Generation of Allogeneic CAR‐T Circumvents Functional Deficits in Patient‐Derived Autologous Product for Glioblastoma

ABSTRACT

Glioblastoma (GBM) is the most common malignant brain tumor in adults, with a poor prognosis despite aggressive standard of care. Chimeric antigen receptor T‐cell (CAR‐T) therapy has shown promising results in liquid malignancies, but clinical trials in GBM targeting various tumor antigens have not shown durable clinical benefit. While this may be attributable to various tumor‐intrinsic immune evasion strategies characteristic of GBM, little work has been done to assess whether the issue is due to the quality of the CAR‐T treatment itself. Currently, CAR‐Ts for GBMs and liquid malignancies are manufactured in an autologous setting in which T‐cells are extracted from patients, engineered ex vivo, and subsequently reinfused back. However, peripheral T‐cells taken from untreated GBM patients have demonstrated qualitative and functional deficits, which may contribute to suboptimal treatment outcomes. Thus, we aimed to establish whether CAR‐Ts generated from GBM patients would show reduced efficacy in comparison to healthy donors using our previously validated CD133 CAR‐T. In this work, we show pre‐treatment exhaustion and reduced survival advantage in autologous, patient‐derived CD133‐targeting CAR‐T cell products using an orthotopic xenograft model of human GBM. To overcome the functional and logistical considerations of autologous therapy, we additionally aimed to generate an “off‐the‐shelf” allogeneic CD133 CAR‐T. Using CRISPR gene editing technology, we generated TCR‐knockout CAR‐T cells with comparable pre‐clinical efficacy to our autologous models. Ultimately, this work highlights the need to reassess autologous CAR‐T therapy for GBM and consider allogeneic approaches as biologically informed therapeutic alternatives.

What's new?

Clinical trials of chimeric antigen receptor T‐cell (CAR‐T) therapies in glioblastoma have shown limited clinical benefits. Whether this may be explained by the basal quality of CAR‐T products, which are currently generated using patient, autologous T‐cells, has been little explored. This study demonstrates that CAR‐T cells derived from patients with glioblastoma are functionally impaired prior to treatment, limiting their therapeutic efficacy. Using CRISPR gene editing, the authors generated allogeneic CAR‐T cells with promising efficacy in vitro and in vivo. This work highlights the need to reassess autologous CAR‐T therapy for glioblastoma and consider allogeneic approaches as potentially feasible therapeutic alternatives.

Clinical trials of chimeric antigen receptor T‐cell (CAR‐T) therapies in glioblastoma have shown limited clinical benefits. Whether this may be explained by the basal quality of CAR‐T products, which are currently generated using patient, autologous T‐cells, has been little explored. This study demonstrates that CAR‐T cells derived from patients with glioblastoma are functionally impaired prior to treatment, limiting their therapeutic efficacy. Using CRISPR gene editing, the authors generated allogeneic CAR‐T cells with promising efficacy in vitro and in vivo. This work highlights the need to reassess autologous CAR‐T therapy for glioblastoma and consider allogeneic approaches as potentially feasible therapeutic alternatives.

Abbreviations

AAV

adeno‐associated virus

alloCART

allogeneic chimeric antigen receptor T‐cell

B2M

beta‐2‐microglobulin

CARCON

chimeric antigen receptor control

CAR‐T

chimeric antigen receptor T‐cell

CD

cluster of differentiation

CM

central memory

CRISPR

clustered regularly interspaced short palindromic repeats

DN

double negative

DP

double positive

Eff

effector

EGFR

epidermal growth factor receptor

EM

effector memory

FACS

fluorescence‐activated cell sorting

FDA

Food and Drug Administration

GBM

glioblastoma

GvHD

graft versus host disease

HDR

homology directed repair

HEK

human embryonic kidney

HER2

human epidermal growth factor receptor 2

HLA

human leukocyte antigen

IFN

interferon

IL

interleukin

IVIS

in vivo imaging software

KLGR1

killer cell lectin‐like receptor G1

KO

knockout

LAG‐3

lymphocyte activation gene 3 protein

MOI

multiplicity of infection

MRI

magnetic resonance imaging

NSG

NOD scid gamma mouse

PCR

polymerase chain reaction

PD‐1

programmed cell death protein 1

rGBM

recurrent glioblastoma

rH

recombinant human

RLU

relative luminescence units

RNP

ribonucleoprotein

S1P1

sphingosine‐1‐phosphate receptor 1

TCR

T‐cell receptor

TIM‐3

lymphocyte activation gene 3 protein

TMZ

temozolomide

TNF

tumor necrosis factor

TRAC

T cell receptor alpha constant

1. Introduction

Glioblastoma (GBM) is the most common malignant primary brain tumor in adults [1]. Despite an aggressive standard of care that includes maximally safe surgical resection, chemo‐ and radiotherapy, median overall survival remains stagnant at 15 months [2]. To further complicate the clinical course, almost all GBMs recur, with no standard therapeutic approach to recurrent or progressive disease [3]. Poor outcome and limited treatment options have thus necessitated the exploration of novel therapeutic approaches in GBM. Since the advent of the chemotherapeutic agent temozolomide (TMZ) in 2005, few drugs have met primary endpoints in GBM in clinical trials [4]. However, immunotherapeutic strategies have provided an exciting avenue of exploration to meet clinical needs. In looking at other malignancies, liquid cancers have seen significant success with chimeric antigen receptor T‐cells (CAR‐Ts). In this strategy, human T‐cells are engineered to express a synthetic receptor that can recognize an antigen and induce killing on the antigen‐expressing tumor cell [5]. Since the earliest clinical application of the CD19‐targeting CAR‐T for follicular lymphoma in 2010, six other CAR‐T products have been approved by the Food and Drug Administration (FDA) as of current to treat an array of hematologic malignancies [6, 7]. As a result of its success in liquid cancers, the exploration of CAR‐T therapy in GBM has exploded. While early clinical trials showed the feasibility and safety of CAR‐T therapy, clinical benefits have been limited [8]. Biologically relevant targets including human epidermal growth factor receptor 2 (HER2) and epidermal growth factor receptor vIII (EGFRvIII) have shown no durable or consistent response in clinical trials for GBM (rGBM) [9, 10, 11, 12, 13]. Other promising targets such as interleukin 13 receptor alpha‐2 (IL13‐Rα2) have only shown transient, non‐durable radiographic responses despite attempts at locoregional delivery or co‐targeting strategies with EGFRvIII [14, 15, 16]. While a plethora of clinical trials are active or recruiting patients for assessment of other targets in GBM [6], the lack of any success emphasizes the importance of investigating the role of autologous, or patient‐derived, T cell immune biology in GBM.

Pre‐clinical works have defined tumor‐mediated immunosuppression as a strong marker of poor immune cell anti‐tumor efficacy [17]. While these effects are known locally, this immunosuppression has additionally been observed peripherally in circulation. Peripheral T‐cell pools show qualitative and quantitative deficits despite confinement of a GBM tumor to the intracranial compartment [18, 19, 20, 21]. As current CAR‐T therapies are sourced from these autologous T‐cells, investigation on the basal quality of these products should be explored. Current assessment of autologous CAR‐Ts in comparison to allogeneic, or healthy donor‐derived, CAR‐Ts has been limited to hematologic malignancies. Interestingly, comparison of autologous and allogeneic anti‐CD7 CAR‐Ts in patients with T‐cell acute lymphoblastic leukemia or lymphoma showed that patients who received allogeneic CAR‐T treatments had a higher remission rate, less recurrence, and more durable CAR‐T survival than those receiving autologous products [22]. While these findings were post‐treatment, it may ultimately be beneficial to assess the efficacy of a patient‐derived CAR‐T in a pre‐clinical autologous setting, a priori to identify predictive value for efficacy in the clinic. Though modifying autologous CARs may be a difficult approach, the above concerns can be circumvented by use of donor‐derived or allogeneic CAR‐Ts. Autologous CAR‐Ts host a myriad of concerns including inconsistency between infusion products of patients and the time to generation for patients who need immediate care. Other questions pertain to the dubious quality of patient‐derived T‐cells after previous lines of myelosuppressive treatment such as chemoradiotherapy. Allogeneic CAR‐Ts thus present a readily available, “off‐the‐shelf” therapy. However, allogeneic products are not innately benign and can cause a potentially life‐threatening toxicity known as “graft‐versus‐host‐disease” (GvHD), in which the donor product attacks foreign, healthy recipient tissue. While many strategies are used to blunt this response, gene editing to knock out the T‐cell receptor (TCR) has been successfully utilized in new allogeneic, T‐cell therapies [23]. While some progress is being made for other solid tumors such as colorectal cancers (NCT03692429), allogeneic CAR‐Ts for use in brain cancer, such as GBM remain in the early preclinical stages and against previously characterized antigens with poor clinical response such as EGFRvIII and IL13‐Rα2 [24, 25]. Thus, in this work, we assess the quality and function of CAR‐Ts generated from patient‐derived blood products using our previously described anti‐CD133 CAR in our pre‐clinical models of GBM [26]. As a tumor‐associated antigen, CD133 represents a functional unit in tumor initiation, maintenance, and treatment resistance, ultimately correlating with poor clinical outcomes. Its expression not only serves as a prognostic marker but also highlights its potential as a therapeutic target for improving treatment strategies [27]. Thus, to address the clinical and functional difficulties of an autologous approach, we additionally generate and evaluate a novel allogeneic anti‐CD133 CAR‐T with comparable anti‐tumor efficacy.

2. Methods and Materials

2.1. Dissociation and Culturing of Primary GBM Tissue

Brain tumor samples are dissociated in PBS containing 0.2 Wünsch unit/mL Liberase Blendzyme 3 (Roche), and incubated in a shaker at 37°C for 15 min. The dissociated tissue is filtered through a 70 μm cell strainer and collected by centrifugation (1200 rpm, 3 min). Red blood cells are lysed using ammonium chloride solution (STEMCELL Technologies). GBM cells are resuspended in Neurocult complete (NCC) media, a chemically defined serum‐free neural stem cell medium (STEMCELL Technologies), supplemented with human recombinant epidermal growth factor (20 ng/mL: STEMCELL Technologies), basic fibroblast growth factor (10 ng/mL; STEMCELL Technologies), heparin (2 μg/mL 0.2% Heparin Sodium Salt in PBS; STEMCELL technologies), antibiotic‐antimycotic (1X; Wisent), and plated on ultra‐low attachment plates (Corning) and cultured as neurospheres. All experiments were performed with mycoplasma‐free cells. All human cell lines have been authenticated using STR profiling within the last 3 years.

2.2. Patient Blood Collection

Recurrent GBM patients who had not received additional medical therapy or surgical intervention for their recurrent disease were identified and consented at the Hamilton Health Sciences for peripheral blood mononuclear cell (PBMC) collection using SepMate (STEMCELL technologies). Healthy donors who were age and/or sex matched were identified with STEMCELL Technology PBMC products (HD_T1, HD_T2) or internally at the Hamilton Health Sciences (HD_T1) with no comorbidities and no history of immunologic and neoplastic disease.

2.3. Fluorescence‐Activated Cell Sorting Analysis

CAR‐T products were characterized by multiparameter flow cytometry using CytoFLEX or MoFlo flow cytometer (Beckman Coulter). Briefly, 0.5 × 10⁶ cells per sample were washed and incubated with fluorochrome‐conjugated antibodies for 15 min at room temperature. Cells were washed and stained with 7AAD viability dye (1:10; Beckman Coulter, A07704) and read on the flow cytometer for 10,000 live events. Compensation was performed using mouse IgG CompBeads (BD Biosciences, 552,843).

2.4. Gating Strategy and Data Analysis

Data were analyzed using FlowJo (version 10.10.0). Lymphocytes were gated by forward/side scatter and dublets were excluded and later dead cells were excluded based on viability dye. Protein expression was defined as positive or negative based on the analysis of regions established by an isotype control.

2.5. T‐Cell Expansion

Human peripheral blood mononuclear cells (PBMCs) were acquired from the above sources. Upon immunomagnetic enrichment of CD3⁺ cells using the SepMate Human T Cell Isolation Kit, isolated T‐cells were seeded into a 24‐well dish in XSFM T‐cell media supplemented with recombinant human (rh) IL‐2 (100 IU/mL), rhIL‐15 (5 ng/mL) at 1.2 × 10⁶ per well. Cells were activated with 10 μL of human T‐cell TransAct in 900 μL of supplemented XSFM T‐Cell media and expanded for 14 days prior to experimentation, as described below. Transduction efficiency of CAR‐T cells was assessed on day 8 by flow cytometry using Anti‐Human EGFR‐AF488 (R&D Systems, Cat# FAB9577G, Clone Hu1).

2.6. Generation of Lentiviral‐Transduced “Autologous” CAR‐Ts (CART133)

Twenty‐four hours after activation, T‐cells were transduced with a CAR‐lentivirus at a multiplicity of infection (MOI) of 2. CAR‐T cell cultures were then expanded using fresh, supplemented XSFM media for 7 days prior to FACS characterization. Transduction efficiency was confirmed by expression of the CAR product tags, EGFR in CART133‐transduced samples, and NGFR in CARCON samples. This was followed by an additional 7 days of expansion prior to experimentation for a total of 14 days of T‐cell expansion.

2.7. Immunophenotyping

T cells were stained with PE‐Cy7 Mouse Anti‐Human CD3 (BD Biosciences, Cat#563423, Clone UCHT1), PE Mouse Anti‐Human CD4 (BD Pharmingen, Cat#555347, Clone RPA‐T4), APC/Cyanine7 anti‐human CD8α (BioLegend, Cat#300925, Clone HIT8a), Alexa Fluor 700 anti‐human CD45RA (BioLegend, Cat# 304119, Clone HI100), and FITC anti‐human CD197 (CCR7) (BioLegend, Cat# 353215, Clone G043H7) for 15 min at room temperature. T cells were gated as CD3⁺ and later classified by CD4 and CD8 expression: CD4⁺CD8⁻ (CD4 single positive), CD4⁻CD8⁺ (CD8 single positive), CD4⁺CD8⁺ (double positive, DP), CD4⁻CD8⁻ (double negative, DN). Within the CD4⁺ and CD8⁺ compartments, different subsets were defined using CD45RA and CCR7 expression: effector (CD45RA⁺CCR7⁻), naive (CD45RA⁺CCR7⁺), central memory (CD45RA⁻CCR7⁺), and effector memory (CD45RA⁻ CCR7⁻).

2.8. Construction of Adeno‐Associated Virus (AAV) Vector

CART133‐HDR repair constructs targeting the TRAC locus were generated by PCR amplification and ligation of the key fragments. First, the TRAC homologous arms were isolated from Jurkat cells (RRID: CVCL_0065; ATCC TIB‐152) with PCR amplification using locus‐specific primer sets with overhangs containing restriction enzyme sites. The TRAC left homologous arm was PCR amplified by the primer set “TRAC LF NotI EcoRI” and “TRAC LR Esp3I”. The TRAC right homologous arm was PCR amplified by the primer set “TRAC RF AgeI” and “TRAC RR NotI”. The CD133‐CAR‐T2A‐EGFR fragment was isolated from the vector containing CD133 CAR [24] using primer sets “Human SecSig F NcoI” and “pMEM201 CytoTail R AgeI.” In brief, the CD133‐CAR consists of a CD133‐binding single‐chain variable fragment, RW03, specific for human CD133, followed by CD8 hinge‐transmembrane regions linked to CD28 intracellular domains and CD3ζ intracellular domain. The CD133‐CAR fragment is followed by the extracellular domain of EGFR separated by a T2A self‐cleavage peptide, which was used as a marker gene to monitor the expression of CD133‐CAR in cells. To generate the T2A self‐cleavage peptide between TRAC left homologous arm and the CD133‐CAR fragment, oligonucleotides “T2A Sense” and “T2A Antisense” were phosphorylated, then annealed to generate the T2A fragment flanked by sticky ends compatible with the Esp3I cut site on TRAC left homologous arm and NcoI cut site on CD133‐CAR. The PCR product of the TRAC left homologous arm was digested with Esp3I. The PCR product of CD133 CAR was digested with NcoI and AgeI. The PCR product of the TRAC right homologous arm was digested with AgeI. These three digested PCR fragments and the annealed T2A fragment were ligated with T4 DNA ligase (NEB). The ligated product was isolated through gel‐purification and further amplified with the primer set “TRAC LF EcoRI NotI” and “TRAC RR NotI” to generate the CART133‐HDR repair construct. The AAV2 transgene backbone was generated by digesting pAAV‐EGFP (Cell Biolabs VPK406) with NotI (NEB) in the presence of shrimp alkaline phosphatase (NEB M0371) and purified through gel purification. CART133‐HDR repair construct was digested with NotI and ligated with the AAV2 transgene backbone to generate the AAV transgene vector pAAV‐CART133.

2.9. AAV Production and Titration

AAV was produced by transfecting HEK293T cells (RRID: CVCL_0063; ATCC CRL‐3216) in 15‐cm tissue culture dishes (Corning 430599) cultured in Dulbecco's modified essential medium (DMEM, Wisent 319‐062‐CL) supplemented with 10% fetal bovine serum (Gibco 12483‐020). Cells were seeded at a density of 10⁷ per plate overnight. The next day, cells were transfected with a mix of AAV2 transgene plasmid, packaging (pDF6) plasmid, and AAV6/9 serotype plasmid (Cell Biolab) at a molar ratio of 1:2:2, using Roche X‐tremeGENE9 transfection reagent at 1:3 ratio DNA:reagent (Millipore Sigma 6365779001). Each 15 cm dish was transfected with 16 μg total plasmid. Seventy‐two hours after transfection, cells were mechanically dislodged from each 15 cm dish using a cell scraper and collected in media. To purify virus from harvested cells, cells were first mixed with pure chloroform (1:10 volume), incubated at 37°C with shaking for 1 h and then treated with 1 M NaCl. The mixture was then centrifuged at 4000 g at 4°C for 1 h. The aqueous layer was collected, while the chloroform layer and interlayer precipitate were discarded. 10% Polyethylene glycol (PEG) 8000 was added to the aqueous layer and shaken until dissolved. The mixture was then incubated at 4°C for 1 h and centrifuged at 4000 g at 4°C for 1 h. The supernatant was discarded. For each 15 cm dish used for AAV production, the pellet was suspended with 0.5 mL AAV lysis buffer (50 mM Tris pH 8.0, 150 mM NaCl, 2 mM MgCl2) and 0.5 U benzonase. The dissolved pellet was incubated at 37°C for 30 min, then mixed with equal volumes of chloroform. The mixture was centrifuged at 4000 g for 1 h at room temperature. The chloroform layer and the interlayer precipitate were discarded, and the aqueous layer was loaded on top of a discontinuous iodixanol gradient consisting of a 60%, 40%, 25% and 15% layer diluted with DPBS. The gradient was centrifuged at 40,000 rpm in the Beckman SW41Ti swing bucket rotor for 6 h at 18°C. After centrifugation, the top half of the 60% iodixanol and the bottom half of the 40% iodixanol layer was collected with a needle and syringe. The collected iodixanol was passed through a 0.45 μm filter and concentrated through a 100‐kDa molecular‐weight cutoff filter ultrafiltration filter (Millipore UFC5100). Virus was titered by quantitative PCR using primers targeted to the AAV inverted terminal repeat.

2.10. Generation of Allogeneic CART133 (alloCART133)

Knockouts were performed by electroporation of complexed Alt‐R As Cas12a (Cpf1) (IDT 10007922) with the TRAC‐targeting guide at a 3:1 guide to protein molar ratio for 15 min at room temperature. Using Lonza Amaxa System Kits, 3 M T‐cells were resuspended in the provided electroporation buffer and gently mixed with complexed ribonucleoprotein (RNP) (Lonza VPA‐1002). The mixture was removed to an electroporation cuvette where the T‐023 electroporation program (Nucleofector IIb, Lonza Biosciences) was used. For sole assessment of knock‐out, cells were gently resuspended in their respective media and incubated in a 12‐well dish for 4 days prior to flow cytometric characterization. For knock‐in, cells were plated at a density of 200,000 cells in 50 μL of respective media with AAV at an MOI of 1E6 genome copies per cell. Cells were then cultured for 2–7 days prior to flow cytometric characterization, and for a total of 14 days of T cell expansion prior to experimentation.

2.11. Activation and Exhaustion Assay

CAR‐T cells were co‐incubated with GBM cells at an effector: target (E:T) ratio of 1:1 for 24 h to assess activation. Cells were analyzed for expression of activation markers Mouse Anti‐Human CD25‐PE (BD Biosciences, Cat#555432, Clone M‐A251) and Mouse Anti‐Human CD69‐APC (BD Biosciences, Cat#555533, Clone FN50), as well as for exhaustion markers PE Mouse Anti‐Human LAG‐3 (CD223) (BD Biosciences, Cat# 565616, Clone TA7‐530), APC Anti‐Human CD366 (Tim‐3) (BioLegend, Cat#364804, Clone A18087E), and Brilliant Violet 421 Anti‐Human CD279 (PD‐1) (BioLegend, Cat#329919, Clone EH12.2H7).

2.12. Cytokine Release Assay

CAR‐T cells (CARCON, CART133, TRAC‐KO or AlloCART133) were co‐incubated with GBM cells (GBM8 and BT935) at a 1:1 ratio for 24 h. Supernatants were collected in duplicate for each condition and stored at −80°C for analysis of cytokines. Human TNF‐α DuoSet ELISA kit (R & D Systems, DY210‐05) and IFN‐γ DuoSet ELISA kit (R & D Systems, DY285B‐05) were used for quantification of the two cytokines by ELISA, according to the manufacturer's recommendation.

2.13. Luciferase‐Based Cytotoxicity Assay

Luciferase‐expressing GBM cells were plated in triplicates at a concentration of 3.0 × 10⁴ cells per well. Effector cells were then added at the E:T ratios 4:1, 2:1, 1:1 and 0:1 in a final volume of 200 μL of XSFM media and 75 μg/mL of D‐luciferin potassium salt in PBS. Bioluminescence intensity (BLI) was measured as relative luminescence units (RLU) by an Omega luminometer for 10 s. Cells were treated with 1% Nonidet P‐40 (NP40) lysis buffer (Thermofisher J60766‐AP) as a measure of maximal lysis. Target cells incubated without effector cells were used to measure spontaneous death RLU. The readings from triplicates were averaged and percent lysis was calculated with the following equation:

$$\begin{array}{c}\%\text{Specific lysis}=100\mathrm{X}\left(\text{spontaneous death}\mathrm{RLU}–\text{test}\mathrm{RLU}\right)/\hfill \\ \left(\text{spontaneous death}\mathrm{RLU}–\text{maximal killing}\mathrm{RLU}\right).\hfill \end{array}$$

2.14. In Vivo Orthotopic Injections of Xenograft Tumors

Animal studies were performed according to guidelines under Animal Use Protocols of McMaster University Central Animal Facility (AUP #22‐12‐38). Intracranial injections were performed as previously described using 0.2−1 × 10⁶ cells of primary, CD133‐expressing GBMs expressing luciferase in PBS in 8–12‐week‐old NSG mice (Chokshi, Savage, Venugopal, and Singh [28]. As previously reported and utilized in [26]), we utilized GBM8, and BT935, two‐patient derived GBM cell lines with over 90% surface expression of CD133 [26]. Animals were subsequently randomized to treatment cohorts a priori of at least triplicates for sex‐balancing, as well as by mean tumor volume captured by in vivo bioluminescence imaging (IVIS) once engraftment was confirmed. All animals were included in the study post‐engraftment. CAR‐T treatment, was initiated upon confirmation of engraftment [26]. Briefly, 1 × 10⁶ Control or treatment CAR‐Ts, suspended in 10 μL of PBS, were injected into the previously generated burr hole and thus intracranially, a week a part for a total of 2 × 10⁶ cells. In the patient‐derived in vivo cohort, a single dose of 0.5 × 10⁶ control or treatment cells were treated to allow for assessment of therapeutic potency at sub‐optimal dosing. For survival studies, mice were monitored daily and euthanized upon reaching humane endpoints, defined as ≥ 20% body weight loss, ruffled fur, hunched posture, or other neurological signs, in accordance with protocol AUP 22‐12‐38. Number of days of survival were noted for Kaplan–Meier analysis. The treatment allocations were blinded to the author who confirmed endpoint.

2.15. In Vivo Bioluminescence Imaging (IVIS)

Tumor volume was assessed using IVIS by sedating mice with isoflurane gas. Mice were subsequently administered D‐luciferin, potassium salt substrate (Thermofisher L2916) at a concentration of 150 mg/kg in PBS subcutaneously. After a 10‐min incubation, intracranial bioluminescent signal (photons per second; p/s) was quantified using Living Image software. Imaging and quantification were performed by a blinded author.

2.16. MRI Imaging

Magnetic resonance imaging (MRI) images were captured using the Bruker PharmaScan 70/16 US MRI system using the magnetization transfer FLASH routine with the following details: echo time = 3 ms, TR = 23 ms, flip angle = 5 degrees, magnetic transfer pulse = 12.00 ms, bandwidth = 228.3, flip angle = 523.0.

2.17. Statistical Analysis

Respective data is represented as mean ± SEM, n values are listed in figure legends for technical replication. Welch's t‐test analyses were performed using GraphPad Prism 5. p < 0.05 is considered statistically significant.

3. Results

3.1. Patient‐Derived Autologous CAR‐Ts Reveal Functional Deficits and Poor Anti‐Tumor Efficacy Compared to Healthy Donor‐Derived CAR‐Ts

To assess the functionality of patient‐derived anti‐CD133 CAR‐Ts (CART133), we extracted peripheral blood mononuclear cells (PBMCs) from recurrent GBM patients not receiving active cytotoxic therapy. The schematic representation of the CART133 generation protocol from patient PBMCs is shown in Figure 1A. Age and sex‐matched healthy donor‐derived CART133 were simultaneously generated to compare pre‐clinical efficacy (Table S1). To identify early putative differences in the generated products, we performed immunophenotyping after 14 days of expansion and prior to antigen exposure. Characterization of T‐cell subset revealed similar CD8, CD4, double positive (DP), and double negative (DN) distributions between healthy donor (black) and patient‐derived CAR products (red), with a skew toward cytotoxic CD8⁺ populations (Figure 1B). Similarly, assessment of both CD4⁺ and CD8⁺ T‐cell subtype revealed few distinctions between groups in subtype distribution. In CD4⁺ populations, preponderance for an effector memory (EM) subtype was observed in both patient‐derived and healthy donor‐derived CAR products (Figure 1C). In CD8⁺ populations, both healthy‐donor and patient‐derived CAR products showed preponderance for EM and effector (Eff) subtypes, with minor skew toward naïve populations (Figure 1D). These phenotypes were additionally consistent between each sample CART133 and a respective control CAR (CAR CON). The similarity in CART133 and CAR CON subtype suggested that any observed changes were unlikely artifacts of CAR generation (Figure S1). Together, early phenotypic characterization failed to identify cell population differences between healthy‐donor and patient‐derived CAR products.

FIGURE 1.

Isolation, production, and characterization of patient‐derived and healthy donor‐derived CD133 CAR T cells. (A) Schematic illustrating the workflow for peripheral blood mononuclear cell (PBMC) isolation and chimeric antigen receptor (CAR) T cell production from recurrent glioblastoma (rGBM) patients and healthy donors of matched age and sex. (B–D) Stacked plots showing the distribution of T‐cell subsets and subtypes following CD133 CAR‐T (CART133) production. (B) represents the overall CART133 T‐cell population without antigen exposure (Multiple T‐tests Healthy donor vs. patient‐derived CD4 p = 0.86; CD8 p = 0.99; DP p = 0.19; DN p = 0.32, n = 3, biological and technical replicates). (C) CD4⁺ (Multiple t‐tests Healthy donor vs. patient‐derived CD4 Naïve p = 0.02; CM p = 0.32; EM p = 0.02; Eff p = 0.12, n = 3, biological replicates) and (D) CD8⁺ (Multiple T‐tests Healthy donor vs. patient‐derived CD8 Naïve p = 0.27; CM p = 0.43; EM p = 0.32; Eff p = 0.24, n = 3, biological replicates) subsets were additionally assessed for subtype from rGBM patients and healthy donors without exposure to antigen. Immunophenotyping was conducted via flow cytometry (FACS) on day 14 post‐transduction. CM, central memory; DN, double negative; DP, double positive; Eff, effector; EM, effector memory. GBMT1‐3, GBM patient‐derived T cells; HDT1‐3, healthy donor‐derived T cells. (E) Comparative analysis of cumulative triple‐positive exhaustion profiles between donor‐derived and patient‐derived CART133 and CARCON cells. Pooled Unpaired student's T test, n = 3, biological replicates. (F, G) Comparative analysis of cumulative CD25 (F) and CD69 (G) expression between donor‐derived and patient‐derived CART133 and CARCON cells. Pooled Unpaired Student's t‐test, n = 3, biological replicates.

We next sought to assess the immune exhaustion profile of the patient‐derived CAR products prior to antigen exposure. Recognizing that the expression of the immune checkpoint molecule PD‐1 alone may not consistently mark chronic exhaustion, we assessed its co‐expression with other co‐inhibitory molecules, namely LAG‐3 and TIM‐3 [29, 30]. While the triple positive expression was not significantly different between each sample and its corresponding control, collectively, patient‐derived CART133 samples showed significantly higher expression of PD‐1/LAG‐3/TIM‐3 as compared to healthy donor samples (Figure 1E). Notably, this trend was not observed in CARCON products, indicating that this phenomenon was specific to patient‐derived CART133 products (Figure 1E). Furthermore, we found no statistical difference in activation markers between patient‐derived vs. healthy donor derived samples (Figure 1F,G), indicating that the T cells exhaustion in patient‐derived samples was not due to activation or tonic signaling.

We next sought to define whether patient‐derived CAR‐T products would show efficacy for clinically relevant outcomes, namely tumor burden and survival, in an in vivo patient‐derived model of GBM. By using a single patient‐model system (GBM8), we sought to eliminate tumor‐mediated differences to identify functional variabilities in our CAR‐T products. Upon confirmation of engraftment, we treated tumor‐bearing mice with a single, subtherapeutic dose of 500 K CAR⁺ cells to additionally assess tumor‐control potency between patient‐derived and healthy‐donor CAR products. Tracking of tumor burden by in vivo luminescence imaging consistently revealed poor early tumor control in patient‐derived CART133 treated mice (red), in comparison to healthy‐donor CART133 (gray) treated mice (Figure 2A,B). Additionally, the pooled flux values at day 21 showed significant tumor control in healthy donor derived CART133 as compared to its control, whereas no significant difference was observed in patient‐derived samples (Figure 2C). This trend can also be seen in time matched MRI images (Figure 2D). As expected, these differences were later reflected in survival outcomes, with patient‐derived CART133 treated mice exhibiting shorter durations of survival benefit in comparison to healthy‐donor CART133 treated mice (Figure 2E,F). Together, the extension of survival compared to CARCON in patient‐derived CART133 treated mice was significantly less (2.5‐fold, p = 0.0207, Welch's t‐test) than in mice treated with healthy donor CART133 (Figure 2G).

FIGURE 2.

Preclinical evaluation of donor‐Derived and patient‐Derived CART133 cells in an orthotopic, intracranial, xenograft model of GBM. (A and B) Evaluation of tumor growth in an in vivo orthotopic xenograft model using GBM8 patient‐derived cells. Tumors were treated post‐engraftment with 500,000 CART133 or CARCON cells derived from patients (red) and age and sex matched healthy donors (gray) (n = 3, technical replicates). (C) Analysis of combined changes in tumor flux measured using the in vivo imaging system on Day 21, comparing patient‐derived and donor‐derived CART133 and CARCON treatments. Pooled Unpaired Student's t‐test, n = 3, biological replicates. (D) Comparison of tumor burden using MRI images between CART133 and CARCON treatments in both patient‐derived and donor‐derived arms. (E and F) Kaplan–Meier survival curves illustrating the survival advantage of patient‐derived (red) and healthy donor‐derived (gray) CART133 compared to CARCON treatment. Log‐rank test, n = 3, technical replicates. (G) Combined survival differences between patient‐derived and donor‐derived CART133 treatment arms. Pooled Unpaired Student's t‐test, n = 3, biological replicates.

3.2. AlloCART133 Reveals Comparable Activation and GBM Cell‐Directed Lysis In Vitro to CART133

To circumvent the functional deficits and molecular changes associated with autologous CAR‐T therapy, we aimed to generate an allogeneic CART133 product with comparable pre‐clinical efficacy. To do so, we employed a one‐step knock‐in, knock‐out (KiKO) system pioneered by Dai et al. [31], to knock out the TCR gene, as well as insert our CAR under control of the endogenous TCR promoter. We first enriched and expanded healthy donor T‐cells for 3 days and then electroporated ribonucleoprotein (Cpf1) targeting the T‐cell receptor α constant (TRAC) locus of the TCR. We subsequently infected these cells with adeno‐associated virus [24] with the CAR sequence and expanded these cells for an additional 5 days prior to immunomagnetic depletion of TCR‐expressing T‐cells (Figure 3A). CART133 and CARCON were prepared using second generation CAR‐T backbone as depicted in Figure 3A. Confirmation of allogeneic CART133 (alloCART133) generation was assessed by flow cytometry for expression of the CAR tag, a truncated EGFR, and TCR negativity after immunomagnetic purification (Figure 3B). We thus compared the conventional, autologous model, or CART133, to the alloCART133, each with its control of a non‐targeting control CAR‐T (CARCON) and TRAC‐knockout (TRAC‐KO), respectively. We then co‐cultured each CAR product and its control with CD133‐expressing patient‐derived GBM lines GBM8 and BT935 to assess cytotoxicity, as well as antigen‐negative HEK293T cells for specificity. In both GBM8 and BT935, we saw significant dose‐dependent cell lysis of GBM cells when co‐cultured with CART133 and alloCART133 (Figure 3C, left and middle panel). Additionally, alloCART133 showed specificity to CD133 comparable to CART133, when co‐cultured with HEK293T cells (negative control), where no cell lysis was observed (Figure 3C, right panel). Together, this data demonstrated on‐target specificity and robust efficiency of alloCART133 comparable to CART133.

FIGURE 3.

Development and therapeutic assessment of allogeneic CART133 Cells. (A) Schematic representation outlining the overall production flow to generate allogeneic CART133, along with the CAR constructs used in the study (CART133, CARCON, AlloCART133). (B) Flow cytometry (FACS) data demonstrating effective knockout of the TRAC locus and successful integration of the CART133 single‐chain variable fragment (scFv), validating the development of AlloCART133. (C) AlloCART133 and CART133 cells significantly induce cytotoxicity in CD133‐expressing GBM lines BT935 and GBM8 and CD133‐negative HEK293T cells as compared to their controls (TRAC‐KO and CARCON). Two‐way ANOVA; n = 3, technical replicates. (D) Assessment of CAR T cell activation in AlloCART133, CART133, CARCON, and TRAC‐KO cells after 24‐h co‐culture with CD133‐expressing GBM lines BT935 and GBM8 and CD133‐negative HEK293T cells. Unpaired Student's t‐test, n = 3, technical replicates. (E) Assessment of basal CAR T cell activation in AlloCART133, CART133, CARCON, and TRAC‐KO cells. Unpaired Student's t‐test, n = 3, technical replicates. (F and G) Assessment of cytokine release in AlloCART133, CART133, CARCON, and TRAC‐KO cells after 24‐h co‐culture with CD133‐expressing GBM lines GBM8 (F) and BT935 (G). Unpaired Student's t‐test, n = 3, technical replicates.

We next sought to assess activation and cytokine release of alloCART133 to further define its functionality. We thus co‐cultured GBM8 and BT935 with either CARCON, CART133, TRAC‐KO, or alloCART133 and assessed co‐expression of the early activation marker CD69 and CD25 after 24 h with flow cytometry. We saw stark elevation of CD25 and CD69 co‐expression in CART133 in comparison to CARCON and alloCART133 in comparison to TRAC‐KO cells (Figure 3D). To confirm specificity, we additionally performed this assay in HEK293T cells. Interestingly, CART133 showed very slight non‐specific activation in comparison to CARCON while alloCART133 showed no activation in comparison to TRAC‐KO. While cytolysis was comparable between alloCART133 and CART133, activation was significantly higher in CART133. Though this activation did not correspond to cytolytic activity as previously described (Figure 3C) we sought to assess whether this activation was due to a basal characteristic of CART133 or due to an allogeneic response from TCR activity. We thus assessed activation of the CAR products in the absence of co‐culture. CART133 retained its activated state comparable to when cultured with HEK293T cells, indicating basal expression. AlloCART133 retained a non‐activated state with low expression of CD25 and CD69 (Figure 3E). Similar observations were seen in cytokine release assay where we assessed each product's' ability to secrete the pro‐stimulatory cytokine tumor necrosis factor‐α (TNF‐α) and the pro‐inflammatory interferon‐γ (IFN‐γ) in response to antigen stimulation. Supernatants from alloCART133 and CART133 treated co‐cultures had significantly higher levels of TNF‐α and IFN‐γ. However, CART133 showed significantly higher cytokine release in comparison to alloCART133 as anticipated from the activation assays in both GBM8 and BT935 (Figure 3F,G). Though this suggests that CART133 activates more potently than alloCART133, higher basal and non‐specific activation was observed. Additionally, reduced population activation did not translate to lower cytolytic activity and functionality compared to CART133, suggesting higher killing efficiency by alloCART133 on a population level.

3.3. AlloCART133 Safely Inhibits Tumor Growth and Prolongs Survival in Patient‐Derived Xenografts of Human Glioblastoma

To further validate the alloCART133 strategy, we tested its pre‐clinical efficacy in vivo using the patient‐derived, orthotopic model of GBM. Using our previously validated in vivo CAR‐T treatment regimen, we first intracranially injected NSG mice with either firefly luciferase‐expressing GBM8 or BT935 and assessed engraftment by bioluminescence imaging. Upon confirmation of half‐maximal engraftment, mice were randomized into cohorts to receive either CARCON, CART133, TRAC‐KO, or alloCART133. Mice were then intracranially treated with a dose of 2 × 10⁶ CAR‐T cells over a two‐week period. We saw significant tumor burden control as quantified and demonstrated by IVIS imaging of CART133 in comparison to CARCON, and alloCART133 to TRAC‐KO treated mice, and a concomitant extension in survival in BT935 (Figure 4A,B) and GBM8 (Figure 4C–E). There was an extension of the median OS of 10 days for alloCART133 vs. TRAC‐KO and 16 days for CART133 versus CARCON in GBM8 (Figure 4D), whereas in BT935 there was an extension of 14 days in median OS for both alloCART133 versus TRAC‐KO and CART133 versus CARCON (Figure 4B).

FIGURE 4.

In vivo validation of allogeneic CART133 cells in a patient‐derived orthotopic animal model. (A and C) Monitoring of (A) BT935 and (C) GBM8 tumor burden as measured by total flux using the in vivo imaging system post treatment with AlloCART133, CART133, CARCON, TRAC‐KO with 2 doses of 1 × 10⁶ CAR T cells over a two‐week period at half‐maximal engraftment. Unpaired Student's t‐test, n = 3, technical replicates. (B) Kaplan–Meier survival curves illustrate the extension of overall survival in treatment arms (AlloCART133, CART133) as compared to the control arms (CARCON and TRAC‐KO) in BT935. Log rank test, **p = 0.0034, Extension of median OS = 14 days for treatment arms (AlloCART133, CART133) as compared to the control arms (CARCON and TRAC‐KO). Log‐rank test, n = 3, technical replicates. (D) Kaplan–Meier survival curves illustrate the extension of overall survival in treatment arms (AlloCART133, CART133) as compared to the control arms (CARCON and TRACKO) in GBM8. (Log rank test, ****p < 0.0001, Extension of median OS = 10 days for AlloCART133 vs. TRAC‐KO, 16 days for CART133 vs. CARCON). Log‐rank test, n = 3, technical replicates. (E) In vivo images depicting overall tumor burden in the GBM8 cohort at various time points across all treatment arms.

4. Discussion

Cellular engineering technology such as CAR‐T therapy has revolutionized therapeutic possibilities in oncological practice. In hematologic malignancies, the success of CAR‐Ts directed against CD19 and other clinically‐relevant antigens has provided a developmental framework to expand into solid tumors such as GBM. However, early clinical trials of CAR‐T therapy have shown limited clinical benefit against biologically relevant targets. Tumor‐mediated immunosuppression from the tumor microenvironment (TME) has long presented as a significant obstacle to the effectiveness of CAR‐T therapy in GBM. Cellular components of the TME, including myeloid‐derived suppressor cells, macrophages, and innate regulatory T‐cells, acquire robust immunosuppressive phenotypes and release immunosuppressive molecules to abrogate T‐cell activity [32, 33]. While local immunosuppression has been widely characterized, a small body of literature has shown that GBMs induce peripheral immunosuppression. Quantitatively, GBM patients show peripheral lymphopenia due to T‐cell sequestration in the bone marrow and T‐cell deficient lymphoid organs. This is mediated by tumor‐imposed loss of sphingosine‐1‐phosphate (S1P1), which governs T‐cell trafficking, from the T‐cell surface [19]. Qualitatively, T‐cells from GBM patients show anergy, senescence and exhaustion [20]. In particular, peripheral T‐cells have previously been observed to show expression of exhaustion/senescence markers PD‐1, KLGR1 and CD57, as well as functional deficits marked by T‐cell hyporesponsiveness irrespective of treatment with the immunosuppressive corticosteroid dexamethasone [34, 35]. Even more so, the use of radiation, TMZ, and glucocorticoids for treatment has been shown to induce significant amounts of immunosuppression in patients with GBM [34, 36]. Pre‐clinical evaluation of CAR‐Ts, which are generated from healthy donors, would thus not adequately predict the efficacy of patient‐derived CAR‐Ts used in clinical trials [37]. Though previous works have observed functional deficits of autologous CAR‐Ts in GBM in an in vitro setting [38], we sought to compare the pre‐clinical efficacy of autologous, patient‐derived CAR‐Ts taken from recurrent GBM patients using our previously validated, and functionally relevant anti‐CD133 CAR‐T targeting strategy for GBM in vivo as a model system [26]. We first sought to identify differences in the immunophenotype of patient‐derived versus healthy donor specimens. Though we saw a general skew toward CD8, which has previously been reported in patient‐derived GBM CAR‐T specimens, we did not find a significant difference in comparison to our cohort of healthy‐donor specimens [39]. Interestingly, we did see a significant difference in CD4 naïve and effector memory populations in healthy donor specimens, in which these early memory subtypes show superior and synergistic antitumor activity in pre‐clinical models for hematologic malignancies [40]. We additionally observed functional differences in which autologous, patient‐derived CAR‐T products had poorer tumor control and reduced survival benefit in our orthotopic models. We believe these findings suggest the need to consider the innate, and permanent biological dysfunction of peripheral T‐cells taken from the apheresis products of GBM patients used for immunotherapies. Despite novel engineering strategies in CAR‐T development such as inducing resistance to exhaustion and dysfunction, using a higher quality base T‐cell product such as donor T‐cells may be more clinically tractable [41, 42]. To circumvent these concerns, the use of allogeneic, or donor‐derived CAR‐Ts presents a readily available, functionally fit product [23, 25, 43]. Considering the functional implications of targeting CD133, we proceeded to generate and compare an allogeneic version of our anti‐CD133 CAR‐T, or alloCART133 using a CRISPR‐mediated “knock‐in, knock‐out” approach. Using a similarly described approach, we knocked‐in the CART133 sequence into the TRAC locus to simultaneously knock‐out the TCR, with expression and gene regulation under the control of the native TCR promoter [44]. In vitro alloCART133 showed specific activation, cytokine release and lysis when cultured with patient‐derived, CD133‐expressing GBMs. Interestingly, alloCART133 additionally showed reduced basal activation in comparison to the conventional CART133, suggesting reduced propensity for tonic signaling and early exhaustion, which have previously been shown to limit antitumor activity [45]. In alloCART133, this mechanism may be secondary to TCR knockout leading to reduced alloreactivity or weak interactions with self‐peptide/MHC complexes. Reduced tonic signaling in alloCART133 may also be multifactorial and related to endogenous control of the CAR. Optimizing CAR placement in the TRAC locus supports stable, low‐tonic baseline expression and balanced post‐antigen recovery, which together limit differentiation and exhaustion and ultimately enhance tumor clearance compared with high‐expression CAR T cells [44]. In vivo, alloCART133 inhibited tumor growth and prolonged survival in patient‐derived xenografts of human glioblastoma, demonstrating significant efficacy comparable to treatment with our autologous CAR‐Ts. As alloCART133 and CART133 were generated from healthy donor T‐cells in the allogeneic vs. autologous part of the study, we did not observe significant differences in survival and tumor burden control. However, alloCART133 presents as an immunologically tolerable, healthy donor product for use in GBM patients with potent preclinical efficacy.

In summary, while CAR‐T therapy has revolutionized oncological practice, its efficacy against solid tumors such as GBM has been limited by possible tumor‐mediated immunosuppression within both the local and peripheral microenvironments. Our investigation into patient‐derived CAR‐Ts highlighted inherent functional and molecular differences compared to healthy donor‐derived counterparts, emphasizing the need to explore these variabilities in larger cohorts, and to develop strategies to address the biological dysfunction present in peripheral T‐cells of GBM patients. Herein we develop an allogeneic anti‐CD133 CAR‐T—alloCART133—which demonstrates promising efficacy in vitro and in vivo, offering a potential solution to the challenges posed by autologous CAR‐T therapy. While this feasibility study presents allogeneic products as plausible therapeutic applications, further engineering may need to be considered, as well as preclinical comparisons directly comparing allogeneic products with patient‐derived CAR‐Ts. Allogeneic products must also be hypoimmunogenic and avoid host recognition and elimination for potent efficacy. As preclinical animal models such as the NSG mouse model used in this study are immunodeficient, immunogenicity of TRAC knockout cells cannot thoroughly be investigated. Thus, genetic inactivation of HLA Class‐I surface markers such as beta‐2‐microglobulin (B2M) needs to be considered.

Moving forward, novel engineering strategies and the use of allogeneic CAR‐Ts hold great promise in overcoming immunosuppressive barriers and advancing the therapeutic landscape for patients with GBM.

Author Contributions

Sabra K. Salim: conceptualization, methodology, validation, data curation, investigation, formal analysis, visualization, writing – original draft, writing – review and editing. Muhammad Vaseem Shaikh: conceptualization, methodology, data curation, investigation, validation, formal analysis, visualization, writing – original draft, writing – review and editing. Jeffrey Wei: methodology, validation, writing – review and editing, formal analysis, data curation, software. William T. Maich: data curation, writing – review and editing. Alisha Anand: data curation, writing – original draft. Oliver Y. Tang: formal analysis, data curation, writing – review and editing, writing – original draft, software. Minomi K. Subapanditha: data curation, visualization, formal analysis, writing – review and editing. Zahra Alizada: data curation. Yujin Suk: data curation. Manoj Singh: data curation, visualization. Kui Zhai: data curation. Aapti Khanna: data curation. Benjamin Brakel: data curation. Vassil Dimitrov: data curation, validation, methodology. Zoya Tabunshchyk: data curation, formal analysis, visualization. Katie Chan: methodology, supervision, project administration, resources. Kevin R. Brown: formal analysis, methodology, validation, resources. Parvez Vora: methodology, supervision, project administration, resources. Donald M. O'Rourke: methodology, resources. Zev A. Binder: methodology, supervision, resources. Chitra Venugopal: conceptualization, methodology, supervision, funding acquisition, project administration, resources, writing – original draft, writing – review and editing. Jason Moffat: conceptualization, methodology, supervision, funding acquisition, project administration, resources, writing – review and editing. Sheila K. Singh: conceptualization, methodology, supervision, funding acquisition, project administration, resources, writing – original draft, writing – review and editing.

Funding

This project was supported by the Genome Applications Partnership Program (GAPP) Grant by Genome Canada, as well as the Brain Tumour Foundation of Canada.

Ethics Statement

This research study was conducted with all relevant ethical regulations according to the Hamilton Health Sciences and McMaster Health Sciences Research Ethics Board. Primary GBM tumors and GBM patient blood were obtained from consenting patients and families (REB #16078). Healthy donor blood was collected from consenting donors (REB #1418). All animal work was conducted in accordance with the McMaster Animal Research Ethics Board (AUP #22–12‐38).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Clinical and Mutational Analysis of Healthy Donor and Patient‐Derived PBMCs.

Figure S1: Characterization of T Cell Subsets and Subtypes in Patient‐Derived and Healthy Donor‐Derived Peripheral Blood Mononuclear Cells (PBMCs). A‐F. Analysis of T cell subtypes in PBMCs obtained from healthy donors (A‐C) and patient derived (D‐F). Double Negative (DN) denotes cells negative for CD4 and CD8 (CD4‐CD8‐CD3+), while Double Positive (DP) represents cells positive for both CD4 and CD8 (CD4 + CD8 + CD3+). G‐L. Subtyping of CD4+ T cells in PBMCs derived from healthy donors (G‐I) and patient derived (J‐L). Subtypes include Central Memory (CM), Effector (Eff), and Effector Memory (EM). M‐R. Characterization of CD8+ T cell subtypes in PBMCs from healthy donors (M‐O) and patients (P‐R). Subtypes include Central Memory (CM), Effector (Eff), and Effector Memory (EM).

Salim S. K., Shaikh M. V., Wei J., et al., “Generation of Allogeneic CAR‐T Circumvents Functional Deficits in Patient‐Derived Autologous Product for Glioblastoma,” International Journal of Cancer 159, no. 6 (2026): 1521–1534, 10.1002/ijc.70509.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.