Why brain cancer is often resistant to immunotherapy

Immunotherapy has revolutionized the treatment of many cancers, but brain tumors such as gliomas remain particularly difficult to treat, in part because they potently suppress immune responses.

New findings from researchers at the Broad Institute of MIT and Harvard and the Dana-Farber Cancer Institute (DFCI) could help make immunotherapies for brain cancer more effective.

The team analyzed almost 200,000 individual immune cells called myeloid cells from tumor samples from patients with glioma, the most common and aggressive type of primary brain cancer. In a new study in Nature, the researchers found four gene expression “programs” — sets of genes with coordinated activity — that either suppress the immune system or make it more active. They also found that patients treated with dexamethasone, a common treatment for brain cancer patients, showed signs of one of the immunosuppressive programs, suggesting that this drug could reduce the effectiveness of immunotherapy.

The team says that defining and understanding what drives these programs could one day help researchers target them with new drugs to dial up or down specific parts of the immune system to improve patient response to immunotherapy.

“This study provides us with the data we need to create myeloid-targeting strategies to modulate these programs and make immunotherapies more effective for brain tumor patients,” said Tyler Miller, co-first author on the study and a resident in clinical pathology at Massachusetts General Hospital when the study began.

Bradley Bernstein, an institute member at the Broad and chair of the cancer biology department at DFCI, was the study’s senior author.

“These gene signatures provide a roadmap that the field can use to study myeloid cells and how they impact the way brain tumors respond to therapy,” Bernstein said.

Chadi El Farran, a postdoctoral researcher in Bernstein’s lab, and Charles Couturier, a postdoctoral researcher in Alex Shalek’s lab at the Broad and MIT, were co-first authors on the work with Miller.

Clustering cells

As a pathology resident, Miller watched treatment after treatment fail in patients with gliomas. He’d seen the difference immunotherapy made for other cancers, and wanted to know how to improve them for brain cancer.

At the Broad, he decided to focus on myeloid cells, which make up to half of the cells in many brain tumors and can suppress the immune system, preventing immunotherapies from working well. Using single-cell RNA sequencing, which probes gene expression in individual cells, Miller and his colleagues examined nearly 200,000 cells from 85 different glioma tumors.

To do this, the team used a different approach to single-cell analysis. Scientists typically analyze single-cell data by grouping similar cells together by gene expression related to cell type. But this approach can cause researchers to overlook features such as gene expression patterns related to a cell’s activities, which are important in myeloid cells. 

So the researchers used a method developed at the Broad Institute in 2019 called consensus non-negative matrix factorization (cNMF), which can define cells by their identity and activity independently. With this approach, the team identified four programs shaping the immune system. In two, the immune system was inflammatory; it was activated and potentially trying to attack the tumor. The other two programs, found in advanced tumors, were immunosuppressive, partially shutting down the immune system and hampering its ability to fight cancer.

Treatment insights

One of the immunosuppressive programs appeared in patients who had been treated with dexamethasone, a steroid often used to reduce swelling in the brain when a patient first develops symptoms and before they receive immunotherapy. Though scientists knew that dexamethasone is immunosuppressive, they’d previously thought this was primarily due to the medication’s effects on T cells. But the new findings suggest that the drug also strongly impacts myeloid cells, and that dexamethasone should be prescribed more sparingly to improve the efficacy of immunotherapies.

“We hope this will spur additional studies to identify ways to tackle edema [brain swelling] using different drugs and also to think about how we design clinical trials based on those results,” Miller said.

He and his colleagues also created organoids, or three-dimensional assemblies of cells in the lab, from pieces of patient tumors, and treated the cultures with dexamethasone. They found that myeloid cells in the organoids continued to express the immunosuppressive programs long after they’d removed the drug, suggesting that the steroid could affect immunotherapy response even if only given to patients for a short time. 

Using the organoids, the researchers also found that cell signaling molecules such as the inflammatory protein IL-1β and the growth factor TGF-β pushed the tumors to express the other immunosuppressive cell program. 

Miller says scientists could one day manipulate the four programs with drugs to make immunotherapies more effective. In the meantime, he hopes that their approach highlights the importance of considering myeloid cells and will inspire other groups to study their roles in other tumors and patient populations.

***

Funding
This work was supported in part by the Brain Tumour Charity, the American Brain Tumor Association, The Brilliant Night Foundation, the National Institutes of Health, the Ludwig Center at Harvard, and the Emerson Collective.

Paper cited
Miller TE, El Farran CA, Couturier CP et al. Programs, origins and immunomodulatory functions of myeloid cells in glioma. Nature. Online February 26, 2025. DOI: 10.1038/s41586-025-08633-8.