We are grateful to Wen-Juei Jeng and colleagues for their thoughtful comments, which we address herein. We are confident that the proportion of patients who had viral remission (ie, a hepatitis B virus DNA concentration of 1Hsu YC Chen CY Chang IW et al.Once-daily tenofovir disoproxil fumarate in treatment-naive Taiwanese patients with chronic hepatitis B and minimally raised alanine aminotransferase (TORCH-B): a multicentre, double-blind, placebo-controlled, parallel-group, randomised trial. and we were not surprised to find seemingly different point estimates from other studies, given differences in study design, patient populations, assay sensitivities, attrition during follow-up, and analytical approach, among other factors. Taking the tenofovir disoproxil fumarate registration trial2Heathcote EJ Marcellin P Buti M et al.Three-year efficacy and safety of tenofovir disoproxil fumarate treatment for chronic hepatitis B. as an example, the proportion of patients with viral remission (defined as a viral DNA concentration of 1Hsu YC Chen CY Chang IW et al.Once-daily tenofovir disoproxil fumarate in treatment-naive Taiwanese patients with chronic hepatitis B and minimally raised alanine aminotransferase (TORCH-B): a multicentre, double-blind, placebo-controlled, parallel-group, randomised trial. Jeng and colleagues also compared the proportions of patients with regression and progression of fibrosis in the tenofovir disoproxil fumarate group in our study with those in two previous studies but overlooked the distinct features of the study populations. In our trial, most patients (129 [81%] of 160) had no or mild fibrosis (ie, an Ishak stage of 0–2), compared with only 136 (39%) of 348 patients in the study by Marcellin and colleagues.3Marcellin P Gane E Buti M et al.Regression of cirrhosis during treatment with tenofovir disoproxil fumarate for chronic hepatitis B: a 5-year open-label follow-up study. Only 12 (8%) of 160 patients in our study presented with advanced fibrosis (ie, an Ishak stage of 4) compared with 133 (38%) of 348 patients in the study by Marcellin and colleagues.3Marcellin P Gane E Buti M et al.Regression of cirrhosis during treatment with tenofovir disoproxil fumarate for chronic hepatitis B: a 5-year open-label follow-up study. Importantly, our trial1Hsu YC Chen CY Chang IW et al.Once-daily tenofovir disoproxil fumarate in treatment-naive Taiwanese patients with chronic hepatitis B and minimally raised alanine aminotransferase (TORCH-B): a multicentre, double-blind, placebo-controlled, parallel-group, randomised trial. examined the effects of treatment on histological progression from a mild disease status, rather than reported regression from a more advanced status following treatment.We noted that the proportion of patients with progression of fibrosis in the tenofovir disoproxil fumarate group might appear unexpectedly high, and discussed this point in the Article.1Hsu YC Chen CY Chang IW et al.Once-daily tenofovir disoproxil fumarate in treatment-naive Taiwanese patients with chronic hepatitis B and minimally raised alanine aminotransferase (TORCH-B): a multicentre, double-blind, placebo-controlled, parallel-group, randomised trial. As commented by Jeng and colleagues, regression to the mean (mediocrity) is difficult to understand. Nevertheless, this counterintuitive phenomenon is ubiquitous and occurs with repeated measurements because random errors and biological variations are inevitable.4Some examples of regression towards the mean. Besides, within-group comparisons in randomised trials are prone to misinterpretation and are not advisable.5Comparisons against baseline within randomised groups are often used and can be highly misleading. In response to their suggestion, we explored potential factors associated with fibrosis progression (table), but caution that this post-hoc analysis is essentially exploratory. Dedicated research is needed to further investigate which factors influence disease progression despite antiviral treatment. Jeng and colleagues themselves compared fibrosis progression between the two treatment groups (ie, the tenofovir disoproxil fumarate and placebo groups) within a subgroup of patients stratified by Ishak stage or HBeAg status, and seemed to overinterpret the subgroup analyses with undue emphasis. In fact, we did not calculate subgroup p values because they are misleading.6Assmann SF Pocock SJ Enos LE Kasten LE Subgroup analysis and other (mis)uses of baseline data in clinical trials.

TableExplorative analysis of baseline factors associated with progression of fibrosis

Given the explorative nature of this analysis, variables with a p value of less than 0·1 during stepwise selection were retained in the logistic regrevssion model.

Finally, HBeAg-positive patients were eligible for this study because these patients were not reimbursed for antiviral therapy in Taiwan, and also because there have been no previous randomised trials evaluating the effects of antiviral treatment on fibrosis progression in HBeAg-positive patients with minimally raised aminotransferase, regardless of their viral load.

Y-CH reports receiving research support from Gilead Sciences; being an advisory committee member for Gilead Sciences; and receiving lecture fees from Abbvie, Bristol-Myers Squibb, Gilead Sciences, Merck Sharp & Dohme, and Novartis. J-TL reports receiving research support from Gilead Sciences. Y-TH declares no competing interests.

References1.Hsu YC Chen CY Chang IW et al.

Once-daily tenofovir disoproxil fumarate in treatment-naive Taiwanese patients with chronic hepatitis B and minimally raised alanine aminotransferase (TORCH-B): a multicentre, double-blind, placebo-controlled, parallel-group, randomised trial.

Lancet Infect Dis. 21: 823-8332.Heathcote EJ Marcellin P Buti M et al.

Three-year efficacy and safety of tenofovir disoproxil fumarate treatment for chronic hepatitis B.

Gastroenterology. 140: 132-1433.Marcellin P Gane E Buti M et al.

Regression of cirrhosis during treatment with tenofovir disoproxil fumarate for chronic hepatitis B: a 5-year open-label follow-up study.

Lancet. 381: 468-4754.

Some examples of regression towards the mean.

BMJ. 309: 7805.

Comparisons against baseline within randomised groups are often used and can be highly misleading.

Trials. 12: 2646.Assmann SF Pocock SJ Enos LE Kasten LE

Subgroup analysis and other (mis)uses of baseline data in clinical trials.

Lancet. 355: 1064-1069Article InfoPublication HistoryIdentification

DOI: https://doi.org/10.1016/S1473-3099(21)00331-5

Copyright

© 2021 Published by Elsevier Ltd.

ScienceDirectAccess this article on ScienceDirect Linked Articles