Endoscopic mucosal resection (EMR) is the standard treatment for large (≥20 mm) non-pedunculated colorectal polyps (LNPCPs). Current guidelines (BSG/ACPGBI/PHE 2020; ESGE 2020/2022; USMSTF 2020) recommend surveillance at approximately 6, 18, and 36 months post-EMR, but these intervals are based on expert consensus rather than interval-specific recurrence evidence. The clinical value of a second surveillance colonoscopy (SC2) after a negative first surveillance colonoscopy (SC1) remains unclear. Clarifying this could inform risk-adapted surveillance strategies. We aimed to quantify recurrence rates at SC1 (3–6 months) and SC2 (12–18 months) following EMR of benign colorectal lesions ≥20 mm.

We systematically searched Ovid MEDLINE, Embase, and the Cochrane Library (1990–2025) for studies reporting recurrence at SC1 and/or SC2 after hot EMR of benign LNPCPs ≥20 mm. We excluded studies without interval-specific outcomes and those involving cold EMR, serrated-only cohorts, or invasive cancer. SC1 was defined as surveillance at 3–6 months and SC2 as 12–18 months. Only lesions without recurrence at SC1 were eligible for SC2 analysis. Due to limited interval-specific data (three studies for SC1; one for SC2), formal quantitative meta-analysis was not feasible, so recurrence estimates are presented descriptively. The Sydney EMR Recurrence Tool (SERT) was synthesised narratively because only one study reported interval-specific SERT outcomes.

Recurrence at SC1 following hot EMR for benign LNPCPs ≥20 mm was consistently reported within the 3–6-month window, typically 10–20%, giving a descriptive estimate of approximately 15%. Among lesions recurrence-free at SC1, the conditional risk of recurrence at SC2 was low at approximately 3%. SC2 follow-up was incomplete (~25% of eligible patients did not undergo surveillance), but in the available multicentre dataset this low recurrence estimate appeared consistent. Most recurrences at both SC1 and SC2 (>90%) were managed endoscopically, with surgery required in only a small minority of cases (~5%). Interval colorectal cancer following benign EMR was very rare (<1%). Across studies reporting risk factors, recurrence was more likely after piecemeal resection (particularly where ≥3 fragments were required) and was also associated with lesion size ≥40 mm, right-sided location, and tubulovillous or villous histology. High-grade dysplasia was associated with recurrence in several, but not all, multivariate analyses.

One study provided recurrence outcomes stratified by SERT. Recurrence at the first follow-up examination, typically performed at 12–18 months, was low among SERT=0 lesions (11.6%) and substantially higher among SERT≥1 lesions (34.9–36.3%). Late recurrence remained uncommon for SERT=0 lesions at a median follow-up of 22 months. Approximately one-third of lesions were SERT=0. Variability in follow-up timing and the absence of additional interval-specific datasets prevented pooling of SERT-based results.

Most recurrence after EMR of benign ≥20 mm colorectal polyps is detected at the first surveillance colonoscopy. Conditional recurrence after a negative SC1 is low (~3%) at 12–18 months, supporting further exploration of risk-adapted surveillance strategies. SERT may help identify a low-risk subgroup suitable for reduced surveillance intensity, but interval-specific evidence is currently limited to a single dataset. Prospective multicentre studies incorporating predefined conditional surveillance pathways and strict follow-up are required before guideline-recommended surveillance intervals can be safely revised.