Detecting Tumors from Shed DNA

Medical institutions across the United States will begin recruiting volunteers next month for a study that its investigators and outside observers describe as groundbreaking. They say it is the first large-scale trial to test the feasibility of using DNA shed by tumors to find early-stage cancer. During the three-year government-funded project, researchers will analyze DNA from stool samples to detect patterns characteristic of colorectal cancer (CRC). The study will have two other notable features. To be cost-efficient, the vast majority of studies that try to associate DNA alterations with cancer look at people at high risk of getting the disease or already suffering from it. But to determine stool DNA's value as a broad screening agent, the new project will instead draw its 4,000 subjects from the general population. In another departure, the study will use DNA changes as a springboard for finding cancer, not for treating it. "There's probably a hundred times more research on the therapeutic implications of [mutations] than on their diagnostic and preventative applications," notes Bert Vogelstein, director of the molecular genetics laboratory at Johns Hopkins Oncology Center. Vogelstein, who is not involved in the upcoming large-scale study, argues that this ratio shouldn't be nearly so lopsided. DNA-based screening "is a really exciting and pretty realistic possibility," he observes. "It's not just stool. You can think about sputum for lung cancer, urine for bladder cancer, seminal fluid for prostate cancer, and mammary effluents for breast cancer." Skepticism, Then Success Scientific advances in recent years may provide the new study's genomic underpinnings, but decades-old medical experience supplies its rationale. CRC is highly curable if caught early. Unfortunately, the commonly used fecal occult blood test has shown sensitivity as low as 25 percent in some clinical trials. And colonoscopy--the gold standard--is invasive and expensive so that few of the 75 million Americans older than 50, who are deemed at "average risk" of getting CRC, undergo the procedure. CRC consequently will kill almost 57,000 people in the United States this year, according to the American Cancer Society. David A. Ahlquist The new study will enroll people over the age of 50 whose colons haven't been structurally evaluated in the past 10 years, says principal investigator David A. Ahlquist, a professor of medicine at the Mayo Clinic. "What we don't want is a group from which the pathology has been culled out," he explains. Subjects will submit three stool samples for DNA testing, take a fecal occult blood test, and undergo a colonoscopy. Results then will be compared. One of the questions that researchers want answered is whether stool DNA testing and flexible sigmoidoscopy, another screening tool, are as accurate in combination as colonoscopy. (Sigmoidoscopies, which view the last third of the colon, won't be performed; data on that segment of the colon will come from the colonoscopies.) Besides Mayo, the participating institutions will include Indiana University in Indianapolis; Kaiser Permanente in Oakland, Calif.; Roswell Park Cancer Institute in Buffalo, N.Y.; Oregon Health Sciences University in Portland; and the North Central Cancer Treatment Group, a consortium of 20 community clinics. The National Cancer Institute's Community Clinical Oncology Program will enlist community physicians and local hospitals in the study as well. Donald E. Henson, the study's project officer at NCI, says that the institute will contribute at least $3.4 million. EXACT Sciences Corp., a small six-year-old biotech firm based in Maynard, Mass., will analyze the DNA at its own expense. The study's roots date back 13 years to David Sidransky's postdoctoral work in Vogelstein's lab. Now a professor of otolaryngology at Hopkins, Sidransky recalls skeptics telling him that he'd never isolate human DNA from stool, where bacterial DNA is thousands of times more prevalent. Vogelstein remembers a more theoretical objection to the research: "It wasn't clear that you were going to have the same gene mutated in lots of different cases of the same kind of cancer." Nevertheless, the two scientists and colleagues succeeded in detecting mutations of the k-ras gene in DNA laboriously purified from the stools of eight people with colorectal tumors.1 This research was carried forward by Jin Jen, first a Vogelstein postdoc, then a Hopkins faculty member, and now a principal investigator at NCI. Biologists in several European countries, Israel, and Japan undertook similar small-scale studies of stool DNA as a source of potential CRC markers. In the mid-1990s, EXACT began developing its own methods to purify DNA from stool. Last fall, a collaborative effort between Mayo and EXACT reported encouraging results from a study of 61 subjects. It found that stool tests for a panel of DNA changes displayed 91 percent sensitivity for CRC and 82 percent for precancerous polyps.2 Ahlquist, the paper's first author, recalls being barraged by thousands of phone calls from patients wanting the tests. In contrast, at the meeting of the American Association for Cancer Research in March, Jen and her colleagues reported that stool DNA testing had only a 71 percent sensitivity rate in 51 patients with CRC but a 92 percent rate in the 39 patients whose tumors contained at least one of the three targeted markers. Markers, Cancers, and Body Fluids From a research standpoint, the two burning questions are: Which DNA changes provide the best markers for early-stage cancer? Which other cancers besides CRC shed detectable DNA into stool or other bodily excretions and secretions? Anthony P. Shuber At first, stool researchers focused on mutations in codons 12 and 13 of the k-ras gene. These well-defined changes occur early in the development of 40 to 50 percent of CRC tumors. But the recent studies by Jen's group and the Mayo-EXACT team both found inexplicable problems--low sensitivity, false positives--with using k-ras as a marker. Anthony P. Shuber, EXACT's vice president of molecular biology, still plans to pursue k-ras by employing a post-PCR technique, modified single-base extension, that he thinks will yield cleaner results than earlier methods. After k-ras, the next most-studied gene in stool is p53. This apoptosis-related gene tends to be mutated during CRC development later than k-ras is, and it bears a greater variety of mutations. Another CRC-linked DNA change is shrinkage of BAT-26, a 26-adenosine-repeat microsatellite. The real prize, though, is the adenomatous polyposis coli (APC) gene, which is overwhelmingly mutated in polyps, precursors of colorectal tumors. This huge gene displays a complex spectrum of alterations that researchers are only starting to get a handle on as potential CRC markers. The Mayo-EXACT team recently stirred debate by proffering lengthy DNA fragments in stool as a robust marker.2 The theory is that tumors shed "long DNA," while DNA in noncancerous colon cells gets chopped up during apoptosis and enters stool in the form of short pieces. Meanwhile, reliable identification of other cancer-associated changes in stool DNA--methylation, telomere reduction, and deletion--either is years away or will require scientific breakthroughs. That doesn't bother Vogelstein. Referring to the currently detectable alterations in microsatellites and in APC, p53, and k-ras, he asserts, "A hundred percent of cancers have mutations in at least one and usually more than one of those genes." Preliminary findings by Ahlquist and Shuber suggest that long DNA in stool could reveal other cancers besides CRC that impinge on the digestive tract: tumors of the esophagus, pancreas, and lung (the hypothesis is that swallowed sputum contains DNA). An earlier, more extensive attempt by Jen and Sidransky to detect lung cancer through various chromosomal DNA markers in bronchoalveolar lavage fluid, however, wasn't very successful. The method flagged only 53 percent of tumors, and it missed the small, peripheral lesions most in need of improved diagnosis.3 Jen now proposes that changes in mitochondrial DNA, which exists in many copies per cell, may better indicate lung cancer. Sidransky, on the other hand, sees blood serum as a promising source for telltale DNA.4 Tumor DNA in serum, he adds, isn't necessarily a sign of metastasis. After stool, urine may be furthest along in development as a source of clinical DNA tumor markers. Sidransky is now wrapping up a 400-subject multicenter study of microsatellites in urine. He and his collaborators are analyzing these markers in high-risk and diseased people to diagnose early-stage bladder cancer and monitor recurrence of the disease. An Enviable Position Don M. Hardison Fitting onto a toilet seat, EXACT's stool DNA test is user-friendly. The company plans to begin marketing it next month to people at risk of hereditary non-polyposis colorectal cancer and in 2003 to average-risk individuals. As a so-called "home brew" test sold by a source certified under the Clinical Laboratory Improvement Act, the test won't need Food and Drug Administration approval, but the company will seek the agency's sanction anyway, says president Don M. Hardison. He foresees people getting stool DNA tests every one to three years. A positive finding would require a follow-up colonoscopy. David R. Lewis, an analyst with Thomas Weisel Partners in San Francisco, estimates the potential U.S. market for DNA stool tests at $6.2 billion. He bases this figure on a three-year interval between tests--meaning that 25 million older Americans would get tested each year--and a $250 price tag, which is about what a sigmoidoscopy costs. (Hardison says that EXACT hasn't yet settled on a price.) Assuming that the results of the NCI-funded study and a large company-funded clinical trial to begin later this year favor stool DNA tests, EXACT appears to be in an enviable position. The firm's head start in this field and its portfolio of 12 issued and 23 pending patents might deter other companies from developing similar tests. Lewis predicts that insurance companies will cover EXACT's stool DNA test if they conclude it will save them money over the long term. Still, researchers have yet to decide conclusively how specific and sensitive stool DNA tests can be. Equally important, notes NCI's Henson, will be the answer to the question "Can you find early cancer? If you find late cancer, so what?" Douglas Steinberg (dougste@attglobal.net) is a freelance writer in New York. References 1. D. Sidransky et al., "Identification of ras oncogene mutations in the stool of patients with curable colorectal tumors," Science, 256:102-5, 1992. 2. D.A. Ahlquist et al., "Colorectal cancer screening by detection of altered human DNA in stool: feasibility of a multitarget assay panel," Gastroenterology, 119:1219-27, Nove.2000. 3. S.A. Ahrendt et al., "Molecular detection of tumor cells in bronchoalveolar lavage fluid from patients with early stage lung cancer," Journal of the National Cancer Institute, 91:332-9, 1999. 4. J. Jen et al., "An overview on the isolation and analysis of circulating tumor DNA in plasma and serum," Annals of the New York Academy of Sciences, 906:8-12, 2000.

Detecting Tumors from Shed DNA

Interested in reading more?

Become a Member of