Endoscopic resection is central to the management of colorectal neoplasia. EMR is widely available and technically simple, whereas ESD allows en-bloc and R0 resection but is more demanding. High-quality, long-term, real-world comparative data from European centers remain scarce.This study aimed to compare technical performance, histopathology, and long-term oncological outcomes of EMR versus ESD over a 10-year period (2015-2025). The primary endpoint was R0 resection. Secondary endpoints included en-bloc resection, recurrence, carcinoma detection, procedural time, fibrosis, complication rates, and follow-up completeness. Additional objectives included identifying independent predictors of carcinoma and local recurrence using multivariate modeling.

This retrospective cohort included 292 consecutive colorectal lesions resected at a single high-volume expert center (EMR n=166, 56.8%; ESD n=126, 43.2%) over 10 years. Demographic variables, lesion morphology (Paris, LST), anatomical location, size, fibrosis severity, use of traction and hood, procedural time, and adverse events were extracted. Histopathological endpoints included dysplasia grade and carcinoma. Follow-up duration, recurrence, and loss to follow-up were recorded. All procedures were performed by a single expert endoscopist to eliminate operator-dependent variability.

Lesions treated with EMR were distributed throughout the colon, most commonly right-sided (38.5%), whereas ESD was strongly concentrated in the rectum (66.7%, p<0.0001).Paris morphology differed significantly (p<0.001): EMR was used for protruding Is/IIa lesions, while ESD was preferentially applied to flat or mixed IIa/IIa+IIc lesions.LST morphology was evenly balanced between granular and non-granular subtypes in both groups, indicating no preferential selection based on LST subtype. Rectal lesions were significantly larger (50.6 vs 32.0 mm, p<10⁻¹¹) and required markedly longer procedure time (57.8 vs 26.5 min, p<10⁻¹²). Recurrence was higher in rectal lesions (20.7% vs 7.9%, p=0.0029), whereas en-bloc and R0 rates were similar between rectal and colonic locations (p=0.82 and p=0.93).

Fibrosis severity differed markedly (p<0.0001): EMR lesions were mostly non-fibrotic (81.3%), whereas ESD targeted lesions with higher fibrosis burden (46.0% moderate/severe vs 18.7%).Hood use was strongly associated with ESD (73.0% vs EMR 3.6%, p<0.0001). Traction was uncommon (1–2%, p=NS).

ESD achieved significantly higher en-bloc (84.1% vs 37.3%, p<0.0001) and R0 resection rates (65.9% vs 49.4%, p=0.006). Median procedure time was longer (45 vs 20 min, p<0.0001).Time-per-mm was almost doubled in ESD (1.17 vs 0.60 min/mm, p<0.0001).In multivariate analysis restricted to ESD, lesion size was the sole independent predictor of procedural time (β=1.16 min/mm, p<0.0001; ρ=0.71), while fibrosis showed a non-significant trend (β=+16.9 min, p=0.12).

ESD identified more high-grade dysplasia (38.9% vs 31.9%, p=0.004) and more than twice as many carcinomas (34.1% vs 12.0%, p<0.0001).Independent predictors of carcinoma were rectal location (OR≈2.1) and use of ESD (OR≈2.9). Lesion size, fibrosis, and LST subtype were not independent predictors.

Adverse events were infrequent and comparable (p=0.21): EMR had 11 intraprocedural bleeds (6.6%, all controlled endoscopically), while ESD had 4 intraprocedural events (3.2%) and one delayed bleed (0.8%). No patient required surgery.

Among patients with follow-up, EMR had longer surveillance (median 24 vs 8 months, p<0.001), while loss-to-follow-up was similar (38.6% vs 40.5%, p=0.73).

Recurrence was significantly higher after EMR (20.5% vs 5.6%, p<0.001).Independent predictors of recurrence were lesion size (p=0.005), rectal location (OR≈3.1), and en-bloc resection, the strongest protective factor (OR≈0.07, p<0.0001). Fibrosis, LST morphology, R0, and technique were not independent predictors.

In real-world practice over 10 years, ESD consistently outperformed EMR across all major oncological outcomes—en-bloc, R0, carcinoma detection, and recurrence—despite being used for significantly more complex lesions.