We thank Dr. Brunt and colleagues for their interest in our article. We agree that validation against clinical outcome is important for any grading and staging system. Our study was designed to examine the relationship of histopathologic features to liver-related mortality in nonalcoholic fatty liver disease (NAFLD), and so among other assessments, we included two widely used scoring systems.1, 2
Regarding NAFLD activity score (NAS), we agree that the original description and subsequent confirmation discouraged its use for diagnosis of nonalcoholic steatohepatitis (NASH),2, 3 and as noted in our article, our results confirmed this.
Regarding the score presented by Brunt et al., we strictly adhered to the definitions in its original description.1 It is true that the score was intended for grading biopsies diagnosed with NASH. However, the article's description of mild NASH is quite clear and precise: “Steatosis (predominantly macrovesicular) involving up to 66% of biopsy; may see occasional ballooned zone 3 hepatocytes; scattered rare intra-acinar polymorphonucleocytes and/or intra-acinar lymphocytes; no or mild portal chronic inflammation.” Steatosis and mild lobular inflammation are mandatory parts of this definition of mild NASH, but ballooning is optional. Elsewhere, the description of mild NASH includes “Ballooning and disarray are minimally present, if at all.” It also noted “these biopsies meet the minimum criteria for steatohepatitis.” By contrast, the description of moderate NASH says “Ballooning and disarray are always present.” Therefore, this means that any biopsy with steatosis and mild lobular inflammation can be interpreted as mild NASH according to the original criteria described in Brunt et al.
It is important to remember that regardless of intended use, all pathologic grading and staging systems are based on the premise that what looks bad to the pathologist will be worse for patients. If the goal of a clinical trial is to improve histologic characteristics, it is with the presumption that histologic improvement will decrease an adverse long-term outcome. A true evidence-based test of that hypothesis requires a study with long-term outcome as the endpoint. If under the rubric of NASH we exclude Brunt's grade 1 or mild NASH, its performance in predicting liver-related mortality improves significantly. In contrast, NAS does not directly translate to a diagnosis of NASH and is more useful for assessing short-term histologic changes in clinical trials, as was its intent.
Finally, we must disagree with the statement that “many” studies have shown that hepatic fibrosis is a predictor of long-term morbidity and/or mortality in NASH. We know of only three other nonalcoholic fatty liver disease studies (combined patients = 302) that assessed histology and long-term outcome. It seems likely that further advances in histologic predictors of natural history will require a much larger number of patients with long-term follow-up.