RESUMO
BACKGROUND AND AIMS: The Baveno VI criteria enable non-invasive screening for esophageal varices. However, these criteria were established based on studies examining a large proportion of patients with viral hepatitis and relatively few patients with non-alcoholic fatty liver disease (NAFLD). Furthermore, because vibration-controlled transient elastography (VCTE) has a high incidence of measurement error, improved criteria are needed. We aimed to develop criteria based on magnetic resonance elastography (MRE) even among patients with NAFLD. METHODS: We performed a cross-sectional analysis of patients who had undergone MRE and/or VCTE as well as an esophagogastroduodenoscopy. The patients were classified as having either a low risk or a high risk of varices. The optimal cut-offs for ruling out esophageal varices were calculated for the MRE and VCTE liver stiffness measurement (LSM), the platelet count in an estimation cohort, and the cut-offs were then evaluated using validation cohorts composed of patients who had undergone only MRE or VCTE. RESULTS: The study included 627 patients (39% with NAFLD). The optimal cut-off values for the MRE-LSM and the platelet count were 4.2 kPa and 18.0 × 104 /µL, respectively. An MRE-LSM of 4.2 kPa plus a platelet count of 18.0 × 104 /µL had a negative predictive value of 1.00 for both low-risk plus high-risk varices as well as for high-risk varices in a validation cohort, enabling the presence of varices to be ruled out. CONCLUSIONS: Magnetic resonance elastography might enable a safer avoidance of screening endoscopy, with a smaller measurement error, among patient populations with a high prevalence of NAFLD.