Prognostic Value of Non-Invasive Fibrosis and Steatosis Tools, Hepatic Venous Pressure Gradient (HVPG) and Histology in Nonalcoholic Steatohepatitis

BACKGROUND & AIMS Non-invasive diagnostic methods for liver fibrosis predict clinical outcomes in viral hepatitis and nonalcoholic fatty liver disease (NAFLD). We specifically evaluated prognostic value of non-invasive fibrosis methods in nonalcoholic steatohepatitis (NASH) against hepatic venous pressure gradient (HVPG) and liver histology. METHODS This was a retrospective cohort study of 148 consecutive patients who met the following criteria: transjugular liver biopsy with HVPG measurement; biopsy-proven NASH; absence of decompensation; AST-to-Platelets Ratio Index (APRI), fibrosis-4 (FIB-4), NAFLD fibrosis score, ultrasound, hepatic steatosis index and Xenon-133 scan available within 6 months from biopsy; a minimum follow-up of 1 year. Outcomes were defined by death, liver transplantation, cirrhosis complications. Kaplan-Meier and Cox regression analyses were employed to estimate incidence and predictors of outcomes, respectively. Prognostic value was expressed as area under the curve (AUC). RESULTS During a median follow-up of 5 years (interquartile range 3-8), 16.2% developed outcomes, including 7.4% who died or underwent liver transplantation. After adjustment for age, sex, diabetes, the following fibrosis tools predicted outcomes: HVPG >10mmHg (HR=9.60; 95% confidence interval [CI] 3.07-30.12), histologic fibrosis F3-F4 (HR=3.14; 1.41-6.95), APRI >1.5 (HR=5.02; 1.6-15.7), FIB-4 >3.25 (HR=6.33; 1.98-20.2), NAFLD fibrosis score >0.676 (HR=11.9; 3.79-37.4). Prognostic value was as follows: histologic fibrosis stage, AUC=0.85 (95% CI 0.76-0.93); HVPG, AUC=0.81 (0.70-0.91); APRI, AUC=0.89 (0.82-0.96); FIB-4, AUC=0.89 (0.83-0.95); NAFLD fibrosis score, AUC=0.79 (0.69-0.91). Neither histologic steatosis nor non-invasive steatosis methods predicted outcomes (AUC<0.50). CONCLUSIONS Non-invasive methods for liver fibrosis predict outcomes of patients with NASH. They could be used for serial monitoring, risk stratification and targeted interventions.