The cancer screening community was jolted by the recently published results of the NORDICC trial: a large, pragmatic, randomized trial comparing colonoscopy screening versus no screening conducted in Poland, Norway, Sweden, and the Netherlands. This huge study randomized 28,220 screening-age men and women to an invitation to colonoscopy and 56,365 to usual care (no invitation to screening). Participants were followed for a median of 10 years. The risk of colorectal cancer at 10 years was 0.98% in the invited group versus 1.20% in the usual-care group, a statistically significant difference. Unfortunately, the risk of dying from colorectal cancer was only 10% lower in the invitation group, a nonsignificant change.¹
Many editorials have speculated on the reasons why a screening program built on colonoscopy—in theory, our most effective cancer screening technology–“failed” to reduce mortality from colorectal cancer. Posited reasons include insufficient follow-up time and missed adenomas due to variable operator skill, among others.²ʼ³ʼ⁴ I am sure that many of these speculations are correct. Here, I want to focus on a critical result from NORDICC that has relevance for the growing field of multi-cancer early detection (MCED) testing: only 42% of the group invited to have a colonoscopy accepted the invitation. How is this statistic relevant to MCED, since the great majority of MCED tests are blood tests? Surely, patients will be far less reluctant to get a simple blood test versus enduring the unpleasant experience of “bowel preparation” (harsh laxatives for 24 hours) and having a large, flexible tube inserted deep into one’s colon (I speak from experience). The answer is easily summed up in the immortal words of Yogi Berra: “You’ve got to be very careful if you don’t know where you are going, because you might not get there.”
In other words, developing, testing, and in some cases marketing MCED tests without knowing how they will work in practice runs the risk of having a number of very expensive tests in practice with only the hope that they are actually helping people.
“Developing, testing, and in some cases
marketing MCED tests without knowing how
they will work in practice runs the risk of
having a number of very expensive tests in
practice with only the hope that they are
actually helping people”
Briefly, MCED tests detect multiple components of a growing cancer, such as circulating tumor cells, tumor DNA, and other analytes, in blood or other body fluids. Multiple cancers can be detected in a single test, in theory, at an earlier stage of cancer than the patient would otherwise have at the time they develop symptoms. At least 20 MCED tests are currently in various stages of development.⁵
For all the criticism leveled at NORDICC, I see one huge benefit: it did not assume that the general public is as enthusiastic about cancer screening as the screening community believes they are. NORDICC investigators, wisely, started randomization with the invitation to screen. This allowed comparison of those who choose to screen versus those who don’t as well as (secondarily) evaluating the impact of colonoscopy among those who actually went through the procedure. The lesson to me for MCED testing is this: taking a blood sample and even showing patients the results of their test does not mean that there will be a benefit: clinicians have to effectively follow-up after a “positive” test, and patients have to agree to that follow-up advice, all the way through diagnosis and treatment. We can’t assume that this will go smoothly.
At this early stage, MCED test developers are understandably concerned with demonstrating the efficacy of their tests: how they perform under idealized conditions. We can expect to see studies showing sensitivity and specificity for these tests among volunteers recruited into carefully managed studies run by expert clinicians, with follow-up evaluations for positive tests approaching 100% and expedient referrals for biopsy-proven cancers. The real world will be far messier. Results will sometimes be difficult to interpret (e.g., indeterminant). It might not be obvious what the correct follow-up test should be. This is a big concern of mine: there are very few well-described protocols for follow-up evaluation in the current batch of MCED screening studies. It is essentially up to the physician to decide what is best. Finally, and most importantly, patients might decide not to follow-up at all. I am particularly concerned about this issue for low-income, poorly insured and underserved populations. As a case in point, researchers have found that only 55% of patients in an integrated safety net system who had abnormal fecal immunochemical test for colorectal cancer had a follow-up diagnostic colonoscopy within one year.⁶
“Health technology assessors and payers
are going to demand that developers
account for the messiness that will come
with implementing MCED screening”
Based on such experience, we should all be concerned that MCED testing will widen the disparity between the haves and have-nots of access to the health care system. Health technology assessors and payers are going to demand that developers account for the messiness that will come with implementing MCED screening. Given that we won’t see a NORDICC-like study for MCEDs for years if not decades, how can we bridge the gap between the evidence we will have and the evidence we will need? This is where simulation modeling has a role. Developers should create models now that account for the expected real-world imperfections of screening programs, and to show how this could influence the effectiveness of MCED testing. It is in their interest to do so: models can help identify the potential “pressure points” of screening, which can then be used to develop programs to study and address them.
Why does this matter? At the expected price of $800-$1500 per test, population-wide MCED screening has the potential to be a massive strain on healthcare budgets. Absent evidence from randomized trials (there currently only one RCT evaluating an MCED test, and the preliminary results are expected in 2023)⁷, payers are going to be very reluctant to make coverage decisions that would be viewed as favorable by manufacturers. In my view, the way in for test companies is to agree to coverage with evidence development programs: agreements to pay for tests only in the context of a prospective study aimed at evaluating their effectiveness in practice. The NORDICC trial provides a template for how such studies might be designed.
“The way in for test companies is to agree to coverage with evidence development programs”
In summary, NORDICC has provided sobering evidence that even the most effective cancer screening tests can yield underwhelming results in the real world. For all their promise, MCED tests are going to enter a cancer screening world that is an expensive, messy business.
We owe the many stakeholders in screening—most importantly patients—a realistic assessment of how these tests will influence cancer and cancer outcomes. Augmenting evidence from trials, high quality models can give the best estimates of what we can expect from these wonderfully promising technologies.
–Scott Ramsey, MD, PhD
Senior Partner and Chief Medical Officer, Curta
REFERENCES
- Bretthauer M, Løberg M, Wieszczy P, et al. Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related Death. N Engl J Med. Published online ahead of print October 9, 2022. doi: 10.1056/NEJMoa2208375.
- Dominitz JA, Robertson DR. Understanding the Results of a Randomized Trial of Screening Colonoscopy. NEJM. Published online ahead of print October 9, 2022. doi: 10.1056/NEJMe2211595
- Vamulapalli R. NordICC Colonoscopy Study Results Should be Viewed with Caution—Longer Follow-up Needed. Cancer Letter. 2022;48(37).
- Dahut WL. What Did We REALLY Learn from the NordICC Colonoscopy Study? Cancer Letter. 2022;48(37).
- Kaiser J. ‘The Complexities are Staggering.’ U.S. Plans Huge Trial of Blood Tests for Multiple Cancers. June 22, 2022. https://www.science.org/content/article. /complexities-are-staggering-u-s-plans-huge-trial-blood-tests-multiple-cancers. doi: 10.1126/science.add6151
- Issaka RB, Singh MH, Oshima SM, et al. Inadequate Utilization of Diagnostic Colonoscopy Following Abnormal FIT Results in an Integrated Safety-Net System. Am J Gastroenterol. 2017;112(2):375-382. doi: 10.1038/ajg.2016.555.
- GRAIL and National Health Service (NHS) England Complete Enrollment of 140,000 Participants in Largest Study of Multi-Cancer Early Detection Test. Press Release. July 18, 2022. https://grail.com/press-releases/grail-and-national-health-service-nhs-england-complete-enrollment-of-140000-participants-in-largest-study-of-multi-cancer-early-detection-test/