EUS-guided portal venous pressure versus portal pressure gradient for detecting esophageal varices in a single-center observational cohort
Francesco Martini, Lucia Salvi, Daniele Balducci, Amine Benmassaoud, Rafael Romero-Castro, Martina Dalla Bella, Annalisa Alessandrini, Michele Montori, Alessandra Mandolesi, Luca Maroni, Antonio Benedetti, Marco Marzioni, Giuseppe TarantinoBackground and Objectives:
EUS-guided portal hemodynamic assessment has focused on the portal pressure gradient (PPG). In an endoscopic setting, the downstream hepatic component used to compute the gradient may introduce additional variability in the measurement. To compare EUS-measured absolute portal venous pressure (PVP) and PPG for clinically relevant manifestations of portal hypertension, with primary focus on esophageal varices.
Methods:
A single-center observational cohort of consecutive adults referred for EUS-based hemodynamic assessment. Procedural and clinical data were recorded prospectively in a dedicated clinical database. PVP and hepatic/inferior vena cava pressures were measured under a standardized protocol (left-lateral position, spontaneous breathing, and moderate sedation), and PPG was calculated. The main diagnostic endpoint was the area under the receiver-operating-characteristic curve (AUC) for esophageal varices of any grade. Secondary analyses evaluated the AUC for advanced portal hypertension (internally derived composite grade PH >2), correlations with noninvasive markers, procedural performance and safety, and subsequent hepatic decompensation.
Results:
Between March 15, 2024, and May 15, 2025, 55 patients were screened and 52 analyzed; technical success was 100%. Median PVP was 23.0 mm Hg (interquartile range 16.8–28.2) and median PPG was 11.0 mm Hg (6.0–15.0). For esophageal varices, AUC was 0.981 (95% confidence intervals 0.953–1.000) for PVP
Conclusion:
In this observational cohort with prospectively collected data, EUS-measured PVP showed higher apparent discrimination than PPG for detecting esophageal varices. These hypothesis-generating findings warrant multicenter validation.