Mesenchymal transition and PDGFRA amplification/mutation are key distinct oncogenic events in pediatric diffuse intrinsic pontine gliomas

Abstract

Diffuse intrinsic pontine glioma (DIPG) is one of the most frequent malignant pediatric brain tumor and its prognosis is universaly fatal. No significant improvement has been made in last thirty years over the standard treatment with radiotherapy. To address the paucity of understanding of DIPGs, we have carried out integrated molecular profiling of a large series of samples obtained with stereotactic biopsy at diagnosis. While chromosomal imbalances did not distinguish DIPG and supratentorial tumors on CGHarrays, gene expression profiling revealed clear differences between them, with brainstem gliomas resembling midline/thalamic tumours, indicating a closely-related origin. Two distinct subgroups of DIPG were identified. The first subgroup displayed mesenchymal and pro-angiogenic characteristics, with stem cell markers enrichment consistent with the possibility to grow tumor stem cells from these biopsies. The other subgroup displayed oligodendroglial features, and appeared largely driven by PDGFRA, in particular through amplification and/or novel missense mutations in the extracellular domain. Patients in this later group had a significantly worse outcome with an hazard ratio for early deaths, ie before 10 months, 8 fold greater that the ones in the other subgroup (p = 0.041, Cox regression model). The worse outcome of patients with the oligodendroglial type of tumors was confirmed on a series of 55 paraffin-embedded biopsy samples at diagnosis (median OS of 7.73 versus 12.37 months, p = 0.045, log-rank test). Two distinct transcriptional subclasses of DIPG with specific genomic alterations can be defined at diagnosis by oligodendroglial differentiation or mesenchymal transition, respectively. Classifying these tumors by signal transduction pathway activation and by mutation in pathway member genes may be particularily valuable for the development of targeted therapies.

Figure 1. DIPG are different from supratentorial high-grade gliomas in children.

Hemispheric, midline/thalamic tumors and DIPG are represented in gold, grey and violet respectively. Panel A: Gene expression of 23 DIPG and 49 supratentorial HGG were compared using a Principal Component Analysis on all 15231 quality control passing probes. Tumors are displayed according to their coordinates on the three first principal components, which describe 29.7% of the variance. Panel B: Heatmap of the 712 most differentially expressed genes between DIPG, midline and hemispheric tumors, selected using the moderated t-test of limma package of Bioconductor. Panel C: Radial plot of the expression of transcription factors and neurogenesis regulators according to the three tumor location, in log2 ratios related to normal brain stem. The zero red line represent the expression level of normal adult brainstem.

Figure 2. DIPG are divided into two groups with different gene expression signatures.

Gene expression levels of 23 DIPG were analysed using a unsupervised procedure. Panel A: K-means algorithm followed by a model selection procedure using BIC defined two separated groups of DIPG that can be also clearly seen with a PCA on all probes that passed the quality control. Panel B: Heatmap of the 643 most differentially expressed genes between the two groups of DIPG, selected using the moderated t-test of limma package of Bioconductor. Panel C: Overall survival curves of the two groups of DIPG defining a group of patients who died early (70% of cases before the median survival time of 10.6 months, light green curve) and a group of patients who died later (90% of cases after the median survival time of 10.6 months, purple curve), (p = 0.004, chi-square test). Panel D: Integrated genomic analysis using DR-Integrator (R package) showing the correlation between probes of copy number and gene expression mapped on the same genomic coordinates (Refseq HG19) of a gene. In the upper panel (resp. the lower panel), colored vertical lines (cyan for group 1, purple for group 2) show probes for which copy number and expression were significantly correlated. The two panels in the middle show the CNA frequencies for the group 1 (resp. for the group 2). Most of the correlations between GE and CGH were found in group 1. Gene set enrichment analysis (GSEA) plot comparing group 1 GE profile to the signatures described for adult type gliomas. Group 1 gene expression profiles were enriched for proneural genes (Panel E) while group A gene expression profiles were enriched for mesenchymal genes (Panel F).

Figure 3. Description of the mesenchymal type of DIPG.

DIPG from group 2 gene expression profile was enriched with genes involved in mesenchymal transition, angiogenesis and stem cell maintenance. Panel A: Heatmap of the transcription factors linked with mesenchymal gene expression signature (MGES) in adult glioblastomas. Biomarkers of mesenchymal phenotype (VIM, CHI3L1 and TNC) and the two master regulators of epithelio-mesenchymal transition, SNAIL1 and SNAIL2/SLUG were added to the list provided by Carro et al (Carro et al., 2010). Panel B: Boxplots comparing the 7 transcription factors driving the MGES in adult glioblastomas (Carro et al., 2010) in the two groups of DIPG (group 1 in cyan, group 2 in purple). Relative expression in log2 ratio compared to normal brainstem control is indicated. Vimentine immunohistochemistry in tumors of group 2 shows the positivity of tumors cells (Panel C) while in group 1 only vessels and reactive astrocytes were positive (Panel D). Panel E: Spearman correlation of the expression of SNAI2 and VEGFA. Group 1 tumors (cyan dots) segregate clearly from tumors of group 2 (purple dots). Panel F: Spearman correlation of the expression of CHI3L1 and VEGFA. Group 1 tumors (cyan dots) segregate clearly from tumors of group 2 (purple dots). Panel G: Gene expression of stem cell and mesenchymal markers in DIPG tumorospheres derived from primary tumors of patients in stem cell medium as previously described (Thirant et al, 2011). Quantitative RT-PCR (qPCR) were performed using normal brain cortectomy as control. The spheroids cultured from three different DIPG were compared to normal neural stem cells (NSC) grown as neurospheres in the same medium.

Figure 4. Description of the oligodendroglial/proneural type of DIPG.

Panel A: Radial plot showing the expression of oligodendroglial markers in the two groups of DIPG, in log2 ratios related each other. The red circle represent the median expression level of the whole population of DIPG. Group 1 expresses higher levels of oligodendroglial markers than group 2 DIPG. Panel B: Morphological oligodendroglial differenciation in group 1 tumors (HES staining, ×40). Panel C: Morphological astrocytic differenciation in group 2 tumors (HES staining ×40). Panel D: Olig2 immunohistochemistry in a group 1 DIPG showing that probably not all cells in the biopsy are tumoral (×40). Panel E: Dual immunohistochemistry for Olig2 and GFAP showing that tumor cells in mitosis are GFAP negative but Olig2 positive (×100). Panel F: Overall survival of 55 DIPG according to the presence (red) or absence (blue) of oligodendroglial differenciation. Median OS was shorter in patients with oligodendroglial type of tumors (7.73 vs 12.37, p = 0.045, log rank test).

Figure 5. PDGFRA amplification/mutation is driving the oncogenesis of the oligodendroglial/proneural type of DIPG.

Panel A: GSEA graph showing the enrichment of group 1 DIPG for the gene set describing the gene expression profile of PDGFRA amplified glioblastomas. Panel B: PDGFRA immunohistochemistry in the infiltrative part of a DIPG. Panel C: PDGFRA immunohistochemistry in the tumoral part of a DIPG. Panel D: FISH analysis of a DIPG using a FIP1L1/PDGFRA probe showing the amplification of the locus encompassing the two genes (most frequently seen). Panel E: Dual-FISH analysis of a DIPG with two probes one for PDGFRA and one for MET showing that the two oncogenes may be gained/amplified in different cells within the tumor. Panel F: Integrative genomic analysis using DR-Integrator (R package). Seven genes are present in the minimal common region (MCR) gained on chromosomal location 4q12 in DIPG. Boxplots represent the distribution of GE data and circles represent CNA data. The circles are centered on the corresponding GE measure on the distribution and their radii are proportional to the absolute value of CNA, red ones being losses and green ones gains. CNA and GE were highly correlated for four of these seven genes (CHIC2, KIT, KDR, PDGFRA). Panel G: Diagram of the PDGFRA gene showing the mutations discovered in DIPG samples and xenografts.

Figure 6. Comparison of gene expression signature of the two DIPG groups with specific neural lineages.

A GSEA analysis was processed using the gene list previously described by Lei et al and derived from the gene sets specifically enriched in astrocytes, oligodendrocytes, neurons, oligodendrocytes progenitors cells and cultured astroglial cells. Heatmap of the enrichment scores of each DIPG sample is represented with a red to blue color scale shows the range from the highest to lowest enrichment score.