Page 110 - GPD-4-2

Gene & Protein in Disease Pediatric glioma circadian clock genes

and the glioma as determined by one-way analysis of To explore whether recurrent and original pediatric
variance (ANOVA) followed by Tukey’s Honest Significant tumors have significantly different expression of the
Difference (HSD) test. Statistical tests were provided by the clock gene set, we examined ependymoma datasets from
GlioVis website interface. Hoffman et al. and Witt et al. and one GBM dataset
52
51
53
Not all genes in the circadian gene set or all glioma from Schwartzentruber et al. in Gliovis. Datasets were
types were present in each dataset examined. The glioma excluded that combined several glioma types into the two
groups being compared.
types evaluated here were glioblastoma multiforme
(GBM), pilocytic astrocytoma (PA), ependymoma (EP), 2.2. Expression index calculation
and medulloblastoma (MB) that had been previously
distinguished according to their histology and other To summarize the significant differences between the tumor
features. Pediatric datasets were selected from the 25 types and non-tumor tissue in the bulk tumor studies, a
available in GlioVis. Datasets that lacked samples from score ranging from -1 to 1 was calculated for each gene in
normal tissue were excluded from this first analysis, in the circadian dataset. We introduced this expression index
which we examined five pediatric datasets 30,36-38 and five (EI), which was calculated as ∑d (1-p)/n, where p is the
i
i
adult datasets. 39-43 The Pomeroy 2002 dataset of pediatric significance by Tukey’s HSD for the difference in expression
gliomas in Gliovis was examined but excluded from the of the gene in the tumor and non-tumor tissue, d is the
study because it did not provide results for eight of the 20 direction of this difference (represented as 1 for positive, -1
genes in the circadian gene set, and there were only four for negative, or 0 for not significantly different), and n is
non-tumor samples. the number of instances when this gene comparison was
available in the datasets, ranging from 1 to 10. (EI was set
To help validate the analysis using GlioVis, a Wilcoxon to zero when n was zero.) The summation is for i = 1 to n
signed-rank test was performed through the TIMER2.0 instances. EI values were calculated for each gene of the
44
website (http://timer.comp-genomics.org/timer/) to circadian gene set in each of three categories – adult GBM,
compare differences between the expression of the clock pediatric GBM, and the low-grade pediatric gliomas (not
genes in adult GBM and non-tumor tissue. This analysis GBM). The solitary adult pilocytic astrocytoma dataset
used data from The Cancer Genome Atlas (TCGA; https:// was not included in the EI calculations.
portal.gdc.cancer.gov/). The effects of age on clock gene
expression in MB were examined using GEO (https:// 3. Results
www.ncbi.nlm.nih.gov/geo/info/geo2r.html), which 3.1. Comparing pediatric and adult glioma clock
provides the statistical tool Linear Models for Microarray gene expression
Analysis (limma) to find differentially expressed genes in
microarray data. We examined the expression of 20 genes that either serve in
the central molecular mechanism of the circadian clock or
We used datasets of microarray and scRNA-seq results are well-established clock-controlled genes whose proteins
to compare clock gene expression in different subtypes control expressed rhythms. We evaluated this set of clock
of pediatric gliomas. Three medulloblastoma datasets genes within published transcriptomic results from five
45
explored through GlioVis were from Cavalli et al., microarray datasets derived from pediatric gliomas and
Robinson et al., and Northcott et al. for each member five from adult gliomas. The focus was on these ten studies
47
46
of the clock gene set, we compared four established tumor where expression was available in normal tissue from at
subtypes with each other in this bulk tumor data. The four least five patients in the same dataset for comparison.
MB subtypes were also examined through the University
of California, Santa Cruz Cell Explorer website (https:// For example, Figure 1 shows PER2 and PER3 expression
48
www.pneuroonccellatlas.org/), which provides interactive levels in individual tumor samples across glioma types
exploration of cluster analyses and original scRNA-seq from one pediatric and one adult glioma dataset. As
19
data generated by various studies. We used UCSC Cell previously described, the PER2 and PER3 genes were
Explorer to examine expression of the core circadian gene significantly suppressed in adult GBM relative to normal
set in MB results from Riemondy et al. and pediatric tissue (Table 1), but this difference was not present in
49
high-grade glioma (pHGG) results from DeSisto et al. to pediatric GBM (Table 2 and Figure 2). Both pediatric
50
compare tumor cell types. Significant differences between and adult GBM were repressed relative to low-grade (LG)
pairwise comparisons were determined using the Dwass- pediatric gliomas.
Steel-Chritchlow-Fligner test after significance was The adult and pediatric gliomas had very similar as well
found by the Kruskal–Wallis test using SysStat v. 13.2.01 as distinctly different expression patterns depending on
(Grafiti). the members of the circadian clock gene set. RORB was

Volume 4 Issue 2 (2025) 4 doi: 10.36922/gpd.4112