While the initiating genetic event in retinoblastoma – biallelic inactivation of the RB1 gene – is well established, little is known about subsequent genetic events that contribute to retinoblastoma formation and progression. A major unexplained question is why loss of Rb does not trigger an apoptotic response that eliminates nascent retinoblastoma cells before clonal expansion can occur. Thus, there are several potential explanations for how retinoblastomas circumvent apoptosis: (1) there could be a heretofore unidentified member of the p53 pathway that functionally ablates the p53 response despite the presence of functional p53; (2) there may be genetic events in other apoptosis-related pathways, such as the Bcl2 pathway; or (3) the retinoblastoma cell of origin may be naturally resistant to apoptosis. In most cancers, there are mutations in the p53 tumor suppressor or other members of the p53 pathway that explain the acquired resistance to apoptosis [6]. Further, in mouse models of retinoblastoma, tumor development is greatly enhanced when p53 is inactivated [27]. However, there is no evidence that p53 is mutated frequently in human retinoblastomas [28]. Further, p53 can be activated in retinoblastomas, suggesting that the protein is functional [29]. Other members of the p53 pathway can be disrupted in cancer, leading to functional inhibition of p53. For example, the p53 inhibitor MDM2 is overexpressed in some forms of cancer [30]. Although there is no evidence to date for MDM2 alterations in human retinoblastoma, other genetic events have been identified, such as E2F3 amplification [31], which may disrupt the p53 response that is normally triggered by loss of Rb. Finally, retinoblastoma may represent a unique exception in which the p53 pathway is intact. There are several lines of evidence to support this possibility. First is the lack of apoptotic mutations described above. Second is the fact that human retinoblastomas typically have a very high rate of apoptosis, suggesting that the apoptotic response is still intact but that proliferation is simply outstripping apoptosis [32]. Third, there is recent evidence suggesting that the retinoblastoma cell of origin may be naturally resistant to apoptosis, which potentially assuages the need to postulate any genetic event subsequent to Rb inactivation that is necessary for retinoblastoma formation [14].

Over the past several years, there has been progress in identifying the pathways that may contribute to our understanding of how retinoblastomas overcome cell death. First, there is evidence that the p53 antagonist called MDM4 is upregulated in retinoblastoma and that there may be alternatively spliced oncogenic forms of MDM4 in retinoblastoma [25, 33]. Indeed, this is therapeutically relevant as the MDM2/MDM4 antagonist (nutlin-3a) can efficiently kill retinoblastoma cells [34]. Second, recent data suggest that the SYK oncogene is epigenetically upregulated in retinoblastoma and this promotes the stabilization of the BCL2 family member called MCL1 (Figs. 6.2 and 6.3). Both SYK and MCL1 are required for retinoblastoma survival, and small-molecule inhibitors of SYK are effective at inducing retinoblastoma cell death in vitro and in vivo [35].

Recent whole-genome sequencing of retinoblastomas showed that some retinoblastomas have very few mutations or chromosomal alterations (Fig. 6.2) [35]. Moreover, the passage of orthotopic xenografts in the eyes of immunocompromised mice retains stable genomes [35]. Taken together, these data suggest that genomic instability and the subsequent genetic lesions that may result from such instability are not required for retinoblastoma progression [35]. While inactivation of the RB1 gene in retinoblastoma may result in defects in sister chromatid cohesion [36], this does not necessarily lead to genomic instability. Instead, it has been shown that inactivation of the RB1 gene leads to massive epigenetic deregulation of known oncogenes and tumor suppressors, and this may contribute to the rapid progression of the disease following the second mutational event in RB1 [35].

Recently, Brenda Gallie proposed that a small subset of retinoblastomas (~1 %) may arise as a result of a single oncogenic event rather than biallelic inactivation of RB1. Specifically, they propose that some retinoblastomas have MYCN amplification and do not have any RB1 mutations [37]. While it is a very small number of patients, it is an intriguing hypothesis that will require independent validation across a broader cohort of retinoblastoma samples. It will also be necessary to rule out other mechanisms of RB1 gene inactivation such as chromosomal events such as chromothripsis. This is important because chromothripsis at the RB1 locus would result in inactivation of the gene but would not be detected by conventional methods of RB1 gene analysis used by Gallie. Specifically, exon sequence analysis, promoter methylation analysis, analysis of LOH, and copy number changes would all appear to be wild type in a tumor where RB1 is inactivated by chromothripsis. All the exons of RB1 would be present, the promoter would be hypomethylated, both copies of RB1 would be present, and any copy number changes would be minimal. Thus, the tumor would appear to be wild type for RB1, but it would actually have a gene inactivation. Whole-genome sequencing combined with fluorescence in situ hybridization is currently the only way to identify retinoblastoma samples with chromothripsis. Future studies will be required to determine how many of the retinoblastomas with MYCN amplification actually have a wild-type RB1 gene and produce functional protein.