Highlights
– In a French multicentre ESD registry of 3,310 large non‑pedunculated colorectal polyps, adjusted submucosal invasive cancer (SMIC) rates were similar in rectal and colonic lesions (9.8% vs 8.9%, p=0.52) after propensity score matching.
– En bloc and R0 resection rates with ESD were equivalent for rectal and colonic lesions; there was a non‑significant trend to higher perforation and surgery for complications in colonic procedures.
– The authors conclude that lesion risk features (size, morphology, optical diagnosis), rather than anatomical location alone, should guide the choice of endoscopic resection technique.
Background
Large non‑pedunculated colorectal polyps are a heterogeneous group of lesions that carry variable risks of submucosal invasion and subsequent cancer. Endoscopic submucosal dissection (ESD) enables en bloc resection of large lesions and precise histopathological staging, which is important when submucosal invasive cancer (SMIC) is suspected. Historically, many endoscopists and guideline documents have favoured ESD for large rectal lesions because the rectum was perceived to have a higher prevalence of SMIC and because surgical salvage for rectal disease may carry particular morbidity. However, whether rectal location independently confers higher inherent SMIC risk compared with the colon remains uncertain.
Study design
This analysis used prospectively collected data from the French ESD registry (FECCo Group) over a three‑year period (September 2019 to September 2022). All large non‑pedunculated colorectal polyps treated with ESD across 13 centres were eligible. After exclusions, 3,310 lesions were analysed. Key lesion descriptors collected included size, morphology (granular versus non‑granular, presence of mixed nodularity), and anatomical location (rectum versus colon). Procedural outcomes recorded included en bloc resection, R0 resection (histologically negative margins), perforation rates, and need for surgery for complications. The primary comparison was SMIC prevalence in rectal versus colonic lesions.
To mitigate selection bias and confounding, the investigators applied propensity score matching (PSM) and inverse probability weighting (IPW) to balance lesion‑level covariates such as size and morphology between rectal and colonic groups. A pre‑planned subgroup analysis examined lesions managed in the three highest‑volume centres where large non‑pedunculated polyps were treated exclusively with ESD; this assessed consistency of findings in high‑volume, uniform‑practice settings.
Key findings
Population and lesion characteristics: Of the 3,310 lesions analysed, rectal lesions were, on average, larger and more often granular with mixed nodularity than colonic lesions (median size 56 mm [IQR 40–75] vs 47 mm [IQR 37–62]; granular 80.0% vs 59.4%; mixed nodular 54.0% vs 32.5%; all p<0.001).
SMIC prevalence (unadjusted): In crude comparisons, rectal lesions appeared to have higher SMIC rates, reflecting their larger size and different morphology. Because these lesion features are established predictors of SMIC, raw comparisons risk confounding.
SMIC prevalence (adjusted): After propensity score matching to balance key lesion attributes, the SMIC rate was 9.8% for rectal lesions and 8.9% for colonic lesions (p=0.52), indicating no statistically significant difference when lesion size and morphology were accounted for. Results were similar after inverse probability weighting.
Procedural outcomes: En bloc resection rates were comparable (rectum 97.7% vs colon 97.3%, p=0.757) as were R0 rates (rectum 89.7% vs colon 89.5%, p=0.937). Perforation rates trended higher for colonic ESD (5.5% rectum vs 7.9% colon, p=0.057), and surgery for complications also showed a non‑significant trend to be greater in the colon (0.1% rectum vs 1.1% colon, p=0.051). The subgroup of three high‑volume centres that used ESD for all large non‑pedunculated polyps reproduced these findings.
Interpretation of effect sizes: The adjusted absolute differences in SMIC prevalence were small and statistically nonsignificant. High en bloc and R0 rates in both locations support the technical feasibility of ESD when performed by experienced teams. The trend toward higher perforation in the colon is consistent with known technical challenges of colonic ESD, including thinner wall and scope maneuverability.
Expert commentary and clinical context
These findings challenge the common practice of using anatomical location (rectum vs colon) as a primary determinant for choosing ESD. International guidelines (for example, ESGE guidance on colorectal endoscopic resection and ESD) emphasise lesion‑level risk stratification—size, surface morphology (granular vs non‑granular), pit pattern, and vascular pattern on advanced imaging—to decide between EMR, piecemeal EMR, and ESD for en bloc resection and accurate staging. The registry data corroborate this principle by showing that when lesion features are balanced, rectal location per se is not associated with higher SMIC prevalence.
Practical implications:
– Preprocedural optical diagnosis (high‑definition white light, NBI/virtual chromoendoscopy, magnification pit/vascular pattern assessment) remains critical to identify high‑risk features that would favour en bloc resection by ESD.
– Institutional and operator expertise remain essential. Although rectal and colonic en bloc/R0 rates were similar in this large registry, the trend to higher perforation in colonic ESD underscores the need for appropriate case selection aligned to operator skill and access to prompt surgical backup.
– Shared decision‑making with patients should include discussion of lesion‑specific risk features, the aims of resection (curative intent vs staging), and local ESD experience and complication rates.
Mechanistic and biological plausibility
There is no compelling biological rationale that the rectum should uniformly harbor more invasive disease independent of lesion features. Differences previously reported between rectal and colonic lesions likely reflect referral patterns (rectal lesions being preferentially referred when large or complex), variation in lesion morphology distribution, and historical selection bias rather than an inherent locational risk. The present registry, by adjusting for key lesion attributes, provides evidence that lesion biology and morphology are the primary drivers of SMIC risk.
Strengths and limitations
Strengths:
– Large, multicentre prospective registry with >3,000 analysed lesions
– Rigorous statistical adjustment for confounding using PSM and IPW
– Subgroup analysis from centres with homogeneous practice patterns
– Detailed procedural and histopathological endpoints
Limitations:
– Registry design is observational and cannot eliminate residual confounding, especially from unmeasured variables (e.g., precise optical diagnosis metrics, inter‑observer pathology differences).
– Operator experience varied across centres and could influence both selection and outcomes; while subgroup analysis mitigates this, generalisability depends on local expertise.
– The analysis focuses on immediate procedural and pathological outcomes; long‑term outcomes such as recurrence, cancer‑specific survival, and quality of life were not reported here.
– Pathology assessment of submucosal invasion depth and grading can vary; standardised central pathology review was not described in the summary.
Clinical takeaways
1. Choice of resection technique for large non‑pedunculated colorectal polyps should be guided primarily by lesion risk features (size, non‑granular morphology, depressed areas, suspicious pit/vascular pattern), not by rectal versus colonic location alone.
2. When performed by experienced endoscopists, ESD achieves high en bloc and R0 rates in both rectum and colon; however, colonic ESD may carry a modestly higher risk of perforation—case selection and access to surgical backup remain important.
3. Centres should continue to develop robust preprocedural optical diagnostic workflows and multidisciplinary pathways so that lesions with high‑risk features are routed to operators and techniques most likely to provide curative resection and accurate staging.
Conclusion
The French ESD registry provides important, large‑scale evidence that rectal location per se is not an independent predictor of submucosal invasive cancer in large non‑pedunculated colorectal polyps when lesion size and morphology are accounted for. These data support a paradigm that emphasises lesion‑level risk stratification over anatomical location for selecting endoscopic resection strategies. Implementation requires high‑quality optical diagnosis, operator expertise, and institutional pathways that balance oncologic safety and procedural risk.
Funding and clinicaltrials.gov
Trial registration number: NCT04592003. Funding details were reported in the original publication (Van der Voort et al., Gut 2025).
References
1. Van der Voort V, Schaefer M, Wallenhorst T, et al; FECCo Group collaborators. Rectal versus colonic submucosal cancer rates and procedural outcomes in large non‑pedunculated polyps: French ESD registry data. Gut. 2025 Oct 12. doi:10.1136/gutjnl-2024-332970. Epub ahead of print. PMID: 41083216.
2. Pimentel‑Nunes P, Dinis‑Ribeiro M, Ponchon T, et al. Endoscopic submucosal dissection: European Society of Gastrointestinal Endoscopy (ESGE) guideline. Endoscopy. 2015;47(9):829–854. (Guidance on indications, technique and training for ESD.)
3. Ferlitsch M, Moss A, Hassan C, et al. Colorectal polypectomy and endoscopic mucosal resection (EMR): European Society of Gastrointestinal Endoscopy (ESGE) Clinical Guideline. Endoscopy. 2017;49(3):270–297. (EMR/ESD selection considerations.)
4. Burgess NG, Hourigan LF, Zanati SA, et al. Local recurrence after endoscopic resection of non‑pedunculated colorectal polyps: systematic review and meta‑analysis. Gut. 2018;67(1):32–40. (Context for recurrence risk and importance of en bloc/R0 resection.)
5. Saito Y, Uraoka T, Yamaguchi Y, et al. Endoscopic submucosal dissection for colorectal neoplasms: clinical outcomes and perspectives. Gut Liver. 2012;6(2):143–149. (Early large series on colorectal ESD outcomes.)