Letter: rates and determinants of significant liver inflammation in chronic hepatitis B virus-infected patients with low ALT levels in the absence of significant fibrosis—authors' reply

Editors,

We thank Huang et al for their interest in our recent article.¹ We used a unique dataset to show that among chronic hepatitis B patients without signs of significant fibrosis, the probability of significant inflammatory activity is extremely low if ALT levels are below two times the upper limit of normal (ULN).² In their letter, Huang et al expressed several potential concerns that we are happy to address.

First, all liver biopsies were taken when patients were off treatment; therefore, our findings cannot be applied to patients on anti-viral therapy. However, not all patients were treatment-naive. Therefore, we performed sensitivity analyses after stratification for previous anti-viral therapy. These analyses showed homogenous results, suggesting that our findings may be applied in both treatment-naive and treatment-experienced patients.

Another potential issue concerned the indication for liver biopsy, and how this might relate to our findings. The patients in our cohort underwent liver biopsy either as part of screening for enrolment in treatment trials or at the discretion of the treating physician. This implies that our population is likely to have been enriched for patients with active disease. The fact that we observed such high diagnostic accuracy even in this high-risk population supports the robustness of our findings.

We agree with Huang et al that other biochemical factors may also be valuable in predicting inflammatory activity, for example in patients with fluctuating ALT levels. However, many routinely assessed factors are inter-related, necessitating careful multivariate analysis before inferences can be made.³ For example, platelet count was previously reported to be associated with inflammatory activity.⁴ A similar relationship was found in our cohort in univariate analysis, but this association was lost upon adjustment for the extent of liver fibrosis. This suggests that the degree of liver fibrosis, rather than platelet count itself, is the driver of the previously proposed association.

Another potential issue identified by Huang et al is the fact that different pathologists applied different methods (i.e., METAVIR, histological activity index) to assess liver inflammation. While this might appear to be a limitation, it is important to consider that both systems are currently being applied across the globe. Since our findings were consistent across cohorts (also shown in Table 2 of our article²), we are confident that the use of these different methods and assessment by different pathologists actually increase the external validity of our results.

Finally, the clinical application of our findings requires the non-invasive assessment of hepatic fibrosis. We previously showed that a FIB-4 score of <0.55 can be used to rule out significant fibrosis.⁵ Among 527 patients with FIB-4 <0.55, only one of 72 with normal ALT and only six of 197 with ALT 1-2 × the ULN had significant inflammation. Since previous studies have shown excellent diagnostic accuracy of liver elastography for ruling out significant fibrosis,⁶ we foresee an important role for this modality in our proposed algorithm. However, this remains to be assessed in future studies.