HBsAg was the first hepatitis B virus (HBV) protein to be discovered (1). Detection of HBsAg in serum, called the ‘Australia antigen’, was the Nobel prize discovery identifying HBV about 40 years ago. Over the years, HBsAg has proven to be a steady, reliable but unspectacular marker of active HBV infection. HBsAg is still the hallmark of overt HBV infection today, and the detection of HBsAg in serum is still the fundamental diagnostic marker of HBV infection (2, 3). HBsAg seroconversion is the ultimate laboratory marker of successful therapy in patients with chronic hepatitis B (CHB). Individuals who become HBsAg negative and develop anti-HBs antibodies can be considered to have resolved hepatitis B (4).
HBsAg is a very important clinical test that might not only indicate active hepatitis B virus (HBV) infection but might also be used to predict clinical and treatment outcome. Clearance of HBsAg in patients with chronic HBV infection is associated with a much better clinical outcome, although surveillance for early detection of hepatocellular carcinoma (HCC) should continue. HBV DNA quantification is currently used for selecting candidates for therapy, monitoring response to therapy and detecting the emergence of drug resistance. Assays for HBsAg quantification are less expensive than HBV DNA and fully automated with a high throughput capacity. HBsAg titering may be a useful tool to manage patients with chronic HBV, to more clearly define which patients may, and more importantly, may not, benefit from treatment. Baseline and on-treatment HBsAg quantification may help to refine future treatment algorithms for both immune-modulator therapy and nucleos(t)ide analogues. Both HBV markers provide complementary information on the status of HBV infection. However, the relevance of serum HBsAg levels and its use as a reliable replacement for both covalently closed circular DNA and HBV DNA remain unclear.
covalently closed circular DNA;
chronic hepatitis B;
hepatitis B virus;
pegylated interferon alfa-2a;
sustained virological response.
Hepatitis B virus is a small DNA virus belonging to the Hepadnaviridae family. The viral genome encodes four open reading frames (ORFs: S, C, P and X) and replicates through an RNA intermediate using a viral polymerase with reverse-transcriptase activity. Upon entry into the cell, HBV sheds its protein coat and the relaxed, partially double-stranded genome is transported into the nucleus. In the nucleus, host and viral polymerase repair the relaxed, partially circular genome to a fully double-stranded, covalently closed circular genome or covalently closed circular DNA (cccDNA). The cccDNA resides in an infected hepatocyte nucleus as a stable, resistant and enduring non-integrated minichromosome, where it acts as a template for the transcription of viral genes (5, 6).
Viral proteins of clinical importance include the envelope protein (HBsAg), the structural nucleocapsid core protein (HBcAg) and a soluble nucleocapsid protein (HBeAg). HBsAg synthesis during the HBV viral life cycle is complex, and usually occurs at the endoplasmic reticulum (ER). The envelope ORF contains three in frame ‘start’ codons, which further divide it into pre-S1, pre-S2 and ORF-S domains. Envelope proteins are generated from two HBV mRNA transcripts, with subsequent translation resulting in small (ORF-S), medium (pre-S2+ORF-S) and large surface envelope proteins (pre-S1+pre-S2+ORF-S). These are also known as L, M and S surface proteins respectively. Newly synthesized envelope proteins interact with mature HBV nucleocapsids at the ER before secretion from the hepatocyte to form the competent virions (42 nm diameter, Dane particles). However, HBsAg production far exceeds that required for virion assembly, and excess surface envelope proteins are covalently linked by intermolecular disulphide bonds and secreted as non-infectious filamentous (20 nm diameter) or spherical (20–22 nm diameter) subviral empty particles (Fig. 1) (7). These subviral particles may play a role in evading the host immune response (8), and may also co-exist with anti-HBs as part of circulating immune complexes (9).
The small (S) protein (226 amino acids) is expressed at the highest levels, predominates in both virions and subviral particles and is secreted without cleavage of amino acid residues during translocation because of its self-assembling capacity with host-derived lipids in the cell ER (10). The middle (M) protein (containing 55 extra residues of the pre-S2 domain) is regulated by the same promoter and is similarly secreted, while the transcription of the large protein (L) is regulated by a specific but weaker promoter (pre-S1) (11). HBV large surface protein (L-HBs) containing both the pre-S2 region and the 108–119 additional residues of the pre-S1 domain, is an essential component of both virions and filaments, and presents 10–20% of their envelope proteins. In contrast, the L-HBs represent only 2% of the 22 nm spherical particles (12, 13).
The complexity of HBsAg production and secretion has been known since early studies showed a greater excess of both filaments and spherical subviral particles in highly viraemic HBeAg-positive carriers compared with low viraemic anti-HBe-positive carriers, in whom the decline of filaments paralleled that of virions while a moderate excess of spherical particles remained (14, 15). Thus, subviral HBsAg particles exceed virions by 102–105 and can accumulate in concentrations of up to several hundred micrograms per millilitre of serum (16).
Serum HBsAg is a result of combinations of these three proteins. It is important to understand that although HBsAg quantification detects all three forms of systemic HBsAg (part of HBV virion, spherical, filamentous), differentiation between the relative proportions is not routine. HBsAg may also be produced from HBV DNA integrated into the host genome. Although viral integration is an essential component of the life cycle of retroviruses such as HIV, it is not required for normal productive hepadnaviral infection. Instead integration of HBV DNA occurs illegitimately through recombination mechanisms using host enzymes from double-stranded linear HBV DNA (Fig. 1) (17, 18). In HBV infection, viral integration seems to occur in the early stages of infection. Although HBV integration is believed to be a random event, a strong preference for integration occurs at the DR1 and DR2 sequences on the HBV genome (19). Integrated sequences cannot provide a template for productive viral replication as a complete genome is not present (20). However, because sequences of the S genes of the enhancer I elements are often present in integrated segments, HBsAg may be produced (20).
HBsAg: clinical relevance
Chronic hepatitis B is a major cause of liver disease worldwide, and a significant cause of cirrhosis and hepatocellular carcinoma (HCC). Several epidemiological studies have shown that seropositivity for HBsAg is an important risk factor for liver disease, including cirrhosis and HCC (21–24).
The natural course of CHB has three chronological phases: an initial immune tolerance phase when patients are positive for HBeAg and have normal alanine aminotransferase (ALT) levels, followed by an immune-clearance phase in which HBeAg-positive patients have elevated ALT levels, and finally HBeAg seroconversion to its antibody with ALT normalization, which marks the transition to the inactive residual phase (25). While up to 25% of these so-called inactive HBsAg carriers may encounter HBV reactivation and develop HBeAg-negative chronic hepatitis, most inactive carriers remain stably inactive throughout their lifetime (26, 27), and some of them eventually clear HBsAg from serum (25, 28).
Spontaneous HBsAg seroclearance, defined as the loss of serum HBsAg on two occasions at least 6 months apart and until the final follow up visit (28), is rare in the natural history of chronic HBV infection. The incidence of spontaneous HBsAg seroclearance varies considerably in different series with an annual incidence of 1–2%, and does not differ much between Caucasian and Asian carriers (28–31). A long-term (>15 years) inactive carrier state is usually required to achieve subsequent HBsAg seroclearance. Older age, HBeAg seronegativity, clinical remission and cirrhosis are factors for HBsAg seroclearance. Age of the patient or duration of chronic HBV infection appears to be the most significant and constant determining factor associated with spontaneous HBsAg seroclearance (4). In contrast, treatment of CHB with interferon has been found to enhance HBsAg seroclearance by approximately three-fold in Western studies and six-fold in Asian studies (32–35).
HBsAg seroclearance is almost always associated with the loss of all serum markers of HBV replication, including HBV DNA by polymerase chain reaction (PCR) and appearance of anti-HBs over time. In general, up to 80% of subjects had detectable anti-HBs after HBsAg seroclearance (4, 28, 31, 36, 37). While only 17% of the carriers had detectable anti-HBs within 1 year after HBsAg seroclearance (36), the proportion with positive anti-HBs increased to 56% at 5 years and 76% at 10 years after HBsAg seroclearance (37). At the time of HBsAg seroclearance, nearly all carriers became HBV DNA negative by hybridization assays, but 30–70% of them were still HBV DNA positive by PCR-based assays (38). The proportion of carriers with detectable HBV DNA by PCR-based assays then decreased to 10–20% between 5 and 10 years and was <10% 10 years after HBsAg seroclearance (36).
In addition to the significant decrease or loss of all HBV replication in the serum and the development of anti-HBs over time, the long-term outcome after HBsAg seroclearance is excellent if there is no pre-existing cirrhosis or viral superinfection. This is supported by natural history studies showing increased survival, lower rates of hepatic decompensation and reduction in the frequency of HCC in patients who cleared HBsAg (30–35). In carriers with cirrhosis at HBsAg seroclearance and with no evidence of viral superinfection, liver function can improve or remain stable and hepatic decompensation rarely occurs. However, the incidence of HCC varied significantly in the different reported series (36, 39, 40). Therefore, HBsAg should probably be tested every 2–3 years to detect HBsAg seroclearance in HBeAg-negative HBsAg carriers with sustained remission, especially in those with cirrhosis. In patients without cirrhosis and no evidence of HCV superinfection, follow-up can be annual after HBsAg seroclearance (4).
Most patients with histological assessment show mild necroinflammation and no significant fibrosis. Immunostaining for HBsAg and HBcAg was negative in all patients (28, 38). However, all tested patients still harboured transcriptionally inactive phase HBV at very low replicative levels in the liver, mainly in the form of cccDNA up to 4 years after HBsAg seroclearance (28, 38). This so-called “occult HBV infection” could explain reactivation of hepatitis B when patients who have lost HBsAg receive immunosuppression or chemotherapy (41).
The loss of HBsAg and the development of anti-HBs antibodies (HBsAg seroconversion) are the ultimate goals of anti-HBV therapy (42). Thus quantitative HBsAg can be a promising prognostic marker of the natural history of HBV infection as well as during antiviral therapy. The first quantitative assays measuring HBsAg levels using enhanced chemiluminescence were introduced over 20 years ago (43, 44) but were limited by a lack of appropriate standardization. Initially, HBsAg levels were expressed as serial dilutions of a reference sample from the Paul Ehrlich Institute (Langen, Germany) (44). Recently, new quantitative HBsAg assays have been developed that fulfill the prerequisites of a biomarker: reproducibility, automated quantification with high-throughput platforms, relatively low cost (<10% of the cost of a serum HBV DNA assay) and standardization (IU/ml) (45, 46).
Indeed, the most recent evidence supports the role of HBsAg as a predictive marker for an anti-HBV treatment response (47–53). In a cohort of 386 HBeAg(−) hepatitis B patients, Brunetto et al. (47) showed that on-treatment HBsAg reduction of >1 log10 IU/ml and levels below 10 IU/ml at the end of treatment with pegylated interferon (PEG-IFN)-α2a were strongly associated with sustained HBsAg clearance 3 years after treatment cessation. In another study of 48 patients with HBeAg-negative CHB treated with PEG-IFN for 48 weeks, an early serological response defined as early HBsAg decline during treatment, was highly predictive of sustained virological response (SVR), defined as undetectable HBV DNA by PCR 24 weeks after treatment (48). The decrease of at least 0.5 log10 and 1 log10 IU/ml in serum HBsAg concentrations at weeks 12 and 24 of treatment, respectively, had negative predictive values of 92 and 97% for SVR. Of note, the kinetics of the decline in HBV DNA throughout treatment were nearly identical for sustained responders and relapsers but were quite different when HBsAg was assessed, suggesting that measurement of HBsAg concentrations may be more reliable to identify patients who will achieve SVR. In a recent study (49) evaluating 102 HBeAg-negative patients treated with PEG-IFN, a solid stopping rule was established at week 12 of treatment with combined declines in serum HBV DNA and HBsAg levels from baseline. Combining HBsAg and HBV DNA declines was the best predictor of a sustained response defined as HBV DNA <10 000 copies/ml 24 weeks after treatment (areas under the curve at week 12=0.74). None of the 20 patients (20% of the study population) whose serum HBsAg levels did not decrease and whose HBV DNA levels decreased <2 log copies/ml achieved a sustained response (negative predictive value=100%).
A recent study (51) in HBeAg (+) patients suggests that a fixed HBsAg level of 1500 IU/ml after 12 weeks of therapy had a better predictive value for treatment response than a logarithmic decline. Half the patients who achieved a decline of <1500 IU/ml after 12 weeks of PEG-IFN therapy achieved HBeAg seroconversion compared with 16% in patients with HBsAg levels above 20 000 IU/ml after 12 weeks. The quantification of HBsAg during long-term treatment with nucleos(t)ide analogues (NUCs) might be useful for monitoring treatment responses to predict HBsAg loss in patients who achieve HBV DNA suppression (54).
Before a specific HBsAg level or a magnitude of change in HBsAg levels can be recommended to predict treatment outcomes, the meaning of these parameters must be understood in the treatment-free setting. Interestingly, two recent studies (55, 56) evaluated the correlation between HBsAg serum levels and the clinical and virological features of chronic HBV carriers at different phases of infection. Overall, 434 chronic carriers were studied: 62 immune-tolerant carriers (IT), 103 HBeAg-positive patients in the immune-clearance phase (IC), 118 HBeAg-negative carriers in the non-/low replicative phase (LC) and 151 patients with HBeAg-negative hepatitis. Two major findings were common to both studies: (i) median HBsAg levels differ significantly during the four phases of HBV infection and decline progressively from IT (4.5–4.96 log10 IU/ml in Asian and European carriers) to LC (2.86–3.09 log10 IU/ml in Asian and European carriers); (ii) HBsAg/HBV DNA ratios are significantly higher in LC (1.05 Asian–1.17 European) compared to other patients (ratio range 0.55–0.64). This suggests that HBsAg secretion is highly dynamic and varies throughout chronic HBV infection both quantitatively and qualitatively.
The substantial variations in total serum HBsAg during the different phases of HBV infection suggest that quantitative HBsAg might be a new diagnostic tool to characterise the HBV carriers in combination with HBV DNA. These two HBV markers provide complementary information on the status of HBV infection and may be very useful in clinical practice to define individual HBV carriers during the highly dynamic phases of chronic HBV infection. This is particularly important to avoid misclassifying an asymptomatic HBeAg-negative CHB patient as an inactive carrier because of a single point serum test with normal transaminases and negative HBV DNA caused by the typical intermittent disease profile of HBeAg-negative CHB. In this respect, in a recent study, a single-point quantification combining HBsAg (<1000 IU/ml) and HBV DNA (≤2000 IU/ml) provided the most accurate identification on inactive carriers (94.3% diagnostic accuracy, 91.1% sensitivity, 95.4% specificity, 87.9% positive predictive value, 96.7% negative predictive value), which was similar to that of long-term regular monitoring (57).
Although HBsAg quantification is a useful predictor of treatment response, its clinical significance has not been completely elucidated. Previous reports have suggested that HBsAg may reflect the content of intrahepatic HBV DNA and/or cccDNA, which serve as an intracellular template for viral RNA transcription, and it has been proposed as a surrogate marker for HBV-infected hepatocytes (58–60). However, other studies using the same HBsAg quantification platform could not confirm the association between HBsAg levels and cccDNA or intrahepatic HBV DNA in HBeAg (−) hepatitis B (61–63).
Most studies of HBsAg and HBV cccDNA have included small cohorts and were limited by the sensitivity of available assays for serum HBV DNA. In addition, studies examining HBsAg quantification did not compare HBeAg-positive and HBeAg-negative populations, which are characterized by distinct immunological milieus and different levels of virion production (60, 64). The correlation between quantitative HBsAg titres and serum and intrahepatic markers of HBV replication differed between patients with HBeAg-positive and HBeAg-negative CHB in a recent study (65). In HBeAg-positive CHB, HBsAg was positively correlated with serum HBV DNA and intrahepatic cccDNA and total HBV DNA (r=0.69, 0.71, 0.76, P<0.01). HBeAg correlated with serum HBV DNA (r=0.60, P<0.0001), although emerging basal core promoter/precore variants reduced HBeAg titres independent of viral replication. In HBeAg-negative CHB, HBsAg correlated poorly with serum HBV DNA (r=0.28, P=0.01) and did not correlate with intrahepatic cccDNA or total HBV DNA. For the first time, quantitative IHC studies have shown that serum HBsAg titres were correlated with the number and intensity of HBsAg-positive cells. However, the number and intensity of hepatocytes staining positive for HBsAg, as well as the pattern of cell distribution, were only related to HBV replication in patients with HBeAg-positive CHB, confirming that the association between HBsAg production and HBV replication breaks down in the HBeAg-negative phase of the disease. This might occur if HBsAg was produced from a source other than intranuclear cccDNA (Fig. 1). In HBV infection, viral integration into the host genome has been shown to begin early in the infection (17, 18). The observation that the number and intensity of HBsAg-positive cells in the liver may be preserved despite low-level HBV replication is consistent with this integration. Although integration is believed to be a random event, a high preference for integration has been observed at the DR1 and DR2 sequences on the HBV genome (19), and sequences of the S genes are often present in integrated segments (20). Although integrated sequences cannot provide a template for productive viral replication, HBsAg may be produced (20). Progressive integration might therefore provide a template for persistent HBsAg production independent of viral load. To date, no method for reliably differentiating truncated HBs peptides in serum has been developed to test this hypothesis. Another explanation might involve preferential control of the viral replication pathway at the post-transcriptional level, sparing HBV cccDNA and HBsAg transcription/secretion. This has been shown to occur in vitro in the setting of cytokine effects (targeting the encapsidation step) (66), and is consistent with the more profound immune pressure that is present in the HBeAg-negative phase of disease. These two hypotheses are not mutually exclusive.
Clinical interpretation of HBsAg and HBeAg titres might therefore be refined by considering the phase of the disease as well as the quasispecies. The variations in predictive HBsAg-values and the discrepancy in studies analysing associations of HBsAg and cccDNA or serum HBV DNA may be because of the highly complex and dynamic nature of HBV infection, as well as the influence of HBV genotypes on HBsAg levels (47, 67). Serum HBsAg levels are thought to depend mainly on the translation of specific messenger RNAs for the ‘S’ gene generated from cccDNA and also from integrated HBV DNA within the host genome. Finally, the decrease in HBsAg levels during transition from the active to the inactive phase of HBV infection suggests that HBsAg levels may reflect ‘transcriptionally’ active cccDNA rather than its absolute amount or HBV DNA integrated sequences, which are thought to increase over time during chronic infection (68).
In conclusion, HBsAg serum levels could be the result of the complex balance between the virus and the host's immune system as well as the product of the transcription of specific mRNAs rather than viral replication. The substantial variations in total serum HBsAg during the different phases of HBV infection suggest that quantitative HBsAg could be a new diagnostic tool for characterization of the HBV carrier in combination with HBV DNA.
Conflicts of interest
Rami Moucari has none to declare. Patrick Marcellin is a consultant for, and is on the speakers' bureau of Roche, Schering-Plough, Gilead, Bristol-Myers Squibb, GlaxoSmithKline and Idenix-Novartis. He is a consultant for and advises Vertex, Valeant, Human Genome Sciences, Cythesis, Intermune, Wyeth and Tibotec. He also advises Coley Pharma.