Christian Sebesta, Mathias Jachs, Lukas Hartl, Michael Schwarz, Lorenz Balcar, Benedikt S Hofer, Nina Dominik, Georg Kramer, Paul Thöne, Annalisa Berzigotti, Josune Cabello Calleja, Juan Carlos García-Pagán, Ioana Morosan, Martin Sebastian McCoy, Bogdan Procopeț, Horia Ștefănescu, Jonel Trebicka, Frank Erhard Uschner, Laure Elkrief, Pierre-Emmanuel Rautou, Bernhard Scheiner, Albert F Stättermayer, Benedikt Simbrunner, Till Schöchtner, Friedrich Haimberger, Nicolas Balutsch, Michael Trauner, Mattias Mandorfer, Thomas Reiberger, David J M Bauer
Reliable SSI-LSM and SSI-SSM provide robust non-invasive assessment of CSPH risk, however, the requirement of reliable (IQR/Med ≤30%) paired measurements may limit applicability. Combined SSI-LSM/SSM-based rules and multivariable models reduced gray-zone, and the SSI-4PH yielded a favorable decision-curve utility for CSPH risk assessment.
BACKGROUND & AIMS: Liver (LSM) and spleen (SSM) stiffness measurements correlate with hepatic venous pressure gradient (HVPG), the reference standard for defining clinically significant portal hypertension (CSPH, i.e. HVPG ≥10 mmHg). LSM, SSM and derived scores have been proposed for non-invasive CSPH risk assessment.
METHODS: Compensated advanced chronic liver disease (cACLD) patients undergoing same-day HVPG, Supersonic (SSI)-LSM, and SSI-SSM were prospectively included. Reliable SSI-LSM and SSI-SSM were defined by IQR/Med ≤30%. SSI-LSM, SSI-SSM and derived Baveno VII rules, Anticipate-SSI, LPS-SSI, and the newly developed SSI-4PH (LSM/SSM, platelet count: PLT, BMI) model were evaluated in a derivation cohort (DeCo) and an external validation cohort (VaCo).
RESULTS: The DeCo comprised 220 patients, paired reliable SSI-LSM/SSI-SSM measurements were available in 118 (53.6%) patients including 74.6% with CSPH. The VaCo comprised 86 eligible patients, of whom 73 (84.9%) had paired reliable measurements including 43.8% with CSPH. In reliable paired analyses, AUROCs for CSPH detection in the DeCo were 0.84 for Anticipate-SSI, 0.87 for LPS-SSI, and 0.90 for SSI-4PH; the corresponding AUROCs in the VaCo were 0.78, 0.77, and 0.73. At the fixed ≥75% predicted-risk threshold, specificity of SSI-4PH was 83.3% in the DeCo and 87.8% in the VaCo.
CONCLUSIONS: Reliable SSI-LSM and SSI-SSM provide robust non-invasive assessment of CSPH risk, however, the requirement of reliable (IQR/Med ≤30%) paired measurements may limit applicability. Combined SSI-LSM/SSM-based rules and multivariable models reduced gray-zone, and the SSI-4PH yielded a favorable decision-curve utility for CSPH risk assessment.