High resolution genome analysis for improved prognostic stratification and candidate gene selection in neuroblastoma

Neuroblastoma (NB) is a rare paediatric tumour with a remarkable clinical heterogeneity. NB can behave very aggressively or can be rather benign and even regress without therapy. Although prognostic features have been identified such as tumour stage, age at diagnosis and MYCN copy number, prediction of tumour behaviour for the individual patient remains a major challenge for therapeutic stratification for children with NB. Misclassifications should be avoided in order to make sure aggressive tumours are not undertreated. It is however equally important to make sure patients with prognostic favourable tumours don't receive intense chemotherapy as they would be unnecessarily exposed to the rather severe short and long term side effects of such treatment. This thesis has two major goals. First we set out to investigate in further detail the possible prognostic significance of genomic alterations (DNA copy number changes) in NB using high resolution array comparative genomic hybridisation (array CGH). A second goal was to contribute to the mapping and identification of genes involved in NB oncogenesis through detailed delineation of regions of loss or gain as determined by array CGH and the application of a new genome wide screening assay for nonsense mutations in putative tumour suppressor genes. In order to achieve the first major goal, I have performed array CGH on 75 primary tumours. Data analysis unequivocally revealed the existence of three major genomic subgroups. In addition other smaller subgroups were identified, in particular tumours showing combined features from two or even all three major genomic subgroups, which seem to be associated with a poor prognosis. This illustrates the potential power of a prognostic classifier based on DNA copy number alterations across the NB genome and pleads for the evaluation of the genomic aberration pattern at diagnosis. Of further importance, we could show that 1q gain was a particularly unfavourable genomic feature in the subset of aggressive NB. This group of patients would therefore represent a prime target for innovative (molecularly oriented) therapy. Given the present lack of insight into the molecular pathways governing NB development, I pursued further studies towards the identification of critically involved genes. First, the same array CGH data set was used in order to refine recurrent regions of 3p and 11q loss and 17q gain. As a result of the use of BAC tiling path arrays and screening of many tumours and cell lines my investigation lead to a significant progress in the mapping of these critical regions. Interestingly, the smallest regions of overlap for 3p deletions coincided with those occurring in common epithelial tumours, indicating the presence of mutual perturbed pathways in seemingly unrelated neoplasms. A similar approach, combined with integrated analysis of mRNA expression data of primary tumours and fetal neuroblasts lead to the identification of a strong candidate tumour suppressor gene CADM1, which possibly exerts its effect through haplo-insufficiency. As no small regions of 17q gain or amplification were identified upon array CGH analysis we followed a new approach in order to hunt for candidate genes. To this purpose a new algorithm was developed called positional gene enrichment analysis (PGE) which lead to the identification of two critical regions of gain at 17q21.32 and 17q24.1 containing a limited number of putatively dosage sensitive oncogenes. Future in vitro and in vivo functional studies need to be performed aiming at elucidating the role of these candidate genes in the pathogenesis of NB. These data illustrate the contribution of array CGH in unravelling the molecular pathology of NB and underline the additive value of implementing array CGH data in a multi-faceted approach aimed at prioritising genes associated with an aggressive phenotype. Summary 153 A second approach to find NB tumour suppressor genes was the genome wide search for the presence of nonsense mutations in NB. This was achieved through the implementation of an innovative high-throughput screening platform called "GINI", which stands for Gene Identification by Nonsense mediated decay Inhibition. To facilitate data analysis, a web based and publicly accessible database was developed to store and retrieve the expression data used in this project. In order to achieve a rational selection of candidate genes for further analysis, different data-mining strategies were developed, resulting in a prioritised list of 78 candidate genes. Mutation analysis did thus far not reveal nonsense mutations in NB and further screening is ongoing. In conclusion, this thesis illustrates the power of array CGH in clinical patient management and future gene discovery efforts in NB. The integrative genomics approach here described serves as a proof-of-principle demonstrating the added value of combining different data sources as an important step towards understanding NB biology. In future, the data obtained in this thesis will be further analysed in conjunction with methylome and miRNA transcriptome data. We are hopeful that insights in the genes and cellular signalling pathways involved in NB obtained through integrative genomics may lead to the identification of targets for molecularly oriented pharmacological interference.

http://ift.tt/2rV8TwJ