Dipg-37. transcriptional, epigenetic, and immunological features of paediatric-type h3 k27-altered diffuse midline gliomas originating from different brain locations

Abstract Diffuse midline gliomas (DMG) are highly aggressive brain tumours predominantly affecting young children. The main oncogenic driver is the substitution of lysine 27 to methionine (K27M) in histone 3 (H3) or overexpression of EZHIP resulting in global loss of H3K27me3. Recent single-cell (epi-)genomic studies have identified developmentally-restricted origins and cell states associated with tumour location and patient age, and we aimed to further explore the underlying biology of distinct subgroups of DMG using orthogonal data modalities. A retrospective reanalysis identified 242 patient genome-wide DNA methylation array profiles with a methylation classification of DMG_K27. H3 status was known for 168 cases (H3.3 K27M n=130, H3.1 K27M n=20, H3 WT n=18), with EZHIP overexpression or loss of H3K27me3 was confirmed in 13/18 H3 WT cases. Unsupervised clustering identified two distinct methylation subtypes, specific to brainstem (62/64) or thalamic (25/29) location, and highly enriched for gain of chromosome 1q or loss of 5q, respectively. Differential expression analysis from bulk RNA-Seq data recapitulated the recently-published overexpression of ZIC1/4 early development genes in thalamic tumours and an enrichment of HOX-cluster genes in those of brainstem origin. Within DMGs of the thalamus, we further observed a significant overexpression and hypomethylation of NR2E1, a regulator of bivalent promoters via LSD1-H3K4, and essential for stem cell renewal; and CD96, an immune checkpoint blockade target in NK/T-cells. We confirmed differential immune response expression between tumour location by GSEA as well as deconvolution by methylation CIBERSORT from bulk samples, which we were able to characterise further by by single-cell RNA-sequencing. We report unique copy-number and differentially methylated loci as well as differential expression that recapitulates previously reported single-cell metaprogrammes. We also highlight involvement of neural stem cell development programme in thalamus-type tumours as well as evidence of these tumours as being potential targets for immune therapeutic targeting.