Abstract
INTRODUCTION: Recurrent medulloblastoma (MB) confers an abysmal prognosis with ~10% 5-year overall survival. Optimal treatment paradigms for relapsed disease are largely unknown. Conservation of molecular subgroup at relapse has been described and divergent clonal evolution implicated, yet multi-dimensional molecular characterization of larger cohorts are warranted to substantiate these findings and to disclose potential mechanisms underlying treatment failure and disease recurrence. METHODS: A multi-institutional series of 85 patient-matched, primary MBs and their relapses was profiled by DNA methylation array (Illumina 450K/850K). Entity and molecular subgroup classifications were assigned using random forest tumor classifiers (Molecular Neuropathology v.11b4). Genome-wide copy-number aberrations were also inferred from these data while parallel next-generation (whole-exome or targeted panel) sequencing on the majority of the cohort facilitated inference of somatic driver mutations. RESULTS: Comprised of WNT (1%), SHH (41%), Group 3 (22%), Group 4 (35%), primary tumors largely retained subgroup affiliation at relapse with the notable exception of 14% of cases. The majority (8/12) of discrepant classifications were determined to be secondary glioblastomas, while three Group 4 primary tumors relapsed as Group 3. Amongst conserved pairs, copy-number analyses suggest somatic clonal divergence between primary MBs and their respective relapses with an average of 0.8 (range 0–5) primary-specific and 2.0 (range 0–11) relapse-specific cytogenetic alterations. Despite high global methylation correlations between primary/relapse pairs (median 0.95), epigenetic evolution at relapse is suggested by the lower degree of conservation amongst promoter-associated probes (median 0.83, p < 2.2e-16). CONCLUSION: Secondary malignancy, particularly glioblastoma, may masquerade as relapsed MB, highlighting the utility of molecular characterization for relapsed disease and necessity for vigilant clinical surveillance. By deciphering the evolution of MB from diagnosis to relapse, a fundamental understanding of disease pathogenesis may be garnered, motivating rational and targeted clinical interventions in early therapy and at the time of relapse.