Identification of patients with decompensated cirrhosis at high risk for death: improving the prediction by hepatic venous pressure gradient?

Background/aims: The incremental usefulness of adding hepatic venous pressure gradient (HVPG) for predicting the risk of death has not, to our knowledge, been evaluated among patients with decompensated cirrhosis. We investigated whether the incorporation of the HVPG in an established model improved the prediction of death in a cohort of decompensated cirrhosis patients.

Methodology: We used data from 106 consecutive patients with decompensated cirrhosis who underwent a hemodynamic study between January 2006 and December 2007, to investigate whether the HVPG improved the risk discrimination of a patient beyond an assessment that was based on the Model for End-stage Liver Disease (MELD) or MELD-Na. We used occurrence vs. non-occurrence of events within 12 months as the outcome for analysis.

Results: For prediction of 12 months survival, the area under the receiver-operating characteristic curve (AUROC) for the MELD-Na was significantly greater than that of the MELD (79.4% vs. 70.5%, p=0.05). The MELD performed similar to the HVPG (70.5% vs. 71.2%, p=0.471). Adding the HVPG measurement to the MELD or the MELD-Na did not result in significant increase of the AUROC,with only a small improvement of about 5% in both cases.

Conclusions: The MELD-Na is the most predictive for 12-month survival in patients with decompensated cirrhosis. The addition of the HVPG to the MELD or the MELD-Na score does not appear to improve the prognostic accuracy of the MELD or the MELD-Na score significantly.