NEW YORK (GenomeWeb) – The research team behind the development of Exact Sciences' ColoGuard colorectal cancer screening test is now exploring whether the types of DNA methylation markers used in that assay can enable organ-specific prediction of tumor sites in gastrointestinal cancer.
In a poster presentation this week at the American Association for Cancer Research annual meeting in Philadelphia, John Kisiel of the Mayo Clinic shared how he and his collaborators at Mayo and Exact identified and used a set of methylated DNA markers to distinguish between different types of GI cancers with an accuracy of around 88 percent using tissue samples and 74 percent using plasma samples.
They hope their research has laid the groundwork to develop a blood- or stool-based test to noninvasively screen patients for GI or other types of cancer without prior knowledge of which organ is affected. To that end, the group is currently planning corroborative studies using a larger number of blood samples and examining several other cancer types, both in the GI tract and in other parts of the body, Kisiel told GenomeWeb.
"The ultimate goal … is a blood- or stool-based test to try to screen for multiple cancers at once," the advantage of which is that "we can then think of starting to screen for the rarer types of cancer," Kisiel said. "We currently don't screen for pancreas, biliary, or esophagus cancer in the general population because the incidence of those is pretty rare. But when you add up the incidence of all cancer within the GI tract, it becomes among the most common."
The team's most recent work was inspired by its previous work developing ColoGuard, which uses quantitative allele-specific real-time target and signal (QUARTS) amplification technology to detect DNA molecular markers in stool samples that contain cells shed from the colon wall. The test searches for aberrantly methylated BMP3 and NDRG4 promoter regions, mutant KRAS, and β-actin, a reference gene used to indicate DNA quantity.
In a pivotal clinical trial published last April in the New England Journal of Medicine, researchers from Exact and Mayo compared ColoGuard to one-time use of a fecal immunochemical test (FIT) for hemoglobin in a cohort of nearly 10,000 people who had an average cancer risk, revealing that the Cologuard test had a 92.3 percent sensitivity for detecting colorectal cancer compared to 73.8 percent for FIT.
Based in part on these data, ColoGuard received FDA clearance last August, and has since garnered coverage from several major payors and found its way into the American Cancer Society's national guidelines on colorectal cancer prevention and early detection.
"The technology for ColoGuard has been something that our research team at Mayo has jointly developed with Exact Sciences very aggressively over about the last five years," Kisiel said. "As we kind of locked down the algorithm for ColoGuard and were [clinically validating] that test … we started thinking about how the same technology might be applied to screen people for other cancers."
Kisiel noted that although biomarkers interrogated by the ColoGuard assay are very good at screening for colon cancer, they're not necessarily unique to the disease and might be represented in other GI tract cancers.
"We had some preliminary data that suggested there was feasibility for detecting cancers of the upper GI tract in stool, and we also had some early pilot data that suggested that we could tell, based on a combination of markers, whether the cancers were coming from the upper or lower part of the GI tract," Kisiel said.
"The extension of those observations was to see if we couldn't comprehensively categorize the fullest panel of markers at each of the major GI tract cancer sites — in this case esophagus, stomach, biliary, pancreas, and colon — and then compile those markers into models that would be able to predict the location of a tumor," he added.
In their most recent study presented at the AACR meeting, Kisiel and colleagues first used reduced representation bisulfite sequencing on frozen or formalin-fixed paraffin-embedded tissue samples and matched controls to identify novel methylation markers from cancers at multiple organ sites.
From a group of around 100 markers, they used bioinformatics approaches to home in on a three-marker panel for the universal detection of GI neoplasms with 95 percent accuracy; two markers to distinguish between upper and lower GI cancers with 94 percent accuracy; and another three-marker panel to identify pancreatic-biliary or gastro-esophageal cancer with 94 percent accuracy. In addition, they identified a two-marker plasma-based panel that was 78 percent accurate in assigning patients to cancer or control groups and 83 percent accurate in assigning cancer origin to colorectal or pancreatic sites.
Subsequently using these tissue biomarker panels with a SYBR Green methylation-specific PCR detection method on more than 300 normal and cancerous tissue samples, the researchers were able to accurately predict the type of cancer with an accuracy of 88 percent. Meanwhile, testing the plasma biomarker panel on 42 independent plasma samples yielded an accuracy of 74 percent in distinguishing colorectal from pancreatic tumors.
"That's just the percentage of the time that we correctly classified the specimen," Kisiel said. "Most of the error in that came from sensitivity rather than specificity, meaning we missed a few of the cancers, probably due to sub-optimal sample conditions. We think probably with larger blood volumes that we collect on our own, we'll do better, as the samples we used for this were convenient; they were not collected according to our specifications; and some of the patients may have already received treatments for their cancer – all of these things could have influenced the result."
Kisiel noted that the next logical step would be for the team to expand its findings in more larger-volume blood samples, and to include additional types of cancer. "We're probably going to have to study additional GI tract cancers and other cancers outside the GI tract to extend these observations to the ultimate goal, which is having a blood- or stool-based test to try to screen for multiple cancers at once," he said.
Should further research yield such an assay, or even a test to distinguish between a few different types of GI cancers, a commercial conduit is already in place through what Exact Sciences Chief Science Officer and study co-author Graham Lidgard told GenomeWeb is a model of public-private collaboration. To wit, in February the organizations announced that they had expanded their prior collaborative agreement, officially begun in 2011, to 2020 with the intent of developing screening, surveillance, and diagnostic tests for both colorectal cancer and other GI cancers.
The most recent work "is really sort of pioneering research on the utility of methylation markers," Lidgard said. "Today it doesn't directly translate into a product; it's really the demonstration that people are coming to realize that methylation is a universal defect or problem in solid tumors, and different tissues have different genes that are methylated. As demonstrated by this study, you can, with some degree of confidence, identify tissue origins [from] methylation markers."
Kisiel further noted that the Mayo and Exact team will likely be presenting data from follow-on studies at various cancer conferences this year.