Genomic classification and analysis of epilepsy-associated glioneuronal tumours

• Main Author: Stone, Thomas John
• Other Authors: Jacques, T. S. ; Ham, J.
• Published: University College London (University of London) 2017
• Subjects: 618.92
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.747011

INTRODUCTION: Glioneuronal tumours are a group of low-grade epilepsy-associated tumours with marked variability in their histological features, resulting in a lack of diagnostic consensus between institutions. This is confounded by a dearth of knowledge regarding their underlying biology, and subsequent lack of robust biologically informed diagnostic tools. This lack of understanding also impedes the development of novel and targeted treatment strategies. METHODS: I have undertaken a comprehensive molecular analysis of the most prevalent glioneuronal tumours: ganglioglioma and dysembryoplastic neuroepithelial tumours. I have used RNA sequencing and Illumina 450K methylation arrays to classify tumours in an unsupervised manner according to their genomic profiles. I then carried out in silico analyses on these datasets to identify genes, gene networks, and pathways that are differentially regulated between groups. Additionally, I have undertaken molecular assays to identify mutations that are specific to each group. Finally, I have used immunohistochemistry to assess a number of potential diagnostic markers revealed by expression profiling. RESULTS: Unsupervised clustering revealed glioneuronal tumours classify into two molecular groups (termed Group 1 and Group 2), which are only partially consistent with histological classification. Group 1 is defined by an astrocytic expression phenotype and an enrichment for BRAF-V600E mutations. Group 2 is defined by an oligodendrocyte precursor phenotype and an enrichment for FGFR1 mutations. A number of disease relevant networks and pathways are differentially regulated between these groups. Additionally, immunohistochemistry against Cyclin-D1 and PDGFRα can be used to distinguish tumour groups from one another. CONCLUSION: This is the first comprehensive genomic investigation of a large cohort of glioneuronal tumours without prior histological bias. I present data suggesting the current histological classification of these lesions is insufficient, and recommend a novel biologically informed strategy. My results also provide insight into the pathways underlying the development of these tumours. This information may assist in the development of novel treatment strategies.