Endogenous immune recruitment in glioblastoma CAR T therapy: cytokine, myeloid, and chemokine circuitry.

Introduction

Glioblastoma (GBM) is the most prevalent and aggressive primary malignant brain tumor in adults with a median overall survival rate around 12–16 months despite aggressive treatment strategies involving maximal surgical resection, radiation therapy and chemotherapy [ 1 ]. Immune checkpoint therapy that has demonstrated significant promise in melanoma and lung cancer have failed to produce meaningful clinical responses in GBM [ 2 – 4 ]. Given the lack of promising therapeutic approaches targeting GBM, novel treatment regimens are being explored including combining multiple checkpoint inhibitors and tumor-specific vaccines [ 5 , 6 ]. However, significant barriers remain owing to the complexity of GBM necessitating the development of innovative treatment strategies.

Chimeric Antigen Receptor (CAR) T-cell therapy represents a groundbreaking approach in cancer immunotherapy leveraging advances in immune engineering and targeted tumor therapy. CAR T-cell therapy involves genetically engineering a patient’s T cells to express receptors that target specific antigens expressed on tumor cells [ 7 , 8 ]. Engineering CAR T-cells begins with the extraction of the patient’s own T cells, followed by their modification using viral vectors. The resultant T cells possess an extracellular domain, usually derived from an antibody’s single-chain variable fragment (scFv), enabling them to recognize tumor antigens directly. Upon binding, these CAR T cells activate and proliferate through activation of signaling pathways associated with the intracellular CD3ζ domain and additional co-stimulatory signals from molecules such as CD28 or 4-1BB, which enhance T cell activation and persistent signaling. In essence, CAR-T cells form a non-classical immune synapse that mediate cytotoxic anti-tumor effects by engaging multiple signaling axes including perforin and granzyme, the Fas and Fas ligand pathway as well as contributing to the release of cytokines that modulate the tumor stroma [ 9 ].

CAR T-cell therapy has demonstrated significant success in hematologic malignancies, particularly targeting CD19 in B-cell neoplasms. Approved therapies include tisagenlecleucel (Kymriah) for B-cell acute lymphoblastic leukemia and axicabtagene ciloleucel (Yescarta) for large B-cell lymphoma [ 9 , 10 ]. In recent years, there has been emerging interest in applying CAR T-cell therapy to various solid tumors, including GBM, with ongoing clinical trials exploring various targets beyond CD19 [ 8 , 11 , 12 ].

While CAR T-cell therapies have shown promise in treating hematological cancers, several challenges persist in their application for effective treatment of GBM. A major hurdle is antigen escape, where tumor cells may downregulate or lose the target antigen after therapy initiation, leading to resistance [ 13 – 17 ]. Tumor heterogeneity in GBM leads to varying, context-dependent antigen expression within a single tumor, complicating effective targeting [ 18 – 20 ]. Furthermore, the immunosuppressive microenvironment of GBM, enriched with myeloid-derived suppressor cells and regulatory T cells, further inhibits T cell activity [ 21 – 24 ]. T cell exhaustion, characterized by diminished cytotoxic functionality following prolonged exposure to antigen, presents a significant barrier [ 25 , 26 ].

To achieve durable clinical responses, it is essential to engage not only CAR T cells but also the host’s endogenous immune system. Combining CAR T-cell therapy with strategies that recruit and activate other immune cells, for instance by enrichment of CAR T-cells with immune stimulatory cytokines, could augment the overall antitumor response [ 27 , 28 ]. This synergistic interaction can promote long-term remission and facilitate antigen spreading [ 29 , 30 ], supporting the maintenance of immune pressure on tumor cells and preventing relapse.

Our review will address the challenges of CAR T-cell monotherapy in treating GBM and elucidate strategies for enhancing CAR T-cell efficacy through recruitment and activation of endogenous immune cells. Augmenting the recruitment of endogenous immune cells may provide a pathway for enhancing the design and application of CAR T-cell therapies in the complex landscape of glioblastoma to improve efficacy and sustained remission in patients.

In addition to summarizing the mechanistic and translational literature, we conducted a structured meta-synthesis of CAR T-cell clinical trials in glioblastoma using publicly available registries to provide context for the current development pipeline. Utilizing a prespecified keyword approach to search the ClinicalTrials.gov database, we screened 91 articles and identified 44 interventional trials of CAR T in patients with GBM, ultimately selecting 23 trials initiated after January 2020 for our analysis. This analysis was conducted in order to provide context for trial motifs of contemporary design, such as those involving novel target antigens, multi-antigen or OR-gated strategies, SynNotch/AND-gated approaches, and inducible safety strategies.

CAR-T therapy in hematologic vs. solid tumors

Hematologic malignancies

CAR-T cell therapy has delivered unprecedented remission rates in B-cell malignancies like acute lymphoblastic leukemia (ALL) and non-Hodgkin lymphoma [ 31 – 33 ]. CAR-T cells targeting CD19 have induced deep, long-term remissions, with many patients remaining disease-free years after treatment [ 34 – 36 ]. Similarly, anti-BCMA CAR-T cells for multiple myeloma have achieved durable responses, including multi-year progression-free survival in a subset of patients [ 37 ]. These outcomes suggest that CAR-T cells can provide sustained immune surveillance against blood cancers. Notably, the host immune system appears to contribute to these lasting remissions [ 37 ]. CAR-T cells secrete inflammatory cytokines (e.g. IFN-γ) and cause tumor lysis, which can recruit and activate endogenous immune effectors. The recruitment of the host immune system secondary to CAR-T cell-mediated killing is termed antigen/epitope spreading [ 38 , 39 ]. In leukemia and lymphoma, which typically lack a profound immunosuppressive niche, such host immunity engagement may eliminate antigen-negative escape variants and reinforce the therapeutic effect. Longitudinal studies in myeloma patients found that sustained remissions were associated with robust CAR-T expansion and a preserved repertoire of host T cells, alongside low levels of suppressive myeloid cells [ 37 , 40 ]. This interplay between engineered T cells and the patient’s immune system underscores how CAR-T therapy in hematologic malignancies can orchestrate a broad, lasting anti-tumor immune response.

Solid tumors

In solid cancers, CAR-T therapy has faced major hurdles due to the tumor’s physical and immunosuppressive barriers. Unlike disseminated blood malignancies, solid tumors form dense masses that are hard for T cells to penetrate [ 41 – 44 ]. Fibrotic stroma, abnormal vasculature, and extracellular matrix components create physical hurdles that impede CAR-T cell infiltration [ 45 – 47 ]. Moreover, the tumor microenvironment (TME) actively suppresses anti-tumoral immune activity through a variety of suppressive mechanisms. Tumor and stromal cells often upregulate immune checkpoint ligands such as PD-L1, which engage PD-1 on T cells to blunt their activity [ 48 – 50 ]. They also exploit metabolic immune evasion: indoleamine 2,3-dioxygenase 1 (IDO1) is frequently expressed, depleting tryptophan and accumulating kynurenine – conditions that starve and paralyze effector T cells [ 51 – 53 ]. Additionally, malignant and stromal cells secrete immunosuppressive cytokines like TGF-β, and chemokines that recruit regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs) into the tumor [ 54 – 56 ]. These components of the TME collectively form a milieu that suppresses T cell–mediated cytotoxicity. The net effect is that CAR-T cells in solid tumors often become functionally exhausted or excluded before eliminating the cancer. Consistent with this, clinical studies report that CAR-T cells exhibit poor expansion and persistence in solid tumor patients compared to hematologic settings [ 15 , 57 ]. Thus, the immunologic landscape of solid tumors, marked by physical exclusion and active immune suppression (via PD-L1, IDO1, TGF-β pathways and others), remains a fundamental barrier to replicating the success of CAR-T seen in blood cancers.

Lastly, an important factor for why CAR therapy has shown the most reproducible efficacy so far in hematologic malignancies is that the target expression is more uniform and clonally preserved compared with most solid malignancies. Antigens like CD19, CD20, CD22, and BCMA are more universally and highly expressed on malignant cells, thereby minimizing the effects of intratumoral antigen heterogeneity and the likelihood of developing antigen loss variants early on. Also, malignant cells are more easily accessible compared with solid malignancies, thereby overcoming significant barriers that make CAR therapy ineffective for solid malignancies [ 15 ].

Lessons from CAR-T therapy

An emerging principle in CAR-T therapy is the importance of antigen spreading for durable cancer control. The resulting inflammatory tumor cell death from CAR-T cell therapy can expose a breadth of tumor antigens to the immune system. This can activate new anti-tumor T cell responses beyond the CAR’s target specificity [ 39 , 58 ]. Additionally, CAR-T cells in these patients often persist as memory-like cells, providing ongoing surveillance. In solid tumors, however, the development of secondary immune responses is curtailed by the suppressive TME. The presentation of released antigens to T cells can be inefficient due to dysfunctional antigen-presenting cells and inhibitory signals in the TME. The contrast is clear: in liquid cancers, CAR-T therapy can catalyze a wider endogenous immune attack and enduring remission, whereas in solid tumors the lack of immune propagation means that CAR-T cells bear the full therapeutic burden. Future CAR-T strategies are therefore focusing on increasing inflammation in the TME (i.e. “immune hot”) by combining CAR-T cells with checkpoint inhibitors or cytokine modulation to promote infiltration, inflammation, and epitope spreading within solid tumors [ 59 – 62 ]. By learning from the immune dynamics of successful CAR-T cases, researchers aim to engineer therapies that not only directly kill tumor cells but also reshape the host-tumor interaction, bridging the efficacy gap between hematologic and solid cancers.

Mechanisms of endogenous immune recruitment

Despite CAR T-cells targeting specific antigens, recruitment of the endogenous immune system is essential for achieving a robust therapeutic response in GBM. Cytokines are key immunomodulatory molecules that play a central role in this process, enabling and enhancing CAR T-cell antitumor activity by mobilizing and activating endogenous immune populations. Interferon-γ (IFN-γ) is a key pro-inflammatory and anti-tumoral cytokine that shapes endogenous myeloid responses by promoting macrophage polarization toward an M1 phenotype and enhancing dendritic-cell antigen presentation through upregulating MHC classes I and II [ 63 ]. IFN-γ signaling is also critical for cross-talk between CAR T-cells and the host immune system. This bidirectional cytokine exchange is central to therapeutic efficacy.

In addition, interferon-γ also directly regulates endogenous immune cell recruitment through a chemokine and adhesion gene program to facilitate the trafficking of immune cells to the tumor. In a tumor microenvironment, IFN-γ signaling can increase the expression of ligands of CXCR3 chemokines, such as CXCL9, CXCL10, and CXCL11, as well as adhesion molecules such as ICAM-1/VCAM-1, to facilitate effector T cell and NK cell infiltration [ 64 , 65 ]. In a GBM microenvironment, in which overcoming blood-brain barriers as well as immune cell exclusion represents a major barrier, this inflamed response induced by IFN-γ secretion can potentially increase this microenvironment’s permeability to a certain degree, shifting it from a poorly permeable to a relatively permeable microenvironment to facilitate a higher degree of antigen presentation following cell death induced by CAR T cell cytolysis [ 63 , 65 , 66 ]. Nevertheless, chronic IFN-γ signals can induce adaptive resistance gene expression, such as PD-L1 expression, to potentially suppress efficacy as well as provide a rationale to combine IFN-γ with strategies to overcome checkpoint blockade resistance.

In syngeneic GBM models, IFN-γ was required not only for CAR T cell–mediated tumor control but also for activating endogenous immune populations to mount complementary antitumor responses [ 65 ]. Interleukin-12 (IL-12) and interleukin-18 (IL-18) are proinflammatory cytokines that synergize to amplify innate antitumor immunity. Together, they activate and recruit NK cells, enhance NK cytotoxicity, and help overcome Treg-mediated suppression. Activated NK cells then produce high levels of IFN-γ, which drives macrophage polarization toward the proinflammatory M1 phenotype and further suppresses Tregs [ 67 , 68 ]. T cells redirected for antigen-unrestricted cytokine-initiated killing (TRUCKs) are CAR T-cells that can release cytokines to modulated the tumor microenvironment (TME) [ 69 ]. TRUCKs enable CAR T cells to remodel the tumor microenvironment and generate productive inflammation, thereby recruiting endogenous immune cells and promoting antitumor activity even against tumor cells not directly targeted by the CAR.

TRUCKs have been engineered to secrete IL-12, thereby amplifying IFN-γ production, enhancing the antitumor activity of T cells and NK cells, and suppressing Treg function [ 70 ]. In GBM models utilizing EGFRvIII specific CAR T-cells, IL-12 coadministration augmented treatment efficacy. This combination also activated endogenous immune populations, shifting the TME from immunosuppressive to proinflammatory by reducing CD4⁺ Tregs and increasing CD4⁺ effector T cells. IL-12 further expanded intratumoral myeloid cells, which the authors proposed contributed to a more inflammatory TME through enhanced antigen presentation [ 71 ]. TRUCKs engineered to release IL-18 have shown success in treating pancreatic and lung tumors that were previously unresponsive to CAR T therapy without supportive cytokines. IL-18 secretion was associated with increased inflammatory macrophage and NK-cell populations, along with reduced Tregs and immunosuppressive dendritic cells. In contrast, IL-12 and IL-21 were less effective at eliciting these responses in the models tested [ 70 , 72 ]. Although these findings are promising, IL-18–based TRUCKs have not yet been thoroughly investigated in the context of GBM.

Interleukin-2 (IL-2) is essential for the proliferation and survival of activated T cells and NK cells. High levels of IL-2 support the expansion of effector lymphocytes and can even reverse Treg-mediated suppression of immune responses [ 67 , 73 ]. IL-2 has been explored as an adjunct in CAR T-cell therapy. In patients receiving EGFRvIII-targeted CAR T cells ( NCT01454596 ), coadministration of intravenous IL-2 produced limited clinical benefit [ 74 ]. Similarly, IL13Rα2-targeted CAR T cells engineered to resist glucocorticoids and delivered with intracranial IL-2 induced only modest responses in patients with GBM [ 75 ], suggesting that IL-2 may provide some support but is insufficient as a standalone adjuvant. Overall, the therapeutic value of IL-2 coadministration in CAR T-cell treatment for GBM remains unclear.

Tumor necrosis factor-α (TNF-α) is a potent pro-inflammatory cytokine that promotes tumor cell apoptosis and drives inflammatory responses by binding to its cell-surface receptors and activating programmed cell-death pathways [ 76 ]. In glioma-bearing mice, treatment with IL13Rα2-targeted CAR T-cells increased intratumoral IFN-γ and TNF-α levels and expanded CD8⁺ T cells, which in turn promoted dendritic-cell trafficking into the CNS and helped establish an inflammatory TME. Notably, this CAR T monotherapy did not reduce Treg populations [ 77 ]. Similarly, in a human patient with recurrent GBM ( NCT02208362 ), IL13Rα2 CAR T-cell therapy induced increases in IFN-γ, TNF-α, and IL-2 but produced only a transient clinical response [ 16 ].

Granulocyte macrophage-colony stimulating factor (GM-CSF) is known for supporting myeloid pro-inflammatory activity and is secreted by a broad range of cells [ 78 , 79 ]. In murine ovarian tumor models, CAR T cells engineered to secrete GM-CSF and IFN-γ increased the abundance of tumor-associated macrophages, which contributed to enhanced antitumor activity [ 80 ]. To our knowledge, this effect has not yet been demonstrated in GBM.

In addition to cytokine-mediated support of endogenous immunity, CAR T cells can promote antigen spread, thereby facilitating a broader antitumor response beyond the specific target antigen. Antigen spread occurs when tumor cells are lysed, releasing a diverse array of tumor-associated antigens that can be captured and presented by antigen-presenting cells, ultimately activating additional endogenous immune populations. This mechanism may help explain how CAR T cells directed against a single antigen can induce a more generalized antitumor response [ 81 – 83 ]. Together, cytokine-driven immune recruitment and antigen spread highlight the capacity of CAR T-cell therapies to reshape the broader antitumor immune landscape.

From the perspective of cytokine support in CAR T-cell therapy, tertiary lymphoid structures (TLSs) represent a promising mechanism for amplifying cytokine-mediated antitumor responses. TLSs are organized aggregates of lymphoid and stromal cells that arise in chronically inflamed tissues, and their presence within tumors is generally associated with improved prognosis [ 84 , 85 ]. TLSs have been identified in human gliomas, and because the brain parenchyma lacks lymph nodes to support localized immune activation, these structures may serve as focal sites for antigen presentation and immune-cell priming [ 85 ]. Sabahi et al. describe an approach to generate inducible TLSs using an implantable biodegradable scaffold designed to retain antigen-containing tissue and recruit immune cells, thereby promoting TLS formation. Once established, such TLSs could provide sustained, localized immune support to enhance CAR T-cell activity [ 86 ]. This strategy offers a novel means of counteracting the immunosuppressive GBM TME by fostering sustained, site-specific antitumor immunity. Such an approach could further augment CAR T-cell therapies that rely on cytokine release to generate a broader and more durable antitumor response. Table 1 recapitulates the endogenous immune recruitment program initiated during GBM CAR T therapy.

Immune cell population regulation during CAR T therapy

Clinical trials in GBM evaluating CAR T cells indicate that these therapies not only directly attack tumor cells but also modulate the patient’s endogenous immune compartment by upregulating and downregulating various lymphocyte populations. For example, O’Rourke and colleagues conducted a 10-patient phase I study of EGFRvIII-CAR T cells ( NCT02209376 ), which showed expansion of endogenous CD8 + T cells within the tumor. This was accompanied by markers of activation and proliferation, namely IFN-γ, granzyme B, and Ki-67. Concurrently, there was a substantial increase in FOXP3 + Tregs and elevated expression of immunosuppressive molecules such as TGF-β, PD-L1, and IL-10 within the tumor along with persistence of myeloid infiltration, suggesting an adaptive resistance response to CAR T therapy [ 15 ].

In a separate phase I trial evaluating locoregional delivery of IL13Rα2 CAR T cells ( NCT01082926 ), cerebrospinal fluid showed transient elevations in IFN-γ and associated chemokines CXCL9, CXCL10, and IL-12, consistent with a Th1-polarized inflammatory response following CAR T infusion [ 75 ]. Although CXCR3 + immune cells, such as CD8 + and NK cells, are known to be responsive to these chemokines, the study did not demonstrate increased bystander immune cell trafficking into tumor tissue.

Another recent phase I trial of EGFRvIII-targeted CAR T cells in newly diagnosed GBM combined with CAR T therapy with PD-1 blockade, which did not result in objective clinical responses [ 90 ]. In paired tumor analyses before treatment and at relapse, investigators observed a post-therapy enrichment of immunosuppressive T cell subsets, specifically an increased infiltration of FOXP3 + Tregs and a higher prevalence of exhausted T cells. These changes, identified through in situ phenotyping and single-cell transcriptomic profiling of resected tumor tissue, demonstrated that post-CAR T specimens contained more PD-1 high /TIM-3 + T cells indicative of T cell exhaustion and expanded Treg populations compared to pre-treatment samples.

Preclinical studies provide mechanistic support for these clinical findings by showing that CAR T-secreted IFN-γ remodels the GBM microenvironment and activates endogenous antitumor immunity. In an immune-competent murine model, IFN-γ produced by CAR T cells drove activation of intratumoral macrophages and microglia (CD86 + MHC-II high ), which in turn supported expansion and functional licensing of endogenous CD8 + T cells characterized by Ki-67 and granzyme expression [ 88 ]. Importantly, endogenous T cells isolated after CAR treatment displayed tumor-specific cytotoxicity, and durable rejection of antigen-negative rechallenge tumors required both CAR-derived IFN-γ and host IFNγR signaling, establishing causality between CAR T cytotoxic activity, IFN-γ-mediated myeloid activation, and the endogenous CD8 + expansion observed in clinical trials. Table 2 summarizes strategies that aim to improve GBM CAR T durability by enhancing endogenous immunity or limiting adaptive resistance.

In summary, CAR T therapy in GBM patients can upregulate endogenous effector immune elements, such as cytotoxic T cells and activated myeloid cells, along with proinflammatory mediators, but this immune activation frequently coincides with the induction of counter-regulatory mechanisms, such as Treg accumulation and checkpoint ligand expression, that may limit efficacy. These insights, derived from early-phase trials with small patient numbers, are largely correlative and must be interpreted cautiously. Further clinical studies are needed to confirm these immune alterations and to determine how to tip the balance toward a more robust, durable anti-GBM immune response in the context of CAR T therapy. Figure 1 recapitulates the Key barriers to CAR T entry and function in the GBM microenvironment, as well as the emerging mechanisms of endogenous immune in CAR-T therapy.

Clinical trial landscape (registry meta-synthesis)

Methods: meta-synthesis from CAR-T clinical trials in GBM

We undertook a structured landscape review (meta-synthesis) of interventional clinical trials of chimeric antigen receptor (CAR) T-cell therapy in glioblastoma (GBM) using the ClinicalTrials.gov trial registry. The search strategy utilized the advanced search features of the ClinicalTrials.gov trial registry and the search query: AREA[ConditionSearch](Glioblastoma Multiforme OR GBM OR glioblastoma OR Glioblastoma) AND AREA[InterventionSearch](CAR-T OR chimeric antigen receptor OR chimeric antigen OR CART OR CAR OR Tcell OR T-cell). Figure 2 recapitulates our search strategy.

Study identification and screening: The keyword search yielded 91 registry entries. Excluding 47 trials not involving CAR T-cell therapies, 44 interventional trials involving CAR T-cell therapies for GBM were obtained. The trials were further filtered to include ongoing trials, leaving 27 trials. Excluding trials not ongoing, according to registry status (Completed, Terminated, Suspended, Withdrawn, or Unknown), 17 trials were removed. The final exclusion involved removing 4 ongoing trials that started before 01/2020, leaving 23 ongoing trials for CAR T-cell therapies for GBM started after 2020.

Eligibility Criteria: Trials had to fulfill all of the following criteria to be included: (1) interventional study design, (2) GBM included in the condition/diagnosis criteria, (3) CAR T-cell therapy included in the intervention criteria, (4) registry status indicating that the trial is currently ongoing at the time of the screening (e.g. Recruiting, Active not recruiting, Not yet recruiting, Enrolling by invitation), and (5) start date on or after 01/2020. Criteria for the exclusion of trials included non-CAR T-cell interventions, non-interventional study designs for the main landscape, non-ongoing studies, and start dates prior to 01/2020.

Data Extraction: For each trial included, we abstracted the registry-reported variables into a standardized table that included the NCT number, trial phase, start date, primary completion date, trial title, trial status, condition(s), age criteria, and intervention description.

Synthesis Method: As we sought to describe rather than combine efficacy results, we performed a descriptive meta-synthesis. Trials were characterized by (i) target antigen(s), (ii) multi-antigen strategies (bispecific or dual-targeting strategies), (iii) conditional logic-gating strategies (AND-gating by synNotch), and (iv) safety and controllability components (inducible suicide or off-switches), as recorded in the registry intervention and title entries, Table 3 .

Results

Chimeric Antigen Receptor (CAR) T cell therapy presents as a unique challenge in glioblastoma therapy, with the low mutational burden of the tumor often cited as the most contributive limitation due the reduced generation of neoantigens that may serve as potential targets for a CAR. Additionally, the substantial cellular and antigen heterogeneity present within an individual GBM tumor as well as within a population of GBM patients has hampered the ability to engineer a consistently efficacious CAR. Indeed, while some notable CAR-T cell trials for GBM in the past decade have demonstrated long term responses in 1–2 patients, statistically meaningful results for larger cohort sizes have been virtually nonexistent.

Following a decade of CAR-T cell clinical trials often targeting a handful of antigens with modest results, next generation CAR-T cells employ a range of novel techniques aimed at boosting efficacy and addressing the well established limitations of previous generation CAR T cells. In the 23 identified ongoing CAR -T cell trials in GBM, these approaches primarily include targeting of newly established antigens with greater percent expression in GBM cells, dual targeting mechanisms/OR gating for increased CAR range, synNotch receptors/AND gating for the targeting and localization of CARs to the TME, and rapidly inducible safety switches to minimize toxicities.

The extensive search for well confined antigens expressed in a substantial portion of GBM cells but not in normal brain parenchyma has paved the way for a host of targets for next generation CAR-T cells, of which B7-H3 (CD276) is the most studied. B7-H3 is an immune checkpoint molecule that, unlike PD-1, CTLA-4, or LAG3, is enriched in tumor cells (particularly glioma stem cells) but not cells of the normal brain. This distinction enables B7-H3 as a promising target for a CAR, as investigated in clinical trials such as NCT06691308 , NCT05241392 , NCT05835687 , NCT05366179 , NCT07193628 , NCT05474378 , and NCT06691308 , of which the majority are phase 1, dose escalation studies with cohort sizes under 50. An additional target of interest for next generation CARs is matrix metalloproteinase (MMP2). MMP2 is an enzyme with expression strongly associated with glioma stem cell markers and highly expressed at infiltrative tumor margins. MMP2 naturally binds in complex with chlorotoxin (CLTX), allowing for CLTX expressing CAR T cells to have strong specificity for MMP2. Trials such as NCT04214392 and NCT05627323 investigate this CAR-T cell mechanism in recurrent or progressive GBM, with cohort sizes of N = 19 and N = 52 respectively.

Beyond the identification of novel targets, next generation CARS employ dual targeting mechanisms (‘OR’ gating) to increase the range of antigens a CAR may recognize and kill, potentially improving the efficacy of CARs particularly in the context of GBM, where established antigens are often variably expressed in the tumor environment or localized to specific cellular subtypes. NCT07193628 is an ongoing dose escalation trial investigating the potential of a bispecific CAR for B7-H3 and IL13Ra2 antigens in recurrent or refractory GBM, with a current cohort size of 14 patients. Of note, in this trial, the dual targeting approach involved combining a less established, ‘next generation CAR’ B7-H3 antigen with a historically trialed ‘previous generation CAR’ antigen IL13Ra2. Other trials such as NCT05168423 and NCT06973096 employ a bi-specific CAR for previously established EGFRvIII and IL13Ra2 antigens, likely with the goal of determining if targeting both antigens may rescue the modest therapeutic benefits demonstrated from previous generation therapies that targeted these antigens individually. One trial, NCT05577091 , is investigating a bispecific CAR targeting glioma stem cell antigens CD44 and CD133 in recurrent GBM, with a current cohort size of 10.

In addition to bi-specific CARs which enable a broader range of antigen recognition, the development of ‘AND’ gating via syn-Notch receptors is perhaps the most advanced frontier in CAR-T cell therapy and has enabled CAR-T cells to have dramatic levels of specificity to the tumor microenvironment. Trial NCT06186401 is currently investigating the safety and ideal dose for a EGFRVIII synNotch receptor induced EphA2/IL-13Ra2 bispecific CAR T cell. Such a design ensures that the expression of the bispecific EphA2/IL-13Ra2 CAR occurs only following a binding and recognition event between the CAR synthetic notch receptor and the EGFRVIII antigen in the tumor microenvironment, allowing not only for the benefits in CAR range from the bispecific receptor, but also an increased degree of tumor localization prior to activation, potentially improving efficacy and minimizing systemic effects of CAR administration.

Concurrent to improvements in CAR targeting and binding to antigens, several advances in next generation CARs have focused on reducing toxicity following administration of therapy, such as through inducible ‘off switches’ for administered CAR T cells. Most notable is the icasp9 suicide switch, as employed in NCT06815432 , which enables rapid killing of the administered CAR T cells if therapy results in severe toxicity or additional side effects.

CRS, ICANS and TIAN

Cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are two major toxicities observed in CAR T-cell therapy clinical trials. Mechanistically, CAR T cells become activated upon encountering their target antigen and secrete inflammatory cytokines such as IFNγ and TNFα, which in turn stimulate host monocytes and macrophages to release IL-1, IL-6 and other mediators [ 91 ]. Clinically, CRS manifests as a systemic hyperinflammatory response characterized by symptoms ranging from fever and fatigue to circulatory shock, vascular leakage, disseminated intravascular coagulation, and multi-organ system failure [ 92 ]. In CD19 CAR T trials for B-cell malignancies ( NCT01626495 , NCT01029366 , NCT01044069 , NCT01593696 , NCT02028455 ), the majority of patients experienced CRS, with a substantial subset developing high-grade (≥ 3) CRS [ 34 , 35 , 93 , 94 ]. An IL-6 receptor–blocking antibody, tocilizumab, has become a standard intervention that often rapidly reverses CRS symptoms within hours without compromising CAR T efficacy [ 16 , 95 ]. By contrast, in CAR T studies for GBM, severe systemic CRS has been infrequent. Only isolated “intracranial CRS” events have been noted. For example, a case report of IL13Rα2-targeted CAR T intracranial infusion noted mild fever and cytokine elevations limited to the CNS [ 16 ].

One small GBM CAR T trial targeting EphA2 did observe two patients with Grade 2 CRS (fever and pulmonary edema) ( NCT03423992 ), but these symptoms resolved with dexamethasone and no other organ toxicities occurred [ 96 ]. In contrast to these typically mild events, a recent phase I trial of GD2-directed CAR T cells in children with diffuse midline gliomas, including DIPG, reported CRS in 15 of 20 patients, with six experiencing grade ≥ 3 toxicity marked by fever, hypotension, and systemic cytokine elevations following intravenous infusion [ 97 , 98 ]. This study stands out as a rare example of high-grade CRS in a solid tumor setting, likely reflecting the immunologically active perivascular environment of midline brain tumors and the robust expansion of CAR T cells following systemic delivery.

Although often overlapping with CRS, Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS) represents a distinct neurotoxic syndrome characterized by CNS dysfunction rather than systemic inflammation. Unlike CRS which primarily affects peripheral organs and vasculature, ICANS manifests with neurologic symptoms driven by cytokine-mediated disruption of the blood-brain barrier and endothelial activation [ 99 ]. In the aforementioned GD2-CAR T trial, neurotoxicity was particularly evident in patients receiving intraventricular infusions, where symptoms included but were not limited to somnolence, encephalopathy, and increased intracranial pressure, alongside imaging evidence of localized peritumoral edema [ 97 ]. Inflammatory toxicities such as CRS and ICANS are markedly rarer in GBM CAR T cell trials compared to hematologic malignancies. While most studies, such as those targeting EGFRvIII and IL13Rα2, report only mild, localized symptoms like transient fever or headache, isolated cases of moderate CRS have been observed, such as in an EphA2-directed trial, confirming that these events, though uncommon, can still occur in glioma patients [ 16 , 96 ]. In hematologic malignancies, ICANS is far more prevalent. In the pivotal ZUMA-1 trial evaluating axi-cel ( NCT02348216 ), a CD19-directed CAR T cell therapy for the treatment of refractory large B-cell lymphoma, ICANS occurred in 67% of patients, with 32% experiencing grade ≥ 3 events such as encephalopathy, aphasia, and somnolence, typically emerging around five days post-infusion [ 100 ].

ICANS management in trials relies on intensive supportive care and corticosteroids to diminish inflammation, with dexamethasone often favored for its high CNS penetration [ 101 ]. These clinical findings support that while CRS and ICANS are dominant toxicities in hematologic CAR T therapies, their occurrence in GBM is significantly less frequent and often limited to low-grade, localized manifestations. Nonetheless, the potential for neuroinflammatory complications, particularly with intracranial delivery, necessitates stringent neuromonitoring in glioma patients receiving CAR T therapy and should inform the design of future clinical trials aimed at optimizing both safety and efficacy for solid CNS tumors.

Lastly, in addition to CRS and ICANS, there has also been the emergence of a distinct and local toxicity phenomenon termed tumor inflammation-associated neurotoxicity (TIAN), which has been noted with the use of cell therapies targeted at CNS tumors [ 102 ]. Unlike CRS and ICANS, which are defined by the presence of systemic inflammation and more generalized encephalopathy, respectively, TIAN refers to the tumor-related and regionally specific phenomenon of neuroinflammation that can manifest with transient episodes of neurologic deterioration and/or inflammation-related edema, hydrocephalus, and increased intracranial pressure, depending on the tumor location and extent. Notably, TIAN subsumes the phenomenon of pseudoprogression but is more inclusive than edema, particularly because inflammatory signals can also mediate neurologic manifestations even when there is limited or no significant tumor progression evident on neuroimaging. A TIAN grading system has also been proposed to provide standardized reporting and guidance on the management and monitoring of patients with CNS tumor immunotherapy clinical trials, particularly with the use of intracranial administration or with high-risk neuroanatomic sites that are more susceptible to inflammatory mass effects.

Therapeutic implications

Glioblastoma forces CAR T-cell therapy to function not only as a cytotoxic drug but as an in situ immune-reprogramming platform, because the tumor microenvironment is typically myeloid-dominant and immunosuppressive, with tumor-associated macrophages/microglia shaping T-cell dysfunction and poor antigen presentation [ 103 ].

Combination strategies

A central translational lesson from early GBM CAR T experience is that on-target activity can be coupled to rapid immunologic counter-adaptation. In an EGFRvIII-directed CAR T trial, CAR T cells trafficked to tumor sites and EGFRvIII expression decreased in post-treatment specimens, yet the post-infusion tumor milieu showed increased expression of inhibitory molecules and greater regulatory T-cell infiltration, consistent with adaptive immune resistance that could blunt both CAR T and endogenous T-cell contributions [ 15 ]. Therefore, checkpoint blockade is a compelling strategy in GBM when framed as a strategy to preserve or amplify antigen spreading–driven host T-cell immunity rather than as a “rescue” for CAR T cells alone. Clinical checkpoint data in GBM remain mixed: neoadjuvant anti–PD-1 therapy produced measurable pharmacodynamic immune remodeling and was initially associated with improved survival in recurrent GBM, while subsequent work highlights heterogeneity and resistance pathways that likely constrain generalized benefit, supporting biomarker-guided, mechanism-matched combinations with CAR T [ 104 , 105 ].

Radiation is also foundational in GBM management, and its immunologic value is increasingly viewed through the lens of antigen release, vascular/TME remodeling, and cross-priming potential. Recent preclinical work using intravital approaches supports true synergy between radiation and CAR T activity in an immunocompetent glioblastoma setting, strengthening the argument that priming the tumor niche can make CAR T cells better recruiters of endogenous immunity rather than solitary effectors [ 106 ]. Concurrently, STING pathway activation is attractive in GBM because it can engage innate sensing programs and reshape the local cytokine landscape. In murine GBM models and human explants, a STING agonist triggered inflammatory outputs and remodeled the tumor microenvironment with prominent innate immune engagement, motivating rational CAR T pairings aimed at converting GBM into a more immunogenic, cross-presenting state [ 107 ].

A key practical implication of locoregional GBM CAR T trials is that CNS-compartment cytokine programs are measurable and therapeutically meaningful. In the largest completed IL-13Rα2 CAR T phase I experience (65 recurrent high-grade glioma patients; most recurrent GBM) ( NCT02208362 ), locoregional dosing was feasible and associated with CNS increases in inflammatory mediators including IFNγ and CXCL9/CXCL10, aligning with a chemokine axis that could recruit and position endogenous effector cells [ 87 ]. The challenge is to harness these programs without unacceptable neuroinflammation. “Armoring” approaches that conditionally deliver supportive cytokines or remodel suppressive myeloid populations are particularly relevant in GBM, where MDSCs and TAMs constrain durable T-cell function. For example, IL-15 modification has been reported to enable CAR T cells to function as a dual-targeting agent against tumor cells and MDSCs in GBM models, conceptually aligning cytokine support with endogenous immune recruitment rather than simple CAR T persistence [ 108 ].

Engineering approaches

GBM clinical signals increasingly suggest that the most durable advances will come from engineering CAR T cells to broaden recognition and actively recruit non-engineered immune effectors, while building in safeguards for CNS toxicity. Locoregional administration is no longer a niche tactic in GBM but rather is becoming a design principle. IL13Rα2 CAR T cells delivered intratumorally/intraventricularly achieved acceptable safety with clear CNS cytokine bioactivity and a subset of meaningful disease control [ 87 ]. The phase I intrathecal bivalent EGFR/IL13Rα2 CAR T trial ( NCT05168423 ) enrolled 18 patients with recurrent GBM and represents a deliberate attempt to broaden antigen recognition while leveraging CSF delivery. This strategy produced frequent radiographic tumor regression but was accompanied by substantial rates of grade 3 neurotoxicity, underscoring that effective trafficking and tumor engagement must be paired with careful control of CNS inflammation [ 108 ].

Antigen heterogeneity and immunoediting remain central failure modes in GBM CAR T therapy, motivating multi-antigen and reach beyond the CAR designs [ 15 ]. While single-antigen CARs can induce meaningful tumor regression, selective pressure frequently results in downregulation or loss of the targeted epitope, leaving antigen-negative clones to drive recurrence. One strategy to address this limitation involves engineering CAR T cells to secrete bispecific T-cell engagers (BiTEs). BiTEs are antibody-based molecules that simultaneously bind CD3 on endogenous T cells and a tumor-associated antigen, thereby redirecting non-engineered T cells toward tumor cells independent of CAR expression. By recruiting bystander T cells against additional tumor antigens, BiTE-secreting CAR platforms extend tumor coverage beyond the original CAR target and reduce dependence on uniform antigen expression [ 109 ].

One approach designed to address antigen heterogeneity while preserving tumor specificity is the use of synthetic Notch (SynNotch) receptor systems. SynNotch receptors are engineered, modular receptors that trigger a programmable transcriptional response only after recognition of a defined “priming” antigen. In CAR T applications, this architecture enables a prime-and-kill circuit where initial engagement of a tumor-restricted antigen activates expression of a secondary receptor targeting tumor-associated antigens. This conditional logic spatially restricts CAR activity to the tumor microenvironment while expanding antigen coverage beyond a single static epitope and has demonstrated efficacy in GBM-relevant contexts [ 110 ]. The emergence of first-in-human SynNotch-inspired GBM trials (e.g., EGFRvIII SynNotch–induced multi-antigen CAR programs) highlights how recruitment of endogenous immunity can be coupled to programmable recognition rather than relying on a single static antigen ( NCT06186401 ).

While EGFRvIII was initially considered an attractive target due to its tumor-specificity, accumulated clinical trial data reveal fundamental limitations. EGFRvIII expression is subclonal and spatially heterogeneous, with antigen loss consistently documented following CAR T exposure [ 14 , 89 ]. These findings establish that EGFRvIII-directed monotherapy is insufficient to durably alter tumor progression. The next generation of GBM CAR targets must address this heterogeneity through three integrated approaches. First, multi-antigen targeting strategies, such as bispecific CARs, synNotch logic-gated circuits, and T-cell engaging platforms (BiTEs and TEAMs), can mitigate antigen escape by broadening recognition beyond single markers [ 109 ]. Second, prioritizing antigens with higher tumor penetrance and more uniform expression, such as B7-H3 which is expressed in > 75% of GBMs with enrichment in glioma stem cells [ 111 ], may provide more durable coverage. Third, integrating single-cell transcriptomics and spatial proteomics can identify surface antigens conserved across malignant cellular states [ 112 ] and resistant to immunotherapeutic selection pressure, representing promising paths toward targets that maintain expression under immune pressure (Table 4 ).

Future directions

GBM is now entering a phase where progress will be determined less by whether CAR T cells can reach the brain, and more by whether they can (i) quantify and induce host immunity, (ii) overcome spatial/antigenic heterogeneity, and (iii) establish durable immune memory without intolerable neuroinflammation. A near-term priority is to operationalize “endogenous immune recruitment” as a measurable clinical endpoint, not an inferred mechanism. GBM uniquely enables longitudinal sampling via CSF reservoirs, creating a practical route to track: (1) TCR repertoire dynamics (CSF and blood), (2) cytokine/chemokine programs that predict recruitment (for example IFNγ–CXCL9/10 patterns seen after locoregional IL-13Rα2 CAR T), and (3) tumor antigen burden and immunoediting. Clinical trials are already pointing to implementable correlates. Intrathecal bivalent EGFR/IL13Rα2 CAR T treatment produced substantial CAR T abundance and cytokine release in CSF in all treated patients in an interim analysis, alongside early-onset neurotoxicity managed with steroids and IL1R blockade, illustrating both the opportunity and the need for compartment-resolved immune monitoring. Finally, “liquid biopsy” approaches in CSF are becoming mechanistically informative: CARv3-TEAM-E reports included CSF-based molecular tracking aligned with rapid radiographic changes, suggesting a path to integrate antigen escape surveillance with endogenous immune readouts in real time.

GBM heterogeneity is not simply an antigen-selection problem; it is an immune-engagement problem.

When CAR T pressure eliminates an antigen-positive compartment, the remaining tumor may be both antigen-divergent and more suppressive. This is clinically foreshadowed by EGFRvIII CAR T data showing both antigen decrease and a shift toward an inhibitory, Treg-enriched environment after infusion. Future work should treat heterogeneity as a design constraint that requires two coupled solutions: (i) broadened recognition (bivalent/multivalent CARs, prime-and-kill circuits) and (ii) deliberate recruitment of non-engineered effectors (e.g., engager-secreting platforms). The intracerebroventricular bivalent EGFR(epitope 806)/IL-13Rα2 CAR T phase I trial (18 treated patients) provides a recent clinical anchor for this approach, showing tumor regression in a majority of patients with measurable disease, but also high rates of grade 3 neurotoxicity, emphasizing that broader targeting must be paired with control of CNS inflammation [ 108 ].

Durable GBM control likely requires a transition from an acute CAR T response to a self-renewing endogenous immune state, potentially involving sustained antigen presentation, myeloid repolarization, and development of functional memory T-cell pools. Mechanistically, IFNγ has emerged as a key mediator linking CAR T activity to host myeloid activation and induction of endogenous immunity, providing a concrete axis to optimize rather than treating immune recruitment as a black box [ 88 ]. However, the same inflammatory circuitry that may enable memory can also drive neurotoxicity. The field therefore needs “immune rheostats”: tunable cytokine programs, on-switches/off-switches, and rational anti-inflammatory co-therapies that preserve cross-priming while preventing runaway CNS inflammation, an issue underscored by the neurotoxicity profiles in CSF-delivered bivalent CAR T trials.

Lastly, in parallel, AI-enabled phenotyping is likely to become a practical lever for GBM CAR T development by improving patient stratification and trial learning. For one, routine pathology and imaging-based prediction models will enable immune states and predictions for response to immunotherapy to inform trials [ 115 ]. These predictions will enable pre-treatment estimates for heterogeneity for antigens, myeloid cell dominance, and adaptive resistance programs to inform CAR-T trials. In the perioperative period, tumor segmentation and surgical planning will enable a more precise relationship between tumor and imaging phenotypes, immune cell states in CSF, and toxicity, including TIAN [ 15 , 116 – 119 ].

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Acknowledgements

NA.

Author contributions

Conceptualization: ML, GL, MA, JL, RM, JC. Investigation: MA, JL, SA. Figures & Tables: MA, SA, Writing - Original Draft: JL, MA, SA, JH, MS, VV, GN, RM, JC, GL,. Writing - Review & Editing: JL, MA, SA, JH, MS, VV, GN, RM, JC, GL. Supervision: ML.

Funding

N/A.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics and Consent to Participate

Not applicable.

Animal and human studies Experimental approval board

Not applicable.

Financial disclosure

None.

Competing interests

No pertinent conflicts of interest relevant to this manuscript. Michael Lim: (Funding from Arbor Pharmaceuticals, Accuray, BMS, Novartis; Consultant: BMS, Merck, SQZ Biotechnologies, Tocagen, VBI; Patents: Combining Focused Radiation and Immunotherapy, Combining Local Chemotherapy and Immunotherapy; Shareholder: Egret Therapeutics). Gordon Li: Consultant for Medtronic, and received funding from Novocure and Targtex.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Justin Liu and Matthew Abikenari contributed equally to this work.

References

1. Price M, Ballard C, Benedetti J et al (2024) CBTRUS Statistical Report: Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2017–2021. Neuro-Oncol 26:vi1–vi85

2. Durga YK, Pusapati RV (2026) Drug resistance in glioblastoma: Challenges, mechanisms and therapeutic strategies (Review). Mol Clin Oncol 24:4

3. Wang X, Guo G, Guan H et al (2019) Challenges and potential of PD-1/PD-L1 checkpoint blockade immunotherapy for glioblastoma. J Exp Clin Cancer Res CR 38:87

4. Abikenari MA, Enayati I, Fountain DM et al (2024) Navigating glioblastoma therapy: A narrative review of emerging immunotherapeutics and small-molecule inhibitors. Microbes Immun 0:5075

5. Dynarowicz K, Smolak B, Bartusik-Aebisher D et al (2026) Current Pharmacotherapeutic Strategies in Diffuse Gliomas: Focus on Glioblastoma, IDH-Wildtype, and Emerging Targeted Therapies for IDH-Mutant Tumors. Pharmaceuticals 19:148

6. Abikenari M, Schonfeld E, Choi J et al (2024) Revisiting glioblastoma classification through an immunological lens: A narrative review. Glioma 7:3–9

7. Wu L, Wei Q, Brzostek J et al (2020) Signaling from T cell receptors (TCRs) and chimeric antigen receptors (CARs) on T cells. Cell Mol Immunol 17:600–612

8. Ershova A, Goldaeva A, Staliarova A et al Future perspectives on novel CAR-T therapeutics beyond CD19 and BCMA in onco-hematology. Front Immunol. 16. Epub ahead of print 14 July 2025. 10.3389/fimmu.2025.1592377.

9. Awasthi R, Maier HJ, Zhang J et al (2023) Kymriah ® (tisagenlecleucel) - An overview of the clinical development journey of the first approved CAR-T therapy. Hum Vaccines Immunother 19:2210046

10. Ahmad A, Uddin S, Steinhoff M, CAR-T Cell Therapies (2020) An Overview of Clinical Studies Supporting Their Approved Use against Acute Lymphoblastic Leukemia and Large B-Cell Lymphomas. Int J Mol Sci 21:3906

11. Gene targets of CAR-T cell therapy for glioblastoma - PMC https://pmc.ncbi.nlm.nih.gov/articles/PMC10136592/ . Accessed 21 Jan 2026.

12. MacKay M, Afshinnekoo E, Rub J et al (2020) The therapeutic landscape for cells engineered with chimeric antigen receptors. Nat Biotechnol 38:233–244

13. Krenciute G, Prinzing BL, Yi Z et al (2017) Transgenic Expression of IL15 Improves Antiglioma Activity of IL13Rα2-CAR T Cells but Results in Antigen Loss Variants. Cancer Immunol Res 5:571–581

14. Sampson JH, Heimberger AB, Archer GE et al (2010) Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant III peptide vaccination in patients with newly diagnosed glioblastoma. J Clin Oncol Off J Am Soc Clin Oncol 28:4722–4729

15. O’Rourke DM, Nasrallah MP, Desai A et al (2017) A single dose of peripherally infused EGFRvIII-directed CAR T cells mediates antigen loss and induces adaptive resistance in patients with recurrent glioblastoma. Sci Transl Med 9:eaaa0984

16. Brown CE, Alizadeh D, Starr R et al (2016) Regression of Glioblastoma after Chimeric Antigen Receptor T-Cell Therapy. N Engl J Med 375:2561–2569

17. Medikonda R, Abikenari MA, Schonfeld E et al (2025) Top advances of the year: The status of chimeric antigen receptor T cells in neuro-oncology. Cancer 131:e35935

18. Skaga E, Kulesskiy E, Fayzullin A et al (2019) Intertumoral heterogeneity in patient-specific drug sensitivities in treatment-naïve glioblastoma. BMC Cancer 19:628

19. Sottoriva A, Spiteri I, Piccirillo SGM et al (2013) Intratumor heterogeneity in human glioblastoma reflects cancer evolutionary dynamics. Proc Natl Acad Sci 110:4009–4014

20. Neftel C, Laffy J, Filbin MG et al (2019) An Integrative Model of Cellular States, Plasticity, and Genetics for Glioblastoma. Cell 178:835–849e21

21. Gabrusiewicz K, Rodriguez B, Wei J et al (2016) Glioblastoma-infiltrated innate immune cells resemble M0 macrophage phenotype. JCI Insight 1:e85841

22. Orrego E, Castaneda CA, Castillo M et al (2018) Distribution of tumor-infiltrating immune cells in glioblastoma. CNS Oncol 7:CNS21

23. Maas AIR, Stocchetti N, Bullock R (2008) Moderate and severe traumatic brain injury in adults. Lancet Neurol 7:728–741

24. Medikonda R, Abikenari M, Schonfeld E et al (2025) The Metabolic Orchestration of Immune Evasion in Glioblastoma: From Molecular Perspectives to Therapeutic Vulnerabilities. Cancers 17:1881

25. Mohme M, Schliffke S, Maire CL et al (2018) Immunophenotyping of Newly Diagnosed and Recurrent Glioblastoma Defines Distinct Immune Exhaustion Profiles in Peripheral and Tumor-infiltrating Lymphocytes. Clin Cancer Res Off J Am Assoc Cancer Res 24:4187–4200

26. Woroniecka K, Chongsathidkiet P, Rhodin K et al (2018) T-Cell Exhaustion Signatures Vary with Tumor Type and Are Severe in Glioblastoma. Clin Cancer Res Off J Am Assoc Cancer Res 24:4175–4186

27. Hu B, Ren J, Luo Y et al (2017) Augmentation of antitumor immunity by human and mouse CAR T cells secreting IL-18. Cell Reports. 20(13):3025–3033. Accessed 21 Jan 2026.

28. Yeku OO, Brentjens RJ (2016) Armored CAR T-cells: utilizing cytokines and pro-inflammatory ligands to enhance CAR T-cell anti-tumour efficacy. Biochem Soc Trans 44:412–418

29. Chen M, Sun R, Shi B et al (2019) Antitumor efficacy of chimeric antigen receptor T cells against EGFRvIII-expressing glioblastoma in C57BL/6 mice. Biomed Pharmacother 113:108734

30. Sampson JH, Choi BD, Sanchez-Perez L et al (2014) EGFRvIII mCAR-modified T-cell therapy cures mice with established intracerebral glioma and generates host immunity against tumor-antigen loss. Clin Cancer Res Off J Am Assoc Cancer Res 20:972–984

31. Denlinger N, Bond D, Jaglowski S (2022) CAR T-cell therapy for B-cell lymphoma. Curr Probl Cancer 46:100826

32. Cappell KM, Kochenderfer JN (2023) Long-term outcomes following CAR T cell therapy: what we know so far. Nat Rev Clin Oncol 20:359–371

33. Wudhikarn K, Flynn JR, Rivière I et al (2021) Interventions and outcomes of adult patients with B-ALL progressing after CD19 chimeric antigen receptor T-cell therapy. Blood 138:531–543

34. Maude SL, Frey N, Shaw PA et al (2014) Chimeric antigen receptor T cells for sustained remissions in leukemia. N Engl J Med 371:1507–1517

35. Park JH, Rivière I, Gonen M et al (2018) Long-Term Follow-up of CD19 CAR Therapy in Acute Lymphoblastic Leukemia. N Engl J Med 378:449–459

36. Porter DL, Hwang W-T, Frey NV et al (2015) Chimeric antigen receptor T cells persist and induce sustained remissions in relapsed refractory chronic lymphocytic leukemia. Sci Transl Med 7:303ra139–303ra139

37. Vieira dos Santos J, Melnekoff DT, Aleman A et al (2025) Long-term remission after cilta-cel in multiple myeloma is linked to diverse t cells and low myeloid suppression. Blood Adv. 2025018078.

38. Upadhyay R, Boiarsky JA, Pantsulaia G et al (2021) A Critical Role for Fas-Mediated Off-Target Tumor Killing in T-cell Immunotherapy. Cancer Discov 11:599–613

39. Ma L, Hostetler A, Morgan DM et al (2023) Vaccine-boosted CAR T crosstalk with host immunity to reject tumors with antigen heterogeneity. Cell 186:3148–3165e20

40. An N, Li J, Luo P et al (2025) Key predictors of long-term outcomes in BCMA-targeted CAR-T therapy for relapsed/refractory multiple myeloma. J Transl Med 23:552

41. Melero I, Rouzaut A, Motz GT et al (2014) T-Cell and NK-Cell Infiltration into Solid Tumors: A Key Limiting Factor for Efficacious Cancer Immunotherapy. Cancer Discov 4:522–526

42. Nicolas-Boluda A, Vaquero J, Vimeux L et al (2021) Tumor stiffening reversion through collagen crosslinking inhibition improves T cell migration and anti-PD-1 treatment. eLife 10:e58688

43. Perez VM, Kearney JF, Yeh JJ (2021) The PDAC Extracellular Matrix: A Review of the ECM Protein Composition, Tumor Cell Interaction, and Therapeutic Strategies. Front Oncol 11:751311

44. Vo M-C, Tran V-D-H, Nguyen V-T et al (2025) Challenges and limitations of chimeric antigen receptor T-cell therapies in solid tumors: why are approvals restricted to hematologic malignancies? J Hematol OncolJ Hematol Oncol 18:91

45. Xiao Z, Todd L, Huang L et al (2023) Desmoplastic stroma restricts T cell extravasation and mediates immune exclusion and immunosuppression in solid tumors. Nat Commun 14:5110

46. Caruana I, Savoldo B, Hoyos V et al (2015) Heparanase promotes tumor infiltration and antitumor activity of CAR-redirected T lymphocytes. Nat Med 21:524–529

47. Deng C, Zhao J, Zhou S et al (2020) The Vascular Disrupting Agent CA4P Improves the Antitumor Efficacy of CAR-T Cells in Preclinical Models of Solid Human Tumors. Mol Ther J Am Soc Gene Ther 28:75–88

48. Juneja VR, McGuire KA, Manguso RT et al (2017) PD-L1 on tumor cells is sufficient for immune evasion in immunogenic tumors and inhibits CD8 T cell cytotoxicity. J Exp Med 214:895–904

49. Dong H, Strome SE, Salomao DR et al (2002) Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion. Nat Med 8:793–800

50. Noguchi T, Ward JP, Gubin MM et al (2017) Temporally Distinct PD-L1 Expression by Tumor and Host Cells Contributes to Immune Escape. Cancer Immunol Res 5:106–117

51. Ninomiya S, Narala N, Huye L et al (2015) Tumor indoleamine 2,3-dioxygenase (IDO) inhibits CD19-CAR T cells and is downregulated by lymphodepleting drugs. Blood 125:3905–3916

52. Campesato LF, Budhu S, Tchaicha J et al (2020) Blockade of the AHR restricts a Treg-macrophage suppressive axis induced by L-Kynurenine. Nat Commun 11:4011

53. Guangzhao L, Xin W, Miaoqing W et al (2025) IDO1 inhibitor enhances the effectiveness of PD-1 blockade in microsatellite stable colorectal cancer by promoting macrophage pro-inflammatory phenotype polarization. Cancer Immunol Immunother CII 74:71

54. Tauriello DVF, Palomo-Ponce S, Stork D et al (2018) TGFβ drives immune evasion in genetically reconstituted colon cancer metastasis. Nature 554:538–543

55. Mariathasan S, Turley SJ, Nickles D et al (2018) TGFβ attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells. Nature 554:544–548

56. Curiel TJ, Coukos G, Zou L et al (2004) Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med 10:942–949

57. Haas AR, Tanyi JL, O’Hara MH et al (2019) Phase I Study of Lentiviral-Transduced Chimeric Antigen Receptor-Modified T Cells Recognizing Mesothelin in Advanced Solid Cancers. Mol Ther J Am Soc Gene Ther 27:1919–1929

58. Kara E, Jackson TL, Jones C et al (2024) Mathematical modeling insights into improving CAR T cell therapy for solid tumors with bystander effects. Npj Syst Biol Appl 10:105

59. Beatty GL, O’Hara MH, Lacey SF et al (2018) Activity of Mesothelin-Specific Chimeric Antigen Receptor T Cells Against Pancreatic Carcinoma Metastases in a Phase 1 Trial. Gastroenterology 155:29–32

60. Avanzi MP, Yeku O, Li X et al (2018) Engineered Tumor-Targeted T Cells Mediate Enhanced Anti-Tumor Efficacy Both Directly and through Activation of the Endogenous Immune System. Cell Rep 23:2130–2141

61. Chmielewski M, Hombach AA, Abken H (2013) Of < scp>CAR s and < scp>TRUCK s: chimeric antigen receptor (< scp>CAR) T cells engineered with an inducible cytokine to modulate the tumor stroma. Immunol Rev 257:83–90

62. Cherkassky L, Morello A, Villena-Vargas J et al (2016) Human CAR T cells with cell-intrinsic PD-1 checkpoint blockade resist tumor-mediated inhibition. J Clin Invest 126:3130–3144

63. Jorgovanovic D, Song M, Wang L et al Roles of IFN-γ in tumor progression and regression: a review. Biomark Res; 8. Epub ahead of print September 2020. 10.1186/s40364-020-00228-x

64. Abikenari M, Liu J, Ha JH et al (2025) Emerging trends in cell-based therapies: contemporary advances and ethical considerations in translational neurosurgical oncology. J Neurooncol 175:1–20

65. Alizadeh AA, Eisen MB, Davis RE et al (2000) Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403:503–511

66. Abikenari M, Sjoholm MA, Liu J et al Molecular and biophysical remodeling of the blood–brain barrier in glioblastoma: mechanistic drivers of tumor–neurovascular crosstalk. Front Phys; 13. Epub ahead of print 18 December 2025. 10.3389/fphy.2025.1723329

67. Zhou H-J, Wang L-Q, Wang D-B et al (2018) Long noncoding RNA MALAT1 contributes to inflammatory response of microglia following spinal cord injury via the modulation of a miR-199b/IKKβ/NF-κB signaling pathway. Am J Physiol-Cell Physiol 315:C52–C61

68. Tokarew N, Ogonek J, Endres S et al (2018) Teaching an old dog new tricks: next-generation CAR T cells. Br J Cancer 120:26–37

69. Gardam B, Gargett T, Brown MP et al Targeting the dendritic cell-T cell axis to develop effective immunotherapies for glioblastoma. Front Immunol; 14. Epub ahead of print 20 October 2023. 10.3389/fimmu.2023.1261257

70. Chmielewski M, Kopecky C, Hombach AA et al (2011) IL-12 release by engineered T cells expressing chimeric antigen receptors can effectively Muster an antigen-independent macrophage response on tumor cells that have shut down tumor antigen expression. Cancer Res 71:5697–5706

71. Agliardi G, Liuzzi AR, Hotblack A et al Intratumoral IL-12 delivery empowers CAR-T cell immunotherapy in a pre-clinical model of glioblastoma. Nat Commun; 12. Epub ahead of print January 2021. 10.1038/s41467-020-20599-x

72. Chmielewski M, Abken H, CAR T, Cells Releasing (2017) IL-18 Convert to T-Bethigh FoxO1low Effectors that Exhibit Augmented Activity against Advanced Solid Tumors. Cell Rep 21:3205–3219

73. Shouse AN, LaPorte KM, Malek TR (2024) Interleukin-2 signaling in the regulation of T cell biology in autoimmunity and cancer. Immunity 57:414–428

74. Goff SL, Morgan RA, Yang JC et al (2019) Pilot Trial of Adoptive Transfer of Chimeric Antigen Receptor–transduced T Cells Targeting EGFRvIII in Patients With Glioblastoma. J Immunother 42:126

75. Brown CE, Rodriguez A, Palmer J et al (2022) Off-the-shelf, steroid-resistant, IL13Rα2-specific CAR T cells for treatment of glioblastoma. Neuro-Oncol 24:1318–1330

76. Jang D, Lee A-H, Shin H-Y et al (2021) The Role of Tumor Necrosis Factor Alpha (TNF-α) in Autoimmune Disease and Current TNF-α Inhibitors in Therapeutics. Int J Mol Sci 22:2719

77. Pituch KC, Miska J, Krenciute G et al (2018) Adoptive Transfer of IL13Rα2-Specific Chimeric Antigen Receptor T Cells Creates a Pro-inflammatory Environment in Glioblastoma. Mol Ther 26:986–995

78. Lee KM, Achuthan AA, Hamilton JA. GM-CSF: A Promising Target in Inflammation and Autoimmunity ImmunoTargets Ther 2020; Volume 9: 225–240

79. Shi Y, Liu CH, Roberts AI et al (2006) Granulocyte-macrophage colony-stimulating factor (GM-CSF) and T-cell responses: what we do and don’t know. Cell Res 16:126–133

80. Spear P, Barber A, Rynda-Apple A et al (2012) Chimeric Antigen Receptor T Cells Shape Myeloid Cell Function within the Tumor Microenvironment through IFN-γ and GM-CSF. J Immunol 188:6389–6398

81. Galluzzi L, Guilbaud E, Schmidt D et al (2024) Targeting immunogenic cell stress and death for cancer therapy. Nat Rev Drug Discov 23:445–460

82. Brossart P (2020) The Role of Antigen Spreading in the Efficacy of Immunotherapies. Clin Cancer Res 26:4442–4447

83. Modulating the glioma (2025) microenvironment with laser interstitial thermal therapy: mechanisms and therapeutic implications. J Neuro Oncol. https://link.springer.com/article/10.1007/s11060-025-05305-5 . Accessed 30 Nov 2025.

84. Chen J, Yang Y, Luan S et al Tertiary lymphoid structures in gliomas: impact on tumour immunity and progression. J Transl Med; 23. Epub ahead of print May 2025. 10.1186/s12967-025-06510-6.

85. van de Walle T, Vaccaro A, Ramachandran M, et al Tertiary lymphoid structures in the central nervous system: implications for glioblastoma. Front Immunol. 12. Epub ahead of print September 2021. 10.3389/fimmu.2021.724739.

86. Sabahi M, Salehipour A, Bazl MSY et al (2023) Local immunotherapy of glioblastoma: A comprehensive review of the concept. J Neuroimmunol 381:578146

87. Brown CE, Hibbard JC, Alizadeh D et al (2024) Locoregional delivery of IL-13Rα2-targeting CAR-T cells in recurrent high-grade glioma: a phase 1 trial. Nat Med 30:1001–1012

88. Alizadeh D, Wong RA, Gholamin S et al (2021) IFNγ Is Critical for CAR T Cell–Mediated Myeloid Activation and Induction of Endogenous Immunity. Cancer Discov 11:2248–2265

89. O’Rourke DM, Nasrallah MP, Desai A et al A single dose of peripherally infused EGFRvIII-directed CAR T cells mediates antigen loss and induces adaptive resistance in patients with recurrent glioblastoma. Sci Transl Med. 9. Epub ahead of print 19 July 2017. 10.1126/scitranslmed.aaa0984

90. Bagley SJ, Desai AS, Linette GP et al (2018) CAR T-cell therapy for glioblastoma: recent clinical advances and future challenges. Neuro-Oncol 20:1429–1438

91. Siegler EL, Kenderian SS (2020) Neurotoxicity and Cytokine Release Syndrome After Chimeric Antigen Receptor T Cell Therapy: Insights Into Mechanisms and Novel Therapies. Front Immunol 11:1973

92. Shimabukuro-Vornhagen A, Gödel P, Subklewe M et al (2018) Cytokine release syndrome. J Immunother Cancer 6:56

93. Lee DW, Kochenderfer JN, Stetler-Stevenson M et al (2015) T cells expressing CD19 chimeric antigen receptors for acute lymphoblastic leukaemia in children and young adults: a phase 1 dose-escalation trial. Lancet Lond Engl 385:517–528

94. Gardner RA, Finney O, Annesley C et al (2017) Intent-to-treat leukemia remission by CD19 CAR T cells of defined formulation and dose in children and young adults. Blood 129:3322–3331

95. Jamois C, Turner DC, Gibiansky L et al (2025) Tocilizumab Dosing for Management of T Cell-Engaging Bispecific Antibody-Related CRS in Patients With R/R B-Cell NHL. Clin Pharmacol Ther 118:917–927

96. Lin Q, Ba T, Ho J et al (2021) First-in-Human Trial of EphA2-Redirected CAR T-Cells in Patients With Recurrent Glioblastoma: A Preliminary Report of Three Cases at the Starting Dose. Front Oncol 11:694941

97. Monje M, Mahdi J, Majzner R et al (2025) Intravenous and intracranial GD2-CAR T cells for H3K27M+ diffuse midline gliomas. Nature 637:708–715

98. Liu J, Ha JH, Abikenari M et al (2025) Adaptive immunotherapeutic paradigms in diffuse midline glioma: integrating epigenetic reprogramming, neuron–glioma interactions, and tumor microenvironment modulation. J Neurooncol 176:144

99. Sterner RC, Sterner RM (2022) Immune effector cell associated neurotoxicity syndrome in chimeric antigen receptor-T cell therapy. Front Immunol 13:879608

100. Locke FL, Neelapu SS, Bartlett NL et al (2017) Phase 1 Results of ZUMA-1: A Multicenter Study of KTE-C19 Anti-CD19 CAR T Cell Therapy in Refractory Aggressive Lymphoma. Mol Ther J Am Soc Gene Ther 25:285–295

101. Rees JH (2022) Management of Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS). In: Kröger N, Gribben J, Chabannon C et al (eds) The EBMT/EHA CAR-T Cell Handbook. Springer, Cham (CH), pp 141–145

102. Mahdi J, Dietrich J, Straathof K et al (2023) Tumor inflammation-associated neurotoxicity. Nat Med 29:803–810

103. Khan F, Pang L, Dunterman M et al (2024) Macrophages and microglia in glioblastoma: heterogeneity, plasticity, and therapy. J Clin Invest 133:e163446

104. Cloughesy TF, Mochizuki AY, Orpilla JR et al (2019) Neoadjuvant anti-PD-1 immunotherapy promotes a survival benefit with intratumoral and systemic immune responses in recurrent glioblastoma. Nat Med 25:477–486

105. McFaline-Figueroa JR, Sun L, Youssef GC et al (2024) Neoadjuvant anti-PD1 immunotherapy for surgically accessible recurrent glioblastoma: clinical and molecular outcomes of a stage 2 single-arm expansion cohort. Nat Commun 15:10757

106. Low-dose radiation (2026) by radiopharmaceutical therapy enhances GD2 TRAC-CAR T cell efficacy in localized neuroblastoma. Sci Adv. https://www.science.org/doi/10.1126/sciadv.adu4417?utm_source=chatgpt.com . Accessed 2 Jan 2026.

107. Berger G, Knelson EH, Jimenez-Macias JL et al (2022) STING activation promotes robust immune response and NK cell-mediated tumor regression in glioblastoma models. Proc Natl Acad Sci U S A 119:e2111003119

108. Bagley SJ, Desai AS, Fraietta JA et al (2025) Intracerebroventricular bivalent CAR T cells targeting EGFR and IL-13Rα2 in recurrent glioblastoma: a phase 1 trial. Nat Med 31:2778–2787

109. Choi BD, Yu X, Castano AP et al (2019) CAR-T cells secreting BiTEs circumvent antigen escape without detectable toxicity. Nat Biotechnol 37:1049–1058

110. Choe JH, Watchmaker PB, Simic MS et al (2021) SynNotch-CAR T cells overcome challenges of specificity, heterogeneity, and persistence in treating glioblastoma. Sci Transl Med 13:eabe7378

111. Nehama D, Di Ianni N, Musio S et al (2019) B7-H3-redirected chimeric antigen receptor T cells target glioblastoma and neurospheres. EBioMedicine 47:33–43

112. Neftel C, Laffy J, Filbin MG et al (2019) An Integrative Model of Cellular States, Plasticity, and Genetics for Glioblastoma. Cell 178:835–849e21

113. Thaci B, Brown CE, Binello E et al (2014) Significance of interleukin-13 receptor alpha 2-targeted glioblastoma therapy. Neuro-Oncol 16:1304–1312

114. Jin L, Ge H, Long Y et al (2018) CD70, a novel target of CAR T-cell therapy for gliomas. Neuro-Oncol 20:55–65

115. Luo X, Wang X, Eweje F et al (2025) Ensemble learning of foundation models for precision oncology. arXiv org, https://arxiv.org/abs/2508.16085v1 . Accessed 8 Jan 2026.

116. Abikenari M, Jain B, Xu R et al Bridging imaging and molecular biomarkers in trigeminal neuralgia: toward precision diagnosis and prognostication in neuropathic pain. Med Res Arch. 13. Epub ahead of print 30 May 2025. 10.18103/mra.v13i5.6605

117. Boretti A (2024) The transformative potential of AI-driven CRISPR-Cas9 genome editing to enhance CAR T-cell therapy. Comput Biol Med 182:109137

118. Reframing clinical AI evaluation in the era of generative models (2026) toward multidimensional, stakeholder-informed, and safety-centric frameworks for real-world health care deployment. Prem Sci. https://premierscience.com/pjs-25-1011/ . Accessed 22 Jan 2026.

119. Full article (2019) Artificial intelligence and immunotherapy. 1623670 https://www.tandfonline.com/doi/full/10.1080/1744666X . Accessed 22 Jan 2026.

Associated Data

Supplementary Materials

Supplementary Material 1

Data Availability Statement

No datasets were generated or analysed during the current study.