A One-Two Punch for Colon Cancer

Colorectal cancer (CRC) is a disease full of challenges. It’s awkward for patients to discuss, difficult to treat if not diagnosed early, and prone to high rates of screening noncompliance, thanks to challenges with stool collection and the necessary bowel preparation for colonoscopy. Even among patients who do submit to noninvasive screening, those who undergo fecal immunohistochemistry testing (FIT) alone run about a one-third risk of false negatives for cancer, almost all of which are early-stage, and a two-thirds risk of false negatives for significant precancerous lesions, especially sessile serrated adenomas (SSAs), which do not bleed. Taken together with noncompliance with annual screening, the outlook for FIT is pretty dismal for CRC prevention and early-stage treatment – but what can be done about it? Barry Berger is part of a team that has developed a noninvasive, in-home test that patients can perform at their leisure. The test uses a multi-target approach that looks for both fecal hemoglobin and DNA markers – and based on their results so far, it might just improve compliance, increase screening success rates, and contribute to decreasing CRC-related mortality and morbidity.

Let’s make a test for that

Stanley Lapidus – the inventor of the thin-prep Pap smear and a good friend of Berger’s – arrived at his laboratory one day in 1995 with a box of equipment in need of a fume hood. He announced, “I know what we’re doing next – we’re going to do stool cytology for CRC screening.” It wasn’t the answer Berger was expecting! “I had two things to say,” Berger recalls, “and the first was ‘No.’” He didn’t want to do stool cytology for two reasons: one, because he suspected that patients wouldn’t participate, and two, because stool samples contain very few cells. “It’s a great need, but not a good approach,” he told Lapidus. “Go think of a different one.” So Lapidus went on a quest and came back after having met Bert Vogelstein at Johns Hopkins. He reported, “Dr Vogelstein has found that colorectal cancer-associated DNA mutations can be found in stool DNA. Why don’t we figure out how to make a test out of that?” And so, teaming up with additional collaborators like the Mayo Clinic’s David Ahlquist, they did exactly that.

Originally, the researchers looked at mutations in genes known to be responsible for chromosomal instability. The trouble was that they needed so many of those mutations in order to get an informative panel that it was too technically challenging for a clinical lab. “We expanded our search to epigenetic changes,” says Berger, “and sure enough, aberrant methylation significantly overlapped with the carcinogenesis pathways. So just two markers allow us more coverage than 50 point mutations – though at the cost of a few more false positives. That was the genesis of Cologuard, the multi-target stool DNA test we have today.”

For Cologuard, a combination assay of DNA markers and fecal hemoglobin analyzed algorithmically in a logistic regression equation, the investigators wanted only the best. “We were agnostic as to the markers we chose, but we were very particular about performance. So we incorporated 11 biomarkers into our assay – the minimal number for maximum detection.” Why 11? Using additional markers resulted in larger decreases in specificity, but contributed only minimal increases in sensitivity – not worth the cost, or the burden on patients. The test included two methylation markers, BMP3 and NDRG4 (chosen because of their ability to discriminate between neoplasia and age-related methylation), as well as more standard markers like KRAS mutations. The test considers the markers as a group, using a regression algorithm to calculate a single score. “We found that optimizing individual marker results as in a ‘marker panel’ approach was less sensitive than looking at the composite logistic regression score,” Berger explains. “The test is optimized based on the score and a cutoff threshold provides a qualitative positive or negative result. Overall, that yields increased sensitivity with only a small reduction in specificity compared with a panel of individually evaluated markers.”

Promising performance

The Cologuard team performed a 10,000-patient study comparing their multi-target testing approach with FIT alone, with colonoscopy as the reference (1). The new test identified 92 percent of CRC, whereas FIT identified only 74 percent – and the difference was even greater when limited to early-stage cancers. Hemoglobin alone tends to miss early disease because it doesn’t ulcerate as reliably, so FIT identified only 70 percent of stage I and II cancers, compared with 94 percent using the multiple-marker test. That performance even held true for precancerous lesions like high-grade dysplasia (with 69 percent detected using Cologuard versus 46 percent using hemoglobin alone) or right-sided flat lesions (SSAs). “Those are particularly striking because they’re so hard for gastroenterologists to find – they don’t bleed, they’re difficult to spot, and FIT often misses them,” says Berger. The new test found 42 percent of SSAs ≥1 cm in size (compared with 5 percent from FIT) and 70 percent of those ≥2 cm (compared with FIT’s 11 percent).

He explains, “False positives do happen – and with 11 biomarkers, you have more than you might with a single marker. We designed our test to have 10 percent false positive results to optimize early-stage CRC detection. This trade-off of sensitivity against specificity allows us to detect 94 percent of actual, curable-stage cancers. That’s what we wanted to see. In the United States, where colonoscopy itself is a screening test, capacity is more than sufficient to accommodate the increase in colonoscopy numbers resulting from a somewhat higher single-application false positive rate. In turn, that increases the prevention and detection of disease – the primary purpose of screening.”

A good day at the office

In terms of application, the new test addresses many of the patient preference factors that keep people from being screened. For those who won’t have a colonoscopy, there’s now a high-sensitivity alternative. “If everybody were screened – either by Cologuard or colonoscopy – we’d catch 19 out of 20 CRC cases in the early stages,” says Berger. “Over a decade, we’d likely achieve significant decreases in CRC incidence and mortality. In addition, the longer test interval of three years decreases the burden of screening on patients, physicians and healthcare systems. We designed the test with higher sensitivity so that we could extend that interval, because we know that patients are often unwilling to do annual fecal blood screening.”

A recent study of 150,000 continuously insured patients over a 10-year period showed that, of those who used fecal occult blood/FIT for screening, only three of every 1,000 patients were compliant throughout the study period (2). “What we have seen is that people who want screening colonoscopies get them, but about half of the population won’t. Those are the people who want everything in a test: safety, sensitivity, convenience, and noninvasiveness. Now, for the first time, we have a test for that population, and I think that’s a big step forward.”

A recent independent study (3) asked whether or not the test actually fulfilled its mission – namely, to appeal to people who were previously noncompliant. The researchers looked at 400 patients 65 and older who were defined as consistently noncompliant (meaning neither colonoscopy within the past decade, nor annual FIT testing). Those patients were offered Cologuard; 88 percent of them came in for screening, and nine out of 10 patients with positive screens went to colonoscopy for diagnosis. “We found four early-stage CRC cases (three stage I, one stage II), all surgically correctable, as well as 21 advanced adenomas and a high-grade dysplasia. All of those were patients who wouldn’t otherwise have been screened, and who would eventually have presented with late-stage, difficult-to-treat disease. So that was a very good day at the office!”

The roll-out phase

“We want Cologuard to provide a standard of care for many patients,” says Berger. “Right now, approximately 40,000 physicians use the test routinely, which means we have another 260,000 to go! It takes time – about three years – to get a new test integrated into medical practice, but so far, the response has been good.” In the new United States Preventive Services Task Force guidelines for colorectal cancer screening published in late June, Cologuard testing every three years was included, placing it on equal footing with all other included strategies – a major milestone for the test (4). In the next few years, Berger and his colleagues intend to focus on rolling the test out first in the United States, and then eventually in Europe (where Cologuard is already CE marked) and Asia. They also want to look at other cancers with an eye to prevention using similar technology.

“I know pathologists are curious about how Cologuard might change their day-to-day work. The good news is that it probably won’t change a thing – at least at first. The test is only performed by Exact Sciences Laboratories, which functions as both a primary and reference lab. In the future, once the roll-out is complete and we’ve addressed any potential logistical issues, we’ll consider the opportunity for centers of excellence to perform the test themselves using our kits. It’s our hope that, at that point, we’ll be able to defeat the great unmet screening need in colorectal cancer.”

Barry M. Berger, MD FCAP is an anatomic, clinical and cytopathologist, as well as Chief Medical Officer, Medical Affairs at Exact Sciences Corporation (Madison, USA).