Global sensitivity analysis of hepatic venous pressure gradient (HVPG) measurement with a stochastic computational model of the hepatic circulation.

Hepatic venous pressure gradient (HVPG) is a widely employed surrogate of portal pressure gradient (PPG) in the diagnosis of portal hypertension (PHT). However, little is known about how HVPG measurement is affected by the complex vascular changes associated with PHT. In this study, we employed a computational method to quantitatively evaluate the sensitivity of HVPG measurement to various vascular factors involved in the development of sinusoidal PHT, aiming to provide a theoretical reference to guide the clinical application of HVPG measurement. The method consisted of developing a lumped-parameter model of the hepatic circulation to simulate HVPG measurement, stochastic parameter sampling used to represent a wide range of pathological conditions, and global sensitivity analysis performed to identify factors that dominate the accuracy of HVPG measurement. The major findings included 1) presinusoidal portal vascular resistance (Rpxs) and splanchnic vascular resistance (Rspl) were the major factors determining the relative difference (EHVPG) between HVPG and PPG; 2) hepatic arteriolar resistance and portosystemic collateral resistance had little influence on EHVPG although they relate closely to the severity of PHT; and 3) postsinusoidal vascular resistance (Rpts) only mildly affected EHVPG, despite its marked influence on HVPG and PPG. Moreover, stochastic simulations calibrated to HVPG/PPG data measured in a patient cohort revealed that misdiagnosis of clinically significant PHT with HVPG was more likely to occur in the presence of high Rspl combined with low Rpxs and Rpts. These findings suggest that understanding patient-specific vascular conditions can help to improve the application or interpretation of HVPG measurement.