Combination Therapy Targeting Distinct Cell States Extends Survival in Mice with Diffuse Midline Glioma
Background: Different drug sensitivities coexist within one tumor
A drug combination targeting distinct cell states in diffuse midline glioma (DMG) extended survival in mouse experiments compared with monotherapy. Published in Nature Genetics, the study analyzed key regulatory proteins that maintain tumor-cell transcriptional states and proposed a strategy for selecting combination therapy candidates.
DMG is a pediatric brain tumor with limited treatment options. Within a tumor, cell populations coexist with different gene-expression patterns and regulatory-protein activity. If a drug effective against one state has little effect on another, the surviving population may contribute to treatment resistance. The researchers focused on drug combinations that could target complementary vulnerabilities across different populations.
Key findings: Identifying combination candidates across seven cell states
The researchers analyzed single-cell RNA sequencing data from 3,032 malignant cells from tumors of six patients. They used ARACNe to analyze regulatory relationships and metaVIPER to infer protein activity, distinguishing seven cell states and identifying the master regulators (MRs) that maintain them. Protein activity was inferred from the expression of target genes; it was not measured directly. The researchers also confirmed the reproducibility of the cell states and regulator patterns in eight independent pediatric DMG samples.
Next, they used CRISPR–Cas9 gene knockout screens in three patient-derived cell lines to assess whether candidate regulators were required for tumor-cell survival. Drug selection drew on transcriptional-response data for 372 compounds, including US Food and Drug Administration (FDA)-approved drugs and anticancer compounds in late-stage development. The researchers predicted which candidates could reverse abnormal regulator activity in each cell state. In subcutaneous xenograft mice, eight of the nine selected drugs selectively reduced the predicted cell states. This result should be distinguished from a treatment success rate across all 372 drugs.
The researchers assessed the survival effect of combination therapy in the DIPG4423 mouse model with tumors implanted in the pons. Median survival was 83 days with avapritinib plus ruxolitinib, compared with 53.5 days with avapritinib alone (P=0.02). The vehicle-control group had a median survival of 25 days. Avapritinib was selected to target primarily an oligodendrocyte-progenitor-cell-like state, while ruxolitinib was selected to target an astrocyte-like state.
Four of the six combinations tested significantly extended survival compared with the corresponding monotherapy. However, not every tumor-suppressing effect translated into improved survival. Avapritinib plus venetoclax inhibited tumor growth but did not extend survival. The difference in tumor volume after two weeks of treatment between the avapritinib–ruxolitinib combination group and the avapritinib-alone group also did not reach statistical significance (P=0.068).
Implications and outlook: Prioritizing combinations for clinical testing
The findings support combination designs that account not only for the dominant cell population but also for other populations that may remain after monotherapy. In cultured cells dominated by a single state, avapritinib and ruxolitinib showed an additive, rather than synergistic, interaction. The researchers interpreted the in vivo combination effect as resulting from targeting distinct cell states together.
The study does not establish a treatment benefit in patients. The in vivo validation used genetically similar models, treatment groups in the survival experiments were small (3–5 mice per group), and assessments were not blinded. A separate dinaciclib–venetoclax combination arm was stopped because of rapid weight loss. Further work must determine whether regulator targeting acts through direct cytotoxicity, differentiation, or changes in the microenvironment, as well as establish appropriate doses and treatment sequence. Additional validation using tumor data from three patients compared drug sensitivity in vitro and should also be distinguished from clinical efficacy.
The study design and results are available in the Nature Genetics paper and the open full text.
Nature Genetics, Published online: 22 April 2026; doi:10.1038/s41588-026-02550-w This paper identifies the ‘master regulators’ that maintain cell transcriptional states in diffuse midline gliomas and shows that targeting them in combination could represent a promising therapeutic avenue for the disease.
The proposed route to application is to infer regulatory-protein activity from RNA analysis of a patient’s tumor, then narrow down drug combinations targeting different cell populations for further validation. Its value lies in accounting for vulnerabilities in minority populations that may be overlooked when selecting a drug that works well only against the dominant population. In pharmaceutical and research settings, the approach could help prioritize candidates with evidence for state-specific activity. Because the tumor’s cell-state composition can shift during culture, efficacy also needs to be tested in vivo. Before use in pediatric patients, clinical trials would need to establish combination safety, drug exposure in the brain, appropriate dosing, and survival benefit.