Stool DNA has potential for multiple cancer screenings

Despite improvements in overall survival for most cancers, survival for patients with pancreatic adenocarcinoma has remained dismally low for decades. Overall, less than 8 percent of patients survive more than five years after diagnosis. This year, pancreatic cancer surpassed breast cancer to become the third-leading cause of cancer deaths in the United States. By 2020, it is expected to replace colon cancer as the second-leading cause of cancer deaths.

High mortality in pancreatic cancer is due to early metastasis, poor response to chemo- and radiotherapy, and especially to late diagnosis. But when detected in its earliest stages before symptoms arise, cure rates approach 75 percent. Although expensive endoscopic and imaging procedures are used to screen for early cancers in high-risk subgroups — those with new abnormalities on imaging studies, hereditary factors or genetic syndromes — no effective and affordable approach to pancreatic screening exists at the population level. There is also a lack of diagnostic tools and risk prediction models that discriminate among pancreatic cysts found incidentally during imaging for symptoms unrelated to the pancreas.

More than a million pancreatic cysts are detected each year in the United States, most of which are not malignant. Nonmucinous cysts (serous cystadenomas) are almost always benign, and small, asymptomatic branch duct cysts have little malignant potential. But mucinous cystic neoplasms (MCNs) and some main duct intraductal papillary mucinous neoplasms (IPMNs) can display a range of neoplastic transformations, from low- to high-grade dysplasia and invasive carcinoma.

Although the overall incidence of cyst-related pancreatic malignancies is relatively low, the challenge is finding the small subgroup of people who have cysts likely to progress to cancer and who therefore will benefit from surgery. Unfortunately, current diagnostic approaches lack accuracy, and an estimated 40 percent of patients undergo unnecessary surgical removal of benign cysts — those that contain no or low-grade dysplasia only.

Mayo researchers are now developing noninvasive, cost-effective molecular tools designed to change this dynamic, says David A. Ahlquist, M.D., a gastroenterologist specializing in the molecular detection and prevention of gastrointestinal cancers at Mayo Clinic's campus in Rochester, Minnesota.

"Our group is looking at several approaches to pancreatic cancer screening and to identification of precursor lesions," he explains. "This effort includes comprehensive studies in various biological media. In early proof-of-concept studies, we demonstrated that detection of pancreatic cancer is feasible by methylated gene markers in stool, blood and pancreatic juice."

"Those proof-of-concept studies used older assay techniques and nonoptimized markers," Dr. Ahlquist says. "Since that time, we have performed rigorous, whole-methylome discovery, in which we identified the best-in-class methylated DNA markers for pancreatic cancer that have greater than 99 percent accuracy at the tissue level. You have to start with highly accurate markers to have the best chance of detecting them in distant media such as stool or blood.

"And we now have a more sensitive assay platform to measure the markers. More discriminate markers and sensitive platforms translate into more sensitive clinical tests, which are in development but not yet ready for release."

Novel discriminatory markers

Dr. Ahlquist's group has also identified and validated novel methylated DNA markers that accurately discriminate high-grade dysplasia (IPMN-high grade, pancreatic intraepithelial neoplasia-3) and invasive cancer from low-grade precursor lesions and normal pancreas.

The markers were identified based on whole-methylome discovery using bisulfite-treated DNA from frozen normal and neoplastic tissues. Candidate markers were selected based on area under the curve (AUC), fold-change and p values. A panel of top markers yielded an AUC of 0.91 and detected 89, 87, 77 and 74 percent of cases with respective specificities of 85, 90, 95 and 100 percent. The findings were presented by Shounak Majumder, M.D., in a Presidential Plenary session talk at Digestive Disease Week 2016.

Although differentiating cysts with high-grade dysplasia from those with low-grade or no dysplasia is a big step forward, more research is needed and is underway. Dr. Ahlquist and colleagues have launched a multi-center study to evaluate markers in pancreatic cyst fluid to better identify which patients with cysts are at high risk and best treated by surgery.

Future directions

"The most exciting thing about molecular tools is that they allow us to reimagine how we go about screening in the first place," Dr. Ahlquist says. "The current state of the art is to screen one organ at a time — colon, breast, cervix and so forth. Molecular testing opens the door to noninvasive detection of multiple cancers with a single test — not only a stool test for upper and lower gastrointestinal tract cancers but also a blood test for all major cancers."

With such an approach, it would be critically important to predict the tumor site so that clinicians could efficiently direct downstream diagnostic evaluations, and this appears to be possible using methylated DNA markers, he says. Results of a Mayo Clinic pilot study presented by John B. Kisiel, M.D., at the American Association for Cancer Research 2015 annual meeting demonstrated the ability of new blood markers to, with high accuracy, correctly classify colon and pancreatic cancers and discriminate them from normal tissue.

"This exciting universal cancer screening approach is not only a revolutionary concept but also a rational and increasingly realizable goal," Dr. Kisiel says. "Every cell has the same amount of DNA; it's the methylation profiles on the DNA that determine tissue specificity. Why not take advantage of the body's inherent biology and look at methylated DNA changes for cancers within various organs as a source of early detection markers for cancer that are also site specific?

"Our group is actively exploring this approach across multiple organ systems. Although it is speculative at this juncture, and it will take a village of committed scientists to accomplish, we are moving toward the day when most cancers can be screened effectively with a single noninvasive test."