Updated USPSTF Guidelines Seek to Increase Screening Rates
Interview by Steve Halasey
In the middle of June, the US Preventive Services Task Force (USPSTF) issued updated recommendations for colorectal cancer screening, reinforcing its previous recommendation that adults aged 50–75 be screened for colorectal cancer.1 This recommendation continues to receive USPSTF’s highest rating—an “A” grade—indicating that the evidence is convincing that screening for colorectal cancer has substantial benefits. The goal of its revised recommendations, says USPSTF, “is to maximize the total number of persons who are screened, because that will have the largest effect on reducing colorectal cancer deaths.”
To find out more about the USPSTF strategy for increasing screening rates for colorectal cancer, CLP recently spoke with David A. Ahlquist, MD, a professor of medicine at the Mayo Clinic College of Medicine, and coinventor of the Cologuard stool-based DNA screening test for colorectal cancer.
CLP: The USPSTF gives an A-level recommendation for average-risk individuals to begin screening for colorectal cancer at age 50 and continue to age 75, but a C-level recommendation to continue screening between ages 76 and 85. What considerations led to this difference in recommendations?
David A. Ahlquist, MD: I can only speculate about what went on in the USPSTF discussions, but the recommendations clearly respond to several acknowledged facts about screening for colorectal cancer. In general, for instance, colon cancer screening by any conventional strategy is more cost-effective than approaches to screening for any other cancer, including breast and cervical cancers. Colon cancer screening is a cost-effective general strategy, and I think the task force wanted to emphasize that.
Parsing the age-related recommendations of the task force ends up being sort of subjective. All would agree that at a certain age, population-level screening is no longer cost-effective, and its benefits no longer outweigh its risks. But there is disagreement over what age offers the best cutoff point. Consequently, the task force has created a sort of “grey area” for patients aged from 76 to 85. Giving a C-level rating to the recommendation for screening patients in this zone may translate into some third parties, and in this case Medicare, not fully covering screening for those patients.
It’s very difficult for clinicians to translate recommendations from population-based studies to the bedside. At age 65, some individuals have so much comorbidity that it would be inappropriate to screen them. While at age 80, other individuals may be very vigorous and have a life expectancy of 20 more years of life—and for those individuals continued screening might make sense. The grey zone on the far end offers an opportunity for physicians to exercise their clinical judgment and decisionmaking at the bedside. But because of these guidelines reimbursement coverage may be an issue for some payors, and this may influence some decisions about treatment.
Initiating screening when a patient reaches age 50 is another recommendation that could be subject to change over time. There’s a hockey stick-like distribution in disease prevalence that takes off after age 50, but during the past 2 decades there’s been a clear increase in colon cancer among individuals younger than 50. In fact, the decade seeing the greatest increase in colon cancer incidence is that of individuals between ages 40 and 50, when about 11% of all colon cancers occur.
The USPSTF guidelines may change in the future to capture this trend toward younger onset of colorectal cancer. This is especially likely if new diagnostic tools are less expensive and highly accurate, and may affect cost-effectiveness models differently from using colonoscopy. The current recommendations represent a reasonable position in light of current knowledge, but they should be subject to change over time.
CLP: From the patient’s point of view, how do the currently available screening tests differ from one another? For instance, USPSTF acknowledges that various screening tests for colorectal cancer have different strengths and limitations, but declines to rank the tests. How should patients decide which test is right for them?
Ahlquist: Depending on how you look at the statistics, only about half of the recommended population is currently being screened regularly. The figure of 60% participation is sometimes cited, but that number is based on self-reported questionnaires; data based on the records of actual procedures confirms that the screening participation rate is about 50%. There are a variety of barriers that may be preventing that number from being higher, including lack of awareness, reticence to undergo screening with current tools, difficulty with access, and so on.
Influencing how patients decide which tests they are willing to undergo requires an effective educational network that may need to go beyond what transpires at the doctor-patient level. This is also a public health education challenge, and it should therefore also target clinicians who need to be aware of the various testing modalities, so that their recommendations to patients can be made with knowledge.
At the end of the day, several factors may influence the recommendations of healthcare providers and the decisions of their patients. At the population level, effective detection equals the product of sensitivity, compliance, and access. You can have a perfectly sensitive test, but if patients are unwilling to use it, or can’t get access to it, the detection rate is zero.
These factors carry different valuations among patients. Some patients are so focused on the sensitivity of the test, that that’s the most important thing. For others, it is the logistics required to undergo testing that’s the biggest concern. If colonoscopy requires a patient to take 2 days away from daily activities, for instance, that could become an insurmountable barrier to patients who are living from paycheck to paycheck, and perhaps do not get reimbursement for their wages or travel expenses. The factors that patients weigh should be based on accurate information and a balanced presentation. Their selection of a test to go forward with depends in part on what the important factors are in their own lives. But in the end, gaining effective detection still boils down to sensitivity, compliance, and access.
With regard to colonoscopy, I’ve already mentioned that it can be difficult for a lot of people to take 2 days off work. Others are unwilling to undergo such an invasive test, or they’re concerned about potential complications. Among all patients that undergo colonoscopy, 1% to 2% end up in the emergency room within 24 hours. That’s a statistic that’s not widely known, but multiple studies have verified the consistency of that complication rate. Moreover, the complications involved are real: bleeding, perforation, and greater risk of mortality related to cardiovascular events, stroke, or heart attack. Those are all factors that have to be balanced; admittedly, they may occur at very low rates, but they nevertheless need to be considered.
We developed the Cologuard test to address each of the key factors of sensitivity, compliance, and access. We’ve shown in multiple studies that the sensitivity of the Cologuard test for cancer is essentially equivalent to what’s been reported for colonoscopy. The two major studies of Cologuard at the population level showed sensitivities of 92% to 100%. And for testing all stage I and stage II cancers—those that are at the most curable stage—the sensitivity of the Cologuard test is greater than 94%. That’s essentially the same as colonoscopy, and maybe better, since it is recommended that colonoscopy be performed every 10 years, whereas we’re recommending every 3 years for Cologuard. In addition, Cologuard also detects large polyps that are at greatest risk of progressing.
Cologuard doesn’t require any preparation, and it can be done in the patient’s home, so there is no need to miss work, Access should be unlimited, because Cologuard goes wherever the mail goes. We hope a test like this will meaningfully increase patient options, but whatever test a patient chooses should perform very well in terms of cancer prevention and early detection.
Too often, the current scenario is that the doctor simply tells his patient, “it’s time for your colonoscopy.” That’s not a balanced discussion, and studies have shown that participation rates are much higher when patients are presented with options. Again, there’s a need for education at the levels of both the patient and the healthcare provider, so that everyone has a more balanced look at the options, and more-informed decisionmaking can take place.
CLP: Patients may require heightened testing as a result of several risk factors, including a family history of colorectal cancer or a personal history of adenomatous polyps. What happens when the USPSTF recommendations don’t apply?
Ahlquist: The task force addressed the population that is at average risk for colorectal cancer, and that’s the large majority of the population. There are other sets of guidelines—by the American Gastroenterological Association and the American Cancer Society, for instance—that address recommendations for patients who fall into those higher-than-average-risk categories.
For patients with previous polyps, with a strong family history of colorectal cancer, or with a genetic syndrome that puts them at very high risk, the guidelines generally fall back to recommending colonoscopy. Despite its disadvantages, colonoscopy is perceived as the most-sensitive approach for cancer and precancers, especially when done initially at an age earlier than age 50, and done more frequently than every 10 years. Patients with clear risk factors are well managed with that approach.
There are some patients who fall into high-risk categories, but still refuse or are unable to undergo a colonoscopy. In those instances, clinicians and patients may choose to have CT colonography or Cologuard, both of which have been shown to have higher accuracy than fecal occult blood testing, which was historically the only other fallback. At this point, the clinical algorithms would favor a fallback to colonoscopy done more intensively than it would be in average-risk patients.
CLP: How concerned should patients be about potential harms related to screening, such as unnecessarily intensive surveillance after mixed test results, or colonic perforations and bleeding as a result of direct visualization tests?
Ahlquist: “Concerned” may not be the right word, but patients should be aware of these risks, and they should take a measured approach in factoring them into how they choose an approach.
For some patient subsets, those risks are actually accentuated, and that should factor into their choice of a screening tool. For instance, patients who are on blood thinners, or who have had cardiac issues, would have higher risks than the general population when undergoing a colonoscopy.
But again, it all boils down to education. Patients should be mindful that colonoscopy is associated with a 1% to 2% likelihood of having to go to the emergency room because of bleeding, or abdominal pain, or cardiac symptoms. But that means that 98% to 99% of patients have no problems.
Some patients like the convenience of having a screening intervention every 10 years. That’s another consideration that has to be factored into the decisionmaking. It should be on the table, though it will weigh differently to different individuals.
Using USPSTF data, the Cologuard test was associated with the highest benefit-to-risk ratio, and when done every 3 years, with the fewest number of colonoscopies over a screening lifetime. That could be important to some individuals.
The question, I think, has to be contextualized. Potential risks are part of the balance of decisionmaking.
CLP: USPSTF calls out the need for randomized trials to compare screening programs. What is the current status of this type of research?
Ahlquist: Colonoscopy, which is the test most commonly used for colon cancer screening, has never undergone a completed randomized controlled trial. Assumptions about it have been imputed into cost-effectiveness models, and that’s why we have a number of modalities in the guidelines now.
The ideal is to conduct a randomized controlled trial of a screening intervention versus no intervention. That type of study has only been done for guaiac fecal occult blood testing—and that study took 20 years. Considering how rapidly technologies are advancing, nowadays that kind of study would be hard to do. By the time you finish conducting the study, the technology that you’ve studied may be outmoded and replaced by something that performs much better.
Randomized controlled trials are extremely expensive and time-consuming, and they have to deal with that issue of technology drift. We need to be as creative from the validation side as we are with the technology development itself.
One approach with potential involves the use of data from very rigorous cross-sectional studies, in which the sensitivity and specificity of a tool can be defined robustly, and then plugged into a health technology assessment model. The model, in turn, is anchored in older randomized controlled trials showing that screening interventions result in reduced cancer mortality and incidence. If a developer has a new tool that is more sensitive and more specific than those older tools, health technology assessment models can reveal pretty well how they would work within a screening application.
Again, the ideal is hard to accomplish. There are a number of comparative studies underway between colonoscopy and fecal immunochemical tests (FIT), the new version of fecal blood testing. Those studies haven’t gotten to the mortality and incidence outcomes yet, but early outcomes look at lesion detection rates, complication rates, and participation rates. We can also learn from reviewing those intermediate end-points.
CLP: You are the coinventor of the Cologuard test, currently the only stool-based DNA screening test available in the United States. Does this test represent a platform that can be refined for improved performance relative to other tests?
Ahlquist: Absolutely. Any technology can be improved or replaced, and that’s true with Cologuard. Can it be miniaturized? Can we detect polyps more accurately? Can it be cheaper? Faster? Can it be more accurate? It’s going to hard to improve on accuracy for cancer when the test is already at 92% to 100%.
All of these considerations represent ways to at least maintain current outstanding performance, while potentially improving some of those other elements. We’re already thinking about next-generation versions, but the current version performs very well, especially for cancer.
Patients with inflammatory bowel disease make up one of those high-risk groups. Our group has published a number of outcomes studies using stool DNA testing to detect precancer and cancer in that high-risk group and, just like for average-risk sporadic cancer, the sensitivities for cancer and high-grade dysplasia have ranged between 90% and 100%.
Notably, that platform uses only methylated DNA markers. It’s a simpler platform, using new markers that we’ve discovered, and we expect the marker combinations will become easier to automate and cheaper to process.
CLP: You are also involved in ongoing research to develop other test methods, including some intended to detect colorectal cancer biomarkers in blood or serum. Are any of these tests close to availability for clinical lab use?
Ahlquist: About 4 years ago, we published a direct comparison between stool DNA testing and plasma testing for septin 9, which is a methylated DNA marker that can now be measured using a commercially available test. The comparison was based on a blinded study using paired samples from the same patients.
In the study, the stool approach detected about twice as many cancers—the overall sensitivity for cancer was 90% versus about 50%. Moreover, the blood test didn’t detect any of the precancers, whereas the stool DNA test detected the majority of the very large polyps used in the trial.
At that point, we thought that the very biology of stool-testing had advantages over blood. Lesions, polyps, and early-stage cancers shed or exfoliate cells into the colon lumen and into the stool, before they invade the bloodstream. So, based on some of those earlier observations, the biology favored stool-testing. That said, we continued to consider what would need to happen to get to a blood test that worked, and we came up with three things:
- First, exquisite analytical sensitivity on a level that wasn’t available even a few years ago. I think it’s possible now to get down to a very low-abundance detection range, which is what needs to happen for early-stage cancer.
- Second, markers that are absolutely discriminant, so that the test doesn’t produce false positives in noncancer, and is uniformly positive with cancer. With new whole-genome, whole-methylome discovery engines, we now have those markers, as do others.
- Finally, researchers who venture into blood without already having markers that are also site-specific are really opening Pandora’s box. Developers need markers that will connect their test results to the organ of origin, the site where the tumor is located. Otherwise, this sort of blood testing can become a very expensive shotgunning method of data collection, without a clear path for downstream evaluation. And, how do you tell a patient, “you have cancer somewhere, but we’re not sure where it’s located, or what we have to do to find it.”
For this whole approach to be attractive and cost-effective, researchers have to demonstrate that they have satisfied each of these three areas of consideration: high analytical sensitivity, high discrimination, and site specificity.
To forward the development of potential blood- and serum-based tests, we’ve been doing very rigorous discovery efforts across multiple cancers, and we are very excited by this research. At last year’s annual meeting of the American Association for Cancer Research, for instance, we reported on our work to sequence all of the gastrointestinal (GI) cancers. Now, we have marker panels that not only universally detect all GI cancers, but also detect marker subsets that accurately point to the individual organ from which a marker arose.
As part of that same presentation, we reported on a pilot study using blood samples. We found that we could almost perfectly discriminate colon cancer, pancreatic cancer, and normal populations.
Our vision—and the biggest idea in all of cancer screening—is to have a universal test for all cancers that would also be individualized to site specificity. We and others are working toward that goal, but there’s a lot of work to do.
CLP: Will the even-handed approach of the USPSTF guidelines encourage more researchers to develop tests for colorectal cancer, hoping that theirs will become “the test that gets done” most often?
Ahlquist: In some ways, the lack of innovation in cancer screening has been due not so much to a lack of imagination or ideas, as to the very cumbersome and arduous regulatory gauntlet that companies have to go through. Validation studies for screening tests take a long time, and going through FDA’s premarket review process can be a big, long, expensive undertaking. But perhaps the USPSTF recommendations can become the first step toward reengineering this whole pathway, in a way that doesn’t compromise quality, but accelerates the translation of meaningful innovation to the patient.
USPSTF and its recommendations could be part of a very large scale conversation among government regulators, professional societies, and other healthcare stakeholders, to discuss how testing can be reengineered, with a view toward doing what’s best for the patient as well as for society. I would love to see a conversation like that taking place at a macro level. I’m hoping it will, and that maybe our institution can play a role in moving that along.
Whatever is done, high quality must continue to be planned, tested, and documented. Companies should be required to pass a minimum performance threshold. But again, I’m hoping that we can speed this whole process.
CLP: FDA recently issued new guidelines on how it intends to handle next-generation sequencing tests and some other molecular tests, suggesting that it might take a more flexible approach. Are you encouraged by that kind of trend at the agency?
Ahlquist: Absolutely. I think that’s a foot in the door to having this larger conversation.
CLP: Are there additional approaches to colorectal cancer screening that you consider promising, but are still in their early stages?
Ahlquist: I’m not sure about promising, but there are some approaches that seem possible and very intriguing.
For instance, there are a number of blood-surrogate markers that are of sufficiently small size that they can passively get into saliva. And if they can be measured from saliva specimens, perhaps they can also be measured from urine specimens. The fragment size of the marker may be small enough to get through the kidney’s filters, but still be detectable in urine. Compared to undergoing venipuncture for a blood draw, such an approach would require samples that are easier for a patient to collect at home and send to the lab. But so far, blood tests have performed better than markers in either of those media.
There’s also some interesting work going on in breath testing—looking at the tens of thousands of aromatic metabolites found in breath. Various groups have looked at using bioinformatic profiling to find metabolites in breath that will enable them to detect lung cancer, esophageal cancer, or colon cancer. The research is still in its early stages, but shouldn’t be disregarded. The technique offers a very simple and attractive approach to sample collection, which is noninvasive, requires no preparation, and could be a very quick turnaround. But there is still a lot of work left to do before such a breath test can ever be brought to market. Much more research data will be needed to determine whether such a breath test can ever be sensitive or specific enough.
Some investigators have proposed the use of nanobiosensors much smaller than a blood cell, which can be injected to circulate with the blood and find tumors. When these nanobiosensors find a tumor, they congregate and send out a beacon indicating that a precancer or cancer is present. Already, some researchers have begun to study this technique in animal models.
Some researchers are exploring scenarios that seem fresh out of Star Trek, like being able to scan patients for disease-related electromagnetic field changes. Imagination may be the only limiting factor on the way that future testing is going to look, but I don’t think most of these ideas are on the near-term horizon.
In the near term, there’s still a great deal that can be done with blood, and with innovations for stool or urine testing. The novel approaches in these areas are already more tangible and more near-term. I think those are the areas where we’re going to see the next generations of innovation—but 20 years from now, who knows?
CLP: Are there other ways that advancing technologies in this field are improving the quality and reliability of cancer screening?
Ahlquist: When I looked at this field several years ago, all of the available tools were largely subjective or operator-dependent, resulting in quality that one would have to consider variable at best. If an endoscopist is having a bad day, their examinations might not be as good as when they’re wide awake and fresh. In fact, there are reams of data indicating such variance. For example, polyp detection rates vary widely across endoscopists in the same institution looking at the same population of patients.
Current approaches are all biased toward detection of left-sided disease. That’s important, because colon cancer is moving increasingly to the right side, and currently about half of all colon cancer occurs on the right side of the colon. Based on the results of a number of large studies, even colonoscopy detects left-side lesions better than right-side lesions.
A nice thing about automated laboratory tests—whether it’s Cologuard or any other test with FDA market authorization—is that they must meet agency requirements for safety and efficacy. Companies must demonstrate reproducibility down to minimal variability, and across the same sample when tested at three different labs. With Cologuard, we were able to meet these high standards. This is a test that is performed with the same high quality every time.
CLP: How do you see advancing technologies changing the role of clinical laboratorians?
Ahlquist: Where laboratorians have a big role in contributing to the quality of colon cancer screening is by performing high-quality laboratory tests that are not susceptible to operator variability, and where there’s high throughput, and ready access. Increasingly, I think, these screening tests for colorectal cancer, as well as for other cancers, will be noninvasive, and often use molecular technologies. This whole field will be laboratory based, and because of that it allows us to reimagine cancer screening.
Over time, the role of the laboratory diagnostician or technologist in screening is only going to increase. If we’re going to screen for multiple cancers, it will make less sense not to screen for low-prevalence lesions simply because they’re not common enough. If you’re screening people rather than individual organs, it’s the aggregate prevalence that’s relevant. So some of these less-common cancers—esophageal, kidney, liver, stomach, and so on—may end up becoming part of our population screening target, because they can all be included in the same test.
Steve Halasey is chief editor of CLP.
- US Preventive Services Task Force. Screening for colorectal cancer: US Preventive Services Task Force recommendation statement. JAMA. 2016;315(23):2564–2575; doi: 10.1001/jama.2016.5989.