A Blood Test for Colon Cancer Sounds Great—But It's Short on Proof

The maker says it can find cancer early, but that’s a far cry from preventing colorectal cancer deaths


It's an eye-catching headline: " 'Revolutionary' Blood Test for Colon Cancer Screening Announced." The press release goes on to say that EDP Biotech Corp., based in Tennessee, will today brief its home state's legislators on what it calls a "breakthrough blood test" that "could save thousands of colon cancer deaths and billions in healthcare costs each year." (It not only briefed them, a later release said; it offered them the chance to participate in a study.) The company is hoping for approval from the Food and Drug Administration in the next couple of years.

In theory, that's wonderful. Routine colon cancer screening is recommended starting at age 50 by the United States Preventive Services Task Force, the American Cancer Society, and other public-health groups. Certain screening methods—including colonoscopy and sigmoidoscopy—can actually be considered a form of prevention, since they find polyps before they become cancerous. Yet, as the report from the February National Institutes of Health State-of-the-Science Conference on colorectal cancer screening said, "screening rates fall short of desirable levels," with about 55 percent of eligible people undergoing some type of screening. More options, particularly inexpensive, noninvasive ones, would be welcome. 

[Read: Colon Screening: 5 Things You Need to Know.] 

But while the ColoMarker test may well have all the potential in the world, based on the information available so far, "it is unproven as a screening measure," writes Durado Brooks, director of prostate and colorectal cancer at the American Cancer Society, by E-mail. He reviewed materials provided by the company. Like the highly controversial prostate-specific antigen (PSA) test, ColoMarker is a blood test that looks for a protein whose levels may be elevated in people with a certain cancer. In this case, it's CA11-19, which EDP says was isolated in 1973. The company says it applied its test to blood samples from more than 400 patients who had been referred to a gastroenterologist for some kind of condition. "All samples from patients with confirmed colon cancer in stages I through III were seen to have elevated levels of CA11-19," reads one slide of today's PowerPoint presentation to legislators. (For some reason, the test doesn't do well at detecting stage IV cancer, the most advanced stage.) That detection ability for stages I through III is encouraging, says Brooks, but not definitive. EDP also looked at CA11-19 levels in people without any symptoms of trouble to define the test's "normal" range, but "because these individuals did not have a colonoscopy or other procedure, there is no way to know how many of these individuals may have had cancer, polyps, or other abnormalities," Brooks says. (In other words, some may not be "normal" after all.) 

Preliminary studies making sure a screening test can find what it's supposed to find are certainly a prerequisite to demonstrating that the test is useful, says Barnett Kramer, associate director for disease prevention at the National Institutes of Health. (Kramer is speaking in general terms; he didn't evaluate this test's claims.) But it's a long way from telling whether a screening test works. "The ultimate goal that I look for, and that I suspect the public is interested in when they hear about any new test, is whether or not it prolongs life or reduces the risk of dying from a particular cancer," says Kramer. The way to do that is through a randomized clinical trial that takes the population in whom screening would be used and randomizes participants either to usual care or to screening (whether it's a blood test, colonoscopy, or imaging test like a mammogram). Researchers follow these people for years and, at the end, count the number of deaths from the disease (or treatment of the disease) in each group and figure out whether the test provided a benefit, says Kramer. 

Short of a randomized trial but a few steps beyond what ColoMarker has been shown to do: seeing if it can pick up cancer earlier than if you'd just gone by symptoms or other exams. That kind of prospective study requires samples taken from people without any symptoms of the cancer, says Kramer. You look to see what the test says about their cancer status, then follow up later to see whether the test accurately identified any cases of cancer before they were diagnosed through the usual means. But even success at early detection doesn't necessarily mean the test is ready for prime time. A test that "finds cancer early doesn't tell whether you help people or not," says H. Gilbert Welch, professor of medicine at the Dartmouth Medical School and codirector of the VA Outcomes Group in White River Junction, Vt., speaking generally rather than about ColoMarker. That's where the controversy over PSA lies: The test definitely finds cancer cases before they'd have been detected by symptoms, but it's as yet unclear whether using the test as a screening tool actually saves lives. That may sound counterintuitive, but in many cases, "the worst cancers are the ones growing fastest and the ones screening tests are most likely to miss," he says. Meantime, a test may be picking up cancers that would have been diagnosed in other ways without affecting the outcome or ones that wouldn't ever have threatened a person's health but, once found, are treated—with possibly harmful side effects.