Diffuse hemispheric glioma with H3 p.K28M (K27M) mutation: Unusual non-midline presentation of diffuse midline glioma, H3 K27M-altered?

Abstract

Diffuse midline glioma, H3 K27-altered (DMG-H3 K27) is an aggressive group of diffuse gliomas that predominantly occurs in pediatric patients, involves midline structures, and displays loss of H3 p.K28me3 (K27me3) expression by immunohistochemistry and characteristic genetic/epigenetic profile. Rare examples of a diffuse glioma with an H3 p.K28M (K27M) mutation and without involvement of the midline structures, so-called “diffuse hemispheric glioma with H3 p.K28M (K27M) mutation” (DHG-H3 K27), have been reported. Herein, we describe 2 additional cases of radiologically confirmed DHG-H3 K27 and summarize previously reported cases. We performed histological, immunohistochemical, molecular, and DNA methylation analysis and provided clinical follow-up in both cases. Overall, DHG-H3 K27 is an unusual group of diffuse gliomas that shows similar clinical, histopathological, genomic, and epigenetic features to DMG-H3 K27 as well as enrichment for activating alterations in MAPK pathway genes. These findings suggest that DHG-H3 K27 is closely related to DMG-H3 K27 and may represent an unusual presentation of DMG-H3 K27 without apparent midline involvement and with frequent MAPK pathway activation. Detailed reports of additional cases with clinical follow-up will be important to expand our understanding of this unusual group of diffuse gliomas and to better define the clinical outcome and how to classify DHG-H3 K27.

INTRODUCTION

Diffuse midline glioma, H3 K27-altered (DMG-H3 K27) is defined as a diffuse glioma that shows loss of H3 p.K28me3 (K27me3) expression by immunohistochemistry and midline location and (1) H3 p.K28M (K27M) or p.K28I (K27I) mutation (for H3 K27-mutant subtypes) or (2) pathogenic mutation or amplification of EGFR (for the EGFR-mutant subtype) or (3) overexpression of EZHIP (for the H3-wildtype with EZHIP overexpression subtype) or (4) methylation profile of one of the subtypes of diffuse midline glioma (1). DMG-H3 K27 is associated with an aggressive clinical course regardless of the presence of high-grade morphological features (ie, microvascular proliferation or necrosis) and is considered central nervous system (CNS) WHO grade 4 (1).

In contrast to diffuse gliomas with an H3.3 p.G35R (G34R) or p.G35V (G34V) mutation, which are primarily hemispheric and fulfill the diagnostic criteria for “diffuse hemispheric glioma, H3 G34-mutant” (DHG-H3 G34), diffuse gliomas harboring an H3 p.K28M (K27M) mutation may rarely present without apparent involvement of midline structures and do not fulfill the diagnostic criteria for DMG-H3 K27. Diffuse gliomas with an H3 p.K28M (K27M) mutation without midline involvement have been reported (2–9) but radiological confirmation of non-midline location without any midline involvement has been documented only in a limited number of cases (7–9). The WHO classification currently recommends reporting such cases as “diffuse hemispheric glioma with H3 p.K28M (K27M) mutation not elsewhere classified (NEC)” (1), herein referred to as DHG-H3 K27. It remains unclear how DHG-H3 K27 relates to DMG-H3 K27 and whether it represents a distinct entity or an unusual non-midline presentation of DMG-H3 K27.

Methylation profiling has emerged as a powerful tool and molecular biomarker to assist in the classification of CNS tumors and delineation of clinically relevant subgroups within well-defined tumor types (1, 10). Herein, we describe 2 radiologically confirmed DHG-H3 K27 with epigenetic and molecular characterization as well as clinical follow-up and summarize other reported cases to further our understanding of this unusual group of diffuse gliomas.

MATERIALS AND METHODS

This study was conducted according to Institutional Review Board-approved protocols.

Case selection

Cases 1 and 2 were identified through a search among diffuse gliomas evaluated by the Mayo Clinic neuro-oncology next-generation sequencing (NGS) clinical test (Rochester, MN) and neuropathology consultation practice (Scottsdale, Arizona) from 2018 to 2022. Clinical, radiological, treatment, and follow-up data were extracted from electronic medical records. Previously reported DHG-H3 K27 were identified via PubMed literature search of studies that reported DMG-H3 K27. Available demographic, clinico-radiological, histopathological, genetic, treatment, and follow-up data were collected. Only cases with available imaging documenting absence of midline involvement and genomic/epigenetic characterization are summarized in the Table. All reported cases are detailed in Supplementary Data Table S1.

Table.

Diffuse hemispheric glioma with H3 p.K28M (K27M) mutation

Authors	Age/sex	Radiology	Histology	H3 K27M* IHC	H3 K27me3† IHC	Molecular alterations‡	Methylation profiling (MGMTp)	Treatment	Outcome (follow-up)
Lopez et al	20/F	Lt insular enhancing	Low-grade diffuse astrocytic glioma	Positive	Loss	H3-3A K27M§, ATRX, EPB41L2::BRAF	NA	NA	NA
Valerio et al	19/M	Lt temporal enhancing	Mitotically active diffuse astrocytic glioma	Positive	Loss	H3-3A K27M§, PIK3CA, TP53	DMG, H3 K27 (unmethylated)	RT+ CT (TMZ), everolimus	DOD (11 months)
Current study	34/F	Lt temporal enhancing	Mitotically active diffuse astrocytic glioma	NA	NA	H3-3A K27M§, PTPN11, PPM1D	NA	RT+CT (TMZ)	DOD (84 months)
			High-grade diffuse astrocytic glioma with necrosis	Positive	Loss	H3-3A K27M§, PTPN11, PPM1D	No match (unmethylated)	Lomustine, bevacizumab
Current study	25/M	Lt parietal enhancing	Mitotically active diffuse astrocytic glioma	Positive	Loss	H3-3A K27M§, TP53, ATRX, NF1, PPM1D, NOTCH1	DMG, H3 K27 (unmethylated)	RT+CT (TMZ)	Alive (14 months)

IHC, immunohistochemistry; MGMTp, MGMT promoter; M, male; F, Female; Lt, Left; NA, not available; DMG, diffuse midline glioma; RT, radiotherapy; CT, chemotherapy; TMZ, temozolomide; DOD, died of disease.

^*H3 p.K28M (K27M).

^†H3 p.K28me3 (K27me3).

^‡Reported clinically relevant sequence variants per AMP/ASCO Tiers I/II.

^§ H3-3A: c.83A>T p.K28M (K27M).

Histopathological, immunohistochemical, and radiological evaluation

All available formalin-fixed-paraffin-embedded (FFPE) hematoxylin and eosin (H&E) stained slides were originally reviewed by a board-certified anatomic pathologist (V.S. for case 1) and a board-certified neuropathologist (K.P.S. for case 2) and re-reviewed by 2 board-certified neuropathologists (K.D., C.M.I.) for morphological diagnostic confirmation. Immunohistochemical stains that were performed at the Mayo Clinic used validated protocols on the Ventana Benchmark Ultra platform (Ventana Medical Systems, Tucson, AZ) with antibodies against histone H3 (mutated K27M) (catalog#31-1175-00-L, clone RM192; 1:1000; RevMAb Biosciences, Burlingame, CA), Tri-Methyl-Histone H3 (K27) (clone C36B11; 1:50; Cell Signaling Technology, Danvers, MA), IDH1-R132H (clone H09; 1:50; Dianova, Hamburg, Germany), ATRX (clone#D-5: sc-55584; 1:1000; Santa Cruz, Dallas, TX), p53 (clone DO-7; no dilution; Ventana Medical Systems), and OLIG2 (clone EPR2673; 1:100; Abcam, Cambridge, United Kingdom). Except for IDH1-R132H immunostain, which shows a cytoplasmic staining pattern, all listed immunostains show a nuclear staining pattern. Nucleolar only staining was considered nonspecific. For all immunostains, a positive result was rendered if staining in expected pattern was observed in at least 25% tumor cells. For Tri-Methyl-Histone H3 (K27) and ATRX immunostains, a negative result was valid only in the presence of internal positive control in endothelial and nonneoplastic cells. All MRI images from both cases were thoroughly reviewed by a board-certified neuroradiologist (D.J.) to ensure absence of radiological evidence of midline involvement.

Molecular testing

The H3-3A (previously H3F3A):c.83A>T p.K28M (K27M) mutation was detected by custom NGS panels performed on genomic DNA extracted from FFPE tumor tissue with tumor purity visually estimated to be at least 60% in both cases. Case 1 was tested using the Mayo Clinic 187-gene amplicon targeted neurooncology NGS panel, which consists of a DNA-based 118-gene mutation subpanel and an RNA-based 81-gene fusion/transcript variant subpanel, as previously described (11). Case 2 was profiled using the University of Washington UW-OncoPlex (version 7), which is a DNA-based capture targeted NGS panel designed to detect mutations, copy number alterations, and select gene fusions involving 376 genes, as well as microsatellite instability and tumor mutation burden (12). Sequence variants were classified using the AMP/ASCO/CAP guidelines (13), and only Tiers I/II variants were considered clinically relevant and reported.

Methylation profiling

In both cases, methylation testing was performed at the Laboratory of Pathology, National Cancer Institute (NCI), as previously described (14). In brief, genomic DNA was extracted from FFPE tumor tissue using the AllPrep DNA/RNA FFPE Kit (Qiagen, Hilden, Germany). Genomic DNA was bisulfite converted by EZ DNA Methylation Kit (Zymo Research, Irvine, CA). Bisulfite converted DNA was processed with the Infinium FFPE DNA Restore kit (Illumina, Carlsbad, CA) and assayed on the Infinium MethylationEPIC kit v1.0 (Illumina), according to the Infinium HD FFPE Methylation Assay automated protocol (Illumina). Methylation data from our 2 cases as well as of a previously reported case (9) were processed using versions 11.b4 and 12.5 of the DKFZ/Heidelberg classifiers (10) using the NCI-Bethesda pipeline and version 2 of the NCI-Bethesda classifier. MGMT promoter methylation status was estimated using the MGMT-STP27 prediction model (15, 16). Copy number profiles were inferred using the R “conumee” package (http://bioconductor.org/packages/conumee/) as implemented in the classifier package. Copy number changes were visually estimated based on distinct gains and losses relative to baseline.

Cases

Case 1

A 34-year-old woman presented with new onset of headaches, memory loss, change in taste and smell, and confusion. Imaging studies revealed a 2.3-cm left temporal minimally enhancing tumor (Fig. 1A). Subtotal tumor resection showed a mitotically active diffusely infiltrative astrocytic glioma without necrosis or microvascular proliferation (Fig. 1C) that was reportedly IDH-wildtype based on targeted IDH1/IDH2 pyrosequencing. Postoperatively, the patient was treated with chemoradiation with temozolomide, followed by additional cycles of temozolomide. After 4 years, a new enhancing area along the surgical cavity gradually increased (Fig. 1B) and was subtotally resected 6 years after the initial surgery. Histologically, the recurrent tumor was a diffuse astrocytic glioma with areas of necrosis (Fig. 1D). By immunohistochemistry, the tumor cells were negative for IDH1-R132H and had preserved expression of ATRX and OLIG2. The Mayo Clinic NGS panel detected an H3-3A: c.83A>T p.K28M (K27M) mutation as well as PPM1D, PTPN11, and SOS1 mutations (Table and Supplementary Data Table S1). Given the unexpected NGS results, H3 p.K28M (K27M) and H3 p.K28me3 (K27me3) immunostains were performed and confirmed the presence of the H3 p.K28M (K27M) mutation (Fig. 1E) and loss of H3 p.K28me3 (K27me3) expression (Fig. 1F). Methylation profiling of the recurrent tumor (ISS-88) did not match in the version 2 of the NCI/Bethesda classifier but clustered within the diffuse midline glioma, H3 K27-altered group in the Uniform Manifold Approximation and Projection (UMAP) (Fig. 2), suggesting an epigenetic profile in keeping with DMG-H3 K27 but distinct from the methylation pattern of DMG-H3 K27 classifier reference class. In the DKFZ/Heidelberg classifier using the NCI/Bethesda pipeline, this case also did not match. NGS testing was retrospectively performed in the initial tumor and detected the H3-3A: c.83A>T p.K28M (K27M) mutation, identical PTPN11 mutation, and a different PPM1D mutation (Supplementary Data Table S1). The patient was treated with Lomustine, (which was discontinued after 3 months due to tumor progression), and bevacizumab therapy with partial improvement until clinical trial enrollment. Six months after enrollment, there was radiological evidence of tumor progression that was associated with a left basal ganglia acute ischemic stroke. Bevacizumab was reintroduced with treatment response but was discontinued after 3 months due decreased renal function. The patient died of disease 7 years after diagnosis.

Figure 1.

Axial and coronal reformatted images of 3D MRI: (A) mildly expansile hyperintense lesion centered in the medial left temporal lobe, with involvement of the uncus and amygdalohippocampal area (T2-weighted FLAIR); (B) 6 years later, after multiple lines of therapy, the tumor remained primarily within the left temporal lobe, with mild extension of the enhancement into the inferior left insula (postgadolinium T1-weighted). Histopathological findings: (C) initial tumor was a moderately cellular infiltrative astrocytic glioma with mitotic activity (circle) and without necrosis or microvascular proliferation (H&E, 200×); (D) recurrent tumor consisted of a diffuse astrocytic glioma with tumor necrosis (H&E, 100×); (E) H3 p.K28M (K27M) immunostain was diffusely positive whereas (F) H3 p.K28me3 (K27me3) immunostain was negative or considerably reduced in the nuclei of the tumor cells, consistent with the presence of an H3 p.K28M (K27M) mutation and loss of expression of H3 p.K28me3 (K27me3) (200×).

Figure 2.

Uniform Manifold Approximation and Projection (UMAP): case 1 (ISS-88) and case 2 (DH32) embedded within the diffuse midline glioma, H3 K27-altered cluster in the version 2 of the NCI/Bethesda classifier.

Case 2

A 25-year-old man presented with a 3-month history of right lower extremity numbness and new-onset seizures. Imaging studies revealed a 3.2 cm left parietal expansile T2/FLAIR hypertense tumor with central contrast enhancement (Fig. 3A, B). The patient underwent gross total resection which revealed a mitotically active diffuse astrocytic glioma with focal multinucleated tumor cells without necrosis or microvascular proliferation (Fig. 3C). By immunohistochemistry, the tumor cells were negative for IDH1-R132H and had loss of ATRX nuclear expression (Fig. 3D), with preserved OLIG2 expression. UW-Oncoplex targeted NGS panel detected H3-3A: c.83A>T p.K28M (K27M), TP53, ATRX, NF1, PPM1D, and NOTCH1 mutations (Table and Supplementary Data Table S1). Because of the unexpected NGS results, H3 p.K28M (K27M) and H3 p.K28me3 (K27me3) immunostains were performed and confirmed the presence of the H3 p.K28M (K27M) mutation (Fig. 3E) and loss of H3 p.K28me3 (K27me3) expression (Fig. 3F). By methylation profiling, this tumor (DH32) matched with high confidence to methylation class diffuse midline glioma, H3 K27-altered (calibrated score: 1.0) and clustered within this group in the version 2 of the NCI/Bethesda classifier (Fig. 2). In the DKFZ/Heidelberg classifier using the NCI/Bethesda pipeline, this case matched with high confidence to methylation class diffuse midline glioma, H3 K27-altered, subtype H3 K27-mutant, or EZHIP expressing (calibrated score: 0.99, version 12b6). Postoperatively, the patient received radiotherapy combined with temozolomide followed by adjuvant temozolomide. At 14 months after surgery, the patient is stable without clinical or radiological evidence of recurrent disease.

Figure 3.

Axial and coronal reformatted images of 3D MRI: (A) relatively well-circumscribed hyperintense lesion centered in the left superomedial parietal lobe, without extension to midline structures (T2-weighted FLAIR); (B) central contrast enhancement within the otherwise T1-hypointense mass (postgadolinium T1-weighted). Histopathological findings: (C) moderately cellular infiltrative astrocytic glioma with focal multinucleated tumor cells and mitotic activity (circles), and without necrosis or microvascular proliferation (H&E, 200×); (D) ATRX immunostain showed loss of expression in tumor cell nuclei with nonspecific nucleolar staining (200×); (E) H3 p.K28M (K27M) immunostain was diffusely positive whereas (F) H3 p.K28me3 (K27me3) immunostain was negative in the nuclei of the tumor cells, consistent with the presence of an H3 p.K28M (K27M) mutation and loss of expression of H3 p.K28me3 (K27me3) (200×).

Summary of DHG-H3 K27

The 2 cases had similar features to the 2 well-documented previously reported examples of DHG-H3 K27 (Table and Supplementary Data Table S1). Collectively, these unusual tumors occurred in patients within the AYA (adolescent and young adult) age range (median, 22.5 years; range, 19–34), in both sexes (F:M ratio, 1:1), and presented as cerebral hemispheric enhancing tumors. Histologically, the tumors were diffuse astrocytic gliomas without high-grade morphological features but often with mitotic activity and were immunohistochemically positive for H3 p.K28M (K27M) with loss of H3 p.K28me3 (K27me3) expression. The H3-3A: c.83A>T p.K28M (K27M) mutation frequently co-occurred with inactivating alterations involving p53 pathway genes (TP53, PPM1D) and ATRX, as well as with activating alterations in MAPK pathway genes (PTPN11, NF1, BRAF). The genomes were unstable with multiple copy number changes and recurrent loss of chromosome 21 (Supplementary Data Table S1). Epigenetically, DHG-H3 K27 clustered with and often matched to methylation class DMG-H3 K27; MGMT promoter was predicted to be unmethylated in all cases. All patients have been treated with chemoradiation with temozolomide. One patient had an aggressive clinical course; one experienced prolonged but progressive disease; and one had stable disease at 14 months after diagnosis. Among the other radiologically documented cases reported as non-midline, there were 2 cases confined to the corpus callosum; the sole case with follow-up data was a patient who received chemoradiation with temozolomide and died of disease 3 months after a stereotactic biopsy (Supplementary Data Table S1).

Frequency of DHG-H3 K27

Case 1 was the single case with an H3-3A: c.83A>T p.K28M (K27M) mutation among 1844 IDH-wildtype tumors with the reported diagnosis of diffuse glioma and non-midline location (<18 years, n = 55; 18–40 years, n = 193; 41–54 years, n = 362; >54 years, n = 1233) tested by the Mayo Clinic targeted neuro-oncology NGS panel (1/1844; 0.05%). Of these 1844 IDH-wildtype non-midline diffuse gliomas, 57 were DHG-H3 G34 (<18 years, n = 21; 18–40 years, n = 31; 41–54 years, n = 5; >54 years, n = 0). In patients 18–40 years of age, the frequency of DHG-H3 K27 was 0.5% (1/193) whereas DHG-H3 G34 accounted for 16% (31/193) of diffuse non-midline gliomas in this age group. All other 51 diffuse gliomas that had an H3-3A: c.83A>T p.K28M (K27M) mutation detected by this NGS panel were reportedly located at or partially involved midline structures.

DISCUSSION

The increasing use of multiomics in the diagnostic workup of CNS tumors has allowed identification of unusual tumors such as our 2 additional examples of DHG-H3 K27. Based on the few radiologically documented and genomically and epigenetically characterized cases, DHG-H3 K27 occurs in AYA patients, and presents as cerebral hemispheric diffuse astrocytic gliomas with overall mutational and epigenetic profile similar to DMG-H3 K27. As DMG-H3 K27 (3–5, 17, 18), DHG-H3 K27 shows loss of H3 p.K28me3 (K27me3) expression with frequent concomitant inactivating alterations that disrupt p53 pathway genes (TP53, PPM1D) and ATRX, and clusters with the DMG-H3 K27 methylation group (9). Of note, DHG-H3 K27 had activating alterations involving MAPK pathway genes (PTPN11, NF1, BRAF) in most cases, at a higher frequency than DMG-H3 K27 (19–21). Taken together, these data suggest that DHG-H3 K27 is closely related to DMG-H3 K27 and may represent an unusual presentation of DMG-H3 K27 with a non-midline epicenter and enrichment for MAPK pathway activation. DMG-H3 K27 can be extensively infiltrating with extension to cerebral hemisphere (7, 22, 23), and seems to more frequently involve multiple sites in adults than in pediatric patients (24). It is unknown whether DHG-H3 K27 harbors radiologically undetectable microscopic involvement of midline structures. DMG-H3 K27 shows signatures that resemble oligodendrocyte precursor cells (25–27), and developing hindbrain epigenetic programs (28). Additional studies are needed to document definitive absence of midline involvement and investigate the underlying biological mechanisms associated with a primarily non-midline tumor growth in DHG-H3 K27.

Available outcome data for DHG-H3 K27 indicate a progressive clinical course despite chemoradiation. Although an aggressive outcome as seen in DMG-H3 K27 has been described (9), our 2 cases have shown longer disease courses. Case 1 had a 7-year clinical course for a subtotally resected tumor. Follow-up for case 2 is limited but so far relatively favorable as this tumor was gross totally resected and the patient has been clinically stable after chemoradiation at 14 months postoperatively. Notably, despite recent conflicting results (20), there are data indicating that mutations in MAPK pathway genes (including PTPN11 and NF1) are associated with long-term survival for DMG-H3 K27M patients and that long-term and short-term DMG-H3 K27M survivors form distinct DNA methylation clusters (19). These findings may relate to the relatively favorable outcomes of our 2 cases as they both had mutations in MAPK pathway genes as well as to the aggressive outcome in the previously reported case that lacked any MAPK pathway gene alteration (9). The more amenable location for surgical resection and the young age that is typically associated with good performance status to tolerate multimodal therapy regiments may also be related to the clinical course of the patients reported in this study. Reporting of cases with radiologic, genomic, and epigenetic characterization and clinical outcome/follow-up is encouraged to expand our understanding and allow better categorization of DHG-H3 K27 in future WHO classification schemes.

Although histone H3 K28M (K27M) testing is currently recommended only for diffuse gliomas that involve the midline (29, 30), routine multigene testing including H3-3A has been endorsed for pediatric and your adult patients with high-grade gliomas (29). Thus, like the cases reported in this study, DHG-H3 K27 would be unexpectedly identified through genetic testing that includes the H3-3A mutational hotspot codons K28 (K27) and G35 (G34). This is a cost-effective testing strategy as DHG-H3 K27 seems rare. Notably, a few examples of diffuse gliomas limited to the corpus callosum and harboring an H3-3A: c.83A>T p.K28M (K27M) mutation with accompanying loss of H3 p.K28me3 (K27me3) expression have been reported (7, 31). The corpus callosum is not considered a classic midline structure even though it is located within the midline plane. This recurrent finding raises consideration for whether corpus callosum may be included within the group of midline structures fulfilling the midline location essential diagnostic criterion for DMG-H3 K27. The promising results of a novel therapy for DMG-H3 K27 patients with reversion of H3 p.K28me3 (K27me3) reduction (32) underscores the importance of precision diagnostics with detection of H3 p.K28M (K27M) mutation as patients with DHG-H3 K27 and with a corpus callosum diffuse glioma with an H3-3A: c.83A>T p.K28M (K27M) mutation may also benefit from such therapy.

In summary, DHG-H3 K27 is an unusual group of diffuse gliomas that shows similar clinical, histopathological, genomic, and epigenetic features to DMG-H3 K27, suggesting that DHG-H3 K27 is closely related to DMG-H3 K27 and may represent an unusual presentation of DMG-H3 K27 without obvious midline involvement and with frequent MAPK pathway activation. As additional examples are identified with the expansion of routine multiomics testing of diffuse gliomas, collaborative efforts to compile cases with clinical follow-up are needed to better define the underlying genomic/epigenetic profile, clinical behavior, and classification of DHG-H3 K27.

CONFLICT OF INTEREST

The authors have no duality or conflicts of interest to declare.

SUPPLEMENTARY DATA

Supplementary Data can be found at academic.oup.com/jnen.