MB-87INTEGRATED GENOMICS REVEALS NOVEL SUBTYPES OF MEDULLOBLASTOMA SUBGROUPS

Abstract

Background: Medulloblastoma is now accepted to comprise four distinct molecular variants, and current clinical trials are stratifying patients using a combined biological and clinical risk stratification. Despite the identi-fication of the 4 core subgroups, there appears to exist tremendous clinical heterogeneity within the four subgroups suggesting additional substructure. Methods: Integrative clustering of 763 primary medulloblastoma samples with gene expression and genome wide methylation data was performed with the Similarity Network fusion Method (SNF), and correlated with clinical features and copy-number aberrations. Results: Integrative clustering faithfully recapitulated the four principal subgroups of medulloblastoma, with a boundary between Group 3 and 4, and intra-subgroup biological heterogeneity more clearly apparent than either expression or methylation alone. The WNT subgroup consists of two subtypes, one defined by monosomy 6 and a second of older patients without monosomy 6.Wefound the highest evidence for four subtypes of SHH, 1) two infants subgroups with clear prognostic differences, pathway aberrations and copy number profiles, 2) childhood group with a poor prognosis and 3) an adult group. MYC amplifications enrich in a distinct cluster of Group 3 with a significantly poor prognosis. Copy number profiles and driver pathways define three subtypes of Group 4. Conclusions: Integrative clustering provides profound insights into the biological heterogeneity within each of the principle medulloblastoma subgroups. As current therapies result in significant longterm sequelae, the identification of substructure within the four subgroups allows for more refinement in biological risk stratification as well as identification of novel agents for future rationale targeted therapies.