B-SYNC CAR-T clears diffuse midline glioma in mouse models
A new CAR-T therapy designed to switch on only inside the brain erased diffuse midline glioma in two mouse models and kept working long after the original tumors disappeared. The engineered immune cells, called B-SYNC CAR-T, were built to remain quiet in the bloodstream and elsewhere in the body, then activate when they encountered a brain-specific signal. In a cancer defined by its location as much as by its biology, that design choice is the key advance.
Diffuse midline glioma is one of the most lethal pediatric brain cancers. It grows into the brainstem, where surgical removal is not feasible and where even small changes in tumor burden can threaten breathing, swallowing, and heart rate. Standard treatment has done little to change the disease course, and nearly all children diagnosed with it die within a year. Against that backdrop, total tumor elimination in every treated mouse is not a minor laboratory signal; it is a sharp departure from the usual ceiling of efficacy in this disease.
The strategy builds on years of work from UCSF investigators trying to make immune cells more precise and less toxic. Conventional CAR-T therapy has transformed some blood cancers, but solid tumors in the brain have resisted that approach. One reason is biological, not merely technical: if the cells are activated too broadly, they can exhaust themselves or cause harm before reaching the tumor; if they are too weak, they never get traction against invasive cancer cells embedded in normal tissue.
A brain-specific switch replaces tumor mutation dependence
Earlier work from the UCSF group used a synNotch system called E-SYNC, which depended on EGFRvIII, a mutant protein found in only a subset of glioblastomas. Once that protein was recognized, the cell was instructed to produce a CAR that targeted EphA2 and IL13Rα2. The logic was elegant, but the dependence on EGFRvIII created a built-in escape route: only about 20% of patients carry tumors with that mutation, and cancers can lose it under treatment pressure.
B-SYNC was designed to solve that problem by using brevican, a protein naturally present throughout the brain, as the trigger for CAR expression. In practical terms, the engineered T cell does not begin its full cytotoxic program until it enters the brain environment. Once activated, it can attack tumor cells bearing EphA2 or IL13Rα2, even if those cells lack EGFRvIII. The logic is straightforward and biologically appealing: use the organ itself as the activation cue, then aim the immune attack at tumor-associated targets that are more broadly shared across disease states.
The concept matters because brain tumors often blend into their surroundings. They do not always present a single reliable mutation to grab onto, and they can spread into tissue that cannot be removed without catastrophic injury. A therapy that waits for the brain before switching on may preserve systemic safety while sharpening local potency.
Complete tumor clearance and durable immune activity in mice
The preclinical testing showed that B-SYNC could work from two different delivery routes. When investigators infused the engineered cells intravenously into mice bearing human DMG, some of the cells crossed into the brain and began producing the cancer-targeting receptor only after reaching that compartment. Cells circulating in blood and present in the spleen did not turn on CAR expression, which is exactly the sort of spatial restriction the platform was built to achieve.
A more direct approach performed even better. When the cells were delivered by intracerebroventricular injection into the brain’s fluid-filled ventricles, the therapy completely eliminated tumors in every mouse tested across two DMG models. Conventional CAR-T cells did not control the tumors. That difference is striking because it suggests the therapeutic gain came not from generic immune activation, but from a more disciplined pattern of activation that matched the tumor’s location.
One sentence captures the result plainly: every treated mouse cleared the tumor.
The study did not stop at initial response. After the original tumors were eradicated, researchers reintroduced DMG cells nearly seven weeks later. Every mouse that had received B-SYNC rejected the new challenge, and the engineered cells remained functional more than 70 days after treatment. Persistence matters in CAR-T therapy because prolonged activation can lead to exhaustion, a state in which immune cells lose their ability to kill effectively. By deferring activation until the cells reach the brain, B-SYNC may avoid burning through their functional capacity before they ever meet the tumor.
Another sentence stands on its own: the cells were still patrolling the brain more than 70 days later.
Human testing is being prepared, with route of delivery still under study
The next step is clinical translation at UCSF Health, with trials being prepared that could begin as early as April 2027. The planned treatment will use each patient’s own T cells, genetically modified into B-SYNC CAR-T cells. Investigators intend to evaluate both intravenous delivery and direct brain administration, a comparison that will be important because route of delivery may determine how much of the therapy actually reaches the disease site and how repeat dosing can be managed.
For patients receiving intracerebroventricular treatment, surgeons will implant a shunt that connects the brain’s ventricles to a reservoir beneath the scalp. That setup would allow repeated infusions without repeated surgery, a practical detail with real bedside implications for children facing a disease that progresses quickly and often leaves little room for delay. The therapy has not yet been tested in humans, so the open questions are not small ones: how well the activation system will behave in patients, whether the immune cells will preserve their selectivity outside the laboratory, and how safety will compare with older CAR-T approaches in a tumor type notorious for infiltrating critical brain tissue.
The preclinical results still matter because they solve several problems at once. They show tumor clearance, durable immune persistence, and a route to limit off-target activation by tying the CAR switch to the brain itself. For diffuse midline glioma, where nearly every existing strategy has failed to produce lasting benefit, that combination creates a credible rationale for early-phase human testing rather than another abstract proof of concept.