3522 TOWARD THE DEVELOPMENT OF DATABASES ON BARRETT'S ESOPHAGUS, THE VARIABILITY OF ROUTINE HISTOPATHOLOGICAL REPORTING ON BIOPSIES IN PATIENTS WITH SUSPECTED BARRETT'S REQUIRES A SYSTEMATIC REVIEW FOR CONSISTENT DATA COLLECTION.

INTRO The malignant potential of Barrett's esophagus (BE) urges highly consistent reporting. In addition, the accuracy of epidemiological research is dependent on the quality of the diagnostic data entered. Inconsistency of histopathological reporting on dysplasia in BE is well documented in the literature. However, few data are available on the diagnostic accuracy of histopathological reporting on tissue samples from gastric appearing mucosa in the esophagus to confirm the presence of BE. METHODS The study consisted of 160 patients (median 52 y, range 26-89, 82 male) with endoscopic features of Barrett's esophagus over a 6 yr period. 160 sample collections were retrieved and systematically reviewed by a single experienced gastrointestinal pathologist. Specific features in glandular (columnar lined) mucosa reviewed: presence of goblet cells- called BE, columnar lined mucosa without description of goblet cells- called probable Barrett's esophagus (PBE), dysplasia, and acute/chronic inflammation. Review results were compared with routine reporting results using the kappa statistic. RESULTS Of the 160 patients with endoscopic Barrett's (52 questionable BE, 63 short and 35 long segment BE) the initial histopathological reports confirmed BE in 41.9 %, suggested PBE in 22.5 %, and rejected Barrett's in 35.6 %. Overall for clinical purpose there was fair to good agreement between the initial reports and the comprehensive review (kappa = 0.46; p