Overdiagnosis of high-grade dysplasia in Barrett's esophagus: a multicenter, international study.

Numerous histological mimics of high-grade dysplasia in Barrett's esophagus predispose to overdiagnosis and potential serious mismanagement, including unnecessary esophagectomy. This study investigates the prevalence and sources of this problem. Biopsies from 485 patients diagnosed with Barrett's high-grade dysplasia were screened for a multi-institutional, international Barrett's endoscopic ablation trial. Screening included review of the original diagnostic slides and an additional protocol endoscopy with an extensive biopsy sampling. Observer variability by the study pathologists was assessed through two blinded diagnostic rounds on 437 biopsies from 26 random study endoscopies. Study diagnostic reassessments revealed significantly lower rates of high-grade dysplasia. Only 248 patients (51%) were confirmed to have high-grade dysplasia. The remaining patients had inflamed gastric cardia without Barrett's (n=18; 7%), Barrett's without dysplasia (n=35; 15%), indefinite change (n=61; 26%), low-grade dysplasia (n=79; 33%), adenocarcinoma (n=43; 18%), and other (n=1; <1%), yielding an alarming total of 194 or 40% of patients who were overdiagnosed with Barrett's high-grade dysplasia. Study pathologists achieved a high-level agreement (90% three-way inter-observer agreement per biopsy, Kappa value 0.77) for high-grade dysplasia. Confounding factors promoting overdiagnosis included Barrett's inflammatory atypia (n=182), atypia limited to the basal metaplastic glands (n=147), imprecise criteria for low grade neoplasia (n=102), tangential sectioning artifact (n=59), and reactive gastric cardiac mucosa (n=38). A total of 194 patients (40%) were overdiagnosed with Barrett's high-grade dysplasia, as affirmed by the extensive screening process and high-level study pathologist agreement. The multiple diagnostic pitfalls uncovered should help raise pathologists' awareness of this problem and improve diagnostic accuracy.