Related: AI screens 6 million molecules to find new antibiotics for drug resistant gonorrhea.
Researchers have designed an immunotherapy that attacks glioblastoma, one of the most lethal and stubborn cancers, by hitting it and its hidden helpers at the same time. In a study published on July 1 in the journal Nature, a team engineered CAR-T cells to recognize a protein called GPNMB that sits both on the tumor cells and on the immune cells the tumor recruits to protect itself. In preclinical models the two pronged approach eliminated detectable tumors and led to long term survival, a striking result against a cancer that has defeated most treatments thrown at it.
Glioblastoma is so hard to treat in part because it does not fight alone. The tumor surrounds itself with macrophages, a type of immune cell that normally hunts threats but that glioblastoma corrupts into allies, using them to blunt the body’s defenses and help the cancer grow. Most therapies aim at the tumor cells and leave that supportive environment intact, which is one reason the disease so often returns. The new work tries to collapse both at once, going after the cancer and the crooked immune cells sustaining it in a single strike.
The key was finding a target common to both. Using a multi omic discovery platform, which combines several layers of molecular data to hunt for vulnerabilities, the researchers identified GPNMB, short for glycoprotein non metastatic melanoma protein B, as a marker present on glioblastoma cells and on the tumor supporting macrophages alike. That dual presence is what makes it valuable, because a therapy pointed at GPNMB can attack the tumor and its protective niche without needing two separate treatments. The team then built CAR-T cells, immune cells taken from the body and re engineered to seek out a chosen target, to recognize GPNMB and destroy whatever carries it.
The results in the laboratory were strong. In several preclinical models, including tumors grown from human patient samples implanted in mice and models designed to mimic a working immune system, the anti GPNMB CAR-T cells showed potent antitumor activity and delivered long term disease control. In many cases the therapy wiped out detectable tumors and the animals survived long term without the cancer coming back, an outcome that is rare in glioblastoma research and that points to the promise of targeting the disease and its environment together.
The appropriate caution is that this is early stage science, not a treatment patients can receive. The work was done in animal and laboratory models, and the long road from a promising preclinical result to a therapy proven safe and effective in people is where many cancer breakthroughs stall. Glioblastoma has a particularly brutal history of treatments that shine in the lab and disappoint in the clinic, so the findings should be read as a strong signal rather than a finished answer. Human trials, with all their uncertainty, are the next hurdle.
What gives the study reach beyond this one cancer is the method behind it. Lead author Professor Sheila Singh and her colleagues frame the discovery of a dual compartment target as a template, a way to systematically find antigens that let a single therapy hit both a solid tumor and the immune cells it exploits. If that framework holds, it could be applied to other hard to treat solid cancers that hide behind a similar shield, extending the idea well past glioblastoma. For a disease that has seen little progress in decades, an approach that treats the tumor and its accomplices as one problem is a genuinely new line of attack, and the coming years of testing will show how far it can go.
Source: study published in Nature (June/July 2026), led by researchers at King’s College London and McMaster University. Credit: Nature / King’s College London.