Combined Genomic and Epigenomic Analysis Improves Classification of Hemispheric Pediatric High-grade Gliomas

Introduction: Midline pediatric high-grade gliomas (pHGGs) are classified based on histone H3 mutations. However, subgrouping of hemispheric pediatric high-grade glioma mutations are less well characterized. The objective of this study was to determine whether combining genomic and methylomic profiling of predominantly hemispheric pHGGs may improve classification using a cohort from a single institution.

Methods: Pediatric patients with a diagnosis of glioblastoma from 1996-2015 under the age of 19 years old were queried. Sixty-six paraffin embedded specimens were collected from 53 patients with a mean age of 10.2 years old. Whole exome sequencing of tumor DNA was done on an Illumina-HiSeq4000. DNA was also prepared using bisulfite conversion for hybridization to Illumina MethylationEPIC microarray (850,000 CpGs). Sequencing files underwent bioinformatic analyses using the Genome Analysis Toolkit (53 samples underwent mutational analysis). Differentially methylated regions were identified using minfi. Uniform Manifold Approximation and Projection (UMAP) was applied to reduce the dimensionality of the methylation data, and HDBSCAN clustering was done for all 66 samples.

Results: The 10,000 most informative methylation probes were calculated based on beta-value variance across samples. Unbiased UMAP and t-SNE clustering revealed distinct sub-grouping and discerned a change in methylation profile at recurrence for certain patients with multiple timepoints. Alterations in genes encoding histone H3.3 and H3.1 (H3F3A, HIST1H3B, HIST1H3C) were found in 42% of samples (22/53 samples). Mutations in SETD2, KDM6B, EGFR, BCOR, ASXL1, ATM, TP53, TOP3A, TERT, PDGFRA, and NF1 were also detected in >40% of samples.

Conclusion : Comprehensive genomic and methylomic profiling reveals unique sub-groups of hemispheric pHGG and can monitor changes in a tumor over time. To our knowledge this is the first study to perform a combined genomic and epigenomic analysis on a large single-institution predominantly-hemispheric pHGG cohort including several patients with serial resections.