Prognostic performance of non-invasive tests for portal hypertension is comparable to that of hepatic venous pressure gradient

Abstract

Non-invasive tests to assess the probability of clinically significant portal hypertension (CSPH) - including the ANTICIPATE±NASH models based on liver stiffness measurement and platelet count±BMI, and the von Willebrand factor antigen to platelet count ratio (VITRO) - have fundamentally changed the management of compensated advanced chronic liver disease (cACLD). However, their prognostic utility has not been compared head-to-head to the gold standard for prognostication in cACLD, i.e. the hepatic venous pressure gradient (HVPG). Patients with cACLD (liver stiffness measurement ≥10kPa) who underwent advanced characterization via same-day HVPG/non-invasive test assessment from 2007-2022 were retrospectively included. Long-term follow-up data on hepatic decompensation was recorded. Four hundred and twenty patients with cACLD of varying etiologies, with a CSPH prevalence of 67.6%, were included. The cumulative incidence of hepatic decompensation at 1 and 2 years was 4.7% and 8.0%, respectively. HVPG, VITRO, and ANTICIPATE±NASH-CSPH-probability showed similar time-dependent prognostic value (AUROCs 0.683-0.811at 1 year and 0.699-0.801at 2 years). In competing risk analyses adjusted for MELD score and albumin, HVPG (adjusted subdistribution hazard ratio [aSHR] 1.099 [95% CI 1.054-1.150] per mmHg; p <0.001), or VITRO (aSHR 1.134 [95% CI 1.062-1.211] per unit; p <0.001), or ANTICIPATE±NASH-CSPH-probability (aSHR 1.232 [95% CI 1.094-1.387] per 10%; p <0.001) all predicted first decompensation during follow-up. Previously proposed cut-offs (HVPG ≥10mmHg vs. <10mmHg, VITRO ≥2.5 vs. <2.5, and ANTICIPATE-CSPH probability ≥60% vs. <60%) all accurately discriminated between patients at negligible risk and those at substantial risk of hepatic decompensation. The prognostic performance of ANTICIPATE±NASH-CSPH-probability and VITRO is comparable to that of HVPG, supporting their utility for identifying patients who may benefit from medical therapies to prevent first hepatic decompensation. Non-invasive tests have revolutionized the diagnosis and management of clinically significant portal hypertension in patients with compensated advanced chronic liver disease (cACLD). However, limited data exists regarding the prognostic utility of non-invasive tests in direct comparison to the gold standard for prognostication in cACLD, i.e. the hepatic venous pressure gradient. In our study including 420 patients with cACLD, the ANTICIPATE±NASH model and VITRO yielded similar AUROCs to hepatic venous pressure gradient for hepatic decompensation within 1 to 2 years. Thus, non-invasive tests should be applied and updated in yearly intervals in clinical routine to identify patients at short-term risk, thereby identifying patients who may benefit from treatment aimed at preventing hepatic decompensation.