Both the number and size of nonadvanced adenomas should determine surveillance colonoscopy intervals, with patients harbouring even one 6-9 mm polyp facing significantly higher risk, a large Korean study has found.

The study, published in Gastroenterology [link here], followed 16,870 adults aged 50 to 75 at a Seoul National University Hospital  health screening centre in South Korea over a median 5.3 years. It is the largest cohort to examine advanced colorectal neoplasm (ACN) risk by both adenoma number and size simultaneously.

The headline finding: five-year ACN risk rose sharply with nonadvanced adenoma (NAA) count, from 1.3% in those with no adenoma, to 4.0% with one or two NAAs, 5.3% with three or four, and 10.1% in those with five to ten.

But the size data added a new layer of clinical nuance. Among patients with one or two NAAs, five-year ACN incidence was 5.5% when at least one adenoma measured 6 to 9 mm, compared with 3.7% when all were 5 mm or less. The gap widened in the three to four NAA group: 8.1% versus 3.0%.

Hazard ratios told a similar story. Compared with those carrying only sub-5 mm adenomas in the one to two NAA group, patients with at least one 6 to 9 mm adenoma had an adjusted HR of 1.38 (95% CI, 1.09 to 1.76; P=0.008). Those with three to four NAAs but at least one 6 to 9 mm adenoma had an HR of 2.39 (95% CI, 1.70 to 3.36; P<0.001).

Critically, three to four all-diminutive NAAs carried no significantly elevated risk compared with one to two all-diminutive NAAs (adjusted HR 1.10; 95% CI, 0.71 to 1.70; P=0.680).

The authors concluded that “both NAA number and size should be considered when determining surveillance intervals.”

The five to ten NAA group carried the heaviest burden regardless of size, with adjusted hazard ratios of 2.80 and 2.58 for all-diminutive and at least one 6 to 9 mm adenoma subgroups respectively, compared with the one to two all-diminutive reference group.

The findings sit in some tension with existing major guidelines. The 2020 US Multi-Society Task Force on Colorectal Cancer recommends three to five year follow-up for patients with three to four NAAs, while the European Society of Gastrointestinal Endoscopy guideline simply returns patients with one to four NAAs to routine screening. Neither guideline formally incorporates NAA size as a stratification variable.

Earlier studies had produced inconsistent results for the three to four NAA group specifically. One single-centre study of 1,414 patients found ACN incidence did not differ between one to two and three to four NAA groups (1.4% vs 1.8%, P=0.666). A separate cohort of 2,570 patients similarly found no significant difference.

The Korean team argued their larger sample and longer follow-up explain the discrepancy, and that differing proportions of 6 to 9 mm versus sub-5 mm adenomas across study populations may have muddied earlier results.

On the question of whether the risk gradient is an artefact of informative censoring, a persistent methodological challenge in surveillance colonoscopy research where adenoma removal during follow-up alters subsequent risk, the authors applied inverse probability of censoring weighting (IPCW) and gamma-imputation sensitivity analyses. In global sensitivity analysis, no tipping point was observed up to a hazard multiplier of r=3, suggesting the main findings are robust.

The accelerated failure time model added further texture, showing progressively shorter times to reach 7.5% cumulative ACN incidence: 11.21 years for no adenoma, 7.67 years for one to two NAAs, 6.05 years for three to four NAAs, and 4.34 years for five to ten NAAs.

There are limitations worth noting. The study was conducted at a single high-performing centre where the adenoma detection rate ran at approximately 40 to 48%, well above average. The authors acknowledged that “high accessibility and low cost of healthcare in Korea may have resulted in more frequent colonoscopies than recommended.” Selection bias towards health-conscious individuals cannot be excluded. Only 24 colorectal cancers were recorded across the entire cohort, too few for reliable subgroup analysis.

The authors also flagged a broader conceptual concern for guideline developers: the traditional “high-risk adenoma” definition, ACA or three or more NAAs, “may have limited discriminatory value and warrants reconsideration,” given that the five to ten NAA group carries risk approximately three times that of the three to four NAA group.