By studying cancer from a network-based perspective rather than a collection of single genes, researchers from Clemson University have characterized uterine cancers in a new way.
“Looking at the whole network of genes involved in uterine cancer rather than a variety of single candidate genes will hopefully lead us to understand it more and develop better therapeutics,” said Allison Hickman, PhD, first author of a paper describing the technique published in G3-Genes Genomes Genetics. ”It is better to see the whole picture rather than target one or two genes out of the network.”
In their paper, the Clemson team describes their method of mining publicly available data sets The Cancer Genome Atlas (TCGA) of genomes from more than 11,000 cancer patients; and the Genotype-Tissue Expression Database (GTEx), a database of donated normal tissue.
“By merging the data sets we try to find biomarkers that are different between normal and cancer states,” ead investigator Alex Feltus, PhD, told Inside Precision Medicine. Feltus’ lab with the Clemson University Center Department of Genetics and Biochemistry has already successfully applied this approach in kidney cancer. Their method relies on statistical algorithms and machine learning from the team’s Knowledge Independent Network Construction (KINC) software to be able to distinguish between the two groups.
“We essentially screen through these huge data sets to find these markers.” As a self-described systems geneticist, Feltus sees his role as finding genes controlling traits, in this case uterine cancer.
Feltus’ team searches for gene expression patterns that differ between normal and diseased tissue. “We want to find the patterns’ biosignatures; not individual genes that are different but packages of genes that are different,” he adds.
With their method, data from a single individual can be injected into the mix and then compared with the normal and disease states. “Once you’ve done that you can see how their expression pattern fits into those groups of other tissues. It is precision biomarker discovery at the RNA level,” says Feltus. His lab is now building the infrastructure to do this insertion of patient data and analyze it.
In their current paper, Feltus and Hickman and colleagues describe discovery of regulatory gene networks in two types of uterine cancer — endometrioid carcinoma and serous adenocarcinoma — that are control points for gene expression.
“We found several candidate control points for normal tissue and cancer expression,” says Feltus, describing them as master regulatory switches.
In particular, the group identified 11 genes that are co-expressed in uterine cancer and appear to be co-regulated by the same transcription factor. “Those 11 genes are uterine cancer specific with evidence of switches that control them that are not found in normal tissue,” says Feltus.
“It appears we have found regulatory targets in uterine cancer that if you can remove those relationships pharmacologically then you might be able to adjust the system.”
The team’s systems biology strategy has confirmed existing relationships and found new ones. And it is knowledge-independent meaning they are studying gene expression patterns and trying to figure out what is going on in cancer. “We are trying not to rely on candidate genes, and we do not build from prior knowledge about gene relationships,” Feltus explains. “Instead, we use it to validate what we discover.”
These new networks add to the growing knowledge of uterine cancer biomarker systems and add to understanding the altered biological pathways that occur in diseased tissue compared with normal. The team hopes this knowledge can be used to better prognose and develop treatments for individuals impacted by these uterine cancers in the future.
