Role of magnetic resonance elastography in compensated and decompensated liver disease

Background & Aims: Non-invasive predictors identifying subjects with compensated liver disease at highest risk for transitioning to a decompensated state are lacking. We hypothesized that liver shear stiffness as measured by magnetic resonance elastography is an important non-invasive predictor of hepatic decompensation. Methods: Among patients with advanced fibrosis undergoing magnetic resonance elastography (2007-2011), a baseline cohort and follow up cohort (compensated liver disease) were established. Cause specific cox proportional hazards analysis adjusting for competing risks was utilized to determine the association between elevated liver shear stiffness and development of decompensation (hepatic encephalopathy, ascites, variceal bleeding). Results: In the baseline cohort (n = 430), subjects with decompensated liver disease had a significantly higher mean liver shear stiffness (6.8 kPa, IQR 4.9-8.5) as compared to subjects with compensated liver disease (5.2 kPa, IQR 4.1-6.8). After adjustment for Model for End Stage Liver Disease score, hepatitis C, age, gender, albumin, and platelet count, the mean liver shear stiffness (OR = 1.13, 95% CI 1.03-1.27) was independently associated with decompensated cirrhosis at baseline. Over a median follow up of 27 months (n = 167), 7.2% of subjects with compensated disease experienced hepatic decompensation. In the follow up cohort, the hazard of hepatic decompensation was 1.42 (95% CI 1.16-1.75) per unit increase in liver shear stiffness over time. The hazard of hepatic decompensation was 4.96 (95% CI 1.4-17.0, p = 0.019) for a subject with compensated disease and mean LSS value >= 5.8 kPa as compared to an individual with compensated disease and lower mean LSS values. Conclusion: Baseline liver shear stiffness assessed by magnetic resonance elastography is independently associated with decompensated liver disease. (c) 2014 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.