Background: The proportion of oesophageal adenocarcinoma that is detected concurrently with initial Barrett's oesophagus diagnosis is not well studied.
Aim: To compare the proportion of prevalent adenocarcinoma vs. incident adenocarcinoma found during surveillance of Barrett's.
Methods: We performed a systematic search of MEDLINE, EMBASE and Web of Science (from their inception to 31 May 2015) for cohort studies in adults with Barrett's (nondysplastic Barrett's ± Barrett's with low-grade dysplasia) with minimum average follow-up of 3 years, and providing numbers of prevalent adenocarcinoma detected (concurrently with Barrett's diagnosis and up to 1 year afterwards) vs. incident adenocarcinoma detected (greater than 1 year after Barrett's diagnosis). Pooled weighted proportions of prevalent and incident adenocarcinoma were calculated, using a random effects model.
Results: On meta-analysis of 13 studies reporting on 603 adenocarcinomas in 9657 Barrett's patients, 85.1% of adenocarcinomas were classified as prevalent [95% confidence interval (CI), 78.1-90.2%) and 14.9% as incident (95% CI, 9.8-21.9%), with substantial heterogeneity (I(2) = 66%). Among nine studies reporting on 787 high-grade dysplasia and oesophageal adenocarcinomas in 8098 Barrett's patients, the proportion of prevalent high-grade dysplasia-oesophageal adenocarcinoma was similar at 80.5% (95% CI, 68.1-88.8%, I(2) = 87%). These results remained stable across multiple subgroup analyses including study quality, setting, duration of follow-up and presence of baseline dysplasia.
Conclusions: In our meta-analysis, four of five patients diagnosed with adenocarcinoma or high-grade dysplasia at index endoscopy or within 1 year of Barrett's follow-up were considered to be prevalent cases. Continued efforts are needed to identify patients with Barrett's before the development of adenocarcinoma.
© 2016 John Wiley & Sons Ltd.