The observation that MB encompasses a number of distinct morphologic variants suggests that these tumors represent different entities arising through alternative mechanisms. Studies of gross chromosomal alterations in MB support this notion and have provided the first clues that different molecular processes underlie the development of MB subtypes (Ellison 2002; Gilbertson 2002; Gilbertson and Ellison 2008).

Deletions of 17p and isochromosome 17q (i17q), which combines loss of 17p and gain of 17q, have long been recognized as the most common chromosomal alterations in MB (Griffin et al. 1988). However, these alterations are not distributed equally among the histologic variants. i17q has been observed in 34 % and 36 % of classic and LCA tumors, respectively, but in only 12 % of desmoplastic MBs (Gilbertson et al. 2001; Lamont et al. 2004; Thompson et al. 2006). Furthermore, the presence of i17q in tumors has been associated with a poor clinical outcome, suggesting that this cytogenetic alteration may contribute to the development of aggressive variants of MB (Batra et al. 1995; Gilbertson et al. 2001).

In contrast, monosomy 6 has recently been shown to occur exclusively in favorable prognosis, mainly classic MBs that contain an intact chromosome 17 and activating mutations in the β-catenin gene (Ellison et al. 2005; Clifford et al. 2006; Thompson et al. 2006) Thus, chromosome 6 may harbor a tumor suppressor gene that cooperates with aberrant Wingless (WNT) signaling to generate an especially curable subtype of MB.

The desmoplastic and LCA variants are also associated with specific chromosomal alterations. Deletions of 9q are observed in up to 40 % of desmoplastic MBs, but occur rarely in tumors of the classic variant (Schofield et al. 1995), and amplifications of the MYCC and MYCN oncogenes occur predominantly in LCA tumors (Ellison 2002). Furthermore, a recent study showed that MB cells transduced with MYCC adopt a severely anaplastic phenotype when grown as xenografts, suggesting a causative relationship between MYCC expression and the LCA phenotype (Stearns et al. 2006).

Tumor formation is in many ways similar to normal development since both involve processes such as cell proliferation, migration, differentiation and apoptosis. Tumorigenesis is in fact ‘development gone wrong’, whereby developmental signaling or growth factor pathways involved in normal development become aberrant and drive tumor formation. Table 9.1 summarizes signaling and growth factors pathway aberrations.

In addition to the developmental pathways described earlier, other pathways have been implicated in MB pathogenesis and these have been investigated for a more precise understanding of the molecular basis of this tumor. Abnormalities in the EGF family of receptor tyrosine kinases, for example, have been detected in MBs. Receptor activation through ligand binding leads to its dimerization, autophosphorylation and activation of downstream PI3K and MAPK signaling cascades (Fig. 9.2). This process is essential for normal development of the CNS. Consequently, aberrant activation of this pathway results in upregulation of downstream signaling elements, which leads to increased cell proliferation and altered cell migration through the activation of transcription factor target proteins. In line with this, upregulation of RAS–MAP kinase downstream components, such as MAP2K1, MAP2K2 and MAPK1/3, has been correlated with metastatic behavior of MBs (MacDonald et al. 2001; Del Valle et al. 2002; Gilbertson and Clifford 2003). Furthermore, overexpression of the EGF receptor family member ERBB2 has been reported in MBs and linked to metastasis. High levels of expression of this protein co-overexpressed with ERBB4 signifies poor clinical outcome in cases of MB.

The HGF/cMET signaling pathway is also known to contribute to MB formation (Fig. 9.2). HGF signaling through the cMET receptor plays a critical role in cerebellar GNPCs proliferation and survival as they migrate from the EGL inward to form the mature IGL. Overactivation of this pathway can thus promote malignancies via increased cell proliferation and cell cycle dysregulation and lead to metastatic behavior by abnormal cell migration and invasion.

Aberrant signaling of the HGF/cMET pathway due to loss of pathway inhibition, ligand or receptor overexpression and/or activating mutations has been implicated in several human malignancies, including renal, lung, breast and CNS tumors. In MB, MET gene copy number gains have been detected by comparative genomic hybridization in 38.5 % of primary tumor samples and the receptor mRNA expression levels have been inversely correlated with patient survival(Tong et al. 2004; Li et al. 2005).

Finally, the amplification of MYC and MYCN proto-oncogenes has been detected in 5–15 % of primary MBs (Herms et al. 2000; Aldosari et al. 2002). MYCN is an early transcriptional target of the SHH pathway and its activation by SHH promotes the expression of the cell cycle proteins CyclinD1 and CyclinD2 leading to GNPCs proliferation. A high expression level of MYC is reported to cause progression of MB to an anaplastic phenotype and has been linked to poor clinical outcome.

Genes that have recently emerged as important suppressors of MB tumorigenesis are those that regulate the DNA damage response. The DNA repair pathway includes proteins such as poly-(ADP-ribose) polymerase (PARP-1) that sense DNA damage and proteins that repair the damage. DNA double-strand breaks activate two major repair pathways, homologous recombination repair and nonhomologous end joining.

The breast/ovarian cancer susceptibility protein BRCA2 is a DNA repair protein with a key role in homologous recombination repair. BRCA2 colocalizes with Partner and localizer of BRCA2 (PALB2) protein, which stabilizes BRCA2 within nuclear structures, facilitating DNA repair. Proteins involved in nonhomologous end joining include the nuclear ligase Lig4 and XRCC4.

Inactivating mutations in BRCA2 and PALB2 cause Fanconi anemia (FA) types D1 and N, respectively. FA includes a collection of disorders characterized by chromosomal instability, growth retardation, congenital malformations, progressive bone marrow failure, cancer predisposition, and cellular hypersensitivity to DNA cross-linking agents. Mutations in 12 genes have been identified in families with the various forms of the disease. FA-D1 and -N each carry a high risk of childhood solid tumors, including MB. Indeed, at least five of seven childhood brain tumors diagnosed in six FA-D1 kindreds were MBs, and seven patients identified with FAN5 had MB (Offit et al. 2003; Hirsch et al. 2004; Reid et al. 2007).

Although suppression of human MB by other components of the DNA damage response pathway remains to be established, mice null for either Lig4, XRCC4, or Parp-1 and p53 develop MB with high penetrance. Interestingly, most XRCC4-null/p53-deficient MBs delete Ptch1 and, in Parp-1 deficient mice, develop within the EGL. These tumors also express markers of GNPC and Shh pathway activation. Thus, defects in DNA repair may cooperate with mutations in the Shh pathway to cause MB.

Similar to what happened in WHO Classifications, MB was re-classified more than once depending on its biologic behavior, recent advances in gene expression profiling techniques have led to the generation of several molecular classification schemes in MBs: