Utility of different Baveno criteria to detect esophageal varices irrespective of their size in patients with compensated cirrhosis.

Abstract

Esophageal varices (EVs), irrespective of size, are the most-reliable indicators of clinically significant portal hypertension (CSPH) in patients with compensated advanced chronic liver disease (cACLD). While non-invasive tools (NITs) accurately identify those with varices needing treatment (VNTs), their role in identifying any EVs in patients with cACLD is not known. Patients with cACLD with reliable liver stiffness measurements (LSM), blood parameters and gastroscopy were retrospectively recruited from a multinational cohort. The performance of Baveno-VI (LSM > 20kPa and platelet count < 150,000/mm3) and expanded Baveno-VI criteria (LSM > 25kPa or platelet count < 110,000/mm3) was assessed to detect VNTs as well as any EVs. This performance was compared with the Baveno-VII possible CSPH criteria (LSM ≥ 15 kPa and platelet ≤ 150,000/mm3) to evaluate its utility in detecting any EVs. Patients with cACLD (n = 1200) of predominantly viral etiology (hepatitis B virus, 269; hepatitis C virus, 564; non-alcoholic fatty liver disease, 145; alcoholic liver disease, 130; other, 92) were included. Any EVs and VNTs were present in 514 (42.8%) and 70 (5.8%) patients, respectively. The Baveno-VI, expanded Baveno-VI and Baveno-VII criteria missed 29/514 (5.6%), 115/514 (22.4%) and 19/514 (3.7%) patients with any EVs, respectively, whereas they misclassified 517/686 (75.4%), 211/686 (30.8%) and 598/686 (87.4%) patients with no EVs as having a high risk of EVs. The Baveno-VI, expanded Baveno-VI and possible CSPH criteria missed 3/70 (4.3%), 15/70 (21.4%) and 0/70 (0%) VNTs, respectively. Both original Baveno-VI and Baveno-VII criteria were highly sensitive in detecting varices in cACLD, albeit with high misclassification rates.