Abstract Current integrated genomic approaches indicate distinct biological variants in medulloblastoma. Comprehensive molecular classification strategies utilize cytogenetic or immunohistochemical biomarkers to refine risk stratification. Novel complementary markers may ameliorate outcome prediction particularly in intermediate or high-risk medulloblastomas. We combined transcriptome and DNA copy-number analysis for 64 primary tumors. Bioinformatic tools were applied to investigate marker genes of molecular variants. Differentially expressed transcripts were evaluated for prognostic value in the entire screening cohort. Immunohistochemical markers were used to determine molecular subtypes in adult and pediatric medulloblastoma samples (n=235). Immunopositivity of FSTL5 was correlated with molecular and prognostic subgroups for 235 non-overlapping medulloblastoma samples on two independent tissue microarrays (TMA). Unsupervised cluster analyses of transcriptome profiles revealed four distinct molecular variants: WNT, SHH, Group C, and Group D. Stable subgroup separation was achieved using only 300 most varying transcripts. Specific distribution of clinical and molecular characteristics was noted for each cluster. Notably, Group C tumors were exclusively present in pediatric medulloblastomas as determined by immunohistochemistry. Delimited expression patterns of FSTL5 in each molecular subgroup were confirmed by quantitative real-time PCR. FSTL5 transcripts were most up-regulated in Group C and Group D tumors with unfavorable prognosis, whereas WNT medulloblastomas showed marked down-regulation. Immunopositivity of FSTL5 identified a large proportion of patients (84 of 235 patients; 36%) at high risk for relapse and death in particular in patients with WNT/SHH-independent tumors. Multivariate analysis revealed that FSTL5 immunopositivity constitutes an independent prognostic marker in pediatric and adult patient cohorts (p<0.0001). Importantly, adding this biomarker to comprehensive outcome prediction schemes substantially reduced the prediction error of the model. Comprehensive analyses of transcriptome and genetic alterations unravel four distinct disease variants. By addition of FSTL5 immunohistochemistry, existing molecular stratification schemes can effectively be complemented and sub-classification of WNT/SHH-independent tumors substantially optimized. This approach may ultimately define clear risk groups to individualize treatment intensities in future clinical trials. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 3447. doi:10.1158/1538-7445.AM2011-3447
Read full abstract