Cancer is a genetic disease—it stems from specific changes in the DNA sequences of the cancer cell genome. Over the years, cancer researchers have been wading through masses of gene data, slowly piecing together the nature of the mutations and the genes involved in cancer. We have now amassed vast genetic databases containing the sequencing information of a large number of cancer genomes. These data libraries serve as a reference point for scientists to pick out potential diagnostic and therapeutic targets.
Within these databases, however, there are regions of the cancer genome that have remained relatively ignored. In a recently published study in Cell Reports, a research team from the Ontario Institute for Cancer Research looked into the deepest, darkest parts of the cancer genome, emerging with a panel of 166 novel prognostic biomarkers.
The team, led by principal investigator Jüri Reimand, was interested in a relatively understudied aspect of the cancer genome: the activity of long non-coding RNAs, or lncRNAs. Over the last decade, genomics studies have uncovered several ways in which lncRNAs interact with DNA, RNA, and proteins to regulate the way cells behave and operate. However, until now, lncRNAs have been largely overlooked as potential diagnostic targets.
With the help of artificial intelligence, Reimand and colleagues evaluated the genomes of 9,500 cancer samples, spanning 30 different cancer types. They were looking specifically for patterns in the activities of 5,600 potential lncRNA biomarkers. The scale and scope of this type of analysis would be close to impossible without leveraging machine learning technologies.
Ultimately, the algorithm found 166 lncRNAs linked with patient survival rates, leading the researchers to hypothesize that these could be used as markers to forecast cancer patient outcomes more accurately. Of these 166, one lncRNA called HOXA10-AS was found to be a particularly robust prognostic biomarker for brain cancer, capable of separating the low-risk patients from the high-risk ones.
Reimand says that the team is excited by this giant leap forward in the field and is motivated to continue exploring the cancer transcriptome in search of more genomic insights. “We have only begun to scratch the surface of the role of RNAs in cancer and are poised for more discoveries as whole-transcriptome sequencing becomes more commonplace in the clinic, and more data is available,” said Reimand.
