I read with interest the study by Shen et al., which investigated the diagnostic accuracy of the cell apoptotic marker cytokeratin-18 M30 and total cell death markers cytokeratin-18 M65/M65ED for the assessment and monitoring of non-alcoholic fatty liver disease (NAFLD). According to the results reported in Table 2, M30 achieved an area under the receiver operating characteristic (AUROC) curve for discriminating non-alcoholic steatohepatitis (NASH) from non-NASH of only 0.66 [95% confidence interval (CI): 0.57–0.75].
This value seems to be significantly lower than that originally reported by Wieckowska et al., who showed that serum M30 accurately predicted ‘definitive NASH’ (according to the NIDDK NASH Clinical Research Network scoring system) in patients with suspected NAFLD with an AUROC of 0.93. Similarly, we have previously reported that M30 levels had an AUROC of 0.83 for distinguishing between ‘simple steatosis’ and ‘definitive NASH’.
The lower diagnostic performance seen in the study by Shen et al. may be attributed to problems inherent in the histological classification of their patients with NAFLD. Indeed, as the results presented in Table S2 seem to suggest, the authors performed a comparison between ‘definite NASH’ vs. ‘borderline NASH’ plus ‘simple steatosis’ (the latter two categories grouped together). To allow comparisons with previous studies, the authors should compare ‘definite NASH’ with ‘simple steatosis’ (i.e. without the inclusion of subjects with ‘borderline NASH’).
In addition, the authors in their article refer to the recent endpoints and clinical trial design for non-alcoholic steatohepatitis. According to this classification system, the presence or absence of steatohepatitis is categorised according to gestalt recognition of the pattern of disease. Therefore, the authors should also reassess the diagnostic accuracy of the biochemical markers using the histological classification proposed by Sanyal et al.