Factors influencing the development of Barrett's epithelium in the esophageal remnant postesophagectomy.

Abstract

Barrett's esophagus results from chronic reflux of both acid and bile. Reflux of gastric and duodenal contents is facilitated through the denervated stomach following esophagectomy, but the development of Barrett's changes in this model and the relationship to gastric and esophageal physiology is poorly understood. To document the development of new Barrett's changes, i.e., columnar metaplasia or specialized intestinal metaplasia (SIM) above the anastomosis, and relate this to the recovery of gastric acid production, acid and bile reflux, manometry, and symptoms. Forty-eight patients at a median follow-up of 26 months (range = 12-67) postesophagectomy underwent endoscopy with biopsies taken 1-2 cm above the anastomosis. The indication for esophagectomy had been adenocarcinoma (n = 27), high-grade dysplasia (n = 2), and squamous cell cancer (n = 19). Physiology studies were performed in 27 patients and included manometry (n = 25), intraluminal gastric pH (n = 24), as well as simultaneous 24-hour esophageal pH (n = 27) and bile monitoring (n = 20). Duodenogastric reflux increased over time, with differences between patients greater than and less than 3 years postesophagectomy for acid (p = 0.04) and bile (p = 0.02). Twenty-four patients (50%) developed columnar metaplasia and of these 13 had SIM. The prevalence of columnar metaplasia did not relate to the magnitude of acid or bile reflux, to preoperative neoadjuvant therapies, or to the original tumor histology. The duration of reflux was most significant, with increasing prevalence over time, with SIM in 13 patients at a median of 61 months postesophagectomy compared with 20 months in the 35 patients who were SIM-negative (p < 0.006). Supine reflux correlated with symptoms. The development of Barrett's epithelium is frequent after esophagectomy, is time-related, reflecting chronic acid and bile exposure, and is not specific for adenocarcinoma or the presence of previous Barrett's epithelium. This model may represent a useful in vivo model of the pathogenesis of Barrett's metaplasia and tumorigenesis.