Clinical prediction model for tumor progression in Barrett’s esophagus

Abstract

Background Individuals with Barrett’s esophagus (BE) are at increased risk of high-grade dysplasia (HGD) and esophageal adenocarcinoma (EAC), but the cost-effectiveness of general surveillance of BE is low. This study aimed to identify a risk prediction model for tumor progression in individuals with BE based on age, sex, and risk factors found at upper endoscopy, enabling tailored surveillance.

Methods This nested case–control study originated from a cohort of 8171 adults diagnosed with BE in 2006–2013 in the Swedish Patient Registry. Cases had EAC/HGD (n = 279) as identified from the Swedish Cancer Registry, whereas controls had no EAC/HGD (n = 1089). Findings from endoscopy and histopathology reports were extracted from medical records at 71 Swedish hospitals and from the Swedish Patient Registry. Multivariable logistic regression provided odds ratios (OR) with 95% confidence intervals (CIs).

Results Older age (OR 1.02 [95% CI 1.01–1.03] per year), male sex (OR 2.8 [95% CI 1.9–4.1]), and increasing maximum BE length (OR 2.3 [95% CI 1.4–3.9] for segments 3–8 cm and OR 4.3 [95% CI 2.5–7.2] for segments ≥ 8 cm) increased the risk of EAC/HGD, while the circumferential extent of the BE, hiatal hernia or reflux esophagitis did not. A model based on age, sex, and maximum BE length predicted 71% of all EAC/HGD cases.

Conclusions A simple combination of the variables age, sex and maximum BE length showed fairly good accuracy for predicting tumor progression in BE. This clinical risk prediction model may help to tailor future surveillance programs.