Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.566964
Title: Molecular subclassification of medulloblastoma and its utility for disease prognostication
Author: Schwalbe, Edward Carl
Awarding Body: University of Newcastle Upon Tyne
Current Institution: University of Newcastle upon Tyne
Date of Award: 2012
Availability of Full Text:
Access through EThOS:
Access through Institution:
Abstract:
Medulloblastoma is the most common malignant brain tumour of childhood. Transcriptomic classification of the disease has indicated the existence of discrete molecular subgroups of medulloblastoma, although the precise number, nature and clinical significance of these subgroups remains unclear. Two groups, characterised by activation of the WNT and SHH signalling pathways, are common to all published studies. An assay for the rapid diagnosis of medulloblastoma subgroups was therefore designed, using transcriptomic gene signatures of pathway activation for the WNT and SHH signalling pathways. The successful validation of these gene signatures in vitro and in silico enabled a meta-analysis of 173 new and published cases to be performed, which defined the molecular and clinico-pathological correlates of the disease subgroups more precisely. WNT subgroup cases were associated with CTNNB1 mutation, chromosome 6 loss and classic histology and were diagnosed > 5 years of age. SHH cases predominated in infants and showed an age-dependent relationship to desmoplastic / nodular histology. WNT / SHH independent tumours showed all histologies, peaked at 3 to 6 years and were associated with chromosome 17p loss. A novel DNA methylation array-based approach was next applied to disease subclassification. Using consensus clustering, based on non-negative matrix factorisation, four methylomic subgroups were identified in a training cohort (n = 100), which were robustly validated in a test cohort (n = 130). The subgroups were characterised by significant relationships to specific clinico-pathological and molecular markers. Two subgroups were characterised by activation of the WNT and SHH signalling pathways and showed equivalent clinico-pathological and molecular characteristics to the previously defined transcriptomic subgroups. For the WNT / SHH independent subgroups, group I was associated with a loss of chromosome 17p, whereas group II was enriched for large cell / anaplastic (LCA) histology. The WNT subgroup was associated with a favourable prognosis, while no survival differences were apparent between the remaining subgroups (SHH, group I, group II). Specific methylation biomarkers were identified for the discrimination of all subgroups. Assays of DNA methylation status were robust in derivatives of FFPE tissues, enabling testing in routinely-collected clinical material. Finally, the prognostic potential of methylomic biomarkers was investigated in a large clinical trials-based cohort (n = 191), with particular focus on the non-WNT subgroups (n = 163), where subgroup membership was not prognostic. Using the Cox Boost algorithm, which adds high dimensional data to mandatory clinical covariates to form cross-validated prognostic Cox survival models, the methylation status of MXI1 and IL8 were each identified as independent prognostic markers. These were incorporated into a novel risk stratification scheme, based on the cumulative assessment of disease risk using clinical (metastatic disease; poor prognosis), pathological (LCA pathology, poor prognosis) and methylomic variables (WNT subgroup, favourable prognosis; MXI1 and IL8 status). Importantly, this scheme assigns 46% of cases to a low risk group of patients (>90% survival) who could potentially be treated less intensively, with the aim of reducing therapy-associated late effects. This model out-performed the current clinical and other state-of-the-art medulloblastoma risk classification schemes. These data provide clear precedent for the utility of DNA methylation biomarkers for disease subclassification and prognostication in medulloblastoma, and their clinical application in diagnostic tumour biopsies.
Supervisor: Not available Sponsor: Katie Trust
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.566964  DOI: Not available
Share: