WASHINGTON -- The Food and Drug Administration is weighing the benefits and risks of two experimental colon cancer screening tests which use DNA from a patient's stool to detect dangerous tumors and growths.

FDA scientists have questions about the accuracy and the potential real-world impact of the kits from Epigenomics and Exact Sciences, according to briefing documents posted online Monday. The agency released its reviews of the tests ahead of a two-day meeting that starts Wednesday.

Doctors have long used stool tests to look for hidden blood that can be a warning sign of tumors and precancerous polyps. Colon cancer is usually treatable if growths are detected and removed before they multiply and spread to other parts of the body.

Both tests under review were more accurate at detecting tumors and worrisome growths than traditional blood stool tests. However, both tests also returned more false positives -- reporting growths when none were actually present.

In addition, the tests were less accurate in patients of certain racial and ethnic groups, including African Americans. The FDA will ask a panel of experts whether the tests should have special warnings for those patients, among on other questions, at its meeting later this week. The agency is not required to follow such expert advice.

Colorectal cancer is the second leading cause of cancer death in the U.S., with over 50,000 deaths expected this year, according to the American Cancer Society. Deaths from the disease have been declining for more than two decades, a development attributed to increased screening in patients ages 50 to 75. Still, only about 60 per cent of people in that age group have had recommended screenings.

Colonoscopy is the most accurate test but many adults are reluctant to undergo the invasive procedure, which requires several hours under sedation as doctors probe the colon with a camera-fitted catheter.

Blood stool tests are less accurate but can be just as effective if used every year, according to a federally-appointed panel that sets testing recommendations. A positive result doesn't automatically mean a patient has cancer but is usually a red flag that leads to confirmatory testing via colonoscopy.

Approving either of the new stool tests could reshuffle current testing practice, a fact reflected in the FDA's review. The agency suggests both tests would require additional follow-up studies to see how accurately they perform over several years. Neither company has offered instructions for how frequently patients should be tested with the technology, which is expected to be more expensive than traditional blood tests.

An editorial by two Veterans Affairs doctors addressed this issue in last week's issue of the New England Journal of Medicine.

"Only through a better understanding of other key factors, such as the screening interval, adherence, cost and diagnostic evaluation of positive results, can we determine the appropriate place for stool DNA testing on the screening menu," stated the authors, Dr. Douglas Robertson and Dr. Jason Dominitz. Neither author has any affiliation with the test makers. The journal published study results for Exact Sciences' Cologuard test ahead of this week's meeting. Exact Sciences Corp. is based in Madison, Wis.

Cologuard appeared to be the more effective of the two tests, according to the FDA's analysis.

The test accurately detected colorectal cancer 92 per cent of the time in patients who had cancerous tumors. That was better than the 74 per cent accuracy of the traditional blood stool test. In patients with precancerous polyps the test was accurate 42 per cent of the time, compared with 24 per cent for the blood test.

However, Cologuard was less accurate at ruling out cancer in healthy patients. When the test cleared patients it was correct about 87 per cent of time, less than the traditional test which accurately identified cancer-free patients 95 per cent of the time.

The FDA will ask its panelists what kind of follow-up instructions should be given to patients who test negative with Cologuard, so they don't presume they are cancer-free. Agency scientists express concern that "patients could defer additional screening after a negative result indefinitely if insufficient advice on follow-up testing is provided."

The panel will ask similar questions about Epigenomics' Epi proColon, which was less accurate in testing.

According to the FDA review, the Berlin-based company's test accurately detected cancer in 68 per cent of patients with cancerous tumors. The test only successfully identified healthy patients 79 per cent of the time, not good enough to meet the study's goal of achieving parity with traditional blood testing. FDA scientists warned that "lower specificity could lead to an increase in the number of avoidable colonoscopies" and "adverse events associated with such invasive procedures."

Neither company tested their product against the only FDA-approved DNA stool test currently on the market, ColoSure from LabCorp, which is based in Burlington, N.C.

The U.S. Preventive Services Task Force -- the federal testing advisory board -- has not endorsed ColoSure, saying there is not enough evidence available about its benefits and harms.

Shares of Exact Sciences Corp. rose 24 cents, or 1.8 per cent, to $13.80 in midday trading . Shares of LabCorp, whose full name is Laboratory Corp. of America Holdings, fell $1.81, or 1.9 per cent, to $95.22.