The Journal of Infectious Diseases MAJOR ARTICLE

Transcription

1 The Journal of Infectious Diseases MAJOR ARTICLE Off-Treatment Hepatitis B Virus (HBV) DNA Levels and the Prediction of Relapse After Discontinuation of Nucleos(t)ide Analogue Therapy in Patients With Chronic Hepatitis B: A Prospective Stop Study Jiawei Cao, 1,a Heng Chi, 2,a Tao Yu, 1 Zhandong Li, 1 Bettina E. Hansen, 2 Xiaoyong Zhang, 1 Chunxiu Zhong, 1 Jian Sun, 1 Jinlin Hou, 1 Harry L. A. Janssen, 2,3,b and Jie Peng 1,b 1 Department of Infectious Diseases, Guangdong Provincial Key Laboratory of Viral Hepatitis Research, Nanfang Hospital, Southern Medical University, Guangzhou, China; 2 Department of Gastroenterology and Hepatology, Erasmus MC University Medical Center Rotterdam, the Netherlands; and 3 Toronto Centre of Liver Disease, University Health Network, Toronto General Hospital, University of Toronto, Canada Background. The optimal management remains unknown after nucleos(t)ide analogue (NA) discontinuation in patients with chronic hepatitis B (CHB). This prospective study investigated the role of off-treatment viral kinetics in predicting relapse after discontinuation of NA therapy. Methods. A total of 82 noncirrhotic Asian patients with CHB who discontinued NA therapy according to international guidelines were prospectively followed. Patients with a hepatitis B virus (HBV) DNA level of >2000 IU/mL and an alanine aminotransferase (ALT) level of >2 times the upper limit of normal (clinical relapse) were retreated. Results. Sixty patients were HBV envelope antigen (HBeAg) positive at the start of treatment, and 22 were HBeAg negative. Clinical relapse developed in 28 patients (2-year rates, 31% among HBeAg-positive patients and 53% among HBeAg-negative patients). Age of 35 years (hazard ratio [HR], 0.37; P =.026) and end-of-treatment HBsAg level of 200 IU/mL (HR, 0.39; P =.078) were independently associated with lower relapse rates. A high risk of biochemical relapse (defined as an ALT level of >2 times the upper limit of normal) was observed if the HBV DNA level was > IU/mL when the level was initially elevated, compared with HBV DNA levels of >2000 to IU/mL (HR, 8.42; P <.001). The risk of biochemical relapse was also high in patients with persistent elevation in the HBV DNA level (confirmed to be >2000 IU/mL within 3 months), compared with the group with transient elevation (HR, 6.87; P <.001). Conclusions. After NA discontinuation, a lower relapse rate was observed in younger patients and in those with low end-oftreatment HBsAg levels. The level and persistence of off-treatment elevated HBV DNA levels were useful in the prediction of a subsequent biochemical relapse and may thus be used to guide off-treatment management. Keywords. Treatment cessation; antiviral agents; hepatitis B; off-treatment management. Nucleos(t)ide analogues (NAs) are inhibitors of hepatitis B virus (HBV) DNA polymerase and directly block the viral replication. It is an effective treatment option for chronic hepatitis B (CHB) but does not eradicate covalently closed circular DNA or the integrated HBV genome [1, 2]. NA discontinuation is not widely practiced and remains a controversial but highly relevant subject [3]. One reason is the lack of high-quality evidence for stopping. Most studies are retrospective with infrequent and irregular off-treatment follow-up Received 10 September 2016; editorial decision 5 January 2017; accepted 10 January 2017; published online January 19, a J. C. and H. C. are co first authors. b H. L. A. J. and J. P. are co senior authors. Correspondence: J. Peng, MD, Hepatology Unit, Nanfang Hospital, Southern Medical University, 1838 Guangzhou Ave North, Guangzhou, Guangdong Province, China (pjie138@163.com). The Journal of Infectious Diseases 2017;215:581 9 The Author Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, journals.permissions@oup.com. DOI: /infdis/jix025 and variable retreatment criteria hampering solid clinical recommendations [4]. In addition, some studies described very high rates of relapse and advised against NA cessation [5 7], whereas other studies demonstrated sustained response and HBV surface antigen (HBsAg) seroclearance in some patients, suggesting that discontinuation is possible [8 10]. Long-term indefinite therapy might not be feasible for all patients, especially in lower-income countries. One should also consider the long-term safety and compliance issues during long-term NA therapy [3, 11], particularly if more than 30 years is needed to achieve HBsAg seroclearance a functional cure [12, 13]. Different stopping criteria have been proposed to increase the likelihood of sustained response [14 16], and yet little is known about how to manage patients after NA cessation. The question remains whether HBV DNA relapse alone after NA cessation should be considered as failure for which retreatment is immediately indicated [4, 17]. One should also consider that a key element of treatment initiation in untreated patients Off-Treatment HBV DNA Level and Relapse JID 2017:215 (15 February) 581

2 as recommended by the international guidelines is the elevation of both serum HBV DNA and alanine aminotransferase (ALT) levels [14 16]. Furthermore, it has been demonstrated that sustained response may occur even after an increase in the HBV DNA level following NA discontinuation [10]. This is probably due to the development of an immune-mediated response after NA cessation, which can be successful in some patients [18]. No studies have investigated the optimal timing of retreatment in terms of the prevention of premature retreatment but also severe flares in the ALT level. The factors associated with ALT relapse after an initial HBV DNA relapse also remain unknown. The HBsAg level has been used as a marker of the complex interplay between the HBV and the host s immune response and may be useful in the guidance of treatment discontinuation [1, 19]. A retrospective study showed promising predictive ability of HBsAg levels for sustained response after NA cessation but only in HBeAg-negative patients [8]. Moreover, preliminary data of a randomized, controlled trial showed that patients who stopped NA treatment demonstrated a greater decrease in the HBsAg level than patients who continued NA therapy [17]. In addition, the important question remains of whether there are any signs of histological progression after NA discontinuation. The aim of this prospective study was to investigate the effect of NA therapy discontinuation in patients with CHB in terms of relapse, off-treatment HBsAg kinetics, liver stiffness measurement by transient elastography, and liver ultrasonography. Moreover, we investigated whether ALT relapse after an initial elevation in the HBV DNA level can be predicted by end-oftreatment and off-treatment factors. METHODS Study Design and Population This was a prospective, observational study of virologically suppressed Asian patients with CHB (defined as HBsAg positivity for >6 months) who discontinued NA therapy and consented to prospective study follow-up. Patients were recruited from Nanfang Hospital (Guangzhou, China) from November 2012 until January 2015 [20]. Based on previous retrospective studies, a minimum of 80 patients were planned to be included to study the rate and risk of relapse [10, 21]. This analysis included follow-up data until July 2015, when all patients had completed at least 6 months of follow-up. Patients who were HBeAg positive at NA initiation were eligible if they fulfilled the stopping criteria of international guidelines: HBeAg seroconversion, undetectable HBV DNA, and normalized ALT level, with a duration of subsequent consolidation therapy of 12 months [14 16]. Patients who were HBeAg negative at NA initiation were eligible if they at least fulfilled the Asian Pacific Association for the Study of the Liver (APASL) stopping criteria: undetectable HBV DNA with a duration of subsequent continued therapy of 18 months [15]. Additional eligibility criteria were age of 18 years, HBsAg positivity and undetectable HBV DNA at the moment of NA discontinuation, and written informed consent with motivation to attend follow-up visits. Exclusion criteria included coinfection with hepatitis C virus, hepatitis D virus, or human immunodeficiency virus; immunocompromised status or malignancy; autoimmune liver disease; alcohol abuse; prior history of liver transplantation; other severe or active disease; history or presence of decompensated liver disease (ultrasonography-confirmed ascites, bleeding esophageal varices, overt encephalopathy, a total bilirubin level of 2 times the upper limit of normal [ULN], an albumin level of <32 g/l, or a prothrombin time prolongation of 3 seconds, compared with the reference value); radiological suspicion of HCC or an α-fetoprotein level of >20 ng/ml; biopsy-proven cirrhosis; or a liver stiffness of >9 kpa as measured by transient elastography (Fibroscan; Echosens, Paris, France). This study was conducted in compliance with the Declaration of Helsinki and approved by the ethics committee (study identifier NFEC K3). Follow-up After treatment discontinuation, patients were followed every month during the first 3 months. Thereafter, the patients were evaluated every 3 months. At every visit, biochemical (determination of ALT, aspartate aminotransferase, total bilirubin, and albumin levels) and virological (quantitative measurement of HBsAg, HBeAg, anti-hbe, and HBV DNA) tests were performed. Ultrasonography and liver stiffness measurements were also performed at treatment discontinuation, year 1, and year 2. End Points, Retreatment Criteria, and Definitions Clinical relapse was defined as an HBV DNA level of >2000 IU/mL, combined with an ALT level of >2 times the ULN. Patients with clinical relapse were retreated and withdrawn from this prospective study follow-up. Biochemical relapse was defined as an ALT level of >2 times the ULN. Other end points included HBsAg seroclearance, off-treatment HBsAg kinetics, ultrasonographic signs of cirrhosis, and liver stiffness measurements. Ultrasonographic signs of cirrhosis were defined as nodules within the hepatic parenchyma, splenomegaly, or a portal vein diameter of >16 mm [22]. A valid liver stiffness value was defined as an interquartile/median ratio of 0.30, 10 successful measurements, and a success rate of 60%. As described by practice guidelines, consolidation therapy was defined as continued NA therapy after HBeAg seroconversion [14, 16], achievement of an undetectable HBV DNA level, and normalization of the ALT level in HBeAg-positive patients and after achievement of an undetectable HBV DNA level in HBeAg-negative patients [15]. Cutoffs in the HBsAg level (ie, 100, 200, and 1000 IU/mL) described in recent studies were further investigated [8, 9, 23]. 582 JID 2017:215 (15 February) Cao et al

3 Persistent elevation of the HBV DNA level was defined as at least 2 tests within 3 months in which the HBV DNA level was >2000 IU/mL. Laboratory Tests Local standardized automated techniques were used to perform biochemical tests. The ULN of the ALT level was 40 U/L for males and 35 U/L for females. Quantitative analysis of HBsAg (lower limit of detection [LLOD], 0.05 IU/mL), HBeAg, and anti-hbe was performed using the Architect assay (Abbott Laboratories, Chicago, IL). Data on pretreatment serum HBV DNA levels were retrospectively collected (by an HBV-specific polymerase chain reaction assay with a LLOD of 1000 copies/ml; Daan Gene, Sun Yat-sen University, Guangzhou, China]). A conversion factor of 5 copies per IU was used for conversion of copies to IU. Serum HBV DNA levels at NA cessation and follow-up were determined using the Cobas HBV-specific TaqMan polymerase chain reaction assay with a LLOD of 20 IU/mL (Roche Diagnostics, Basel, Switzerland). Data Analysis The date of NA cessation was defined as baseline. Follow-up was calculated from baseline until a particular study end point, loss of follow-up, censorship (retreatment), or the last follow-up visit, when appropriate. Characteristics of HBeAg-positive patients and HBeAgnegative patients were compared using the χ 2 test, for categorical variables, and the Student t test or Mann Whitney test, for continuous variables, when appropriate. Follow-up data were assessed with Kaplan-Meier analysis. The log-rank test was used to compare Kaplan-Meier curves. Associations between factors and study end points were examined by Cox proportional hazards regression. Multivariable Cox models were adjusted for start-of-treatment HBeAg status and factors that were statistically different between HBeAg-positive and HBeAg-negative patients. To investigate the predictors of biochemical relapse after an initial elevation in the HBV DNA level to >2000 IU/ ml, a time-reset approach was used in which the time of the first elevation in the HBV DNA level was considered as t = 0. Persistent elevation in the HBV DNA level was analyzed as a time-dependent covariate, allowing patients to switch to the group with persistent elevation in the HBV DNA level during follow-up. The decrease in the HBsAg level was calculated from baseline to the time of the last available level while considering the period of decrease. Declines were compared using the Student t test. Liver stiffness measurements at year 1 and year 2 were compared to baseline values, using the paired Student t test. Statistical tests were 2-sided, and a P value of <.05 was considered statistically significant. All tests were performed with SPSS, version 21.0 (IBM, Armonk, NY). RESULTS Patient Characteristics In total, 87 patients gave informed consent and were enrolled. Five patients were negative for HBsAg at week 0 and were excluded from this analysis. Therefore, this analysis included 82 patients who were HBsAg positive and HBV DNA negative at the moment of NA discontinuation. At the initiation of NA therapy, 60 patients (73%) were HBeAg positive, and 22 (27%) were HBeAg negative (Table 1). Characteristics of HBeAgpositive and HBeAg-negative patients were largely comparable except for age and duration of consolidation therapy: HBeAgnegative patients were older and had received consolidation for a longer period than HBeAg-positive patients. Overall, only 2 of 82 patients (2%) were lost to follow-up (at weeks 24 and 60). Patients who did not develop clinical relapse (ie, those who were not retreated) were followed for a median duration of 1.9 years (interquartile range, years). Clinical Relapse Twenty-eight patients developed a clinical relapse, of whom 16 had an elevated ALT level that was >5 times the ULN (Figure 1A). Among patients with an ALT level of >5 times the ULN, the median peak ALT level was 9.5 times the ULN (maximum, 40 times the ULN). The cumulative rates of clinical Table 1. Patient Characteristics According to Hepatitis B Virus (HBV) Envelope Antigen (HBeAg) Status at the Start of Treatment HBeAg Positive Characteristic (n = 60) Baseline (end of treatment) HBeAg Negative (n = 22) Age, y 34 ± ± Male sex 53 (88) 20 (91) 1.00 Transient elastography, kpa 5.1 ( ) 5.9 ( ).19 NA experienced 19 (32) 5 (23).43 (Pegylated) interferon experienced 15 (25) 3 (14).37 NA therapy a First line 25 (42) 7 (32).42 Second line 35 (58) 15 (68) Therapy duration, y 4.0 ( ) 3.9 ( ).41 Consolidation therapy duration, y 2.1 ( ) 2.9 ( ).027 ALT level, ULN 0.5 ( ) 0.7 ( ).15 HBV DNA level, log IU/mL UD UD HBsAg level, log IU/ ml 2.8 ± ± Start of treatment ALT level, ULN 6.3 ( ) 4.9 ( ).83 HBV DNA level, log IU/mL 6.6 ± ± Data are mean ± SD, no. (%) of participants, or median (interquartile range). Abbreviations: ALT, alanine aminotransferase; HBsAg, hepatitis B virus surface antigen; NA, nucleos(t)ide analogue; UD, undetectable (<20 IU/mL); ULN, upper limit of normal. a Entecavir was first-line treatment, and lamivudine, adefovir, and telbivudine were second-line treatment. P Off-Treatment HBV DNA Level and Relapse JID 2017:215 (15 February) 583

4 undetectable HBV DNA and normal ALT levels. One patient was lost to follow-up after the start of retreatment, and 1 patient was lost to follow-up after 24 weeks of retreatment, with an HBV DNA level of 35 IU/mL and a normal ALT level. No patients developed signs of decompensated liver disease. Only age was significantly associated with clinical relapse (Table 2). Patients with end-of-treatment HBsAg levels of 200 IU/mL appeared to have a lower risk of clinical relapse than patients with HBsAg levels of >200 IU/mL (Table 2). The effect of age and end-of-treatment HBsAg levels remained comparable in the multivariable model adjusted for pretreatment HBeAg status and consolidation therapy duration (Table 2). At year 2, 48% of patients aged >35 years experienced clinical relapse, compared with 25% of patients aged 35 years (P =.007); 22% of patients with end-of-treatment HBsAg levels of 200 IU/mL experienced clinical relapse, compared with 44% of patients with higher levels (P =.042; Figure 1B and 1C). Figure 1. Cumulative rate of clinical relapse after nucleos(t)ide analogue discontinuation in 82 patients with chronic hepatitis B, according to start-of-treatment hepatitis B virus (HBV) envelope antigen (HBeAg) status (A), age (B), and end-oftreatment HBV surface antigen (HBsAg) level (C). relapse among patients who were HBeAg positive before treatment were 14%, 26%, and 31% at years 0.5, 1, and 2, respectively. For patients who were HBeAg-negative before treatment, the rates were 14%, 27%, and 53% at years 0.5, 1, and 2, respectively. After initiation of retreatment, 26 of 28 patients achieved Off-Treatment HBV DNA Levels and Prediction of Biochemical Relapse In total, 58 of 82 patients developed an elevated HBV DNA level of >2000 IU/mL at least once. Of these 58 patients, 31 experienced a biochemical relapse. None of the 24 patients with sustained off-treatment HBV DNA levels of 2000 IU/mL developed biochemical relapse during a median follow-up of 2 years (interquartile range, years). At the time the HBV DNA level first increased to >2000 IU/ ml, biochemical relapse was very likely to develop if the HBV DNA level was > IU/mL, compared with values of >2000 to IU/mL (hazard ratio [HR], 8.42; 95% confidence interval [CI], ; P <.001). Biochemical relapse developed in all 7 patients with an HBV DNA level of > IU/ ml, compared with 5 of 11 patients with an HBV DNA level of > to IU/mL and 19 of 40 patients with an HBV DNA level of >2000 to IU/mL (year 2 rates of relapse, 100%, 65%, and 51%, respectively; Figure 2A). Additionally, at the time the HBV DNA level first increased, the direction of the change in the HBsAg level from that at the end of treatment did not significantly predict biochemical relapse (increase vs decrease: HR, 1.31; 95% CI, ; P =.47). To assess the effect of a persistently elevated HBV DNA level on the development of biochemical relapse, the 7 patients with an initial HBV DNA level of > IU/mL were excluded, as were 4 additional patients who had biochemical relapse concurrent with the first elevation in the HBV DNA level. In the remaining 47 patients, a persistently elevated HBV DNA level was observed in 21 patients (45%), whereas a transiently elevated HBV DNA level was observed in 26 patients (55%) (Supplementary Figure 1). Patients with a persistently elevated HBV DNA level were almost 7 times as likely to develop a biochemical relapse as compared to patients with a transient 584 JID 2017:215 (15 February) Cao et al

5 Table 2. Cox Proportional Hazards Regression Analysis of Clinical Relapse Clinical Relapse (n = 28) Univariable Multivariable a Variable HR (95% CI) P HR (95% CI) P End of treatment Age Per year increase 1.08 ( ) y (vs >35 y) 0.33 ( ) ( ).026 Female sex 1.28 ( ).68 First-line NA therapy b 0.95 ( ).89 Consolidation duration Per month increase 1.00 ( ) ( ) 0.54 >3 y (vs 3 y) 0.64 ( ).28 HBsAg level Per log IU/mL increase 1.43 ( ) IU/mL (vs >100 IU/mL) 0.32 ( ) IU/mL (vs >200 IU/mL) 0.35 ( ) ( ) IU/mL (vs >1000 IU/mL) 1.02 ( ).96 ALT level, per fold increase relative to ULN 0.58 ( ).37 Start of treatment HBeAg positivity (vs negativity) 0.53 ( ) ( ).29 HBV DNA level, per log IU/mL increase 0.96 ( ).78 ALT level, per fold increase relative to ULN 0.99 ( ).66 Abbreviations: ALT, alanine aminotransferase; CI, confidence interval; HBeAg, hepatitis B virus envelope antigen; HBsAg, hepatitis B virus surface antigen; HBV, hepatitis B virus; HR, hazard ratio; NA, nucleos(t)ide analogue; ULN, upper limit of normal. a Adjusted for start-of-treatment HBeAg status and factors that were statistically different between HBeAg-positive and HBeAg-negative patients (age and consolidation therapy duration). b Entecavir. Figure 2. Cumulative rate of biochemical relapse after initial elevation in the hepatitis B virus (HBV) DNA level according to the HBV DNA level at the first elevation (A) and the development of persistent HBV DNA elevation (B). Persistent elevation in the HBV DNA level was analyzed as a time-dependent covariate. Off-Treatment HBV DNA Level and Relapse JID 2017:215 (15 February) 585

6 elevation in the HBV DNA level, after adjustment for startof-treatment HBeAg status (adjusted HR, 6.87; 95% CI, ; P <.001). In total, 73% of patients developed biochemical relapse after a persistent elevation within 2 years, compared with 17% of patients after a transient elevation (P <.001; Figure 2B). Serum HBsAg Kinetics Five patients cleared serum HBsAg during off-treatment follow-up, resulting in a cumulative HBsAg seroclearance rate of 10% at year 2 (Figure 3A). The off-treatment serum HBsAg, HBV DNA, and ALT levels of these patients are presented in Figure 3B. Patients with end-of-treatment HBsAg levels of 100 IU/mL were 7-fold as likely to clear serum HBsAg (HR, 7.08; 95% CI, ; P =.033). The change in the HBsAg level was significantly different between persons with a sustained response and those with clinical relapse, with a mean HBsAg change per year of 0.48 vs log IU/mL, respectively (P =.004; Figure 4). After exclusion of the 5 patients with HBsAg seroclearance, the decrease in the HBsAg level among persons with a sustained response was more modest but still significantly different than that for persons with clinical relapse ( 0.16 vs log IU/ ml; P =.016). Longitudinal Liver Stiffness and Ultrasonographic Assessments Because patients who developed a clinical relapse were retreated and not followed further in this study protocol, liver stiffness and ultrasonographic assessments were only available during off-treatment follow-up. At year 1, 56 patients were still in off-treatment follow-up, compared with 32 patients at year 2. Compared with baseline values, the liver stiffness measurements were not significantly different at year 1 (5.7 vs 6.0 kpa; P =.33) or year 2 (5.6 vs 5.8 kpa; P =.57; Supplementary Figure 2). Ultrasonography was performed for 49 of 56 patients (88%) at year 1 and for 23 of 32 patients (72%) at year 2. At years 1 and 2, none of the patients developed ultrasonographic signs of cirrhosis. DISCUSSION In this prospective follow-up study of Asian patients with CHB who stopped NA therapy, we demonstrated that clinical relapse occurred within 2 years, despite adequate long-term viral suppression, in 31% of patients who were positive and 53% who were negative for HBeAg at the start of treatment. Importantly, this is one of the first studies to show that the level and the persistence of an off-treatment elevated HBV DNA level of >2000 IU/mL can be used to predict subsequent biochemical relapse and may therefore guide the management of stopped patients. A high percentage of our patients developed elevated HBV DNA levels of >2000 IU/mL after stopping treatment. Because NAs inhibit HBV DNA production directly at the level of the DNA polymerase, the reactivation of HBV DNA production appears to be logical and almost unavoidable after stopping [5, Figure 3. A, Cumulative rate of off-treatment hepatitis B virus (HBV) surface antigen (HBsAg) seroclearance. B, Off-treatment serum HBsAg, HBV DNA, and alanine aminotransferase (ALT) kinetics of the 5 patients with posttreatment HBsAg seroclearance. ULN, upper limit of normal. 6, 10]. Consequently, when NA discontinuation is considered, an off-treatment increase in the HBV DNA level is a part of the process in most cases. NA discontinuation is therefore not feasible for all patients, particularly when active viral replication is highly unwanted, as in cirrhotic patients [9, 24]. For such patients, HBsAg seroclearance appears to be the best end point for treatment discontinuation [25]. 586 JID 2017:215 (15 February) Cao et al

7 Figure 4. Individual off-treatment serum hepatitis B virus (HBV) surface antigen (HBsAg) profiles after nucleos(t)ide analogue discontinuation in patients with a sustained response (A) and patients with clinical relapse (B). *P =.004 for the comparison between the group with sustained response and the group with clinical relapse. We showed that it is highly recommended to immediately retreat patients who have an off-treatment HBV DNA level of > IU/mL, owing to the high rates of biochemical relapse. Additionally, compared with patients with a transiently elevated HBV DNA level after NA cessation, patients with HBV DNA levels of >2000 IU/mL at consecutive visits had almost a 7-fold higher chance to develop a biochemical relapse. Retreatment is thus highly recommended if persistent elevation has been confirmed, whereas a wait-and-see strategy may be followed in case of a transient elevation. Comparably, Wang et al showed that serum HBV DNA levels 6 months after stopping treatment may be used to predict relapse [26]. Importantly, our study showed that, regardless of which off-treatment month is evaluated, the level and persistence of serum HBV DNA elevation were predictive of subsequent biochemical relapse. Our study provided important data regarding the effect of retreatment criteria on the off-treatment outcome. Owing to the prospective design of our study, all patients were subjected to the same retreatment criteria (an HBV DNA level of >2000 IU/ ml, combined with an ALT level of >2 times the ULN). With these retreatment criteria, different patterns of off-treatment elevated HBV DNA levels were observed with more-favorable outcomes among patients with transiently elevated HBV DNA levels. Therefore, retreatment may not be directly indicated in such patients. Seto et al observed very low rates of sustained off-treatment response and HBsAg loss when retreatment was initiated the first time at which the HBV DNA level increased to >2000 IU/mL [6]. In contrast, when retreatment was started in the case of concurrently elevated HBV DNA and ALT levels, a substantial number of patients developed sustained response and HBsAg loss, even after an initial elevation in the HBV DNA level [10]. This phenomenon of immunological control after a virological relapse has also been observed in an interim analysis of a randomized trial [17]. However, this retreatment strategy is at risk of further increase of HBV DNA levels and flares in ALT levels. Therefore, to prevent flares in the ALT level but also premature retreatment, we propose an algorithm that uses the off-treatment HBV DNA levels to guide the off-treatment management (Supplementary Figure 3). This algorithm warrants further validation. Our results showed that low end-of-treatment HBsAg levels were associated with relapse and HBsAg loss. This is not surprising because serum HBsAg levels are correlated with the number of infected hepatocytes and intrahepatic covalently closed circular DNA [19, 27]. Serum HBsAg level has been associated with the response after NA cessation [8 10, 23, 28] but also during NA or pegylated interferon treatment [29, 30]. Our finding that older age was associated with relapse is in line with results of earlier studies [9, 31, 32]. The exact mechanism is unknown. In addition, in our study pretreatment HBeAg status was not significantly associated with relapse, in line with previous retrospective studies [8, 9, 26]. Individuals with a sustained response demonstrated a greater decrease in HBsAg level than patients who developed a clinical relapse. Despite this, the off-treatment HBsAg kinetics were not predictive of relapse in individual patients, probably because Off-Treatment HBV DNA Level and Relapse JID 2017:215 (15 February) 587

8 of the steady course of serum HBsAg levels. Furthermore, the annual HBsAg level decrease of 0.48 log IU/mL among persons with a sustained response appeared to be high, compared to reported decreases of 0.07 and 0.11 log IU/mL in NA-suppressed HBeAg-positive and HBeAg-negative patients, respectively [12, 13]. Moreover, a randomized trial showed that patients who stopped therapy demonstrated a more prominent decrease in the HBsAg level as compared to patients continuing NA therapy [17]. These results together suggest the presence of an immunomediated beneficial response in some patients after stopping therapy. There were no signs of histological progression, based on liver stiffness and ultrasonographic assessments, among patients continuing off-treatment follow-up. However, no histological data were available after the occurrence of clinical relapse. To our knowledge, this is also one of the first studies to investigate the liver stiffness measurements and ultrasounds after NA discontinuation. Long-term follow-up studies are needed to determine the risk of HCC development after NA cessation. Our study has a few limitations. Long-term follow-up studies are needed to determine the long-term sustainability of response, and therefore the current prospective follow-up protocol is being extended with ongoing recruitment. Nevertheless, protocol adherence was good, with only 2 patients lost to follow-up. In addition, HBV genotype data were not available. However, because all our patients were of Chinese origin, most patients were probably infected with genotype B or genotype C, which were not associated with relapse [24]. Important to note is that the cutoff thresholds of predictors (such as age, HBV DNA level, and HBsAg level) were not prespecified. However, thresholds used in previous studies were considered in the analysis. Finally, the APASL stopping criteria were used for HBeAg-negative patients [15]. These criteria are controversial and were written with consideration of the economic burden of indefinite NA therapy [3, 14, 16]. Despite this, our HBeAgnegative patients had undetectable HBV DNA for a median of 2.9 years before NA discontinuation, and it has been shown that prolonged virological suppression was associated with a lower risk of relapse risk [9]. In conclusion, we showed that relapses were common after NA discontinuation in HBsAg-positive patients with CHB, even when stopping criteria were fulfilled. Importantly, different patterns of off-treatment elevations in the HBV DNA level were observed. The level and persistence of off-treatment elevations in the HBV DNA concentration were useful in the prediction of a subsequent biochemical relapse and may thus be used to guide off-treatment management of patients, allowing early retreatment of those at high risk of biochemical relapse. Supplementary Data Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. Notes Acknowledgments. We thank all the patients in this study and the nurses who assisted in the patient management and the collection of serum samples. J. C. was responsible for study concept and design, patient recruitment and follow-up, data collection, analysis and interpretation of data, drafting of the article, and finalizing the article. H. C. was responsible for study concept and design, analysis and interpretation of data, drafting of the article, and finalizing the article. T. Y. and Z. L. were responsible for patient recruitment and follow-up, data collection, and critical revision of the article. B. E. H. was responsible for analysis and interpretation of data, drafting of the article, and finalizing the article. X. M. was responsible for patient recruitment and follow-up, data collection, and critical revision of the article. C. Z. was responsible for patient recruitment and follow-up, data collection, and critical revision of the article. J. S. and J. H. were responsible for patient recruitment, data collection, and critical revision of the article. H. L. A. J. was responsible for study concept and design, analysis and interpretation of data, drafting of the article, and finalizing the article. J. P. was responsible for study concept and design, drafting of the article, finalizing the article, and critical review of the final article. Disclaimer. The funding sources did not have any influence on the study design, data collection, analysis and interpretation of the data, writing of the report, or the decision to submit for publication. Financial support. This work was supported by the National Science Foundation of China (grant ) and the Rotterdam Foundation for Liver Research. Potential conflicts of interest. B. E. H. has received grants from and is a consultant for Intercept Pharmaceuticals. H. L. A. J. has received grants from and is a consultant for Abbott, Anadys, Bristol-Myers Squibb, Gilead Sciences, Innogenetics, Merck, Novartis, Roche, Santaris, Tibotec, and Janssen. J. H. has received consulting fees from Roche, Novartis, GSK, and Bristol-Myers Squibb and has received grant/research support from Roche, Novartis, and GSK. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed. References 1. Janssen HL, Sonneveld MJ, Brunetto MR. Quantification of serum hepatitis B surface antigen: is it useful for the management of chronic hepatitis B? Gut 2012; 61: Seeger C, Mason WS. Molecular biology of hepatitis B virus infection. Virology 2015; : Reijnders JG, Janssen HL. Relapse of chronic hepatitis B after discontinuation of nucleos(t)ide analogs: is the glass half full or half empty? Hepatology 2013; 58: Chang ML, Liaw YF, Hadziyannis SJ. Systematic review: cessation of long-term nucleos(t)ide analogue therapy in patients with hepatitis B e antigen-negative chronic hepatitis B. Aliment Pharmacol Ther 2015; 42: Chaung KT, Ha NB, Trinh HN, et al. High frequency of recurrent viremia after hepatitis B e antigen seroconversion and consolidation therapy. J Clin Gastroenterol 2012; 46: Seto WK, Hui AJ, Wong VW, et al. Treatment cessation of entecavir in Asian patients with hepatitis B e antigen negative chronic hepatitis B: a multicentre prospective study. Gut 2015; 64: Buti M, Casillas R, Riveiro-Barciela M, et al. Tenofovir discontinuation after longterm viral suppression in HBeAg negative chronic hepatitis B. Can HBsAg levels be useful? J Clin Virol 2015; 68: Chen CH, Lu SN, Hung CH, et al. The role of hepatitis B surface antigen quantification in predicting HBsAg loss and HBV relapse after discontinuation of lamivudine treatment. J Hepatol 2014; 61: Chi H, Hansen BE, Yim C, et al. Reduced risk of relapse after long-term nucleos(t) ide analogue consolidation therapy for chronic hepatitis B. Aliment Pharmacol Ther 2015; 41: Hadziyannis SJ, Sevastianos V, Rapti I, Vassilopoulos D, Hadziyannis E. Sustained responses and loss of HBsAg in HBeAg-negative patients with chronic hepatitis B 588 JID 2017:215 (15 February) Cao et al

9 who stop long-term treatment with adefovir. Gastroenterology 2012; 143: e Chotiyaputta W, Peterson C, Ditah FA, Goodwin D, Lok AS. Persistence and adherence to nucleos(t)ide analogue treatment for chronic hepatitis B. J Hepatol 2011; 54: Chevaliez S, Hézode C, Bahrami S, Grare M, Pawlotsky JM. Long-term hepatitis B surface antigen (HBsAg) kinetics during nucleoside/nucleotide analogue therapy: finite treatment duration unlikely. J Hepatol 2013; 58: Zoutendijk R, Hansen BE, van Vuuren AJ, Boucher CA, Janssen HL. Serum HBsAg decline during long-term potent nucleos(t)ide analogue therapy for chronic hepatitis B and prediction of HBsAg loss. J Infect Dis 2011; 204: European Association for the Study of the Liver. EASL clinical practice guidelines: Management of chronic hepatitis B virus infection. J Hepatol 2012; 57: Liaw YF, Kao JH, Piratvisuth T, et al. Asian-Pacific consensus statement on the management of chronic hepatitis B: a 2012 update. Hepatol Int 2012; 6: Terrault NA, Bzowej NH, Chang KM, Hwang JP, Jonas MM, Murad MH; American Association for the Study of Liver Diseases. AASLD guidelines for treatment of chronic hepatitis B. Hepatology 2016; 63: Berg T, Simon K, Mauss S, et al. Stopping tenofovir disoproxil fumarate (TDF) treatment after long term virologic suppression in HBeAg-negative CHB: week 48 interim results from an ongoing randomized, controlled trial ( finite CHB ). J Hepatol 2015; 62:S Levrero M, Pollicino T, Petersen J, Belloni L, Raimondo G, Dandri M. Control of cccdna function in hepatitis B virus infection. J Hepatol 2009; 51: Thompson AJ, Nguyen T, Iser D, et al. Serum hepatitis B surface antigen and hepatitis B e antigen titers: disease phase influences correlation with viral load and intrahepatic hepatitis B virus markers. Hepatology 2010; 51: Peng J, Cao J, Yu T, et al. Predictors of sustained virologic response after discontinuation of nucleos(t)ide analog treatment for chronic hepatitis B. Saudi J Gastroenterol 2015; 21: Fung J, Lai CL, Tanaka Y, et al. The duration of lamivudine therapy for chronic hepatitis B: cessation vs. continuation of treatment after HBeAg seroconversion. Am J Gastroenterol 2009; 104:1940 6; quiz Arends P, Sonneveld MJ, Zoutendijk R, et al.; VIRGIL Surveillance Study Group. Entecavir treatment does not eliminate the risk of hepatocellular carcinoma in chronic hepatitis B: limited role for risk scores in Caucasians. Gut 2015; 64: Liang Y, Jiang J, Su M, et al. Predictors of relapse in chronic hepatitis B after discontinuation of anti-viral therapy. Aliment Pharmacol Ther 2011; 34: Jeng WJ, Sheen IS, Chen YC, et al. Off-therapy durability of response to entecavir therapy in hepatitis B e antigen-negative chronic hepatitis B patients. Hepatology 2013; 58: Kim GA, Lim YS, An J, et al. HBsAg seroclearance after nucleoside analogue therapy in patients with chronic hepatitis B: clinical outcomes and durability. Gut 2014; 63: Wang CC, Tseng KC, Hsieh TY, Tseng TC, Lin HH, Kao JH. Assessing the durability of entecavir-treated hepatitis B using quantitative HBsAg. Am J Gastroenterol 2016; 111: Manesis EK, Papatheodoridis GV, Tiniakos DG, et al. Hepatitis B surface antigen: relation to hepatitis B replication parameters in HBeAg-negative chronic hepatitis B. J Hepatol 2011; 55: Chan HL, Wong GL, Chim AM, Chan HY, Chu SH, Wong VW. Prediction of off-treatment response to lamivudine by serum hepatitis B surface antigen quantification in hepatitis B e antigen-negative patients. Antivir Ther 2011; 16: Sonneveld MJ, Hansen BE, Piratvisuth T, et al. Response-guided peginterferon therapy in hepatitis B e antigen-positive chronic hepatitis B using serum hepatitis B surface antigen levels. Hepatology 2013; 58: Marcellin P, Buti M, Krastev Z, et al. Kinetics of hepatitis B surface antigen loss in patients with HBeAg-positive chronic hepatitis B treated with tenofovir disoproxil fumarate. J Hepatol 2014; 61: Lee HW, Lee HJ, Hwang JS, et al. Lamivudine maintenance beyond one year after HBeAg seroconversion is a major factor for sustained virologic response in HBeAg-positive chronic hepatitis B. Hepatology 2010; 51: Jung KS, Park JY, Chon YE, et al. Clinical outcomes and predictors for relapse after cessation of oral antiviral treatment in chronic hepatitis B patients. J Gastroenterol 2016; 51: Off-Treatment HBV DNA Level and Relapse JID 2017:215 (15 February) 589