O4.04 * 'FROM THE CORE TO BEYOND THE MARGIN': A GENOMIC PICTURE OF GLIOBLASTOMA INTRATUMOR HETEROGENEITY

Abstract

INTRODUCTION: Glioblastoma (GB) is a very agressive brain tumor that almost systematically recurs. This recurrence is linked to the highly invasive behavior of this tumor. Major challenges in therapy of GB are therefore associated with controlling this infiltration sustained by a strong heterogeneous biology of GB tumors. In the past decade, molecular studies of GB have been focused on inter-individual heterogeneity. These studies have identified gene mutations and molecular signatures with putative prognostic or predictive significance. However, accumulating evidence suggests that intratumor heterogeneity is the key to understanding treatment failure. MATERIAL AND METHOD: To explore GB intratumor heterogeneity we developed a surgical multisampling scheme to collect tumor fragments from 10 GB patients in four spatially distinct areas defined on 3D MRI ‘from the core to beyond the margin’: necrotic zone, tumor zone, interface, and peripheral brain zone. These samples were studied genome-wide at three molecular levels: genome, transcriptome and methylome. We used Principal Component Analysis (PCA) to classify the samples and to select genes. We constructed a co-expression network on the basis of the expression data of the most informative genes related to the data structure revealed by PCA. RESULTS AND DISCUSSION: At the genome level, we identified common GB alterations (loss/partial loss of chromosome 10, polysomy of chromosome 7, focal deletion of the CDKN2A/B and focal high-level amplifications of EGFR) and a strong inter-individual molecular heterogeneity. Transcriptome analysis highlighted a pronounced intratumor architecture reflecting the surgical sampling plan of the study and identified gene modules associated with hallmarks of cancer. We provide a master gene signature highly correlated with the intratumor gradient and associated with the tumor infiltrative behavior. In this signature, the percentage of genes presenting an anti-correlation between expression and methylation levels was significantly higher than in the whole genome. This indicated that their abnormal expression might be linked to cis-acting epimutations in their promoter. We also evidenced that a single tumor can harbor different transcriptional GB subtypes depending on its spatial architecture. Results were validated on published data (Sottoriva et al., PNAS, 2013). CONCLUSION: Our study confirms that GBs are highly diverse at the three molecular levels, but not randomly so. GB intratumor heterogeneity is an important issue for the potential use of GB molecular subclasses in predictive algorithms and clinical trial stratification. We also evidenced a strong epigenetic determination of GB intratumor heterogeneity.