1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

The profile and clinical significance of serum hepatitis B surface antigen (HBsAg) levels during long-term nucleoside analogue (NA) therapy in chronic hepatitis B (CHB) is undetermined. From 1994 to 2002, 322 Chinese CHB patients were started on lamivudine in our center. Patients were recruited if they were continuously treated with lamivudine for at least 10 years and maintained favorable virologic responses throughout therapy (HBV DNA <2,000 IU/mL). HBsAg and HBV DNA levels were measured serially, and the predictability of HBsAg kinetics in determining NA-related HBsAg seroclearance was determined. Seventy patients were recruited, of which 43 (61.4%) were hepatitis B e antigen (HBeAg)-positive. Fifty-two (74.3%) patients had undetectable viremia (HBV DNA <20 IU/mL) during therapy. Fifteen (21.4%) patients were followed up for 15 years. The median rate of HBsAg reduction was 0.104 log IU/mL/year, with no significant difference found when comparing patients who were HBeAg-positive versus HBeAg-negative, were genotype B versus C, and had detectable versus undetectable viremia during therapy (all P > 0.05). Seven (10%) patients achieved HBsAg seroclearance, and when compared with the remaining 63 patients, had significantly lower median baseline HBsAg levels (P = 0.012) and a greater median rate of HBsAg reduction (P < 0.001). Baseline HBsAg levels and the rate of HBsAg reduction achieved an area under the receiver operating characteristic curve of 0.860 (P = 0.004; 95% confidence interval [CI], 0.742-0.978) and 0.794 (P = 0.018; 95% CI, 0.608-0.979), respectively. Baseline HBsAg <1,000 IU/mL and on-treatment reduction of HBsAg >0.166 log IU/mL/year were optimal cutoff levels in predicting subsequent HBsAg seroclearance (negative predictive values, 98.1% and 97.8%, respectively). Conclusion: Low baseline HBsAg levels and greater rate of HBsAg reduction achieved high predictive values for predicting HBsAg seroclearance, strengthening the prognostic role of HBsAg measurements during NA therapy. (Hepatology 2013;53:923–931)


area under the curve


chronic hepatitis B


confidence interval


hepatitis B e antigen


hepatitis B surface antigen


human leukocyte antigen


nucleoside analogue


receiver operating characteristic.

The introduction of nucleoside analogue (NA) therapy has revolutionized the management of patients with chronic hepatitis B (CHB). Since the introduction of lamivudine in 1998[1] and subsequently other more potent antiviral agents, including entecavir[2, 3] and tenofovir,[4] CHB patients are able to achieve continuous virologic suppression with NA therapy, reducing the chances of disease progression.[5, 6]

The quantification of serum hepatitis B surface antigen (HBsAg) has been recently advocated as another marker of disease activity in CHB. Unlike the fluctuating nature of serum HBV DNA,[7] natural history studies have found serum HBsAg to decrease very gradually with time.[8] Serum HBsAg levels have been shown to play a role in identifying inactive carriers with genotype D infection,[9] anticipating histologic severity,[10] determining risk of hepatocellular carcinoma (HCC),[11] and predicting HBsAg seroclearance.[12]

Although serum HBsAg levels have been demonstrated to have a predictive value in pegylated interferon therapy in CHB,[13] the role of serum HBsAg measurement in NA therapy has not been well defined. Recent studies have shown that serum HBsAg levels decline slowly despite persistent virologic suppression with NA therapy[14, 15] and could be used to predict virologic suppression during entecavir therapy.[16] Nonetheless, the duration of follow-up in these studies is short (1-2 years). An Italian study reported the changes in HBsAg kinetics during lamivudine therapy among CHB patients with a median follow-up duration of 66 months, but only included six patients with satisfactory virologic response.[17] In all these studies, the relationship of serum HBsAg kinetics with hepatitis B e antigen (HBeAg) seroconversion, HBV genotype, and HBsAg seroclearance during NA treatment have not been analyzed thoroughly.

In addition, the results of our recent study demonstrated human leukocyte antigen (HLA)-DP polymorphisms to be associated with spontaneous HBsAg seroclearance in CHB18; the effect of HLA-DP polymorphisms during NA treatment on serum HBsAg kinetics hence also warrants further investigations.

In our current study, we propose to investigate serum HBsAg kinetics among CHB patients who had a favorable virologic response to NA therapy and had been on NA therapy for a prolonged duration of time of more than 10 years. Because of this requirement, we primarily recruited CHB patients who responded favorably to lamivudine (the first approved nucleotide analogue for CHB since 1998) and therefore had treatment duration for at least 10 years.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

This was a retrospective study analyzing Han Chinese CHB patients started on lamivudine from different sources as depicted in Fig. 1. The first source was from three previous clinical trials conducted in the Department of Medicine, the University of Hong Kong, Queen Mary Hospital, Hong Kong (NUCB 3009, NUCB 3018, and NUCB 4003), with the date of lamivudine commencement between June 1994 and January 1998. The second source was from the Liver Clinic, Department of Medicine, the University of Hong Kong, Queen Mary Hospital, in which included patients were started on lamivudine between May 1998 and August 2002. All patients were HBsAg-positive for at least 6 months prior to commencement of lamivudine. Patients with decompensated liver function on presentation (albumin <35 g/L or bilirubin >1.5× upper limit of normal) or with coexisting liver disease, including chronic hepatitis C and D infection, primary biliary cirrhosis, autoimmune hepatitis, and Wilson disease, or with significant regular alcohol intake (>30 g/day for men, >20 g/day for women), were excluded. All patients consented to sera storage, with serum samples collected at each visit stored at −20°C until tested.


Figure 1. Flow chart showing the recruitment process of 70 patients in the current study. *NUCB 3009 and 3018 randomized patients to lamivudine 100 mg/day, lamivudine 25 mg/day, and placebo for 2 years, followed by open-label lamivudine 100 mg/day. ‡NUCB 4003 randomized patients to lamivudine 100 mg/day versus famciclover 500 mg three times per day for 12 weeks followed by open-label lamivudine 100 mg/day. #Patients developed either virologic breakthrough (increase of HBV DNA by ≥1 log from nadir) or HBV DNA persistently >2,000 IU/mL. Different therapeutic changes include switching to adefovir, entecavir 1 mg/day or tenofovir, and addition of adefovir or tenofovir. ^Included three patients intially on lamivudine 25 mg/day and three patients initially on placebo from NUCB 3009/3018, and seven patients initially on famciclovir 500 mg three times per day.

Download figure to PowerPoint

All patients were followed up regularly at our clinic with regular measurements of liver biochemistry, alpha-fetoprotein, and HBV serology and with assessment of drug adherence. Because the aim of the study was to investigate the change in HBsAg levels during prolonged successful NA therapy, we only included patients who had been on continuous lamivudine monotherapy for at least 10 years up to August 2012. Patients who stopped therapy, had documented drug incompliance, defaulted follow-up or responded suboptimally to lamivudine requiring a change in therapy within this 10-year period were excluded.

Serum HBsAg and HBV DNA levels were checked at baseline, year 5, year 10, and year 15 (for patients with 15 years of follow-up). A favorable virologic response was defined as achieving and maintaining serum HBV DNA <2,000 IU/mL after the commencement of lamivudine. Virologic breakthrough was defined as an increase of HBV DNA levels by at least 1 log IU/mL from the nadir for patients with detectable HBV DNA levels, or HBV DNA levels increasing to ≥20 IU/mL for patients with undetectable HBV DNA levels during therapy. The study was approved by the Institutional Review Board of the University of Hong Kong and West Cluster of Hospital Authority, Hong Kong.

Laboratory Assays

Serum HBeAg, antibody to HBeAg and antibody to HBsAg were measured by Abbott Laboratories (Chicago, IL). Serum HBsAg levels were measured using Elecsys HBsAg II assay (Roche Diagnostics, Gmbh, Mannheim), with a linear range of 0.05 to 52,000 IU/mL. Samples with levels higher than 52,000 IU/mL were retested at a dilution of 1:100 according to the manufacturer's instructions. Serum HBV DNA levels were performed using Cobas Taqman assay (Roche Diagnostics, Branchburg, NJ) with a lower limit of detection of 20 IU/mL. HBV genotype was determined in all patients using the INNO-LiPA HBV genotyping assay, which was performed according to the manufacturer's instructions (Innogenetics, Gent, Belgium). Resistance profile was performed using a line probe assay (Innogenetics, Gent, Belgium) for year 5 and 10 samples with detectable viremia. Genotypic resistance to lamivudine was defined by the presence of rtM204V/I with or without rtL180M.

HLA-DP rs3077 Polymorphisms

We genotyped HLA-DP single nucleotide polymorphism (SNP) rs3077, located in the HLA-DPA1 region of chromosome 6, for all subjects. SNP rs3077 was noted to be associated with HBsAg seroclearance in CHB in our previous study,[18] and was genotyped using TaqMan SNP genotyping assay (Life Technologies, Carlsbad, CA). Briefly, free circulating DNA was extracted from 200 μL of serum samples using a Purelink Genomic DNA Mini Kit (Life Technologies). The concentration of extraction DNA was then measured using a NanoDrop 2000c spectrophotometer (Thermo Scientific, Wilmington, DE). SNP genotyping was then performed using 5 ng of template DNA and a QuantiFast Probe PCR Kit (QIAGEN GmbH, Hilden, Germany), together with SNP-specific primers and FAM- and VIC-labeled probes, followed by real-time PCR reaction and SNP analysis using the RotorGene Q PCR System (QIAGEN). The possible genotypes for these bi-allelic polymorphisms are: CC, CT, and TT (T = minor allele).

Statistical Analysis

Continuous values are expressed as the median (range). For subjects with undetectable serum HBV DNA or HBsAg, the results were taken as the lower limit of detection (20 IU/mL and 0.05 IU/mL respectively). The annual rate of HBsAg reduction was expressed in logarithms (log IU/mL/year). The genotypic distribution of rs3077 polymorphism was tested for Hardy-Weinberg equilibrium. For statistical comparison, a Mann-Whitney U test or Kruskal-Wallis test was used as appropriate for continuous variables; a chi-squared test was used for categorical variables. Correlation between serum HBsAg levels and other variables was performed using Spearman's correlation coefficient. The prediction of HBsAg seroclearance during NA therapy was first examined by the construction of corresponding receiver operating characteristic (ROC) curves, then using areas under the curves (AUCs) to assess overall accuracy. The optimal level of prediction was attained using the Youden index, defined as sensitivity plus specificity minus one.[19]

Statistical analyses were performed using SPSS version 19.0 (SPSS Inc, Chicago, IL). A two-sided P value of <0.05 was considered statistically significant.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

Between June 1994 and August 2002, 322 CHB patients were started on lamivudine 100 mg/day at our center. After applying the selection criteria mentioned above, 70 patients (21.7%) maintained satisfactory virologic suppression (HBV DNA <2,000 IU/mL) during at least 10 years of continuous lamivudine therapy and were included in our current study. The details of patient deposition are illustrated in Fig. 1. Fifteen patients (21.4%) had been on lamivudine at least 15 years. The baseline demographics are depicted in Table 1. All 43 HBeAg-positive patients achieved HBeAg seroconversion after a median period of 2.82 (range, 0.31-10.8) years. Three patients (6.9%) experienced hepatitic flares before HBeAg seroconversion, with maximum alanine aminotransferase levels being 159, 198, and 279 U/L, respectively. There were no cases of HBeAg seroreversion.

The changes in HBV DNA levels during the 10-year period are depicted in Fig. 2. Fifty-two patients (74.3%) had undetectable HBV DNA (<20 IU/mL) after commencement of lamivudine. For patients with detectable viremia during follow-up (n = 18), the median HBV DNA levels at years 5 (n = 12) and 10 (n = 10) were 416 IU/mL (range, 28.8-1,949 IU/mL) and 437 IU/mL (range, 34.7-1,819 IU/mL), respectively. Of these, seven (38.9%) patients had virologic breakthrough, although two of these patients redeveloped undetectable viremia. The resistance testing for six of these seven patients showed wild-type HBV, with the remaining patient developing genotypic lamivudine resistance after 10 years of lamivudine. This patient had undetectable viremia during the first nine years of therapy but developed a HBV DNA level of 1,044 IU/mL at year 10. Serum alanine aminotransferase remained normal. Virologic suppression was subsequently achieved after switching from lamivudine to tenofovir.


Figure 2. Median HBsAg and HBV DNA levels throughout the 10-year study period. HBsAg/HBV DNA measurements are expressed in logarithms.

Download figure to PowerPoint

HBsAg Titers

Baseline median HBsAg levels were 10,800 and 1,590 IU/mL for HBeAg-positive and HBeAg-negative patients, respectively (Table 1). Baseline HBsAg levels showed a moderate inverse correlation with age (r = −0.385; P = 0.001) and moderate correlation with baseline HBV DNA levels (r = 0.296; P = 0.014).

Table 1. Baseline Parameters of All 70 Patients
  1. Data are presented as the median (range) or no. (%).

  2. Abbreviation: ALT, alanine aminotransferase.

Age starting lamivudine 100 mg/day, years38.5 (13.1-66.7)
Male sex56 (77.8)
HBV genotype 
B26 (37.1)
C41 (58.6)
Undetermined3 (4.3)
Fifteen years of follow-up15 (21.4)
HBeAg-positive43 (61.4)
Time to HBeAg seroconversion, years2.82 (0.31-10.8)
Albumin, g/dL42 (23-54)
Bilirubin, μmol/L15 (6-141)
ALT, U/L191 (22-1,079)
HBV DNA, IU/mL6.38 ×106 (56.8-6.14 ×108)
HBsAg, IU/mL4,790 (6.01-1.09 ×105)
HBeAg-positive10,800 (66.0-1.09 ×105)
HBeAg-negative1,587 (6.01-9.29 ×104)
Number of patients with HLA-DP rs3077 genotype 
CC38 (54.3)
CT27 (38.6)
TT5 (6.9)

The changes in HBsAg levels during the 10-year study period are depicted in Fig. 2. There was an overall statistically significant decline in median HBsAg titers from baseline to year 5 to year 10 (4,780 to 1,332 to 603 IU/mL, respectively; P < 0.001). The median rates of HBsAg reduction from baseline to year 5 and from year 5 to year 10 were 0.104 log IU/mL/year (range, −0.094-1.487 log IU/mL/year) and 0.072 log IU/mL/year (range, −0.241-1.190 log IU/mL/year) (P = 0.426). The decline in HBsAg levels was also significant when stratified by HBeAg status (HBeAg-positive, P < 0.001; HBeAg-negative, P = 0.002) and HBV genotype (genotype B, P = 0.006; genotype C, P < 0.001). The three patients with hepatitic flares before HBeAg seroconversion had a rate of HBsAg reduction of 0.151, 0.209, and 0.246 log IU/mL/year, respectively.

For the 15 patients (21.4%) on at least 15 years of lamivudine, the decline in median HBsAg titers were also significant (31,840 to 1,837 to 1,026 to 830 IU/mL for baseline, year 5, year 10, and year 15, respectively; P < 0.001).

The changes in HBsAg levels in relation to HBeAg status, HBV genotype, and HBV DNA detectability are shown in Fig. 3A-C. The median rates of HBsAg reduction for the different patient groups are also shown in Table 2. Rates of HBsAg reduction had no association with baseline HBeAg status, HBV genotype or HBV DNA detectability during the 10 years of treatment (all P > 0.05). During the 10-year study period, 18.6% (8/43) HBeAg-positive patients and 33.3% (9/27) HBeAg-negative patients achieved a >50% reduction of HBsAg titers in logarithm from baseline (P = 0.162). Among the 18 patients with detectable viremia (HBV DNA ≥20 IU/mL), there was no difference in the median rate of HBsAg reduction when comparing the four patients with persistently detectable viremia versus the remaining 14 patients (0.165 and 0.125 log IU/mL/year, respectively; P = 0.671)

Table 2. Median Annual Rates of HBsAg Reduction for Different Patient Groups
Patient GroupBaseline to Year 5 (log IU/mL/year)Year 5 to Year 10 (log IU/mL/year)Baseline to Year 10 (log IU/mL/year)Pa
  1. a

    Based on values obtained from baseline to year 10.

All patients (n = 70)0.104 (−0.094-1.487)0.072 (−0.308-1.190)0.104 (−0.026-0.843) 
HBeAg status   0.527
HBeAg-positive (n = 43)0.122 (−0.016-1.487)0.065 (−0.241-0.763)0.111 (−0.003-0.843) 
HBeAg-negative (n = 27)0.054 (−0.094-0.595)0.106 (−0.308-1.190)0.081 (−0.026-0.802) 
Genotype   0.120
B (n = 26)0.118 (−0.052-1.487)0.059 (−0.308-1.190)0.089 (−0.026-0.802) 
C (n = 41)0.099 (−0.094-0.612)0.117 (−0.241-0.763)0.112 (−0.014-0.843) 
HBV DNA   0.202
Undetectable HBV DNA (n = 52)0.104 (−0.094-1.487)0.072 (−0.308-1.190)0.089 (−0.014-0.843) 
Detectable HBV DNA (n = 18)0.103 (−0.052-0.364)0.155 (−0.063-1.121)0.133 (−0.026-0.622) 
HBsAg status   <0.001
HBsAg seroclearance (n = 7)0.364 (−0.048-1.487)0.942 (0.232-1.190)0.683 (0.341-0.843) 
No HBsAg seroclearance (n = 63)0.070 (−0.094-0.595)0.063 (−0.308-0.451)0.093 (−0.026-0.349) 

Figure 3. Median HBsAg levels in different patient groups stratified by (A) HBeAg status, (B) genotype, (C) HBV DNA detectability, and (D) subsequent HBsAg seroclearance. (C) Patients with undetermined genotype (n = 3) were excluded from analysis. (D) HBsAg measurements are expressed in logarithms. Error bars denote interquartile ranges.

Download figure to PowerPoint

Seven (10%) patients (genotype C, n = 4; genotype B, n = 2; undetermined genotype, n = 1) achieved HBsAg seroclearance during lamivudine therapy after a median duration of 7.59 years (range, 3.59-12.2 years). At the time of writing, two patients (28.6%) had developed antibody to the hepatitis B surface antigen. Four were baseline HBeAg-positive, with HBeAg seroconversion occurring after 1.32, 3.30, 3.91, and 7.35 years, respectively. These seven patients with HBsAg seroclearance, when compared with the remaining 63 patients without HBsAg seroclearance, had a significantly greater median rate of HBsAg reduction (0.683 and 0.093 log IU/mL/year, respectively; P < 0.001) (Fig. 3D and Table 2). They also had a significantly lower median baseline HBsAg level when compared with the remaining 63 patients (531 and 6,390 IU/mL, respectively; P = 0.012).

Prediction of HBsAg Seroclearance During NA Therapy

The ROC curves and the AUC values of different parameters predicting NA-related HBsAg seroclearance are shown in Fig. 4. Baseline serum HBsAg achieved an AUC of 0.860 (P = 0.004; 95% confidence interval [CI], 0.742-0.978), better than the AUC for the rate of HBsAg reduction (0.794; P = 0.018; 95% CI, 0.608-0.979). The optimal baseline HBsAg level and rate of HBsAg reduction to predict NA-related HBsAg seroclearance were 1,000 IU/mL (Youden index, 6.75; sensitivity, 85.7%; specificity, 84.1%; positive predictive value, 37.5%; negative predictive value, 98.1%) and 0.166 log IU/mL/year (Youden index, 5.63; sensitivity, 83.3%; specificity, 73.0%; positive predictive value, 26.1%; negative predictive value, 97.8%), respectively.


Figure 4. ROC curves of different parameters in predicting HBsAg seroclearance during NA therapy. The rate of HBsAg reduction is based on HBsAg levels at baseline and at year 5. *One patient with HBsAg seroclearance within the first 5 years was excluded from analysis.

Download figure to PowerPoint

Both baseline serum HBsAg <1,000 IU/mL (P = 0.006; odds ratio, 11.8; 95% CI, 2.02-68.97) and HBsAg reduction >0.166 log IU/mL/year (P = 0.003, odds ratio 17.3, 95% CI, 1.93-154.45) were significantly associated with subsequent HBsAg seroclearance.

Three patients (4.3%) with both baseline HBsAg <1,000 IU/mL and HBsAg reduction >0.166 log IU/mL/year did not achieve HBsAg seroclearance. After 10 years, their HBsAg levels had declined from 73.1, 210, and 980 IU/mL to 0.127, 128, and 151 IU/mL, respectively.

HLA-DP rs3077 Genotype

The genotypic distribution of rs3077 genotypes is shown in Table 1 and was in Hardy-Weinberg equilibrium (chi-squared, 0.005; P = 0.945). All seven patients achieving HBsAg seroclearance had the dominant C allele (CC genotype, n = 4; CT genotype, n = 3). Among patients with the dominant C allele (n = 65), 16 (24.6%) had baseline serum HBsAg <1,000 IU/mL; all five patients with the TT genotype had baseline serum HBsAg >1,000 IU/mL. When comparing the three rs3077 genotypes (CC versus CT versus TT), there was no significant difference noted in the median rate of HBsAg reduction (0.104, 0.117, and 0.081 log IU/mL/year, respectively; P = 0.884).

Among patients with the rs3077-dominant C allele (n = 65), the rate of HBsAg reduction achieved a better AUC in predicting NA-related HBsAg seroclearance (0.825; 95% CI, 0.655-0.996). The AUC of baseline serum HBsAg in predicting HBsAg seroclearance was similar (0.853; P = 0.005; 95% CI, 0.773-0.974).


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References

HBsAg seroclearance remains the ultimate therapeutic endpoint in the treatment of CHB. Few studies have investigated the factors influencing NA-related HBsAg seroclearance, not only because of its rarity in clinical practice, but also because potent NAs, including entecavir and tenofovir, have only been in clinical use for CHB for 7 and 4 years, respectively. Lamivudine was the first NA introduced for use in CHB, and although resistance is common, 25% to 30% of patients were able to maintain good virologic suppression with long-term therapy.[6, 20] A recent study showed such patients achieving a cumulative HBsAg seroclearance rate of 21.5% after 10 years of therapy.[21] Our current study excluded patients with fluctuating viremia due to either resistance or drug noncompliance and only included patients on decade-long therapy who had responded favorably. Although excluding patients with poor virologic control meant we were unable to study the relationship between serum HBsAg titers and drug resistance, our current cohort of patients would be ideal in investigating the changes in serum HBsAg kinetics during long-term NA therapy and its association with NA-related HBsAg seroclearance.

In our current study, with the long periods of low HBV DNA levels maintained during NA therapy, serum HBsAg decreased at approximately 0.1 log IU/mL/year, which is similar to the HBsAg decline rate seen in natural history studies.[12, 22] Our study results illustrate the different kinetics of both serum HBV DNA and HBsAg during NA therapy. HBsAg production exceeds that of virions, because HBsAg can be secreted by viral integration, which is a nonessential component in the life cycle of HBV.[23] In addition, NA therapy, despite suppressing the process of reverse transcription, only has a small effect in reducing intrahepatic covalently closed circular DNA (cccDNA) levels.[24] cccDNA is the template for the transcription of HBs messenger RNAs, and its continuous presence would result in the persistent secretion of serum HBsAg despite NA therapy.

Our current study showed that the best indicator of subsequent NA-related HBsAg seroclearance was the baseline HBsAg level (AUC 0.860) with an optimal predictive level of <1,000 IU/mL, strengthening the clinical applicability of serum HBsAg measurements during NA therapy. Low HBsAg levels of <1,000 IU/mL could reflect a moderate degree of immune control toward HBV,[9] although not to the extent of viral clearance. This level is higher than the optimal level used to predict spontaneous HBsAg seroclearance in CHB, which is identified within the range of 10 to 200 IU/mL,[8, 12, 25] suggesting that long-term NA therapy may increase the chance of HBsAg seroclearance in patients with modest levels of serum HBsAg between 200 and 1,000 IU/mL, levels in which the chance spontaneous HBsAg seroclearance is less likely. It will be interesting to have a further long-term study comparing the rates of HBsAg seroclearance with these modest HBsAg levels in patients with and without NA therapy. In addition, serum HBsAg <1,000 IU/mL in predicting NA-related HBsAg seroclearance achieved a high negative predictive value (98.1%), and hence performing HBsAg measurements at the commencement of NA therapy could identify patients with low probability of subsequent HBsAg seroclearance even when successful virologic suppression is achieved.

Another method of predicting NA-related HBsAg seroclearance is by observing the rate of HBsAg reduction (AUC 0.794). A decline rate of 0.5 log IU/mL/year was found to be predictive of both spontaneous and NA-related HBsAg seroclearance in previous studies.[12, 26] Based on our study results, an HBsAg reduction rate of >0.166 log IU/mL/year was predictive, again achieving a high negative predictive value (97.8%), and could be reflecting substantial restoration of host immune control that precedes HBsAg seroclearance. Although HBsAg levels were only measured at 5-year intervals in our study, the slow rate of decline in HBsAg levels with NA therapy in the initial 1 to 2 years[14, 15] suggests that the interval change over a period of several years might be more predictive of subsequent HBsAg seroclearance, with HBsAg reductions within the first year less predictive.[26] This is unlike the usage of serum HBsAg levels in pegylated interferon therapy, where an on-treatment decline of HBsAg levels over a period of months was already useful.[13]

The low positive predictive values (<40%) for both baseline HBsAg levels and rate of HBsAg reduction in this study would suggest there are other factors influencing NA-related HBsAg seroclearance. It is possible the host genome has a role in this; SNPs identified in genome-wide association studies have been shown to be associated with both spontaneous[18] and pegylated interferon-related HBsAg seroclearance.[27] Although our study did demonstrate patients with the HLA-DP rs3077 TT (T = minor allele) allele failing to achieve NA-related HBsAg seroclearance, such patients only constitute approximately 10% of the Han Chinese population.[28] Hence the identification of human genomic factors associated favorable outcomes in CHB for different ethnicities would require more in-depth sequencing studies.

As for patients with a high baseline HBsAg (≥1,000 IU/mL) or failing to achieve a significant HBsAg decline, HBsAg seroclearance during NA therapy would be an improbable treatment endpoint, with long-term NA therapy warranted. Nonetheless, novel treatment options (e.g., HBsAg release inhibitors) are currently undergoing clinical trials,[29] thus treatment-related HBsAg seroclearance could still be a reachable target for such patients in the future.

Our current study results did not find HBV genotype, HBeAg status, or the detectability of HBV DNA to influence the rate of HBsAg decline. Concerning HBV genotype, our study only investigated genotypes B and C, the common genotypes in the Asian CHB patients. Because most cases of reported NA-related HBsAg seroclearance are of genotypes A and D,[4, 30] it is possible that HBsAg levels undergo different kinetics in these different genotypes. Validation studies in CHB patients of European descent are thus needed to determine the applicability of the cutoff HBsAg levels found in our study.

In addition to the lack of more frequent measurements of HBsAg mentioned above, our study is limited by the relatively small number of patients with decade-long therapy and good virologic control (n = 70) and the small number of patients achieving HBsAg seroclearance (n = 7). As the number of CHB patients and the duration of continuous entecavir and tenofovir therapy increases, there should be additional data in the future to illustrate more detailed changes in HBsAg kinetics during long-term NA therapy. The prediction of HBsAg seroclearance in patients with different baseline HBsAg levels can then be more accurately assessed by these most potent NAs, in which the probability of drug resistance is expected to be minimal. Nevertheless, the results of the present study are likely applicable to patients receiving the more potent antiviral agents in the long term, because these agents should have more than 90% patients achieving undetectable HBV DNA levels (74.3% patients in the current study).

In conclusion, serum HBsAg levels decreased gradually during decade-long NA therapy (0.1 log IU/mL/year). Based on their high negative predictive values, baseline serum HBsAg <1,000 IU/mL and rate of HBsAg reduction >0.166 log IU/mL/year were the optimal levels in predicting NA-related HBsAg seroclearance. Serum HBsAg measurements hence have a role in the clinical monitoring of CHB patients, and in prognosticating CHB patients for the probability of eventual HBsAg seroclearance during NA therapy.


  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. References