Letters to the Editor
Letter: diagnostic accuracy of M30 levels for identifying patients with non-alcoholic steatohepatitis – authors’ reply
Version of Record online: 17 DEC 2012
© 2012 Blackwell Publishing Ltd
Alimentary Pharmacology & Therapeutics
Volume 37, Issue 2, pages 283–284, January 2013
How to Cite
Shen, J. and Wong, V. W.-S. (2013), Letter: diagnostic accuracy of M30 levels for identifying patients with non-alcoholic steatohepatitis – authors’ reply. Alimentary Pharmacology & Therapeutics, 37: 283–284. doi: 10.1111/apt.12156
- Issue online: 17 DEC 2012
- Version of Record online: 17 DEC 2012
- Manuscript Accepted: 31 OCT 2012
- Manuscript Received: 30 OCT 2012
We thank Dr Yilmaz for the comments. According to his suggestion, we compared 32 ‘definite non-alcoholic steatohepatitis (NASH)’ patients (NAS ≥ 5) with 42 ‘simple steatosis’ (NAS ≤ 2). If we excluded patients with ‘borderline NASH’, the area under the receiver operating-characteristics curve (AUROC) of CK-18 M30, M65 and M65ED in differentiating ‘definite NASH’ was 0.77 (95% confidence interval: 0.66–0.88), 0.82 (0.72–0.92) and 0.85 (0.76–0.94) respectively. These results would be comparable to Yilmaz's report. However, we have a strong concern that it may lead to misinterpretation.
As shown in Figure S3 of our paper, all biomarkers increased in a stepwise fashion with increasing features of NASH. The difference between ‘definite NASH’ and ‘simple steatosis’ would certainly be more dramatic than comparing ‘definite NASH’ with all other non-alcoholic fatty liver disease (NAFLD) patients. However, we have a full spectrum of patients in real life, not just extreme cases. Including only extreme cases will overestimate the accuracy of non-invasive tests, an effect similar to evaluating liver stiffness measurement by transient elastography only in patients with advanced cirrhosis.
On the other hand, our AUROC of M30 still appears to be lower than those in some other studies. Feldstein and Tamimi both reported that the AUROC for M30 was over 0.80 in differentiating NASH. We want to emphasise again that the better diagnostic performance may be explained by their inclusion of patients without NAFLD in the ‘non-NASH’ group. If we included both NAFLD patients and control subjects in the analysis, the AUROC for M30, M65 and M65ED in differentiating NASH would be increased to 0.80, 0.83 and 0.83 respectively.
Finally, we have evaluated the biomarkers using both NAS and the global histological assessment according to current recommendations. Relevant information can be found in Tables 1, 2, 5, S3 and Figures 1, 2B, S4B, S5B.
The authors' declarations of personal and financial interests are unchanged from those in the original article.4