HONG KONG – A new collaborative study by the Riken Center for Life Science Technologies (CLST) in Japan and the Harry Perkins Institute for Medical Research in Australia has identified multiple new genes that are up-regulated in different cancer types, opening the door for the development of new biomarker tests for early cancer detection and more effective new treatments.

Biomarkers allow cancers to be diagnosed and staged, ideally based on relatively non-invasive blood tests. Although reliable biomarkers have been identified for several cancers, including prostate-specific antigen for prostate cancer and human chorionic gonadotropin for germ cell tumors such as testicular and ovarian cancers, they remain elusive for most malignancies.

However, while many different cancer biomarkers have been identified and published in the medical literature, very few of those actually seem to have had clinical utility, which may be for a number of reasons.

For example, "pretty much every gene and protein that has ever been used as a cancer biomarker has a time and a place where it should be normally expressed and very few, if any, markers are completely cancer-specific," project leader Alistair Forrest told BioWorld Today, when asked why that might be the case.

"Thus when we look at cancer markers, we are looking for something expressed either at a higher level than it should be or in a place where it shouldn't normally be expressed," explained Forrest, professor of systems biology and genomics at the Harry Perkins Institute. "Also, when you look at many people you often see that different people have different basal levels of these proteins."

In addition, "I would say that false-positive results are a major problem," said the study's first and co-correspondent author, Bogumil Kaczkowski, a postdoctoral research fellow at Riken CLST.

In order to avoid false-negative or -positive diagnoses "you need a very sensitive and specific test at a reasonable price, but meeting all those criteria is extremely difficult in practice," Kaczkowski explained. "We also hope that recent advances in RNA technology will help overcome the limitations of protein-level biomarkers."

The findings of a new collaborative study, which were based on the results of two different transcriptome (RNA) technologies, were reported in the Nov. 9, 2015, early online issue of Cancer Research.

The researchers first looked at data generated by Cap Analysis of Gene Expression (CAGE), a highly sensitive technique developed at Riken CLST by Piero Carninci and Yoshihide Hayashizaki as part of the ongoing Functional Annotation of the Mammalian (FANTOM) genome project, an international consortium led by Riken. (See BioWorld Today, April 1, 2014.)

Using the CAGE technology, the researchers examined RNA expression profiles of 225 cancer cell lines and 339 normal cells, and they looked for differences in gene expression between cancerous vs. normal states.

They then supplemented that information by looking at RNA sequencing data from 4,055 primary tumors and 563 healthy tissues from the U.S. NIH's The Cancer Genome Atlas (TCGA) database, then compared the two sets of data, in order to identify changes that were common to both. "Using the two different datasets allowed us to make use of the strengths of each," explained Kaczkowski.

"The TCGA data [are] complicated by the fact that tumor samples are composed of a mixture of different cells, while the FANTOM CAGE data [are] from cells grown in a cell culture, where changes might arise from the culture process," said Kaczkowski. "By putting the two together and looking at changes found in both, we have been able to make a robust catalog."

The researchers were able to identify 128 markers that were consistently perturbed in both datasets in a variety of tumor types. Some of them, such as TOP2A and MKI67, are well-known candidates for cancer biomarkers.

However, using data from the FANTOM project 5th edition (FANTOM5), the researchers identified a number of new markers, including noncoding RNAs, RNAs derived from repetitive elements and enhancer elements that CAGE technology could identify. In particular, they found that a little known repetitive element called REP522, was up-regulated in many cancers.

"About half of the human genome is made up of sequences that are repeated, many of which are retrotransposons and represent ancient viruses," noted Piero Carninci, director of the Division of Genomic Technologies and deputy center director at Riken CSLT, who was one of the leaders of FANTOM5 and a co-author of the Cancer Research study.

"These [repetitive elements] are usually considered either useless or even dangerous to the genome in the most classic view, but we have started to discover that often they provide important regulatory elements, said Carninci.

"Understanding the good and the bad role of these elements is important. Although many studies have discarded the results of the retrotransposon elements, because they consider they 'should' not do anything, we take them into account," he told BioWorld Today.

"We found hundreds of promoters, the parts of the genome that initiate expression of a gene, that are up-regulated in cancer, and in particular, promoters that overlap with repetitive elements in the genome seem to be up-regulated. This is an interesting insight into cancer development and we are especially interested in the REP522 element, and would like to determine what role it plays," said Carninci.

"We observed a very clear activation of transcription from REP522 elements in cancer cell lines," said Kaczkowski. "This is an interesting observation that has not been previously reported and we are now planning further studies to learn more about it," he added, noting that the significance of REP522 in terms of cancer therapy and/or diagnosis is currently unknown.

"We hope that researchers will use our findings to identify markers for the many cancer types that currently have no useful markers," said Carninci. "It may also be possible to target the genes we have identified as potential drug targets. These targets could potentially help many cancer patients as they are up-regulated in many different types of tumors."

Regarding new cancer drug development, "As we focus on molecules that are recurrently up-regulated in multiple cancer types, then a drug that targets one of these may not only work in one cancer type but in many cancer types, including those with currently poor prognosis," said Forrest.

"The other translational aspect is that if you diagnose patients earlier, then you generally have a better survival rate. Some cancers such as mesothelioma have a latent period of 20-40 years and, once diagnosed, patients typically die within nine months. If it is possible to detect these patients years earlier, we [would] have a better chance of treating them successfully."

"I believe that this is the first step of the many needed to validate these RNAs as biomarkers in a large population," said Riken's Carninci. However, he added, "we must not give false impression that everything is solved with this first discovery. In science, many steps like this one are needed and we should keep the funding high for follow-up studies, such as validations, broadening to many patients, etc. Our study is comprehensive, but not done in a clinical environment."

The FANTOM project began at Riken in 1998, with the first FANTOM aiming to build a complete gene catalog with complementary (cDNA) technologies, while FANTOM5 aimed to provide the first holistic view of transcriptional regulatory network models for the most human cell types that make up a human. "FANTOM6 has just started, which aims to assign functions to long noncoding RNAs," said Forrest.

"We also plan to elucidate the function of REP522 in cancer, and further validate candidate biomarkers, which will require extensive testing before they meet the requirements for clinical application," concluded Kaczkowski.