Potential conflict of interest: Nothing to report.
The impact of prolonged direct antiviral therapy on hepatitis B surface antigen (HBsAg) levels in patients with chronic hepatitis B is poorly understood. We quantitatively assessed serum HBsAg levels during 3 years of telbivudine treatment, as well as their relationship with virologic and biochemical characteristics in 162 hepatitis B e antigen–positive patients who maintained undetectable serum hepatitis B virus (HBV) DNA long-term. Telbivudine treatment progressively reduced serum HBsAg levels (mean ± SD) from baseline (3.8 ± 0.6 log10 IU/mL) to treatment week 24 (3.4 ± 0.7 log10 IU/mL), treatment year 1 (3.3 ± 0.8 log10 IU/mL), and treatment year 3 (3.0 ± 1.4 log10 IU/mL) (P <0.0001). In this patient population, HBsAg loss was observed in nine (6%) of 162 patients through year 3. During the first year of treatment, three patterns of HBsAg decline were observed: rapid (≥1 log10 IU/mL) in 32 patients, slow (0-1 log10 IU/mL) in 74 patients, and steady levels in 56 patients. These findings were associated with different likelihoods of HBsAg loss during long-term telbivudine therapy. Eight of 32 patients with rapid HBsAg decline versus none of 56 patients with steady HBsAg levels achieved HBsAg loss at year 3 (P = 0.0024). HBV genotype was a significant determinant for HBsAg kinetics, with the fastest decline in genotype A patients. In patients with subsequent HBsAg loss, viral antigens were already undetectable in liver biopsy samples after 1 year of treatment. This was associated with markedly enhanced antiviral T cell reactivity. Conclusion: In patients who have effective suppression of viral replication during telbivudine treatment, a rapid decline in serum HBsAg levels during the first year may identify those with a greater likelihood of achieving HBsAg clearance. (HEPATOLOGY 2010
Hepatitis B surface antigen (HBsAg) is the most abundant viral protein produced during hepatitis B virus (HBV) replication, and its presence in the serum for >6 months is a hallmark of chronic HBV infection.1-3 During the course of chronic HBV infection, spontaneous clearance of serum HBsAg is rare, occurring at a rate of approximately 0.1%-0.8% per year, particularly in patients who acquired HBV through vertical transmission.4, 5 HBsAg clearance is considered the ultimate goal in the treatment of patients with chronic hepatitis B (CHB), because several studies have established that spontaneous or treatment-induced HBsAg loss is associated with sustained disease remission and improved survival.6-8
Recent data indicate that quantitation of serum HBsAg during therapy with pegylated interferon (PEG-IFN) may be useful in predicting the likelihood of HBsAg loss.9-13 The information regarding HBsAg kinetics during treatment with direct antiviral agents is limited.9, 14, 15 The nucleoside/nucleotide analogs are potent inhibitors of HBV replication, targeting HBV polymerase without a direct effect on HBsAg transcription and translation.16 Thus, HBsAg loss during treatment with these agents can occur only after a prolonged period of inhibition of HBV replication with undetectable serum HBV DNA and hepatitis B e antigen (HBeAg) clearance.17 The GLOBE trial has demonstrated that treatment with telbivudine leads to a rapid and profound inhibition of HBV replication. A subanalysis of this study identified undetectable HBV DNA at week 24 as the strongest predictor of optimal outcomes at 2 years.18, 19 Patients in the GLOBE trial who received 2 years of telbivudine therapy had the option to continue treatment in an extension study for 2 additional years.20 In the present study, we quantitated serum HBsAg levels in serial samples of patients with HBeAg-positive CHB who underwent up to 3 years of telbivudine treatment and maintained undetectable serum HBV DNA. Our aim was to investigate the kinetics of serum HBsAg levels and the relationship with virologic, biochemical, and immunologic characteristics. We further assessed the potential use of HBsAg monitoring to identify a subset of patients with effective on-treatment suppression of HBV replication who were likely to achieve HBsAg clearance during long-term telbivudine therapy.
ALT, alanine aminotransferase; CHB, chronic hepatitis B; HBcAg hepatitis B core antigen; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; PEG-IFN, pegylated interferon.
Patients and Methods
Of the 440 patients with detectable hepatitis B e antigen (HBeAg) at baseline in the telbivudine arm of the GLOBE study,18 395 patients completed 2 years of treatment, and 264 patients completed a third year of treatment in the subsequent 2-year extension study20 (Fig. 1). This group consisted of patients who had either completed 3 years of continuous telbivudine treatment (205 patients) or had discontinued telbivudine after 1-2 years of treatment due to efficacy (HBeAg loss and HBV DNA suppression) and followed an off-treatment protocol (59 patients).
The present analysis included 162 of these patients who completed 3 years of study and had a potential for HBsAg clearance, based on the criterion of maintained undetectable serum HBV DNA. Maintained undetectable serum HBV DNA was defined as HBV DNA (i.e., <300 copies/mL or 60 IU/mL) at treatment year 2 (for patients on continuous treatment) or the last on-treatment visit (for patients who had discontinued treatment before) and undetectable HBV DNA at 3 years or an additional visit after 3 years. The study was conducted in compliance with the Declaration of Helsinki and in accordance with Good Clinical Practice guidelines and local regulations.
HBV DNA determination was made using the COBAS Amplicor HBV Monitor polymerase chain reaction assay (Roche Molecular Systems, Pleasanton, CA) with a lower limit of detection of 300 copies/mL. HBV genotyping was performed by a central laboratory as described.20 HBsAg levels were quantified retrospectively using the ARCHITECT HBsAg assay (Abbott Diagnostics, Abbott Park, IL) from frozen serum samples at baseline and weeks 24, 52, 104, and 156 of telbivudine treatment or off-treatment follow-up. The lower limit of detection of this assay is 0.05 IU/mL. HBeAg levels were quantified retrospectively in frozen serum samples at baseline and weeks 24, 52, 104, and 156 of telbivudine treatment or off-treatment follow-up. The ARCHITECT HBeAg assay was used with modifications for quantitative use (see Supporting Information) with a lower limit of detection of 0.17 PEIU/mL.
Hepatic Expression of HBsAg and Hepatitis B Core Antigen in Patients with Seroclearance of HBsAg.
The hepatic expression of HBsAg and hepatitis B core antigen (HBcAg) (nuclear and cytoplasmic) was detected by immunostaining formalin-fixed, paraffin-embedded liver tissue from patients with HBsAg loss at 3 years. Antibodies to HBsAg (Zymed Laboratories, South San Francisco, CA) and HBcAg (Dako, Carpinteria, CA) were used, and staining was detected using the Dako EnVision Dual Link (Dako). A semiquantitative evaluation using a six-point ordinal scale was applied to estimate the number of positive hepatocytes. Histologic assessments were evaluated under blinded conditions by an independent pathologist (ZDG). Routine evaluation of liver biopsy samples to determine the Knodell histologic activity index and Ishak fibrosis score has been described.21
Evaluation of T Cell Activity Through Enzyme-Linked Immunosorbent Spot and Lymphocyte Proliferation Assays.
The enzyme-linked immunosorbent spot protocols used in this study have been described.22 Lymphocyte proliferation was measured by way of quantitation of bromodeoxyuridine intake and flow cytometry (see Supporting Information).
Statistical analyses were based on patients with HBeAg-positive CHB who demonstrated a positive response during up to 3 years of telbivudine therapy (i.e., patients who discontinued treatment early due to efficacy and patients who demonstrated undetectable HBV DNA at year 2 and the last available assessment). Only patients with at least 1 postbaseline HBsAg quantitation assessment were included. Missing observations were imputed with the last-observation-carried-forward method for assessing patient outcomes at year 3. For HBV DNA analyses, HBV DNA levels below the detectable limit of the polymerase chain reaction assay were assigned half the lower limit of quantitation (i.e., 150 copies/mL or 2.176 log10 copies/mL or 28.3 IU/mL). Subgroup analyses were performed according to the pattern of HBsAg decline at week 24 and year 1, and by HBeAg and HBsAg status at year 3. Two sample t tests or analysis of variance were used to compare continuous end points between any two subgroups, and a Fisher's exact test was used to compare discrete end points. A linear regression model was used to investigate the relationship between HBsAg and HBV levels.
During the 3-year study period, 115 (71%) of the 162 patients achieved HBeAg loss and 93 patients (57%) achieved HBeAg seroconversion. Nine patients (6%) lost serum HBsAg, and five of these nine patients developed antibody to HBsAg as well. A weak but significant positive correlation was noted between pretreatment HBsAg and serum HBV DNA levels (r2 = 0.2337; P < 0.0001). No correlation was observed between baseline HBsAg and alanine aminotransferase (ALT) levels.
HBsAg Kinetics During Treatment.
The kinetics of serum HBsAg differed from the rapid kinetics of serum HBV DNA, ALT, and HBeAg decline during telbivudine therapy (Fig. 2). Telbivudine treatment reduced mean HBsAg levels slowly but significantly (mean ± SD) from baseline (3.8 ± 0.6 log10 IU/mL) through treatment week 24 (3.4 ± 0.7 log10 IU/mL [P <0.0001]), with further declines through treatment year 1 (3.3 ± 0.8 log10 IU/mL), treatment year 2 (3.2 ± 1.1 log10 IU/mL), and treatment year 3 (3.0 ± 1.4 log10 IU/mL).
Patterns of HBsAg Kinetics.
Three patterns of HBsAg kinetics from baseline to year 1 were identified: rapid decline (≥1 log10 IU/mL), slow decline (0-1 log10 IU/mL), and steady HBsAg levels (≤0 log10 IU/mL) (Fig. 3A, Table 1). The evaluation of HBsAg kinetics patterns at year 1 showed that eight (89%) of the nine patients with HBsAg loss at year 3 demonstrated a rapid decline of ≥1 log10 IU/mL in HBsAg levels during the first year of telbivudine therapy (Fig. 3B). Of note, none of the patients with steady HBsAg levels achieved HBsAg loss by year 3.
Table 1. HBsAg Loss in Total Population at Year 3 of Telbivudine Therapy According to HBsAg Decline at Week 24 and Year 1 (N = 162)
Patients With HBsAg Loss
Patients With HBsAg Loss
Abbreviations: CI, confidence interval.
≥0.5 log10 IU/mL
≥1 log10 IU/mL
0–0.5 log10 IU/mL
0–1 log10 IU/mL
≤0 log10 IU/mL (no change)
≤0 log10 IU/mL (no change)
Similar analysis of the HBsAg kinetics patterns from baseline to treatment week 24 was conducted using thresholds of ≥0.5 log10 IU/mL and <0.5 log10 IU/mL declines for the rapid decline and slow decline groups, respectively. A rapid decline in HBsAg levels at treatment week 24 was observed in the majority of patients (seven [78%] of nine) who experienced HBsAg loss at year 3; only two (22%) of the nine patients demonstrated a slow decline in HBsAg levels.
Rapid HBsAg decline at week 24 (≥0.5 log10 IU/mL) and year 1 (≥1 log10 IU/mL) was a good predictor for subsequent HBsAg clearance, whereas steady HBsAg levels from baseline through week 24 identified patients who had little or no chance to clear HBsAg.
Characteristics of Patients with HBsAg Loss.
Analysis of baseline characteristics of all 162 patients indicated that HBV genotype, race, and HBsAg levels were linked to the consecutive probability of HBsAg clearance (Table 2). The overall probability of HBsAg loss in patients with HBV genotype A or D (four [19%] of 21 patients) was significantly higher than that of patients with genotype B or C (five [3.5%] of 141 patients; P < 0.0174). HBsAg loss occurred more frequently in Caucasian than in Asian patients (P = 0.0376).
Table 2. Demographics, Baseline, and On-Treatment Characteristics of Patients with and Without HBsAg Loss at Year 3
P values were for comparisons between patients with and without HBsAg loss at 3 years. For continuous variables, two-tailed t tests were used; for categorical variables, Fisher's exact tests were used.
Conversion factor (COBAS Amplicor): 1 IU = 5.6 copies/mL.
All nine patients had undetectable HBV DNA (<300 copies/mL) at year 1.
Biopsy samples were available for 130 of 153 patients who did not have HBsAg loss at year 3.
The majority of patients (115 [71%] of 162) in the study cohort experienced HBeAg loss by year 3. On-treatment declines in HBsAg levels from baseline were greater in patients with HBeAg loss than in patients without HBeAg loss at 3 years (Fig. 3C). A significant difference in HBsAg levels between the 2 groups was noted early, beginning at week 24 (3.3 ± 0.8 versus 3.7 ± 0.50; P = 0.0006) and was maintained at year 1 (3.2 ± 0.93 versus 3.59 ± 0.52; P = 0.002), year 2 (3.0 ± 1.25 versus 3.60 ± 0.50; P < 0.0001), and year 3 (2.7 ± 1.6 versus 3.5 ± 0.52; P < 0.0001). HBeAg seroconversion did not influence HBsAg decline; the HBsAg decline pattern was similar in patients with and without HBeAg seroconversion.
HBsAg Kinetics by Genotype.
Following the finding that HBsAg loss rates differed by HBV genotype, we analyzed HBsAg kinetics by genotype and found significant differences in HBsAg levels both at baseline (P < 0.001) and year 3 (P < 0.0031), with genotype A associated with the fastest decline (Fig. 4).
HBsAg decline was faster in patients with HBsAg loss than in the overall population. Of the nine patients who lost HBsAg, six had HBsAg loss before the second year of telbivudine treatment, and the other three had HBsAg loss at the end of year 3 (Fig. 3B). A rapid decline in HBV DNA was observed in all nine patients, with five reaching undetectable levels at week 24 and all reaching undetectable levels by year 1. There was no significant difference in the rate of undetectable HBV DNA at week 24 (<300 copies/mL or 60 IU/mL) between patients with and without HBsAg loss. Six patients had discontinued telbivudine therapy due to HBeAg seroconversion between treatment years 1 and 2; four of these patients lost HBsAg after discontinuation of the treatment.
Histologic Evaluation and Hepatic Expression of HBsAg and HBcAg.
We analyzed the declines in Knodell necroinflammatory and Ishak fibrosis scores for eight of the nine patients who lost HBsAg. These declines were comparable with those observed in the overall patient population (Table 2).
Liver tissue samples from eight patients with HBsAg loss at year 3 were available for immunostaining of HBsAg and both nuclear and cytoplasmic HBcAg (Fig. 5A). At baseline, the prevalent immunostaining corresponded to cytoplasmic HBcAg (median score, 4 [equivalent to 51%-75% of hepatocytes]), whereas HBsAg and nuclear HBcAg were detected at a lower intensity. At year 1 of telbivudine therapy, neither hepatic HBsAg nor nuclear or cytoplasmic HBcAg could be detected in any of the liver biopsy specimens of the patients with HBsAg loss. A representative section of a biopsy sample comparing the degree of HBsAg and HBcAg immunostaining at baseline and year 1 from patient 1 is shown in Fig. 5B.
Antiviral T Cell Responses.
T cell reactivity to HBV antigens was determined at baseline and during the first year of telbivudine therapy for two of the nine patients who lost HBsAg (Fig. 6), from whom peripheral blood mononuclear cells were available. In both of these patients, IFN-γ enzyme-linked immunosorbent spot analysis showed increased CD4-positive T cell reactivity to HBcAg and HBsAg at year 1 compared with earlier time points. T cell proliferation to HBcAg- and HBsAg-specific peptides was observed in patient 1 at week 24 and year 1, with markedly increased reactivity at year 1 compared with week 24.
Quantitative serology is emerging as an effective tool for the identification of on-treatment predictors of response to PEG-IFN treatment, in addition to monitoring serum HBV DNA levels.11, 12 The present study is the first to characterize HBsAg kinetics during prolonged oral treatment with a direct antiviral agent such as telbivudine and to define the relationship between different patterns of on-treatment HBsAg decline with subsequent HBsAg clearance.
Our results showed that in patients with HBeAg-positive CHB, telbivudine treatment significantly and progressively reduced serum HBsAg levels over a 3-year period. The reduction of serum HBsAg levels was slower in comparison with the rapid decline of HBV DNA and HBeAg levels, as well as the quick normalization of ALT levels. In the present study, nine (6%) of 162 HBeAg-positive patients who maintained undetectable HBV DNA on telbivudine therapy achieved HBsAg loss over 3 years of observation.
Three patterns of early on-treatment HBsAg kinetics were observed: rapid decline, slow decline, and steady levels. This is similar to patterns of HBsAg decline described in patients with HBeAg-positive CHB during PEG-IFN treatment.11, 12 Our results demonstrate that similar patterns were observed during treatment with the nucleoside analog telbivudine, which is also supported by the finding that PEG-IFN showed similar on-treatment HBsAg decline in a direct comparison.23 Importantly, patients who lost HBsAg by year 3 had considerably faster declines in HBsAg levels than the overall population. A rapid on-treatment HBsAg decline of ≥1 log IU/mL by year 1 was found to be associated with a high rate of consecutive HBsAg loss. In contrast, the steady levels of serum HBsAg during the first 24 weeks were an early indication that there was little chance of treatment-induced HBsAg reduction. Immunostaining of liver biopsy samples at 1 year showed that undetectable hepatic expression of viral antigens preceded the loss of serum HBsAg in most cases. Furthermore, we have shown that patients who cleared HBsAg had enhanced T cell reactivity with HBV-specific proliferation and IFN-γ production early during telbivudine treatment. These results indicate that long-term clearance of HBsAg would require effective immune control in addition to treatment-induced viral suppression. The implications of the different patterns of HBsAg decline identified in the present study for application in routine clinical practice would require future prospective investigations; however, the data support HBsAg quantitation as an additional tool for monitoring antiviral therapy.
Another important finding in the present study is that patients with HBV genotypes A and D, Caucasian race, and high pretreatment HBsAg levels have a higher probability of HBsAg clearance, which is in line with the findings of previous reports.12, 24-26 It is not known whether HBV genotype, race, duration of HBV infection, and route of transmission are independent determinants or are linked for this correlation. In a subanalysis, we found that HBV genotype is a significant determinant of HBsAg kinetics during telbivudine treatment, similar to the relationship recently shown during PEG-IFN treatment.27 This is consistent with reports showing that HBV genotype has an influence on the HBsAg/HBV DNA ratio, with positive correlations with genotypes B, C, and D, but no such correlation with genotype A.28-30 Based on these findings, it is apparent that cross-study comparisons of HBsAg clearance rates in patients undergoing different treatment regimens must be considered with caution, due to differences in study design and patient- and virus-related factors, especially race and HBV genotype.
In the present study involving patients with HBeAg-positive CHB, we found a low but significant correlation between baseline serum HBsAg and HBV DNA levels, similar to previous observations.9, 13 However, the correlation between serum HBV DNA and HBsAg levels is weak. Apart from the association with virions, HBsAg is translated from additional mRNAs and forms the noninfectious HBsAg particles that outnumber HBV virions in the serum by 1,000:1 to 10,000:1.16 Importantly, in patients with chronic HBV infection, HBsAg is produced not only in hepatocytes with HBV replication, but in hepatocytes with integrated HBV DNA fragments.31 Double immunostaining has shown the expression of HBcAg and HBsAg in different hepatocytes, with an accumulation of HBsAg-positive hepatocytes in patients with antibody to HBeAg–positive CHB derived from integrated HBV DNA.32 These data on HBV replication and HBsAg production also help explain the present findings that the previously identified strong predictors of favorable virologic or serologic response (e.g., HBV DNA levels <9 log10 copies/mL, undetectable HBV DNA at treatment week 24) showed only a weak correlation with the rate of HBsAg clearance.19 In our study, eight of nine patients with HBsAg loss had HBV DNA levels >9 log10 copies/mL at baseline, and in four patients HBV DNA was still detectable by treatment week 24. All nine patients experienced HBeAg loss before clearance of HBsAg, and even in the overall population, HBsAg levels declined markedly faster in patients with subsequent HBeAg loss than in patients without HBeAg loss, indicating that both virologic and serologic response might be a necessary (albeit insufficient) prerequisite for HBsAg clearance. Furthermore, six of the nine patients had discontinued telbivudine treatment due to HBeAg seroconversion between 1 and 2 years, and in four patients HBsAg clearance had occurred during the follow-up phase. These data highlight the fact that HBsAg levels are not only associated with HBV DNA levels, but might be controlled by additional mechanisms.
The results from this first analysis of HBsAg kinetics in a large cohort of patients during telbivudine treatment show that prolonged and effective viral suppression with a nucleoside/nucleotide analog may lead to HBsAg clearance. The present study also revealed that rapid on-treatment declines in HBsAg levels occurring up to week 24 and by year 1 are associated with future HBsAg clearance. The failure to induce HBsAg reduction in approximately 30% of this patient cohort with maintained effective HBV suppression might be overcome with a longer treatment period, but it might also be an indication that achieving the ultimate treatment goal of HBsAg clearance is not feasible in these patients. However, the overall rates of HBsAg clearance with currently available treatment strategies are low, and new treatment strategies are needed to achieve a long-term cure for patients with CHB. HBsAg monitoring, together with HBV DNA levels, may be helpful to define and evaluate new strategies.
The authors thank Kathleen Covino, PhD, for her editorial assistance in the development of this manuscript, which was supported by Novartis Pharma AG.