Subclassification of diffuse glioma tumours based on gene expression and allele copy number variation

The development of targeted cancer treatment adapted to individual patients requires the identification of different tumour classes according to their biology and prognosis. Gliomas are the most frequent primitive cerebral tumours and in general are divided into 2 histological subtypes: astrocytic and oligodendroglial. Most of these tumours are diffuse and infiltrating. Relapses are inevitable, with a usually fatal outcome. The histological criteria allowing the distinction between glioma tumours remain difficult to use. For about ten years, expression microarrays have been used to refine the diagnosis of gliomas and to develop a new classification. Our work focused on genetic and transcriptomic modifications that could affect the tumour behavior in a set of 30 diffuse gliomas, including astrocytomas, oligodendrogliomas, and mixed gliomas of WHO (World Health Organization) grade II and III. We linked Mapping data from Affymetrix 250k_Nsp arrays and Expression data from Affymetrix HG_U133_Plus2 arrays through a Biological Pathways analysis. The hypothesis is that common chromosomal aberrations should affect gene expression levels in commonly involved pathways. Thus, sub-classes of gliomas are defined based on statistically affected pathways and genes. Mapping studies allowed us to confirm the discriminative status of the chromosomal 1p and 19q regions for the oligodendrogliomas. Most of the chromosomal aberrations are associated with oligodendrogliomas. Astrocytomas only shared an uniparental disomy on chromosome 17p as a common aberration. We found 3 sub-groups of gliomas characterized by distinct signaling pathway alterations (more than 50% of the genes involved in those pathways are differentially ex- pressed, p-value! 0.05). These results are currently under validation by immunohistochemistry. Combining major biological parameters in a single analysis is a powerful way to create a first step of classification. Mapping studies aim to focus on interesting genomic regions. Biological pathways led us to the main genes involved, and by transcriptomics we assessed the statistical relevance of both pathways and groups of samples. These results will lead us to refine the glioma classification based on genetic pattern, and will allow us to discriminate these tumours at a new genetic level.