The stakes could hardly be higher. Diffuse midline glioma, including diffuse intrinsic pontine glioma, accounts for roughly 10 to 20 percent of pediatric central nervous system tumors and is the leading cause of cancer-related death in children. Median overall survival is a mere 11 months, and fewer than 10 percent of patients survive beyond two years. For decades, radiotherapy has been the only standard-of-care option, offering transient benefit at best. In August 2025, the US Food and Drug Administration granted accelerated approval to dordaviprone, also known as ONC201, for previously treated patients aged one year and older with H3K27M-mutant disease, and preclinical work supporting combination with the PI3K/mTOR inhibitor paxalisib provided the rationale for the adaptive Phase II PNOC022 trial. Yet responses remain variable and often fleeting, raising the question that sits at the heart of the review: which molecular programs allow these tumors to adapt, evolve and resist?

The answer begins with the tumor’s origin. Diffuse midline glioma is defined by alterations in histone H3 lysine 27, most commonly the H3K27M mutation, or by overexpression of EZHIP. Both events converge on disruption of polycomb repressive complex 2, the cellular machinery that maintains repressive chromatin marks, producing a global collapse in H3K27 trimethylation and a profoundly dysregulated epigenetic landscape. The consequence is aberrant reactivation of developmental programs that should have been silenced after cells committed to a lineage. Single-cell studies show the tumor arises within a developmentally restricted glial progenitor context and recapitulates an oligodendroglial differentiation hierarchy, dominated by proliferative oligodendrocyte precursor cell-like states. This extensive heterogeneity and plasticity means therapy can select for, or actively induce, resistant cellular states that remain poorly characterized.

To bring order to this complexity, the authors organize tumor adaptation into four axes: compositional shifts in cellular populations, functional state remodeling, malignant cell plasticity, and intercellular communication networks. Each axis represents a different way a tumor can respond to treatment, and each demands its own computational approach. The framework matters because resistance is not a single phenomenon. A tumor may expand a pre-existing resistant population, reprogram its transcriptional programs without changing cellular identity, transition cells between differentiation states, or rewire the signaling conversations between malignant cells and their microenvironment. Disentangling these possibilities from high-dimensional single-cell and spatial transcriptomic data requires sophisticated mathematics and machine learning.

On the first axis, the review highlights methods that model differential abundance while accounting for the compositional and technical pitfalls that plague single-cell data. Because cell populations are measured as relative proportions, depletion of one group can masquerade as enrichment of another, and conventional clustering can force cells into discrete groups even when states exist along a continuum. Tools such as Milo represent cellular states as neighborhoods within a k-nearest neighbors graph, enabling detection of responsive intermediate states, while newer machine learning approaches like Cellograph use graph neural networks to identify treatment-responsive and resistant states across conditions. Spatially informed methods such as NicheCompass extend this to tissue architecture, relevant because in diffuse midline glioma, proliferating OPC-like and OC-like cells cluster within mitotic niches, while mesenchymal-like cells colocalize with vascular cells and SPP1-positive, GPNMB-positive myeloid cells in an immunosuppressive niche.

Functional state remodeling captures a subtler phenomenon: a cell population may persist after treatment while undergoing profound reprogramming. The H3K27M mutation drives extensive metabolic rewiring, creating both vulnerabilities and escape routes. Inhibiting oxidative phosphorylation with phenformin extends survival in preclinical models, but tumors adapt by activating alternative pathways including cholesterol synthesis. Similarly, these tumors depend heavily on purine metabolism, and after radiotherapy they upregulate the purine salvage pathway; targeting hypoxanthine-guanine phosphoribosyltransferase alongside radiation produces durable responses in mouse models. Computational tools such as RECODR can detect shifts in pathway activity and identify the transcription factors driving adaptation, while TransMarker and CancerTrace reconstruct dynamic regulatory networks. Spatial methods including C-SIDE and SPATA2 then correlate these programs with anatomical position, from tumor core to infiltrative margin.

Plasticity, the third axis, may be the defining feature of this disease. H3K27 alterations elevate α-ketoglutarate, a cofactor for histone demethylases, reinforcing the hypomethylated state. Dordaviprone disrupts mitochondrial homeostasis and alters α-ketoglutarate availability, partially restoring H3K27 methylation, proof that epigenetic states are therapeutically modifiable. To track how cells move between states, the review describes trajectory inference methods such as Monocle and Slingshot, which have revealed multiple OPC-like states in diffuse midline glioma and a resistance-associated trajectory in DIPG models, with resilient mesenchymal-like and mitotic populations emerging after triple therapy. Because pseudotime from static snapshots cannot establish directionality, RNA velocity approaches and deep generative models such as PRESCIENT, GRNvelo and CytoTRACE2 add temporal and potency information. Optimal transport methods including STIFT, Moscot and STORIES reconstruct cellular trajectories while preserving spatial context, potentially revealing whether resistant states pre-exist or are induced by therapy.

The fourth axis acknowledges that tumor cells do not adapt in isolation. The microenvironment of diffuse midline glioma is dominated by tumor-associated myeloid cells with sparse lymphocyte infiltration, and neuron-glioma connectivity is increasingly implicated in tumor genesis. Ligand-receptor inference tools such as CellChat and CellPhoneDB have exposed extensive tumor-myeloid communication, including the IGSF11-VISTA immune checkpoint axis. Newer frameworks push further: scComm uses supervised contrastive learning to reconstruct communication at single-cell resolution, CELLetter integrates large language models to predict novel ligand-receptor pairs, Tensor-cell2cell models interactions across multiple conditions simultaneously, and COMMOT and CellNEST exploit spatial proximity and graph attention to map signaling between neighboring cells. DualCellChat even infers directionality, identifying senders and receivers. These networks highlight actionable pathways that could be disrupted to restore antitumor immunity or enhance treatment efficacy.

Translating these models into the clinic remains formidable. True longitudinal sampling is largely unfeasible in patients, with matched specimens typically limited to diagnostic biopsies and post-mortem autopsies, confounded by cumulative therapy effects. Deep learning demands patient-level replication, careful separation of training and test data, and robust generalization across institutions. The authors point to preclinical models, organoids, patient-derived xenografts and liquid biopsies of blood or cerebrospinal fluid as bridges toward prospective modeling, and note that biomarkers such as TP53 mutation status, linked to radioresistance and reduced dordaviprone response, could stratify patients in adaptive trials. Foundation models like NicheFormer, pretrained on large single-cell atlases, have yet to be applied to resistance trajectories but represent a promising direction. If the field succeeds, the authors conclude, artificial intelligence integrated with scientific and ethical rigor could make diffuse midline glioma care predictive, personalized and proactive, a future in which resistance is anticipated before it emerges rather than discovered after treatment fails.

Subject of Research: Machine learning and computational modeling of therapeutic response and resistance in diffuse midline glioma using single-cell and spatial transcriptomics

Article Title: Machine learning and computational approaches to model therapeutic response and resistance in diffuse midline glioma

Article References: Dubois, O. A., Loughnan, O. E. L., Jiang, Y., Petraello, B., Douglas, A. M., McLachlan, T., Findlay, I. J., Vo, T., Savary, C., & Dun, M. D. (2026). Machine learning and computational approaches to model therapeutic response and resistance in diffuse midline glioma. Journal of Neuro-Oncology, 179(3), Article 104. https://doi.org/10.1007/s11060-026-05817-8

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05817-8

Keywords: diffuse midline glioma, DIPG, machine learning, single-cell RNA sequencing, spatial transcriptomics, therapeutic resistance, H3K27M, tumor plasticity, tumor microenvironment, dordaviprone, computational biology, pediatric oncology

Diffuse midline glioma, H3 K27-altered, is one of the most aggressive tumor entities recognized by the World Health Organization Classification of Central Nervous System Tumors. These tumors arise in midline structures of the brain and spinal cord, including the brainstem, thalamus and spinal cord, and they carry a prognosis that remains grim despite decades of clinical trials. The defining molecular feature of the entity is disruption of polycomb repressive complex 2 function at lysine 27 of histone H3, which manifests either as the canonical K27M oncohistone substitution in H3.3 or H3.1, or as overexpression of EZHIP, a protein that mimics the inhibitory effect of the mutant histone on the methyltransferase complex. The downstream consequence is a characteristic global reduction of tri-methylated lysine 27, detectable in diagnostic sections as loss of nuclear H3K27me3 immunostaining in tumor cells, with retained staining in internal control elements such as endothelial cells and infiltrating lymphocytes.

The original study by Tauziède-Espariat and colleagues, which prompted the commentary, reported cases in which this hallmark loss of H3K27me3 staining was present even though the tumors harbored neither H3K27M mutations nor EZHIP expression. The authors of that study flagged the finding as a potential diagnostic pitfall, because current diagnostic algorithms often treat the combination of midline location and H3K27me3 immunoloss as presumptive evidence of H3 K27-alteration, frequently in the absence of confirmatory sequencing. In the absence of the two canonical drivers, the mechanism underlying the epigenetic silencing of the mark remained opaque, leaving open the question of whether such tumors truly belong to the H3 K27-altered category or represent a separate, still-undefined process.

The Moscow-based commentators propose an explanation grounded in the biophysics of the histone H3 amino-terminal tail. The lysine 27 residue sits within a short stretch of the tail that also contains serine 28, immediately adjacent to the methylated lysine, and serine 31, located just a few residues downstream. Both serines are functionally important phosphorylation sites. Serine 28 phosphorylation by Aurora kinase B is classically associated with mitotic chromosome condensation, while phosphorylation of serine 31 in the H3.3 variant has been linked to chromosome segregation fidelity, p53-dependent cell cycle arrest after missegregation, enhancer acetylation through activation of the p300 coactivator, and the regulation of nucleosome dynamics during transcription. Recent work summarized in the commentary indicates that K27M oncohistones suppress serine 31 phosphorylation and that this suppression can itself contribute to gliomagenesis, underscoring how tightly the biology of these neighboring residues is interwoven with tumor formation.

Against this background, the commentators draw attention to tumor sequencing data in which H3 K27-altered diffuse midline gliomas carried additional, concurrent substitutions converting serine 28 to cysteine, or serine 31 to cysteine or tyrosine. These are not random passengers, they argue, but may reflect a mutation bias inherent to the biology of the tumor. One plausible mechanism is the altered local chromatin environment created by K27-alteration itself: the loss of polycomb-mediated repression and the resulting transcriptional deregulation can expose the H3-encoding genes, particularly the replication-independent H3.3 variant genes H3F3A and H3F3B, to error-prone repair or to the activity of error-prone polymerases. Because codons for serine 28 and serine 31 are chemically primed for specific single-base substitutions, cysteine and tyrosine represent favored conversion products, and their recurrent appearance in the same tumors that alter lysine 27 is unlikely to be coincidental.

The functional implications of such substitutions are potentially profound. Replacing a phosphorylatable serine with cysteine or tyrosine permanently abolishes that phosphorylation site, mimicking a non-phosphorylatable state that cannot be dynamically regulated by kinases. In the case of serine 31, evidence from model systems shows that loss of H3.3 S31 phosphorylation disrupts mitotic fidelity and promotes chromosomal instability, a hallmark of high-grade gliomas. In the case of serine 28, studies in Drosophila and in mammalian cells demonstrate that this residue is essential for efficient polycomb-mediated gene repression and that its modification state influences heterochromatin formation through effects on H3K9 and H3K36 demethylases. A tumor cell in which both lysine 27 methylation is disabled and neighboring serines are locked into non-phosphorylatable states would therefore accumulate multiple converging disruptions of the same regulatory hub, potentially reinforcing the oncogenic chromatin program.

The diagnostic dimension of the argument is where the commentary acquires its practical urgency. Commercial diagnostic antibodies against H3K27M and against other mutant H3 epitopes depend on precise recognition of short amino acid sequences surrounding the substituted residue. Substitutions in immediately adjacent residues, such as S28C, which sits directly beside lysine 27, or S31C and S31Y, can alter the local epitope landscape. Depending on the antibody clone and its epitope, such a change could either weaken binding and produce a false-negative result, or in some assay contexts generate aberrant or ambiguous staining patterns. A laboratory that interprets an unexpected immunoprofile at face value, without sequencing, could misclassify a tumor, and given that diffuse midline glioma is frequently diagnosed on small stereotactic or open biopsies where tissue is scarce, the margin for error is narrow.

The commentators therefore recommend a disciplined diagnostic pathway. When a midline glioma shows loss of H3K27me3 but is negative for H3K27M by immunostaining or sequencing, and EZHIP overexpression is absent, the possibility of concurrent histone tail substitutions should be actively considered. Targeted sequencing of the H3F3A and H3F3B loci, ideally with coverage capable of detecting co-occurring variants on the same allele or on separate histone gene copies, can resolve the ambiguity. Methylation profiling, now regarded as an increasingly essential adjunct in central nervous system tumor diagnostics, can provide independent support for assigning such tumors to the H3 K27-altered diffuse midline glioma class, since the methylation signature of the entity appears to be preserved even when individual driver assays behave unpredictably. Integration of histopathology, immunohistochemistry, sequencing and methylation classification, rather than reliance on any single test, emerges as the central lesson.

Beyond the diagnostic clinic, the observation raises broader questions about how oncohistone tumors evolve. If K27-alteration biases the histone tail toward further substitutions at serine 28 and serine 31, then the canonical model of a single dominant oncohistone mutation may be incomplete, and the H3 tail may function as a recurrent mutational hotspot whose successive hits progressively dismantle polycomb regulation and mitotic control. The commentators, whose work was supported by the Foundation for support and development in the field of pediatric hematology, oncology and immunology Science for Children and approved by the ethics committee of the Dmitry Rogachev Center, frame their argument as a call to the neuropathology community to recognize that H3K27me3 loss without canonical drivers is not necessarily an artifact but a window into an additional layer of histone biology. For a disease in which molecular classification currently drives eligibility for targeted trials, distinguishing true biological complexity from mere diagnostic noise may determine which patients receive the right investigational therapy, and the serine residues flanking the infamous lysine 27 have now moved from biochemical footnotes to the center of that conversation.

Subject of Research: Concurrent histone H3 S28C and S31C/Y substitutions in H3 K27-altered diffuse midline gliomas and their impact on diagnosis

Article Title: Concurrent substitutions H3 S28C or S31C/Y in H3 K27-altered diffuse midline gliomas may represent mutation bias, apart from being a source of potential diagnostic pitfall. Scientific commentary on: “H3K27ME3 loss in diffuse midline gliomas lacking H3K27M or EZHIP expressions, a potential diagnostic pitfall!”

Article References: Zaytseva, M., Papusha, L., & Druy, A. (2026). Concurrent substitutions H3 S28C or S31C/Y in H3 K27-altered diffuse midline gliomas may represent mutation bias, apart from being a source of potential diagnostic pitfall. Scientific commentary on: “H3K27ME3 loss in diffuse midline gliomas lacking H3K27M or EZHIP expressions, a potential diagnostic pitfall!”. Acta Neuropathologica, 152(1), Article 33. https://doi.org/10.1007/s00401-026-03083-6

Image Credits: AI Generated

DOI: 10.1007/s00401-026-03083-6

Keywords: diffuse midline glioma, histone H3, H3K27me3, H3K27M, EZHIP, oncohistone, serine 28, serine 31, polycomb repressive complex 2, diagnostic pitfall, neuropathology, brain tumor