Multiple approaches converge on three biological subtypes of meningioma and extract new insights from published studies [Methylation array]

ABSTRACT: Purpose: Clustering of meningiomas using DNA methylation and RNA-sequencing data yields groups which predict clinical behavior than the current clinical standard of histopathologic grading. Both techniques segregate many common genomic features seen in meningiomas similarly, raising the question of whether they were identifying the same underyling groups. Methods: DNA methylation profiling (EPIC 850K array) and RNA-sequencing of 110 primary meningiomas was performed. Unsupervised clustering was performed using each type of data, followed by computational modeling which explored correlations of promoter methylation and copy number variation (CNV) with gene expression. Results: We performed unsupervised clustering of the DNA methylation data and gound three clusters to be optimal. These clusters aligned closely with our prior transcriptional classification (Patel et al. PNAS 2019). These groups also closely aligned with groups defined by large-scale cytogenetic changes and merlin expression/genomic instability, suggesting the presence of three biologic groups of meningioma that can be identified by multiple methods. Computational modeling demonstrated that both promoter methylation and CNV correlated closely with gene expression differences seen in meningioma and our biologic groups, although cytogenetic changes (particularly chr 1p loss) was predominant in the clinically aggressive tumors. Conclusions: Our analysis suggests the presence of three biologic groups of meningioma (MenG A, B, and C) which can be accurately identified through DNA methylation, RNA-seq, and cytogenetic profiling.

ORGANISM(S): Homo sapiens

PROVIDER: GSE189521 | GEO | 2021/11/30

REPOSITORIES: GEO