Integrative genomic analysis of medulloblastoma identifies a molecular subgroup that drives poor clinical outcome

Abstract

Purpose: Medulloblastomas are heterogeneous tumors that collectively represent the most common malignant brain tumor in children. To understand the molecular characteristics underlying their heterogeneity and to identify whether such characteristics represent risk factors for patients with this disease, we performed an integrated genomic analysis of a large series of primary tumors.

Patients and methods: We profiled the mRNA transcriptome of 194 medulloblastomas and performed high-density single nucleotide polymorphism array and miRNA analysis on 115 and 98 of these, respectively. Non-negative matrix factorization-based clustering of mRNA expression data was used to identify molecular subgroups of medulloblastoma; DNA copy number, miRNA profiles, and clinical outcomes were analyzed for each. We additionally validated our findings in three previously published independent medulloblastoma data sets.

Results: Identified are six molecular subgroups of medulloblastoma, each with a unique combination of numerical and structural chromosomal aberrations that globally influence mRNA and miRNA expression. We reveal the relative contribution of each subgroup to clinical outcome as a whole and show that a previously unidentified molecular subgroup, characterized genetically by c-MYC copy number gains and transcriptionally by enrichment of photoreceptor pathways and increased miR-183∼96∼182 expression, is associated with significantly lower rates of event-free and overall survivals.

Conclusion: Our results detail the complex genomic heterogeneity of medulloblastomas and identify a previously unrecognized molecular subgroup with poor clinical outcome for which more effective therapeutic strategies should be developed.

Fig 1.

(A) Non-negative matrix factorization consensus clustering of mRNA expression array data from 194 primary medulloblastomas and nine atypical teratoid/rhabdoid tumors (ATRTs) reveals seven (k = 7) stable subgroups (six medulloblastoma subgroups, c1 through c6, plus an additional ATRT subgroup). (B) Silhouette widths indicate a strong similarity of samples to others within their subgroup relative to samples from other subgroups. (C) Heat map of the top 25 upregulated genes for each subgroup shows significant overlap between the c2 and c4 subgroups, which are characterized by enrichment of neuronal/glutamatergic signatures, and overlap to a lesser extent between c1 and c5 subgroups, which both have enrichment of photoreceptor transcriptional signatures; normal cerebellum samples are included to assess for the degree of stromal contamination. coph coeff, cophenetic coefficient; WNT, Wingless signaling pathway; SHH, Sonic Hedgehog signaling pathway.

Fig 2.

Copy number (CN) profiles of samples arranged by non-negative matrix factorization subgroup reveals statistically significant CN alterations associated with each. Detailed files of significant CN lesions within each subgroup, as determined by genomic identification of significant targets in cancer (GISTIC) analysis, are provided in the Data Supplement. ch, chromosome.

Fig A1.

miRNA expression profiles of samples arranged by non-negative matrix factorization subgroup reveal statistically significant enrichment of the miR-17∼92 cluster and miR-21 in all tumors relative to normal cerebellum; miR-183∼96∼182 in c1, c5, and c4 tumors; miR-23∼27∼24 in c6 (WNT pathway) tumors; miR-592 in c2 and c4 tumors; and miR-199a/b in c3 (SHH) tumors. P value less than .001 for each miRNA-subgroup association.

Fig A2.

(A-F) Kaplan-Meier survival analysis of patients in each non-negative matrix factorization subgroup reveals (A) decreased survival (event-free survival [EFS] and overall survival [OS]) of patients having c1 subgroup tumors relative to all other tumors (REST). (F) Increased OS is noted for c6 subgroup tumors. Survival analyses were based on clinical risk, histologic subtype, and M stage and are provided as figures in the Data Supplement.

Fig A2.

(A-F) Kaplan-Meier survival analysis of patients in each non-negative matrix factorization subgroup reveals (A) decreased survival (event-free survival [EFS] and overall survival [OS]) of patients having c1 subgroup tumors relative to all other tumors (REST). (F) Increased OS is noted for c6 subgroup tumors. Survival analyses were based on clinical risk, histologic subtype, and M stage and are provided as figures in the Data Supplement.

Fig A3.

Validation of molecular subgroups in three independent data sets. Samples were assigned to corresponding non-negative matrix factorization (NMF) subgroups by using metagene projection. (A) Heat map of the top 25 upregulated genes for each NMF subgroup in the Thompson et al data set and (B) a merged data set containing the Kool et al and Fattet et al expression array data. (C) Kaplan-Meier survival analysis of patients from data sets by Thompson et al, Kool et al, and Fattet et al confirms poor overall survival associated with the c1 NMF subgroup. REST, all other tumors.

Fig A3.

Validation of molecular subgroups in three independent data sets. Samples were assigned to corresponding non-negative matrix factorization (NMF) subgroups by using metagene projection. (A) Heat map of the top 25 upregulated genes for each NMF subgroup in the Thompson et al data set and (B) a merged data set containing the Kool et al and Fattet et al expression array data. (C) Kaplan-Meier survival analysis of patients from data sets by Thompson et al, Kool et al, and Fattet et al confirms poor overall survival associated with the c1 NMF subgroup. REST, all other tumors.