Methylation Markers Distinguish Aggressive from Manageable Prostate Cancer

Scientists in the U.K. and Canada have identified a DNA methylation signature that they claim can distinguish aggressive prostate cancer from more manageable disease, with up to 92% accuracy. The researchers, headed by Norman Maitland, Ph.D., at the University of York’s department of biology, hope that the pattern of 17 epigenetic markers could be used to assess each case of prostate cancer before deciding on whether the patient will need to undergo surgery or radiotherapy, or whether they can be monitored.

“Unnecessary prostate treatment has both physical consequences for patients and their families, but is also a substantial financial burden on the NHS [National Health Service], where each operation will cost around £10,000,” Dr. Maitland comments. “Cancers that are contained in the prostate, however, have the potential to be ‘actively monitored’ which is not only cheaper, but has far fewer negative side-effects in patients with non-life threatening cancer.”

The University of York researchers, together with colleagues in the U.K. and with Davide Pellacani’s team at the BC Cancer Agency in Vancouver, report on the new methylation signature in the British Journal of Cancer, in a paper titled, “Phenotype-independent DNA methylation changes in prostate cancer.”

Treatment-naïve prostate cancer (PCa) is characterized by an abnormal accumulation of proliferating cells that are similar to the luminal cells in normal prostate, the authors write. However, PCa also harbors small numbers of tumor cells with basal features, which display “cancer stem cell” features. These cells appear more likely to be treatment-resistant, and are proposed to serve as the reservoir for tumor recurrence after castration therapy,” the researchers continue.

Studies in bulk PCa samples have indicated that changes to DNA methylation patterns are involved in cancer development, but these studies have looked at epigenetic changes in luminal and basal cells combined, which may not be representative of tumor samples, they comment. Whereas normal prostate epithelial tissue is composed of similar numbers of luminal and basal cells, most treatment-naïve prostate cancers comprise primarily luminal featured cells. “This shift in favor of a transcriptional and epigenomic program of normal epithelial cells might mask or complicate the identification of cancer-specific features in prostate cancer when bulk analyses are performed on this type of tumor,” the authors state. In fact, “… very little is known about the specific DNA methylation features of PCa cells with basal and luminal phenotypes in comparison to their normal counterparts.”

To address this the team carried out genome-wide DNA methylation profiling of separate FACS-purified populations of basal and luminal cells, isolated from more than 500 patient-matched tumor and normal samples. Using a computer algorithm to eliminate the background of methylation patterns that are unique to individuals, the researchers were able to use their comparative approach to home in on cancer-related methylation profiles in each cell type.

Importantly, this method made it possible to look specifically at changes in the luminal cell fraction of PCa. The DNA methylation changes in these cells were associated with genes involved in metabolic processes, cell proliferation, and epithelial development, “all functions clearly deregulated in prostate cancer, therefore potentially containing major cancer driver events.” Their results highlighted two different classes of PCa-specific epigenetic changes: “… we were able to identify two separate classes of PCa-specific DNA methylation changes: those specific to cancer luminal cells that can distinguish both normal from cancer samples and organ-confined cancers from those with extraprostatic extensions; those common to basal and luminal cancer cells that are able to distinguish PCa efficiently from normal samples,” the team writes.

They used computer modeling to identify a panel of 17 cancer-related methylation changes which, when tested against a prostate cancer data within The Cancer Genome Atlas (TCGA) dataset, could distinguish cancers from normal samples, but could also differentiate between cancers that were confined to the prostate, and extraprostatic extensions, which are indicative of highly aggressive, invasive cancers. “Using this computer analysis, not only could we see which tissue samples had cancer and which didn’t, but also which cancers were dangerous and which ones less so,” comments Dr. Pellacani, who started the studies in York, before moving to the University of British Columbia. “Out of almost a million markers studied, we were able to use our new tools to single out differences in cancer potency.”

The researchers are continuing their studies in new cancer samples, and hope to team up with industry to generate a test that could be used with cancer patients in a clinical setting.