The role of peginterferon in nucleos(t)ide‐analogue‐treated chronic hepatitis B patients: A review of published literature

Chronic hepatitis B infection (CHB) causes up to 1.0 million deaths annually. Currently, more than 90% of CHB patients worldwide are receiving indefinite nucleos(t)ide analogue (NA) therapy. New strategies for optimizing hepatitis B surface antigen (HBsAg) loss are required for NA‐treated patients as the majority are unable to achieve HBsAg loss and may require lifelong therapy. In hepatitis B e antigen (HBeAg)‐positive patients, switching from NAs to finite peginterferon (PegIFN) therapy can double HBeAg seroconversion rates. One in five patients who switch to PegIFN can achieve HBsAg loss, whereas patients who continue NA therapy typically do not. In HBeAg‐negative NA‐treated patients, add‐on PegIFN therapy achieves higher, albeit modest, HBsAg loss rates compared with continued NA monotherapy and offers the opportunity for NA‐treated patients to achieve the inactive carrier state. In the absence of curative therapies, PegIFN represents a valuable, finite option for NA‐treated patients who would otherwise require potentially lifelong therapy.     

Histological response to combination therapy with nucleos(t)ide analogs and peginterferon alpha in treatment‐naïve chronic hepatitis B patients

Although nucleos(t)ide analog (NA) monotherapy is effective in hepatitis B virus suppression and fibrosis regression, serological response rates are not satisfactory. Studies assessing the benefits of combination therapy with NAs and peginterferon alpha (PegIFNα) in patients with chronic hepatitis B (CHB) have produced conflicting results and mainly focused on serological outcomes. Histological changes in response to combination therapy have not been evaluated in real‐world practice. This study aimed to evaluate the histological changes in response to NA‐PegIFNα combination therapy in CHB patients and to comprehensively compare the efficacy of NA‐PegIFNα combination therapy and NA monotherapy. We conducted a retrospective analysis of data from 40 CHB patients who underwent either NA‐PegIFNα combination therapy or NA monotherapy. Changes in histology at 48 weeks after treatment initiation were evaluated. Serological characteristics were also analysed and compared between the NA‐PegIFNα combination therapy and NA monotherapy groups and between histological responders and nonresponders. Compared to baseline biopsies, both fibrosis staging and necroinflammatory grading scores were significantly lower in the second biopsies examined post‐treatment in both groups. Nearly all patients experienced a reduction in inflammation (87.5% in both groups), but there was a subgroup of patients who exhibited either no significant improvement or fibrosis progression (33.3% and 31.2% in the NA monotherapy and NA‐PegIFNα combination therapy groups, respectively). Nearly, all patients achieved ALT normalization and sustained virological response (SVR) after 48 weeks of antiviral treatment. Approximately one‐third of individuals (36.8% and 30% in the two groups, respectively) achieved HBeAg loss at 48 weeks after treatment initiation. Although there were no significant differences in overall rates of histological, biochemical, virological and serological responses between the two groups, an earlier virological response and a higher cumulative SVR rate over time were observed during long‐term follow‐up in patients treated with NA‐PegIFNα combination therapy (P = 0.0129). Trends of more rapid HBeAg loss and a higher cumulative HBeAg loss rate throughout long‐term follow‐up were also observed but were not statistically significant. The ALT normalization rates at 24 and 48 weeks after treatment initiation were associated with the histological response. Significant regression of fibrosis and resolution of necroinflammation were induced with either NA‐PegIFNα combination therapy or NA monotherapy. Significant biochemical, virological and serological responses were observed in both groups, and the response rates at 48 weeks were similar in the two groups. Over time during long‐term follow‐up, the virological and serological responses were faster and superior following the combination regimen.     

Pronounced decline of serum HBsAg in chronic hepatitis B patients with long-term effective nucleos(t)ide analogs therapy

Aims: This study aims to investigate the kinetics of serum HBsAg levels in chronic hepatitis B patients with long-term nucleos(t)ide analogs (NAs) therapy.

Methods: This was a retrospective clinical study. Serum HBsAg in serial samples of 94 patients, who received at least 8 years of NAs therapy, were measured using Elecsys^® HBsAg II Quant Assay.

Results: In this cohort, serum HBsAg levels reduced from 3.80 log10?IU/mL at baseline to 2.72 log10?IU/mL at year 8 (p?p?p?=?.001). As compared to patients with slow (0–1 log10?IU/mL) or steady HBsAg(≤0 log10?IU/mL) decline at year 1, patients with a rapid HBsAg (≥1 log10?IU/mL) decline had a significantly lower HBsAg levels from year 2 to 8. However, Cox regression analysis showed that only absolute HBsAg levels at year 1 was an independent predictor of subsequent HBsAg <1000?IU/mL at year 8 of antiviral therapy(HR 0.242, p?=?.004).

Conclusion: Pronounced HBsAg declines could be achieved in patients after long-term effective therapy with NAs, and on-treatment low serum HBsAg level at year 1 might be a predictor of serum HBsAg <1000?IU/mL at year 8.     

Kinetics of serum O-glycosylated M-hepatitis B surface antigen with hepatocellular carcinoma history and nucleos(t)ide analogue therapy in hepatitis B patients

A newly developed O-glycosylated M-hepatitis B surface antigen (HBsAgGi) measurement system can detect hepatitis B surface antigen (HBsAg) associated with infectious particles. We investigated the association of HBsAgGi levels with clinical parameters and a history of hepatocellular carcinoma (HCC) development in a cross-sectional cohort analysis (Study 1) as well as the quantitative changes in HBsAgGi during nucleos(t)ide analogue (NA) therapy in a longitudinal cohort analysis (Study 2). A total of 124 patients with genotype C chronic HBV infection were analysed in Study 1 to evaluate correlations of HBsAgGi with conventional HBV markers and HCC history. Among those, 36 patients receiving NA therapy were enrolled in Study 2 for quantitative comparisons between pre-treatment baseline and 48 weeks of NA therapy. In Study 1, serum HBsAgGi was significantly associated with HBsAg (r = .5857, p < .00001) and weakly but significantly correlated with HBV DNA (r = .2936, p = .001). Although HBsAgGi (p = .111) was comparable between HCC history (+) group and HCC history (−) group, the HBsAgGi/HBsAg ratio (p = .011) was significantly higher in HCC history (+) patients. In Study 2, HBsAgGi was significantly decreased after 48 weeks of NA therapy (p < .001). HBsAg findings were similar (p = .005) along with an HBV DNA reduction (p < .001). In the baseline hepatitis B e antigen (HBeAg) (+) subgroup, HBsAgGi decreased significantly between baseline and 48 weeks of NA (p = .005), while HBsAg was comparable (p = .051). Low HBsAg and high HBsAgGi were associated with a history of HCC development. HBsAgGi decreased significantly by 48-week NA therapy.     

Long‐term entecavir therapy results in falls in serum hepatitis B surface antigen levels and seroclearance in nucleos(t)ide‐naïve chronic hepatitis B patients

Entecavir (ETV) is reported to result in suppression of hepatitis B virus DNA (HBV DNA) replication with minimal drug resistance. However, information on the long‐term effect of such therapy on serum hepatitis B surface antigen (HBsAg) level and elimination of HBsAg is not available. ETV therapy was started in 553 nucleos(t)ide‐naïve patients with chronic hepatitis B infection (HBeAg positive: 45%) in our hospital. Serum HBsAg levels were measured serially by the Architect assay. The median baseline HBsAg was 2180 IU/mL (0.12–243 000 IU/mL), and median follow‐up period was 3.0 years, with 529, 475, 355, 247 and 163 patients followed‐up for 1, 2, 3, 4 and 5 years, respectively. At year 5, the mean log HBsAg decline from baseline was −0.48 log IU/mL, and the cumulative HBsAg clearance rate was 3.5%. Multivariate analysis identified HBV DNA level at baseline (<3.0 log copies IU/mL, odd ratio = 10.2; 95% confidence interval = 1.87–55.5, P = 0.007) and HBsAg level (<500 IU/mL, odd ratio = 29.4; 95% confidence interval = 2.80–333, P = 0.005) as independent predictors of HBsAg seroclearance. These results indicate that although serum HBsAg level declines gradually during ETV therapy, HBsAg seroclearance remains a rare event.     

Association of on‐treatment serum hepatitis B surface antigen level with sustained virological response to nucleos(t)ide analog in patients with hepatitis B e‐antigen positive chronic hepatitis B

Aim: This study evaluated the on‐treatment serum hepatitis B surface antigen (HBsAg) level during nucleos(t)ide analog (NUC) therapy and the correlation with off‐treatment sustained virological response (SVR). Methods: Fifty‐one consecutive patients with hepatitis B e‐antigen (HBeAg) positive chronic hepatitis B who achieved HBeAg loss/seroconversion after NUC therapy and completed 12 months or more of additional therapy were included. Serum HBsAg and hepatitis B virus (HBV) DNA levels were determined at baseline, 3, 6, 9 and 12 months, and at the end of treatment. SVR was defined as HBV DNA levels of less than 10 000 copies/mL until 6 or 12 months off‐treatment without reappearance of HBeAg. Results: Twenty‐two (43.1%) and 13 (25.5%) patients maintained SVR at 6 and 12 months off‐treatment, respectively. In univariate analyses, a decline of HBsAg of 0.5 log₁₀ IU/mL or less at 6 months (P = 0.006) and 12 months (P = 0.013), the mean change in HBsAg level at 6 months (P = 0.024), and lamivudine or entecavir treatment (P = 0.019) were significant predictive factors for SVR at 6 months off‐treatment. A decline of HBsAg of 0.5 log₁₀ IU/mL or less at 6 months and lamivudine or entecavir treatment were independent factors on multivariate analyses (odds ratio [OR], 16.67; 95% confidence interval [CI], 1.86–142.86 [P = 0.012]; and OR, 14.83; 95% CI, 1.18–185.73 [P = 0.036]; respectively). Conclusion: On‐treatment serum HBsAg level predicted early off‐treatment SVR to NUC therapy in patients infected with genotype C.     

The suppressive effect of nucleos(t)ide analogue treatment on the incidence of hepatocellular carcinoma in chronic hepatitis B patients

The development of nucleos(t)ide analogues (NA) has influenced hepatitis B virus management. However, the annual incidence rate during NA treatment has been reported to be 0.3–1.2% in non‐cirrhosis cases and 1.8–6.0% in cirrhosis cases, indicating that the suppressive effect of NA treatment on hepatocellular carcinoma (HCC) would be insufficient. Past studies, including one randomized control trial that compared lamivudine treatment with placebo, have revealed that NA treatment could suppress the incidence of HCC in patients with advanced fibrosis. However, it remains unknown whether NA treatment can suppress the incidence of HCC in chronic hepatitis patients without advanced fibrosis. The HCC incidence in patients treated with entecavir was similar to that of those treated with lamivudine, although entecavir exhibits a stronger viral suppression than lamivudine. The following risk factors related to the incidence of HCC during NA treatment have been identified: older age, male gender, pre‐existing cirrhosis, a family clustering of hepatitis B virus, lower platelet counts, and higher hepatitis B core‐related antigens as baseline factors and higher alpha fetoprotein levels as an on‐treatment factor. Conversely, the loss of the hepatitis B surface antigen (HBsAg) by interferon or NA was correlated with a lower HCC incidence rate. Because interferon treatment has much more effects on reducing HBsAg levels compared with NA treatment, a combination treatment with NA and pegylated interferon can bring additional reduction of HBsAg levels compared with NA monotherapy. Further study is needed to clarify this.     

核苷（酸）类似物抗病毒治疗慢性乙型肝炎的优化策略

目前我国有4种核苷（酸）类似物可用于慢性乙型肝炎的抗病毒治疗,即拉米夫定、阿德福韦酯、恩替卡韦和替比夫定。但由于患者的性别、年龄、遗传背景、感染乙型肝炎病毒（HBV）的途径、病毒基因型、病程长短、肝脏病变程度和对治疗药物敏感性等不同,即使有同样治疗适应证的患者按同样的规范方案治疗后,仍有     

Optimal management of chronic hepatitis B patients receiving nucleos(t)ide analogues

Management of chronic hepatitis B (CHB) is still considered a challenge in clinical practice. Patients must be carefully evaluated before starting therapy. This includes virology and laboratory assessments, an estimation of fibrosis by invasive and/or noninvasive methods, and an estimation of the risk of hepatocellular carcinoma (HCC). Nucleos(t)ide analogues (NAs) with a high barrier to resistance (tenofovir disoproxil fumarate [TDF], entecavir [ETV] and tenofovir alafenamide [TAF]) are the most frequently used treatments because of their good long‐term efficacy and tolerability. None of these options has been shown to be more effective than the other, but certain factors should be considered when selecting the best therapy for specific populations. Most patients achieve a virological and biochemical response to these agents, with a low rate of emerging resistance during long‐term treatment. However, the rate of hepatitis B surface antigen (HBsAg) loss is low and in most cases NAs therapy is lifelong. Safety concerns for long‐term NA use have become a priority in the management of CHB, in particular, the risk of impaired kidney function and bone marrow density loss described with TDF regimens. The risk of HCC is not completely eliminated by NAs. Thus, patients at higher risk should be identified and provided with appropriate surveillance.     

Treatment of chronic hepatitis B with nucleos(t)ide analogues

Recently antiviral therapies for chronic hepatitis B using nucleos(t)ide analogues have become standard treatment modalities on the basis of several independent guidelines, starting with those of the American Association for the Study of Liver Diseases (AASLD) and other such organizations and bodies, including the European Association for the Study of the Liver (EASL), the Asian Pacific Association for the Study of the Liver (APASL), and the Japanese Ministry of Health, Labour and Welfare (MHLW)'s research team. The philosophies underlying such treatment strategies are considered basically equivalent. MHLW's guidelines define subjects for medical intervention to be cases measuring alanine aminotransferase (ALT) ≥31 IU/L, with serological hepatitis B virus (HBV) DNA level ≥5 log copies/mL for hepatitis B e antigen (HBeAg)‐positive cases, and serological HBV DNA level ≥4 log copies/mL for HBeAg‐negative cases. These Japanese guidelines advocate entecavir as the first‐line treatment option for nucleos(t)ide‐naïve patients, and combination treatment of lamivudine and adefovir as the basis of treatment for patients with lamivudine‐ and/or entecavir‐resistant viruses. Of particular note for patients undergoing lamivudine treatment with persistent HBV DNA level < 2.1 log copies/mL is the recommendation of a switch to entecavir. Early detection of drug‐resistant virus is desirable after initiation of nucleos(t)ide analogue treatment, but such a procedure is not uniformly available at all medical institutions. Nevertheless, timely estimation of potential early‐stage drug‐resistant virus development is crucial for getting a head start on treatment. HBV core‐related antigen (HBcrAg) level or HBV DNA level are considered useful markers for the appearance of such drug‐resistant viruses.     

HBsAg and HBcrAg as predictors of HBeAg seroconversion in HBeAg‐positive patients treated with nucleos(t)ide analogues

HBeAg seroconversion marks an important spontaneous change and treatment end‐point for HBeAg‐positive patients and is a pre‐requisite for HBsAg loss or functional cure. In this retrospective analysis, we aimed to identify predictors of seroconversion using serum quantitative HBsAg and HBcrAg, in HBeAg‐positive patients treated with nucleos(t)ide analogues (NA). Data and samples from 118 HBeAg‐positive adults (genotypes A‐G) started on NA between Jan 2005 and Sept 2016 were retrospectively analysed at several time‐points. The predictive power of on‐treatment levels of HBsAg and HBcrAg was determined using receiver operating curve (ROC) analysis and cut‐off values determined by maximized Youden's index. About 36.4% of patients achieved HBeAg seroconversion after a median of 39 months’ treatment. On treatment kinetics of HBV DNA, HBsAg and HBcrAg differed between HBeAg seroconverters and nonseroconverters. A combination of HBsAg and HBcrAg had the greatest predictive value for HBeAg seroconversion: at 6 months, HBsAg of 3.9 log₁₀ IU/mL and HBcrAg of 5.7 log₁₀ U/mL had a sensitivity of 71.4%, specificity of 79.5%, positive predictive value (PPV) of 65.2% and negative predictive value (NPV) of 83.8%, with AUROC of 0.769 (0.668, 0.869; 95%CI), and at 12 months, HBsAg 3.8 log₁₀ IU/mL and HBcrAg 5.5 log₁₀ U/mL had a sensitivity of 73.7%, specificity of 79.5%, PPV of 63.6% and NPV of 86.1%, with AUROC 0.807 (0.713, 0.901; 95% CI). In conclusion, our results may be used to identify patients who are unlikely to achieve treatment end‐points, which will be important as the future management of chronic hepatitis B looks to therapies that offer functional cure.     

Exploration of nucleos(t)ide analogs cessation in chronic hepatitis B patients with hepatitis B e antigen loss

BACKGROUNDNucleos(t)ide analogs (NAs) cessation in chronic hepatitis B (CHB) patients remains a matter of debate in clinical practice. Current guidelines recommend that patients with hepatitis B e antigen (HBeAg) seroconversion discontinue NAs after relatively long-term consolidation therapy. However, many patients fail to achieve HBeAg seroconversion after the long-term loss of HBeAg, even if hepatitis B surface antigen (HBsAg) loss occurs. It remains unclear whether NAs can be discontinued in this subset of patients.AIMTo investigate the outcomes and factors associated with HBeAg-positive CHB patients with HBeAg loss (without hepatitis B e antibody) after cessation of NAs.METHODSWe studied patients who discontinued NAs after achieving HBeAg loss. The Cox proportional hazards model was used to identify predictors for virological relapse after cessation of NAs. The cut-off value of the consolidation period was confirmed using receiver operating characteristic curves; we confirmed the cut-off value of HBsAg according to a previous study. The log-rank test was used to compare cumulative relapse rates among groups. We also studied patients with CHB who achieved HBeAg seroconversion and compared their cumulative relapse rates. Propensity score matching analysis (PSM) was used to balance baseline characteristics between the groups.RESULTSWe included 83 patients with HBeAg loss. The mean age of these patients was 32.1 ± 9.5 years, and the majority was male (67.5%). Thirty-eight patients relapsed, and the cumulative relapse rate at months 3, 6, 12, 24, 36, 60, 120, and 180 were 22.9%, 36.1%, 41.0%, 43.5%, 45.0%, 45.0%, 45.0%, and 52.8%, respectively. Twenty-six (68.4%) patients relapsed in the first 3 mo after NAs cessation, and 35 patients (92.1%) relapsed in the first year after NAs cessation. Consolidation period (≥ 24 mo vs < 24 mo) (HR 0.506, P = 0.043) and HBsAg at cessation (≥ 100 IU/mL vs < 100 IU/mL) (HR 14.869, P = 0.008) were significant predictors in multivariate Cox regression. In the PSM cohort, which included 144 patients, there were lower cumulative relapse rates in patients with HBeAg seroconversion (P = 0.036).CONCLUSIONHBeAg-positive CHB patients with HBeAg loss may be able to discontinue NAs therapy after long-term consolidation, especially in patients with HBsAg at cessation < 100 IU/mL. Careful monitoring, especially in the early stages after cessation, may ensure a favorable outcome.     

Prevalence and risk factors for viral blipping in chronic hepatitis B patients treated with nucleos (t) ide analogues

The clinical relevance of viral blipping during nucleos (t) ide analogue (NA) treatment is unclear in chronic hepatitis B (CHB). We investigated the prevalence, risk factors and clinical outcomes for those with viral blipping during NA treatment. A retrospective cohort study investigated consecutively treated CHB patients from May 2008 to February 2015 on the NAs such as entecavir (ETV), tenofovir (TDF) and lamivudine (LAM). Included patients were previously treatment naive. Viral blipping was defined as serum HBV DNA >20 IU/mL on one occasion, and not >200 IU/mL, with subsequent measurement returning to undetectable levels, that is <20 IU/mL. A total of 242 treatment‐compliant CHB patients were included with 44 (18.2%) experiencing viral blipping. In multivariable Cox regression, Asian race (HR=7.40, 95% CI 1.01–54.29, P<.049), LAM therapy (vs ETV/TDF, HR=2.53, 95% CI 1.29–4.95, P<.007), higher creatinine (per SD, HR=1.47, 95% CI 1.21–1.79, P<.001), HBeAg positivity (HR=2.68, 95% CI 1.39–5.03, P<.003) and longer time to achieve undetectable HBV DNA (per month, HR=1.05, 95% CI 1.02–1.08, P=.001) were associated with an increased risk of viral blipping. Viral blipping did not show any significant association with viral breakthrough, HBsAg loss, ALT flares or disease progression. Viral blipping is a frequent event during NA therapy; however, it did not lead to any clinically significant outcomes. Thus, it may not require more frequent blood work and patient visits in clinical practice.     

Effect of nucleos(t)ide analogues on blood lipid profiles in patients with chronic hepatitis B: A cross-sectional survey

This study aimed to explore the effects of the 3 nucleos(t)ide analogues (NAs) on lipid levels. We retrospectively included patients treated with NAs at 2 centers and collected their clinical data at their visiting points. Differences in blood lipid levels were analyzed by statistical methods, and factors related to hyperlipidemia were discussed. In these 2 centers, the prevalence rates of hypercholesterolemia were 12/181 (6.6%) for tenofovir alafenamide fumarate (TAF)-, 0/158 (0%) for tenofovir disoproxil fumarate (TDF)-, and 13/182 (7.1%) for entecavir (ETV)-treated individuals (P = .003). The prevalence rates of hypertriglyceridemia were 30/181 (16.6%) for TAF-, 11/158 (7.0%) for TDF-, and 26/182 (14.3%) for ETV-treated individuals (P = .025). In TAF (n = 181, 10 [6, 15] months), TDF (n = 158, 18 [7.5, 45] months), and ETV (n = 182, 24 [10, 60] months) groups, total cholesterol (TC) levels were 4.63 ± 0.91 mmol/L, 3.86 ± 0.61 mmol/L, and 4.53 ± 0.87 mmol/L, respectively; triglyceride (TG) levels were 1.27 ± 0.76 mmol/L, 0.87 ± 0.51 mmol/L, and 1.14 ± 0.67 mmol/L, respectively (P < .001). In multivariate regression analysis, factors associated with hypercholesterolemia were age (adjusted hazard risk [HR] = 1.055 [1.018–1.094]; P = .003) and body mass index (BMI) (adjusted HR = 0.817 [0.669–0.998]; P = .048). Factors associated with hypertriglyceridemia were TAF group (vs. TDF group) (adjusted HR = 0.405 [0.167–0.980]; P = .045), age (adjusted HR = 1.028 [1.002–1.055]; P = .038), and sex (adjusted HR = 0.190 [0.079–0.456]; P < .001). Among the patients treated with TAF (10 [6, 15] months), TDF (18 [7.5, 45] months), and ETV (24 [10, 60] months), the blood lipid levels in the TDF group were lower than those in the TAF group and ETV group, and the occurrence of hyperlipidemia was associated with age, sex, BMI, and different treatment.     

Immunological biomarkers as indicators for outcome after discontinuation of nucleos(t)ide analogue therapy in patients with HBeAg‐negative chronic hepatitis B

The optimal duration of treatment with nucleos(t)ide analogues (NAs) for patients with HBeAg‐negative chronic hepatitis B (CHB) is unknown. The aim of this study was to identify an immune signature associated with off‐treatment remission to NA therapy. We performed microarray analysis of peripheral blood mononuclear cell (PBMCs) from six patients with chronic hepatitis B who stopped NA therapy (three with off‐treatment remission, three with relapse) and five patients with chronic HBV infection (previously termed ‘inactive carriers’) served as controls. Results were validated using qRT‐PCR on a second group of 21 individuals (17 patients who stopped treatment and four controls). PBMCs from 38 patients on long‐term NA treatment were analysed for potential to stop treatment. Microarray analysis indicated that patients with off‐treatment remission segregated as a distinct out‐group. Twenty‐one genes were selected for subsequent validation. Ten of these were expressed at significantly lower levels in the patients with off‐treatment remission compared to the patients with relapse and predicted remission with AUC of 0.78‐0.92. IFNγ, IL‐8, FASLG and CCL4 were the most significant by logistic regression. Twelve (31.6%) of 38 patients on long‐term NA therapy had expression levels of all these four genes below cut‐off values and hence were candidates for stopping treatment. Our data suggest that patients with HBeAg‐negative CHB who remain in off‐treatment remission 3 years after NA cessation have a distinct immune signature and that PBMC RNA levels of IFNγ, IL‐8, FASLG and CCL4 may serve as potential biomarkers for stopping NA therapy.     

Combination of hepatitis B viral antigens and DNA for prediction of relapse after discontinuation of nucleos(t)ide analogs in patients with chronic hepatitis B

Aim: The factors associated with hepatitis recurrence after discontinuation of nucleos(t)ide analogs (NAs) in patients with chronic hepatitis B were analyzed to predict the risk of relapse more accurately. Methods: A total of 126 patients who discontinued NA therapy were recruited retrospectively. The clinical conditions of a successful discontinuation were set as alanine aminotransferase (ALT) below 30 IU/L and serum hepatitis B virus (HBV) DNA below 4.0 log copies/mL. Results: Relapse of hepatitis B were judged to occur when maximal serum ALT became higher than 79 IU/L or when maximal serum HBV DNA surpassed 5.7 log copies/mL following NA discontinuation since these values corresponded with mean values of ALT (30 IU/L) and HBV DNA (4.0 log copies/mL), respectively. At least 90% of patients with either detectable hepatitis B e antigen or serum HBV DNA higher than 3.0 log copies/mL at the time of NA discontinuation relapsed within one year. In the remaining patients, higher levels of both hepatitis B surface and core‐related antigens at the time of discontinuation, as well as a shorter course of NA treatment, were significantly associated with relapse by multivariate analysis. Conclusions: It appears that negative results for hepatitis B e antigen and serum HBV DNA lower than 3.0 log copies/mL are essential for successful NA discontinuation, which may be attained by a longer treatment period. Levels of hepatitis B surface and core‐related antigens are also significant factors independently associated with relapse of hepatitis.