Persistent indefinite for dysplasia in Barrett's esophagus is a risk factor for dysplastic progression to low-grade dysplasia

Patients with Barrett's esophagus (BE) are at increased risk of esophageal adenocarcinoma (EAC). The risk is largely based on the degree of dysplasia. Dysplasia cannot always be differentiated from inflammatory changes, and therefore may be classified as indefinite for dysplasia (IND). The risk of progressive dysplasia in patients with IND is unclear. Our aim is to characterize the risk of progression in US veterans with BE-IND. We performed a single-center retrospective cohort study of patients with BE-IND between 2006 and 2016. All IND was diagnosed by consensus conference with an expert gastrointestinal (GI) pathologist or review by an expert GI pathologist and persistence was defined as IND present on subsequent endoscopic biopsy. The primary outcome was the incidence rate of high-grade dysplasia (HGD)/EAC. Secondary outcomes included any progression including incident low-grade dysplasia (LGD), any prevalent dysplasia and risk factors for dysplastic progression, namely persistent IND. Risk factors for progression were assessed using univariate and multivariate analysis with logistic regression. Among 107 patients with BE-IND, there were no incident cases of HGD/EAC. Twenty patients (18.7%) developed incident LGD during a median follow-up of 2.39 years (interquartile range, 1.13-5.17). The annual rate of progression to LGD was 5.95 per 100 patient-years (95% CI, 3.73-9.02). Prevalent dysplasia was common (9.3%). Eight patients had prevalent LGD, one patient had prevalent HGD and one patient had prevalent EAC. Twenty-eight patients (30.1%) were found to have persistent IND. Among those with persistent IND, 10 (36%) patients progressed to LGD (none to HGD/EAC). The progression rate to LGD for patients with persistent IND was 7.86 (95% CI, 3.99-14.02) cases per 100 patient-years versus 4.78 (95% CI, 2.48-8.52) for nonpersistent IND (PY=0.036). The odds ratio for progression to LGD in persistent IND was 3.06 (95% CI, 1.08-8.64). In multivariate analysis adjusting for age, smoking history, presence of hiatal hernia and BMI>30, persistent IND remained significant (OR 3.23; 95% CI, 1.04-9.98). Regression to nondysplastic BE was very common. Seventy-one (61%) patients developed complete and sustained regression of all dysplastic changes at last follow-up. Persistent IND, present in one-third of patients with IND, is an independent risk factor for progression to LGD. Although no patients in this cohort developed HGD/EAC, prevalent dysplasia was common (9.3%). Taken together, patients with IND should receive close surveillance for both prevalent and incident dysplasia especially if IND is persistent.