Predicting Progression in Barrettʼs Esophagus: Development and Validation of the Barrettʼs Esophagus Assessment of Risk Score (BEAR Score)

To develop and validate a scoring tool capable of accurately predicting which patients with Barrett's esophagus (BE) will progress to dysplasia and/or esophageal adenocarcinoma. Endoscopic therapies have emerged capable of eradicating BE with high efficacy and low complication rates, but which patients should receive treatment is still debated. Current knowledge of risk factors is insufficient to allow for the accurate prediction of which patients will progress to dysplasia or adenocarcinoma. We retrospectively collected data from a cohort of BE patients over a 13-year period. A multivariable logistic regression model was constructed to predict progression. A simplified risk of progression (ROP) score was developed from weighted beta coefficients. Internal validation was performed using bootstrap analysis, and model discrimination was assessed using k-fold cross-validation. The cohort included 2591 BE patients of which 133 progressed to dysplasia/adenocarcinoma. Multivariable analysis with bootstrap internal validation resulted in 5 variables associated with an increased ROP (age ≥70 years, male sex, lack of proton-pump inhibitor use, segment greater than 3 cm, and history of esophageal candidiasis). Using this model, we developed a simple ROP score between 0 and 8. Receiver operating characteristic analysis showed a cutoff of 3 or higher to have a sensitivity and specificity of 70% and 79%, respectively. Patients with a score of 3 or higher had an odds ratio of 9.04 (95% confidence interval 6.06–13.46). The c-statistic obtained from 10-fold cross-validation was 0.76 (95% confidence interval 0.72–0.79), indicating good overall discrimination. Our data show the development and internal validation of the Barrett's Esophagus Assessment of Risk Score as capable of quantifying the likelihood of progression to dysplasia/adenocarcinoma. The Barrett's Esophagus Assessment of Risk Score can be used clinically to guide treatment decisions in nondysplastic BE patients.