The review, led by George W. Koutsouras and Michael C. Dewan of Vanderbilt University Medical Center together with colleagues at the Barrow Neurological Institute, Princess Margaret Cancer Centre, and The Hospital for Sick Children, synthesizes the contemporary literature on molecular classification, imaging, and treatment outcomes for posterior fossa ependymoma. Its central message is that the World Health Organization’s shift from a purely histologic framework to molecularly integrated entities has finally caught up with the biology. Posterior fossa ependymoma group A (PFA) and group B (PFB) are not grades of the same tumor but separate diseases with different epigenetic engines, different age distributions, different imaging signatures, and sharply different prognoses.

The biological fault line runs through the epigenome. PFA tumors, which predominate in younger children with a median age of about five years, show a global loss of the repressive histone mark H3K27me3. This loss stems from inhibition of the polycomb repressive complex 2, the molecular machine that normally silences genes by trimming histone proteins, and the inhibition is commonly driven by overexpression of a protein called EZHIP, which mimics the notorious K27M oncohistone mechanism. The result is widespread derepression of developmental gene programs that fuel tumor growth. PFB tumors, by contrast, retain their H3K27me3 mark, preserve PRC2 function, and instead carry numerous arm-level chromosomal gains and losses, a pattern of somatic aneuploidy that paradoxically accompanies a more indolent clinical course. Within PFA itself, the review highlights an even more ominous stratum: tumors with chromosome 1q gain and/or 6q loss form a very high-risk group marked by increased recurrence, metastatic spread, and poor survival even after gross total resection and upfront radiotherapy.

Clinically, the two subgroups announce themselves differently. Children with posterior fossa ependymomas typically present with headache, nausea, vomiting, lethargy, and gait disturbance, the consequences of obstructive hydrocephalus as the tumor blocks cerebrospinal fluid flow from the fourth ventricle, often accompanied by truncal ataxia and sometimes lower cranial nerve deficits. Symptom onset can be insidious, and a long history of vomiting without headache may delay diagnosis. But PFA tumors tend to arise in younger children, carry a larger tumor burden, and produce more pronounced hydrocephalus, while PFB tumors appear in adolescents and adults and behave less aggressively. Even magnetic resonance imaging reflects the molecular divide: PFA tumors more often show larger volume, hydrocephalus, lateral extension through the skull base, and lower apparent diffusion coefficient values, whereas PFB tumors are typically more homogeneous, circumscribed, and cystic. The authors recommend craniospinal MRI at diagnosis, during surveillance, and at recurrence, noting that thin-cut T2-weighted spinal imaging is essential because metastatic deposits may not enhance.

When it comes to treatment, the review is unambiguous about the hierarchy of interventions. Surgery remains the single most important component of care, and the extent of resection is the key modifiable determinant of disease control. The evidence is striking. In the Children’s Oncology Group ACNS0121 study, gross total resection followed by immediate conformal radiotherapy yielded a five-year event-free survival of 68.5 percent and overall survival of 86.2 percent, compared with 37.2 percent and 70.2 percent for subtotal resection followed by adjuvant therapy. A posterior fossa-specific study by Silva da Costa and colleagues found median overall survival of 136 months after gross total resection versus 38 months after subtotal resection, with progression-free survival of 40 versus 7 months. The authors caution, however, that most surgical series are retrospective and that extent of resection partly reflects tumor anatomy, biology, and institutional experience rather than a pure treatment effect. Notably, adverse PFA biology, particularly 1q gain and 6q loss, may attenuate the survival benefit of complete resection, meaning that even a perfect operation cannot fully overcome hostile molecular terrain.

The surgical challenge itself is formidable. These tumors arise from the fourth ventricle and frequently extend through the foramina of Luschka and Magendie, wrapping around cranial nerve nuclei, the brainstem, and the vertebrobasilar circulation. In a recent molecularly defined pediatric cohort, cranial neuropathies occurred in 60 percent of patients after primary resection, though 56 percent of those resolved within a year. The review describes operative adjuncts in detail: prone positioning with individualized head fixation, since pin fixation in children under two risks calvarial injury; intraoperative neuromonitoring with spontaneous EMG recordings of cranial nerves; neuronavigation, ultrasound, and intraoperative MRI; and external ventricular drain placement, given that hydrocephalus affects up to 80 percent of children with a posterior fossa mass. When residual disease remains, early second-look surgery at an experienced center is favored, and complex ventral extension may occasionally require staged operations, always balanced against the morbidity of delaying adjuvant therapy.

Radiation therapy forms the second pillar of treatment. Postoperative focal radiotherapy is recommended for most children with localized disease after maximal safe resection, typically at a dose of 59.4 Gy, reduced to 54 Gy in selected very young patients after complete resection. Contemporary conformal techniques, including proton therapy, have progressively reduced the exposure of healthy tissue: Children’s Oncology Group protocols shrank clinical target volume margins from 10 millimeters in ACNS0121 to 5 millimeters in ACNS0831, limiting dose to normal brain and cerebellum. Proton therapy offers dosimetric advantages with disease control comparable to photon treatment, though the authors note there is no high-quality data directly comparing toxicity between modalities, and prescription doses for protons still vary from 54 to 59.4 GyE because of concerns about brainstem toxicity. Craniospinal irradiation is generally reserved for metastatic disease or recurrence, and failure after treatment remains predominantly local, underscoring that radiation efficacy depends on dose, technique, target definition, and the extent of resection achieved.

Chemotherapy, by contrast, has little to no definitive role. The randomized ACNS0831 trial found that adding four cycles of vincristine, cyclophosphamide, etoposide, and cisplatin to radiotherapy did not improve outcomes across the overall cohort, and the ACNS1021 trial showed that the targeted agent sunitinib produced no sustained objective responses. Historical baby-brain protocols that deferred radiation in children under three achieved five-year progression-free survival of at most 42 percent, while early conformal radiotherapy in comparable populations pushed seven-year progression-free survival toward 77 percent, a comparison that drove the field away from radiation-deferral strategies in children older than 12 months. One emerging exception is molecularly directed therapy: preclinical work by Griesinger and colleagues identified 1q-gain PFA as the highest-risk subgroup and demonstrated sensitivity to 5-fluorouracil and all-trans retinoic acid, a finding the review flags as a promising avenue for future investigation.

Recurrence remains the disease’s cruelest feature, occurring in up to 48 percent of cases with mortality approaching 50 percent despite multimodal therapy, and mean survival after first recurrence of roughly 31 months. More than 80 percent of recurrent infratentorial ependymomas are PFA tumors, with post-recurrence survival of 24.7 months compared with 48 months for PFB. Tumor biology evolves at relapse: 1q gain and 6q loss are enriched at recurrence, and 1q gain shortens the latency to relapse from 25.3 to 14.7 months. Recurrence can appear up to a decade after treatment, sometimes with spinal or supratentorial dissemination, which is why the authors emphasize prolonged surveillance with serial brain and spine MRI tailored to molecular risk. At relapse, management hinges on resectability: repeat surgery, even for multiply recurrent disease, is associated with improved survival, and selective re-irradiation, including craniospinal re-irradiation for posterior fossa primaries, may extend control. The review’s overall vision is a risk-adapted framework in which methylation profiling, copy number analysis, imaging biomarkers, and modern radiotherapy converge to individualize care, with treatment de-escalation for favorable PFB tumors tested only in prospective trials and molecularly directed therapies offering the clearest hope for the children whose tumors, until now, have defied every conventional weapon thrown at them.

Subject of Research: Molecular classification and management of pediatric posterior fossa ependymoma, focusing on the distinct PFA and PFB subgroups

Article Title: Posterior fossa ependymoma: a comprehensive review of molecular classification, management guidelines, and clinical outcomes (Part I of ependymomas across compartments)

Article References: Koutsouras, G. W., Rivera, F., Price, A. M., Tsang, D. S., Esbenshade, A., Ramaswamy, V., Dirks, P. B., & Dewan, M. C. (2026). Posterior fossa ependymoma: a comprehensive review of molecular classification, management guidelines, and clinical outcomes (Part I of ependymomas across compartments). Journal of Neuro-Oncology, 179(3), Article 96. https://doi.org/10.1007/s11060-026-05808-9

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05808-9

Keywords: posterior fossa ependymoma, PFA, PFB, H3K27me3, EZHIP, pediatric brain tumor, gross total resection, conformal radiotherapy, proton therapy, chromosome 1q gain, 6q loss, molecular classification

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