Objectives: The use of the hepatic venous pressure gradient (HVPG) to assess the efficacy of the pharmacological treatment of portal hypertension in cirrhosis is controversial. Our aim was to establish whether target HVPG reduction predicts variceal bleeding in cirrhotic patients receiving variceal bleeding prophylaxis.
Methods: Data sources were MEDLINE, EMBASE, Cochrane Controlled Trials Register, citation lists, and abstracts (most recent search March 2006). Cohorts of patients on drug therapy from randomized and nonrandomized studies correlating variceal bleeding and HVPG change were used. Heterogeneity was explored by metaregression analysis.
Results: Ten studies totaling 595 patients undergoing two HVPG measurements were identified. The RR of bleeding was lower in patients achieving an overall (HVPG <or=12 mmHg or decrease >or=20%) (0.27, 95% CI 0.14-0.52), complete (HVPG <or=12 mmHg) (0.48, CI 0.28-0.81), or partial (HVPG decrease >or=20%) (0.41, CI 0.20-0.81) response, with significant heterogeneity. Regression analysis identified the interval between the HVPG measurements significantly associated with the RR of bleeding. Heterogeneity was no longer significant after exclusion of an outlier trial, which showed the longest interval to HVPG remeasurement and the lowest quality score. Even considering nonvaluable patients because of bleeding as HVPG responders, the RR of bleeding was lower in overall responders than in nonresponders (0.66, CI 0.51-0.86). Overall response was associated with lower liver-related mortality (RR 0.58, CI 0.37-0.91).
Conclusions: Current evidence supports the validity of HVPG end points to monitor drug therapy efficacy for variceal bleeding prophylaxis. HVPG monitoring also provides valuable prognostic information.