Columnar islands in Barrett's esophagus: Do they impact Prague C&M criteria and dysplasia grade?

Background and Aim: The standard for classifying Barrett's metaplasia on endoscopy, the Prague C&M criteria, ignores all islands of metaplastic-appearing tissue. The aims of the present study were to measure the prevalence of columnar islands, quantify their impact on metaplasia extent, and determine if they harbor advanced dysplasia. Methods: Data from two prospective patient cohorts were retrospectively analyzed. They included adults who underwent upper endoscopy to evaluate for gastroesophageal reflux disease, Barrett's esophagus (BE), dysplasia, or adenocarcinoma between 2003 and 2012 at tertiary care centers in the USA and Germany. The BE pattern, location, and pathology were examined. The extent of BE as defined by the Prague criteria (disregarding the location of islands) was compared with the complete maximal extent of BE (incorporating the location of islands). Results: A total of 555 patients underwent endoscopy (mean age 60.1 years, 67.2% male, 91.9% white). Among those patients, 191 (34.4%) showed metaplastic-appearing mucosa in islands. Endoscopically, in 101 (52.9%) cases, islands were proximal to the farthest segment of BE as defined by the Prague M location. Histologically, intestinal metaplasia was confirmed in 60 (58.8%) of the 102 esophagogastroduodenoscopies (EGDs) where islands were biopsied. In 41 (40.2%) cases, the histologically confirmed BE islands extended farther than the maximal segment based on the Prague criteria. Pathology from biopsies of islands either changed the diagnosis or worsened the BE dysplasia grade in 16 (15.7%) of the 102 patients. Conclusions: Columnar islands are commonly seen on EGD. The Prague C&M criteria may underestimate the maximal extent of BE and overlook the area of highest dysplasia grade.