We thank Miller and Dillon for their comments. Their first point was on the definition of non-alcoholic steatohepatitis (NASH) and whether some NASH patients might have non-alcoholic fatty liver disease activity score (NAS) below 5 points. Among 69 patients classified as NASH in our study, 32 (46%) had NAS over 5 points.
While we agree with Miller and Dillon that the inclusion of less florid cases in the NASH group might lead to lower apparent diagnostic accuracy of the cytokeratin-18-based cell death biomarkers, the meaning of NAS deserves clarification. According to current guidelines, NASH is a histological diagnosis based on global features of hepatic steatosis, inflammation and hepatocyte ballooning. In contrast, NAS is the unweighted sum of the scores of these 3 histological parameters. NAS was originally developed by the NASH Clinical Research Network (CRN) to assess treatment response in NASH trials. However, since its introduction, many hepatologists have mistaken its meaning and used a cut-off of 5 points to diagnose NASH. A number of subsequent studies suggest that this approach is inappropriate. When pathologists from the NASH CRN reviewed the histology and clinical data of 976 adult NAFLD patients, only 75% of patients with definite NASH had NAS above 5, and at least 29% of patients with NAS below 5 had NASH. In an independent study of 386 adult patients, NAS was again shown to have only moderate correlation with the diagnosis of NASH.[6, 7]
Finally, we agree that one of the important roles of new biomarkers of NASH is to monitor disease activity over time because serial liver biopsies are unrealistic outside research setting.[8, 9] Further studies should be conducted to evaluate the performance of biomarkers against serial biopsies. This will eventually reduce the need for liver biopsy and improve patient care.