Do we need a histological score to diagnose non-alcoholic steatohepatitis? authors’ reply

Abstract

Sirs, We whole-heartedly agree with Drs Wong and Chan’s observations.1 The non-alcoholic fatty liver disease (NAFLD) activity score (NAS) was not intended and should not be utilised for the sole purpose of diagnosing non-alcoholic steatohepatitis (NASH). The NAS was intended to assess histological changes over time for use in clinical research trials and natural history studies2 and should continue as such. However, the NAS as a tool for assessing histological changes should correlate with the spectrum of disease in NAFLD; thus, assessment of the NAS agreement with the histological diagnosis is important. The discussed study3 demonstrates that the NAS correlates well with the histological spectrum of disease activity. In response to Dr Wong’s proposed question, do we need NAS to diagnose NASH? The short answer is no. We agree, however, that if this were the purpose of the NAS, the NAS should be more accurate. In the discussed study, even using the optimal cut-off (NAS ≥ 4), the sensitivity and specificity were 85% and 81%.3 In a separate study with biopsies reviewed by the pathology committee of the NASH Clinical Research Network (CRN), 14% of biopsies with an NAS ≥ 5 did not have steatohepatitis, and 30% of NAS ≤ 4 had definite steatohepatitis.4 In another NASH CRN study, 16% of biopsies did not have NASH, yet had an NAS ≥ 5.5 Adding histological features such as fibrosis would probably improve the diagnostic accuracy, but diagnostic accuracy is not the intention of the NAS. The NASH CRN chose the histological features scored in the NAS (ballooning degeneration, necroinflammatory activity and steatosis), because they are potentially reversible in the short term, whereas fibrosis is a consequence of disease activity and less modifiable.2 We believe the NAS is sufficiently reproducible with low interrater variability. Within the NAS CRN, agreement between pathologists for the various NAS features had weighted kappa values >0.5.2 Another study by Younossi et al. also demonstrate good interrater agreement between hepatopathologists for features of fatty liver disease.6 The NAS is a good scoring system to compare liver biopsies in clinical trials, because it is simple to use, reproducible and correlates closely with the spectrum of disease activity, but it should not be used as a substitute for histological diagnosis. Further studies are needed to evaluate the effectiveness of the NAS at assessing changes in histopathology over time and predicting prognosis. Declaration of personal interests: S. A. Harrison has served as a speaker for Merck, Bristol-Myers Squibb and Onyx, and advisory board member for Merck and Amylin. He has received research support from Genentech, Merck and Rotterpharm. Declaration of funding interests: None.