Molecular subclassification of medulloblastoma and its utility for disease prognostication

摘要: Medulloblastoma is the most common malignant brain tumour of childhood. Transcriptomic classification disease has indicated existence discrete molecular subgroups medulloblastoma, although precise number, nature and clinical significance these remains unclear. Two groups, characterised by activation WNT SHH signalling pathways, are to all published studies. An assay for rapid diagnosis medulloblastoma was therefore designed, using transcriptomic gene signatures pathway pathways. The successful validation in vitro silico enabled a meta-analysis 173 new cases be performed, which defined clinico-pathological correlates more precisely. subgroup were associated with CTNNB1 mutation, chromosome 6 loss classic histology diagnosed > 5 years age. predominated infants showed an age-dependent relationship desmoplastic / nodular histology. independent tumours histologies, peaked at 3 17p loss. A novel DNA methylation array-based approach next applied subclassification. Using consensus clustering, based on non-negative matrix factorisation, four methylomic identified training cohort (n = 100), robustly validated test 130). significant relationships specific markers. pathways equivalent characteristics previously subgroups. For subgroups, group I 17p, whereas II enriched large cell anaplastic (LCA) favourable prognosis, while no survival differences apparent between remaining (SHH, I, II). Specific biomarkers discrimination Assays status robust derivatives FFPE tissues, enabling testing routinely-collected material. Finally, prognostic potential investigated trials-based 191), particular focus non-WNT 163), where membership not prognostic. Cox Boost algorithm, adds high dimensional data mandatory covariates form cross-validated models, MXI1 IL8 each as These incorporated into risk stratification scheme, cumulative assessment (metastatic disease; poor prognosis), pathological (LCA…