EMBR-25. GENOME-WIDE GENETIC AND EPIGENETIC ASSESSMENT OF GROUP 4 MEDULLOBLASTOMA FOR IMPROVED, BIOMARKER DRIVEN, PROGNOSTICATION AND RISK-STRATIFICATION

INTRODUCTION: Medulloblastoma (MB) is the most common malignant brain tumour in children. The most frequent molecular subgroup, Group 4 (MB(Grp4)) accounts for ~35/40% of cases, however it has the least understood underlying biology. Clinical outcomes are heterogeneous in MB(Grp4) and are not accounted for by established clinico-pathological risk factors. There is now a requirement for a comprehensive study of MB(Grp4), considering established clinico-pathological features and novel molecular biomarkers to enhance risk-stratification and identify novel therapeutic targets. METHODS: A clinically-annotated, retrospective MB(Grp4) discovery cohort (n = 420) was generated from UK CCLG institutions, collaborating European centres and SIOP-UKCCSG-PNET3 and HIT-SIOP-PNET4 clinical trials. Contemporary, multi-omics profiling was performed. Focal and arm level copy number aberrations (CNAs) were determined from molecular inversion probe (MIP) or DNA methylation array which additionally provided next generation non-WNT/non-SHH (Grp3/Grp4) subtype classifications. Targeted next-generation DNA sequencing was performed to overlay the mutational landscape. Survival modelling was carried out with patients >3 years old who received craniospinal irradiation. RESULTS: MB(Grp4) subtypes were assigned to 88% of tumours with available data. Subtype VIII was strongly associated with i17q (p<0.0001). The favourable-risk cytogenetic signature (2 or 3 of; chromosome 7 gain, chromosome 8 loss and/or chromosome 11 loss) associated with both subtypes VI and VII (p<0.0001). MYCN amplifications were strongly associated with subtype V (p<0.0001) in addition to 16q loss (p<0.0001). The high-risk CNA group was enriched for mutations in genes involved in chromatin remodelling (p<0.0001). Risk factors were identified from multivariate survival modelling. Subtype and CNA groups contributed to improved risk-stratification models that outperformed current clinical schemes. CONCLUSION: Comprehensive genetic and epigenetic profiling in this large retrospective cohort has improved our understanding of the molecular and clinical heterogeneity within MB(Grp4). Incorporation of molecular biomarkers improved risk-stratification for MB(Grp4).