Low-grade dysplasia diagnosis ratio and progression metrics identify variable Barrett's esophagus risk stratification proficiency in independent pathology practices

Abstract

Background and aims: The diagnosis of low-grade dysplasia (LGD) in Barrett's esophagus (BE) is subject to substantial interobserver variation. Our central aim in this study is to compare independent pathology practices using objective measures of BE risk stratification proficiency, including frequency of diagnosis and rate of progression, with high-grade dysplasia (HGD) or adenocarcinoma (EAC) after the first diagnosis of LGD.

Methods: We retrospectively evaluated over 29,000 endoscopic biopsy cases to identify 4734 patients under endoscopic biopsy surveillance for BE in a healthcare system with multiple independent pathology practices: a subspecialized GI pathology group (SSGI; 162 BE cases per pathologist annually), 3 high BE volume general surgical pathology practices (GSPs; >50 BE cases per pathologist annually), and multiple low BE volume GSPs (10.6 BE cases per pathologist annually). We measured LGD diagnosis frequencies and rates of diagnostic progression to HGD or EAC in patients diagnosed with LGD.

Results: The proportion of all BE cases diagnosed as LGD (LGD/BE diagnosis ratio) ranged from 1.1% to 6.8% in the different hospital settings (P < .001). The cumulative proportion of patients with HGD or EAC within 2 years of the first diagnosis of LGD was 35.3% in the SSGI and ranged from 1.4% to 14.3% in the GSPs (P < .001). LGD diagnosed by the GSP with the lowest LGD/BE diagnosis ratio had an adjusted risk of progression similar to LGD diagnosed by subspecialists (hazard ratio, .42; 95% CI, .06-3.03). However, when LGD was diagnosed by other generalists, the adjusted risk of progression was 79% to 91% lower than subspecialists (P < .001). When LGD was diagnosed in a low-volume GSP practice, the risk of progression was not significantly increased relative to patients with nondysplastic BE (hazard ratio, 1.3; 95% CI, .4-3.9).

Conclusions: General surgical pathologists and subspecialists show highly significant differences with respect to LGD/BE ratio, risk of progression relative to nondysplastic BE, crude annual progression rates, and the cumulative 2-year progression rate after LGD. These metrics can be used to assess proficiency in BE risk stratification in historical cases. Some general practitioners were able to achieve results similar to subspecialists. General surgical pathologists with little annual experience evaluating BE biopsy specimens did not successfully risk stratify patients with BE.