Endoscopic eradication therapy (EET), combining endoscopic mucosal resection (EMR) and radiofrequency ablation (RFA), is the standard treatment for Barrett’s oesophagus (BO) with dysplasia and early oesophageal adenocarcinoma. Nevertheless, the long-term durability of complete remission and the risk of recurrence remain only partially defined. This systematic review and meta-analysis evaluated the long-term outcomes of EET for BO-related neoplasia.

We systematically searched MEDLINE (n=1364), Embase (n=443), and the SCOPUS (n=2805) from inception to 20 July 2025 for studies reporting long-term outcomes following EET for BO neoplasia. Primary endpoint was recurrence rates of IM and dysplasia/cancer during surveillance. Proportions were pooled using a random-effects model (DerSimonian–Laird) with logit transformation.

Eight studies including 3,367 patients and ~7,500 person-years of follow-up met inclusion criteria; 75% were high-quality studies. Median follow-up was 39.5 months. Pooled CR-IM was 86.4% (95% CI: 81.2–90.5%; 8 studies, N=3,245) and CR-D was 94.0% (95% CI: 91.2–96.1%; 8 studies, N=3,245). Among patients achieving CR-IM, intestinal metaplasia recurred in 14.0% (95% CI: 8.0–23.5%; 7 studies, N=1,997) and dysplasia/cancer in 5.6% (95% CI: 3.5–8.9%; 8 studies, N=3,028). Any recurrence occurred in 18.5% (95% CI: 12.3–26.8%; 8 studies, N=3,028), corresponding to an annual dysplasia recurrence rate of ~1.5 per 100 person-years. Heterogeneity was considerable across outcomes (I² = 76–94%). In pooled treatment comparisons, dysplasia/cancer recurrence was lower after EMR+RFA than RFA alone (4.2% [95% CI: 2.8–6.3%] vs 8.9% [95% CI: 5.1–15.1%]; relative risk [RR] 2.1, 95% CI: 1.3–3.4; p=0.022). Baseline low-grade dysplasia was associated with lower dysplasia/cancer recurrence than HGD/T1a (3.1% [95% CI: 1.8–5.2%] vs 6.8% [95% CI: 4.1–11.0%]; RR 2.2, 95% CI: 1.1–4.4; p=0.043). Progression to advanced disease (T1a) occurred in 6–12% of recurrences, mostly managed endoscopically. Leave-one-out analyses showed pooled dysplasia/cancer recurrence remained within 4.8–6.2%, and omitting the most influential study (Cotton 2017) changed estimates by only 0.8%, indicating no single study drove the results. Temporal analyses suggested a learning-curve effect, with CR-IM improving from ~75% to 90–95% and dysplasia recurrence falling from ~12% to 3–5% between 2009 and 2025. Baujat plots identified Cotton 2017 as contributing 42.5% of total heterogeneity, while large recent cohorts (Wolfson 2022, van Munster 2022) strongly influenced but also stabilised pooled estimates. Meta-regression bubble plot for relationship between BO length and dysplasia recurrence rate showed weak positive but non-significant association (R²=0.18, p=0.23).

EET has shown long-term durability following successful eradication of BO with neoplasia. Although recurrence is not uncommon, it is usually detected early and amenable to further endoscopic treatment. Our findings do not support frequent surveillance within the first year and suggest a personalised surveillance strategy based on the patient’s risk of recurrence, in order to minimise the burden of surveillance gastroscopy.