Hepatic venous pressure gradient (HVPG)-3P score as predictor tool of endoscopic signs of portal hypertension and liver-related events in patients with advanced chronic liver disease

Background

Clinically significant portal hypertension (CSPH) in patients with advanced chronic liver disease (ACLD) is high-risk condition for liver-related events (LRE). According with BAVENO VII, non-invasive tests (NIT) should be preferred to rule in or out for CSPH to avoid endoscopy. The novel score HVPG-3Parameters gave a prediction of HVPG showing good accuracy in CSPH prediction.

Aim

Aim is to evaluate the ability of HVPG-3P in PH and LRE prediction.

Methods

A retrospective study was conducted in two different Italian centers: University of Naples and University of Marche. In 10 years, consecutive compensated-ACLD patients who underwent upper-GI endoscopy for PH assessment were enrolled. NIT (FIB-4,ANTICIPATE,PLT/spleen-diameter ratio and HVPG-3P) were calculated. During the follow-up any LRE were recorded.

Results

Two-hundred and ninety-one ACLD subjects (with no differences between centers) were recruited (62.1% males, mean age 60 years).

Mean LSM was 23.6±11.7kPa; HVPG-3P was 13.3±2.0mmHg, ANTICIPATE was 1.0±1.6, FIB-4 was 5.0±3.2, platelets/spleen-diameter ratio was 906.5±568.2.

One-hundred and forty-two (48,9%) subjects showed PH endoscopic signs. They had higher LSM (26.6±12.4vs20.8±10.3kPa, p=0.00002), HVPG-3P (14.2±1.6vs12.4±2.0mmHg, p<0.001), ANTICIPATE (1.66±1.4vs0.31±1.5, p<0.001), FIB-4 (5.87±3.28vs4.13±2.84, p=0.001) and lower platelets/spleen-diameter ratio (652.4±323.6vs1148.6±641.8, p<0.0001) compared to subjects without PH endoscopic signs.

During a median follow-up of 41.5months (IQR 17.8-74.0) LRE occurred in a median of 18.9months (IQR 7.5-37) in 126/291 (43.2%).

Patients who experienced LRE showed higher LSM (26.0±12.3 vs21.9±11.0kPa, p=0.003), HVPG-3P (14.1±1.7vs12.6±2.0mmHg, p<0.0001), ANTICIPATE (1.54±1.57vs0.55±1.56, p<0.0001), FIB-4 (6.02±3.36vs4.19±2.79, p<0.0001) and lower platelets/spleen-diameter ratio (694.9±385.7vs1068.0±630.0, p<0.0001).

The HVPG-3P seems to have a good performance in PH and LRE prediction (considering the results from Random Forest model: MDA 4.52 and 6.55, higher than the majority of variables).

Conclusions

HVPG-3P is a simple tool for non-invasive prediction of PH and could be used to stratify the risk of LRE in patients with ACLD. However, this score needs to be validated in larger cohorts.