337PMolecular subtypes of gliomas defined by gene expression profiling

Background: Gliomas are the most common primary brain tumors in adults. The heterogeneity of tumors, the lack of reliable criteria for identifying different subtypes make their histopathological diagnosis and their management complex. The molecular classification is one of the most promising approaches to better characterize gliomas. The aim of this study was to characterize molecular markers from the differential transcriptomic analysis of high and low-grade gliomas. Methods: Tumor samples were obtained from 81 patients diagnosed with gliomas between 1998 and 2013 at Limoges and Montpellier University Hospitals. Transcripts from brain tumor frozen samples were analyzed by Taqman Low Density Array. Cluster and principal component analyses were performed on a list of 96 selected genes belonging to glioma markers, genes coding for neurotrophins and their receptors or involved in different mechanisms such as glycosylation, autophagy, RTK signaling pathways, hypoxia and angiogenesis. Protein expressions from selected genes were achieved by immunohistochemical staining on Tissue MicroArray. Results: Firstly, variations in gene expression were noted between the primary tumor and its recurrence and between biopsy and resected surgical specimen for a same patient. To have homogeneous series, only 64 primary brain tumors obtained from resected surgical specimens and free from radiotherapy and/or chemotherapy were presented in this work. Brain tumors were diagnosed as grade II oligo-astrocytoma (n = 9), grade III oligo-astrocytoma (n = 10), grade III astrocytoma (n = 8), glioblastoma (n = 17), grade II oligodendroglioma (n = 9) and grade III oligodendroglioma (n = 11). Using the hierarchical cluster method, we identified gene expression patterns specific of low or high grades and a set of genes of interest appeared significantly overexpressed or under-expressed according to tumor grade (p < 0.05). Some of them were correlated to prognosis (p < 0.05). Immunohistochemistry analysis confirmed the changes of protein expression between low and high grades. Conclusions: Our results showed that high and low-grade gliomas differ in their gene expression profiles and several genes might act as new biomarkers for differential diagnosis and prognosis in gliomas.