Caractérisation des cancers de vessie par l’analyse intégrative des données de puces exons

The development of microarray technology in the late 1990’s served as an essential tool to comprehend the scope of transcriptomic deregulations occurring in cancer cells. Signals generated from the first generation of transcriptomic microarrays gave simultaneous measures of expression from a large number of genes, therefore enabling to identify candidate genes involved in cancer progression and putative therapeutic targets. In 2006, through a fast de- velopment of high-throughput technologies, the available large scale analysis tools became enriched with a new generation of high resolution microarrays measuring expression signals both at the gene-level and at the exon-level of each gene. The advent of this high-resolution microarray, commonly called exon array, provided the opportunity to get a more accurate meas- ure of transcriptomic changes affecting cancer cells by enabling to consider relative expression changes of the exons from a same gene.Alternative splicing and alternative transcription are the two main biological mechanisms accounting for the production of several transcripts from a same gene. Although these bio- logical processes have been known for a long time, their regulation in normal cells and their deregulation in cancer still remain challenging to well-characterize, mainly due to the complex- ity of the involved mechanisms. Through their design, exon arrays enable to identify variable expression patterns within several potential transcripts of a same gene, therefore bringing new insight into these biological processes.Based on a large dataset of bladder cancer samples that were profiled on exon arrays, we focused on the study of alternative splicing changes and alternative promoter usage in bladder tumours. Analysis of these exon arrays through the use of adapted statistical and mathemat- ical tools initially resulted in the identification of numerous genes showing differential relative expression patterns of their transcripts between cancer and normal samples. These transcripts represent a new opportunity to define tumour-specific therapeutic targets. We demonstrated that using an approach targeted on relative expression changes of transcripts from a same gene, it was possible to build up a panel of potential tumour-specific markers enabling the development of new urinary test to detect bladder cancer and monitor its evolution.Through an unsupervised analysis of putatively deregulated exon profiles, we observed that the partitioning of bladder tumours was similar to the classification resulting from the study of classical gene microarray expression profiles, consequently confirming the existence of a bladder subgroup with peculiar transcriptomic properties. For this subgroup of bad prognosis, we established a signature based on the differential alternative inclusion of several exons. This signature relates to 19 genes and enables to accurately identify tumours from this subgroup, therefore providing a powerful tool to be used in clinical practice.By studying a specific pathway often deregulated in cancer, we highlighted an overall dereg- ulation of the relative expression of alternative transcripts from genes involved in cell prolifer- ation, and identified potential actors involved in the underlying regulatory process. Eventually, the analysis of exon arrays in the light of DNA methylation array data enabled us to identify an epigenetic mechanism regulating the use of alternative promoters in a subgroup of bladder tumours.Together, the results obtained from exon array analysis consequently provided a character- ization at the transcript level of bladder tumour specific deregulations and brought insight into the underlying mechanisms. The highlighted deregulations not only allow to accurately identify two subgroups of tumours, of which one has a bad prognosis, but also offer new possibilities regarding the definition of urinary markers for patient monitoring.