The EPoS II Trial Just Made the 3-Year Colonoscopy Standard Look Like a Reflex, Not a Rule

A gastroenterologist in Oslo pulls up her post-polypectomy surveillance schedule on a Tuesday morning in September 2026 and sees what she has always seen: high-risk adenoma patients slotted for a return colonoscopy at three years. That interval has been the anchor of post-polypectomy care for so long that most clinicians treat it not as a guideline but as a biological truth. Then she opens the September 17 issue of the New England Journal of Medicine and reads the EPoS II interim analysis. Her three-year assumption does not survive the morning.

The EPoS II trial, a randomized controlled study examining colonoscopy surveillance intervals after the removal of high-risk adenomas, found that initiating surveillance at five years is noninferior to surveillance at three years in terms of cumulative colorectal cancer incidence. The trial, published in NEJM Volume 395, Issue 11, published in NEJM Volume 395, Issue 11, delivers this finding not from a registry or a claims database but from a prospective, multicenter RCT with pre-specified noninferiority margins. That distinction matters enormously, both for clinical practice and for anyone designing the next generation of colorectal cancer prevention trials.

The reflex has been wrong. Now comes the harder question: what do sponsors, protocol designers, and regulators do with that knowledge?

Why the Three-Year Interval Survived This Long

The durability of the three-year post-polypectomy interval reflects a phenomenon familiar to anyone who has watched screening medicine evolve: conservative defaults calcify into standards before the confirmatory evidence arrives. The interval was never derived from a powered RCT comparing three years against five. It emerged from expert consensus, biologic plausibility arguments about adenoma growth kinetics, and the understandable clinical instinct that returning sooner is safer. That instinct is not unreasonable. It is just untested.

The European Society of Gastrointestinal Endoscopy addressed this directly in its Post-Polypectomy Colonoscopy Surveillance: ESGE Guideline, updated in 2020, which stratified surveillance recommendations by adenoma risk category. High-risk patients received a three-year return recommendation under the ESGE guideline. The guideline acknowledged the limited RCT evidence base but defaulted to the shorter interval on precautionary grounds. The USPSTF, whose 2021 colorectal cancer screening recommendations focus on average-risk adults rather than post-polypectomy surveillance, left the adenoma interval question to specialty societies, creating a guidance gap that EPoS II now fills with prospective data for the first time.

The cost consequences of that gap have been substantial. A modeling study by Saini, Schoenfeld, and Vijan, reviewed by the VA Health Services Research and Development program, found that a strategy involving routine screening every ten years for average-risk patients carried an incremental cost-effectiveness ratio of $5,743 per quality-adjusted life year gained, which the authors flagged as highly cost-effective. Post-polypectomy surveillance at three-year intervals for high-risk patients multiplies procedure volume far beyond that baseline. If the five-year interval produces equivalent cancer incidence outcomes, the system has been spending endoscopy capacity, patient time, and payer dollars on a schedule that the evidence never required.

But cost efficiency is not the headline. Noninferiority in cancer incidence is. The distinction matters because a cost argument alone invites the countercharge that the system is rationing safety. EPoS II does not make a cost argument. It makes an outcomes argument, which is a categorically different kind of evidence.

What EPoS II Reveals About RCT Design in Screening Medicine

Open any standard colorectal cancer prevention trial protocol from the last decade and you will find surveillance intervals treated as fixed background conditions, not as variables to be tested. The trial assumes an interval, builds a follow-up schedule around it, and hands investigators a colonoscopy window that has never been formally validated for the population being studied. EPoS II treats the interval itself as the experimental variable, which requires a design logic most prevention trials actively avoid.

Noninferiority trials in screening medicine carry a particular methodological burden. The margin must be pre-specified before unblinding, the margin must be clinically motivated rather than statistically convenient, and the primary endpoint must be hard enough that a null result cannot be dismissed as a surrogate problem. Cancer incidence satisfies that last requirement. A surrogate endpoint like adenoma recurrence rate at surveillance colonoscopy would not have settled the clinical question EPoS II is answering, because adenoma recurrence at five years versus three years says nothing definitive about cancer risk in the interval between procedures. The EPoS II investigators chose the harder endpoint and designed a trial long enough to observe it. That design choice is itself a lesson for sponsors thinking about the next wave of chemoprevention or surveillance optimization studies.

The multicenter prospective cohort study registered to develop a prediction model for colorectal adenoma recurrence and malignant transformation risk is pursuing a complementary approach: rather than testing a fixed alternative interval, it aims to identify patient-level predictors that could support individualized surveillance timing. That model-building approach and the EPoS II RCT represent two ends of the evidence pipeline. EPoS II tells you that five years is as safe as three years at the population level. The prediction model work aims to tell you which individual patients within that population might still benefit from the shorter interval. Both are necessary, and neither is sufficient alone.

Sponsors who understand that distinction will build better trials. Those who treat EPoS II as a simple “five years is fine” signal and stop there will design protocols that the next prediction-model paper immediately complicates.

The Operational Reckoning for Trial Sponsors and Sites

Consider the immediate protocol implications for any ongoing colorectal cancer prevention trial that uses post-polypectomy surveillance as an eligibility criterion or a stratification variable. If investigators enrolled patients on the assumption that high-risk adenoma patients return at three years, and EPoS II now establishes five-year equivalence, the sponsor faces a materially changed background standard. An FDA reviewer evaluating a prevention trial’s control arm design has every reason to ask whether the surveillance interval embedded in the protocol reflects current clinical practice, especially if that protocol was written against pre-EPoS II guidance.

The FDA Breakthrough Device Designation granted to Amadix’s PreveCol blood test on January 23, 2024, for detection of precancerous colorectal lesions offers a concrete illustration of where this evidence gap creates downstream regulatory complexity. A blood-based screening tool validated against a surveillance paradigm that assumes three-year colonoscopy intervals for high-risk patients will need its clinical utility claims re-examined if the standard of care shifts to five years. The comparator changes. The performance threshold changes. Any pending or in-preparation 510(k) or De Novo submission for a colorectal screening technology should be stress-tested against EPoS II findings before submission, because a reviewer who has read the NEJM paper will ask that question even if the sponsor does not anticipate it.

The adherence problem sharpens this further. A large-scale analysis of over 2.5 million outpatient screening colonoscopies in the United States between 2016 and 2019 found significant variability among endoscopists in adherence to recommended surveillance intervals. If the actual real-world interval for many high-risk patients already drifts toward or beyond five years due to scheduling gaps, patient non-adherence, and capacity constraints, then EPoS II is not just validating a longer interval, it is validating what is already happening in practice. That has direct implications for RWE studies using colonoscopy registries: the exposure classification between “three-year” and “five-year” patients in observational data may be far noisier than the protocol on paper suggests.

Sponsors designing surveillance optimization trials should now treat interval compliance as a primary analysis variable, not a protocol deviation category. Build your stratification around actual time-to-surveillance, not the assigned interval. Pre-specify a sensitivity analysis that censors patients whose actual surveillance occurred more than six months outside their randomized window. If you do not, a reviewer will demand it post-submission, and you will be reconstructing your statistical analysis plan under time pressure.

The gastroenterologist in Oslo closes the NEJM paper and looks again at her surveillance schedule. The three-year column is still there, populated with names and appointment dates. She knows that changing it requires a guideline update, a payer reimbursement decision, and probably a conversation with her hospital’s quality committee. EPoS II did not give her a switch to flip. It gave her the evidence to start that conversation, which is a different and considerably more durable kind of power. The sponsors who recognize that distinction before the guideline committees do will be the ones writing the next trial protocols, not reacting to them.

References

1. New England Journal of Medicine, “Colonoscopy Intervals and Colorectal Cancer Incidence after Adenoma Removal” (EPoS II), Volume 395, Issue 11, September 17, 2026
2. PubMed, “Post-polypectomy colonoscopy surveillance: European Society of Gastrointestinal Endoscopy (ESGE) Guideline – Update 2020”
3. U.S. Preventive Services Task Force, Colorectal Cancer Screening Recommendations, May 18, 2021
4. VA Health Services Research and Development, Saini, Schoenfeld, and Vijan cost-effectiveness modeling of colonoscopy screening intervals
5. Amadix, FDA Breakthrough Device Designation for PreveCol blood test, January 23, 2024
6. PMC, Adherence to colonoscopy surveillance intervals among U.S. endoscopists, 2016–2019 registry analysis
7. PMC, Multicenter prospective cohort study protocol: prediction model for colorectal adenoma recurrence and malignant transformation risk