Introduction Barrett's oesophagus (BO) is a premalignant lesion of the oesophagus that predisposes to oesophageal adenocarcinoma (OAC). Currently, widespread endoscopic surveillance of BO is undertaken in an attempt to detect and treat OAC at an early stage. The cost effectiveness of such surveillance is highly dependant on OAC risk in BO. However, the reported incidence of OAC in BO varies widely due to a lack of robust population-based studies and controversy over BO definition. The aim of this study was to examine the risk of OAC in one of the largest population based registers of BO worldwide.
Methods The Northern Ireland Barrett's oesophagus Register (NIBR) includes every adult diagnosed with BO (columnar lined oesophagus) in Northern Ireland between 1993 and 2005. The register was constructed by examining pathology reports from all oesophageal biopsies taken during this period. Information was collected on length of BO segment, presence of endoscopically visible segment, presence of specialised intestinal metaplasia (SIM), and degree of dysplasia seen. Incident OAC cases were identified by matching the NIBR with the Northern Ireland cancer registry. Deaths were identified by matching with death files from the Registrar General's Office. BO patients were followed up until the end of December 2005 and patients with <12 months follow-up were excluded from analysis.
Results 7585 BO patients were followed up for an average of 5.02 years. The incidence (95% CI) of OAC per 100 person years was 0.18 (0.14, 0.23) and for combined events (OAC and high-grade dysplasia) the rate was 0.23 (0.19, 0.28). In robustly defined BO patients with evidence of a clinically visible segment and histologically confirmed SIM, OAC incidence was higher at 0.29 (0.21, 0.40) and 0.41 (0.31, 0.54) for combined events. OAC and HGD risk was further elevated in males and in patients with long segment BO, hiatus hernia or low-grade dysplasia (LGD). However, if the analysis was restricted to the robust definition of BO, only patients with LGD were at a heightened risk of these outcomes.
Conclusion The risk of OAC in BO in this study is lower than previously reported, indicating that the currently recommended surveillance strategies for the detection of OAC in BO may not be cost effective. Limiting surveillance to those with LGD may improve the cost effectiveness of surveillance, however, this would miss a substantial portion of cancers. Further research into tissue biomarkers, optical recognition of dysplasia, endoscopic therapy for dysplasia and neoplasia is eagerly awaited.