Non-invasive model predicting clinically-significant portal hypertension in patients with advanced fibrosis

Hepatic venous pressure gradient (HVPG) has been established as a predictor for the development of varices, clinical decompensation and death. In the present study, the primary objectives were to determine the diagnostic accuracy of the model developed by using readily-available data in predicting the presence of significant portal hypertension and esophageal varices.

This study included a total of 61 consecutive treatment-naive patients with advanced fibrosis (METAVIR F3, F4), established by liver biopsy. All patients underwent subsequent HVPG measurement and upper gastrointestinal endoscopy within 1 week of liver biopsy.

Seventeen patients (F3, 2/26; F4, 15/35) had clinically-significant portal hypertension (HVPG > or = 10 mmHg). The Risk Score for predicting significant portal hypertension was 14.2 - 7.1 x log(10) (platelet [10(9)/L]) + 4.2 x log(10) (bilirubin [mg/dL]). The area under the receiver-operator curve (AUC) curve was 0.91 (95% confidence interval [CI], 0.84-0.98). The optimized cut-off value (Risk Score = -1.0) offered a sensitivity of 88% (95% CI, 62-98%) and a specificity of 86% (95% CI, 72-94%). The AUC of the Risk Score in predicting varices was 0.82 (95% CI, 0.67-0.98). The cut-off had a sensitivity of 82% (95% CI, 48-97%) and a specificity of 76% (95% CI, 62-86%).

A predictive model that uses readily-available laboratory results may reliably identify advanced fibrosis patients with clinically-significant portal hypertension as well as esophageal varices. However, before accepted, the results of the current study certainly should be validated in larger prospective cohorts.