Values and limitations of serum aminotransferases in clinical trials of nonalcoholic steatohepatitis

Bacjkground/aims: Choosing endpoints in nonalcoholic steatohepatitis (NASH) trials is challenging because of the lack of validated surrogates and the trade-off between accuracy and invasiveness. In this study, we assessed diagnostic accuracy of serum aminotransferase changes in predicting histological changes in NASH trials. Methods: We conducted a longitudinal cohort study by using 102 participants in ursodeoxycholic acid-NASH trial who had both baseline and 2-year liver biopsy and multiple measurements of serum aminotransferases. We calculated rates of alanine aminotransferase (ALT) [or aspartate aminotransferase (AST)] change as slopes of linear regression over 2 years (IU/l/month) and changes in each histological feature as differences in Brunt's scores of two biopsies in each individual. Results: Rate of aminotransferase change correlated with changes in inflammation and fibrosis, but not steatosis and only with change in inflammation in multivariable analysis. In each histological feature, changes were inversely correlated with baseline histological grade. In predicting improvement of inflammation, areas under the receiver-operating characteristic curve of aminotransferase information alone were 0.72 for ALT and 0.73 for AST and were improved to 0.88 and 0.89, respectively, when baseline histology were taking account of. Conclusions: Serum aminotransferase changes could be useful as surrogates in screening therapies for NASH in clinical trials with appropriate consideration of baseline aminotransferase and histology.