Gut 64:667-672 doi:10.1136/gutjnl-2014-307237
  • Hepatology
  • Original article

Treatment cessation of entecavir in Asian patients with hepatitis B e antigen negative chronic hepatitis B: a multicentre prospective study

  1. Henry Lik-Yuen Chan3
  1. 1Department of Medicine, The University of Hong Kong, Hong Kong
  2. 2Department of Medicine, Alice Ho Miu Ling Nethersole Hospital, Hong Kong
  3. 3Institute of Digestive Disease, The Chinese University of Hong Kong, Hong Kong
  1. Correspondence to Professor Man-Fung Yuen, Department of Medicine, The University of Hong Kong, Hong Kong; mfyuen{at} Professor Henry Lik-Yuen Chan Institute of Digestive Disease, The Chinese University of Hong Kong, Hong Kong; hlychan{at}
  • Received 13 March 2014
  • Revised 23 April 2014
  • Accepted 2 May 2014
  • Published Online First 15 May 2014


Background and objective The off-treatment durability of nucleos(t)ide analogue therapy in Asian hepatitis B e antigen (HBeAg) negative chronic hepatitis B (CHB) and the role of hepatitis B surface antigen (HBsAg) levels in predicting off-treatment durability has not been well investigated.

Methods Following Asia-Pacific Association for the Study of the Liver guidelines, entecavir was stopped in Asian HBeAg negative patients treated for ≥2 years with undetectable HBV DNA levels on ≥3 separate occasions 6 months apart before treatment cessation. HBsAg and HBV DNA levels were prospectively monitored every 6–12 weeks for 48 weeks. Entecavir was restarted if there was virologic relapse (defined as HBV DNA >2000 IU/mL).

Result 184 patients (mean age 53.9 years, 67.9% male) were recruited. The cumulative rate of virologic relapse at 24 and 48 weeks was 74.2% and 91.4%, respectively. The median HBV DNA level at virologic relapse was 11 000 (range 2115 to >1.98×108) IU/mL. 42 (25.8%) patients had elevated alanine aminotransferase (median level 97 U/L, range 37–1058 U/L) during virologic relapse. Mean rate of off-treatment HBsAg decline was 0.018 (±0.456) log IU/mL/year. No patients cleared HBsAg. There was no correlation between off-treatment serial HBsAg and HBV DNA levels (r=−0.026, p=0.541). HBsAg levels at the time of entecavir commencement, entecavir cessation and the subsequent rate of HBsAg reduction were not associated with virologic relapse (all p>0.05).

Conclusions Entecavir cessation in Asian HBeAg negative CHB resulted in high rates of virologic relapse, suggesting nucleos(t)ide analogue therapy should be continued indefinitely until the recognised treatment endpoint of HBsAg seroclearance.

Significance of this study

What is already known on this subject?

  • First-line nucleos(t)ide analogue therapy achieves high rates of virologic suppression in chronic hepatitis B (CHB).

  • Discrepancies exist in the possible treatment endpoints for hepatitis B e antigen (HBeAg) negative CHB.

  • The off-treatment durability of nucleos(t)ide analogue therapy in HBeAg negative CHB has not been well investigated.

  • The role of serum hepatitis B surface antigen (HBsAg) level monitoring after treatment cessation is undetermined.

What are the new findings?

  • Despite good virologic suppression before treatment discontinuation, HBeAg negative Asian CHB patients had high rates of virologic relapse (91.4% in 48 weeks).

  • 25.8% of subjects had biochemical hepatitis during virologic relapse.

  • Mean serum HBsAg levels declined slowly and gradually after treatment cessation.

  • Baseline and subsequent serial HBsAg levels had no association with virologic relapse.

How might it impact on clinical practice in the foreseeable future?

  • With the high rates of virologic relapse noted, nucleos(t)ide analogue therapy should be continued indefinitely until the recognised treatment endpoint of HBsAg seroclearance.


Nucleos(t)ide analogue therapy has revolutionised the treatment paradigm of chronic hepatitis B (CHB). Long-term nucleos(t)ide analogue therapy suppresses viral replication, reverses liver fibrosis1 ,2 and reduces the risk of liver-related complications and deaths.3 ,4 For hepatitis B e antigen (HBeAg) negative CHB, which is currently the predominant type of CHB worldwide,5 ,6 continuation of nucleos(t)ide analogue therapy for an indefinite duration might be required.7 ,8

An important question that remains is whether nucleos(t)ide analogue treatment can be discontinued in HBeAg negative disease after a finite period of viral suppression with the abovementioned benefits still maintained. Hepatitis B surface antigen (HBsAg) seroclearance, the established treatment endpoint for nucleos(t)ide analogue therapy, rarely occurs.2 ,9 Without HBsAg seroclearance, virus replication would undoubtedly resume in the majority of patients after treatment discontinuation.10 Nonetheless, if serum HBV DNA rebounds to <2000 IU/mL, that is, the level below which the rate of development of long-term complications is significantly lower,11 ,12 the benefits of nucleos(t)ide analogue treatment could still be sustainable. Numerous studies in different HBeAg negative CHB populations have found variable rates of relapse after treatment discontinuation.1318 These studies all differ in the definition of relapse, type of nucleos(t)ide analogue used, duration of therapy and the criteria used for therapy cessation, with the majority of these studies having a comparatively small sample size.

The quantification of HBsAg has been recently advocated as a marker for disease monitoring in CHB.19 Low serum HBsAg level, for example <1000 IU/mL, has been suggested as a marker for good immune control in HBV genotype D,20 and low serum HBsAg levels or a marked reduction in HBsAg titres could play a role in predicting virologic remission after lamivudine cessation.21 These findings need confirmation in larger prospective studies involving newer nucleos(t)ide analogues.

We aimed to determine primarily the off-treatment sustainability of entecavir in HBeAg negative CHB, and secondarily the suitability of HBsAg levels in predicting virologic relapse after treatment discontinuation.


Patients and study design

This was a prospective, multicentre, observational study initiated in Asian CHB patients achieving good virologic suppression with entecavir. Subjects were recruited from three tertiary liver centres in Hong Kong from February 2012 to September 2012. All enrolled subjects were aged 18–65 years, had been on entecavir monotherapy for at least 24 months, were HBeAg negative at commencement of entecavir therapy, and had normal alanine aminotransferase (ALT) levels documented on two separate occasions 6 months apart. All subjects fulfilled the Asian-Pacific consensus statement on nucleos(t)ide analogue cessation in HBeAg negative CHB, with undetectable HBV DNA (<20 IU/mL) documented on three separate occasions of at least 6 months apart.22 Ultrasonography of the liver was performed at baseline (ie, at entecavir cessation), with cirrhosis defined as ultrasonographic evidence of small-sized or nodular-shaped liver with or without splenomegaly. Patients with any radiologic suspicion of hepatocellular carcinoma (HCC) were excluded. In addition, patients with concomitant HCV infection, prior interferon or nucleos(t)ide analogue therapy other than entecavir, decompensated liver disease (defined as Child's B to C cirrhosis), cirrhotic complications, prior history of HCC or prior history of organ transplantation were excluded. This study was approved by the respective institutional research boards of all three study centres.

After discontinuation of entecavir, we evaluated all subjects at baseline, and week 6, 12, 24, 36 and 48, with liver biochemistry, α-fetoprotein, HBV DNA and HBsAg levels determined at each clinic visit. In addition, HBsAg levels were also retrospectively measured in subjects with available saved serum samples at the time of entecavir commencement. The primary outcome of this study was virologic relapse, defined as serum HBV DNA >2000 IU/mL regardless of serum ALT levels, in which entecavir would be started. Secondary outcomes include undetectable HBV DNA (<20 IU/mL) and HBsAg seroclearance at week 48. Immune control was defined as HBV DNA persistently ≤2000 IU/mL until week 48.

Sample size

We aimed to compare the degree of virologic relapse as stratified by baseline HBsAg levels, that is, HBsAg <1000 IU/mL (or <3 log IU/mL) versus HBsAg ≥1000 IU/mL (or ≥3 log IU/mL). If we hypothesise the off-treatment sustainability in the first group of subjects to be 70%15 and the difference of response between the two groups to be at least 25%,21 with an alpha risk of 5% and a power of 90%, a total of 164 subjects would be required. Allowing a dropout rate of 10%, we planned to recruit a minimum of 182 subjects.

Laboratory assays

The upper limit of normal for serum ALT was based on the standards set by our local laboratory (58 U/L for men, 36 U/L for women). Serum HBV DNA levels were measured using Cobas Taqman assay (Roche Diagnostics, Branchburg, New Jersey, USA) with the lower limit of detection of 20 IU/mL. Serum HBsAg levels were measured using Elecsys HBsAg II assay (Roche Diagnostics, Mannheim), with the lower limit of detection of 0.05 IU/mL.

Statistical analyses

Continuous variables were expressed as mean (±SD) or median (range) when appropriate. Serum HBsAg levels were expressed in logarithms. The rate of HBsAg reduction was expressed as log IU/mL/year, and was calculated based on the HBsAg results at baseline and at the date of last follow-up. The χ2 test, or the Fisher exact test when appropriate, was used in the comparison of categorical variables; the independent Student t test, one-way analysis of variance, or the Mann–Whitney U test when appropriate, were used for comparison of continuous variables. Multivariate analysis was performed by binary logistic regression. Correlations of the repeated measurements of serum HBsAg levels with serum HBV DNA were calculated using Pearson's weighted correlation coefficient. The cumulative rate of virologic relapse was calculated using the Kaplan–Meier method, with the log-rank test used for comparisons. All statistical analyses were performed using SPSS V.19.0. A two-sided p value of <0.05 was considered statistically significant.


A total of 184 Asian CHB patients were recruited in the present study; their baseline (time of entecavir cessation) characteristics are depicted in table 1. Baseline mean HBsAg level was 2.86 log IU/mL, with 54.3% (n=100) and 5.4% (n=10) having baseline HBsAg levels of <3 log IU/mL and <2 log IU/mL, respectively. Six patients (3.26%) withdrew from the study and resumed entecavir at weeks 6 (2 patients), 12 (2 patients), 24 (1 patient) and 36 (1 patient), respectively. These six patients did not experience virologic relapse prior to entecavir resumption.

Table 1

Baseline* clinical characteristics of all 184 subjects

Virologic kinetics after treatment cessation

The cumulative rate of virologic relapse (HBV DNA >2000 IU/mL) at weeks 24 and 48 were 74.2% and 91.4%, respectively (figure 1). Altogether, 163 subjects experienced virologic relapse, with the most common time point of virologic relapse being week 24 (n=117, 71.8%), followed by week 12 (n=24, 14.7%). When considering all time points together, the median HBV DNA level at virologic relapse was 11 000 (range 2115 to >1.98×108) IU/mL. The changes in median serum HBV DNA levels stratified by the time point of virologic relapse are depicted in table 2. Once HBV DNA became detectable, when comparing serial HBV DNA values at each time point, a continued increasing trend in subsequent median HBV DNA levels was noted in subjects who eventually developed virologic relapse (all p<0.001).

Table 2

Median serum HBV DNA levels after treatment cessation stratified by time of virologic relapse

Figure 1

Cumulative rate of virologic relapse up to week 48. Virologic relapse defined as HBV DNA >2000 IU/mL. Relapse rate calculated using Kaplan–Meier method.

Forty-two (25.8%) subjects had elevated ALT and 19 (11.7%) had ALT >2× the upper limit of normal during virologic relapse. None of these 42 subjects had an increase in serum bilirubin. The median ALT level among those with ALT elevation was 97 (range 37–1058) U/L, with 88.1% (n=37) occurring in either week 12 or 24.

The cumulative rate of virologic relapse at 48 weeks stratified by baseline HBsAg levels (at the time of commencement of entecavir) is depicted in table 3. Regardless of the baseline HBsAg level used for stratification, there was no significant difference in the cumulative rate of virologic relapse (all p>0.05). Baseline serum HBsAg <1000 IU/mL, when compared to HBsAg ≥1000 IU/mL, was not associated with a reduced rate of virologic relapse (p=0.445).

Table 3

Cumulative rate of virologic relapse at 48 weeks stratified by baseline HBsAg levels

Fifteen (8.4%) patients remained in immune control (HBV DNA ≤2000 IU/mL) throughout 48 weeks off-treatment. Although median HBV DNA levels became detectable in week 24, there was no further significant increase up to week 48 (table 2, p=0.653). The median HBV DNA level at 48 weeks among these 15 subjects was 193 IU/mL. In five subjects (2.9%) HBV DNA remained <200 IU/mL up to week 48; one of these patients (0.6%) had persistently undetectable HBV DNA (<20 IU/mL).

HBsAg kinetics after treatment cessation

Mean HBsAg levels after entecavir cessation are depicted in figure 2. The mean rate of HBsAg decline was 0.018 (±0.456) log IU/mL/year. There was no significant difference in serial mean HBsAg levels during the follow-up period (p=0.571). No subjects achieved HBsAg seroclearance. Ten subjects (5.6%) had a significant HBsAg reduction of >0.5 log IU/mL/year after entecavir cessation; all 10 subjects however, had virologic relapse after 24 weeks off-treatment. When comparing the mean rate of HBsAg decline off-treatment and the mean rate of HBsAg decline during prior entecavir therapy among the 125 subjects with available data, there was no significant difference noted (0.009 and 0.068 log IU/mL/year, respectively, p=0.164).

Figure 2

Mean HBsAg (hepatitis B surface antigen) levels after cessation of entecavir. Mean represented by square markers, SD represented by error bars.

When considering all time points from treatment discontinuation to the date of last follow-up, there was no correlation between serial HBsAg levels and serial HBV DNA levels after entecavir cessation (r=−0.026, p=0.541).

Association between off-treatment virologic relapse and HBsAg levels

Univariate analysis of baseline factors associated with virologic relapse is shown in table 4A. There was no association between virologic relapse and serum HBsAg level at entecavir commencement (p=0.922), serum HBsAg levels at entecavir cessation (p=0.618), serum HBsAg levels <3 log IU/mL (p=0.591) or any of the remaining baseline factors (all p>0.05).

Table 4

Clinical characteristics of subjects with virologic relapse (n=163) versus subjects maintained in immune control (n=15) after cessation of entecavir stratified by baseline (A) and off-treatment (B) variables

Comparative analysis of off-treatment factors associated with virologic relapse is shown in table 4B. After multivariate analysis, off-treatment HBV DNA detectability at week 12 (p<0.001, OR 10.57) and higher ALT levels at relapse (p=0.033, OR 1.06) were significantly associated with virologic relapse; 76.1% (35/46) subjects with undetectable HBV DNA at week 12 still developed virologic relapse. The rate of HBsAg reduction after entecavir cessation had no significant association with virologic relapse (p=0.236).

Other clinical events

Among the six subjects who withdrew from the study, two were at the discretion of the investigators. One patient (50-year-old man), who had been on entecavir for 4 years and 10 months prior to recruitment, was noted to have a 1.5 cm hypoechoic shadow on ultrasonography 5 months after treatment cessation. He was referred for hepatectomy, with subsequent histology confirming moderately differentiated HCC. Entecavir was resumed in this patient after 24 weeks off-treatment. Another patient (62-year-old man) developed metastatic cholangiocarcinoma 7 months after treatment cessation; entecavir was resumed after 36 weeks off-treatment in anticipation of subsequent chemotherapy. These two patients were still in immune control (HBV DNA <20 and 667 IU/mL, respectively) when entecavir was resumed.

In addition, a 57-year-old woman developed a persistent rise in HBV DNA levels after resumption of entecavir. Resistance profile using a line probe assay (LiPA, Innogenetics NV, Gent, Belgium) found the mutations rtL180M, rtM204 V and rtT184S/C/G/A, confirming entecavir genotypic resistance. Resistance profile was also performed for the serum sample at entecavir commencement, confirming wild-type virus. This patient was switched to tenofovir monotherapy and achieved HBV DNA undetectability 3 months later. The remaining subjects achieved HBV DNA undetectability after reintroduction of entecavir; there were no cases of hepatic decompensation.


Our study results demonstrated a high rate of virologic relapse, 91.4%, within 48 weeks after cessation of entecavir in Asian HBeAg negative patients who had achieved excellent viral suppression with undetectable serum HBV DNA. Similar to treatment discontinuation in HBeAg positive disease after HBeAg seroconversion,23 ,24 treatment cessation in HBeAg negative CHB showed poor off-treatment sustainability of nucleos(t)ide analogue effectiveness.

Nucleos(t)ide analogue therapy, regardless of differences in antiviral potency, only has a small effect on the levels of intrahepatic covalently closed circular DNA (cccDNA),25 the template of HBV DNA transcription. Moreover, replenishment of cccDNA26 and the reinfection of new hepatocytes via residual integrated viral DNA27 persist during nucleos(t)ide analogue therapy. Therefore, despite the prior powerful antiviral effect of entecavir and a mean therapy duration of more than 3 years, a high rate of virologic rebound was seen in our cohort after treatment discontinuation. Although off-treatment HBV DNA detectability at week 12 was independently associated with virologic relapse, 76.1% of subjects with undetectable HBV DNA at week 12 still had virologic relapse. Hence, our study results would suggest that nucleos(t)ide analogue therapy should be continued indefinitely in Asian CHB patients, at least until the definitive endpoint of HBsAg seroclearance.28

The relapse rate was higher than that seen in similar studies.14 ,16 ,18 We analysed all recruited subjects using an intention-to-treat principle and avoided the potential bias from excluding subjects with an inadequate off-treatment follow-up duration.18 We also believe the results observed in the present study were accurate because of the large number of HBeAg negative subjects (n=184) who were prospectively recruited and closely monitored. Given that the mean duration of entecavir treatment was more than 3 years, the high rate relapse was unlikely due to inadequate treatment duration. It is however possible that the duration of infection could have a bearing on the relapse rate after treatment cessation. Asian CHB patients were infected at a very young age,29 which could result in poorer off-treatment durability; patients of European descent, on the other hand, are usually infected later in life and could thus have better rates of virologic control after treatment discontinuation.14

In the present study, instead of using biochemical hepatitis to define relapse, we chose a stringent yet appropriate HBV DNA level of >2000 IU/mL, the level associated with increased long-term risk of liver-related complications.11 ,12 The off-treatment virologic kinetics seen in our study also support our current criteria used for retreatment. Once HBV DNA becomes detectable after treatment cessation, subjects with subsequent relapse, unlike those remaining in immune control, demonstrated a continued serial increase in HBV DNA levels (table 2); thus without retreatment, it is very likely the continued escalation of HBV DNA levels could eventually lead to biochemical hepatitis and hyperbilirubinaemia. And in spite of adopting our current stringent definition of relapse, we still observed a relatively high rate of biochemical hepatitis (25.8%) accompanying virologic relapse, including a case with ALT reaching 1058 U/L only after 12 weeks off-treatment, highlighting the potential hazard of stopping therapy.

Another potentially negative consequence of treatment cessation would be the possible introduction of drug resistance, as highlighted by a case of primary entecavir resistance noted after drug reintroduction. Drug non-adherence would lead to suboptimal drug concentration levels which could be linked to the development of antiviral resistance,3032 and could occur even when a drug with high genetic barrier to resistance like entecavir were used.

A key strength of our study is the large number of recruited subjects (n=184), allowing adequate power in assessing the role of HBsAg levels in predicting virologic relapse. Nonetheless, our study results failed to demonstrate any association between HBsAg levels and virologic relapse. This can be explained by the decoupling of serum HBsAg and HBV DNA production in HBeAg negative disease.19 Serum HBsAg secretion surpasses the required amount for virion assembly; it is often secreted as empty subviral particles that remain in circulation even when viral load decreases. HBsAg, in the absence of viral replication, can also be produced via viral integration, a non-essential event in the life cycle of HBV.33 With HBsAg and HBV DNA undergoing different viral kinetics, HBsAg would not be expected to play a major role in determining outcomes related to changes in viral replication brought about by treatment cessation.

Our study found off-treatment mean HBsAg levels to decline very gradually (0.018 log IU/mL/year), similar to the slow rate found in natural history studies34 and during nucleos(t)ide analogue therapy,9 ,35 and unlike the fluctuating nature of HBV DNA levels. Significant HBsAg decline is usually only found in patients experiencing strong immune clearance of HBV, for example in patients with severe reactivation of HBeAg negative hepatitis.36 Hence in the treated CHB population with good virologic control, any major variations in HBsAg levels would not be expected.

On the other hand, a certain degree of linkage could still exist between HBsAg and HBV DNA, but possibly only at very low HBsAg titres when there is good immune control. Treatment cessation has been shown to be durable after reaching the standard treatment endpoint of HBsAg undetectability (ie, <0.05 IU/mL).28 In addition, low HBsAg levels (ie, 10–200 IU/mL) in treatment-naïve subjects were predictive of continued HBV DNA undetectability and subsequent HBsAg seroclearance.37 ,38 ,39 Our study only recruited a small number of subjects with baseline HBsAg <100 IU/mL (n=10), the level suggested to be predictive of subsequent HBsAg seroclearance in treated patients.21 Future studies could concentrate on specifically recruiting subjects with HBsAg <100 IU/mL to investigate if an adequately low HBsAg threshold could predict off-treatment durability.

Our study is limited by the lack of HBV genotyping, which might not be feasible due to the undetectable HBV DNA at the time of patient recruitment. Nonetheless, as shown previously, there was no difference in virologic relapse noted when comparing genotypes B and C, the two most common HBV genotypes found in our locality.18 HBV genotyping did not seem to be a major determinant of HBsAg kinetics.9 ,34 ,38 The high proportion of patients with relapse from the present study also suggests that genotyping is unlikely to play any significant role. In addition, a longer follow-up period of more than 48 weeks would be needed to robustly conclude if subjects with off-treatment HBV DNA <2000 IU/mL would continue to maintain good immune control long term.

In conclusion, a high rate of virologic relapse (91.4% within 48 weeks) was seen after discontinuation of entecavir in Asian HBeAg negative CHB patients. Serum HBsAg levels did not have any significant bearing in predicting off-treatment virologic relapse. Given the poor off-treatment sustainability of nucleos(t)ide analogue therapy, treatment of indefinite duration should be recommended, at least until the recognised treatment endpoint of HBsAg seroclearance has been reached.


The authors would like to thank Ms Angel Chim, Mr Ringo Wu, Mr Charles Cheng and Ms Rosita Chan for their efforts in data acquisition and arrangement of patient follow-up, and also Mr John Chi-Hang Yuen and Mr Chi-Hang Tse for their assistance in different laboratory measurements.


  • Contributors All authors have participated in the preparation of the manuscript and have seen and approved the final version. W-KS was involved in study concept and design, acquisition of data, analysis and interpretation of data and drafting of manuscript. AJH was involved in study concept and design and acquisition of data. VW-SW, GLHW and KS-HL were involved in acquisition of data. C-LL was involved in study concept and design and in critical revision of manuscript. M-FY and HL-YC were involved in study concept and design, analysis and interpretation of data, critical revision of manuscript and overall study supervision.

  • Funding This study was supported by an unrestricted grant from Roche Diagnostics.

  • Competing interests HL-YC is an advisory board member of Bristol Myers Squibb, Gilead Sciences, Roche Pharmaceutical and Roche Diagnostics. C-LL and M-FY are advisory board members of Bristol Myers Squibb, Gilead Sciences and Roche Diagnostics. VW-SW is an advisory board member of Gilead Sciences and Roche Pharmaceuticals. W-KS and GL-HW are advisory board members of Gilead Science and have also served as speakers for Bristol Myers Squibb.

  • Ethics approval The respective institutional research boards of all three study centres.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Relevant Article

Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of Gut.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.