CRISPR-Edited ‘Off-the-Shelf’ CAR-T Cells Show Early Promise Against Recurrent Glioblastoma in Small China Trial

Neurosurgeons in eastern China have tested a radically engineered cell therapy in one of medicine’s deadliest cancers, reporting that CRISPR-edited “off-the-shelf” immune cells shrank recurrent brain tumors when injected into patients’ spinal fluid.

In a phase 1 trial published Jan. 6 in Nature Communications, five adults with recurrent high-grade glioma—mostly glioblastoma—received repeated doses of donor-derived CAR-T cells altered with CRISPR gene editing and delivered by lumbar puncture into the cerebrospinal fluid.

The experimental treatment, known as MT026, was primarily tested for safety. Investigators at the Fourth Affiliated Hospital of Soochow University in Suzhou reported no severe treatment-related side effects and early signs that the tumors were responding, including one complete response and three partial responses on MRI scans.

A new kind of CAR-T for the brain

Chimeric antigen receptor T-cell therapies, or CAR-T, reprogram a patient’s T cells to recognize and kill cancer cells. They have transformed outcomes in certain blood cancers but have largely failed to make similar inroads against solid tumors, including brain cancers.

MT026 takes the approach a step further. Instead of using each patient’s own cells, the product is manufactured from healthy donors, then heavily engineered with CRISPR-Cas9 to function as a universal, or allogeneic, therapy.

The team disrupted the TRAC gene, which encodes the T-cell receptor, to prevent graft-versus-host disease—an immune attack against the patient. They also knocked out genes encoding HLA-A, HLA-B and HLA-C molecules, key components of immune “self” recognition, to reduce rejection of the donor cells while leaving some signals intact to avoid destruction by natural killer cells.

The engineered cells carry a receptor targeting IL‑13 receptor alpha 2 (IL‑13Rα2), a protein overexpressed on most glioblastoma cells but largely absent from normal brain tissue.

“The absence of therapy-related serious adverse events and the potential clinical benefit warrant further investigation,” the authors wrote.

High-risk patients, limited options

All five participants had recurrent high-grade glioma that had progressed after standard treatment with surgery, radiation and temozolomide chemotherapy. Their tumors carried genetic features—IDH-wildtype and unmethylated MGMT promoter status—associated with aggressive disease and resistance to chemotherapy.

Most had significant disability, with four of the five scoring below 70 on the Karnofsky Performance Scale, a measure of a cancer patient’s ability to carry out daily activities.

For recurrent glioblastoma, median survival is typically six to 12 months, and there is no widely accepted standard salvage therapy. Options such as lomustine, bevacizumab, re-irradiation or reoperation often offer modest benefit.

Against that backdrop, the team ran an investigator-initiated, open-label phase 1 study registered as ChiCTR2000028801. Between August 2020 and July 2022, they enrolled five patients—fewer than the original target of 12, a shortfall they attributed partly to COVID-19 disruptions.

Injected into the spinal fluid, not the vein

Unlike approved CAR-T treatments infused into the bloodstream, MT026 was given intrathecally—injected into cerebrospinal fluid via lumbar puncture. The intent was to bathe the brain and spinal cord in therapeutic cells, bypassing the blood-brain barrier and targeting multifocal or diffuse lesions that are difficult to reach with local catheters.

Patients received four to nine intrathecal infusions spaced about four weeks apart, at doses ranging from 10 million to 30 million CAR-T cells per treatment. Notably, they did not receive the lymphodepleting chemotherapy that often precedes CAR-T therapy for blood cancers.

Two neuroradiologists, blinded to clinical data, evaluated MRI scans using specialized immunotherapy response criteria to reduce the risk of misreading treatment-related inflammation as tumor growth.

Safety profile: local inflammation, no severe neurotoxicity

The trial’s main goal was safety. According to the published data, no patient experienced grade 3 or higher treatment-related adverse events, no cases of immune effector cell-associated neurotoxicity syndrome (ICANS), and no graft-versus-host disease.

The most common side effects were mild to moderate fever (four of five patients), along with transient hypoxia, vomiting and headache. All were classified as grade 1 or 2.

Laboratory monitoring detected CAR-T DNA in cerebrospinal fluid but not in peripheral blood, suggesting the cells largely stayed within the central nervous system compartment. Levels of the inflammatory cytokine interleukin-6 rose sharply in spinal fluid after dosing but remained much lower in blood, consistent with strong local immune activation without a systemic cytokine storm.

Tumor shrinkage in all five patients

Although not designed to prove efficacy, the study tracked standard cancer endpoints. All five patients showed measurable reductions in contrast-enhancing tumor volume on MRI.

• One patient achieved a complete response.
• Three patients met criteria for partial response.
• One patient had shrinkage that did not meet partial-response thresholds and was classified as non-complete response/non-progressive disease.

Across the group, the objective response rate was 80%. The mean duration of response was 3.4 months, with progression-free survival of 4.7 months and average overall survival from recurrence of 13.1 months. One patient lived about 33 months after recurrence, though some participants later received other investigational CAR-T therapies, complicating attribution of longer-term outcomes to MT026 alone.

In at least one case, biopsy of a recurrent tumor after treatment showed loss of IL‑13Rα2 expression, suggesting antigen escape as a mechanism of resistance.

Building on a decade of glioma CAR-T attempts

The study builds on more than a decade of work with IL‑13Rα2-targeted CAR-T approaches. Early trials at City of Hope in California using autologous CAR-T cells delivered directly into tumor cavities or brain ventricles suggested feasibility and occasional dramatic responses, but in a minority of patients.

Other groups have tested allogeneic IL‑13Rα2 CAR-T products and bivalent CAR-T designs that target both IL‑13Rα2 and EGFR. In a small phase 1 trial published in 2024, six patients who received intrathecal bivalent CAR-T therapy developed neurotoxicity consistent with CAR-T-associated encephalopathy, including one case of grade 3 toxicity, even though some tumors shrank.

In contrast, the Chinese team reported no severe neurologic events with MT026, though experts caution the current dataset is limited to five patients.

A platform aimed at global trials

MT026 is part of a broader allogeneic CAR-T platform developed by T-Maximum Pharmaceutical, a Suzhou-based biotechnology company focused on solid tumors and autoimmune disease. The company has also advanced MT027, an off-the-shelf product targeting B7‑H3, another molecule expressed on gliomas.

Early clinical data for MT027 in recurrent high-grade glioma—using similar intrathecal or intracerebroventricular delivery—showed responses in roughly one-third of evaluable patients, according to company presentations. In December 2025, the U.S. Food and Drug Administration cleared an investigational new drug application for a phase 2 trial of MT027 in recurrent glioblastoma.

Open questions on CRISPR safety and access

The Nature Communications report includes laboratory assessments of off-target editing and chromosomal rearrangements in the engineered T cells using multiple sequencing-based methods. The authors reported relatively few off-target sites and low-frequency structural variants, arguing the results support the product’s “clinical translatability.”

Still, long-term risks of multiplex CRISPR editing in cells that may persist for months or longer remain uncertain, and regulators typically require extended follow-up for gene and cell therapies—particularly when the central nervous system is involved.

There are also practical questions about access and cost. Allogeneic CAR-T products could be manufactured at scale and stored, potentially reducing production time compared with bespoke autologous therapies. But intrathecal administration, intensive MRI monitoring and specialized neuro-oncology care will likely limit near-term availability to major centers.

For now, MT026 offers a proof of concept rather than a new standard of care. In a small group of patients facing a disease that is often fatal, heavily engineered donor immune cells delivered into the spinal fluid appeared safe in the short term and were associated with tumor shrinkage.

Larger, controlled trials will be needed to determine whether that early signal translates into durable benefit in one of oncology’s most challenging arenas.