Potential conflict of interest: The author has been involved in clinical trials and has served as a global advisory board member for Roche, Bristol-Myers Squibb, Novartis, and Gilead Sciences.
The author thanks the Chang Gung Medical Research Fund for its long-term grant support (SMRPG1005 and BMRPG380061) and Roche Diagnostics (Rotkreuz, Switzerland) for its financial support.
This clinically relevant review focuses on recent findings concerning hepatitis B surface antigen (HBsAg) quantitation in untreated patients and treated patients with chronic hepatitis B. Recent studies and emerging data have shown that both HBsAg and hepatitis B virus (HBV) DNA levels decline during the natural course of a chronic HBV infection; they are lowest in the inactive phase, which is also characterized by the highest HBsAg/HBV DNA ratio. It has been demonstrated that the combined use of HBsAg and HBV DNA levels might help in the identification of true inactive carriers with high accuracy. Retrospective analyses of HBsAg levels in patients undergoing therapy have suggested a role for HBsAg quantitation in monitoring the response to therapy. In comparison with nucleos(t)ide analogues (NAs), interferon-based therapy results in greater overall declines in serum HBsAg levels. A rapid on-treatment decline in HBsAg levels appears to be predictive of a sustained response. With the aid of HBsAg quantitation, it appears that we can anticipate an individualized approach to tailoring the treatment duration. The proposal of early stopping rules for patients not responding to pegylated interferon (according to a lack of any HBsAg decline) represents a step toward a response-guided approach. The development of stopping rules for patients treated with NAs is desirable for reducing the need for lifelong therapy. However, before stopping rules for antiviral therapy can be applied, we need to learn more about the kinetics of HBsAg declines during the natural history of the infection and as a response to therapy so that we can better define the best timing, the relevant HBsAg cutoff levels, and the best ways to apply these rules in clinical practice. (HEPATOLOGY 2011;)
The detection of hepatitis B surface antigen (HBsAg) in serum was pivotal to the discovery of hepatitis B virus (HBV) more than 4 decades ago and remains the cornerstone of diagnosis today.1-3 HBsAg seroclearance is considered to be the closest thing to a cure for chronic hepatitis B (CHB): it reflects immunological control of the infection and confers an excellent prognosis in the absence of preexisting cirrhosis or concurrent infections with other viruses.2-6 Not surprisingly, HBsAg seroclearance has attracted considerable attention in both natural history studies and therapeutic trials.
The incidence of spontaneous HBsAg seroclearance is low, especially in younger patients. Interferon (IFN) therapy appears to be able to enhance the rate of HBsAg seroclearance from 0.72% (controls) to 2.25% per year in European patients and from 0.07% to 0.43% per year in Asian patients.6 A greater understanding of the factors influencing HBsAg levels might enable us to improve this still further. Recently, a wealth of new data on HBsAg quantitation has emerged, and it is becoming apparent that information on HBsAg levels can add to our understanding of both the natural history of the disease and its response to therapy. This is a good time to review and discuss issues concerning the clinical utility of HBsAg quantitation and the ways in which this may help us with patient management in the future.
ALT, alanine aminotransferase; anti-HBe, antibody to hepatitis B e antigen; cccDNA, covalently closed circular DNA; CHB, chronic hepatitis B; ETV, entecavir; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCV, hepatitis C virus; IFN, interferon; LAM, lamivudine; LdT, telbivudine; NA, nucleos(t)ide analogue; NPV, negative predictive value; PEG-IFN, pegylated interferon; PPV, positive predictive value; TDF, tenofovir.
HBsAg and HBV DNA Levels
Our understanding of the pathogenesis and natural history of CHB has been facilitated by technological advances that have improved the sensitivity of both serological assays for quantifying antigens (including HBsAg) and polymerase chain reaction assays for measuring HBV DNA. Several independent groups have compared HBsAg and HBV DNA levels during different phases of the disease, and their findings have been rather consistent. To put these findings into context, we must consider the HBsAg production pathway and the ways in which this is related to serum HBV DNA levels and intrahepatic covalently closed circular DNA (cccDNA).
HBsAg is produced by more than one pathway (Fig. 1): the translation of transcriptionally active cccDNA molecules, which serve as a template for replication, and the translation of viral genes transcribed from integrated HBV DNA sequences in the host genome.7, 8 In addition to being the envelope of infectious HBV particles, HBsAg is also found in the form of noninfectious spheres or filaments, which exceed infectious virions in number by 102 to 105.9 Serum HBsAg appears to correlate with transcriptionally active cccDNA and is considered a surrogate marker of infected cells.10-14 Although cccDNA is the most accurate reflection of the number of infected hepatocytes, it can be assessed in tissue only with complex techniques that are restricted to specialized research centers. This excludes the analysis of cccDNA levels from general clinical applications.
The quantitation of HBV DNA by polymerase chain reaction is now a standard part of the diagnostic workup for CHB. A serum HBV DNA decline reflects a reduction in viral replication. In contrast, a serum HBsAg decline represents a reduction in the translation of messenger RNAs produced from transcriptionally active cccDNA or integrated sequences.14 Thus, HBsAg quantitation provides different but complementary information that may aid us in the characterization of an individual's infection status.
HBsAg and HBV DNA Levels During Different Phases of CHB
Several cross-sectional studies have compared HBsAg and HBV DNA levels during different phases of CHB (Table 1). The results are encouragingly similar, even though the studies were conducted in different patient populations and, therefore, with different genotypes. Both HBsAg and HBV DNA levels vary during the natural course of the infection, and they are highest in the initial immune tolerance phase when the serum alanine aminotransferase (ALT) level is normal with no or minimal hepatitis activity. HBsAg levels become lower during the immune clearance phase and decrease slowly and progressively in those who maintain persistently normal ALT levels after hepatitis B e antigen (HBeAg) seroconversion.10 All groups have observed the lowest levels of HBsAg and HBV DNA during this inactive phase, which is also characterized by the highest HBsAg/HBV DNA ratio.7, 10, 15 A Hong Kong follow-up study of 68 HBeAg-negative patients over a median period of 8 years showed a slow overall decrease in HBsAg levels, and a >1 log10 IU/mL HBsAg decline between the initial and last visits reflected improved immune control, which was associated with a higher HBsAg seroclearance rate and stronger viral suppression.10 Two European studies of inactive HBsAg carriers showed that those with subsequent HBsAg seroclearance had a significantly greater HBsAg decline than those who remained HBsAg-seropositive (0.28-0.29 versus 0.054-0.058 log10 IU/mL/year).16, 17 A longitudinal study of 47 Taiwanese HBeAg-negative carriers of HBsAg with subsequent HBsAg seroclearance showed that the median HBsAg level decreased to <2 log10 IU/mL; 82% of the patients reached a level < 200 IU/mL, and 67% reached a level < 100 IU/mL 3 years before HBsAg seroclearance (Y. C. Chen and Y. F. Liaw, unpublished data March 2011).
Table 1. Levels of HBsAg and HBV DNA During the Natural Course of CHB and Their Relationship
Immune Tolerance Phase
Immune Clearance Phase
Immune Control/ Inactive Carrier
Reactivated HBeAg-Negative Disease
Abbreviation: anti-HBe, antibody to hepatitis B e antigen.
There is still debate about the HBV DNA cutoff level used to define the inactive HBsAg carrier state. According to information from natural history studies, it may be possible to better differentiate between true inactive carriers and those with active disease. In comparison with inactive carriers, HBeAg-negative patients who experience reactivation have higher HBsAg and HBV DNA levels.7, 10, 15, 16 Several groups have proposed cutoff levels of HBsAg and HBV DNA that, when used together, reliably identify patients with inactive disease.15-19 Although the exact values differ slightly, they are approximately 1 to 2 × 103 IU/mL for HBsAg and 2 × 103 IU/mL for HBV DNA. With these values, inactive carriers can be identified with 94% to 100% accuracy. The cutoff values derived from large studies by Brunetto et al.,16 Martinot-Peignoux et al.,17 and Manesis et al.19 seem to be most applicable (Table 2). However, the results of retrospective analyses require further validation by prospective studies of patients infected with all the major genotypes. Although we can anticipate some differences according to the genotype, further studies will likely confirm that HBsAg levels have potential value in managing CHB patients because they can be used to define more clearly who requires treatment and who does not. Their use could even reduce the need for liver biopsy in those who concurrently have mildly elevated ALT levels and low levels of both HBsAg and HBV DNA.20 For patients with values above these cutoff levels, more frequent monitoring would be advised for the detection of reactivation.
Table 2. HBsAg and HBV DNA Cutoffs Proposed for Differentiating True Inactive Carriers From Patients With Reactivated HBeAg-Negative Disease
Decline of HBsAg During Therapy With Various Agents
The suggestion that the measurement of HBsAg levels might be valuable for monitoring responses to IFN therapy in HBeAg-positive patients was first proposed in 1994 when a significant HBsAg decline was observed in patients who responded to IFN with HBeAg seroconversion but not in patients without HBeAg seroconversion (P < 0.001); thus, HBsAg quantitation was proposed as a simple means of monitoring patients with CHB.21 However, the lack of commercially available assays precluded its widespread application until recently. Reports of HBsAg quantitation in HBeAg-negative patients with HBV infections or HBV/hepatitis delta virus dual infections who were undergoing therapy again suggested the potential of this marker for monitoring the response to therapy.22, 23 It was also proposed that HBsAg monitoring could predict eventual HBsAg clearance23, 24 after approximately 5.4 years of a sustained response to IFN or after 10.6 years of viral suppression with lamivudine (LAM) maintenance therapy.23 Subsequent studies have clearly demonstrated that IFN-based therapy results in a greater overall HBsAg decline than treatments with a nucleos(t)ide analogue (NA), as summarized in Table 3.22, 25-34 This suggests that the HBsAg decline is affected more by immune modulation than an antiviral effect.
Table 3. Decline of HBsAg Levels During Therapy
PEG-IFN and NA
For comparison only (this was not a randomized controlled trial). : Comparison of data with same marker.
Predicting Responses to IFN-Based Therapy With HBsAg Levels
Because a sustained response to pegylated interferon (PEG-IFN) is achieved in only approximately 35% of HBeAg-positive patients and 25% of HBeAg-negative patients, identifying a potential treatment success is valuable for both the patient and the physician. A number of groups have retrospectively analyzed response rates in PEG-IFN cohorts with respect to on-treatment HBsAg declines.
Predicting a Sustained Response.
In HBeAg-positive patients, one study showed that a baseline HBsAg level < 10,000 IU/mL was associated with a higher rate of response to PEG-IFN therapy.35 Other studies have not confirmed this observation but have reported a significant association between on-treatment levels of HBsAg and responses to PEG-IFN. A large European study of 202 patients treated with PEG-IFNα2b with or without LAM for 52 weeks showed that responders (response was defined as an HBeAg loss with HBV DNA levels < 1 × 104 copies/mL 26 weeks after treatment) experienced a more profound HBsAg decline at week 52 (3.3 versus 0.7 log10 IU/mL) and week 78 (3.4 versus 0.35 log10 IU/mL, P < 0.001). Moreover, any HBsAg decline at week 12 had a positive predictive value (PPV) of 25% for a response and a PPV of 15% for HBsAg loss up to 3 years after treatment.26 A Hong Kong study of 92 patients who were treated with PEG-IFNα2b with or without LAM for 32 to 48 weeks found that HBsAg levels < 1500 IU/mL at month 3 and HBsAg levels < 300 IU/mL at month 6 (21% of the patients) could predict a sustained response 12 months after treatment (the PPVs were 46% and 62%, respectively). In addition, the combination of an HBsAg level ≤ 300 IU/mL and a >1 log reduction at month 6 had a PPV of 75%.35 A small study from China showed that an HBsAg level < 1500 IU/mL at week 12 of IFNα/PEG-IFNα therapy had a PPV of 33% for HBeAg seroconversion after end of 24 weeks of treatment.36 Piratvisuth et al.37 reported that HBsAg levels < 1500 IU/mL at week 12 of PEG-IFNα2a treatment (23% of the patients) were associated with an HBeAg seroconversion rate of 57% 6 months after treatment; 18% of these patients experienced HBsAg clearance.
In HBeAg-negative patients, the baseline HBsAg level could not predict the response to PEG-IFN therapy,32, 38, 39 but sustained responders had marked decreases in their serum HBsAg levels at the end of treatment (2.1 ± 1.2 log10 IU/mL) and at week 72.38 Brunetto et al.32 further indicated that both an HBsAg level ≤ 10 IU/mL at week 48 (12% of the patients) and an on-treatment HBsAg decline > 1.1 log10 IU/mL (22% of the patients) were significantly associated with HBsAg clearance 3 years after treatment (relative risks of 22.8 and 10.8, respectively, P < 0.0001). Moucari et al.38 also found a significant association between an HBsAg decline and a sustained response; they reported that decreases of 0.5 and 1.0 log10 IU/mL at week 12 (19% of patients) and week 24 (25% of patients) of PEG-IFNα2a therapy had high PPVs (89% at week 12 and 92% at week 24). A study of 120 patients showed that a decline ≥ 10% at week 12 of PEG-IFNα2a therapy was associated with a 1-year off-therapy sustained response of 47% and an HBsAg seroclearance rate of 23% 5 years after treatment.33 In contrast, the HBV DNA declines throughout treatment were nearly identical for sustained responders and relapsers; this suggests that measurements of the HBsAg concentration may more reliably distinguish those destined to have a sustained response.38 Furthermore, the association of HBsAg reductions with sustained responses was observed across the major genotypes (A-D).40
Although these low on-treatment levels were reached by less than 25% of the treated patients, these encouraging data suggest a potential role for HBsAg levels in predicting the response to PEG-IFN. This could encourage or motivate patients to complete their course of therapy.
The ability to determine who is most unlikely to achieve a sustained response to PEG-IFN might be of more practical value for patient management. The early identification of nonresponders would allow the discontinuation of therapy and/or changes in the treatment strategy for these patients. High negative predictive values (NPVs) for response have been reported for both HBeAg-positive and HBeAg-negative patients. Sonneveld et al.26 reported an NPV of 97% for 202 PEG-IFNα2b–treated, HBeAg-positive patients (74% with genotype A or D HBV), which was based on any decline in HBsAg levels at week 12. An HBsAg decline at week 12 had an NPV of 82% in another large study of PEG-IFNα2a therapy (88% with genotype B or C HBV).41 The Hong Kong study reported an NPV of 86% for HBsAg levels < 1500 IU/mL at month 3 and an NPV of 89% for levels < 300 IU/mL at month 6.35 The Chinese study also showed that an HBsAg level < 1500 IU/mL at week 12 had an NPV of 91%, whereas the NPV was 95% when the cutoff level was 2890 IU/mL at week 24.36
For HBeAg-negative patients, Moucari et al.38 reported an NPV of 90% for an HBsAg decline of 0.5 log10 IU/mL at week 12 and an NPV of 97% for a decline of 1 log10 IU/mL at week 24 in a mixed-genotype population. In a population that mainly had genotype D, Rijckborst et al.39 reported an NPV of 100%, which was based on a combination of an HBsAg decline and a 2 log10 IU/mL decline in HBV DNA levels from the baseline to week 12. This proposed stopping rule was recently validated in another cohort of patients treated with PEG-IFN.42
These apparently robust early stopping rules with high NPVs could help with the management of patients and may even encourage patients to consider PEG-IFN as first-line therapy. This may be particularly applicable when the alternative is most likely lifelong therapy with NAs, especially for patients with HBeAg-negative disease. On the basis of these studies in different populations with different genotypes, week 12 of IFN-based therapy seems to be the right time for assessing an HBsAg decline (Table 4). However, the most appropriate degree of this decline still needs to be established before it can be adopted by the community as a guide for clinical practice.
Table 4. On-Treatment Prediction of Responses to PEG-IFN Therapy
HBsAg profiles were also analyzed retrospectively in hepatitis C virus (HCV)/HBV-coinfected individuals treated with PEG-IFN and ribavirin for their predominant HCV infection. The HBsAg levels tended to be lower in this patient cohort (median = 120 IU/mL) versus HBV-monoinfected patients and decreased gradually during treatment. Low baseline HBsAg levels were associated with HBsAg clearance (40% for baseline HBsAg levels ≤ 20 IU/mL and 2% for levels > 20 IU/mL, P < 0.05). Interestingly, a 50% decrease in the HBsAg level from the baseline to week 12 was associated with a reduced likelihood of HBV DNA reactivation in patients with serum HBV DNA levels that were undetectable at the baseline (PPV = 89.5%).43 This raises the possibility that we might be able to identify those HCV/HBV-coinfected patients who are likely to experience HBV reactivation and may need additional therapy with NAs.
Predicting Responses to NAs With HBsAg Levels
HBsAg declines during NA therapy appear more apparent in HBeAg-positive patients than HBeAg-negative patients, at least in the short term (Table 3). Although differences in the inclusion criteria preclude a comparison of HBsAg reduction across studies, one study involving both HBeAg-positive patients and HBeAg-negative patients showed that a substantial HBsAg decline during entecavir (ETV) therapy was restricted to HBeAg-positive patients.29 A small study from Korea showed that a decrease > 1 log10 IU/mL in serum HBsAg levels during therapy (5 of 28 patients) was associated with a much higher cumulative incidence of HBeAg loss (80% versus 30%, P = 0.034) after 1 year of ETV therapy.28 Notably, one study showed that the HBsAg decline during ETV therapy was restricted to patients with baseline ALT levels greater than or equal to 2 times the upper limit of normal (P = 0.007), and it was most profound in those who lost HBeAg.29 This suggests that the HBsAg decline might be linked to increased immunological activity.
HBsAg seroclearance is sometimes reported during NA therapy in HBeAg-positive patients. With tenofovir (TDF), HBsAg seroclearance rates of 3%, 6%, and 8% were reported after 1, 2, and 3 years of therapy in HBeAg-positive patients, but seroclearance was not observed in HBeAg-negative patients.44 HBsAg seroclearance rates for HBeAg-positive patients treated with ETV or LAM after 96 weeks on treatment and 24 weeks off therapy were 5% and 3%, respectively.45 Recent studies have shown that a rapid decline in HBsAg levels during the first year of NA therapy in HBeAg-positive patients is associated with a higher probability of HBsAg seroclearance: 8 of 32 patients with a rapid HBsAg decline > 1 log10 IU/mL during telbivudine (LdT) therapy achieved HBsAg clearance in year 3, whereas 0 of 56 patients with steady HBsAg levels achieved this (P = 0.0024).30 Similarly, patients with HBsAg loss during TDF therapy, who were most frequently infected with genotype A or D, exhibited a rapid HBsAg decline, and the HBV DNA decline pattern in patients with or without HBsAg loss was similar.31
There would be a considerable advantage in being able to determine whether a patient could stop NA therapy after a successful response. The current recommendation is to consider stopping after 12 months of consolidation therapy following HBeAg seroconversion, whereas NAs should be stopped only after HBsAg clearance is achieved in HBeAg-negative patients.3 However, the durability of NA-induced HBeAg seroconversion is at best 80% in LdT-treated patients during 2 years of off-therapy follow-up.46 A small study of 17 patients who showed a sustained response to LdT (defined as HBV DNA levels < 300 copies/mL, HBeAg seroconversion, and ALT normalization 2 years off treatment) found that the HBsAg decline rates at weeks 24 and 52 were better predictors of the off-treatment response than the HBV DNA decline rates. Moreover, an HBsAg level < 2 log10 IU/mL at treatment week 104 was highly predictive of a sustained off-treatment response (PPV = 93%, NPV = 100%).47 If this is confirmed by other appropriate studies, HBsAg quantitation may help us to identify patients who are able to stop NAs with only a very low chance of relapse.
Currently, data on HBsAg levels during NA therapy are insufficient. Although a rapid HBsAg decline (>1 log10 IU/mL) during NA therapy appears predictive of an off-treatment response, the reports to date are based on only small numbers of patients, and predictors have not been clearly defined. Although a stopping rule for NA-treated patients is highly desirable from a clinical perspective, further studies are clearly required to explore the potential of HBsAg in this context.
Conclusions and Perspectives
In summary, recent studies and emerging data have shown that HBsAg levels change during the natural course of a chronic HBV infection, and a rapid decline in HBsAg levels indicates a strong response to therapy, regardless of which treatment approach is being used. It appears that the information provided by the monitoring of HBsAg levels (in addition to HBV DNA levels) may help us to determine the best management strategy for a considerable proportion of patients. An overview of the potential clinical applications of HBsAg quantitation is provided in Table 5. Proposed early stopping rules for PEG-IFN that are based on HBsAg declines can increase the appeal of trying this approach first and represent a step toward a response-guided approach. The development of a stopping rule for highly potent NAs would be a desirable step for reducing the burden of lifelong therapy.
Table 5. Potential Applications of HBsAg Level Monitoring
During natural history
Identification of true inactive carriers
Reassurance that treatment is not required
Identification of patients who need therapy now or whose disease is likely to be reactivated in the near future
Identification of patients who need closer monitoring and possible identification of patients who need treatment
Early identification of patients who are unlikely to respond to PEG-IFN
Early stopping rule for avoiding unnecessary and ineffective therapy
Early identification of patients who are responding to PEG-IFN
Motivation for patients to continue therapy
Early identification of patients who experience a rapid decline in HBsAg levels during NA therapy (LdT or TDF)
Identification of patients who have a high chance of HBsAg clearance and development of a stopping rule that enables patients with a low chance of relapse to stop NA therapy
The accessibility of an assay contributes considerably to its value and its likelihood for implementation in routine clinical practice. If HBsAg levels are to be used as a potential guide to patient management, there is a clear need for standardized commercial assays providing accurate results that are traceable to a standard reference serum. The Architect HBsAg assay (Abbott Diagnostics) was used in most of the studies reviewed here. This assay requires a 1:100 to 1:1000 dilution of patient sera in most instances. An excellent correlation between the Architect assay and the Elecsys HBsAg II assay (Roche Diagnostics) has been demonstrated.48 Using an automated onboard dilution step, the latter has a broad linear range that covers the HBsAg levels most frequently encountered and thus reduces the need for manual dilutions, which are potential sources of error.49 Currently, the cost of these assays is not reimbursed in many countries, and they are not commercially available in the United States, so research-only tests are the only option at present. However, this can be expected to change in the future.50
Undoubtedly, there is still more to learn about the kinetics of the HBsAg decline and the ways to best use this in practice to optimize therapy. It remains to be confirmed whether HBsAg levels can reliably predict HBeAg seroconversion or HBsAg seroclearance. Studies in regions other than Europe and Asia are needed because the HBsAg kinetics for different HBV genotypes may differ during the natural course of the disease or in response to anti-HBV therapy. The on-treatment predictive value of HBsAg quantitation also needs to be studied in a sufficiently large number of patient with consistent time points (e.g., weeks 12 and 24 of therapy) and with the same definition of response. The optimal HBsAg cutoff with the ideal PPV and NPV also awaits clarification. Prediction models combining the quantitation of HBsAg with HBV DNA and ALT levels should also be explored. Until these issues are resolved, HBsAg quantitation will not be ready for clinical practice. Nevertheless, with the assistance of HBsAg quantitation, we may be on our way to establishing an individualized approach that might enable us to tailor anti-HBV treatments.
The author thanks Karen Searle (Elements Communications, Ltd.) for her editorial assistance and Su-Chiung Chu for her secretarial assistance.