A Markov model of fibrosis development in Non-Alcoholic Fatty Liver Disease predicts fibrosis progression in clinical cohorts.

Disease progression in non-alcoholic steatohepatitis (NASH) is highly heterogenous and remains poorly understood. Fibrosis stage is currently the best predictor for development of end-stage liver disease and mortality. Better understanding and quantifying the impact of factors affecting NASH and fibrosis is essential to inform clinical study design. We developed a population Markov model to describe the transition probability between fibrosis stages and mortality using a unique clinical non-alcoholic fatty liver disease (NAFLD) cohort with serial biopsies over three decades. We evaluated covariate effects on all model parameters and performed clinical trials simulations to predict fibrosis progression rate for external clinical cohorts. All parameters were estimated with good precision. Age and diagnosis of type 2 diabetes (T2D) were found to be significant predictors in the model. Increase in hepatic steatosis between visits was the most important predictor for progression of fibrosis. Fibrosis progression rate (FPR) was 2-fold higher for fibrosis stage 0 and 1 (F0-1) compared to fibrosis stage 2 and 3 (F2-3). A 2-fold increase in FPR was observed for T2D. A 2-point steatosis worsening increased the FPR 11-fold. Predicted fibrosis progression was in good agreement with data from external clinical cohorts. Our fibrosis progression model shows that patient selection, particularly initial fibrosis stage distribution, can significantly impact fibrosis progression and as such the window for assessing drug efficacy in clinical trials. Our work highlights the increase in hepatic steatosis as the most important factor in increasing FPR, emphasizing the importance of well-defined lifestyle advise for reducing variability in NASH progression during clinical trials.