Article Text

Download PDFPDF

Limited sustained response after stopping nucleos(t)ide analogues in patients with chronic hepatitis B: results from a randomised controlled trial (Toronto STOP study)
  1. Kin Seng Liem1,2,
  2. Scott Fung1,
  3. David K Wong1,
  4. Colina Yim1,
  5. Seham Noureldin1,
  6. Jiayun Chen1,
  7. Jordan J Feld1,3,
  8. Bettina E Hansen1,4,
  9. Harry L A Janssen1
  1. 1 Toronto Centre for Liver Disease, Toronto General Hospital, University Health Network, Toronto, Ontario, Canada
  2. 2 Department of Gastroenterology and Hepatology, Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlands
  3. 3 Sandra Rotman Centre for Global Health, Toronto, Ontario, Canada
  4. 4 Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada
  1. Correspondence to Professor Harry L A Janssen, Toronto Centre for Liver Disease, Toronto General Hospital, University Health Network, Toronto, ON M5T 2S8, Canada; harry.janssen{at}


Objective Although most patients with chronic hepatitis B (CHB) reach effective virological suppression with long-term nucleos(t)ide analogues (NA) therapy, some might not need to continue treatment for life. In this randomised, controlled, phase IV trial, we evaluated off-therapy outcomes in patients after discontinuing long-term NA therapy.

Design Patients who had received NA therapy for ≥1 year and achieved virological suppression (hepatitis B e antigen (HBeAg) seroconversion combined with undetectable hepatitis B virus (HBV) DNA ≥12 months in HBeAg-positive patients or undetectable HBV DNA ≥36 months in HBeAg-negative patients) were randomised 2:1 to stop or continue NA therapy for 72 weeks. Sustained disease remission (HBeAg negative, HBV DNA <2000 IU/mL and normal alanine aminotransferase (ALT)) was evaluated at 72 weeks after stopping NA therapy.

Results Among 67 enrolled patients, sustained disease remission was observed in 13/45 (29%) stop versus 18/22 (82%) continue patients. Hepatitis B surface antigen (HBsAg) loss occurred in two patients (one in each group). The median HBsAg decline from randomisation to week 72 was similar in both groups (0.2 (0.0–0.4) vs 0.1 (0.0–0.2) log IU/mL in stop vs continue patients). Among patients who stopped, 15/45 (33%) had virological or biochemical relapse and 17/45 (38%) were retreated according to predefined criteria. A total of 11/18 (61%) pretreatment HBeAg-positive versus 6/27 (22%) HBeAg-negative patients required retreatment (p=0.01). Fourteen (31%) patients developed ALT >10× upper limit of normal (ULN) and another 7 (16%) had ALT >5× ULN. No patients experienced liver decompensation or died.

Conclusion The findings of this prospective study suggest limited benefit of stopping NA therapy in chronic hepatitis B.

Trial registration number NCT01911156.

  • viral hepatitis
  • discontinuation
  • nucleos(t)ide analogue
  • HBeAg negative
View Full Text

Statistics from

Significance of this study

What is already known on this subject?

  • Discontinuation of nucleos(t)ide analogues (NA) therapy in patients with chronic hepatitis B may lead to functional cure, defined as HBsAg loss.

  • However, previous studies have highlighted contradictory findings because of widely ranging rates of relapse and sometimes dangerous flares.

What are the new findings?

  • Stopping NA therapy led to a high rate of relapse and retreatment and a low rate of HBsAg loss.

  • A post hoc analysis showed that off-therapy outcomes were worse for pretreatment HBeAg-positive than HBeAg-negative patients.

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

  • These findings in an Asian majority cohort suggest limited benefits of stopping NA therapy, especially if patients were HBeAg positive at the start of therapy. At this moment, NA treatment should only be stopped in highly selected, well-monitored patients or in the setting of clinical research.


The goals of treatment for patients with chronic hepatitis B (CHB)—improving survival and quality of life—can be achieved by nucleos(t)ide analogues (NA).1 However, since NA interfere late in the viral life cycle and do not directly inhibit or degrade cccDNA, even with long-term therapy, patients rarely achieve functional cure.2 Because of this, long-term NA therapy has become a prerequisite for most patients to improve their clinical outcomes. However, prolonged NA therapy increases risks of treatment non-adherence and adverse events (AEs) (renal impairment, bone loss and virological resistance), as well as greater healthcare expenditures and inconvenience to patients.3 There has thus been great interest in determining whether patients can stop NA therapy before achieving HBsAg loss.

The early studies had ambiguous results, showing the potential of stopping NA therapy, with relatively high HBsAg loss rates, but also highlighting a wide range of virological relapse that sometimes led to dangerous flares.4–7 The current hepatitis B virus (HBV) clinical practice guidelines have included considerations for NA discontinuation that were based on low-level evidence from several retrospective or small prospective studies.8–10

To clarify the potential benefit of NA treatment withdrawal, we performed a randomised controlled trial to evaluate the safety and efficacy of stopping long-term NA therapy in HBeAg-negative patients.

Materials and methods

Trial design

This was an investigator-initiated, single-centre, randomised, open-label, phase IV clinical trial (Toronto STOP study; registered at, Identifier: NCT01911156) conducted at the Toronto Centre for Liver Disease (University Health Network, Canada) from May 2016 to May 2018. Patients were randomised (baseline) 2:1 to discontinue (stop group) or continue NA monotherapy (continue group) after which patients were followed for 72 weeks (end of follow-up (EOF)).

The random allocation sequence was computer generated in blocks of six by a biostatistician. After an eligible patient provided written informed consent, a research nurse who was not involved in the trial opened the next numbered sealed envelope with assigned intervention. Due to the nature of the intervention, patients, clinicians and study team could not be blinded to treatment allocation.

Patients in the stop group had study visits at week 4, 6, 12, 18, 24, 36, 48, 60 and 72, whereas patients in the continue group had visits at week 24, 48 and 72. The latter group continued using the NA they had been prescribed or received tenofovir disoproxil fumarate (TDF) 300 mg once daily if treatment continuation was not possible for logistic or financial reasons.

Patients in the stop group were retreated if they fulfilled one of the following criteria: HBeAg seroreversion (reappearance of HBeAg in serum); HBV DNA >2000 IU/mL and alanine aminotransferase (ALT) >600 IU/mL at any visit; HBV DNA >2000 IU/mL and ALT >5× upper limit of normal (ULN: 40 IU/mL) on two consecutive visits; HBV DNA >2000 IU/mL and ALT >200 IU/mL but <600 IU/mL for >6–8 weeks; or HBV DNA >20 000 IU/mL on two consecutive visits at least 4 weeks apart. The final decision to restart antiviral treatment was at the discretion of the treating physician.

ALT flare management was also at the discretion of the treating physician and was based on the severity of flares and signs of (impending) liver failure. In an ALT flare event, clinical assessments were done at least once a month until ALT levels reached <200 IU/L. During these visits, patients underwent physical examination and routine biochemical, virological and haematological testing. Retreated patients then continued follow-up according to the scheduled study visits.

An independent Data Safety Monitoring Board reviewed all safety data on a regular basis and advised the study team on requirements to amend or terminate the study. This study was approved by the research ethics board of University Health Network in Toronto and performed in concordance with Good Clinical Practice guidelines and the Declaration of Helsinki (2013). All patients provided written consent.

Study population

Patients with CHB (HBsAg positive >6 months), aged 18 years or above, were eligible if they had received NA monotherapy for ≥1 year prior to screening, had documented HBeAg status at start of NA therapy and achieved virological suppression on NA therapy. Pretreatment HBeAg-positive patients were included if they were HBeAg negative, anti-HBe positive and HBV DNA undetectable for at least 12 months prior to screening, and pretreatment HBeAg-negative patients if they had undetectable HBV DNA for at least 36 months prior to screening. Patients were excluded for the following reasons: treatment with any investigational drug within 30 days of screening; ALT >10× ULN; creatinine clearance <70 mL/min; presence of cirrhosis as documented by biopsy within 5 years, liver stiffness measurement (FibroScan) >9 kPa and/or Fibrotest >0.48; neutropenia (neutrophils <1000/mm3); coinfection with hepatitis C or D virus and/or HIV; other acquired or inherited causes of liver disease (alcoholic liver disease, fatty liver disease, drug-related liver disease, autoimmune hepatitis, haemochromatosis, Wilson’s disease or α−1 antitrypsin deficiency); α-fetoprotein >50 ng/mL; hyperthyroidism or hypothyroidism; immune suppressive treatment within the previous 6 months; pregnancy and/or lactation; significant pulmonary dysfunction in the previous 6 months, malignancy other than skin basocellular carcinoma in previous 2 years, immunodeficiency syndromes (eg, HIV positivity, autoimmune diseases, organ transplants other than cornea and hair transplant); any medical condition requiring, or likely to require chronic systemic administration of steroids, during the course of the study; substance abuse (alcohol (>80 g/day) and intravenous or inhaled drugs (past 2 years); and any other condition which, in the opinion of the investigator, would make the patient unsuitable for participation or could interfere with the patient participating in and completing the study.

Efficacy analysis

The efficacy analysis comprised all patients who were randomised. The primary endpoint of sustained response was defined as HBV DNA <2000 IU/mL 48 weeks after baseline. Secondary outcomes included virological outcomes (HBV DNA <20/200/2000 IU/mL); serological endpoints (HBeAg seroreversion, sustained disease remission (HBeAg negative, HBV DNA <2000 IU/mL and ALT normalisation), HBsAg loss, HBsAg decline from baseline); biochemical measures (ALT normalisation, clinical relapse (HBV DNA >2000 IU/mL+ALT >1.5 x ULN)); and histological responses (liver stiffness).

Study follow-up and measurements

At every study visit, a physical exam and routine biochemical and haematological tests were performed. Plasma and serum research samples were collected and stored at −80°C. An assessment of liver fibrosis was done at baseline and week 48 by FibroScan or Fibrotest measurement. Serum ALT values were standardised by dividing the value by the ULN (40 IU/mL). Virological and serological tests were analysed at the core laboratory of University Health Network (Toronto, Canada). Serum HBV DNA was measured by Cobas TaqMan 48 PCR assay (lower limit of detection: 20 IU/mL; Roche Diagnostics, Basel, Switzerland). Serum HBeAg, anti-HBe and HBsAg were analysed by Architect (Abbott Laboratories, North Chicago, Illinois, USA; lower limit of detection: 0.05 IU/mL). If HBV genotype could not be assessed due to undetectable HBV DNA levels at baseline, historic HBV genotype results were used, where possible. The presence of cirrhosis was defined by Ishak stage 6 on liver biopsy.

Safety analysis

The safety analysis included all patients who were randomised. Safety measures included recording and grading of AEs (vital signs, and chemistry and haematology data, analysed according to the modified WHO grading system, adapted for chronic liver disease). The causality of AEs was determined by the investigator.

Statistical analysis

The power analysis was based on the primary endpoint, which was estimated to occur in 60% of patients who discontinued NA monotherapy and in 95% of patients who continued NA monotherapy. After correction for duration of HBeAg loss and 2:1 weighted randomisation, power fixed at 80% and a two-sided α-level of 0.05, the required sample size was 58 patients. Taking into account a 10% drop-out rate, a total of 66 patients were required to reach the minimum sample size: 44 patients in stop group and 22 patients in the continue group.

A data analysis plan was specified prior to freezing the database. No imputation was performed for missing data for the primary endpoint or any secondary outcomes. At the time of a missing visit or at the time of retreatment, patients were coded as not having achieved a virological, serological or biochemical outcome. Variables are summarised by mean±SD or frequency (percentage). Non-normally distributed variables were log-transformed. Outcomes were compared by χ2 test, Student’s t-test or Mann-Whitney test, where appropriate. Cumulative rates of relapse were evaluated by Kaplan-Meier method and tested by log-rank test and Cox proportional hazards regression. Analyses were performed in SPSS (V.25.0) and SAS V.9.4. Two-sided p values <0.05 were considered significant.


Study cohort

A total of 159 patients were screened, of which 88 patients declined participation or were ineligible (figure 1). One patient was excluded due to HBsAg loss at baseline and two patients due to withdrawing consent prior to randomisation. Of the 67 enrolled patients, 45 (67%) patients were randomly allocated to stop and 22 (33%) to continue NA therapy. The baseline characteristics were balanced between the treatment arms (table 1). The mean (SD) age was 49 (10) years, 60% of patients were male and all but two patients were of Asian ethnicity. Forty (60%) patients were HBeAg negative, and 27 (40%) patients HBeAg positive at the start of NA therapy. Sixty-one (92%) patients were anti-HBe positive at baseline. After inclusion of the first 11 subjects who were allowed to be anti-HBe positive or negative at baseline, the study protocol was amended to only include patients if they were anti-HBe positive at baseline. The mean duration of NA therapy was 7.3 (2.9) years and HBsAg level was 3.0 (0.7) log IU/mL at baseline.

Figure 1

Flow chart of patient inclusion.

Table 1

Characteristics per treatment arm of 67 HBeAg-negative patients with chronic hepatitis B

Sustained response, HBsAg loss and retreatment

The primary endpoint of HBV DNA <2000 IU/mL at week 48 was observed in 12/45 (27%) stop versus 21/22 (95%) continue patients (p<0.005; table 2). A total of 12/45 (27%) stop patients versus 19/22 (86%) continue patients remained in disease remission with HBV DNA <2000 IU/mL and ALT <1.5× ULN (p<0.005) at week 48. At week 72, HBV DNA <2000 IU/mL was observed in 13/45 (29%) stop patients and in 20 (91%) continue patients (p=0.04). Throughout 72 weeks of follow-up, 28/45 (62%) stop patients remained off-treatment. These patients had used NA therapy on average for 6.5 (2.3) years, 25% were HBeAg positive at start of therapy and 96% was anti-HBe positive at baseline.

Table 2

Primary and secondary outcomes at week 24–72 (n=67)

HBsAg loss occurred in two patients, one in each randomised arm. Of these two, only the stop patient developed antibodies to HBs. This patient was a 66-year-old Asian woman who had been treated with TDF for 11.5 years, was HBeAg negative at start of therapy, anti-HBe positive at baseline and had a HBsAg at baseline of 2.8 log IU/mL. In comparison, HBsAg loss in the NA continuation group occurred in a 59 year-old Caucasian man who had used TDF for 9.4 years. The patient was HBeAg negative at start of therapy, anti-HBe negative at baseline and had a very low HBsAg value at baseline (0.1 log IU/mL). Both patients exhibited small fluctuations in quantitative HBsAg after an initial negative HBsAg test.

Among patients who stopped NA therapy, 17/45 (38%) required retreatment by week 72 (figure 2). The median time to retreatment was 12 weeks. One (6%) patient was retreated due to HBeAg seroreversion on two consecutive visits, 12 (71%) patients for virological relapse >20 000 IU/mL twice and 4 (24%) patients because of combined virological and biochemical relapse, of which 2 had concurrent imaging suggestive of fibrosis/cirrhosis development. The median (IQR) time to retreatment was 12 (10–30) weeks. Retreatment led in all 17 stop patients to prompt virological suppression, HBeAg negative status and ALT normalisation. Six (13%) other stop patients developed clinical relapse (HBV DNA >2000 + ALT>1.5× ULN) by the EOF that did not meet the study requirements for retreatment and for which the investigator did not restart therapy.

Figure 2

Beneficial and adverse outcomes in 45 NA stop patients (A). Panels B and C show results of pretreatment HBeAg-positive and HBeAg-negative patients, respectively. The last-value-carried-forward method was used in case of missed study visits. ALT, alanine aminotransferase; NA, nucleos(t)ide analogue.

Virological, serological and ALT kinetics

Virological, serological and biochemical kinetics are depicted in figure 3. Virtually all NA stop patients had a virological relapse within the first 12 weeks of NA discontinuation with peak HBV DNA values of 8 log IU/mL. Continuously suppressed HBV DNA was observed in the NA continuation group throughout follow-up.

Figure 3

Virological, serological and biochemical kinetics in patients who stopped or continued NA treatment (n=67). NA, nucleos(t)ide analogue.

The median HBsAg decline from randomisation to week 72 was 0.2 (0.0–0.4) in the stop group and 0.1 (0.0–0.2) log IU/mL in the continuation group and was not associated with peak ALT or HBV DNA values after stopping (p>0.05). Similarly, among the 17/45 retreated patients, the development of ALT flares did not influence HBsAg decline from randomisation to week 72. The median (IQR) HBsAg decline was 0.4 (0.3–0.7) versus 0.2 (−0.1–0.3) log IU/mL in 13 retreated stop patients with and four without ALT flare (p=0.09). Among these retreated patients, HBsAg declines were not associated with ALT values at the time of retreatment or peak ALT values throughout off-therapy follow-up (p>0.05). Four (8.9%) stop patients achieved HBsAg decline >1 log IU/mL from baseline to week 72 compared with none of the continue patients (p=0.16). Six (13%) stop and 2 (10%) continue patients achieved HBsAg <100 IU/mL at the EOF (p=1.00).

Because retreatment may have altered the disease course of patients stopping therapy, we compared the difference in peak HBsAg decline and its rate by retreatment. Neither the mean (SD) peak HBsAg decline nor the proportion of patients with >1 log IU/mL HBsAg decline was significantly different according to retreatment (data not shown).

Since only 5/67 (7.5%) patients had baseline HBsAg <100 IU/mL, the small number of patients with low baseline HBsAg values reduced the strength of a stratified analysis on HBsAg decline. After determining category thresholds for baseline HBsAg based on the histogram, 11/27/25 patients were categorised into baseline qHBsAg groups 0.0–2.3, 2.3–3.3 and >3.3 log IU/mL. HBsAg decline at week 72 was 0.4 (0.5), 0.4 (0.7) and 0.1 (0.1) log IU/mL, respectively, and was not significantly different between groups (p=0.11).

Liver stiffness measurements

At week 72, liver stiffness measurements by FibroScan were 5.2 (1.6) versus 5.3 (1.6) kPa in the NA stop versus continue group (p=0.90; table 2); the liver stiffness decline at week 72 from baseline was 0.0 (−0.9–1.0) versus 0.1 (−0.5–0.9; p=0.83) kPa.

Outcomes according to pretreatment HBeAg status

Baseline values according to pretreatment HBeAg status are shown in online supplementary table 1. The primary endpoint was achieved by 3/18 (17%) initially HBeAg-positive versus 7/24 (29%) initially HBeAg-negative stop patients (p=0.10; figure 2). A greater proportion of pretreatment HBeAg-positive compared with HBeAg-negative stop patients was retreated (11/18 (61%) versus 6/27 (22%); log-rank p=0.01). The median time to retreatment was similar for pretreatment HBeAg-positive and HBeAg-negative patients (12.3 (10.7–49.9) vs 12.7 (9.3–23.1) weeks; p=0.37). In a post hoc, multivariable, logistic regression analysis that included relevant baseline factors, HBeAg-positive status at start of treatment was the only independent predictor of relapse (OR (95% CI): 7.4, (1.3–42.6); p=0.03). Importantly, end-of-therapy qHBsAg, HBV genotype, ALT, METAVIR fibrosis score or the duration of NA consolidation therapy were not associated with relapse (p>0.05).

Supplementary data

Supplementary file

Furthermore, HBeAg seroreversion occurred at least transiently in 8/45 (18%) NA stop patients. These patients met retreatment criteria other than HBeAg seroreversion first but had HBeAg seroreversion after retreatment. All were HBeAg positive at start of therapy, and 3/8 (38%) patients were anti-HBe negative at end-of-therapy. HBeAg seroreversion persisted for at least two consecutive visits in four stop patients, of which three had simultaneous clinical relapse. The four other stop patients experienced HBeAg seroreversion at a single visit with concurrent clinical relapse, all of whom achieved HBeAg negative status, HBV DNA <20 IU/mL and ALT normalisation on retreatment.

The median HBsAg change from randomisation to week 72 was 0.0 (−0.1–0.2) versus 0.2 (0.0–0.3) log IU/mL in HBeAg-positive versus HBeAg-negative stop patients (p=0.04) (see online supplementary figure 1). The median peak ALT (13.7 (2.8–21.2) vs 5.0 (1.7–8.5) × ULN; p=0.03) and mean peak HBV DNA values (6.1 (1.8) vs 5.0 (1.3) log IU/mL; p=0.02) were significantly different between pretreatment HBeAg-positive and HBeAg-negative stop patients. However, these differences disappeared when the results were stratified according to retreatment (p>0.05 in all groups).

We also analysed outcome according to anti-HBe status at treatment discontinuation. However, the small number of patients (n=5) who were anti-HBe negative at stopping treatment limited a comprehensive analysis. A subgroup analysis that excluded these five baseline anti-HBe negative patients showed results similar to the overall group.


Safety data are reported in table 3. The vast majority of grade 3 AEs comprised ALT elevations >5× ULN. Among patients who stopped therapy 14 (31%) developed ALT >10× ULN (peak ALT: 41× ULN) and another 7 (16%) patients had ALT >5× ULN. No flares occurred in those who continued NA therapy. One of the flares occurred in a pretreatment HBeAg-positive patient whose bilirubin increased to 68 μmol/L after stopping NA therapy without developing signs of decompensation. One other patient in the stop group underwent an elective cataract surgery during posttreatment follow-up. None of the patients developed hepatic decompensation, serious AEs or died.

Table 3

Safety profile up to week 72


In this largest prospective RCT to date on NA withdrawal in CHB, 27% of patients had sustained response, 71% relapsed and 2% achieved HBsAg loss 72 weeks after stopping therapy. HBsAg levels in the stop group declined only marginally and were not different from the NA continuation group. These findings suggest that stopping NA therapy confers little benefit in our population of mainly Asian patients.

Our findings that very few patients achieved HBsAg loss and had minimal HBsAg decline contradicts results from previous studies. In a RCT that randomised 42 HBeAg-negative mainly Caucasian patients to stop or continue NA therapy, 19% of stop patients achieved HBsAg loss by week 144 off-therapy (6.3% annually), which was significantly more than in the continuation group. The median HBsAg change was −0.59 log IU/mL.4 A large retrospective study from Taiwan reported HBsAg loss in 13% of 691 patients at 6 years off-therapy follow-up (1.8% annually).5 Although the follow-up of our study was 1.5 years, the minute HBsAg decrease over time in all patients suggests that few will achieve HBsAg loss during longer follow-up. In addition to ethnicity, the discrepant HBsAg findings between studies may be due to differences in end-of-therapy HBsAg values, HBV genotype, the duration of NA consolidation therapy and retreatment criteria.11 12

Remarkably, pretreatment HBeAg-positive patients were three times more likely than HBeAg-negative patients to require retreatment for relapse after stopping. These results challenge previous claims of a more sustained off-therapy response for HBeAg-positive than HBeAg-negative patients.13 Importantly, even though pretreatment HBeAg-positive patients only required 1 year of NA therapy following HBeAg seroconversion for study entry, the mean (SD) duration of consolidation therapy was 3.4 (1.7) years and did not differ between those who did and did not require retreatment. In a large, prospective cohort study from China, the cumulative virological relapse rate at 10 years off-therapy was 31% versus 62% in 138 pretreatment HBeAg-positive and 85 HBeAg-negative patients.14 In this study, however, 218/223 (98%) patients had used older generation NA therapies (adefovir dipivoxil, emtricitabine, lamivudine or telbivudine); the end-ff-therapy fibrosis stage was unknown, and since HBV DNA was measured with a less sensitive assay (lower limit of quantification: 200 IU/mL), a low residual viral load could have influenced relapse rates. Notably, different studies have used different definitions for relapse and for retreatment criteria, which could account for the discrepant results. Furthermore, at least transient HBeAg seroreversion occurred in 8/45 (18%) patients in our study, raising concern that stopping in this population is not helpful and potentially harmful.

The host’s immune system plays an integral role in viral control during and after NA therapy. Long-term NA consolidation treatment partially restores the liver-specific cellular immune response and reduces the risk of relapse off-therapy.12 15 16 The patients in our study had received substantially longer NA consolidation therapy (3.8 years for HBeAg-positive patients; 6.0 years for HBeAg-negative patients) than the 6–12 months as recommended by the current guidelines.8–10 Nonetheless, the high rate of relapse and of retreatment that occurred predominantly within the first 24 weeks suggests that the recommended duration of consolidation therapy might be too short or not useful at all. A previous study from our group found that prolonging consolidation therapy beyond 3 years reduced the risk of persistent virological relapse.12 We were not able to confirm these findings in the current study because almost all patients had greater than 3 years of consolidation therapy at study entry, and a relatively high rate of relapse was still seen.

Some authors have suggested that virological rebound off-therapy precipitates HBsAg loss.17 This pattern was also visible in the single stop patient who achieved HBsAg loss after a simultaneous decline in HBsAg and rise in viral load. However, overall, we saw no change in HBsAg levels after stopping therapy, and HBsAg decline was not associated with peak ALT or HBV DNA values. The overall lack of HBsAg decline in our study precluded any further exploration.

A critical point to discuss with patients who consider stopping NAs is the small but significant risk of adverse outcomes. In a systematic review of NA withdrawal studies hepatic decompensation occurred in 0.8%–3% of patients.18 At least 11 deaths (0.2%–1.6% of patients) have been reported due to off-therapy flares in various studies, at least some in the presumed absence of cirrhosis.7 13 19 Since none of the published studies have been able to find reliable predictors of relapse, a visit-intensive monitoring schedule must be reinforced. Essentially all episodes of retreatment in our study were preceded by lone virological relapse in the previous study visit. Further exploration of the predictive value of HBV DNA and new virological biomarkers, such as HBcrAg or HBV RNA, holds promise to improve identifying those at highest risk for AEs.

An important strength of the present study, which is the largest randomised controlled trial on NA discontinuation, was the inclusion of both pretreatment HBeAg-positive and HBeAg-negative patients. Furthermore, the retreatment criteria of our study were not as stringent as in other studies, which allowed us to better evaluate the off-therapy viral and biochemical kinetics during close-visit monitoring. We balanced the retreatment criteria such that we allowed the viral load to rebound but not at the costs of dangerous flares.

A limitation to our study was the inclusion of a predominantly Asian cohort. This prohibited us from studying the influence of HBV genotype or race on off-treatment outcomes. In addition to further exploring ethnicity-related questions in a heterogeneous population, future studies should ideally be designed with well-balanced, predefined retreatment criteria that are not based on current guideline recommendations and investigate long-term clinical outcomes.

In conclusion, the findings from this predominantly Asian cohort suggest that NA discontinuation before HBsAg loss has limited benefits, especially for pretreatment HBeAg-positive patients.


Data collection and data management at the trial centre was supported by Victor Lo and Doinita Vladutu. Feng Fei Huang and Lianne Thai helped with data collection and patient care. We would like to thank Dr Tony Mazzulli for providing part of the quantitative HBsAg data.


View Abstract


  • Correction notice This article has been corrected since it published Online First. The abstract has been corrected.

  • Contributors HLAJ and BEH: study design and study supervision; JC, KSL: data acquisition; KSL, BEH and HLAJ: analysis and interpretation of data; DKHW, SF, CY and JJF: study clinician advisers and contributed to data interpretation; BEH: study statistician; KSL: drafting of the manuscript; all authors: critical revision and approval of the final manuscript.

  • Funding This investigator-initiated study was organised and sponsored by University Health Network (Toronto, Canada). Gilead Sciences, Inc. (Foster City, US) has provided funding support and study medication for patients who required but did not have access to tenofovir disoproxil fumarate.

  • Disclaimer The funding sources did not have any influence on study design, data collection, analysis and interpretation of the data, writing of the report or the decision to submit for publication.

  • Competing interests BEH has received grants from and is consultant to Intercept Pharmaceuticals. HLAJ has received grants from and is consultant to Abbott, Bristol-Myers Squibb, Gilead Sciences, Merck, Novartis, Roche and Janssen. JJF reports receiving support for research or scientific consulting from Abbott, Abbvie, Contavir, Enanta, Gilead Sciences, Janssen and Roche. SF has received research support from Gilead Sciences and speaking and teaching and/or consulting fees from Gilead Sciences, Merck and AbbVie. DKHW has received speaking and teaching fees from AbbVie and Merck.

  • Patient consent for publication Not required.

  • Ethics approval This study was approved by the University Health Network Research Ethics Board (protocol number: 13-6159).

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

  • Data availability statement Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles