Evolving CAR T cell therapy to overcome the barriers in treating pediatric central nervous system tumors

Abstract

Central nervous system (CNS) tumors are a leading cause of pediatric cancer related death. CAR T cells are an innovative approach for these affected children who are in desperate need of novel therapies, but CNS-directed cellular therapies have only recently advanced to the clinic. While early phase trials have begun to demonstrate the feasibility of manufacturing fractionated doses and the tolerability of repeated infusions for children with CNS tumors, major challenges remain. Here, we will take an inventory of the current state of the pediatric CNS CAR T cell field through the lens of translational obstacles to broader clinical success.

Introduction

In recent years, CNS tumors have surpassed leukemia as the leading cause of death among children with cancer, in part due to the therapeutic success of chimeric antigen receptor (CAR) T cells against hematologic malignancies(1,2). CAR T cells are a targeted immunotherapy in which T cells, engineered to express a CAR, can directly recognize and eliminate specific cells. CAR constructs typically include an extracellular antigen-binding domain - often a single-chain variable fragment (scFv) derived from monoclonal antibodies - able to bind specifically to the target. Upon engagement, the signaling domains within the CAR trigger the CAR T cell to proliferate, release cytokines, and exert cytotoxic effects.

Following multiple reports of preclinical efficacy(3–9), cellular therapies have advanced to early phase clinical trials for children with CNS tumors (Table 1) and provided a foundation of feasibility and safety, along with preliminary evidence of local immune activation((7,8,10), https://doi.org/10.21203/rs.3.rs-3454977/v1)). While there are reports of clinical responses, this benefit has not been universal and significant anatomical and biological obstacles remain. Beyond a paucity of surface targets with clinically relevant therapeutic windows, the hurdles of delivering CAR T cells to children with CNS tumors include insufficient CAR T cell trafficking to the tumor; limited persistence; inflammation-related neurologic adverse events; tumor heterogeneity; a tumor microenvironment ranging from inert to inhospitable; and lack of appropriate monitoring of outcomes. Efforts to address these obstacles include testing variable routes of delivery to improve tumor trafficking, engineering multi-antigen targeting to overcome antigen loss, crafting multimodal regimens, and modulating the tumor immune microenvironment (TIME) (Figure 1). Ultimately, the first generation of pediatric CNS CAR T cell studies has formed the platform upon which future laboratory and clinical advancements can be built. Therefore, it is imperative to take an assessment of what we have learned in the context of the critical hurdles that remain.

Table 1.

Completed and ongoing CAR T cell therapy clinical trials for pediatric CNS tumors

NCT Number	Site	Indication	Target	Route	Dose	Conditioning	Status	Ref.
03500991	Seattle Children’s Hospital	Pediatric and young adult R/R CNS tumors, excluding DIPG	HER2	ICT (Arm A) or ICV (Arm B)	Repeated doses of 1e7 to 1e8 CAR T cells	No	Active, not recruiting	Vitanza et al. (2021)(8)
03638167	Seattle Children’s Hospital	Pediatric and young adult R/R CNS tumors, excluding DIPG	EGFR806	ICT (Arm A) or ICV (Arm B)	Repeated doses of 1e7 to 1e8 CAR T cells	No	Active, not recruiting	N/A
04185038	Seattle Children’s Hospital	Pediatric and young adult R/R CNS tumors	B7-H3	ICT (Arm A) or ICV (Arm B and C)	Repeated doses of 1e7 to 1e8 CAR T cells	No	Recruiting	Vitanza et al. (2025)(7)
05768880	Seattle Children’s Hospital	Pediatric and young adult R/R CNS tumors	HER2, EGFR806, B7-H3, and IL13Rα2	ICV	Repeated doses of 1e7 to 1e8 CAR T cells	No	Recruiting	N/A
02442297	Texas Children’s Hospital	Pediatric and adult R/R primary CNS tumors and tumors metastatic to the CNS, excluding DIPG	HER2	ICT or ICV	Repeated doses of 1e7 to 1e8 CAR T cells	No	Active, not recruiting	N/A
04099797	Texas Children’s Hospital	Pediatric and young adult R/R CNS tumors	GD2	IV and ICV (Cohort 1) or IV only (Cohort 2)	One IV dose of 1e7/m2 followed by ICV doses of 2e6 to 5e6 (Cohort 1) or repeated IV doses of 1e7/m2 CAR T cells	Yes	Recruiting	Lin et al. (2024)(46)
04196413	Lucile Packard Children’s Hospital	Pediatric and young adult DIPG and spinal DMG	GD2	IV and ICV (Arm A) or ICV only (Arm B&C)	One IV dose of 1e6 to 3e6/kg followed by ICV doses of 1e7 to 3e7 (Arm A) or ICV 1e7-1e8 (Arm B&C) CAR T cells	Yes (Arm A&C) or No (Arm B)	Recruiting	Monje et al. (2024)(10)
04510051	City of Hope Medical Center	Pediatric and young adult R/R CNS tumors	IL13Rα2	ICV	Repeated doses of 1e7 to 5e7 CAR T cells	Yes (except for the first three patients)	Recruiting	Wang et al. (https://doi.org/10.21203/rs.3.rs-3454977/v1)
04903080	Multi-center trial by PBTC	Pediatric and young adult R/R ependymoma	HER2	IV	Up to three doses of 8e7/m2 CAR T cells	Yes	Recruiting	N/A
05835687	St. Jude Children’s Research Hospital	Pediatric and young adult R/R CNS tumors	B7-H3	ICT or ICV	Repeated doses of 1e7 to 1e8 CAR T cells	No	Recruiting	N/A
05298995	Bambino Gesù Hospital	Pediatric and young adult R/R CNS tumors	GD2	IV	Single dose of 1e6 to 6e6/kg CAR T cells	Yes	Recruiting	N/A
06221553	King Chulalongkorn Memorial Hospital	Pediatric DIPG	B7-H3	ICV	Single dose of 1e7 to 1e8 CAR T cells	No	Recruiting	N/A

Abbreviations: R/R, recurrent/refractory; ICT, intracerebrotumoral; ICV, intracerebroventricular; IV, intravenous; PBTC, pediatric brain tumor consortium.

Figure 1: Challenges faced by CAR T cells in pediatric CNS tumors.

Key obstacles encountered by CAR T cells in targeting pediatric CNS tumors include (1) Limited trafficking to the tumor: the selectivity of the blood-brain barrier (BBB) restricts CAR T cell infiltration into the tumor site, limiting their persistence and activity; (2) Immunosuppressive tumor immune microenvironment (TIME): the TIME, characterized by regulatory T cells (T_regs), abnormal vasculature, and immunosuppressive molecules, inhibits CAR T cell function after crossing the BBB; (3) CAR T cell exhaustion: upregulation of immune checkpoint molecules, such as PD-L1, B7, CTLA-4, and TIM-3, suppresses CAR T cell activation, proliferation, and cytotoxicity; (4) Tumor heterogeneity and antigen escape: reduced expression of tumor-associated antigens (TAAs) leads to immune evasion, allowing resistant tumor cell populations to persist and invade other CNS regions. Created in BioRender. Timpanaro, A. (2025) https://BioRender.com/u18j236

Anatomical barriers and toxicities: from systemic to locoregional CAR T cell delivery

Current strategies for administering CAR T cells greatly influence both effectiveness and potential toxicity of the therapy, and various delivery routes - including intravenous (IV), intrathecal (IT), intracerebroventricular (ICV), and intracerebrotumoral (ICT) - have been explored.

IV CAR T cell dosing has several advantages, including being the least invasive and most often the easiest to manufacture and to store, as fractionated dosing used in some ICV and ICT clinical trials (11) is generally not required. Most current CNS CAR T cell trials using locoregional delivery require a CNS catheter (e.g. Ommaya), and while this is a relatively safe neurosurgery and the complication risk is low (12), it is easier to place a central line for IV delivery. CAR T cells also demonstrate greater expansion in the peripheral blood where a single dose can provide relatively long-term surveillance, while intracranial dosing most often requires repeated dosing to account for the relatively shorter viability of CAR T cells in the CSF(7,13,14).

Effective trafficking of CAR T cells into the CNS is crucial for therapeutic benefit. While IV dosed CAR T cells have been shown to infiltrate glioblastoma (GBM) in adults, one study found that less than 5% of the T cell clonotypic repertoire in the pre-infusion CAR T cell product are identified in the post-treatment tumor biopsies(15). This may indicate limited trafficking of IV-dosed CAR T cells to the CNS tumor. Another study suggests that repeating IV infusions may have limited efficacy for adult GBM(16). Of note, GBM often has a disrupted BBB, a distinct situation from some malignant pediatric CNS tumors such as diffuse intrinsic pontine glioma (DIPG) and diffuse midline glioma, H3K27M-altered (DMG), which are among the most common pathologies enrolled on phase 1 trials. Preclinical studies also support that locoregional dosing may be superior to systemic dosing, but this phenomenon may not be universal or clinically equivalent (6,17). If limitations in IV dosing do exist, they could be mitigated by incorporating other evolving technologies, including focused ultrasound (FUS) or low-intensity pulsed ultrasound (LIPU) to promote intratumoral penetration(18). Additionally, CAR T cells can be engineered to express specific chemokine receptors, such as CXCR1 or CXCR2, that promote migration to the specific tumor(19). However, the range of biological subgroups of CNS tumors and the challenging anatomical locations (e.g. thalamus, pons) may require trafficking enhancements distinct for the site and tumor subtype.

Another potential downside to IV dosing is the risk of cytokine release syndrome (CRS) and immune cell-associated neurotoxicity syndrome (ICANS)(20). CRS typically occurs within the first two weeks after CAR T cell infusion and can range in severity from mild symptoms, such as tachycardia and fever, to life-threatening complications, while ICANS is a spectrum of neurologic changes that can follow CRS. These risks may be mitigated with intracranial delivery where limited presence of CAR T cells in the periphery may limit the biologic potential to cause the extensive avalanche of cytokines that drive those adverse events.

Currently, multiple centers have adopted intracranial CAR T cell dosing (Table 1). While some centers such as Seattle Children’s have used intracranial dosing exclusively(7,8,21), others, such as Stanford’s trial of GD2-targeting CAR T cells (GD2 CAR T cells), have used both methods - potentially providing objective opportunities to compare routes of administration(10). On that trial, CRS was dose-limiting for systemic GD2 CAR T cell dosing, while locoregional dosing resulted in no DLTs. The study did report that both IV and ICV dosed CAR T cells frequently caused tumor inflammation-associated neurotoxicity (TIAN), characterized by symptoms such as fever, headache, worsening of baseline neurologic deficits, or neurologic events(22). The more localized inflammatory changes, compared to the global cerebral edema sometimes seen with IV dosing of CD19-targeting cellular therapies, may be a more tolerable adverse event profile. A recently completed phase 1 trial of repeatedly intracerebroventricular B7-H3-targeting CAR T cells (B7-H3 CAR T cells) for children with DIPG at Seattle Children’s did not include TIAN grading but found no CRS or ICANS(21). Repeatedly-dosed ICV B7-H3 CAR T cells resulted in 1 DLT, which was an intratumoral hemorrhage. Hemorrhage is considered part of the natural history of DIPG(23) so on some trials it is not listed as an adverse event related to therapy, making it challenging to determine if intracranial CAR T cells carry an increased risk of hemorrhage. While some of the field is coalescing around locoregional delivery for potentially improved efficacy and more tolerable adverse events, it is important to note that the routes are also not mutually exclusive and that both efficacy and tolerability will almost certainly vary across different CAR T cell products. It is also important to note that some pediatric cancers are cured with intensive, life-threatening therapies, so if CAR T cells demonstrate substantial benefit, then even profound toxicities may be manageable. Ultimately, while early clinical results suggest ICV administration of CAR T cells may be the most promising therapeutic strategy, the goal will remain to identify the ideal route of delivery and subsequent toxicity management for each CAR T cell product and tumor subtype.

Tumor targeting: the search for optimal antigen targeting

As developmentally regulated tumors are often driven by few molecular aberrations(24), most pediatric CNS tumors express a paucity of neoantigens, limiting the number of targets for immunotherapies. The ideal cellular therapy target is an antigen absent in healthy tissues (to mitigate on-target off-tumor side effects) while uniformly present on tumor cells. Currently, the five antigens most commonly evaluated as CAR T cell targets preclinically and targeted on clinical trials in the United States are B7 Homolog 3 (B7-H3, also known as CD276)(25), epidermal growth factor receptor (EGFR) (26), the disialoganglioside GD2(27), human epidermal growth factor receptor 2 (HER2)(28), and interleukin-13 receptor alpha-2 subunit (IL13Rα2)(29) (Table 1).

B7-H3 is expressed on most pediatric CNS tumors with minimal expression on most normal tissues(25,30–32). B7-H3 is a B7 superfamily checkpoint protein that regulates complex immune interactions including inhibition of NK and T cell activation(33). One caveat to this target is that B7-H3 may be also upregulated on tumor-infiltrating activated T cells, raising the possibility of fratricide(34). However, B7-H3 CAR T cells have shown consistent preclinical efficacy against a range of CNS tumors with curative survival benefit in multiple mouse models(7,30,31). Therefore, fratricide may have a minimal effect especially as both systemic and intracranial trials have shown detectability of viable CAR T cells post-infusion on timelines similar to other CAR T cells against solid tumors(7,21,35). B7-H3 CAR T cells have been detected in the CSF days to weeks post infusion in GBM and DMG(7,36). Regarding tolerability, a recent report by Vitanza et al. showed the completed results from a first-in-human phase 1 trial of repeated ICV dosed B7-H3 CAR T cells for children and young adults with DIPG (BrainChild-03; NCT04185038)(21). While manageable grade 3 toxicities occurred, the only DLT was an intratumoral hemorrhage in one patient following dosing of the lowest CAR T cell dose (1x10⁷). There were no other DLTs or incidences of hemorrhage through escalation to the highest planned dose of 10x10⁷, which was confirmed as the highest tolerated dose regimen. The median survival for patients with DIPG following their initial CAR T cell infusion was 10.7 months and the median survival from diagnosis was 19.8 months with 3 patients still alive at 44, 45, and 52 months from diagnosis. Furthermore, there was evidence of cumulative tolerability as the 21 treated patients received a total of 253 doses, with one patient receiving 81 doses and still on protocol therapy nearly 4 years from diagnosis at the time of publishing. St. Jude also has presented preliminary findings from their Loc3CAR phase 1 trial (NCT05835687) of B7-H3 CAR T cells, showing a manageable safety profile and potential antitumor activity, as evidenced by neurologic changes, circulating-tumor-DNA clearance, and local immune responses(37). While adverse events like headache and fever were common, B7-H3 CAR T cells demonstrated early signs of efficacy with stable disease in some patients, warranting further exploration. These findings underscore the promise of intracranial B7-H3 CAR T cell therapy and plan for upcoming phase 2 trials.

EGFR overexpression and mutations in EGFRvIII result in an untethered surface EGFR protein identified on multiple pediatric CNS tumors, including ependymoma, high-grade glioma (HGG), and medulloblastoma (MB)(38,39). Expression of EGFRvIII promotes cell proliferation, angiogenesis, and invasion in different model systems(40,41). Preclinically, EGFR-targeting CAR T cells (EGFR CAR T cells) have shown efficacy against orthotopic CNS tumor models via either systemic or locoregional injection(42–44). Clinically, despite the tolerability of IV EGFRvIII CAR T cells in a phase 1 trial for adult GBM, but long-term clinical benefit may be affected by loss of antigen expression(15). Therefore, 806-based CAR T cells have been developed, incorporating the antigen-binding domain derived from the monoclonal antibody 806 that is able to recognize a unique conformational epitope present on EGFRvIII as well as a subset of overexpressed wild-type EGFR. This distinction allows 806-based CAR T cells to target a broader range of tumor cells while sparing normal tissues with low EGFR expression(5). Of note, variability in EGFR expression patterns between adult and pediatric CNS tumors may be a challenge, though there are clearly a subset of pediatric CNS tumors expressing the target.

GD2 is an overexpressed surface antigen on multiple pediatric solid and CNS tumors and promotes cell survival, invasion, antibody-mediated cell death, and T cell dysfunction(45). Strong interest for the high GD2 levels in solid tumors led to the development of GD2 CAR T cells, which have been subsequently tested in CNS tumors(3,46). Detection of GD2 expression in normal brain(47) and peripheral nerve cells(48), as well as a preclinical study in which GD2 CAR T cell-treated mice developed fatal encephalitis(49), initially raised concern for potential toxicity from GD2 CAR T cells. However, multiple clinical trials using GD2-targeting cellular therapies have not shown definitive evidence of off-tumor, on-target toxicity (10,50–56). Instead, peritumoral inflammation observed in GD2 CAR T cell-treated DMG mouse models may have predicted TIAN, the dominant toxicity observed in clinical trials. To minimize neurotoxicity risks, specific clinical pathways including the monitoring of intracranial pressure (ICP) or restriction to particular anatomical locations may be required. A recent report by Monje at al. reviewed their experience with patients receiving IV and ICV GD2 CAR T cells(10). Following IV dosing, 3 patients experienced DLTs related to CRS; however, there were no DLTs in the nine patients following ICV GD2 CAR T cells. Four patients had substantial radiographic responses, and one patient has had a complete response ongoing at time of publication 30 months from enrollment. A separate report by Lin et al. showed no DLTs following IV delivery of GD2 CAR T cells both with and without a constitutive IL-7 receptor, underscoring safety differences amongst different CAR T cell products even when the targeted antigen is the same(46). They also demonstrated multiple radiographic responses, supporting the potential clinical efficacy of CAR T cells against pediatric CNS tumors. Given the success of GD2 CAR T cells, additional international trials have been initiated (Table 1).

HER2 was one of the earliest CAR T cell targets identified on CNS tumors(9,57). When activated by ligand binding, HER2 can promote cell proliferation, survival, and invasion, contributing to tumor growth(58). Its expression in pediatric CNS tumors, in particular medulloblastoma and ependymoma(9), drove the development of HER2-targeting CAR T cells (HER2 CAR T cells)(8,59–61). Low levels of HER2 in vital organs initially raised safety concerns including a possible increased risk of pulmonary toxicities after systemic infusions(62). To mitigate side effects, HER2 CAR T cells were modified with an FRP5scFv-based exodomain, showing no dose-limiting toxicities in a phase 1 clinical trial for GBM including pediatric patients(59). Of note, HER2 was the target of the first intracranial CAR T cell clinical trial for children, which demonstrated the feasibility of fractionated dosing(8). Ultimately, preclinical efficacy and preliminary clinical experience prompted the initiation of further HER2 CAR T cell clinical trials (Table 1).

IL13Rα2 is a receptor upregulated on pediatric CNS tumors (63) and may be associated with a worse prognosis(10,46,64). Identified as a downstream target of EGFRvIII, IL13Rα2 has been shown to enhance tumor cell invasion and activate pathways that drive tumor cell proliferation(65). CAR T cells were developed to target IL-13Rα2 by utilizing an antigen binding domain consisting of membrane-tethered mutated IL-13 ligand, which allows preferential binding to IL-13Rα2 while reducing the binding affinity to IL-13Rα1. These IL13Rα2-targeting CAR T cells (referred to as IL13-zetakine+ T cells”), demonstrated effectiveness in orthotopic GBM mouse models(6,66), as well as remarkable radiographic tumor regression in adult GBM patients via locoregional delivery(67,68). These CAR T cells have also been ICV dosed in a phase 1 clinical trial for pediatric CNS tumors (NCT04510051), and the initial results showed that, although only three of five evaluable patients experienced temporary benefits that did not meet the study response criteria, no DLTs have been observed, making this therapy a potentially well-tolerated treatment (https://doi.org/10.21203/rs.3.rs-3454977/v1). Other important observations included the clonal expansion of endogenous CAR-negative CD8⁺ T cells in the CSF, but not in the peripheral blood (https://doi.org/10.21203/rs.3.rs-3454977/v1). These findings highlight the potential of IL13-zetakine+ T cell therapy as a promising and well-tolerated approach for GBM treatment, while underscoring the need for further optimization to enhance clinical efficacy and achieve sustained therapeutic responses.

Bridging insights: emerging adult CNS tumor targets in pediatric research

Further targets not extensively investigated in pediatric CNS tumors yet, but considered putative for the development of novel CAR T cell therapies include CD70, CD99, CD133, ephrin A2 (EphA2), ephrin A3 (EphA3), glypican-2 (GPC2), natural killer group 2 member D ligands (NKG2DLs), platelet-derived growth factor receptor alpha (PDGFRα), and protegenin (PRTG). These antigens play diverse roles in CNS tumor biology, ranging from promoting proliferation and immune evasion to maintaining cancer stem cell populations. For example, CD70, overexpressed in GBM, contributes to immune evasion by modulating T cell responses(29,69), while CD133 marks cancer stem cells in pediatric CNS tumors associated with treatment resistance and recurrence(70,71). In both cases, CAR T cell approaches revealed promising efficacy in eliminating tumor cells and reducing tumor burden in preclinical models(69,71,72). EphA2 and EphA3, members of the ephrin family of receptor tyrosine kinases, drive tumor growth, survival, and angiogenesis. Recent preclinical studies have shown successful targeting of EphA2 and EphA3 with CAR T cell therapy, making them attractive targets for selective CNS tumor disruption(73–76). GPC2, a glypican proteoglycan overexpressed across multiple pediatric CNS tumor types, has been targeted with GPC2-targeting CAR T cells, demonstrating substantial tumor regression in MB models and significantly prolonged survival in DMG models(4). Protegenin (PRTG) is a neural stem cell-associated protein implicated in the initiation and maintenance of aggressive pediatric CNS tumors, particularly Group 3 MB(77). PRTG-positive cells reside in a specialized neurovascular niche, making PRTG a promising therapeutic target for disrupting tumor growth and progression. PRTG-targeting CAR T cells have recently shown preclinical diminished tumor burden and longer survival in MB models(77), increasing the interest for possible clinical trials. Immune evasion mechanisms are also addressed by CAR T cells targeting NKG2DL, stress-induced ligands that facilitate immune escape in GBM(78). Preclinical studies involving these two targets have demonstrated significant tumor regression and prolonged survival in various CNS models(79,80). While these approaches have limited clinical experience, their compelling preclinical success underscores their potential as candidates for future CAR T cell therapeutic strategies against pediatric CNS tumor patients.

Tumor heterogeneity: advanced engineering strategies and combinatorial targeting

To address the issue of tumor heterogeneity, different CAR T cell strategies have been developed to target multiple antigens. Tandem-CAR T cells, characterized by two scFvs in tandem in the CAR antigen binding domain to recognize two independent targets, have been tested in preclinical GBM models targeting HER2-IL13Rα2, EGFRvIII-IL13Rα2, IL13Rα2-EphA2, or CD70-B7-H3, demonstrating diminished tumor antigen escape and enhanced efficacy(72,81–83). Recently, dual-CAR T cells targeting EGFR and IL13Rα2 (CART-EGFR-IL13Rα2) demonstrated preliminary safety, bioactivity, and early signs of efficacy in treating adults with GBM (NCT05168423)(16). The phase 1 study also reported manageable neurotoxicity(16), making this potentially relevant against pediatric CNS tumors.

CAR T cells with multiple antigen specificities can also be produced by expressing polycistronic CAR cassettes or co-transducing T cells with a mixture of viral particles encoding for various CAR constructs. For instance, trivalent-CAR T cells expressing a tricistronic cassette of CAR genes to simultaneously target EphA2, HER2, and IL13Rα2 have been preclinically tested by two different groups against MB, ependymoma, and GBM models(84,85). Clinically, a phase 1 trial (BrainChild-04; NCT05768880) has been initiated to assess the effectiveness of quad-targeting CAR T cells, generated with pooled viral vectors that provide CARs targeting B7-H3, EGFRvIII, HER2, and IL-13Rα2 (Table 1).

In addition to these multi-targeting approaches, another possible strategy is engineering CAR T cells to secrete bispecific T cell engagers (BiTE) to locally target an additional antigen. Choi et al. generated BiTE-expressing CAR T cells specifically targeting EGFRvIII and secreting a BiTE against the more broadly expressed EGFR(44). The BiTE-based CAR T cells, which recruited both CAR T cells and unmodified bystander T cells to target EGFR-positive GBM cells, effectively eliminated heterogeneous tumors in mouse models without causing systemic toxicity or side effects(86). This work has been translated into the clinical setting and demonstrated patient responses(14).

Combining CAR T cells with small molecule drugs to synergistically enhance bystander tumor death or using logic-gated CAR T cells have been preclinically shown to combat antigen escape. For example, inhibitor of apoptosis (IAP) antagonists have been combined with CAR T cells to treat antigen-heterogeneous GBM models, in which IAP antagonists could sensitize the tumor cells to apoptosis induced by CAR T cell-derived cytokines, such as TNF, therefore enhancing the bystander tumor death of antigen-negative tumor cells(87).

Logic-gated circuits aim to mitigate on-target off-tumor toxicity by providing a more precise mechanism that discriminates between healthy tissues and tumor cells(88) so that more homogeneously expressed tumor antigens could be safely targeted. Logic-gated strategies include: (1) ‘AND-Gates’, requiring the presence of two or more antigens on the cancer cell surface to activate and direct CAR T cells towards cells exhibiting both targets; (2) ‘NOT-Gates’, designed to inhibit CAR T cell activation in the presence of certain antigens, usually expressed on healthy tissues; and (3) ‘IF/THEN-Gates’, involving conditional activation of CAR T cells upon recognition of a specific antigen, which triggers a secondary further response for a better specificity control. Current studies of DIPG and ependymoma aim to target CD99/B7-H3 through an AND-Gated strategy to exploit high CD99 expression on DMG tumors and reduce fratricide killing among CAR T cells(89). Ultimately, future clinical trials will almost definitely require enhanced effector function, gated technology for maximum targeted tumor killing, or combinatorial targeting to produce a cure against such aggressive tumors.

The tumor immune microenvironment: overcoming a potentially hostile landscape

Many factors contribute to the TIME favoring tumor resistance to immune surveillance, such as the upregulation of immune checkpoint ligands (e.g., PD-L1(15)), recruitment of immunosuppressive immune cells (e.g., T_regs(15)), immunoinhibitory cytokines (e.g., TGF-β(90)), and stressful metabolic environments (e.g., nutrient deprivation(91)). The TIME is a complex network involving tumor cells as well as resident and trafficking non-tumor cells, such as endothelial cells, pericytes, fibroblasts, resident microglia, and other immune cells, each shaping the response to CAR T cell therapies(92). Importantly, the TIME varies among tumor types and between adult and pediatric patients. For example, adult HGG often harbors more infiltrated immune cells along with increased immunoinhibitory factors, while pediatric HGG may be more immunologically inert harboring fewer lymphocytes(93). For many CNS tumors, the TIME drives a cytokine milieu that fosters immune suppression and tolerance(94). T_regs and tumor-associated macrophages (TAMs) promote CNS tumor growth by producing immunosuppressive cytokines like IL-10 and TGF-β, which could diminish the efficacy of CAR T cell therapy(95,96) and result in T cell senescence(97). Ultimately, strategies to overcome the barrier of the TIME should be tailored based on the patient and tumor type and may require combination of multiple approaches.

As demonstrated in preclinical studies, CAR T cells directed against TAMs or resistant to TGF-β may increase antitumoral activity and shift other immune cells, such as microglia, from a pro-tumorigenic to a pro-inflammatory phenotype(98). Optimized CAR T cell designs, also including T cells redirected for universal cytokine-mediated killing (TRUCKs), are being developed to improve the efficacy and safety of treatments for CNS tumors. TRUCKs, designed to release high levels of cytokines like IFN-γ and IL-15, can remodel the CNS TIME by activating immune cells and reducing suppression, thereby maintaining a stem-like memory phenotype of CAR T cells in favor of CAR T cell persistence(99). Similarly, GD2 CAR T cells that constitutively express IL-7 receptor augmented cytokine responses in pediatric CNS tumor patients and may be more clinically efficacious than their standard counterparts(46). These studies suggest promising strategies that enable CAR T cells to adapt to the suppressive pediatric CNS TIME, supporting their persistence and enhancing their antitumoral activity.

Beyond “lymphodepletion”: TIME modulation

Against hematologic malignancies, lymphodepletion is leveraged for maximal efficacy of systemically-delivered CAR T cells, and there has been substantial interest in whether this would also be useful for children with CNS tumors. Preliminary data from the Stanford’s GD2 CAR T cell trial and from the City of Hope IL13-zetakine+ T cell trial suggest that lymphodepletion prevents or delays onset of anti-CAR immune responses detectable in the CSF ((10), https://doi.org/10.21203/rs.3.rs-3454977/v1). While lymphodepletion prior to systemic CAR T cell dosing may have clinical value, against CNS tumors receiving locoregional CAR T cells, the systemic lymphocyte population may not be the best target of immunomodulation. Therefore, we will refer to conditioning regimens targeting the TIME as “TIME modulation” rather than “lymphodepletion.” Pre-CAR T cell TIME modulation via CNS-penetrant drugs or focal radiation ideally would deplete immunosuppressive TIME elements or repolarize antagonistic components into CAR T cell allies. Additionally, TIME modulation may abrogate or mitigate the development of anti-CAR immune responses. For systemic CAR T cell therapy against CNS tumors, lymphodepletion may augment peripheral CAR T cell proliferation, although there may be increased systemic toxicity. However, TIME modulators that are not systemically lymphodepleting may enhance locoregional CAR T cell activity without systemic risk. Further studies need to define which systemic agents appropriately modulate the TIME to induce clinical beneficial and, if so, in what settings. Additionally, data on the optimal frequency of TIME modulation in the setting of the repeated CAR T cell dosing schedules does not exist and ongoing clinical approaches have varied (e.g. Stanford University’s GD2 CAR T cell trial is administering repeated “lymphodepletion”, whereas City of Hope’s IL13-zetakine+ T cell trial is not). Ultimately, as with all treatments especially for children, more targeted solutions would clearly be preferable; understanding the role of TIME modulation more precisely will help to develop more elegant approaches to abolishing or leveraging the TIME.

Combinatorial approaches to address immune suppression in CNS tumors

Given the scarcity of pediatric CNS tumor-specific TIME studies, some strategies used to counteract immune suppression in adult patients can offer valuable insights for developing combinatorial approaches to overcome barriers in pediatric CNS tumors. Combining CAR T cell therapy with other treatments, such as immune checkpoint inhibitors (ICIs), may have a combinatorial cytotoxic effect, promote antigen spread, and induce a more amenable TIME for CAR T cell activity.

ICIs are a class of immunotherapies leveraging the patien’s immune system via inhibited interaction between immune checkpoint molecules on immune cells and their ligands. Some of these have highlighted remarkable improvement on CAR T cell persistence via CRISPR/Cas9-based knockout of the immune checkpoint regulator genes PD-1, LAG3, or CTLA-4(100–102), secretion of a programmed death ligand 1 (PD-L1) blocking antibody(103), and engineering of T cells with switch receptors to convert immunosuppressive signals into stimulatory ones(104). Upregulation of the immune checkpoint ligand PD-L1 has been observed in adult GBM(15), justifying combinatorial strategies with anti-PD-1 monoclonal antibodies (pembrolizumab and nivolumab) which have shown safety and anti-tumor activity in preclinical models(105). Notably, a recent phase 1 trial in adult EGFRvIII-positive GBM patients (NCT03726515) evaluated the combination of IV EGFRvIII CAR T cells with the anti-PD1 monoclonal antibody pembrolizumab, demonstrating safety and biological activity, despite lacking substantial efficacy(106). However, exploratory analyses revealed substantial evolution of the TIME after treatment, including an increase in exhausted, regulatory, and interferon-stimulated T cells at relapse(106), highlighting the need for alternative strategies to overcome immune suppression.

An alternative inhibition of other immune checkpoints, such as TIM-3 and CTLA-4, holds promise for pediatric CNS tumors, as demonstrated by encouraging results against preclinical models. For example, targeting TIM-3 has been shown to promote tumor regression in DIPG(107), while also inducing a durable immune memory response through increased microglia activity and enhanced CD8⁺ T cell anti-tumor response(108). Similarly, CTLA-4 inhibition has improved T cell functionality within the TIME of glioma models(109). These results highlight the potential of feasible combinatorial strategies integrating ICI with CAR T cell therapy and represent a logical trend and a promising approach for pediatric CNS tumors.

Additional strategies driving CAR T cell persistence and enhanced efficacy

Overcoming the suppressive TIME in CNS tumors requires innovative approaches that address both immune cell exhaustion and phenotypic alterations. Epigenetic modulators have emerged as potential tools to reprogram CAR T cell functionality, enhance persistence, and overcome barriers within the TIME. Although these technologies have not been extensively applied to clinical trials against pediatric CNS tumors yet, their feasibility demonstrated in multiple CNS models suggests their potential to improve CAR T cell efficacy in the brain.

Targeted agents such as inhibitors of histone deacetylase (HDACi) can provide direct anti-tumor cytotoxicity while also potentially shaping the favorable phenotypes of both native and engineered immune cells in pediatric CNS tumors(110–113). For instance, the pan-HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) has been shown to upregulate B7-H3 expression in GBM, thereby enhancing the antitumor activity of B7-H3 CAR T cells(114). Complementing this, DNA methyltransferase inhibitors (DNMTi) have shown the ability to boost CD8⁺ T cell anti-tumor efficacy by upregulating MHC class I antigen presentation on glioma cells while simultaneously preventing CAR T cell exhaustion(115,116). Preclinical data in CNS tumor models also revealed that pre-treating B7-H3 CAR T cells with DNMTi can promote memory T cell phenotype during production(117). To address the challenge of maintaining T cell fitness in the immunosuppressive TIME, BET protein inhibitors (BETi) may ameliorate CAR T cell exhaustion by reducing inhibitory receptor expression and improving metabolic fitness, leading to enhanced proliferation and T cell reinvigoration(118). Similarly, LSD1 inhibitors (LSD1i) can alleviate CD8⁺ T cell exhaustion, enhance persistence, and increase effector cytokine production, resulting in greater antitumor efficacy in mouse models(119). A recent study also demonstrated that priming CAR T cells with LSD1i promotes their persistence and antitumor efficacy(119). Furthermore, preliminary data in DMG models suggest that combination of HDACi and LSD1i may increase GD2 expression on the tumor cells, making them more susceptible to GD2-directed CAR T cell treatment, thus improving therapeutic efficacy(120). Similar effects were observed also in MB models in which GD2 levels increased following treatment with tazemetostat, a EZH2 inhibitor, improving cytotoxic activity of anti-GD2 CAR T cells(121). In addition to this repertoire, EZH2 inhibitors (EZH2i) enhance T cell tumor infiltration and boost the efficacy of ICIs in preclinical models(122,123), while ablation of histone methyltransferases (HMT) such as SUV39H1 improves CAR T cell persistence and promotes the generation of stem-like memory T cells(124–126).

These epigenetic strategies (Table 2) represent a promising avenue to reprogram CAR T cells, enhancing their activation, persistence, and memory phenotype while simultaneously exerting direct anti-tumor effects. While many combinatorial approaches remain unexplored in pediatric CNS tumors and might induce immune modulation(127), the observed benefits of CAR T cell priming with epigenetic therapeutics underscore the potential to investigate such combinations in this context. By targeting the functional barriers imposed by the TIME, these tools have the capacity to significantly enhance CAR T cell efficacy and durability in CNS malignancies.

Table 2.

Main inhibitors used in combination with CAR T cell therapy

Class	Full name	Primary target(s)	Mechanism of action
HDACi	Histone Deacetylase Inhibitors	Histone deacetylases (HDACs)	Inhibit histone deacetylases, leading to increased histone acetylation and transcriptional activation.
DNMTi	DNA Methyltransferase Inhibitors	DNA methyltransferases (e.g., DNMT1, DNMT3A, DNMT3B)	Inhibit DNA methyltransferases, leading to DNA demethylation and reactivation of silenced genes.
BETi	Bromodomain and Extra-Terminal Inhibitors	BET proteins (e.g., BRD2, BRD3, BRD4, BRDT)	Block the bromodomains of BET proteins, disrupting their role in chromatin remodeling and gene transcription.
LSD1i	Lysine Demethylase 1 Inhibitors	Lysine demethylase 1 (LSD1/KDM1A)	Inhibit LSD1, preventing the removal of methyl groups from histone H3 and affecting gene expression.
EZH2i	Enhancer of Zeste Homolog 2 Inhibitors	EZH2 (a component of the Polycomb Repressive Complex 2, PRC2)	Block EZH2-mediated H3K27 trimethylation, leading to repression of silenced genes.
HMTs	Histone Methyltransferase Inhibitors	Various histone methyltransferases (e.g., SUV39H1, G9a, DOT1L)	Target specific histone methyltransferases, altering histone methylation patterns to regulate gene expression.

In addition to epigenetic approaches, commonly dosed agents such as dexamethasone, bevacizumab, and temozolomide must be considered for potentially antagonist or beneficial effects within the TIME. Dexamethasone, used to treat cerebral edema, may hinder CAR T cell activity and has demonstrated negative effects in vaccine trials for adult GBM(128). By contrast, bevacizumab, an antibody targeting VEGF, may reduce cerebral edema while enhancing lymphocyte trafficking, potentially improving CAR T cell therapy(129). Similarly, temozolomide, a cytotoxic chemotherapeutic agent, has been shown to support CAR engraftment and CAR T cell persistence, leading to tumor regression in preclinical GBM models(130). These promising results have contributed to start an EGFRvIII CAR T cell phase 1 trial against adult glioma (NCT02664363).

Ultimately, the integration of CAR T cell therapy with adjunctive treatments like radiation, chemotherapy, and epigenetic modulators offers a promising strategy to enhance the efficacy of immunotherapy for CNS tumors. By modifying the tumor microenvironment or improving CAR T cell functionality, these combinatorial strategies may be critical in maximizing the potential of CAR T cell therapies. Direct cytotoxic effects by these treatments will likely be needed as well, as such aggressive cancers are unlikely to be cured with single therapeutic regimen. While these combinatorial immunotherapeutics thus far are more commonly tested against adult tumors, they offer valuable insights for translation into pediatric CNS tumors.

Patient-centered metrics: experience and equity

Once children are enrolled on trials, it is important to measure the myriad aspects of the patient experience in parallel with other correlative data. Paired assessments of CAR T cell activity or presence can be recorded in the context of individual patient-level symptoms and outcomes especially as a range of components that have a substantial impact on quality of life, such as word-finding, clarity of thought, and coordination, are almost never captured reliably in children. Thus, attention to these patient-centered outcomes including physical symptom burden, psychosocial aspects of care, access to therapy, and potential biobehavioral implications (131–133) can be important to the implementation of timely and targeted supportive care interventions, to making these potentially toxic though beneficial therapies more tolerable, and to advance from phase 1 highly monitored trials into more sustainable and accessible treatment regimens.

Patient reported outcomes (PROs) include any information coming directly from patients, without secondary interpretation from others about how they feel or function. PRO instruments (or proxy reported instruments when children are developmentally or medically unable to self-report) can capture anticipated and unanticipated effects of any intervention. In the context of early phase clinical trials, PROs are becoming more established as clinical trial endpoints and are a stated priority for regulatory bodies(134,135). Naturally, there will be unique considerations for the collection of PROs in young patients receiving CNS CAR T cells. Developmental and cognitive abilities of the patient, nature of potential symptoms, and overall burden to patients and families must be considered when selecting and integrating PRO measures. Validated surveys capturing neurocognitive, psychosocial, functional, and pain-related symptoms can be delivered electronically in shortened form, which is well-suited to CNS CAR T cell trials(136). Given the dynamic quality of CAR T cell-related symptoms, sequential PRO data, which even on a weekly schedule has been shown to be feasible in children(133), can provide valuable data that can improve the frequency of CAR T cell dosing and the timing of supportive measures.

In order to make early phase CNS CAR T cell trials as generalizable as possible, demographic diversity in enrollment is critical. Unfortunately, many factors including access to care, resources to travel across the country for clinical trials, and often the requirement to relocate during treatment worsen disparities in access to CAR T cell therapies. For example, it is known that a lower proportion of publicly insured and Hispanic patients are referred for pediatric CAR T cell trials(137). Adult studies have shown similar findings; African American patients are less likely to receive CAR T cell therapy than other racial/ethnic groups, as are patients belonging to lower socioeconomic strata(138). Inequitable access to CAR T cell trials not only exacerbates healthcare disparities among historically marginalized groups but also introduces bias that may obscure our understanding of these therapies.

Validating the preliminary experience: novel trials and cooperation

Over the past 5 years, advanced pediatric CNS CAR T cell programs - in particular, those at City of Hope, Seattle Children’s, Stanford, and Texas Children’s - have driven the CNS CAR T cell field from encouraging preclinical results to a series of foundational early phase clinical trials (Table 1). Through this cumulative early pediatric experience, the field can set its eyes on the next phase. Future clinical trials should iterate on the recent wealth of preclinical testing and correlative studies from the first wave of cellular therapy trials for children with CNS tumors. Recent discoveries should be aimed at both building toward engineering enhancements but also delivering the optimal sequence and most effective direct combinatorial regimens. This may include combined delivery of systemic and locoregional agents and optimized sequences of dosing. Investigation of the TIME will be critical but will be a significant challenge in the setting of different CNS tumor biology and locations. There is already an immense amount of correlative data from individual trials including CAR T cell detection, secondary metrics of effector function and trafficking, tumor and TIME sequencing, and imaging files that are of much less utility when siloed but could be powerful when reviewed together. This cooperation will hopefully lead to future multi-site trials, critical for validating current single-site experiences and to identify opportunities to expand access to these therapies. A combined mission that encompasses preclinical evaluation of advanced CAR T cells, biologic understanding of the TIME, improved correlative data sharing, leveraging of patient reported outcomes, and equitable access to care on future clinical trials and novel therapeutics will provide a path to more consistent success as the pediatric CNS CAR T cell field enters its next phase.

Summary

CAR T cell therapy for pediatric CNS tumors have undergone rapid advancement from the bench to the bedside. Multiple trials are ongoing or recently completed, building a body of data on which to design the next generation of trials. Important information has already been harnessed from these trials, including the potential advantages of locoregional delivery, unique toxicities such as TIAN, and deeper understanding of the microenvironment. Additional work to define tumor targets and factors affecting T cell function in the CNS continue, with a goal to design rational combinatorial approaches. The next iteration of trials must also seek to ensure the entire complexity of the patient experience is considered, as traditional outcome measures fall short of capturing the full spectrum of improvement and burdens. Our knowledge of effective CNS CAR T cells will continue to evolve and hopefully improve the outcomes for children with CNS tumors.

Significance.

Central nervous system (CNS) tumors are the leading cause of cancer related death in children, highlighting the dire need for new treatment strategies. CAR T cells represent a unique approach, distinct from the cytotoxic chemotherapies and small molecule inhibitors that have dominated the clinical trial space for decades. Phase 1 CAR T cell trials have shown feasibility and possible efficacy against pediatric CNS tumors; however, many challenges must be overcome if these therapeutics are going to be beneficial to most affected children. While rapid translational development and early phase trials have quickly evolved our understanding, the pediatric CNS CAR T cell community now yearns for critical assessments and open dialogue about overcoming the remaining obstacles ahead.