Molecular heterogeneity of glioblastomas: does location matter?

authors

  • Denicolaï Emilie
  • Tabouret Emeline
  • Colin Carole
  • Metellus Philippe
  • Nanni Isabelle
  • Boucard Celine
  • Tchoghandjian Aurélie
  • Meyronet David
  • Baeza-Kallee Nathalie
  • Chinot Olivier
  • Figarella-Branger Dominique

keywords

  • Tumor location
  • Glioblastoma
  • Mesenchymal profile
  • Proneural profile

abstract

Glioblastomas in adults are highly heterogeneous tumors that can develop throughout the brain. To date no predictive-location marker has been identified. We previously derived two glioblastoma cell lines from cortical and periventricular locations and demonstrated distinct transcriptomic profiles. Based on these preliminary results, the aim of this study was to correlate glioblastoma locations with the expression of ten selected genes (VEGFC, FLT4, MET, HGF, CHI3L1, PROM1, NOTCH1, DLL3, PDGFRA, BCAN). Fifty nine patients with newly diagnosed glioblastomas were retrospectively included. Tumors were classified into cortical and periventricular locations, which were subsequently segregated according to cerebral lobes involved: cortical fronto-parietal (C-FP), cortical temporal (C-T), periventricular fronto-parietal (PV-FP), periventricular temporal (PV-T), and periventricular occipital (PV-O). Gene expression levels were determined using RT-qPCR. Compared to cortical glioblastomas, periventricular glioblastomas were characterized by a higher expression of two mesenchymal genes, VEGFC (p = 0.001) and HGF (p = 0.001). Among cortical locations, gene expressions were homogeneous. In contrast, periventricular locations exhibited distinct expression profiles. PV-T tumors were associated with higher expression of two proneural and cancer stem cell genes, NOTCH1 (p = 0.028) and PROM1 (p = 0.033) while PV-FP tumors were characterized by high expression of a mesenchymal gene, CHI3L1 (p = 0.006). Protein expression of NOTCH1 was correlated with RNA expression levels. PV-O glioblastomas were associated with lower expression of VEGFC (p = 0.032) than other periventricular locations, whereas MET overexpression remained exceptional. These data suggest a differential gene expression profile according to initial glioblastoma location.

more information