Abstract

Background: The non-invasive models stratifying clinically significant portal hypertension (CSPH) are limited. We aim to develop a novel non-invasive model for predicting CSPH in patients with compensated advanced chronic liver disease, and investigate whether carvedilol could prevent hepatic decompensation in high-risk CSPH patients stratified by the novel model.

Methods: In this study, the derivation cohort (n=819) from a meta-analysis of 6 studies was used to identify risk factors and develop a novel noninvasive model for predicting CSPH. The novel model was validated in hepatic venous pressure gradient (HVPG) cohort (n=151) and was further assessed for the ability of predicting hepatic decompensation in follow-up cohort (n=1,102). The carvedilol-treating cohort (n=51) was included to prove that carvedilol could prevent hepatic decompensation in high-risk CSPH patients stratified by the novel model.

Results: In derivation cohort, liver stiffness measurement and platelets were identified as independent risk factors of CSPH and fitted to develop the novel CSPH risk model. A novel CSPH model was established as follows: 0.093510*LSM (kPa) - 0.01005*PLT (×10^9/L)-0.11. The novel model performed significantly better (all p<0.05) than other methods in HVPG cohort (Figure 1A). The risk of CSPH was stratified by the cut-off values at -0.68 and 0. The cumulative incidences (1.7% vs 2.5% vs 15.8%) of decompensation were significantly different in the low-, middle- and high-risk (p<0.001, Figure 1B) groups in follow-up cohort. The high-risk CSPH patients stratified by the novel model from carvedilol-treating cohort had significantly lower rates of decompensation than those of non-selective beta-blockers untreated high-risk CSPH patients from follow-up cohort (p<0.05, Figure 1 C&D).

Conclusion: Treatment with carvedilol among high-risk CSPH patients stratified by the novel model significantly reduces the risk of hepatic decompensation.