Understanding the molecular complexity of medulloblastoma

Abstract

The understanding and treatment of medulloblastoma, the most common childhood malignant brain tumour, is rapidly evolving. Three complementary deep-sequencing studies that were recently published in Nature add to our knowledge of this disease, further refine risk stratification, and identify potential druggable targets.

Despite recent studies that have demonstrated an improvement in the survival of children with medulloblastoma, the most common childhood malignant brain tumour, one-third of patients will still succumb to this disease, and survivors often have substantial treatment-related life-altering and life-limiting sequelae.^1–3 Newer molecularly targeted therapies have been sought, but their identification has been limited by a lack of understanding of the biology of medulloblastomas.

Recently, primarily due to the work of research collectives investigating relatively large numbers of specimens of this comparatively rare tumour, substantive insights have been gained into the molecu lar genetics of medulloblastoma. These efforts have resulted in a new proposed risk stratification of the disease into four mini mally overlapping, molecularly based sub groups with distinct clinical, biological and gen etic profiles: WNT tumours (activated wingless pathway signalling) with CTNNB1 point mutations, chromosome 6 loss and a favourable prognosis (Group 1); sonic hedgehog (SHH) tumours demon strating SHH pathway mutations, somatic loss of chromosome 9 and an excellent prognosis in very young chil dren (Group 2); Group 3 tumours, which are frequently disseminated at diagnosis, exhibit MYC amplification, and have a poor prognosis; and Group 4 tumours, which exhibit MYCN and CDKG amplification and have an inter mediate prognosis.^4,5 Three papers recently published in Nature further deline ate the subgroups, elucidate the pathogenic mecha nisms of genetic alterations, and identify potential druggable targets.

Pugh and colleagues utilized whole-exome hybrid capture and deep sequencing to identify somatic mutations across the coding regions of 92 medulloblastoma–normal pairs.⁶ The Pugh et al. paper confirmed many of the previous genetic abnormalities found in the four proposed subtypes. Medullo blastomas had a relatively low load of somatic mutations compared with other cancer types, and the highest load was seen in the tumours from the oldest patients. As previously reported by Parsons et al.,⁷ MLL2, which encodes a histone-lysine N-methyltransferase, was found to host recur ring activating mutations, thereby substantiating the evidence for dysregulated histone modification in some medulloblastomas. Mutations in histone-modifying genes such as MLL3 and HDAC2 were found in Group 4 tumours. DDX3X, which encodes an RNA helicase involved in transcription and RNA splicing, transport and translation, was found in both WNT and SHH tumours (Groups 1 and 2). This gene is thought to code for a component of pathogenic WNT– β-catenin signalling. Novel findings of this study were mutations in several genes encoding components of the nuclear co-repressor (N-CoR) complex, especially in SHH-driven tumours. Novel mutations in another gene, CTDNEP1, which encodes a component of the mammalian target of rapamycin (mTOR) complex, were found in Group 3 tumours with isochromosome 17q.

Jones and colleagues deep sequenced 125 medulloblastoma–normal pairs as part of the International Cancer Genome Consortium PedBrain Tumor project, and integrated their findings with RNA-seq data.⁸ These authors confirmed the relatively low somatic mutation load within medulloblastomas as a whole, compared with other malignant solid tumours. They found that tetraploidy was most commonly observed in Group 3 and Group 4 tumours, and that tetraploid SHH tumours tended to harbour mutations in the tumour suppressor gene TP53. The authors concluded that tetraploidy followed by genomic instability may be an early driving event in the Group 3 and 4 tumours. They also suggested that tumours arising in older patients may derive from more-differentiated cells and require a greater number of alterations to undergo malignant transformation.

The RNA-seq data permitted greater sensi tivity for identification of gene fusion events, and genes previously associated with medulloblastoma, such as ERBB4, were found fused to novel transcription units, thereby dramatically increasing their expres sion.⁸ Jones et al. aligned the somatic DNA mutations with the RNA expression data, and discovered that 129 of 268 (48%) nonsynonymous mutations in the DNA were found in expressed RNAs. Moreover, only about half of these RNAs were present at sufficient levels to be likely to hold biological relevance. Thus, the already remarkably low load of somatic DNA mutations observed by all three papers is reduced by another 75%, supporting the hypothesis that paediatric medullo blastomas are driven by a surprisingly short list of genetic hits. In this paper, the majority of mutated genes were found to be unique to a single case, unlike the situation in another paediatric brain tumour, glioblastoma multiforme. Jones et al. found that the most mutated gene in Group 3 tumours was SMARCA4. As in the Pugh et al. study, CTDNEP1 mutation was found to accompany 17q abnormality, suggesting that CTDNEP1 is a good candidate for a tumour suppressor gene on 17q.^7,8

The paper by Northcott et al. for the MAGIC Consortium focused on somatic copy number alterations (SCNAs) in 1,087 medulloblastoma tumours, as determined by single nucleotide polymorphism arrays.⁹ The authors pointed out that many of the recurrent targetable SCNAs, such as those involving IGF1R, KIT, PDGFRA and PTEN, have already been targeted with small molecules in the treatment of other malignancies, suggesting rapid translation for the SHH subset of patients. They found that the oncogene OTX2 was recurrently deleted in Group 3 tumours. OTX2 is a prominent target of transforming growth factor β (TGF-β), suggesting that TGF-β signalling could be a rational target for this subset of patients, who typically have a poor prognosis. Northcott et al. also found that PTV1 fusion genes were highly restricted to Group 3. In addition, they detected abnormalities in the α-synuclein-interacting protein (SNCAIP) gene in Group 4 tumours, indicating that SNCAIP may be a driver gene for this subgroup of patients.

Overall, these three papers further confirm previous findings, including the distinctive ness of Group 1 (WNT) and Group 2 (SHH) tumours.^6,8,9 They clarify aspects that separate Group 3 and 4 tumours, and identify possible driver mutations. The studies, especially that of the MAGIC group, delineate important potential therapeutic targets for medullo blastomas. However, the potential applications for this new knowledge remain unclear. With the excellent prognosis in Group 1 tumours (over 90% dis ease control in some studies), incorporating novel approaches and proving their efficacy and the benefits of their inclusion into treatment regimens for newly diagnosed patients will be difficult. In view of the more variable prognosis of patients with SHH-driven tumours, and given that PI3K gene amplifications and deletions of PTEN were restricted to this tumour type, subsets of patients (especially older children and adults) may be excellent candidates for PI3K antagonists, in addition to SHH antagonists already in clinical trials.⁹ Outcomes of Group 3 tumours are relatively poor, and the identification of TGF-β or other modifiers of cellular signalling as potential targets is of importance. Therapy for Group 4 tumours also remains problematic, as does determination of the most rational interventions, although therapeutic targets are beginning to be elucidated.

Without this deep sequencing of medulloblastoma and the investigations of ‘large’ number of patients, such insights could not have been gained. The next challenge will be to determine which genetic abnormalities are truly drivers of disease and what molecular abnormalities are best targeted, with which agents and with what consequences to the tumour and patient.