Non-invasive prediction of esophageal varices by stiffness and platelet in non-alcoholic fatty liver disease cirrhosis

[Display omitted] •Expanded Baveno VI works better than Baveno VI criteria for ruling out the presence of VNT in NAFLD-cirrhosis.•New NAFLD-cirrhosis criteria based on LSM and platelet values and optimized for M and XL FibroScan probes are optimal.•The accuracy of noninvasive scores for VNT is lower in non-obese patients. Baveno VI and expanded Baveno VI criteria can avoid the need for esophagogastroduodenoscopy (EGD) to screen for varices needing treatment (VNT) in a substantial proportion of compensated patients with viral and/or alcoholic cirrhosis. This multicenter, cross-sectional study aims to validate these criteria in patients with compensated cirrhosis due to non-alcoholic fatty liver disease (NAFLD), accounting for possible differences in liver stiffness measurement (LSM) values between M and XL probes. We assessed 790 patients with NAFLD-related compensated cirrhosis who had EGD within six months of a reliable LSM, measured by FibroScan® using M and/or XL probe. Baveno VI and expanded Baveno VI criteria were tested. The main variable used to optimize criteria was the percentage of endoscopies spared, keeping the risk of missing large VNT below a 5% threshold. LSM was measured by both M and XL probes (training set) in 314 patients, while only M or XL probe (validation sets) were used to measure LSM in 338 and 138 patients, respectively. In the training set, use of Baveno VI and expanded Baveno VI criteria reduced the number of EGD by 33.3% and by 58%, with 0.9% and 3.8% of large esophageal varices missed, respectively. The best thresholds to rule-out VNT were identified as platelet count >110,000/mm3 and LSM 110,000/mm3 and LSM