Prevalence and Predictors of NAFLD in a Population Based BE Cohort

Abstract

Introduction: Central obesity, increased visceral abdominal fat and its sequelae such as metabolic syndrome and diabetes have been shown to be reflux independent risk factors for Barrett's esophagus (BE) and esophageal adenocarcinoma (EA). There are limited data prevalence and risk factors for non-alcoholic fatty liver disease (NAFLD) in BE subjects. Our aim was to assess the proportion of BE subjects in a population based BE cohort who had NAFLD and the predictors of NAFLD in those with BE. Methods: The Olmsted County BE cohort consists of all individuals with incident BE in Olmsted County, MN, from 1977 to 2014. All individuals with BE had histologic evidence of intestinal metaplasia and at least 1 cm of columnar mucosa in the tubular esophagus. The BE database was utilized to obtain additional demographic and clinical details. An initial NAFLD diagnosis was made using standard ICD9 and HICDA codes. The diagnosis was then confirmed by review of laboratory, radiology and/or histology reports. Those with evidence of substantial alcohol intake and/or hepatitis B or C infection were excluded. Logistic regression analysis was used to identify predictors of NAFLD. Results: The BE cohort consisted of 522 patients. Of these, 72 (13.8%) had a diagnosis of NAFLD. Table 1 presents baseline data of these cohorts. The diagnosis of NAFLD was made by radiologic studies (ultrasound, CT, or MRI) in the majority of subjects and by histology in one case. BE patients with NAFLD were more obese and had shorter BE segments than those without NAFLD (table 1). Multivariable regression modeling also confirmed this association (Table 2) adjusting for age and gender. The prevalence of dysplasia or cancer was not statistically different between these two cohorts.Table 1Table 2: Logistic Regression Models with outcome of NAFLDConclusion: In this retrospective study, 14% of BE subjects in a population based cohort from a midwestern US state had a clinical diagnosis of NAFLD. This may reflect an underestimate given lack of systematic screening for NAFLD in all BE subjects. BE subjects with NAFLD had a higher BMI and shorter BE segments than those without NAFLD.Figure 1Figure 2