Clinical outcome of patients with medulloblastoma can be highly variable, irrespective of parameters routinely used in the clinic to predict patient risk (i.e., patient age, extent of resection, and metastatic stage). This is often exemplified by the very disparate therapeutic responses observed among patients with histologically identical disease. Furthermore, survival patterns reported for different age groups of medulloblastoma patients have strongly suggested underlying biological differences that must account for the distinct age of onset and therapeutic response of these patient subgroups [6, 22]. Much of this biological and clinical heterogeneity observed in medulloblastoma is now being explained by the recognition of unique molecular subgroups of the disease, a concept that has been supported for more than a decade now [23].

Hybridization of moderate-to-large series of RNAs isolated from primary medulloblastoma specimens to gene expression microarrays has revolutionized our concept of medulloblastoma as a disease. Based primarily on expression array profiling, numerous studies performed by independent laboratories have reported the existence of discrete molecular subgroups of medulloblastoma [24–28], each characterized by distinct genetics, cytogenetics and transcriptional profiles, as well as patient demographics, tumor phenotype, and clinical behavior (Figs. 9.3 and 9.4) [29]. In late 2010, a consensus meeting attended by leading groups in the medulloblastoma community was held where it was proposed that medulloblastoma should be regarded as consisting of a four-subgroup structure: WNT, SHH, Group 3, and Group 4 [30]. The impact of this consensus report and the studies that precipitated it have changed the way medulloblastoma is both viewed and studied in the basic research setting. Moreover, the implications for the development of clinical trials, revamping patient risk-stratification, and the administration of targeted therapy have been dramatic, all of which are now beginning to take molecular subgroup status into consideration.

The least common of the four consensus medulloblastoma subgroups are those belonging to the WNT subgroup, accounting for just 10–15 % of cases. This subgroup affects a higher than expected proportion of female patients (male:female ratio of ~1:1 compared to an expected ratio of ~1.5:1 for all medulloblastomas) and is predominantly diagnosed in childhood and adolescence, almost never encountered in infants. WNT medulloblastomas are nearly without exception of classic histology and non-metastatic, currently having the best overall prognosis of any patient subgroup with cure rates of almost 100 % [4, 24, 27, 28, 31, 32].

Somatic missense mutations in exon 3 of CTNNB1 (which encodes β-Catenin) are present in >90 % of WNT medulloblastomas [33]. These mutations constitutively activate β-Catenin and prevent it from being degraded, resulting in its nuclear accumulation and consequent deregulation of WNT target genes. Monosomy 6 is another characteristic genetic feature of WNT subgroup medulloblastomas, highly enriched in this subgroup and found in the vast majority of cases [32, 34]. Immunohistochemistry (IHC) for β-Catenin and fluorescence in situ hybridization (FISH) for chromosome 6 constitute two commonly used “readouts” for WNT subgroup assignment, with either nucleo-positivity for β-Catenin or monosomy 6 acting as capable surrogates for identifying WNT subgroup patients (Fig. 9.5) [35]. More specific assays such as re-sequencing of CTNNB1 exon 3 performed in combination with additional state-of-the-art molecular profiling methods (discussed below) are now being recommended as more specific and sensitive methods for identifying this important subgroup of medulloblastoma patients.

Medulloblastomas exhibiting aberrant activation of the SHH signaling pathway account for ~25–30 % of cases [32]. In infants (age 3 years or less) and older patients (age 16 years and up), SHH-activated cases predominate, accounting for up to 70 % of patients in these age groups [36]. The gender distribution for SHH subgroup medulloblastomas is comparable to what is seen among all medulloblastoma patients, with an observed male:female ratio of ~1.5:1. True desmoplastic and MBEN histologies appear to be largely restricted to this subgroup [35]; however, large cell/anaplastic (LCA) disease is also encountered, particularly in childhood cases. Metastatic disease (M+) is also observed but relatively uncommon. With respect to patient outcome, SHH cases appear to represent an intermediate prognosis class of patients [32], although specific patient subsets may have highly favorable or disparate clinical outcomes, depending on their underlying genotypes (as will be discussed later) [22, 37].

As was first verified in Gorlin Syndrome patients that develop medulloblastoma, PTCH1 is the prototypical tumor suppressor in the SHH subgroup, with somatic mutations inactivating PTCH1 confined to this subgroup. SUFU, another tumor suppressor functioning as a regulatory component of the SHH signaling pathway, is likewise mutated in this subgroup, both in the germline and somatically [38–40]. Loss of heterozygosity (LOH) on chromosomes 9q and 10q is enriched in SHH medulloblastomas [27, 34], presumably as a mechanism for inactivating the remaining wild-type allele in cases exhibiting PTCH1 or SUFU mutations, respectively. Similar to the use of β-Catenin nucleo-positivity as a biomarker for WNT medulloblastoma, comparable IHC-based methods have been proposed and implemented with modest success for the identification of SHH medulloblastomas, including SFRP1, GLI1, and GAB1, all of which exhibit apparent specificity for this subgroup [27, 28, 35, 41, 42].

The generically named Group 3 medulloblastomas account for ~25 % of all cases and appear to be restricted to pediatric patients, with only extremely rare instances reported in adults [42]. There is a male gender bias in Group 3, with an observed male:female ratio of ~2:1. M+ disease is common in Group 3 and has been documented in up to half of these cases. Group 3 subgroup affiliation currently carries with it the most dismal overall survival of the four subgroups, with only ~50 % of these patients or less alive at 5 years from the time of initial diagnosis [32].

Amplification of MYC is a characteristic oncogenic event observed in this subgroup, an event that is by and large restricted to Group 3 and found in 15–20 % of cases [34]. Cytogenetic aberrations such as gain of chromosomes 1q and 7, and loss of chromosomes 8p, 11, 16q and i[17]q are commonly observed in Group 3. Although preliminary biomarkers for Group 3 have been suggested (i.e., MYC amplification), there are currently no gold-standard single gene/marker assays for assigning medulloblastomas to this subgroup.

Group 4 medulloblastomas represent the most common patient subgroup, accounting for 35–40 % of all cases [32]. These tumors occur across all age groups but constitute the most predominant form of the disease in childhood and adolescence. There is a strong gender bias in Group 4, with an observed male:female ratio of ~3:1. Metastatic disease is observed in approximately one-third of Group 4 patients. Similar to the SHH subgroup, Group 4 patients tend to comprise an intermediate outcome subgroup [32], although there is growing evidence for clinical heterogeneity within this large fraction of patients [43].

Cytogenetically, Group 4 medulloblastomas share some commonalities with Group 3, most notably being the highly prevalent i[17]q which is noted in up to 70–80 % of Group 4’s. Chromosome X loss in female Group 4 patients is also a frequent occurrence. Compared to other subgroups, Group 4 medulloblastomas remain the least well understood with regards to oncogenic driver genes, although a few interesting novel candidates have emerged from recent genomic studies, as will be discussed in detail below.

A number of genes previously implicated in medulloblastoma have now been accurately placed in the context of the molecular subgroups. Several of these candidates were once thought to be rarely affected when studying medulloblastoma as a single entity, but are now considered of higher relevance given their subgroup-specificity and increased frequency within a particular subgroup.

The TP53 tumor suppressor, classically reported as being somatically mutated in only ~5 % of medulloblastomas, has now been observed to be almost exclusively mutated in WNT and SHH subgroup cases, affecting ~10–15 % of cases from each subgroup [37, 46]. Moreover, germline mutations in TP53, the hallmark genetic event causing Li-Fraumeni Syndrome (LFS) [47, 48], a condition predisposing to the development of a variety of different cancers including medulloblastoma, have now been confirmed to be restricted to the SHH subgroup [49], especially in childhood and adolescent patients.

In contrast to the MYC proto-oncogene which is amplified exclusively in Group 3, high-level amplifications of MYCN are found in both SHH and Group 4 but rarely in Group 3 and never in WNT medulloblastomas [34, 43]. Likewise, copy number gains of OTX2, a developmental transcription factor previously implicated in medulloblastoma pathogenesis [50–54], appear to be restricted to Groups 3 and 4 [34, 55], suggesting it plays an important role in the biology of these tumors.

Additional oncogenic copy number alterations showing enrichment in SHH-driven medulloblastoma now include (but are not limited to) amplification of GLI2, MYCL1, PPM1D, YAP1, IGF1R, IRS2, MDM4, and miR–17/92 and focal homozygous deletion of PTEN and PTCH1 [34, 56, 57]. These events collectively suggest that at least three main pathways contribute to the majority of SHH-driven medulloblastomas: SHH signaling, RTK/PI3K signaling, and TP53 signaling [34].

Identified as a common target of recurrent mutation in three parallel NGS studies of medulloblastoma [58–60], DDX3X is among the newest candidates to be implicated as an important medulloblastoma driver gene. DDX3X is a DEAD-box RNA helicase that has been shown to play a role in a variety of cellular processes, ranging from chromosome segregation to transcription and translation [61–64]. Mutations in DDX3X are confined to either of its two helicase domains and are always non-truncating variants [33], suggesting that these mutations alter the function of DDX3X rather than causing loss-of-function [59, 60]. Approximately half of all WNT cases harbor a DDX3X mutation, whereas 10–15 % of SHH cases are likewise mutated. Recent NGS of adult SHH medulloblastomas has revealed a high proportion of DDX3X SNVs, suggesting this candidate is particularly important in the biology of adult SHH cases. In contrast, mutations in DDX3X are seldomly observed in Groups 3 and 4.

One of the most unexpected findings disclosed from recent genomic studies of medulloblastoma concerns the high frequency of mutations and copy number alterations affecting chromatin modifiers [33, 65, 66]. In 2009, Northcott and colleagues reported a series of infrequent but recurrent SCNAs targeting histone lysine methyltransferases, histone lysine demethylases, histone acetyltransferases, and chromatin remodelers [67]. A subsequent landmark exon re-sequencing study performed by Parsons et al. identified recurrent and mutually exclusive mutations in histone 3, lysine 4 (H3K4) methyltransferases, MLL2 and MLL3, collectively mutated in ~16 % of surveyed cases [68]. Since these two initial reports implicating deregulation of the chromatin machinery in medulloblastoma, this theme has been further substantiated in all recent medulloblastoma NGS studies, now in the context of molecular subgroups [33, 58–60]. Interestingly, MLL2 mutations have been confirmed to be more common in WNT and SHH tumors, whereas MLL3 mutations are more prevalent in Groups 3 and 4. The SWI/SNF family gene SMARCA4 that encodes BRG1, a component of a multi-protein chromatin-remodeling complex, is recurrently inactivated in WNT and Group 3 medulloblastomas. Similarly, chromatin-modifying genes LDB1, BCOR, and LMO4 are targeted either by mutations, copy number alterations, or both exclusively in SHH-driven cases. Finally, KDM6A, a histone 3, lysine 27 (H3K27) demethylase, appears to be inactivated by either somatic mutation or focal deletion specifically in Group 4. Moreover, EZH2, which imposes the opposite function of KDM6A, catalyzing the trimethylation of H3K27 (H3K27me³) is aberrantly over-expressed in Groups 3 and 4, suggesting a propensity for an aberrant H3K27 methylation state in these subgroups [60, 69]. Collectively, this series of recent observations strongly supports the notion that deregulation of the histone code is a key event in medulloblastoma pathogenesis, with somatic alterations occurring in a high proportion of cases across the four subgroups.

Medulloblastoma subgroups exhibit highly disparate molecular genetics and clinical characteristics, suggesting they should be treated as different diseases in the clinic. Before such a concept is put into general practice, robust, highly accurate, and efficient methods that are accessible to treating physicians for establishing subgroup assignments are necessary. Novel assays that are gaining an appreciation in this arena include the use of DNA methylation arrays and platforms for measuring the expression of custom gene panels. Both Schwalbe et al. and Hovestadt et al. have recently demonstrated that DNA methylation arrays can be used to assign medulloblastoma subgroups with high confidence, including samples derived from formalin-fixed paraffin embedded (FFPE) material [104, 105]. RNA-based methods such as the nanoString assay have also shown utility at subgrouping of samples preserved in FFPE [106] (Fig. 9.9). Further validation of these methods and potentially others in the setting of medulloblastoma clinical trials is expected in the near future as the interest to subgroup patients in a prospective manner increases.

In light of the excellent prognosis associated with WNT medulloblastoma patients, plans to de-escalate craniospinal radiation or even eliminate it in these patients will be implemented in forthcoming clinical trials. Similarly, prospective stratification of all Group 3 patients into a high-risk treatment category is likewise being considered.

Another subset of patients that appears to be of significant clinical relevance are those with TP53-mutated SHH medulloblastomas. Using large retrospective patient cohorts, Zhukova et al. demonstrated that the dismal outcome sometimes attributed to medulloblastomas harboring TP53 mutation [107] can be explained by considering patient subgroup information [37]. TP53-mutated cases within the SHH subgroup exhibit a significantly worse outcome compared to either subgroup-matched non-mutated counterparts or WNT cases likewise harboring TP53 mutation. Furthermore, the prevalence of chromothripsis and oncogene amplification observed in TP53-mutated SHHs suggest these patients should be stratified as a unique risk-group and possibly subjected to treatments tailored for their genotype.

One of the major goals motivating the comprehensive genomic characterization of medulloblastoma is the identification of targets that can be specifically exploited for future treatment of the disease. To date, antagonists of the SHH pathway, acting mainly at the level of SMO, have demonstrated the most promise [108–112]. Compounds such as GDC-0449 from Genentech have shown dramatic although transient tumor regression when administered to patients with metastatic medulloblastoma, with patients eventually becoming resistant to the targeted therapy [111, 113]. Similar acquired resistance has been noted when related SMO antagonists (i.e., LDE-225) have been used to treat mouse models of the disease [110]. Ongoing efforts aim to combine SHH antagonists with additional inhibitory agents targeting cooperating pathways in hopes of achieving an improved and more sustained response to treatment. Furthermore, genomic analysis of human SHH medulloblastomas suggests that not all SHH-driven cases are likely to respond to inhibitors acting at the level of SMO, as subsets of cases such as those exhibiting amplification of downstream pathway components (i.e., GLI2) are likely to have primary resistance to these agents [34, 67]. As such, screening patients for both SHH subgroup affiliation and their mutation/copy number status in select SHH pathway genes prior to treatment with the current generation of SHH pathway inhibitors could improve their likelihood of response in the future. In addition, novel approaches targeting pathway components downstream of SMO, including the use of agents inhibiting GLI family transcription factors, are currently being evaluated and may increase the likelihood of response when combined with conventional SHH antagonists [114, 115].

The frequent deregulation of chromatin modifiers in medulloblastoma and the potential consequences of these events on the underlying epigenome make the prospect of epigenetic therapy an attractive possibility for medulloblastoma patients [65, 66]. Histone deacetylase (HDAC) inhibitors such as Vorinostat are currently being evaluated in medulloblastoma clinical trials [116, 117]. Similarly, 3-deazaneplanocin A (DZNep), a potent inhibitor of EZH2, is now being prioritized as an agent to be tested in upcoming clinical trials for medulloblastoma [118]. Similar agents targeting histone-modifying enzymes are presently being evaluated in the research setting and undoubtedly will enter the clinical trials arena for medulloblastoma patients in the near future.

Several studies have used methods of varying resolution to investigate SCNAs in CNS-PNET over the last 10–15 years. The general picture that has emerged is that these tumors do not typically display any of the common changes seen in medulloblastoma, such as i[17q], but may harbor other changes. Recently, a region of focal amplification on chromosome 19q has been identified as a highly recurrent alteration in certain subsets of CNS-PNET, as will be discussed in more detail in a later section.

One of the earliest studies comparing the cytogenetics of CNS-PNETs to medulloblastoma, for example, noted a higher frequency of chromosome 14q and 19q loss in supratentorial compared with infratentorial tumors, and no 17q gain in the supratentorial cases [126]. Another early report described an amplification of the TERT gene in a recurrence of a medulloepithelioma, which was not seen in the primary tumor, suggesting a role in tumor progression [127]. The authors also described increased expression of telomerase as a common finding in CNS-PNET. A lack of 17q gain but recurrent loss of 13q was noted in an array-based study of CNS-PNETs, which also reported amplifications of PDGFRA/KIT, MYB and 19q, as well as homozygous CDKN2A/B deletion [128]. One subsequent report did identify chromosome 17 alterations in 2/10 CNS-PNETs, but again noted that this change was significantly less frequent than in medulloblastoma [129]. The same study reported regions of loss on 1p and gain of 19p as being recurrent in CNS-PNET, and described deletions of CDKN2A/B in 7/21 cases examined [129]. Recurrent gain of 19p was confirmed in a more recent study of 29 CNS-PNETs, along with gains on 2p and 1q, which were seen in more than 20 % of cases [130]. Focal loss of CADPS on 3p was observed in 28 % of samples, and tumors showing this loss carried a worse prognosis [130]. In addition to CDKN2A/B deletions, SCNAs at other key cell-cycle regulatory genes seem to be a relatively common event in CNS-PNET, with 5/20 cases in one study displaying focal amplifications of CDK4, CDK6, CCND1, or CCND2 [131].