Scientists at UC San Francisco have built a new kind of cancer treatment that stays switched off until it reaches the brain. In experiments with mice, the treatment completely removed diffuse midline glioma, or DMG, an aggressive childhood brain cancer that usually kills within a year of diagnosis. The results were published in Science Translational Medicine and described in a UCSF news release.
The therapy is called B-SYNC CAR-T. CAR-T therapy means taking a patient's own immune cells, called T cells, and giving them a new sensor, called a CAR, that recognizes cancer cells. Standard CAR-T treatments work well for some blood cancers but have mostly failed against brain tumors. The UCSF team added a second sensor called synNotch, which detects brevican, a protein found throughout the brain. Only when the cell senses brevican does it start making the cancer-fighting CAR, which then targets two proteins, EphA2 and IL13Rα2, found on many tumor cells. In effect, the brain itself acts as the switch that turns the therapy on.
Tumors cleared in every mouse tested
Researchers gave the engineered cells to mice carrying human DMG tumors, either through a vein or by injection directly into the brain's fluid-filled spaces. With direct delivery, the treatment cleared the tumors in every mouse in two different tumor models, while conventional CAR-T cells failed to control the cancer. Cells outside the brain stayed inactive, as designed. Nearly seven weeks later, the scientists gave the same mice a second round of cancer cells, and every treated mouse fought them off. The engineered cells were still active more than 70 days after treatment. The team thinks that keeping the cells dormant until they reach the brain protects them from exhaustion, a state in which immune cells lose their killing power after being active too long.
This matters because DMG grows into the brainstem, which controls breathing, heartbeat, and swallowing, and no current treatment stops it. The researchers are now preparing clinical trials at UCSF Health that could begin as early as April 2027, testing both delivery methods in patients. The results so far come only from mice, and the therapy has not yet been tested in people, so whether it works safely in humans remains unknown.