Poor Agreement for Detection of Goblet Cells in Esophageal and GEJ Biopsies
Helen Wang, Ian Brown, Priyanthi Kumarasinghe, Cord Langner, Gregory Lauwers, Neil Shepherd, Michael Vieth, Amitabh Srivastava, Robert Odze. Beth Israel Deaconess Hospital, Boston; Royal Brisbane Hospital, Brisbane, Australia; PathWest QEII/UWA, Perth, Australia; Medical University of Graz, Graz, Austria; Massachusetts General Hospital, Boston; Cheltenham General Hospital, Cheltenham, United Kingdom; Institute of Pathology, Bayreuth, Germany; Brigham and Women's Hospital, Boston
Background: The American Gastroenterological Association (AGA) criteria for Barrett's esophagus (BE) includes endoscopic recognition of columnar mucosa and histologic confirmation of goblet cells in mucosal biopsies. Goblet cells are often few in number, and may be difficult to differentiate from pseudogoblet cells. This study was performed to determine the interobserver variability for detection of goblet cells, and other cell types, in biopsies from the distal esophagus and GEJ.
Design: Digitally scanned mucosal biopsies (Aperio system) from 34 patients, obtained from either the distal esophagus (N=16, 5 long segment BE, 11 short segment BE) or GEJ (N=18) were evaluated by 7 GI pathologists for a variety of histologic features, such as goblet cells, multilayered epithelium (ME), pseudogoblet cells, and type of glands, and then the pathologists were asked whether they believed the biopsies were diagnostic of BE. Overall, 18 cases with few, or only rare, goblet cells diagnosed by the original signout pathologist, who did not participate further in the interobserver study, were selected for study. The data were analyzed by Kappa statistics.
Results: Overall interobserver agreement for detection of goblet cells was poor (Kappa = 0.35), and agreement was worse for esophageal versus GEJ biopsies. All 7 reviewing pathologists agreed on the presence of goblet cells in only 12 cases. Kappa values for detection of ME and pseudogoblet cells were very poor (0.29 and 0.10, respectively). The highest levels of agreement were obtained for detection of gland type (Kappa = 0.54). However, even regarding gland type, all 7 pathologists agreed on the presence of mucous glands (vs. oxyntic vs. mixed mucous/oxyntic) in only 29% of cases. Agreement regarding whether the observers believed the biopsy fulfilled the histologic criteria for BE was extremely poor (Kappa=0.15).
Conclusions: Based on the results of this study, the current AGA criteria for BE is problematic, since interobserver agreement for detection of goblet cells when the latter are few in number is poor.
Monday, March 19, 2012 1:00 PM
Poster Session II # 81, Monday Afternoon