UCSF's 'Brain-Activated' CAR-T Cells Wipe Out Deadly Childhood Tumors in Every Mouse Tested

A childhood brain cancer that kills nearly every patient within eleven months may have met its match in a new class of engineered immune cells — ones that lie dormant in the bloodstream and switch on only after reaching the brain. In every mouse tested across two separate tumor models, the therapy wiped out diffuse midline glioma completely. No conventional CAR-T treatment has managed that.

The therapy, developed at UC San Francisco and called B-SYNC CAR-T, was published October 28 in Science Translational Medicine. Researchers are now preparing for clinical trials at UCSF Health, potentially beginning as early as April 2027.

It's been a struggle to make CAR-T cells that overcome DMG without causing unacceptable side effects. By limiting B-SYNC's CAR expression to when and where it matters — in the brain — we saw consistent, total shutdown of DMG tumors in the lab.— Hideho Okada, MD, PhD, Professor of Neurological Surgery, UCSF, and co-corresponding author

Diffuse midline glioma is one of the few brain cancers that strikes children. It grows into the brainstem — the region that controls breathing, heartbeat, and swallowing — and spreads into surrounding healthy tissue in ways that make surgical removal impossible. Existing treatments have had little success, and the median prognosis is just eleven months, according to UCSF.

Standard CAR-T therapy works by removing a patient's T cells, engineering them to carry special receptors called CARs that recognize proteins on cancer cells, and reinfusing them to trigger an immune attack. The approach has succeeded against certain blood cancers but has repeatedly struggled against brain tumors — partly because the cancer-fighting machinery can activate throughout the body before the cells ever reach their target, leading to toxicity or exhaustion.

B-SYNC solves that problem with a molecular sensing system called synNotch. The engineered cells carry synNotch receptors tuned to detect brevican, a protein naturally distributed throughout the brain. Only when a B-SYNC cell encounters brevican — meaning it has entered the brain — does synNotch direct it to produce the cancer-targeting CAR receptors. Outside the brain, in the blood or the spleen, the cells remain inert.

That was just as we designed it.— Senthilnath Lakshmanachetty, PhD, postdoctoral researcher, Okada Laboratory, UCSF

Once activated by brevican, B-SYNC cells hunt for tumor cells carrying either of two proteins — EphA2 or IL13Rα2 — that are commonly displayed on both DMG and glioblastoma. This two-target approach matters: it means the therapy does not depend on a single tumor mutation that cancer cells can shed to escape detection.

Lakshmanachetty, who led the DMG experiments in Okada's laboratory, first tested B-SYNC delivered intravenously. Some cells traveled through the bloodstream into the brain, where they activated and began attacking tumors — while cells circulating elsewhere showed no CAR production. The team then tested intracerebroventricular injection, which delivers cells directly into the brain's fluid-filled ventricles. That method produced the most striking results: complete tumor elimination in every mouse across two models implanted with human DMG cells. Conventional CAR-T cells, given the same conditions, could not bring the tumors under control, according to Science Daily.

The protection did not stop when the tumors disappeared. When researchers reintroduced cancer cells several weeks later, the B-SYNC cells continued fighting and prevented new tumors from forming, suggesting the therapy confers lasting immune memory.

B-SYNC builds on earlier work by Okada and co-developer Wendell Lim, PhD, a professor of Cellular and Molecular Pharmacology at UCSF. In 2021, the pair reported laboratory results with a predecessor therapy called E-SYNC, designed for glioblastoma. E-SYNC used the same synNotch architecture but relied on EGFRvIII — a mutated protein present in only about 20 percent of glioblastoma patients — as its activation signal. Tumors could also lose EGFRvIII during treatment, allowing cancer cells to escape. Switching to brevican, a stable brain-tissue protein rather than a tumor mutation, was designed to make the system work for a far broader patient population.

The researchers note that the precise reason B-SYNC outperforms conventional CAR-T in the brain remains unclear — understanding that mechanism will be part of ongoing work as the team moves toward human trials.

Why it matters — DMG kills nearly every child it strikes within a year and has resisted all existing treatments, making B-SYNC's complete preclinical elimination of the cancer — and its path toward a 2027 human trial — a meaningful development for one of pediatric oncology's hardest problems.

Reported by sciencedaily.com, scitechdaily.com, ecancer.org, ucsf.edu, futurity.org