Tumor Cell State Predicts Who Responds to Immunotherapy in Glioblastoma
A large genomic study finds mesenchymal glioblastoma responds better to immune checkpoint blockade, opening a path to smarter patient selection.
Summary
Glioblastoma is one of the deadliest brain cancers, and immune checkpoint blockade therapies have shown inconsistent results. This study from Dana-Farber and the Broad Institute analyzed 181 glioblastoma tumors using multiple sequencing methods to identify which tumor features predict response. The key finding: tumors classified as mesenchymal subtype responded significantly better to immunotherapy, while tumors with PDGFRA or CDKN2A mutations fared worse. Intriguingly, tumor mutational burden — a standard biomarker used in other cancers — was not predictive here. Mesenchymal tumors had higher immune cell infiltration and greater HLA-I expression, suggesting they are more visible to the immune system. When tumors became resistant to immunotherapy, they tended to shift away from the mesenchymal state, revealing a likely mechanism of acquired resistance unique to immunotherapy rather than standard treatment.
Detailed Summary
Glioblastoma remains one of the most lethal cancers, with median survival under 15 months even with aggressive treatment. Immune checkpoint blockade, which has transformed outcomes in melanoma and lung cancer, has largely disappointed in glioblastoma — but no one has fully understood why some patients still benefit. This study set out to identify the tumor-intrinsic features that predict immunotherapy response in this disease.
Researchers from Dana-Farber Cancer Institute, the Broad Institute, and Weizmann Institute profiled 181 ICB-treated glioblastoma samples using bulk DNA sequencing, bulk RNA sequencing, and single-nucleus RNA sequencing — one of the largest and most comprehensive molecular analyses of this patient population to date.
The central finding is that the tumor's transcriptional subtype at baseline is a strong predictor of survival after immunotherapy. Mesenchymal (MES) glioblastoma was associated with improved outcomes specifically following ICB, but not after standard chemoradiation, suggesting the benefit is immunotherapy-specific. Tumors with non-MES-associated genetic alterations, including PDGFRA and CDKN2A lesions, predicted worse ICB outcomes. Notably, tumor mutational burden — widely used as an immunotherapy biomarker in other cancers — showed no predictive value here.
At the cellular level, MES tumors showed greater T cell infiltration and elevated HLA class I expression on malignant cells, making them more recognizable to the immune system. Paired pre- and post-treatment samples revealed that acquired resistance was linked to a shift from mesenchymal toward non-mesenchymal tumor states, suggesting immune escape via transcriptional reprogramming.
These findings carry real implications for clinical trial design and patient stratification. Transcriptional subtyping could serve as a practical eligibility filter for future immunotherapy trials in glioblastoma. However, the study is based on abstract-level data only, and validation in prospective cohorts is needed before clinical implementation.
Key Findings
- Mesenchymal glioblastoma subtype predicts improved survival after immune checkpoint blockade but not standard chemoradiation.
- PDGFRA and CDKN2A mutations in non-MES tumors are associated with worse immunotherapy outcomes.
- Tumor mutational burden was not predictive of ICB response in glioblastoma.
- MES tumors show higher HLA-I expression and T cell infiltration, suggesting immune visibility drives response.
- Acquired ICB resistance is linked to mesenchymal-to-non-mesenchymal tumor state transition.
Methodology
The study profiled 181 ICB-treated wild-type IDH glioblastoma samples using bulk DNA sequencing, bulk RNA sequencing, and single-nucleus RNA sequencing. Paired pre- and post-treatment samples were analyzed to assess acquired resistance mechanisms. The multi-modal genomic approach allowed simultaneous assessment of genetic lesions, transcriptional subtypes, and tumor microenvironment composition.
Study Limitations
This summary is based on the abstract only, as the full paper is not open access, which limits assessment of statistical methods, cohort characteristics, and effect sizes. The retrospective nature of the analysis and potential selection bias in ICB-treated cohorts should be considered. Prospective validation of transcriptional subtyping as a clinical biomarker is required before adoption in practice.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.