A novel image analysis method can detect dynamic changes in Barrett's esophagus surface area: A retrospective...

Barrett's esophagus (BE) is seen in individuals with GERD. It can progress into dysplasia and cancer, so detection and surveillance are necessary. We report on a retrospective analysis of patient endoscopy reports identifying factors influencing surface progression of BE.

Patient demographics: age, sex, Body Mass Index, Proton-Pump Inhibitor treatment and BE segment length at index endoscopy were collected for analysis. Prague criteria circumferential (C) and longitudinal (M) extent of the BE lesion progression were evaluated. Long-segment BE was defined as M > 3 cm. BE lesion surface area was quantified using ImageJ on 81 patients' endoscopic images. 200 patients were identified with average time to follow-up of 2.4 years. Multivariate regression analysis was used to determine features associated with dysplasia.

Surface progression of C (dC) or M (dM) had no significant association with age, sex, or BMI, however, differences were seen between initial endoscopy segment length. Both mean dM in long and short segment groups (t = 4.61, df=195, p < 0.0001) and mean dC (t = 3.343, df=195, p = 0.0010) were found to be significantly different. On quantitative image analysis, long-segment BE showed significant decrease (t = 1.999, p < 0.05). Any increase in circumferential BE (C) was associated with increased odds of dysplasia OR 3.0 (95% CI: 1.0-9.0, p = 0.05).

This preliminary work suggests that BE is a dynamic entity with poorly understood surface progression patterns. Surface progression risk appears to be independent of many risk factors associated with BE. While additional work is required to better elucidate this link, our work suggests that surface lesion changes are noteworthy.