Germline variants influence immune checkpoint inhibitor responses, but their role in engineered immune cell therapies, such as chimeric antigen receptor T cells (CAR T cells), remains unclear. We integrated whole germline sequencing from patients with lymphoma treated with axicabtagene ciloleucel CAR T cell products in the ZUMA-1 and ZUMA-7 clinical trials with detailed biomarker and functional analyses to identify variants influencing clinical toxicity and pharmacokinetics. Putative deleterious variants in STXBP2 (syntaxin binding protein 2) were enriched among patients with toxicity in ZUMA-1, although not confirmed in ZUMA-7. Mechanistically, STXBP2-deficient or variant-expressing T cells triggered increased inflammatory cytokine production and macrophage activation. Conversely, variants in ADAMTSL3, a TGFβ (transforming growth factor-β) signaling regulator, correlated with protection from toxicity across both trials. Furthermore, variants in PTPN22, a negative regulator of T cell receptor signaling, strongly associated with enhanced CAR T cell expansion, a key determinant of efficacy. Together, these findings demonstrate that germline genetics shape the safety and activity of engineered immune cell therapies, affecting future design and patient management.

ABSTRACT:Intractable diarrhea is a recently described complication following B-cell maturation antigen (BCMA)-targeted chimeric antigen receptor (CAR) T-cell therapy for multiple myeloma with reported mortality rates of 36% to 50%. The optimal clinical management is unknown. Here, we report a series of 5 patients who presented with severe diarrhea after BCMA CAR T-cell treatment. We hypothesized that the Janus kinase inhibitor, ruxolitinib, might be an effective therapy based on its success in graft-versus-host-disease after allogeneic bone marrow transplant and other immune-driven diarrhea syndromes. Three patients received ruxolitinib, all of whom experienced rapid clinical improvement. Among the 2 patients with matched pretreatment and posttreatment biopsies, both showed signs of histopathologic response, including 1 with CAR T-cell-associated indolent T-cell lymphoproliferative disease of the gastrointestinal tract.

Characterizing the m6A epigenetic landscape is essential for understanding glioma biology, yet transcriptome-wide mapping of these modifications at isoform resolution across specific tumor subtypes has remained limited. Conventional short-read approaches lack the capacity to resolve full-length transcript isoforms or assign m6A modifications to individual transcripts, representing a critical gap in glioma where alternative splicing is pervasive. Methods: We performed direct RNA nanopore sequencing and transcriptome-wide m6A analysis in 14 glioma tumor tissues, including IDH1-mutant astrocytoma, oligodendroglioma, and IDH1 wild-type glioblastoma, enabling isoform-resolved profiling not accessible by conventional short-read approaches. m6A sites were predicted computationally using the m6Anet deep learning framework, which has been independently benchmarked against MeRIP-seq-derived sites, and high-confidence calls were defined at a probability threshold of ≥0.9 and required detection across multiple patients within each subtype. Results: IDH1-mutant gliomas showed a higher overall burden of computationally inferred m6A-modified sites, transcripts, and genes than IDH1 wild-type glioblastoma, along with variation in transcript biotypes, regional distribution of m6A sites, and extent of isoform methylation. Differential methylation analysis identified subtype-specific patterns of m6A localization, many of which were observed without corresponding changes in gene-level expression, indicating that m6A variation represents a post-transcriptional regulatory layer not captured by gene-level analysis alone. Integration of gene expression, isoform usage, and m6A status further identified variation in isoform composition and transcript features between astrocytoma and glioblastoma. Analysis of m6A regulators showed subtype-associated expression patterns among readers, writers, and erasers, and exploratory analyses identified isoform-level associations with survival that were not apparent at the gene level. Conclusions: Overall, these data describe subtype-specific patterns of m6A marking and isoform architecture across glioma tissues, derived from computational inference using direct RNA sequencing in a modestly sized cohort and warrant validation by orthogonal methods in larger studies. These findings are consistent with concurrent independent evidence that isoform-specific m6A deposition is evolutionarily conserved across mammals and that long-read isoform resolution reveals transcript diversity in glioma not captured by gene-level analysis. While cohort size and the absence of orthogonal site-level validation suggest that the data require cautious interpretation, this work provides a hypothesis-generating resource and methodological framework for future mechanistic and translational investigation of the glioma epitranscriptome.

CD19-directed chimeric antigen receptor T-cell therapy (CD19-CAR) has yielded encouraging efficacy in CNS lymphomas (CNSL), but most patients ultimately experience progressive disease (PD). Risk factors, progression patterns as well as optimal salvage therapies remain unclear. Clinical and radiological characteristics of CD19-CAR failure were therefore retrospectively defined in CNSL treated at Massachusetts General Hospital from 2018 to 2024. PD patterns were defined as local or distant. CNS-progression-free survival from CD19-CAR infusion (CNS-PFS1) and first subsequent progression (CNS-PFS2) were analyzed. CD19-CAR achieved a 60

BACKGROUND:Mechanisms driving aggressive meningiomas remain poorly understood. Given the pivotal role of the immune microenvironment in tumor progression, we developed a comprehensive atlas of the meningioma microenvironment, with a view toward identifying modifiable opportunities. METHODS:The immune microenvironment of 2,727 meningiomas was profiled using orthogonal methodologies, including 24 with mass cytometry, 24 single-cell RNAseq, 1,437 bulk RNAseq, 1,125 DNA methylation, 117 multiplex immunofluorescence, as well as that of 5 paired peripheral blood samples and 10 human meninges. Patient-derived organotypic tumor spheroids (PDOTS) were established to assess the effect of STING stimulation combined with anti-PD-1 treatment. RESULTS:We revealed a rich immune infiltration in meningioma, among the highest across 34 human cancer types (n = 12,188). Macrophages predominated in meningioma microenvironment, in contrast to the lymphoid dominance of peripheral blood, with meninges exhibiting an intermediate immune profile between meningiomas and peripheral blood. Cellular states and phenotypes of both immune and tumor cells shifted during tumor progression, with aggressive meningiomas possessing earlier-stage, immunosuppressive immune cells and proliferative tumor cells. Using ex vivo meningioma PDOTS, we demonstrated inducible responses to STING activation, marked by elevated cytokine release, which were synergistic when combined with PD-1 blockade. CONCLUSIONS:These findings offer an extensive resource on the cellular heterogeneity of the meningioma microenvironment and provide a framework for rational therapeutic modeling and strategy development.

2059 Background: Chimeric Antigen Receptor (CAR) T cells for glioblastoma (GBM) have been limited by the challenge of targeting a single tumor antigen in a heterogeneous disease. To address this barrier, we generated a novel engineered T-cell product (CARv3-TEAM-E) that targets the EGFRvIII antigen while also secreting T-cell-Engaging Antibody Molecules (TEAMs) against wild-type EGFR. Methods: The INCIPIENT clinical trial is a first-in-human study of CARv3-TEAM-E in patients with GBM (NCT05660369). Patients with recurrent GBM were treated with intraventricular CARv3-TEAM-E T cells (10E6 cells per infusion) and one of three pre-treatment regimens: no lymphodepletion (N=3), lymphodepleting chemotherapy with cyclophosphamide and fludarabine (LDC) (N=7), or cyclophosphamide, fludarabine, and rituximab (LDC+R) (N=3). The primary objective was safety and tolerability. Immune cells were profiled in the cerebrospinal fluid (CSF) and peripheral blood by flow cytometry. Results: CAR T manufacturing was successful for all patients. There were no dose-limiting toxicities (DLT). Three patients were treated without LDC; all developed anti-CAR and/or anti-TEAM antibodies (i.e., anti-therapy antibodies) after a single infusion of CARv3-TEAM-E. Subsequently, 7 patients were treated with LDC prior to CARv3-TEAM-E, 5 of which underwent serial infusions (range 2-5 infusions). Four patients had reinfusions after which CAR T cells were not detected in the CSF. We therefore added rituximab to the LDC regimen. In the 3 individuals pre-treated with LDC+R, CAR T cells were detected in CSF not only after initial infusion (range 21-28 days) but also following repeat infusion in all patients (range 3-15 days post-reinfusion). Anti-therapy IgG was not detected in patients treated with LDC+R, as compared to 9/10 patients who developed an antibody response when rituximab was not administered. As of 12/15/2025, 10/13 patients are alive 6-30 months after first infusion, with 7 alive >14 months. Conclusions: Intraventricular CARv3-TEAM-E infusions were well-tolerated after LDC+R pre-conditioning and no DLTs were noted. The addition of Rituximab abrogated anti-therapy antibody formation and prolonged CAR T-cell persistence in 3/3 patients. Survival data reflect continued promise of CARv3-TEAM-E in patients with recurrent glioblastoma. Clinical trial information: NCT05660369 .

Neuronal-glioma interactions are increasingly recognized as critical in the development and progression of central nervous system tumors. Recent research highlights that gliomas can integrate into neural circuits through various mechanisms, including the synaptogenic factor thrombospondin-1 (TSP-1). This new mechanistic understanding of cancer neuroscience allows for novel insights into target discovery. Critically, therapies that modulate neuron-tumor interactions remain agnostic to other oncogenic changes within tumor cells yet may still target fundamental drivers of tumor growth. In line with these findings and controlling for critical confounding variables, we demonstrate a survival benefit associated with gabapentin (an antagonist of TSP-1) following surgical resection of newly diagnosed glioblastoma. This retrospective, multi-institutional cohort study included 1,072 patients, with a discovery cohort of 693 patients and an additional 379 patients from a separate site for external validation. Furthermore, our findings indicate that gabapentin administration is associated with reduced serum TSP-1 levels, suggesting its potential as a future biomarker.

ABSTRACT:Tumor inflammation-associated neurotoxicity (TIAN) was recently proposed as a unique complication of immunotherapy in patients with brain tumor. Here, we report a first comprehensive characterization of TIAN in patients with central nervous system (CNS) lymphoma (CNSL) treated with CD19-directed chimeric antigen receptor (CD19-CAR) T cells. TIAN occurred in 10 of 56 (17.9%) patients with CNSL, with clinical onset at a median 3.5 days (range, 1-9) after CD19-CAR T-cell infusion. It was less frequently associated with cytokine release syndrome (60% vs 100%; P = .009) than immune effector cell-associated neurotoxicity syndrome (ICANS). Although symptoms were usually transient and fully reversible, TIAN was associated with a fatal outcome in 1 patient. Larger CNS tumor volume at baseline allowed the identification of patients at risk for TIAN (area under the curve, 0.847; P = .002). Maximizing Youden J statistics, a discriminatory tumor volume threshold of >3.4 cm3 was determined, which carried 87.5% sensitivity and 80.5% specificity. TIAN correlated with higher overall response rates to CD19-CAR T cells (90% vs 52%; P = .036) and improved progression-free survival (hazard ratio, 0.22; 95% confidence interval, 0.07-0.61; P = .006) on multivariate Cox proportional hazard regression. Postmortem histopathological evaluation of a TIAN lesion revealed a dense macrophage population with central necrosis and peripheral reactive gliosis, accompanied by loss of white matter and intracytoplasmic myelin in foamy macrophages. Collectively, our work supports TIAN as a localized on-tumor, on-target neurotoxicity syndrome, closely related to preexisting CNSL lesions and distinct from ICANS. CNS tumor volume at baseline may allow to identify patients at risk and may guide management.

Chimeric antigen receptor (CAR)-T cell therapies demonstrate potent anti-tumor efficacy in hematologic malignancies, yet clinical outcomes remain unpredictable due to the bespoke nature of the treatment, which is manufactured from each patients own T-cells. While germline variants are known to influence response to immune checkpoint inhibitors their role in CAR-T cell therapy is unknown. Here, we pair whole-genome germline sequencing of lymphoma patients from the ZUMA-1 and ZUMA-7 clinical trials of axicabtagene ciloleucel, along with correlative biomarkers and functional assays, to ascertain the impact of germline variants on CAR-T cell behavior. Hypothesizing shared mechanisms of the most common toxicities of CAR-T cells, namely cytokine release syndrome (CRS) with hemophagocytic lymphohistiocytosis (HLH), a hyperinflammatory syndrome driven by T cell overactivation, we first looked within 17 canonical HLH-associated genes, and identified putative deleterious STXBP2 variants in 15% of ZUMA-1 patients with toxicity, which were absent in control subjects who did not experience high grade toxicity. Subjects with these variants had elevated baseline IFN-g; and inflammatory cytokines, findings that were recapitulated in engineered STXBP2-deficient and STXBP2-variant-expressing primary CAR-T cells derived from healthy donors. However, STXBP2 variant enrichment was absent in ZUMA-7 for this toxicity phenotype, possibly reflecting differences in underlying disease burden and evolving clinical management between the trials. A more expansive genome-wide analysis revealed ADAMTSL3 (a negative regulator of TGFB) as the only gene nominally enriched for putative deleterious variants in both ZUMA-1 and ZUMA-7 among control subjects, suggesting a protective effect. Finally, we focused on associations between germline variants and CAR-T cell expansion after infusion, a more objective and granular continuous variable that is strongly associated with clinical response across most CAR-T products8. We found a strong association between PTPN22, a known negative regulator of T-cell activation and an autoimmune risk gene variant status and CAR-T cell expansion in both ZUMA-1 and ZUMA-7, with the patients having the highest level of CAR-T expansion across clinical trials harboring variants in the gene. Together, these findings demonstrate the first clear association between germline variants and the clinical behavior of engineered immune cell therapies, which has implications for cellular therapy design, monitoring, testing, clinical trial design, and patient care. ### Competing Interest Statement JB, SF, and RRS are Kite Pharma employees. M.V.M is an inventor on patents related to adoptive cell therapies, held by Massachusetts General Hospital and the University of Pennsylvania (some licensed to Novartis). Dr. Maus holds equity in TCR2, Century Therapeutics, Genocea, Oncternal, and Neximmune, serves on the Board of Directors of 2Seventy Bio, and has served as a consultant for multiple companies involved in cell therapies. MVM interests were reviewed and are managed by Massachusetts General Hospital, and Mass General Brigham in accordance with their conflict-of-interest policies. BDC, MVM, and MBL are inventors of patents related to the use of engineered cell therapies. BDC received commercial research grants from ACEA Biosciences. R.G. has equity in Moderna, Pfizer, and Vertex Pharmaceuticals. ### Funding Statement This study was funded by Kite Pharma. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: We obtained available genomic DNA from patients in the ZUMA-1 and ZUMA-7 clinical trials. All patients had provided written informed consent as part of the ZUMA-1 and ZUMA-7 studies at their respective institutions. None of the authors of this manuscript had access to identifiable patient information and nor did they participate in the consent process for the individuals. For primary T-cells used in experiments: Human T cells were purified (Stem Cell Technologies, #15061) from healthy donor leukopaks whose use was determined to be non-human subjects research by the Institutional Review Board (IRB) at the Massachusetts General Hospital (MGH). I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The data that support the findings of this study are available from Kite Pharma, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available due to patient privacy concerns. Data are however available from the authors upon reasonable request and with permission of Kite Pharma.

ABSTRACT:CD70 has emerged as a promising target in acute myeloid leukemia (AML), and we have previously demonstrated the potency of an optimized CD70-targeted ligand-based chimeric antigen receptor (CAR). However, here, we identify in vivo CD70 antigen escape as a limitation of single-antigen targeting. Combination targeting of CD70 and CD33 may overcome AML antigen heterogeneity. We hypothesized that modifying our CD70 CAR platform to secrete a bispecific T-cell engaging antibody molecule (TEAM) targeting CD33 (7033) would create a therapeutic window whereby AML heterogeneity could be addressed without increasing tissue toxicity. We found that CD33 TEAMs mediated specific cytotoxicity across AML cell lines, including CD33 or CD70 single-antigen knockout tumors. 7033 CAR T cells eradicated tumor in an in vivo mixed tumor model of CD70 antigen escape and outperformed the previously optimized CD70 CAR in a patient-derived xenograft. In vivo gene expression profiling of CAR T cells revealed enhanced 7033 CAR T-cell pathway scoring for persistence, activation, and T-cell receptor signaling. Additionally, CD33 TEAMs successfully redirected T cells isolated from patients with AML to activate, secrete cytokines, and kill tumor targets despite exposure to substantial prior cytotoxic therapies. In summary, our findings demonstrate the feasibility of our 7033 CAR to overcome AML heterogeneity and leverage the bystander T cells of patients; this approach warrants further study in patients with this dire clinical need.

Mutations in isocitrate dehydrogenase (IDH) are a significant prognostic and biological factor leading to slower growth and T cell suppression within diffuse gliomas. However, the effect of IDH and its downstream metabolites specifically on intratumoral myeloid cells remains underexplored. Utilizing patient tumor samples, we performed RNA-sequencing and quantitative immunofluorescence on IDH-wildtype glioblastoma and IDH-mutant grade 4 astrocytoma cases. We then engineered the murine GL261 glioma cell line to harbor mutant IDH, comparing transcriptomic and cell-level changes in IDH-wildtype versus IDH-mutant murine tumors. We identified greater hallmarks of productive inflammation in IDH-mutant tumors compared to IDH-wildtype tumors. We also saw transcriptomic enrichment of suppressive macrophage and myeloid-derived suppressor cell (MDSC) signatures in IDH-wildtype tumors, which was confirmed at the cellular level. Furthermore, engineering the IDH mutation into murine tumors appeared sufficient to recapitulate many of the transcriptomic and cellular shifts observed among patient samples. Our data show that mutant IDH is associated with greater inflammatory signatures and fewer suppressive myeloid cells in human gliomas, and that delivering mutant IDH to murine tumors is sufficient to drive these microenvironment changes. This work advances our understanding of key myeloid cell populations that may be targeted by future immunotherapy strategies.

Brain metastases (BMs) affect approximately 10-30 % of cancer patients, and their prevalence is growing as patients live longer with controlled primary disease. Surgical resection remains a cornerstone of treatment for both solitary and multifocal lesions. Since the advent of intracranial tumor surgery, neurosurgery has trended towards less invasive surgical approaches, facilitated by a proliferation of surgical innovations ranging from intraoperative MRI to tubular retractors. Minimally invasive cranial surgery (MICS) incorporates approaches such as keyhole craniotomies and tubular retraction with the goal of maximizing extent of resection and reducing iatrogenic tissue injury. Supramarginal resection builds upon this approach, expanding the boundaries of the resection cavity to ensure removal of microscopic tumor fragments and decrease recurrence. Because MICS is generally performed through craniotomies< 5 cm in diameter with limited ability to change predefined surgical corridors intraoperatively, meticulous attention must be given to the preoperative workup. Imaging modalities, including CT, MRI, DWI, and DTI, may reveal characteristics of the intra-tumoral environment and are important in defining the anatomical relationship of BMs to surrounding functional tissue and neurovascular structures. Intraoperatively, neuronavigation helps maintain alignment within predefined surgical corridors, and adjunctive modalities such as intraoperative ultrasound and brain mapping help compensate for brain shift. Advancements in visual augmentation tools such as fluorescence, endoscopes, and exoscopes further enable intraoperative delineation of tumor boundaries and allow for expanded utilization of MICS in deep-seated, complex BMs. The ever-growing armamentarium of minimally invasive surgical tools has made neurosurgery an increasingly safe and effective option for patients with BMs.

Glioblastoma (GBM) is a devastating primary brain tumor of adults with few treatment options. While there has been long-standing interest in engaging the immune system to combat this disease, monoclonal antibody therapies – including checkpoint modulating agents – have yet to result in widespread clinically meaningful outcomes. By contrast, adoptive cell therapies including chimeric antigen receptor (CAR)-T cell therapies have demonstrated promising early indications of efficacy in select GBM patients. However, heterogeneity of target molecule expression remains a significant barrier to long-term disease control. In a recent report, a first-in-human trial of CARv3-TEAM-E T cells engineered to target the epidermal growth factor receptor variant III tumor-specific antigen (EGFRvIII) as well as wild-type EGFR through secretion of a T-cell-engaging antibody molecule (TEAM) showed preliminary evidence of CAR activity. We now report the results of single cell RNA-sequencing (scRNA seq) of lymphocytes isolated from the cerebrospinal fluid (CSF) of CARv3-TEAM-E treated GBM patients. Longitudinal CSF sampling in these patients was performed via an Ommaya reservoir. Sampled cells were interrogated by scRNA Seq. We demonstrate the ability to detect transduced CARv3-TEAM-E expressing T cells in the CSF from these patients. Unsupervised analysis revealed evidence of the expression of cytotoxic effector gene expression programs, which we explore over the course of longitudinal sampling in this initial cohort. Additionally, we compare expression signatures in the CSF-sampled T cells to those present in the CARv3-TEAM-E infusion products. Taken together, our findings offer insight into the longitudinal dynamics of T cell gene expression programs in the setting of this first-in-human trial of CARv3-TEAM-E T cells in GBM patients. Christopher Mount, Demi Gerovasilis, Sophia Kovatsis, Jun Zhong, Md Raihan Chowdhury, Maxx King, William T. Curry, Elizabeth R. Gerstner, Kathleen M. Gallagher, Bryan D. Choi, Mario Suva, Marcela V. Maus. Single cell RNA sequencing of cerebrospinal fluid lymphocytes in CARv3 TEAM E treated glioblastoma patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr LB358.

Background:The glioma immune repertoire has emerged as a vital point of interest, particularly in the context of immunotherapeutics development and as a key player for prognostic and diagnostic biomarker identification. Methods:Tumor tissue was collected from glioma patients and targeted immune repertoire sequencing of tumor infiltrating lymphocytes (TIL) from each of the four collected glioma tumor subtypes was performed. Gliomas were stratified based on WHO21 classification to map the TCR landscape of astrocytomas (grade II/III, grade IV), glioblastomas, and oligodendrogliomas. Results:Following stratification of TCR repertoires and complete clonotype, V-J cassette, and CDR3 analysis, we identified cohort-specific levels of diversity, clonotype sharing, and conservation. Partitioning of these repertoires based on TCR diversity revealed significant influence on patient survival. Furthermore, mapping of CDR3 binding regions to antigens and their origins highlighted prognostic biomarkers and identified sequences binding to viral signatures associated with patient clinical outcomes. Conclusion:These findings underscore the importance of characterizing TCR repertoires in the context of the patient clinical condition. These unique repertoire signatures and correlated antigens may facilitate patient outcome prognostication and serve as a potential foundation for immunotherapeutic applications.

2008 Background: Chimeric Antigen Receptor (CAR) T cells for glioblastoma (GBM) have been limited by the challenge of targeting a single tumor antigen in a heterogeneous disease. To address this barrier, we generated a novel engineered T-cell product (CARv3-TEAM-E) that targets the EGFRvIII antigen while also secreting T-cell-Engaging Antibody Molecules (TEAMs) against wild-type EGFR. Methods: The INCIPIENT clinical trial is a first-in-human study of CARv3-TEAM-E in patients with recurrent GBM (NCT05660369). Patients were treated with intraventricular CARv3-TEAM-E T cells (10E6 cells per infusion). A subset of patients were conditioned with lymphodepleting chemotherapy (LDC) consisting of cyclophosphamide and fludarabine. Immune cells were profiled in the cerebrospinal fluid (CSF) and peripheral blood of patients by flow cytometry. Results: CAR T cells were detected in the CSF of all patients for an average of 33.6 days ( SD = 10.33). Granulocytes, NK cells, B cells, and monocytes appeared in the CSF immediately after infusion, decreasing to low levels over the course of several weeks. TEAM-positive T cells persisted in CSF until (median) day 33.6 ( SD = 10.8) with a range of 21-56 days. CAR T cells were transiently detected in the peripheral blood of 9/10 patients at an average of 14 days ( SD = 3.5) after infusion. Prior to infusion, CAR T cells were predominantly CD4-positive and remained as such in the CSF over time. Those in the periphery exhibited CD4-to-CD8 polarization. Of patients who received multiple infusions, 3 out of 6 had CAR-positive T cells in the CSF after a second infusion, although their persistence was short-lived and was not detected in the periphery following repeat infusions. LDC increased engraftment of CAR T cells in CSF but not in peripheral blood. Patients with poor CAR persistence demonstrated the development of anti-CARv3-TEAM-E antibodies in the CSF and serum, which increased with reinfusion. Conclusions: Following initial infusion, intraventricularly delivered CARv3-TEAM-E T cells were detected in the CSF and peripheral blood in patients with recurrent GBM. Reduced persistence was observed with subsequent infusions. This corresponded with the emergence of anti-CARv3-TEAM-E antibodies in treated patients. Clinical trial information: NCT05660369 .

Treatment resistance in glioblastoma (GBM) is largely driven by the extensive multi-level heterogeneity that typifies this disease. Despite significant progress toward elucidating GBM's genomic and transcriptional heterogeneity, a critical knowledge gap remains in defining this heterogeneity at the spatial level. To address this, we employed spatial transcriptomics to map the architecture of the GBM ecosystem. This revealed tumor cell states that are jointly defined by gene expression and spatial localization, and multicellular niches whose composition varies along the tumor core-edge axis. Ligand-receptor interaction analysis uncovered a complex network of intercellular communication, including niche- and region-specific interactions. Finally, we found that CD8 positive GZMK positive T cells colocalize with LYVE1 positive CD163 positive myeloid cells in vascular regions, suggesting a potential mechanism for immune evasion. These findings provide novel insights into the GBM tumor microenvironment, highlighting previously unrecognized patterns of spatial organization and intercellular interactions, and novel therapeutic avenues to disrupt tumor-promoting interactions and overcome immune resistance.

Figure S2 | Post-craniotomy stress map for each individual patient involved in this study.