• Potential conflict of interest: Nothing to report.


We write in reply to the letter of Dr. Y.F. Liaw. In our original article we included all the relevant studies mentioned in the meta-analyses of Sung et al. and Yang et al.[1] The study of Benvegnu et al.[2] was excluded because the main conclusion of the study was on hepatitis C-related cirrhosis.

Concerning the two studies from Taiwan,[3, 4] Dr. Liaw misses our point about the conflicting conclusions. Briefly, the first study showed no reduction in the development of new cirrhosis after interferon treatment. The second study showed that interferon was effective in prevention of hepatocellular carcinoma (HCC) only in patients with preexisting cirrhosis at study recruitment. In the second study, presumably some of the treated patients with no cirrhosis on recruitment would have developed new cirrhosis if the first study was correct. This did indeed occur, mainly but not solely, in the “nonresponders” But interferon did not decrease HCC in this group of patients despite the shorter duration of the newly developed cirrhosis. This is the basic paradox of the studies.

Dr. Liaw mentions the differences in the results obtained from two “studies” from Hong Kong[5, 6] with our published article on interferon treatment.[7] However, he is citing two preliminary abstracts with incomplete patient follow-up (n = 48 and 138). The database of our interferon studies was exhaustively reviewed. In the final article we included all interferon-treated patients (n = 208). A control group (n = 203), recruiting all patients who fulfilled the treatment criteria during the study period but were not treated, was added to demonstrate the effectiveness or otherwise of interferon treatment. Furthermore, Dr. Liaw erroneously quotes that the interferon-treated group was found to have a high risk for HCC. Our conclusion was that interferon treatment did not lower the risk of development of HCC, there being no statistical difference between the treated and untreated patients.

Dr. Liaw also mentioned that our study mostly recruited immunotolerant patients, and therefore it was excluded from one meta-analysis.[8] However, of the 278 patients in the four original trials, only 89 were in the children cohort. 14.3% of the adult population was cirrhotic. The hepatitis B e antigen (HBeAg) seroconversion rates of the treated population and the control group were comparable to each other as well as to other studies in the review.

Dr. Liaw mentions that the proportion of patients with cirrhosis in his study of lamivudine in cirrhosis patients[8] should be 60% and 65% for the lamivudine-treated group and control group, respectively. In our review article the respective proportions quoted were 31% and 39%. These were the proportions of patients with Ishak fibrosis score of 6; that is, patients with established, and therefore definite, cirrhosis. The proportions mentioned by Dr. Liaw included patients with Ishak fibrosis score 5, which were patients with incomplete cirrhosis. There is in fact debate as to whether an Ishak fibrosis score of 5 should or should not be included as “cirrhosis” under the more recent METAVIR system of fibrosis scoring.[9] We therefore quoted only those with Ishak fibrosis score of 6.

Finally, for the study showing entecavir is superior to lamivudine, it was published after we submitted and proof-read our review.

  • Ching-Lung Lai, M.D.

  • Man-Fung Yuen, M.D., Ph.D., MBBS

  • Department of Medicine University of Hong Kong Queen Mary Hospital Hong Kong