Engineered Immune Cells That Switch On Only Inside the Brain Cleared a Deadly Childhood Tumor in Mice

Childhood brain tumors are among the hardest cancers to treat, and one of them, diffuse midline glioma (DMG), is close to the bleakest diagnosis in pediatric oncology. Nearly all children with DMG die within a year, and existing treatments have done little to change that. A new study from the University of California, San Francisco (UCSF) describes an engineered immune cell therapy that cleared these tumors in mice, and the design behind it is as interesting as the result.

The work was published on October 28, 2026 in Science Translational Medicine. It is preclinical, which means it was done in mice, not people. The researchers are now preparing for human trials that could begin as early as April 2027.

Why DMG is so difficult

DMG grows into the brainstem, the region that controls breathing, heartbeat and swallowing. The tumor weaves into healthy tissue instead of forming a clean mass, so surgery is not an option and there is little room for treatments that damage surrounding brain.

CAR-T therapy, which has worked well against some blood cancers, takes a patient’s own T cells and equips them with a receptor (a chimeric antigen receptor, or CAR) that recognizes proteins on cancer cells. Against brain tumors it has struggled. A CAR that is “always on” can cause unacceptable side effects, and T cells that stay activated for long periods can become exhausted and lose their ability to attack.

A switch that only flips inside the brain

The therapy, called B-SYNC, builds on earlier work by neurosurgeon Hideho Okada and cell biologist Wendell Lim at UCSF. In 2021 they reported E-SYNC, a glioblastoma therapy built on a sensor system called synNotch. E-SYNC detected EGFRvIII, a mutated protein found on only about 20% of patients’ tumors. It could also fail when tumors lost that protein under treatment.

For B-SYNC, the team changed the trigger. Instead of a tumor-specific mutation, the synNotch sensor responds to brevican, a protein found throughout the brain. Once a T cell senses brevican, it begins producing a CAR aimed at EphA2 and IL13Rα2, two proteins that appear on both glioblastoma and DMG cells. The logic works like a two-step lock: be in the brain first, then look for the tumor.

What happened in the mice

·         Activation stayed in the brain. When B-SYNC cells were given intravenously to mice carrying human DMG cells, some reached the brain and began making the CAR. B-SYNC cells in the blood and spleen did not.

·         Complete tumor clearance. When the cells were injected directly into the brain’s fluid-filled ventricles (an intracerebroventricular, or ICV, injection), the treatment eliminated the tumors in every mouse tested, across two mouse models. Conventional CAR-T cells did not bring the tumors under control.

·         Protection against recurrence. Nearly seven weeks after treatment, the researchers gave the mice a second round of DMG cells. Every treated mouse rejected them.

·         Long-lived, functional cells. B-SYNC cells were still present and working more than 70 days after treatment.

Why might it work better?

The researchers suggest that holding back CAR production until the cells reach the brain may protect them from exhaustion, so they arrive at the tumor with their strength intact. The researchers have not yet established that this is the actual reason, and it remains a hypothesis. The press materials describe the precise cause of the improved performance as unclear.

How to read these results

“Every mouse tested” is a striking phrase, but it deserves context. These were mouse models in which human tumor cells were implanted, which cannot fully reproduce how DMG behaves in a child. Treatments that look excellent in mice often perform far less impressively in people, and the safety of B-SYNC has not yet been tested in humans. The planned first trial will be led by UCSF physicians Marie Jaeger-Krause and Sabine Mueller and will use each patient’s own T cells. It will compare intravenous delivery with direct delivery into the brain. For the direct route, surgeons would place a shunt connecting the ventricles to a reservoir under the scalp, allowing repeat doses without repeat surgery.

In practical terms, this is a strong reason to move into a clinical trial, not evidence yet that the therapy works in children.

This post is a summary based on the UCSF press release. For full methods, statistics and limitations, see the original paper.

Sources

·         Press release: Smarter CAR-T cells could beat a deadly pediatric brain cancer (University of California, San Francisco, via EurekAlert)

·         Original paper: Lakshmanachetty S. et al., Brain-sensing synNotch-CAR T cells enable precise targeting and clearance of diffuse midline glioma in multiple preclinical models, Science Translational Medicine 2026; 18(873) (DOI: 10.1126/scitranslmed.aed9994)

Source: Engineered Immune Cells That Switch On Only Inside the Brain Cleared a Deadly Childhood Tumor in Mice