Objective Emerging data suggest that a 30% relative decline in liver fat, as assessed by MRI-proton density fat fraction (MRI-PDFF), may be associated with Non-Alcoholic Fatty Liver Disease Activity Score improvement, but the association between decline in MRI-PDFF and fibrosis regression is not known. Therefore, we aimed to examine the association between ≥30% relative decline in MRI-PDFF and fibrosis regression in non-alcoholic fatty liver disease (NAFLD).

Design This prospective study included 100 well-characterised patients with biopsy-proven NAFLD with paired contemporaneous MRI-PDFF assessment at two time points. MRI-PDFF response was defined as ≥30% relative decline in MRI-PDFF. The primary outcome was ≥1 stage histological fibrosis regression.

Results The median (IQR) age was 54 (43–62) years and body mass index was 31.9 (29–36) kg/m². In multivariable-adjusted logistic regression analysis (adjusted for age, gender, diabetes status, race/ethnicity, interval between biopsies, gamma-glutamyl transferase, liver stiffness by magnetic resonance elastography and change in platelet counts), MRI-PDFF response was an independent predictor of fibrosis regression with an adjusted OR of 6.46 (95% CI 1.1 to 37.0, p=0.04). The proportion of patients with MRI-PDFF response with fibrosis regression, no change in fibrosis and fibrosis progression was 40.0%, 24.6% and 13.0%, respectively, and the proportion of patients with MRI-PDFF response increased with fibrosis regression (p=0.03).

Conclusion ≥30% reduction in MRI-PDFF in early phase trials can provide a useful estimate of odds of ≥1 stage improvement in fibrosis. These data may be helpful in sample size estimation in non-alcoholic steatohepatitis trials.