Abstract 4347: Medulloblastoma comprises four distinct diseases

Prior attempts to subgroup medulloblastoma (MB) using genomics have identified 4-6 distinct molecular subtypes, including two subgroups driven by activated Wnt and Shh signaling. We performed an integrative analysis on a cohort of 103 primary MBs using a combination of Affymetrix expression and SNP genotyping arrays to determine how many subgroups of the disease exist, how they differ, and the extent of overlap between subgroups. Both unsupervised hierarchical clustering and principal component analysis of expression data reveals very high confidence for the existence of four medulloblastoma subgroups: WNT, SHH, Group C, and Group D. Additional bioinformatic analyses using non-negative matrix factorization (NMF) and subclass mapping (SubMap) further strengthens the support for the four subgroups. These subgroups exhibit distinct demographics: SHH tumors occur in infants and adults, Group C tumors are restricted to children, and WNT and Group D tumors are found across all age groups. While the sex ratio for the entire cohort is ∼1.5:1 (M:F), the sex ratio for WNT group tumors was ∼1:3. We identified a number of previously uncharacterized, subgroup-specific regions of chromosomal abnormality including 9p, 3q, 20q, and 21q gain in SHH tumors, 1q gain and 5q, 16q, and 10q loss in Group C tumors, and near universal isochromosome 17q and frequent loss of the X chromosome in Group D tumors. These regions likely harbor subgroup-specific oncogenes and tumor suppressor genes that could be targets for rationale therapy. Our demographic, transcriptional, and genetic data support the non-overlapping character of these four subgroups of MB. We identified ‘signature’ genes over-expressed in each subgroup for which there are high quality commercial antibodies available. Staining two separate MB tissue microarrays containing 294 non-overlapping tumors for DKK1 (WNT), SFRP1 (SHH), NPR3 (Group C), and KCNA1 (Group D) demonstrated that 288/294 (98%) tumors stained positive for only a single marker. Analysis of the demographics in these patients validated the results observed in our discovery series studied at the RNA level. Leptomeningeal dissemination was highly over-represented in Group C (47%) followed by Group D (30%) patients. A multivariate analysis that included age, extent of resection, histology, M stage, and subgroup revealed that only LCA histology and Group C affiliation were prognostic. As M0 Group C tumors have a very poor prognosis, we suggest that Group C patients include many of the children with ‘average-risk’ MB who relapse after current therapies. Our data highly support the existence of four independent subtypes of MB that differ in their demographics, transcription, genetic events, rate of metastases, and clinical outcome. Our novel ‘4 antibody’ technique is capable of determining MB subgroup through immunohistochemistry on formalin-fixed, paraffin-embedded material suggesting that it will be broadly generalizable across the globe. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 4347.