Meta analysis: Cancer risk in Barrett's oesophagus

Aliment Pharmacol Ther. 2007 Dec;26(11-12):1465-77. doi: 10.1111/j.1365-2036.2007.03528.x. Epub 2007 Sep 26.


Background: Risk of cancer in Barrett's oesophagus is reported to vary between studies and also between countries, where the studies were conducted as per several systematic reviews. Cancer incidence has implications on surveillance strategies.

Aim: To perform a meta-analysis to determine the incidence of oesophageal cancer in Barrett's oesophagus.

Methods: Articles retrieved by MEDLINE search (English language, 1966-2004). Studies had to necessarily include verified Barrett's oesophagus surveillance patients, documented follow-up and cancer identified as the outcome measure. A random effects model of meta-analysis was chosen and results were expressed as mean (95% CI).

Results: Forty-one articles selected for conventional Barrett's oesophagus (length >3 cm); eight included short segment Barrett's oesophagus (one additional article including only short segment Barrett's oesophagus). Cancer incidence was 7/1000 (6-9) person-years duration of follow-up (pyd), with no detectable geographical variation [UK 7/1000 (4-12) pyd, USA 7/1000 (5-9) pyd and Europe 8/1000 (5-12) pyd]. Cancer incidence in the UK was 10/1000 (7-14), when two large studies were excluded. Cancer incidence in SSBO was 6/1000 (3-12) pyd. When short segment Barrett's oesophagus compared to conventional Barrett's oesophagus, there was a trend towards reduced cancer risk [OR 0.55, (95% CI: 0.19-1.6), P = 0.25].

Conclusion: We found no geographical variations in Barrett's oesophagus cancer risk, but observed a trend towards reduced cancer risk in short segment Barrett's oesophagus. There is a time trend of decreasing cancer incidence.

Publication types

  • Meta-Analysis

MeSH terms

  • Adenocarcinoma / etiology
  • Aged
  • Barrett Esophagus / complications*
  • Esophageal Neoplasms / epidemiology
  • Esophageal Neoplasms / etiology*
  • Female
  • Humans
  • Incidence
  • Male
  • Middle Aged
  • Prevalence
  • Publication Bias
  • Risk Factors