules limit autoimmunity by regulating T cell responses to self-antigens 62. However, they also limit attacks on cancer cells 62. ICIs block these molecules, enhancing T cell recognition and killing of cancer cells 62 (Figure 2). Between 1980 and 2000, key immune checkpoints, including CTLA-4 and PD-1 were identified, establishing the foundation for ICIs 63, 64. In 1996, Allison and colleagues first demonstrated that CTLA-4 blockade enhances antitumor immunity in preclinical models 65. PD-1 was cloned by Honjo's team in 1992, and its role as an immune checkpoint was elucidated by 2000 63. Following regulatory approval of the PD-1 inhibitors Nivolumab and Pembrolizumab in 2014, clinical trials in glioma were initiated. The first-in-human Phase I study of Nivolumab in recurrent glioblastoma (CheckMate 143) established its safety and tolerability 66. However, the subsequent Phase III CheckMate 143 trial failed to demonstrate an OS benefit for Nivolumab over Bevacizumab in recurrent glioblastoma, underscoring the limitations of ICI monotherapy in this disease 67. After 2015, ICI trials and combination therapies for glioma increased. According to the literature, ICIs used in trials in the past 10 years include: 1) anti-CTLA-4 (Ipilimumab) 42, 43, 68; 2) anti-PD-1 (Nivolumab, Pembrolizumab) 42, 69; 3) anti-PD-L1 (Atezolizumab) 70. The efficacy of ICI monotherapy is limited, and clinical studies often combine multiple ICIs or other drugs or as adjuvant therapy 71. As shown in the figure, immune checkpoint molecules such as CTLA-4, PD-1, PD-L1, and TIM-3 negatively regulate T cell activation and T cell-mediated tumor cell killing. Immune checkpoint inhibitors block these inhibitory signaling pathways, relieve T cell exhaustion, and subsequently restore the capacity of T cells to recognize and clear tumor cells. 2.1 Clinical Trials and Combination Therapies Representative clinical studies from the past decade include: 1) Ipilimumab + Nivolumab for nGBM (two studies) 42, 43; 2) Atezolizumab + Radiotherapy + TMZ Phase I/II trial (NCT03174197) 70; 3) Evaluation of neoadjuvant Pembrolizumab in rGBM 69; 4) Nivolumab + Ipilimumab before radiotherapy for nGBM Phase I trial 68; 5) Bevacizumab + anti-PD-1 therapy (NCT05502991, NCT05540275) 72, 73. The safety and efficacy of Ipilimumab + Nivolumab for nGBM were demonstrated in the NRG Oncology BN002 (Phase I) trial 42. However, the randomized Phase II/III NRG Oncology BN007 trial for MGMT-unmethylated nGBM found that Ipilimumab + Nivolumab did not improve Progression-Free Survival (PFS) compared to TMZ 43. Subsequently, multiple clinical trials have been completed, and despite the use of various immune checkpoint inhibitor regimens, none have demonstrated superiority of ICI monotherapy or combination therapy over standard treatment 43, 74, 75. Thus, without molecular selection or biomarker guidance, ICIs confer extremely limited survival benefit 43. Nonetheless, the identification of patients with specific genetic features or immune phenotypes sensitive to ICIs remains of great value, and relevant biomarker analyses are currently ongoing 43. Atezolizumab + Radiotherapy + TMZ was well-tolerated in a Phase I/II trial (NCT03174197). Unlike DC vaccine sensitivity, this therapy was more effective in PTEN-mutant, mesenchymal subtype GBM patients, less effective in EGFR-mutant, proneural subtype, and significantly better in MGMT-methylated vs. unmethylated patients 70. Additionally, GBM immune enrichment correlated highly with pre-treatment tumor mesenchymal subtype and patient gut microbiota 70. Moreover, longer-surviving patients had enriched Bacillota phylum bacteria in stool samples 70. A trial of neoadjuvant Pembrolizumab in rGBM showed that treatment activity correlated with shortened/reduced cell cycle and proliferation genes and increased T cell/interferon-related genes 69. This suggests that combined transcription factors that reduce cell cycle/proliferation gene expression or increase T cell/interferon gene expression may improve efficacy 69. Also, ICIs may have inherent resistance due to chronic inflammation in the TME, suggesting combining anti-inflammatory drugs might restore neoadjuvant ICI activity 69. Phase I clinical trials of Nivolumab + Ipilimumab in the treatment of nGBM before radiotherapy found that advanced tumors have high LAG-3 levels, downregulation of PPAR signaling and up-regulation of TGF-β and ERBB signaling, and dysregulation of these pathways is associated with ICI resistance 68. This suggests combining LAG-3 inhibitors, PPAR agonists, and TGF-β/ERBB pathway inhibitors might overcome resistance and improve efficacy 68. Two studies (NCT05502991, NCT05540275) found that Bevacizumab plus anti-PD-1 significantly improved OS in patients with rGBM. Furthermore, tumor in situ fluid circulating tumor DNA technology can monitor therapeutic response and guide therapy 72, 73. Collectively, these findings are promising, suggesting pre-treatment screening for patients likely to respond to Bevacizumab + anti-PD-1, refining the target population and reducing medical risk and cost. 2.2 New Preclinical Advances ICIs show efficacy in many solid tumors but face challenges in glioma. Core challenges include the BBB limiting brain delivery, highly immunosuppressive TME, heterogeneity, scarce tumor-infiltrating lymphocytes (TILs), and dominant TAMs 11. Recent research focuses on novel delivery systems, combination therapies, TME remodeling, and biomarker identification/patient stratification to improve ICI efficacy. 2.2.1 Novel Delivery Strategies The BBB is a major obstacle that limits the access of ICIs to GBM tumor sites 5. To overcome this, researchers utilized low-intensity pulsed ultrasound combined with intravenous microbubbles to achieve transient, reversible, and localized opening of the BBB, thereby enhancing the concentration of aPD-1 in the sonicated regions 76. Additionally, various delivery systems have been developed, including extracellular vesicles (EV) 77, redox-responsive micelles 78, lipid nanocarriers 79, and nanovesicles 61. These systems successfully crossed the BBB, achieving efficient brain delivery and tumor-specific accumulation of ICIs, improving treatment safety 61, 78, 79. However, these delivery systems are still in the research phase and face translational challenges. For instance, although extracellular vesicles possess good biocompatibility and hold promise as therapeutic carriers, they remain limited by heterogeneity, low yields, inefficient drug loading, and rapid clearance by the mononuclear phagocyte system 80. No EV-based therapy has yet received regulatory approval. Redox-responsive micelles face complex preparation and clinical translation challenges, including amphiphile amount 81, ligand density 82, linkage type 83, polymerization degree 84, reaction conditions 85, and branching degree. These factors collectively influence micelle formation, stability, and scalability. The clinical translation of nanocarriers is hindered by complex physiological barriers, off-target effects of targeting ligands, inherent issues of stability and toxicity, as well as challenges in scalable manufacturing and batch-to-batch consistency 86. 2.2.2 Combination Therapy Strategies Single ICI efficacy is limited 71. Researchers explored combinations with chemotherapy 79, radiotherapy 87, photodynamic therapy 88, CAR-NK therapy 89, etc., showing synergy. Moreover, dual/multiple immune checkpoint blockade (ICB) (e.g., PD-1/CTLA-4 87, αVβ8/PD-1 90) enhances immune responses 87. Furthermore, combining ICIs with epigenetic regulators such as Zinc Finger Protein 638 (ZNF638) inhibition 91 and CpG-STAT3ASO 81 has demonstrated potential to restore immune sensitivity in preclinical settings 91. 2.2.3 Tumor Microenvironment Remodeling The GBM TME is highly immunosuppressive, dominated by TAMs and MDSCs 61. Improving this state is key to enhancing ICI efficacy. Strategies include 1) TAM reprogramming (M2 to M1) via inhibiting difluoromethylornithine (DFMO) 61 or Actin Related Protein 2/3 Complex Subunit 1B (ARPC1B) 92; 2) inhibiting immunosuppressive molecules (Nitric Oxide (NO), ROS, IL-10) to protect T cell function 93; 3) modulating B cell function to overcome Transforming Growth Factor β (TGFβ)-mediated immunosuppression 90. 2.2.4 Biomarker Identification Identifying predictive biomarkers is crucial for patient stratification and personalized therapy. Potential biomarkers explored include 1) molecular markers (PD-L1, ZNF638, ARPC1B) 91, 92; 2) cellular markers (TAMs, B cells, NKT cells) 89, 90, 92; 3) immune microenvironment features (double-stranded RNA (dsRNA), IFN signaling) 91; 4) spatial transcriptomics revealing cell interaction networks 90. 3. Adoptive Immune Cell Therapy Adoptive immune cell therapy is a treatment method in which immune cells with anti-tumor activity are screened, expanded, activated, or genetically engineered in vitro, and then infused back into the patient's body to help the patient's immune system recognize and attack cancer cells 94. During the 1980s to 2000s, lymphokine-activated killer (LAK) cells and tumor-infiltrating lymphocytes (TILs) were explored clinically for glioma treatment 95. Early studies by Jacobs et al. demonstrated the feasibility of intratumoral LAK cell injection 96; however, clinical efficacy was limited and responses were transient, with subsequent trials confirming these therapeutic constraints 97, 98. TIL-based approaches faced challenges including low TIL abundance in glioma and functional exhaustion within the immunosuppressive TME 99. However, a recent study using IL-2/IL-15/IL-21-expanded TILs reported complete tumor regression in a glioblastoma patient, suggesting renewed potential for this strategy 100. Around 2000s-mid 2010s, engineered T cells emerged. CAR-T therapy genetically modifies patient T cells to express CARs targeting specific TAAs, enabling tumor cell recognition and killing 101. T cell receptor-engineered T cells (TCR-T) differ by introducing optimized TCR genes to recognize intracellular antigens presented by major histocompatibility complex (MHC) molecules 102. From late 2010s-present, new generation therapies arose, including multi-target CAR-T, logic-gated CAR, armored CAR-T, CAR-NK, CAR-M, neoantigen-targeting CAR-T/TCR-T, combination therapies, and gene-edited universal products 103. According to the data collected, most of the literature considers adoptive cell therapy to have a good safety profile, with the most serious adverse event usually grade 3 fever 104, 105. However, recent literature has reported that this therapy is associated with the occurrence of cognitive impairment in patients, suggesting that its safety needs to be further reviewed. In glioma patients specifically, neurological events occur in approximately 41% of cases across early-phase trials 106. Mechanistic studies in mouse models have demonstrated that CAR-T therapy can disrupt hippocampal neurogenesis and oligodendroglial homeostasis, leading to cognitive deficits 107, and similar reactive microglial states have been confirmed in human brain tissue from patients who received CAR-T therapy for brain tumors 107. Furthermore, authoritative guidelines from the EBMT and reviews in Nature Reviews Neurology have identified movement and neurocognitive toxicity as emerging concerns across both hematologic and solid malignancies, including CNS tumors 108, 109. These converging lines of evidence underscore the need for further scrutiny of the long-term safety of adoptive cell therapy. 3.1 Clinical Trials Since 2015, CAR-T cell therapies for glioma have entered clinical trials. Targets in CAR-T trials primarily include 1) TAAs: EGFRvIII 13, 14, 110; 2) Interleukin-13 Receptor subunit alpha-2 (IL13Rα2) 111, 112; 3) Disialoganglioside GD2 (GD2) 113; 4) Matrix Metalloproteinase-2 (MMP-2) 114. 3.1.1 EGFRvIII Target EGFRvIII results from an in-frame deletion of exons 2-7 and mutation at the exon 1/8 junction 14. It is a tumor-specific, oncogenic, immunogenic epitope 14. Expressed in ~25-30% of nGBM cases, its expression is spatiotemporally heterogeneous, can re-emerge, and varies greatly within a tumor 14. The efficacy of anti-EGFRvIII CAR-T cell therapy alone has proven to be limited 14. After treatment, the upregulation of inhibitory molecules, especially PD-L1, in the TME is accompanied by increased infiltration of Tregs, suggesting that the combination of ICIs may improve the efficacy 14. However, a Phase I trial of repeated anti-EGFRvIII CAR-T infusions + Pembrolizumab was ineffective 13. While safe and tolerable, disease control was limited, with most patients progressing within months 13. Reasons included limited peripheral CAR-T cell expansion/persistence, difficulty infiltrating the tumor, EGFRvIII heterogeneity/antigen loss, and increased Tregs/TAMs post-treatment, indicating Pembrolizumab alone couldn't reverse the highly suppressive TME 13. A Phase I trial of bivalent CAR-T cells targeting EGFR epitope 806 and IL-13Rα2 via intracerebroventricular delivery for EGFR-amplified rGBM showed safety and feasibility 110. 62% had tumor shrinkage, a few had durable control 110. The trial enrolled EGFR-amplified patients but not based on IL-13Rα2 expression, potentially affecting efficacy 110. In five re-infused patients, CAR-T expansion/persistence was worse than the first infusion, with short PFS, suggesting immune clearance or resistance mechanisms 110. 3.1.2 IL13Rα2 Target IL13Rα2 is a high-affinity IL-13 decoy receptor, negatively regulating IL-13 signaling 111. It is expressed in ≥ 75% of GBM patients, overexpressed in > 50%, correlating with shorter survival 111. It is expressed on GSCs, differentiated tumor cells, and TAMs/MDSCs, but not significantly in normal brain 111. Its expression correlates with the mesenchymal subtype 111. This first-in-human trial evaluated an off-the-shelf, steroid-resistant allogeneic CAR-T therapy targeting IL13Rα2 in rGBM 112. By using healthy donor T cells, it overcomes the manufacturing time, cost, and feasibility limitations of autologous CAR-T 112. This therapy uses zinc-finger nucleases to knock out the glucocorticoid-receptor gene in T cells so that they remain functional in the presence of dexamethasone 112. In terms of clinical outcomes, treatment was well tolerated, with local tumor necrosis or shrinkage in most patients 112. However, antitumor activity was transient, and all patients eventually experienced disease progression 112. The lack of durability of CAR-T cells is a major factor limiting efficacy, in part because of the use of first-generation CAR constructs (Figure 3) containing only the CD3ζ signaling domain, which lack the costimulatory domain and result in limited expansion and survival 112. In addition, because HLA matching was not performed, one of the six patients developed anti-CAR antibodies, suggesting the presence of host immune rejection 112. Future improvements include the use of second-generation or higher-generation CAR designs with costimulatory domains, optimized gene editing (e.g., knockout of the TCR to avoid graft-versus-host disease), and the exploration of manufacturing processes that preserve the stem-like properties of T cells and prolong their survival 112. This figure shows the engineered structure of the first to fourth-generation CAR-T. The core structure of first-generation CAR-T is an extracellular single-chain variable fragment (scFv) plus a single intracellular signaling domain (usually CD3ζ). The core structure of the second-generation CAR-T is extracellular scFv + a costimulatory signal domain + CD3ζ. The core structure of the third-generation CAR-T is extracellular scFv + two or more costimulatory domains + CD3ζ. The core structure of the fourth-generation CAR-T integrates additional genetic elements on the basis of the second-generation CAR, so that it can secrete specific cytokines or express additional receptors. The structures of the fourth-generation CAR-T are diverse, and only the cytokine secreting fourth-generation CAR-T is shown in the figure. 3.1.3 GD2 Target GD2 is a disialoganglioside, and its expression is highly specific 113. In normal tissues, it is mainly distributed in the neuronal cell membrane 113. However, it is highly expressed in a variety of malignant tumors 113. In GBM, GD2 is enriched on tumor cells and specifically overexpressed on GSCs. ~80% of diffuse intrinsic pontine gliomas express GD2 113. A trial of GD2-specific 4SCAR-T cells for GD2-positive GBM used a fourth-generation CAR design with an inducible suicide gene 113. The results showed that both intravenous and intracavitary injections were safe and tolerable 113. Median survival from infusion was 10 months 113. Regarding CAR-T cell kinetics, they expanded in peripheral blood for 1-3 weeks, then persisted at low frequency 113. Importantly, post-infusion tumor resection showed GD2 antigen loss (mechanism unknown), suggesting need for multi-target therapy 113. Additionally, after treatment, TAMs increased and tumor-infiltrating NK cells decreased, suggesting that the immunosuppressive microenvironment still needs to be improved 113. 3.1.4 MMP-2 Target Metalloproteinase MMP-2 is widely expressed in GBM cells, and its main role is to degrade collagen Ⅳ, the main component of the basement membrane of the extracellular matrix 114. In gliomas, overexpression of MMP-2 is closely related to the malignant behavior of tumors 114. It accelerates the malignant progression of glioma by destroying the BBB, inhibiting immune cell infiltration, and mediating immune evasion 114. The CLTX-CAR T-cell study is the first to exploit the ability of CLTX, a peptide derived from scorpion venom, to treat patients with recurrent, MMP-2-expressing GBM through the mechanism of broad MMP-2 binding to glioma cells 114. Pharmacokinetically, CAR T cells were persistently detected in tumor cavity fluid (TCF) and at low levels peripherally, proving intratumoral delivery allows transport 114. In terms of safety and efficacy, intratumoral infusion was tolerable with manageable adverse events (AEs), achieving a disease control rate of 75%, although the small sample size precluded assessment of survival benefit 114. However, several limitations were identified. MMP-2 expression on tumor vasculature, fibroblasts, and myeloid cells poses a potential risk 114. Furthermore, the inhibitory TME may limit efficacy, as stromal and myeloid components may drive T cell exhaustion, suggesting that combination strategies targeting the TME could improve outcomes 114. Future directions include modifying peptide ligand/CAR structure for better targeting, enhancing T cell persistence/proliferation, and combining therapies 114. 3.2 New Preclinical Advances CAR-T therapy has demonstrated significant clinical efficacy in hematologic malignancies, but its application in glioma remains limited. Major challenges: 1) High antigen heterogeneity; 2) Immunosuppressive TME factors (TGF-β, PD-L1); 3) CAR-T cell exhaustion from chronic antigen exposure; 4) BBB limiting migration; 5) Antigen loss/downregulation post-treatment 94. Recent innovations to overcome these include: 1) Universal targeting platforms (e.g., monomeric streptavidin 2-based Chimeric Antigen Receptor T cells (mSA2 CAR-T)) 115; 2) Dual/multi-target CAR designs 116; 3) Armored CAR-T cells 116; 4) Sonogenetically controlled CAR-T 117; 5) Combination therapies (e.g., with oncolytic viruses, small molecules) 118, 119; 6) New targets (e.g., Glycoprotein A Repetitions Predominant, Human Leukocyte Antigen-E) 120, 121. Conversely, several other CAR-T strategies have demonstrated limited efficacy or encountered significant challenges in preclinical glioma models. These include permanently expressed CAR constructs associated with off-tumor toxicity 122, and B7-H3-targeted CAR-T cells that have yet to overcome intratumoral heterogeneity 123. Promising preclinical studies are listed below. 3.2.1 Universal CAR-T Platform: mSA2 System The mSA2 CAR-T platform is a revolutionary universal strategy 115. It uses mSA2 as the CAR's antigen-binding domain, mediated by biotinylated antibodies 115. Its advantage is that the targeting specificity of CAR-T cells can be rapidly changed by simple replacement of biotinylated antibodies, without redesign and manufacture of new CAR constructs, so that rapid target switching, multi-target combination, overcoming antigen escape, and reducing manufacturing costs can be achieved 115. Researchers developed biotinylated antibodies for GBM markers CD276 (B7-H3), EPHA2, CD70, IL13Rα2 115. In vitro, mSA2 CAR-T cells specifically recognized primary GBM cell lines antigen-dependently, simultaneously targeting heterogeneous subpopulations, clearing GBM cells within 20 hours via live imaging 115. In immunodeficient mouse orthotopic GBM models, intracranial injection of mSA2 CAR-T pre-loaded with anti-CD70 or anti-CD276 antibodies significantly suppressed tumor growth, induced apoptosis, and extended survival 115. This therapy provides a good safety foundation for clinical translation. In a mouse model of GBM, mSA2 CAR-T cells did not interact with brain endogenous biotin, thereby reducing the risk of off-target cross-reactivity 115. 3.2.2 Checkpoint Antibody Receptor-modified (ARMed) CAR-T Cells The BBB blocks antibody entry into the CNS, a major reason for limited efficacy in anti-PD-1 + CAR-T trials 94. To overcome this, researchers developed fourth-generation ARMed CAR-T, combining EGFRvIII-targeting CAR with anti-PD-1 minibody secretion 118. The advantage of this therapy is that ARMed CAR-T is able to cross the BBB and secrete microantibodies at the tumor site, overcoming the BBB limitation of systemic delivery 118. Mice treated with EGFRvIII ARMed CAR-T had significantly smaller brain tumors versus controls, with a median OS of 34 days and 10% surviving to day 50 (study end) 118. Additionally, no dose-limiting toxicities were observed, and human T cells remained detectable in peripheral blood at day 29 post-infusion, comparable to CAR T cells alone 118. 3.2.3 NR4A3 Deficiency + FOS Overexpression Synergy Overcomes CAR-T Exhaustion NR4A family genes (NR4A1, NR4A2, NR4A3) are upregulated in TILs, especially terminally exhausted T cells 124. FOS expression is highest in CD8⁺ naive populations and is co-expressed with stemness genes including TCF7, CCR7, IL7R, and LEF1 124. Based on this, researchers knocked down NR4A3 and overexpressed FOS (FOS OE/NR4A3 KD) in CAR-T cells 124. Modified CAR-T cells showed long-term tumor suppression, significantly extended mouse survival, and widespread tumor infiltration 124. This provides a novel genetic modification strategy for solid tumor CAR-T, enhancing stemness and anti-exhaustion capacity by targeting NR4A3 and FOS, offering a preclinical basis for delaying progression 124. 3.2.4 CD44/CD133 Dual-Target IL7Rα Armored CAR-T Cells IL7Rα is key for T cell survival/proliferation 116. Armoring CAR-T with ΔIL7Rα intracellular domain enhances tumor-killing 116. Targeting both CD44 and CD133, classic GSC biomarkers, aims to clear GSCs comprehensively, preventing antigen escape 116. Since CD44 is also on T cells, to prevent CAR-T fratricide, researchers modified the CAR structure; Tanζ-T28 became the ideal base, successfully constructing Tanζ-T28 CAR-T cells 116. Evaluating Tanζ-T28 CAR-T in glioblastoma organoid (GBO) models showed decreasing GBO volume, successful CAR-T infiltration, memory T cell enrichment in brain tissue, reduced checkpoint marker expression, and upregulated effector function genes 116. In addition, a clinical trial (NCT05577091) is ongoing 116. In recent years, with the continuous advancement of innovative research on CAR-T therapy, the field has encountered challenges in clinical translation. Overall, the clinical translation of CAR-T therapy is hindered by high costs, complex regulatory compliance, substantial quality variability in manufacturing processes, and a widespread shortage of infrastructure and specialized personnel 125. However, universal platforms such as mSA2 CAR-T may help reduce costs and accelerate the clinical translation process 115. 4. Oncolytic Virus Therapy Oncolytic virus therapy uses genetically engineered, weakly pathogenic viruses to selectively induce ICD in cancer and stromal cells 7 (Figure 4). It works via multiple mechanisms: direct oncolysis, immune activation, TME remodeling, and vascular disruption 126. In the early 21st century, clinical trials initially verified the safety and feasibility of herpes simplex virus type 1 (HSV-1) G207 127, reovirus adenovirus 128, NDV-HUJ oncolytic virus 129 and other modified viruses. A Phase I trial of G207 combined with radiotherapy demonstrated acceptable safety and radiographic responses in rGBM patients, with six of nine patients achieving stable disease or partial response 127. By 2010, the safety and feasibility of DNX-2401 (Delta-24-RGD) and recombinant poliovirus (PVSRIPO) were established 130, 131. A pivotal Phase I study of intratumoral DNX-2401 in rGBM reported a 20% 3-year survival rate, including one patient who achieved complete remission lasting three years 130. Notably, post-treatment tumor analysis revealed CD8+ T-cell infiltration, providing clinical evidence of "cold"-to-"hot" tumor conversion 130. Concurrently, a Phase I trial of PVSRIPO in rGBM demonstrated a 21% 3-year survival rate versus 4% in historical controls, highlighting its potential to induce durable antitumor immunity 131. Since the late 2010s, emerging strategies have focused on combination approaches with ICIs, chemoradiotherapy, and the design of next-generation oncolytic viruses 126, 132-135. The figure illustrates the oncolytic effect mediated by oncolytic viruses following intratumoral injection. This effect promotes the release of a large amount of antigens from the tumor, including tumor-associated antigens (TAAs), damage-associated molecular patterns (DAMPs), pathogen-associated molecular patterns (PAMPs), and viral antigens, thereby facilitating the activation of immune cells and their capacity to kill tumor cells. 4.1 Clinical Trials Oncolytic virus platforms in trials over the past decade include: 1) measles virus 136; 2) adenovirus 132, 137, 138; 3) herpes virus 139, 140. 4.1.1 Oncolytic Measles Virus Measles virus is a single-stranded, negative-sense RNA virus. It naturally targets cells via receptors (CD46, Nectin-4), which are often overexpressed on cancer cells 136. Harnessing this property, oncolytic measles viruses are engineered for enhanced efficacy/safety, overcoming natural limits, and allowing replication monitoring 136. A prime example is the Measles Virus-based Carcinoembryonic Antigen-expressing Oncolytic Virus (MV-CEA), derived from the safe Edmonston vaccine strain 136. It causes minimal normal tissue damage, uses CD46 for selective infection, has a strong bystander effect, and expresses soluble CEA for monitoring replication 136. In clinical testing, this therapy was well tolerated and safe, with a median OS of 11.6 months and a 12-month OS rate of 45.5% in patients with refractory rGBM, which was superior to historical controls 136. Notably, pre-existing immunity did not prevent virus replication within the tumor 136. Instead, tumor-intrinsic interferon-stimulated gene (ISG) levels emerged as the key determinant of viral replication 136. Specifically, high ISG signaling inhibited viral replication, thereby limiting therapeutic benefit 136. Therefore, this therapy may be best suited for low ISG patients; for high ISG, combining with JAK and STAT inhibitors could block interferon response and enhance replication 136. Additionally, CD8+ T cell infiltration observed post-treatment waned in subsequent samples from some patients, suggesting single/limited treatments may not sustain long-term immune activation 136. Future directions include developing measles strains expressing H. pylori neutrophil-activating protein for enhanced immunostimulation or combining measles virus with ICIs 136. 4.1.2 Oncolytic Adenovirus Adenovirus is a non-enveloped dsDNA virus. Genetic modification enables tumor-selective replication 141. Several adenoviral platforms have been clinically assessed for GBM, including Ad-TD-nsIL12 137, DNX-2401 132, and NSC-CRAd-S-pk7 138. Among these, Ad-TD-nsIL12 is a third-generation adenovirus with multiple genetic modifications 137. Specifically, deletion of E1ACR2 and E3gp19K enabled tumor-specific replication, whereas deletion of E1B19K enhanced antitumor immunity, whereas E1B55K and E3B overcame previous limitations of impaired viral replication capacity and M2-type macrophage infiltration 137. Furthermore, the IL-12 signal peptide was modified to create a non-secreting form (nsIL12), enabling sustained, low-level IL-12 expression within the TME while avoiding systemic IL-12 toxicity such as lethal inflammatory syndrome 137. Clinically, a Phase I trial demonstrated that Ad-TD-nsIL12 was tolerable at 1×10¹⁰ vp, with dose-limiting toxicity occurring at 5×10¹⁰ vp. Most adverse events were grade 1-2 and resolved within 48 hours 137. Median OS was 5.1 months (range: 3.1-21.2 months) 137. Notably, one patient achieved complete response and one achieved partial response; both responders had IDH1 mutation and MGMT expression, suggesting that exploring IDH mutant sensitivity to oncolytic viruses warrants further investigation 137. Mechanistically, late TME analysis revealed dominance by M2 macrophages (CD163⁺), which may promote tumor progression 137. Therefore, combining Ad-TD-nsIL12 with M2 inhibitors or reprogrammers to promote M1 polarization could enhance therapeutic efficacy 137. Another platform, DNX-2401 (tasadenoturev/Delta-24-RGD), is a conditionally replicating adenovirus featuring a 24-bp deletion in E1A for selective replication and an RGD peptide insertion in the fiber knob to improve GBM cell infectivity via integrin binding 132. In a multicenter Phase I/II trial (NCT02798406), DNX-2401 combined with Pembrolizumab was safe and tolerable 132. Median survival in GBM was 12.5 months, with an objective response rate (ORR) of 10.4%, which was not statistically different from the preset 5% control rate 132. Importantly, post-treatment analysis revealed upregulation of several immune checkpoints (TIGIT, LAG3, B7-H3), suggesting that combining multiple ICIs post-treatment might improve efficacy 132. Furthermore, an exploratory analysis suggested that patients with a TMEmedium subtype characterized by moderate immune infiltration, medium PD-1 expression, and absence of multiple immunosuppressive factors may be more likely to derive clinical benefit from this approach 132. 4.1.3 Oncolytic Herpes Virus HSV is a neurotropic dsDNA virus with a large genome (~150kb) 142. Over the past decade, G47Δ has emerged as a key oncolytic HSV-1 variant 139, 140. It is a third-generation, triple-mutated virus constructed by deleting the α47 gene and overlapping the US11 promoter from its parental virus G207, which enhances tumor-specific replication and immunogenicity 139. In clinical studies, previous Phase I/II trials confirmed the safety of G47Δ in progressive GBM, with patients achieving a median survival of 7.3 months (range: 1-14 months) 140. A subsequent Phase II trial for residual or recurrent GBM showed more impressive results: median OS from G47Δ start was 20.2 months, with a 1-year survival rate of 84.2%—far exceeding the 15% historical control 139. Mechanistically, biopsy after treatment revealed a sustained increase in CD4⁺ and CD8⁺ T cells without an increase in Foxp3⁺ Tregs, suggesting that G47Δ can effectively improve the immunosuppressive microenvironment and promote immune infiltration 139. Based on these findings, G47Δ was approved in Japan in June 2021, becoming the country's first oncolytic virus product 139. However, distant new lesions, CSF dissemination, and local recurrence around the target were common, indicating that local therapy struggles to control systemic or diffuse progression 139. Additionally, imaging proved inadequate for accurately evaluating efficacy, underscoring the critical need for novel assessment methods 139. 4.2 New Preclinical Advances Oncolytic virus therapy shows great potential but faces challenges: poor targeting, high toxicity, poor drug control, systemic delivery limited by BBB, weak virus infiltration, host immune clearance, and immunosuppressive TME limiting immune activation. Recent strategies in preclinical research have made important progress. Innovations in virus platforms, delivery strategies, combination therapies, and metabolic regulation are summarized below. 4.2.1 New Oncolytic Virus Platforms New platforms focus on improving targeting, reducing toxicity, and enabling drug control. 1) Zika virus (ZIKV), 2) infectious bursal disease virus (IBDV), 3) human cytomegalovirus (HCMV), and 4) myxoma virus (MYXV) platforms provide strong support 126, 133, 143-145. 1) ZIKV shows selective targeting of GSCs, with natural neuroprogenitor tropism and SOX2⁺ cell preference 126, 143. 2) IBDV is a dsRNA non-human virus, avoiding pre-existing immunity; it infects tumor and myeloid cells 133. 3) HCMV based on AD169 strain: deleted UL1-UL20/UL/b regions to reduce virulence; restored pentamer complex (gH/gL/pUL128-131) function to enhance tropism; integrated paramyxovirus glycoprotein H/F complex for EGFR targeting; added tet-off system for drug-controlled IL-12 expression 144. 4) MYXV is an enveloped dsDNA poxvirus. A MYXV construct lacking the M011L gene (vMyx-M011L-KO/EGFP) induced apoptosis in infected brain tumor-initiating cells 145. 4.2.2 Innovative Delivery Strategies Cell carrier-mediated delivery, surface modification for BBB penetration, and combination with photodynamic therapy (PDT) improve virus delivery to tumors 145-148. 1) Macrophage carriers protect virus from neutralizing antibodies, effectively infiltrate glioma tissue, and have strong tumor tropism 146. 2) Adipose-derived stem cell carriers have low immunogenicity (allowing allogeneic use), strong tumor tropism, protect virus from antibodies/T cells, and maintain migration ability post-infection 145. 3) Cholesterol-modified adenovirus (OA@Cho) uses LDL receptor-mediated transcytosis across BBB for precise glioma targeting/delivery 147. 4) PDT-generated ROS damage cancer cells, increase tumor vascular permeability, induce ICD, thereby increasing virus susceptibility, promoting virus entry, and enhancing anti-tumor immunity 148. 4.2.3 Combination Therapy Strategies Combining oncolytic viruses with chemotherapy, ICIs, metabolic inhibitors, PDT, etc., shows synergistic potential for improved efficacy 133-135. For example, when IBDV is combined with TMZ, IBDV can enhance the cytotoxic activity of TMZ, and TMZ not only does not impair the susceptibility of GBM cells to IBDV, but can enhance the ability of viral replication and promote viral release 133. Triple therapy (a bromodomain-containing protein 9 (BRD9) inhibitor + oHSV + anti-PD-1/CTLA-4) showed potent anti-tumor effect, nearly eradicating tumors; long-term survivors showed protective immunity upon re-challenge 134. 4.2.4 Metabolic Regulation Intervention Metabolic reprogramming and targeting metabolic pathways can reduce virus-induced lipid peroxidation 135. For example, combining oHSV with an IDH R132H inhibitor restored the sensitivity of IDH-mutant glioma to oHSV, inducing ferroptosis and anti-tumor immunity 135. Mechanistically, this combination blocked glycolysis, cleared ROS, inhibited glutamine metabolism, and suppressed Protein Kinase C in immunocompetent mouse IDH-mutant tumor models, significantly improving virus therapy efficacy 135. However, the use of metabolic reprogramming inhibitors must ensure selective toxicity to tumor cells, thereby preventing metabolic disruption in normal cells that could render them susceptible to oncolytic virus infection 149. 4.3 Biomarkers and Patient Stratification Analysis of GBM trial (NCT03152318) data revealed several predictive biomarkers: 1) Low BRD9 mRNA expression correlated with significantly longer PFS 134. 2) METTL14-low gliomas are more sensitive to oHSV-1 150. 3) Wild-type IDH gliomas are more sensitive to oHSV-1 134. 4) High PD-L1/PD-1 expression suggests potential benefit from combination ICB 135. 5) High glycolysis patients may be sensitive to metabolic reprogramming. In these cases, consider combining glycolysis inhibitors 135. 6) Patients with high oxidative phosphorylation may benefit from oHSV monotherapy 135. 7) Immune "hot" tumors with rich T cell infiltration suit viral immunotherapy and may benefit from ICIs 134. 8) Immune "cold" tumors with scarce T cell infiltration (especially GBM) need virus-induced immune activation. For such patients, consider higher doses, multiple administrations, or combining immunomodulators 134. In recent years, advancements in oncolytic virotherapy have provided new strategies for the treatment of glioma. However, the translation of these therapies from research to clinical application remains hindered by multiple challenges, including safety risks and heterogeneity associated with animal-derived components, difficulties in maintaining viral activity and stability, and the urgent need to upgrade purification technologies toward large-scale, automated, and intelligent manufacturing 151. 5. Common Challenges Shared across Different Immunotherapies Although the immunotherapeutic strategies described above have shown promise in preclinical models, they have largely failed to translate into durable survival benefit in clinical trials (Table 1). This suggests common barriers across different therapeutic modalities. This section systematically analyzes the core mechanisms underlying these failures, including 1) immunosuppression in the TME 152, 153, 2) tumor antigen heterogeneity and escape 13, 154, 3) blood-brain barrier limitations for drug delivery 155, 156, and 4) adaptive resistance under therapeutic pressure 153, 157. Table 1Summary of key clinical trials of immunotherapy for gliomaRepresentative Agent/StrategyTrial PhaseEnrollment (n)Patient PopulationMedian OS(months)Median PFS(months)Key Efficacy/Survival EndpointKey Toxicities/Safety ProfileReferenceCancer VaccinesWT1-mRNA/DCPhase I/II40 (13 GBM)Advanced solid tumors (incl. GBM)GBM: 43.7Not reported• DCR: 69.2% (T1/T2)• ORR: 23.1% (T2)• Type 1 T-cell responses linked to improved OS• Well tolerated• Local injection site reactions only• No systemic toxicity 21 AV-GBM-1Phase II60 (57 treated)nGBM (pre-RT/TMZ)16.010.7• PFS 57% longer vs historical controls• 3-yr PFS: 12.3%• 8 doses correlated with OS (HR 0.09, p <0.0001)• Well tolerated• Local injection site reactions only• No treatment discontinuations 23 ICT-107Phase II (RCT, DB)124 (81 vaccine, 43 control)nGBM (HLA-A1⁺/A2⁺, post-RT/TMZ)17.0 (vaccine) vs 15.0 (control); p=0.58 (NS)11.2 (vaccine) vs 9.0 (control); HR 0.57, p=0.011• PFS significantly improved• HLA-A2 subgroup: greater immune benefit• Methylated HLA-A1: OS 47.6 vs 25.8 mo (p = 0.049)• Well tolerated; no difference in AEs vs control• Most common: fatigue, convulsions, nausea 25 GSC-DCVPhase II (RCT, DB)43 (22 vaccine, 21 placebo)nGBM/rGBM, post-resection (≥95%)13.7 (DCV) vs 10.7 (placebo); p=0.05Multivariate HR 2.5, p=0.027.7 (DCV) vs 6.9 (placebo); p=0.75• IDH1ʷᵗTERTᴹᵀ subgroup: OS (p<0.01), PFS (p=0.03)• Low B7-H4 subgroup: OS (p=0.02)• Well tolerated• Mild fever (1 pt), injection site erythema (1 pt) 26 pp65-DC + DI-TMZ + GM-CSFPhase I11nGBM (post-resection >90%, post-RT/TMZ)41.1 (vs 19.2 historical, p=0.0001)25.3 (vs 8.0 historical, p=0.0001)• 4/11 pts PFS at 59-64 mo• 100% exceeded RPA predicted survival (median gain 30 mo)• pp65 immune response correlated with OS>40 mo (p=0.031)• Well tolerated• 1 Grade 3 vaccine-related immunologic reaction (GM-CSF sensitization) 29 CMV-ATCT-DCPhase I/II (RCT, single-blind)22 enrolled, 15 completed (8 DC, 7 saline)nGBM (CMV-seropositive, KPS ≥80, post-RT/TMZ)Not reported (underpowered)Not reported (underpowered)• Primary safety endpoint met• Polyfunctional T cells: significant increase in CMV-ATCT-DC vs saline (p=0.0078)• Polyfunctional T cells correlated with OS (R=0.74, p=0.037)• Well tolerated• No SAEs related to treatment• AEs consistent with standard of care 27 Td toxoid + CMV pp65-DCPhase I/II (RCT, blinded)12 (6 Td, 6 DC)nGBM (CMV-seropositive, KPS ≥80, post-RT/TMZ)Td: not reached (>36.6)DC: 18.5Td: not reached (>36.6)DC: 10.8• DC migration: significant increase in Td vs DC (p=0.049)• Td significantly improved PFS (p=0.013) and OS (p=0.013)• DC migration correlated with survival• Well tolerated• No vaccine- or Td-related AEs 28 Rindopepimut (ACT III)Phase II65nGBM (EGFRvIII+, post-GTR, post-CRT)21.8 (from study entry)9.2 (from study entry)• PFS at 5.5 mo: 66% (primary endpoint met, p=0.017)• 36-mo OS: 26%• 85% ≥4x Ab titer• 67% target loss in evaluable samples• Grades 1-2 injection site reactions (nearly all)• Grade 3/4 events rare; no fatal AEs 32 Rindopepimut (ACT IV)Phase III745 (405 MRD)nGBM (EGFRvIII+, MRD <2 cm² post-CRT)20.1 (vaccine) vs 20.0 (control); HR 1.01, p=0.93Not reported• Primary endpoint not met (futility stopped)• 2-yr OS (SRD): 30% vs 19% (exploratory, p=0.029)• Grade 3-4: thrombocytopenia (9% vs 6%), fatigue (2% vs 5%), brain edema (2% vs 3%)• Injection site reactions (80% vs 41%, mostly grade 1-2) 31 IDH1-vacPhase I33 enrolled, 32 vaccinatednGBM WHO grade 3/4 IDH1(R132H)+, post-RT/TMZ3-yr OS rate: 0.843-yr PFS rate: 0.63• Immune response rate: 93.3% (across multiple HLA alleles)• PsPD rate: 37.5% (linked to immune response)• MSS correlated with antigen presentation• No RLT/DLT; AEs grade 1 only• Most common: injection site induration (65.6%), erythema (46.9%) 33 SurVaxM (Phase I)Phase I9 enrolled, 8 evaluablerGBM/AA (survivin⁺, HLA-A*02/03/24)86.6 weeks (~21.7) (from study entry)17.6 weeks (~4.4) (from study entry)• Immune response: 75% (cellular & humoral)• 3 pts PR/SD >6 mo• 7/9 pts survived >12 mo• 1 pt CR, alive NED at 174 weeks• No SAEs related to vaccine• Mostly grade 1: injection site reactions, fatigue, myalgia• Grade 3 seizure (unrelated) 41 SurVaxM + adj TMZ (Phase IIa)Phase IIa64 enrolled, 63 evaluablenGBM (survivin⁺, HLA-A*02/03/11/24, post-RT/TMZ)25.9 (from first vaccine)28.4 (from diagnosis)11.4 (from first vaccine)14.4 (from diagnosis)• PFS-6 (from diagnosis): 95.2% (primary endpoint met, p<0.0001 vs 54% historical)• 36-mo OS: 41.4%• Ab titer >30,000 correlated with OS (HR 0.41, p = 0.015)• Methylated MGMT: mOS 41.4 mo; Unmethylated: 16.5 mo• No SAEs related to SurVaxM• Most common: grade 1 injection site reactions (37.5%)• AEs attributable to TMZ (leukopenia, etc.) 34 WT2725Phase I62 (20 GBM)Advanced WT1+ malignancies (HLA-A*0201+/0206+)Solid tumors: 13.0GBM: 10.2Solid tumors: ~2GBM: ~2• irCR (GBM): 13.3% (2/15); irPR: 6.7%• CTL response: dose-dependent (0 mg → 27 mg: 0% → 88.9%)• 2 GBM pts irCR, PFS ≥30.6 and ≥37.0 mo• No DLT; MTD not reached• Most common: grade 1-2 injection site reactions• No discontinuations 36 IMA950Phase I45nGBM (HLA-A*02+, post-resection, receiving CRT+adj TMZ)15.3 (overall)Not reported (PFS-6: 74%, PFS-9: 31%)• Immune response: 90% TUMAP responders• ISR correlated with OS (ISR+ mOS 26.7 mo vs ISR- 13.2 mo; HR 0.33, p=0.0001)• MGMT methylated: mOS 28.3 mo vs unmethylated 14.8 mo (p=0.025)• Well tolerated• Most common: grade 1-2 injection site reactions (26/45)• 2 DLTs (grade 3 fatigue, anaphylaxis) 19 IMA950 + poly-ICLCPhase I/II19 (16 GBM, 3 AA)nGBM/AA (HLA-A2+, post-RT/TMZ)GBM: 19 (from surgery)Whole cohort: 21GBM: 9.5 (from surgery)PFS-6/9 (GBM): 81%/63%• Primary immunogenicity endpoint met• CD8 multi-peptide responses: 46.2%• CD4 responses: 84.6%• Well tolerated• Most common: injection site reactions, fatigue, headache• 1 Grade 4 interstitial pneumonia (unrelated) 37 WT1 cocktail vaccinePhase I14 (12 GBM, 2 AA)rGBM/AA (HLA-A*24:02+, WT1+, refractory to TMZ)24.7 weeks (~5.7)Not reported• SD at 6 weeks: 43%• 1-year OS rate: 36%• Immune responses: WT1-specific CD8+ (64%), CD4+ (91%)• No DLT; No grade 3/4 AEs• Grade 1 injection site reactions (all patients) 38 VEGFRs peptide vaccine + TMZPhase I/II4nGBM (HLA-A*24:02+, post-surgery, receiving RT/TMZ)Individual: 31.6, 41.8, >46.8, >31.6Individual: 18.2, 41.8, >46.8, >31.6• 2/4 patients achieved CR (incl. 1 MGMT unmethylated)• Both CR patients alive at analysis• Histopathological evidence of CTL-mediated killing of tumor vessels, cells, and Tregs• No grade 3/4 AEs• Grade 1 local skin reaction in 1 patient 39 GAA peptide vaccine + poly-ICLCPhase I23LGG (high-risk), HLA-A2⁺Not reportedCohort 1 (no prior RT): 17Cohort 3 (recurrent): 12• Cohort 1: 91% responded to ≥1 GAA, 82% to ≥3 GAAs• Cohort 3: 56% responded to ≥1 GAA, 33% to ≥3 GAAs• 3 pts in Cohort 1 remain PFS (37, 42, 47+ mo)• Grade 1-2 injection site reactions (100%)• One case of Grade 3 fever/fatigue (RLT) 40 Personalized mutIDH1 multi-peptide vaccinePhase I/II52IDH1-mutant glioma (grade 2-4), newly diagnosed or recurrentNot reached (median f/u 36 mo)Not reported• mutIDH1 immune response: 89%• Immunological responders had significantly longer OS (HR 0.09; p=0.0057)• mutIDH1-specific CD4+ T cells were polyfunctional and polyclonal• Identical TCRβ sequences found in patients with shared HLA-DQ alleles• Well tolerated• Grade 1 injection site reactions only• No grade ≥3 AEs 20 Personalized peptide vaccination (PPV) based on pre-existing IgG levelsPhase III (RCT, DB)88 (58 PPV, 30 placebo)HLA-A24+ recurrent GBM, refractory to TMZ/RTPPV: 8.4Placebo: 8.0 (p=0.62)Not shown (no significant difference)• Primary endpoint not met• PPV detrimental with SART2-93 peptide (HR 15.9), age ≥70 yr (HR 7.9), weight >70 kg (HR 4.1), PS3 (HR 2.8)• PPV beneficial without SART2-93 + age <70 yr (HR 0.49, p=0.03)• Lower baseline CD11b+CD14+HLA-DRlow monocytes and higher CD3+CD4+CD45RA- T cells correlated with improved OS• Well tolerated• Most common: injection site reactions (PPV 71%, placebo 53%)• Grade ≥3 AEs: PPV 40%, placebo 37%• One grade 3 pulmonary embolism (PPV-related) 44 Autologous formalin-fixed tumor vaccine (AFTV)Phase IIb (RCT, DB, placebo-controlled)57 (30 AFTV, 27 placebo)Newly diagnosed supratentorial GBM, age 16-75 yr, KPS ≥60%AFTV: 25.6Placebo: 31.5 (p=0.64)AFTV: 13.3Placebo: 13.3 (p=0.98)• Primary endpoint not met• Total resection subgroup: 3-yr OS 80% vs 54% (p = 0.16); 3-yr PFS 81% vs 46% (p = 0.067)• p53-negative subgroup: 3-yr OS 79% vs 43% (p = 0.072)• DTH-2 response not associated with survival• Met predefined criteria for Phase III trial• Well tolerated; no significant difference in AEs• Grade ≥3 AEs: AFTV 23%, placebo 30%• No serious AEs attributed to AFTV 46 APVAC1 + APVAC2 + poly-ICLC + GM-CSFPhase I16 enrolled, 15 vaccinatednGBM (HLA-A*02:01⁺/24:02⁺, post-GTR, receiving CRT)29.0 (n=15 evaluable)14.2• APVAC1 immune response: 92.3%• APVAC2 immune response: 80% (Th1-type CD4⁺)• Low baseline Tregs correlated with higher memory cell induction• Well tolerated• Most common: grade 1-2 injection site reactions• Two cases of anaphylaxis (potentially GM-CSF-related)• One grade 3 cerebral edema 45 Immune Checkpoint InhibitorsIpilimumab + Nivolumab + adj TMZPhase I32 enrolled, 31 evaluablenGBM (unifocal, supratentorial, post-GTR/NTR, post-CRT)Combination cohort (n=19): 20.7 (from treatment start)Combination cohort (n=19): 16.1• Primary safety endpoint met: 1 DLT in Cohort 1, 1 DLT in Cohort 2, 0 DLTs in Cohort 3• 1-year OS rate (combination): 68.4%• 2-year OS rate (combination): 31.6%• Well tolerated; no Grade 5 AEs• No unexpected toxicity• Most common Grade 3-4: skin disorders (38.5%), lab abnormalities (23.1%) 42 Ipilimumab + Nivolumab + RT (no TMZ)Phase II/III (RCT)159 (79 immunotherapy, 80 TMZ)nGBM (MGMT-unmethylated, KPS ≥70, post-resection)Immature (~13 mo each)HR 0.95, p=0.367.7 (immunotherapy) vs 8.5 (TMZ)HR 1.47, p=0.96• Primary endpoint not met (PFS not improved)• Trial terminated• 2 treatment-related deaths on immunotherapy arm (autoimmune disorder, colitis)• No significant difference in AEs between arms (p = 0.69)• Grade ≥3 AEs: 37.8% (TMZ) vs 41.0% (immunotherapy)• Grade 5 events: 0 vs 2 43 Atezolizumab + RT/TMZPhase I/II60nGBM (WHO 2016 criteria, KPS ≥60)All pts: 18.0IDH-wt: 16.1MGMT methylated: 25.4MGMT unmethylated: 14.6Not reported• ORR: 23.3% (CR 10% + PR 13.3%)• DoR: 13.91 mo• High ESTIMATE Immune Score tumors: mOS 24.8 vs 14.5 mo (HR 0.45, p = 0.02)• Mesenchymal subtype: mOS 26.5 mo• Well tolerated; no new safety signals• Grade ≥3 treatment-related AEs: 56.7% (most common: lymphopenia)• 3 patients removed due to hepatitis/pneumonitis (DLTs) 70 Nivolumab + Ipilimumab (pre-radiation)Phase I15nGBM (Grade 3/4, post-surgery, pre-RT)19.3 (95% CI, 12.9-NA)1.3 (95% CI, 0.92-2.99)• Feasibility demonstrated (treatment started median 38 days post-op)• MGMT methylated: mOS 35.7 mo vs unmethylated: 12.6 mo (p = 0.004)• Paired tumor analysis (n = 5): post-progression tumors showed ↑ TGF-β/ERBB, ↓ PPAR signaling, ↑ monocytes, ↓ B cells.• Growing tumors trended toward higher baseline LAG-3.• Well tolerated; no DLTs• Most common: rash, pruritus, fatigue, nausea, anorexia• Grade 3: lipase ↑ (2), anorexia (1), pruritus (1), rash (3)• One Grade 4 cerebral edema• No Grade 5 events 68 Neoadjuvant + Adjuvant PembrolizumabPhase II (single-arm expansion)25 (expansion) + 32 (initial) = 57 pooledrGBM (surgically accessible, 1st/2nd relapse, KPS ≥70)Expansion: 6.8Initial neoadjuvant: 13.7Initial adjuvant: 7.5Expansion: 2.5Initial neoadjuvant: 3.3Initial adjuvant: 2.5• Primary pharmacodynamic endpoint met: Cell cycle gene signature decreased in neoadjuvant vs adjuvant tumors (p=0.009)• IFN-high/CC-low subgroup (n=13): mOS 13.8 mo• No confirmed OS benefit for neoadjuvant regimen• Well tolerated; no new safety signals• Grade ≥3 AEs in 6 pts: cerebral edema (2), headache, fatigue, adrenal insufficiency, hyperthyroidism 69 Tislelizumab + Low-Dose BevacizumabPhase II32rGBM14.38.2• ORR: 56.3%• 12-month OS: 43.8%• >20% reduction in MAF/TMB post-treatment correlated with improved PFS and OS (p = 0.0005-0.008)• Baseline MUC16 mut, H3F3B amp, SRSF2 amp associated with worse OS• Any grade AEs: 90.6% (most common: anemia, fatigue, increased ALT)• Grade ≥3: 6.2% (incl. one grade 4 acute pancreatitis)• No grade 5 events 73 Adoptive Immune Cell TherapyCART-EGFRvIII cells + PembrolizumabPhase I7nGBM (EGFRvIII+, MGMT-unmethylated)11.8 (90% CI, 9.2-14.2)5.2 (90% CI, 2.9-6.0)• Primary safety endpoint met: no DLTs• Antigen reduction: EGFRvIII decreased post-treatment in 6/7 pts• ↑ IFN-stimulated T cells in post-treatment tumors correlated with time from relapse to death (Spearman p=0.03)• Safe and well-tolerated• No CRS, no ICANS• One patient experienced severe (grade 3-4) immune-related AE (liver/kidney injury) likely related to Pembrolizumab 13 CART-EGFRvIII cells (single infusion)Phase I10rGBM (EGFRvIII+, MGMT-unmethylated)251 days (~8)Not evaluable• Primary safety & feasibility endpoint met• On-target activity: EGFRvIII expression decreased in 5/7 assessed pts (p = 0.03)• Trafficking: Higher CAR T concentration in brain vs blood in 2 pts• TME changes: ↑ immunosuppressive molecules (IDO1, PD-L1, FoxP3, IL-10)• No EGFR-directed toxicity, no systemic CRS, no neurotoxicity typical of CD19-CARs• 3 subjects with clinically significant neurologic events (possibly related to disease or local inflammation)• No DLTs 14 CART-EGFR-IL13Rα2 (bivalent, ICV)Phase I18rGBM (EGFR-amplified, IDH-wt)Not reached (f/u 8.1 mo)1.9 (90% CI, 1.1-3.4)• ORR: 8% (1/13)• Tumor regression: 62% (8/13)• Durable disease: 2 pts PFS >6 mo (7.7, 16.6)• MTD/RDE: 2.5 × 10⁷ cells• CRS: 100% (all grades 1-2)• Grade 3 neurotoxicity: 56% (no grade 4-5)• One DLT (grade 3 anorexia, weakness, fatigue) at MTD 110 IL-13(E13Y)-zetakine CD8+ CTLsPhase I3rGBMNot reported (mean ~11)Not reported• Feasibility demonstrated• Antigen loss: In one patient, IL13Rα2 expression decreased post-treatment• Radiographic response: In one patient, ↑ necrotic volume on MRI post-retreatment• Well tolerated at 10⁷ and 5×10⁷ cell doses• At 10⁸ cell dose: Grade 3 headache (1 pt), Grade 3 neurological event (shuffling gait, tongue deviation; 1 pt); both transient and manageable 111 GRm13Z40-2 cells (allogeneic, GR-knockout)Phase I6rGBM (unresectable, IL13Rα2+, on dexamethasone)2.9 (post-treatment)Not reported• Proof-of-Concept: First-in-human allogeneic, ZFN-modified, steroid-resistant CAR T product• Anti-tumor activity: Transient tumor reduction/necrosis at infusion site in 4/6 pts• Well tolerated• All treatment-related AEs ≤ grade 3, transient, manageable• No DLTs, GVHD, or severe treatment-related neuroinflammation 112 GD2-specific 4SCAR-T (IV ± intracavitary)Phase I8rGBM (GD2+, IDH-wt; 4 adults, 4 children)10 (range, 3-24) from infusionNot reported• ORR: 50% (4/8 PR, 3-24 mo duration)• Disease control (PR+SD): 62.5%• Antigen loss & TME modulation: ↓ GD2, ↑ CD8+, ↓ CD163+ M2 macrophages (1 pt)• Safe and well-tolerated• No severe AEs (grade ≥4) or CAR-T-related neurotoxicity• One patient: grade 2 seizures and grade 3 headache probably related to infusion 113 CLTX-CAR T cells (intracavitary/intratumoral)Phase I (Interim)4rGBM (MMP-2+, heavily pretreated)<6 (3 pts)1 pt >20Not reported• Disease Control Rate: 75% (3/4 SD)• Regional control: No recurrence at infusion site for >9 weeks (1 pt)• Proof-of-Concept: First-in-human toxin-based CAR platform• Well tolerated; no DLTs or CRS• No grade ≥3 CNS events attributed as probable/definite• Grade 3 cerebral edema (possibly related) in 2 pts• No humoral immunogenicity detected 114 Oncolytic VirotherapyMV-CEAPhase I22rGBM (anti-MV immune at baseline)11.63.4• 12-month OS: 45.5% (favorable vs historical controls)• Disease control rate (SD): 88.9% (Group A), 92.3% (Group B)• Biomarker (ISG signature): DLDA score inversely correlated with viral replication (R=-0.6, p=0.04)• TME remodeling: ↑ CD8+ and CD68+ infiltration (day 5)• Well tolerated; no DLTs up to max dose (2×10⁷ TCID50)• Only grade 1-2 AEs• No viremia, viral shedding, or treatment-induced immunosuppression 136 DNX-2401 + PembrolizumabPhase I/II49rGBM/rGS (1st/2nd recurrence, unresected)12.53.4• Primary ORR endpoint (10.4%) not met (p=NS vs 5%)• Secondary 12-mo OS endpoint (52.7%) met (vs 20% control)• Clinical benefit rate (SD or better): 56.2%• 3 patients alive at 45, 48, and 60 mo• Biomarker: Responses only in the "TMEmedium" subtype (exploratory)• Well tolerated; no DLTs• Most common: brain edema (37%), headache (31%), fatigue (29%)• Grade 3 AEs: 22%; one grade 4 event• No treatment-related deaths 132 Ad-TD-nsIL12Phase I8rHGG (with ventricular system connection)5.1 (range, 3.1-21.2) from first injectionNot reported• Primary MTD endpoint: 1×10¹⁰ vp (grade 3 seizures as DLT at 5×10¹⁰ vp)• Antitumor activity: 1 CR, 1 PR (both in Cohort 2)• TME remodeling: ↑ CD4+/CD8+ T cell infiltration, confirmed viral replication• Well tolerated at ≤1×10¹⁰ vp• Most common: fever, cognitive disturbance, nausea/vomiting• Grade 3 seizures (DLTs) at highest dose in 2/2 pts• No viral DNA in blood/CSF 137 NSC-CRAd-S-pk7 (neural stem cell-loaded)Phase I12nGBM (WHO grade III/IV)18.49.05• Primary safety endpoint met: No DLT up to highest dose• MGMT unmethylated subgroup (exploratory): mOS 18.0 mo, mPFS 8.8 mo (favorable vs historical)• TME remodeling: ↑ CD8+ T cell infiltration, ↑ PD-1 expression, ↓ survivin/syndecan-1• Safe and well-tolerated; no DLT• One patient: grade 3 viral meningitis (probably related, due to inadvertent intraventricular injection)• No treatment-related deaths• RP2D: 1.50×10⁸ NSCs / 1.875×10¹¹ vp 138 G47Δ (triple-mutated HSV-1)Phase II19Residual/rGBM (post-RT/TMZ)20.2 (from 1st injection)28.8 (from initial surgery)4.7 (from 1st injection)• Primary endpoint (1-yr OS): 84.2% (met, trial terminated early)• TME remodeling: ↑ CD4+/CD8+ TILs with repeated dosing, low Foxp3+ Tregs• Regulatory outcome: First approved oncolytic virus for malignant glioma (Japan, 2021)• Safe and well-tolerated; no DLT• Most common: fever (89.5%), vomiting (57.9%), nausea (52.6%)• Grade ≥3 lymphocytopenia: 26.3% (resolved without treatment) 139 G47Δ (triple-mutated HSV-1)Phase I/II13rGBM (post-RT/TMZ)7.3 (from last dose)30.5 (from initial surgery)8 days (from last dose)• 1-year OS rate (from last dose): 38.5%• Long-term survival: 3 patients >46 mo• Best overall response (2-yr, exploratory): CR 7.7%, PR 7.7%• TME remodeling: ↑ CD4+/CD8+ T cell infiltration post-treatment• Safe and well-tolerated• Most common: fever (61.5%), headache (61.5%), vomiting (46.2%)• Grade 3 events: leukopenia, headache, vomiting, neurological disorder (few pts)• No DLT 140 Abbreviations: AA, anaplastic astrocytoma; Ab, antibody; adj, adjuvant; AE, adverse event; AFTV, autologous formalin-fixed tumor vaccine; ALT, alanine aminotransferase; amp, amplification; APVAC, actively personalized vaccination; CAR, chimeric antigen receptor; CBR, clinical benefit rate; CI, confidence interval; CMV, cytomegalovirus; CNS, central nervous system; CR, complete response; CRT, chemoradiotherapy; CRS, cytokine release syndrome; CSF, cerebrospinal fluid; CTL, cytotoxic T lymphocyte; DB, double-blind; DC, dendritic cell; DCR, disease control rate; DI-TMZ, dose-intensified temozolomide; DLT, dose-limiting toxicity; DLDA, diagonal linear discriminant analysis; DoR, duration of response; DTH, delayed-type hypersensitivity; dx, diagnosis; ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; EGFRvIII, epidermal growth factor receptor variant III; ESTIMATE, Estimation of Stromal and Immune cells in Malignant Tumors using Expression data; f/u, follow-up; FACS, fluorescence-activated cell sorting; FoxP3, forkhead box P3; GAA, glioma-associated antigen; GBM, glioblastoma; GM-CSF, granulocyte-macrophage colony-stimulating factor; GR, glucocorticoid receptor; GSC-DCV, glioma stem cell-derived dendritic cell vaccine; GTR, gross total resection; GVHD, graft-versus-host disease; H3F3B, H3.3 histone B; HLA, human leukocyte antigen; HR, hazard ratio; HSV, herpes simplex virus; ICANS, immune effector cell-associated neurotoxicity syndrome; ICB, immune checkpoint blockade; ICI, immune checkpoint inhibitor; ICV, intracerebroventricular; IDH, isocitrate dehydrogenase; IDO1, indoleamine 2,3-dioxygenase 1; IFN, interferon; Ig, immunoglobulin; IHC, immunohistochemistry; IL, interleukin; incl., including; IQR, interquartile range; irCR, immune-related complete response; irPR, immune-related partial response; iRANO, immunotherapy Response Assessment in Neuro-Oncology; ISG, interferon-stimulated gene; ISR, IMA950-specific response; ITT, intent-to-treat; IV, intravenous; KPS, Karnofsky Performance Status; LAG-3, lymphocyte activation gene 3; LGG, low-grade glioma; MAF, mutant allele fraction; MGMT, O⁶-methylguanine-DNA methyltransferase; mo, months; MRS, magnetic resonance spectroscopy; MRD, measurable residual disease; MSS, mutation-specificity score; MTD, maximum tolerated dose; MUC16, mucin 16; mut, mutation; MV, measles virus; MVAF, maximal somatic variant allelic frequency; NA, not available; NED, no evidence of disease; nGBM, newly diagnosed glioblastoma; NR, not reached; NS, not significant; NTR, near total resection; ORR, objective response rate; OS, overall survival; PD, progressive disease; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; pfu, plaque-forming units; PLR, platelet-to-lymphocyte ratio; poly-ICLC, polyinosinic-polycytidylic acid stabilized with poly-L-lysine and carboxymethylcellulose; post-, post-treatment; PPV, personalized peptide vaccination; pre-, pre-treatment; PR, partial response; PsPD, pseudoprogression; PS, performance status; pt, patient; RANO, Response Assessment in Neuro-Oncology; RCT, randomized controlled trial; RDE, recommended dose for expansion; rGBM, recurrent glioblastoma; rGS, recurrent gliosarcoma; rHGG, recurrent high-grade glioma; RLT, restricted limiting toxicity; RPA, recursive partitioning analysis; RP2D, recommended Phase 2 dose; RT, radiotherapy; SAE, serious adverse event; SD, stable disease; SRD, significant residual disease; SRSF2, serine and arginine rich splicing factor 2; T1, time point 1; T2, time point 2; TCID50, 50% tissue culture infectious dose; Td, tetanus-diphtheria; TERT, telomerase reverse transcriptase; TGF-β, transforming growth factor beta; TIL, tumor-infiltrating lymphocyte; TISF, tumor in situ fluid; TMB, tumor mutational burden; TME, tumor microenvironment; TMZ, temozolomide; TNF, tumor necrosis factor; Treg, regulatory T cell; TUMAP, tumor-associated peptide; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor; vp, viral particles; vs, versus; wk, week; WT1, Wilms tumor 1; yr, year; ZFN, zinc finger nuclease. Summary of key clinical trials of immunotherapy for glioma Abbreviations: AA, anaplastic astrocytoma; Ab, antibody; adj, adjuvant; AE, adverse event; AFTV, autologous formalin-fixed tumor vaccine; ALT, alanine aminotransferase; amp, amplification; APVAC, actively personalized vaccination; CAR, chimeric antigen receptor; CBR, clinical benefit rate; CI, confidence interval; CMV, cytomegalovirus; CNS, central nervous system; CR, complete response; CRT, chemoradiotherapy; CRS, cytokine release syndrome; CSF, cerebrospinal fluid; CTL, cytotoxic T lymphocyte; DB, double-blind; DC, dendritic cell; DCR, disease control rate; DI-TMZ, dose-intensified temozolomide; DLT, dose-limiting toxicity; DLDA, diagonal linear discriminant analysis; DoR, duration of response; DTH, delayed-type hypersensitivity; dx, diagnosis; ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; EGFRvIII, epidermal growth factor receptor variant III; ESTIMATE, Estimation of Stromal and Immune cells in Malignant Tumors using Expression data; f/u, follow-up; FACS, fluorescence-activated cell sorting; FoxP3, forkhead box P3; GAA, glioma-associated antigen; GBM, glioblastoma; GM-CSF, granulocyte-macrophage colony-stimulating factor; GR, glucocorticoid receptor; GSC-DCV, glioma stem cell-derived dendritic cell vaccine; GTR, gross total resection; GVHD, graft-versus-host disease; H3F3B, H3.3 histone B; HLA, human leukocyte antigen; HR, hazard ratio; HSV, herpes simplex virus; ICANS, immune effector cell-associated neurotoxicity syndrome; ICB, immune checkpoint blockade; ICI, immune checkpoint inhibitor; ICV, intracerebroventricular; IDH, isocitrate dehydrogenase; IDO1, indoleamine 2,3-dioxygenase 1; IFN, interferon; Ig, immunoglobulin; IHC, immunohistochemistry; IL, interleukin; incl., including; IQR, interquartile range; irCR, immune-related complete response; irPR, immune-related partial response; iRANO, immunotherapy Response Assessment in Neuro-Oncology; ISG, interferon-stimulated gene; ISR, IMA950-specific response; ITT, intent-to-treat; IV, intravenous; KPS, Karnofsky Performance Status; LAG-3, lymphocyte activation gene 3; LGG, low-grade glioma; MAF, mutant allele fraction; MGMT, O⁶-methylguanine-DNA methyltransferase; mo, months; MRS, magnetic resonance spectroscopy; MRD, measurable residual disease; MSS, mutation-specificity score; MTD, maximum tolerated dose; MUC16, mucin 16; mut, mutation; MV, measles virus; MVAF, maximal somatic variant allelic frequency; NA, not available; NED, no evidence of disease; nGBM, newly diagnosed glioblastoma; NR, not reached; NS, not significant; NTR, near total resection; ORR, objective response rate; OS, overall survival; PD, progressive disease; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; pfu, plaque-forming units; PLR, platelet-to-lymphocyte ratio; poly-ICLC, polyinosinic-polycytidylic acid stabilized with poly-L-lysine and carboxymethylcellulose; post-, post-treatment; PPV, personalized peptide vaccination; pre-, pre-treatment; PR, partial response; PsPD, pseudoprogression; PS, performance status; pt, patient; RANO, Response Assessment in Neuro-Oncology; RCT, randomized controlled trial; RDE, recommended dose for expansion; rGBM, recurrent glioblastoma; rGS, recurrent gliosarcoma; rHGG, recurrent high-grade glioma; RLT, restricted limiting toxicity; RPA, recursive partitioning analysis; RP2D, recommended Phase 2 dose; RT, radiotherapy; SAE, serious adverse event; SD, stable disease; SRD, significant residual disease; SRSF2, serine and arginine rich splicing factor 2; T1, time point 1; T2, time point 2; TCID50, 50% tissue culture infectious dose; Td, tetanus-diphtheria; TERT, telomerase reverse transcriptase; TGF-β, transforming growth factor beta; TIL, tumor-infiltrating lymphocyte; TISF, tumor in situ fluid; TMB, tumor mutational burden; TME, tumor microenvironment; TMZ, temozolomide; TNF, tumor necrosis factor; Treg, regulatory T cell; TUMAP, tumor-associated peptide; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor; vp, viral particles; vs, versus; wk, week; WT1, Wilms tumor 1; yr, year; ZFN, zinc finger nuclease. 5.1 Immunosuppression in the Tumor Microenvironment Regardless of the immunotherapeutic strategy employed, its ultimate antitumor efficacy depends on the effective activation of a functional immune system 158 (Figure 5). However, the TME of glioma, particularly glioblastoma, is characterized by profound immunosuppression, rendering immune cells—whether vaccine-induced T cells 159 or CAR-T cells 13, 14—prone to inactivation or exhaustion upon entry into the tumor site. Schematic overview of four major immunotherapeutic strategies for glioma. 1) Peptide vaccines and dendritic cell (DC) vaccines: Peptide vaccines deliver tumor-associated antigens (TAAs) in combination with adjuvants, thereby inducing DC maturation in vivo. DC vaccines are generated by differentiating patient-derived CD14⁺ monocytes into immature DCs, followed by loading with tumor antigens. Mature DCs promote the activation and differentiation of T cells. Activated T cells cross the blood-brain barrier (BBB) and exert anti-glioma effects through Fas/FasL-mediated apoptosis and the perforin/granzyme release pathway. 2) Oncolytic virus therapy: Oncolytic viruses selectively infect and lyse glioma cells, releasing TAAs and damage-associated molecular patterns (DAMPs), which in turn activate DCs and initiate T cell responses, ultimately leading to tumor cell killing. 3) Immune checkpoint inhibitors: These inhibitors block immune checkpoint signaling pathways, reverse T cell exhaustion, and thereby eliminate glioma cells. 4) Chimeric antigen receptor T (CAR-T) cell therapy: Patient-derived peripheral blood mononuclear cells (PBMCs) are isolated, genetically modified to express chimeric antigen receptors (CARs), expanded ex vivo, and then infused back into the patient. CAR-T cells cross the BBB and specifically target and kill glioma cells. 5.1.1 Impact on Vaccines Vaccines require the presentation of antigens to DCs to subsequently activate T cells 160. However, in the immunosuppressive TME, dendritic cell function is impaired, and activated T cells are also unable to effectively infiltrate and proliferate 160. 5.1.2 Impact on ICIs The PD-1/PD-L1 immune checkpoint pathway is just one of many immunosuppressive mechanisms 161. The TME also contains abundant Tregs, MDSCs, and inhibitory cytokines such as TGF-β and IL-10 161. Even upon PD-1/PD-L1 blockade, these additional mechanisms continue to effectively suppress T cell function 162. 5.1.3 Impact on CAR-T Cells Even if CAR-T cells successfully reach the tumor site, they encounter the same immunosuppressive microenvironment, leading to poor persistence, rapid exhaustion, and an inability to sustain long-term antitumor effects 103. 5.1.4 Impact on Oncolytic Viruses The primary mechanism underlying the antitumor effect of oncolytic viruses is a T cell-mediated, tumor-specific immune response 163. However, the TME can inhibit T cell function through various mechanisms, thereby compromising the efficacy of this therapeutic approach 163. 5.2 Tumor Antigen Heterogeneity and Escape Glioblastoma exhibits significant spatial and interpatient heterogeneity, which is a key factor contributing to immune evasion 164. The expression of recognizable antigens on tumor cells is a prerequisite for vaccine-induced immune responses and for ICIs to relieve T cell suppression 165. For oncolytic viruses, however, direct oncolysis depends on the expression of surface receptors on target cells, yet it does not require that all tumor cells express such receptors 166. As long as a sufficient number of cells express the receptors and become infected, releasing ample damage-associated molecular patterns and tumor-associated antigens, an immune response can be activated to eliminate tumor cells that do not express the receptors 167. Strategies such as multi-target vaccines 19, dual-target CAR-T cells 116, and "plug-and-play" CAR-T platforms (e.g., mSA2 CAR-T) 115 hold promise for overcoming immune evasion driven by tumor heterogeneity by enabling simultaneous targeting of multiple tumor-associated antigens or flexibly adapting to changes in antigen expression, positioning them as a key direction for future research. 5.3 Blood-Brain Barrier Constraints Glioblastoma is located within the central nervous system, where the unique BBB not only acts as a physical obstacle to drug delivery but also restricts the entry of immune cells from the periphery into the tumor site 168. 5.3.1 Impact on Macromolecular Therapeutics and Oncolytic Viruses Monoclonal antibodies (such as PD-1/PD-L1 antibodies), CAR-T cells, and oncolytic viruses are all macromolecules or nanoscale particles that are largely incapable of crossing an intact BBB under normal conditions 145-148. Even though tumor growth can disrupt the local barrier, such disruption is heterogeneous and insufficient, resulting in suboptimal drug accumulation 156. 5.3.2 The Impact on Vaccines Vaccines activate T cells in peripheral lymph nodes, and these activated T cells must successfully cross the BBB to exert their antitumor effects 169. The presence of the BBB significantly limits the infiltration efficiency of these T cells 169. Various delivery strategies, such as nanocarriers 61, 79 and focused ultrasound 76, offer potential approaches to overcome this physical barrier. 5.4 Adaptive Resistance and Tumor Evolution Most studies suggest that tumors undergo dynamic evolution under therapeutic pressure. Even when an initial treatment is effective, tumors can acquire adaptive resistance through multiple mechanisms, including upregulation of alternative immune checkpoint molecules, metabolic reprogramming, or activation of bypass growth signaling pathways 68, 132. However, a few studies have also demonstrated that tumors can evolve even in the absence of pharmacological intervention, under natural conditions 31. For instance, following treatment with PD-L1 inhibitors, tumors may upregulate the expression of other checkpoint molecules (such as LAG-3, TIM-3, TIGIT, and CTLA-4), thereby achieving "bypass escape" 170. For CAR-T therapy, in addition to direct antigen loss, tumors can also evade immune recognition by downregulating antigen processing and presentation machinery components, such as MHC class I molecules, thereby impairing the antitumor response of endogenous T cells and indirectly promoting immune evasion 171. For vaccines, under the immune pressure induced by treatment, tumor cells may escape immune recognition by directly downregulating target antigen expression or by downregulating antigen processing and presentation machinery components, such as MHC class I molecules, thereby hindering the killing of tumor cells by CD8+ T cells 172. Oncolytic virus infection of tumor cells depends on specific cell surface receptors 173. Under therapeutic pressure, tumor cells can resist viral infection by downregulating or losing these receptors 173. This mechanism reveals the underlying reasons for the transient efficacy of monotherapy and highlights the importance of combination therapy. The advantages of combination therapy can be summarized as follows: 1) Broader target coverage. Whether through the combination of ICIs 70, multi-target vaccines 19, or universal CAR-T platforms 115, the range of target coverage exceeds that of monotherapy, enabling more effective elimination of heterogeneous tumor cell populations. 2) Broader coverage of mechanisms of action. Under therapeutic pressure, tumors can acquire adaptive resistance through multiple mechanisms 68, 132. To circumvent this, combining therapies with distinct mechanisms—such as ICIs with epigenetic modulators 91 or with oncolytic viruses 134—makes it more difficult for tumors to escape. 3) Synergistic effects. The combination of certain therapies, such as chemotherapy with immunotherapy, can produce synergistic effects that achieve "1+1 > 2" therapeutic outcomes, surpassing the simple additive effects of individual agents 39. 4) Improved safety. In a combination therapy setting, the dosage of each therapeutic agent can be reduced accordingly, thereby alleviating target organ toxicity and enhancing the safety and tolerability of treatment 174. In line with the growing recognition of combination strategies, several ongoing clinical trials are evaluating the efficacy of immunotherapy combined with other modalities (Table 2). Table 2Ongoing clinical trials of immunotherapy combination strategies in gliomaCombination StrategySpecific RegimenTrial PhaseTarget PopulationPrimary Endpoint / ObjectiveNCT NumberVaccine + ChemoradiotherapyTamavaq™ personalized neoantigen vaccine + Radiotherapy + TemozolomideEarly Phase INewly diagnosed gliomaPrimary: Safety, efficacyNCT07077616Immuno-immuno combinationAutologous TIL therapy + Pembrolizumab (anti-PD-1)Phase I/IIAdvanced brain cancer (gliomas, meningiomas, etc.)Primary: Efficacy, safetyNCT06640582Immunotherapy + ChemotherapyTarlatamab (DLL3-targeted BiTE) + Metronomic TemozolomidePhase I/IIAdolescents and adults with recurrent high-grade CNS tumors (including IDH-mutant glioma, ependymoma, medulloblastoma)Primary: Safety;Secondary: Clinical activityNCT07243470Immunotherapy + Anti-angiogenicCTX-009 (DLL4/VEGF-A bispecific antibody) + CTX-471 (anti-CD137 agonist antibody)Phase Ib/IIRecurrent glioblastoma (CNS WHO grade 4, IDH-wildtype)Primary: Safety;Secondary: Efficacy (preliminary)NCT07392957Dendritic cell vaccine + Standard of careDOC1021 (autologous dendritic cell immunotherapy) + Radiotherapy + Temozolomide + Tumor Treating Fields (TTFields)Phase IINewly diagnosed glioblastoma (WHO grade 4)Primary: Safety, efficacyNCT06805305Peptide vaccine + ChemotherapyPEP-CMV vaccine + Adjuvant temozolomide (one cycle)Phase INewly diagnosed MGMT-unmethylated glioblastomaPrimary: Safety, tolerabilityNCT06132438ICI + Anti-angiogenic + RadiotherapyRetifanlimab (anti-PD-1) + Bevacizumab + Hypofractionated RadiotherapyPhase IIRecurrent glioblastomaPrimary: EfficacyNCT06160206Gene therapy + Immunotherapy + SurgeryTGX-007 (AAV-HSV-tk + IL-12) intratumoral injection + oral valacyclovir + standard of care surgeryPhase I/IINewly diagnosed or recurrent glioblastomaPrimary: Safety, OBD;Secondary: EfficacyNCT07346144CAR-T+ChemotherapyAutologous GD2-C7R CAR-T cells (GD2 CAR + constitutively active IL-7 receptor) + Cyclophosphamide/Fludarabine lymphodepletionPhase IGD2-expressing Brain Tumors (DMG, HGG, DIPG, medulloblastoma)Primary: MTD; Secondary: CAR-T persistence, anti-tumor activityNCT04099797Dual ICI + SurgeryIpilimumab (anti-CTLA-4) + Nivolumab (anti-PD-1) administered IV (neo-adjuvant) and intracranial (adjuvant via Ommaya reservoir) + maximal safe resectionPhase IResectable recurrent glioblastomaPrimary: Safety, feasibilityNCT06097975Bispecific antibody-armed T cells + Focused ultrasoundEGFR BATs (anti-CD3 × anti-EGFR bispecific antibody-armed autologous T cells) + Low-intensity focused ultrasound (LIFU) with microbubbles for BBB openingPhase INewly diagnosed MGMT unmethylated IDH-wildtype glioblastomaPrimary: Safety, feasibility; Secondary: T cell traffickingNCT07343986bispecific antibody-immunotoxin + immunomodulatorD2C7-IT (bispecific antibody) + 2141-V11 (immunomodulator)Phase I/IIResected recurrent glioblastomaPrimary: Safety; Secondary: EfficacyNCT06455605Radioimmunotherapy + ICIActimab-A (anti-CD33 monoclonal antibody conjugated with actinium-225) + Cemiplimab (anti-PD-1)Phase IRecurrent glioblastomaPrimary: Safety, tolerability;Secondary: EfficacyNCT07422363Allogeneic γδ T cell therapy ± Chemo/RadiotherapyAllogeneic Vγ9Vδ2 T cells (intraventricular via Ommaya reservoir) ± Temozolomide or radiotherapyPhase IWHO grade IV malignant glioma (glioblastoma, DIPG, medulloblastoma, ependymoma)Primary: Safety, feasibility; Secondary: EfficacyNCT06396481Bispecific antibody-immunotoxin+ immunomodulator + RadiotherapyD2C7-IT (bispecific antibody-immunotoxin) + 2141-V11 (immunomodulator)Phase I/IINewly diagnosed MGMT unmethylated glioblastomaPrimary: Safety;Secondary: EfficacyNCT05734560CAR-T + Dual ICIs + Neoadjuvant/AdjuvantIL13Rα2-CAR T cells (intracranial via Rickham catheter) + Nivolumab (anti-PD-1) ± Ipilimumab (anti-CTLA-4)Phase IResectable recurrent glioblastomaPrimary: Safety;Secondary: EfficacyNCT04003649IL-7 agonist + PD-1 inhibitor + SurgeryNeoadjuvant Efineptakin Alfa (NT-I7) and PembrolizumabPhase IIRecurrent glioblastomaPrimary: Safety; Secondary: EfficacyNCT05465954PD-L1 inhibitor + Multi-kinase inhibitorAtezolizumab (anti-PD-L1) + Cabozantinib (VEGFR/MET/AXL inhibitor)Phase I/IIRecurrent glioblastomaPrimary: Safety; Secondary: EfficacyNCT05039281Dendritic cell vaccine + PD-1 inhibitorHybrid dendritic cell vaccine + Pembrolizumab (anti-PD-1)Phase IIIAdult glioblastoma (post standard-of-care surgery, chemotherapy, radiotherapy)Primary: Efficacy, safetyNCT06749925Dual ICIs + Chemotherapy + Ultrasound BBB openingBalstilimab (anti-PD-1) + Botensilimab (anti-CTLA-4) + Liposomal doxorubicin + Soncloud-9 ultrasound device for BBB openingPhase IIaNewly diagnosed glioblastoma (post-radiotherapy)Primary: Safety, feasibility;Secondary: EfficacyNCT05864534PD-L1/VEGF bispecific antibody + ChemotherapyBNT327 (Pumitamig, anti-PD-L1 x VEGF-A bispecific antibody) ± TemozolomidePhase IIRecurrent glioblastoma (WHO grade IV, post-radiotherapy + Temozolomide)Primary: Safety; Secondary: EfficacyNCT07297212CAR-NK + IL-15 agonist + Anti-VEGF + TTFieldsNogapendekin alfa inbakicept (IL-15 agonist) + PD-L1 t-haNK (CAR-NK cells) + Bevacizumab + Tumor Treating Fields (TTFields)Phase II/IIBRecurrent or progressive glioblastomaPrimary: Safety;Secondary: EfficacyNCT06061809Allogeneic virus-specific T cells + PD-1 inhibitorAllogeneic CMV-specific T cells + Pembrolizumab (anti-PD-1)Phase I/IINewly diagnosed and recurrent glioblastoma/grade 4 astrocytomaPrimary: MTD, RP2D; Secondary: Anti-tumor activityNCT06157541Abbreviations: AAV, adeno-associated virus; anti-CD137, anti-cluster of differentiation 137; anti-CTLA-4, anti-cytotoxic T-lymphocyte-associated protein 4; anti-PD-1, anti-programmed cell death protein 1; anti-PD-L1, anti-programmed death-ligand 1; AXL, AXL receptor tyrosine kinase; BBB, blood-brain barrier; BiTE, bispecific T-cell engager; CAR, chimeric antigen receptor; CAR-NK, chimeric antigen receptor natural killer; CAR-T, chimeric antigen receptor T-cell; CMV, cytomegalovirus; CNS, central nervous system; DIPG, diffuse intrinsic pontine glioma; DLL3, delta-like ligand 3; DLL4, delta-like ligand 4; DMG, diffuse midline glioma; EGFR, epidermal growth factor receptor; GD2, disialoganglioside 2; HGG, high-grade glioma; HSV-tk, herpes simplex virus thymidine kinase; ICI, immune checkpoint inhibitor; IDH, isocitrate dehydrogenase; IL, interleukin; IL13Rα2, interleukin-13 receptor alpha 2; LIFU, low-intensity focused ultrasound; MET, mesenchymal-epithelial transition factor; MGMT, O⁶-methylguanine-DNA methyltransferase; MTD, maximum tolerated dose; OBD, optimal biological dose; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; RP2D, recommended Phase II dose; TIL, tumor-infiltrating lymphocyte; TTFields, tumor treating fields; VEGF, vascular endothelial growth factor; VEGF-A, vascular endothelial growth factor A; VEGFR, vascular endothelial growth factor receptor. Ongoing clinical trials of immunotherapy combination strategies in glioma Abbreviations: AAV, adeno-associated virus; anti-CD137, anti-cluster of differentiation 137; anti-CTLA-4, anti-cytotoxic T-lymphocyte-associated protein 4; anti-PD-1, anti-programmed cell death protein 1; anti-PD-L1, anti-programmed death-ligand 1; AXL, AXL receptor tyrosine kinase; BBB, blood-brain barrier; BiTE, bispecific T-cell engager; CAR, chimeric antigen receptor; CAR-NK, chimeric antigen receptor natural killer; CAR-T, chimeric antigen receptor T-cell; CMV, cytomegalovirus; CNS, central nervous system; DIPG, diffuse intrinsic pontine glioma; DLL3, delta-like ligand 3; DLL4, delta-like ligand 4; DMG, diffuse midline glioma; EGFR, epidermal growth factor receptor; GD2, disialoganglioside 2; HGG, high-grade glioma; HSV-tk, herpes simplex virus thymidine kinase; ICI, immune checkpoint inhibitor; IDH, isocitrate dehydrogenase; IL, interleukin; IL13Rα2, interleukin-13 receptor alpha 2; LIFU, low-intensity focused ultrasound; MET, mesenchymal-epithelial transition factor; MGMT, O⁶-methylguanine-DNA methyltransferase; MTD, maximum tolerated dose; OBD, optimal biological dose; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; RP2D, recommended Phase II dose; TIL, tumor-infiltrating lymphocyte; TTFields, tumor treating fields; VEGF, vascular endothelial growth factor; VEGF-A, vascular endothelial growth factor A; VEGFR, vascular endothelial growth factor receptor. In summary, glioma immunotherapy faces a series of interrelated and overlapping challenges, including a complex immunosuppressive TME, high tumor heterogeneity, the restrictive nature of the BBB, and tumor adaptive resistance and evolution. In the future, strategies such as multi-target approaches, optimization of delivery systems, remodeling of the TME, and the development of rational combination therapies hold promise for overcoming these obstacles one by one, ultimately achieving breakthroughs in glioma immunotherapy. 6. Biomarkers: From Classification to Combination Strategies The biomarkers discussed in this manuscript have distinct functional roles in clinical practice. To establish a unified classification framework, this section categorizes them into three groups: predictive biomarkers, prognostic biomarkers, and biomarkers guiding combination therapy selection. 6.1 Predictive Biomarkers Predictive biomarkers are defined as biological characteristics that can be objectively measured and indicate the likelihood of response to a specific therapeutic intervention 175. These markers are critical for patient selection in targeted therapy and immunotherapy, enabling the identification of patients most likely to derive clinical benefit while minimizing unnecessary toxicity 175. Table 3 summarizes key predictive biomarkers discussed in this review, along with their associated therapies and current evidence. Table 3Predictive biomarkers for therapeutic responseTherapeutic StrategyBiomarkerPredictive RoleCorresponding StudyCancer VaccineHLA-A2, HLA-A1HLA typing for patient selectionICT-107 25IDH1, TERT, B7-H4IDH1ʷᵗTERTᴹᵀ subgroup showed OS/PFS benefit; low B7-H4 subgroup showed OS benefitGSC-DCV 26MGMT methylationMethylated patients: mOS 41.4 months vs 16.5 months in unmethylatedSurVaxM + adj TMZ 34MGMT methylationAmong ISR⁺ patients, methylated: mOS 28.3 months vs 14.8 months in unmethylatedIMA950 19HLA-A*24:02, WT1Patient selection based on HLA typing and WT1 expressionWT1 cocktail vaccine 38HLA-A*24:02Patient selection based on HLA typingVEGFRs peptide vaccine + TMZ 39IDH1 mutationTargeting IDH1-mutant glioma patientsPersonalized mutIDH1 multi-peptide vaccine 20Immune Checkpoint InhibitorMGMT methylationMethylated patients: mOS 35.7 months vs 12.6 months in unmethylatedNivolumab + Ipilimumab (pre-radiation) 68CAR T-Cell TherapyEGFRvIII, MGMT unmethylatedPatient selection based on EGFRvIII expression; MGMT unmethylated status for patient selectionCART-EGFRvIII + Pembrolizumab 13; CART-EGFRvIII cells 14EGFR amplification, IDH wild-typePatient selection based on EGFR amplification and IDH wild-type statusCART-EGFR-IL13Rα2 110IL13Rα2Patient selection based on IL13Rα2 expressionGRm13Z40-2 cells 112GD2, IDH wild-typePatient selection based on GD2 expression; IDH wild-typeGD2-specific 4SCAR-T 113MMP-2Patient selection based on MMP-2 expressionCLTX-CAR T cells 114Oncolytic VirotherapyISG signatureISG signature score inversely correlated with viral replication, predicting treatment responseMV-CEA 136TMEmedium subtypeTreatment response observed only in the "TMEmedium" subtype (exploratory)DNX-2401 + Pembrolizumab 132Abbreviations: adj TMZ, adjuvant temozolomide; B7-H4, B7 homolog 4; CAR-T, chimeric antigen receptor T-cell; EGFR, epidermal growth factor receptor; EGFRvIII, epidermal growth factor receptor variant III; GD2, disialoganglioside 2; GSC-DCV, glioma stem cell-derived dendritic cell vaccine; HLA, human leukocyte antigen; IDH, isocitrate dehydrogenase; IL13Rα2, interleukin-13 receptor alpha 2; ISG, interferon-stimulated gene; ISR, immune response; MGMT, O⁶-methylguanine-DNA methyltransferase; MMP-2, matrix metalloproteinase-2; mOS, median overall survival; OS, overall survival; PFS, progression-free survival; TERT, telomerase reverse transcriptase; TME, tumor microenvironment; TMZ, temozolomide; VEGFR, vascular endothelial growth factor receptor; WT1, Wilms tumor 1. Predictive biomarkers for therapeutic response Abbreviations: adj TMZ, adjuvant temozolomide; B7-H4, B7 homolog 4; CAR-T, chimeric antigen receptor T-cell; EGFR, epidermal growth factor receptor; EGFRvIII, epidermal growth factor receptor variant III; GD2, disialoganglioside 2; GSC-DCV, glioma stem cell-derived dendritic cell vaccine; HLA, human leukocyte antigen; IDH, isocitrate dehydrogenase; IL13Rα2, interleukin-13 receptor alpha 2; ISG, interferon-stimulated gene; ISR, immune response; MGMT, O⁶-methylguanine-DNA methyltransferase; MMP-2, matrix metalloproteinase-2; mOS, median overall survival; OS, overall survival; PFS, progression-free survival; TERT, telomerase reverse transcriptase; TME, tumor microenvironment; TMZ, temozolomide; VEGFR, vascular endothelial growth factor receptor; WT1, Wilms tumor 1. 6.2 Prognostic Biomarkers 6.2.1 IDH-wildtype glioblastoma MGMT promoter methylation is one of the most well-established prognostic biomarkers in IDH-wildtype glioblastoma, with its methylation status associated with significant improvements in OS and PFS 176, and it retains prognostic value even in the context of adjunctive therapies such as tumor-treating fields 177. TERT promoter mutations occur in approximately 80% of IDH-wildtype glioblastomas 178; however, their prognostic significance remains controversial. Some studies have associated TERT promoter mutations with poorer survival 179, whereas others suggest a synergistic interaction between TERT promoter mutations and MGMT methylation status, with patients harboring both alterations exhibiting the longest survival 180. In addition, chromosome 7 gain and chromosome 10 loss, along with EGFR amplification, are characteristic genomic alterations in IDH-wildtype glioblastoma 181 and are closely associated with worse OS and PFS 182. PTEN mutations, observed in approximately 40% of cases, are likewise associated with poor prognosis and treatment resistance 183. 6.2.2 IDH-mutant astrocytomas In IDH-mutant diffuse astrocytomas, tumor grade is a key prognostic factor, with grade 3 tumors exhibiting a median OS of 8.1 years, whereas grade 4 tumors show a significantly shorter median OS of 4.7 years (p < 0.05) 184. Homozygous CDKN2A/B deletion is strongly associated with worse survival outcomes; patients with this deletion have a median OS of only 1.8 years compared with 5.5 years in those without the deletion (p < 0.001) 2. The presence of IDH mutation has now been established as a favorable prognostic factor in diffuse astrocytomas and oligodendrogliomas 185. MGMT promoter methylation is more prevalent in IDH-mutant gliomas and is independently associated with improved OS and PFS when treated with TMZ; however, its predictive value appears to be largely limited to grade 4 tumors 186. In addition, TP53 mutations and ATRX loss frequently co-occur, indicating genomic instability, and are generally associated with favorable prognosis 187; patients with ATRX mutations are younger and have longer OS 188. 6.2.3 IDH-mutant and 1p/19q-codeleted oligodendroglioma Oligodendrogliomas have long been recognized as having a favorable prognosis, with their defining feature being the concurrent presence of IDH1/2 mutations and 1p/19q codeletion 2. The molecular landscape of oligodendrogliomas includes frequent TERT promoter mutations 189, retained ATRX expression, and absence of p53 accumulation, which is consistent with the mutual exclusivity between 1p/19q codeletion and TP53 or ATRX alterations 190. In a subset of cases, CDKN2A/B deletions may be detected, an alteration associated with more aggressive tumor behavior and poorer prognosis 191. 6.3 Biomarkers for Combination Therapy Selection The rational design of combination therapies increasingly relies on biomarkers that identify synergistic vulnerabilities. Table 4 outlines representative biomarkers and their associated combination approaches. Table 4Biomarker guiding combination therapy selectionTherapeutic StrategyBiomarkerPredictive RoleCorresponding StudyCancer VaccineMGMT methylationMethylated patients: mOS 41.4 months vs 16.5 months in unmethylatedSurVaxM + adj TMZ 34MGMT methylationAmong ISR⁺ patients, methylated: mOS 28.3 months vs 14.8 months in unmethylatedIMA950 19Immune Checkpoint InhibitorMGMT methylationMethylated patients: mOS 35.7 months vs 12.6 months in unmethylatedNivolumab + Ipilimumab (pre-radiation) 68ESTIMATE Immune Score, molecular subtypeHigh immune score tumors: mOS 24.8 vs 14.5 months; mesenchymal subtype: mOS 26.5 monthsAtezolizumab + RT/TMZ 70PLR, MUC16, H3F3B, SRSF2Low PLR correlated with improved PFS/OS; MUC16 mutation, H3F3B amplification, SRSF2 amplification associated with worse OSTislelizumab + Low-Dose Bevacizumab 72, 73CAR T-Cell TherapyEGFRvIII, MGMT unmethylatedPatient selection based on EGFRvIII expression; MGMT unmethylated status for patient selectionCART-EGFRvIII + Pembrolizumab 13; CART-EGFRvIII cells 14EGFR amplification, IDH wild-typePatient selection based on EGFR amplification and IDH wild-type statusCART-EGFR-IL13Rα2 110IL13Rα2Patient selection based on IL13Rα2 expressionGRm13Z40-2 cells 112Oncolytic VirotherapyISG signatureISG signature score inversely correlated with viral replication, predicting treatment responseMV-CEA 136TMEmedium subtypeTreatment response observed only in the "TMEmedium" subtype (exploratory)DNX-2401 + Pembrolizumab 132Abbreviations: adj TMZ, adjuvant temozolomide; CAR-T, chimeric antigen receptor T-cell; EGFR, epidermal growth factor receptor; EGFRvIII, epidermal growth factor receptor variant III; ESTIMATE, Estimation of Stromal and Immune cells in Malignant Tumors using Expression data; H3F3B, H3.3 histone B; IDH, isocitrate dehydrogenase; IL13Rα2, interleukin-13 receptor alpha 2; ISG, interferon-stimulated gene; ISR, immune response; MGMT, O⁶-methylguanine-DNA methyltransferase; MMP-2, matrix metalloproteinase-2; mOS, median overall survival; MUC16, mucin 16; OS, overall survival; PFS, progression-free survival; PLR, platelet-to-lymphocyte ratio; RT, radiotherapy; SRSF2, serine and arginine rich splicing factor 2; TME, tumor microenvironment; TMZ, temozolomide. Biomarker guiding combination therapy selection Abbreviations: adj TMZ, adjuvant temozolomide; CAR-T, chimeric antigen receptor T-cell; EGFR, epidermal growth factor receptor; EGFRvIII, epidermal growth factor receptor variant III; ESTIMATE, Estimation of Stromal and Immune cells in Malignant Tumors using Expression data; H3F3B, H3.3 histone B; IDH, isocitrate dehydrogenase; IL13Rα2, interleukin-13 receptor alpha 2; ISG, interferon-stimulated gene; ISR, immune response; MGMT, O⁶-methylguanine-DNA methyltransferase; MMP-2, matrix metalloproteinase-2; mOS, median overall survival; MUC16, mucin 16; OS, overall survival; PFS, progression-free survival; PLR, platelet-to-lymphocyte ratio; RT, radiotherapy; SRSF2, serine and arginine rich splicing factor 2; TME, tumor microenvironment; TMZ, temozolomide. 7. Conclusions Recent progress in glioma immunotherapy has been substantial, particularly in efficacy 136, 139, BBB penetration 47-50, 76, TME remodeling 47, 49, 55, 57, 60, overcoming heterogeneity 46, 115, and controlled release with reduced toxicity 144. However, several critical research gaps remain to be addressed. 1) Target selection. Many targets, such as EGFRvIII, exhibit unstable expression and are prone to loss following treatment, leading to resistance 31. An ideal target should be closely associated with tumor cell growth, proliferation, and invasion, while maintaining stable expression that is less likely to be lost 192. Therefore, the identification and screening of optimal targets constitute a critical priority, as this will help mitigate off-target effects and reduce the incidence of therapeutic resistance. 2) Biomarker identification and patient stratification. The identification of reliable biomarkers in glioma faces unique challenges compared with many other solid tumors. (a) The presence of the blood-brain barrier (BBB) severely restricts the release of tumor-derived biomolecules into the peripheral circulation, resulting in extremely low abundance of circulating biomarkers, including circulating tumor DNA (ctDNA) and circulating tumor cells (CTCs) 193, 194. Studies have shown that the detection rate of plasma ctDNA in patients with glioma is currently below 20%, which is considerably lower than that reported for other malignancies 194. (b) Glioma exhibits high tumor heterogeneity, making it difficult for a single biopsy or a single biomarker to fully capture the overall tumor status 195. This characteristic often manifests in clinical trials as a lack of statistically significant overall outcomes or statistically significant efficacy only in specific subgroups, suggesting that current patient stratification strategies remain insufficiently refined 49. (c) The TME of glioma is highly immunosuppressive, characterized by a low tumor mutational burden (TMB) and a paucity of high-quality neoantigens, which further limits the identification and application of immune-related biomarkers 196. (d) The expression of certain biomarkers in glioma is dynamic, whereas clinical detection still relies primarily on pathological biopsy, making it difficult to capture real-time changes during treatment 197. (e) Owing to the infiltrative growth pattern of glioma, obtaining tissue samples carries significant risks; although liquid biopsy holds promise, standardized methods for its application have not yet been established 197. 3) Preclinical model selection. The representativeness of certain models is limited. In immunotherapy studies in particular, a fully competent immune system is essential, yet the establishment of humanized models often relies on immunodeficient mice, which compromises the representativeness of efficacy assessments 48. Selecting models that are reproducible, amenable to visualization, biomimetic, and easy to establish is crucial for facilitating clinical translation 198. 4) Drug design. Drug design should take into account patient compliance, prioritizing convenient routes of administration. From the early stages of design, the involvement of pharmaceutical analysis experts is essential to develop analytical protocols that support preparation, purification, and scalable manufacturing. Additionally, aligning drug development with real-world manufacturing processes and patient needs is critical. The establishment of versatile platforms, such as the mSA2 system, can help shorten the research and development cycle and reduce costs 115. 5) Drug development. The safety of CAR-T therapy requires further validation 106-109. Enhancing the persistence of CAR-T cells in vivo 112, 114 and preventing the rapid clearance of oncolytic viruses by the host 145 remain key challenges. Currently, glioma immunotherapy continues to be constrained by the BBB, high tumor heterogeneity, the immunosuppressive microenvironment, and adaptive resistance accompanied by tumor evolution 13, 152-157. Future progress is expected through improved delivery routes, combination strategies, the identification of optimal targets, and remodeling of the immune microenvironment. 6) Clinical trial considerations. (a) Patient selection bias. Enrolling patients without disease progression following standard therapy as trial subjects, compared with general patient populations, may lead to an overestimation of treatment efficacy 32. (b) Control arm selection. With the advancement of standard-of-care treatments, OS in historical control groups has improved, making it difficult to reproduce promising results from early-phase trials in subsequent randomized studies 31. (c) Control arm design. Immunotherapy trials should use standard therapy—at minimum TMZ—as the comparator to determine whether the investigational treatment offers improvements over existing therapies 66. (d) Insufficient sample size. The low incidence and high mortality rate of gliomas make patient recruitment challenging, limiting the feasibility of many trials and resulting in underpowered analyses 26, 46, 114, 132. Multicenter international collaboration is urgently needed, along with efforts to standardize treatment protocols. (e) Contradictory findings. Some studies suggest that immunotherapy requires minimal residual disease 46, whereas others have observed long-term survival benefits in patients without complete resection 31. The rindopepimut trial found no correlation between antibody titers and survival benefit, suggesting that T-cell responses may serve as a more relevant selection criterion 31; conversely, most trials, including AFTV, have reported that higher antibody levels correlate with better outcomes 46. 7) Efficacy evaluation. Current imaging modalities are insufficient for assessing treatment response, highlighting the need for novel techniques 139. Certain therapies, such as AFTV, contain theoretically all tumor antigens, making it impossible to evaluate efficacy based on specific antibody titers, necessitating the establishment of new evaluation criteria 46. 8) Clinical translation. The production, quality control, and management of novel biological products require standardization 52. Furthermore, biomarker-guided stratification or individualized treatment approaches are costly and require policy support 7. Although each of the four major immunotherapy strategies faces distinct translational challenges 51, 52, 86, 125, 151, oncolytic virus therapy currently demonstrates the most promising translational potential. This approach has garnered significant attention due to its unique dual mechanism of "oncolysis plus immune activation" 199, 200. G47Δ has been approved in Japan for the treatment of malignant glioma, with a Phase II trial reporting a one-year survival rate of 84.2% and a median OS of 20.2 months, marking a milestone in the field 139. Furthermore, the combination of oncolytic viruses with ICIs has shown synergistic effects in preclinical and early-phase clinical studies, with the potential to convert "cold" tumors into "hot" tumors 134. In contrast, ICIs are unlikely to confer additional survival benefits in the absence of molecular screening or specific biomarker guidance 43, 74, 75. Tumor vaccines are limited in monotherapy due to tumor heterogeneity and antigen loss; however, their value as components of combination regimens should not be overlooked 31. Adoptive cell therapies similarly face challenges such as antigen loss and limited in vivo persistence, resulting in modest efficacy as monotherapies 13, 112, 114. Nonetheless, the recently developed mSA2 universal platform offers a promising strategy to address antigen loss by enabling simple exchange of biotinylated antibodies, providing new opportunities for this therapeutic modality 115. Acknowledgments All figures in this review were created using Figdraw. Author Contributions JG: Conceptualization, Writing-review & editing, Supervision, Funding acquisition, Visualization. YH: Writing-review & editing, Supervision, Funding acquisition, Visualization. JL: Writing-original draft, Writing-review & editing, Formal analysis, Validation, Project administration. GY: Writing-review & editing, Formal analysis. ZC, HQ, and MZ: Writing-review & editing, Software. CZ and XL: Writing-review & editing, Data curation, Investigation. Abbreviations Adverse Events Autologous Formalin-fixed Tumor Vaccine Checkpoint Antibody Receptor-modified Chimeric Antigen Receptor T-cell Actin Related Protein 2/3 Complex Subunit 1B Autologous Tumor Antigen Blood-Brain Barrier Bromodomain-containing Protein 9 Chimeric Antigen Receptor T-cell Clofazimine Central Nervous System Damage-associated molecular patterns DNA Damage Response Difluoromethylornithine Dendritic Cells Double-stranded RNA Epidermal Growth Factor Receptor Variant III Extracellular Vesicles Glioma-associated Antigen Glioma Active Personalized Vaccine Consortium Glycoprotein A Repetitions Predominant Glioblastoma Glioblastoma Organoid Glioma stem cell Human Cytomegalovirus Human Leukocyte Antigen A2 Herpes Simplex Virus Type 1 Infectious Bursal Disease Virus Immune Checkpoint Blockade Immunogenic Cell Death Immune Checkpoint Inhibitors Isocitrate Dehydrogenase Isocitrate Dehydrogenase 1/2 Isocitrate Dehydrogenase 1 Mutation-targeted Vaccine Interferon-stimulated Gene intravenous delivery Janus Kinase Keyhole Limpet Hemocyanin Lymphokine-activated Killer Lipid Nanoparticles Multi-cationic Protein Myeloid-derived Suppressor Cells Methyltransferase-like 14 O-6-Methylguanine-DNA Methyltransferase Major histocompatibility complex Monomeric Streptavidin 2-based Chimeric Antigen Receptor T-cells Measles Virus-based Carcinoembryonic Antigen-expressing Oncolytic Virus Myxoma Virus Nanoshield Vaccine Newly Diagnosed Glioblastoma Nitric Oxide Non-secreting Interleukin-12 Cholesterol-modified Adenovirus Oncolytic Herpes Simplex Virus-1 Overall Survival Pathogen-associated molecular patterns Peripheral blood mononuclear cells Photodynamic Therapy Progression-Free Survival Personalized Peptide Vaccine Poliovirus (Recombinant Nonpathogenic Poliovirus: Rhinovirus Recombinant) Recurrent Glioblastoma Reactive Oxygen Species Single-chain variable fragment Survivin Peptide-CpG Nanovaccine Sonodynamic Nanovaccine Signal Transducer and Activator of Transcription Survivin-targeting Peptide Vaccine Tumor-associated Antigens Tumor-associated Macrophages Tumor Cavity Fluid T Cell Receptor-engineered T Cells Telomerase Reverse Transcriptase Transforming Growth Factor Beta 2 Transforming Growth Factor Beta Tumor-infiltrating Lymphocytes Tumor microenvironment Temozolomide Regulatory T Cells Tumor-specific Antigens World Health Organization Wilms Tumor 1 Zika Virus Zinc Finger Protein 638 References