Figure 1.
Correlation matrix of the gene expression signatures of the recurrences.
After a low-stringent initial filtering (p≤0.01 in at least 20% of experiments), a subset of 29783 probes was used to measure the correlations between relapses (Pearson correlation). As expected, and because each relapse was co-hybridized with its own reference, 6 of the 7 multi-recurrence patients clusterized together. There was however no similar evolution of profiles according to relapse locations or to the treatments received before the recurrence. PF = posterior fossa, ST = supratentorial, RT = radiotherapy, CT = chemotherapy, S = surveillance.
Figure 2.
Comparison of functions associated with location of relapses.
A Gene most frequently upregulated at relapse for supratentorial (upper yellow panel) are represented with the genes most frequently upregulated in infratentorial tumors (lower blue panel). Bars indicated the percentage of tumors in each location with upregulation of the specific gene. B The -Log10(p-values) of the most discriminatory functions in each group are represented. The p-value for a given function was calculated using the right-tailed Fisher Exact Test by considering 1) the number of uploaded functional analysis molecules that participate in that function, and 2) the total number of molecules that are known to be associated with that function in Ingenuity's knowledge base.
Figure 3.
Supervised hierarchical clustering of differentially expressed genes in ependymoma relapse compared to diagnosis.
Heatmap showing the 87 genes signature of the genes statistically up- or down-regulated in more than 50% of relapses with a fold change >2. Notice the homogeneity of the underexpression of the metallothioneins.
Figure 4.
Confirmation analyses (internal validation of gene expression).
A qPCR Heatmap showing expression of 3 candidate genes (MT3, KIF11, ASPM) in gene expression array and RT-PCR side to side. Pearson correlation coefficients between the two analyses are indicated. B Evolution of MT3 expression throughout progression. RT-PCR levels are given as Log scale compared to normal brain.
Figure 5.
MT3 and ASPM immunostains differ at diagnosis and relapse.
(A) Strong nuclear and cytoplasmic staining for MT3 at diagnosis. (B) At relapse the same patient shown at A displayed only weak MT3 staining. Another example of paired tumours, for which ASPM immunostaining was negative at diagnosis (C) and positive at relapse (D). Please also note paranuclear dots and a marked cell in mitosis (D, inserts), two patterns typically observed in ASPM immunostains, together with weak cytoplasmic staining.
Table 1.
Metallothionein 3 immunohistochemical expression in ependymomas at diagnosis and relapse.
Table 2.
ASPM immunohistochemical expression in ependymomas at diagnosis and relapse.
Figure 6.
A: Methylation status of 74 CpG sites at MT3 promoter and intron 1.
Each row of circles represents one EP sample sequenced from PCR products generated from amplification of bisulfite-treated DNA. Empty circles = unmethylated cytosines; Dotted circles = hemimethylated cytosines; Dark circles = methylated citosines. B: MT3 expression analysis by quantitative PCR in the corresponding ependymoma tumors. Samples with results under 1.0 are downregulated and those over 1.0 are upregulated compared to normal brain. The black bars correspond to sample at diagnosis, and the grey bar to relapse. Each histogram represent the corresponding sample studied for methylation. C: Epigenetic modulation of MT2A and MT3 expression on short term ependymoma cultures EP1 and EP2 and medulloblastoma cell line DAOY as control. Demethylation by 5-Aza-Deoxycytidine (left panel). Histone deacetylation inhibition by Trichostatin A (middle panel). Combined treatments (right panel). D: Treatment with zinc sulfate restores the expression of MT3. MT2A and MT3 expression level after 24 hours of 200 mM of ZnSO4 (left panel) and 5 microM of dexamethasone (right panel) treatments in the ependymoma primary culture cells EP1 and EP2 and in the medulloblastoma cell line DAOY.