Hepatitis B surface antigen quantification: Not what it seems on the surface


  • Potential conflict of interest: Dr. Locarnini consults for and received fees from Bristol-Myers Squibb. He received royalties from and holds intellectual property rights with Melbourne Health. He consults for Gilead. He received fees from Merck.

See Article on Page 434

The World Health Organization (WHO) defines chronic hepatitis B virus (HBV) infection as persistent hepatitis B surface antigen (HBsAg) positivity in blood for at least 6 months and acute HBV infection as the transient presence of hepatitis B surface antigenemia.1 The HBsAg can be found in the blood of infected individuals in a number of different particulate forms: complete virions of ∼42 nm in diameter2 and subviral particles that are either spherical and 20-22 nm in diameter or filamentous forms of various lengths with a width of ∼20 nm.2 These noninfectious subviral particles are typically produced in excess over the infectious virions by several orders of magnitude and can accumulate in blood to concentrations ranging from 50-300 μg/mL.3 The HBsAg found in serum is in fact a mosaic of viral envelope proteins and can contain all three forms of the surface proteins of the mature HBV virion, the large (L), medium (M), and small (S) proteins. These proteins are encoded by the S-open reading frame (ORF) which contains three in-frame “start” codons divided into Pre-S1, Pre-S2, and S domains (Fig. 1A). The proteins are translated from two subgenomic messenger RNA (mRNA) transcripts, the longer Pre-S1 mRNA encodes the L protein, whereas the Pre-S2/S mRNA encodes the M (Pre-S2) and S (S) proteins from separate initiation codons (AUG in Fig. 1A). The S protein (226 amino acids in length) is expressed in the highest amount and is the main protein present in both virions and subviral particles. The M protein contains an extra 55 amino acid extension at the amino-terminus of the S protein, whereas the L protein includes both the Pre-S2 and S regions and has an additional 108-119 amino acids (depending on the HBV genotype) domain at the amino-terminus of the Pre-S2 protein (Fig. 1A). The production of the S and M protein is regulated from a strong TATA-less, nonliver-specific promoter (Sp), whereas transcription of the L protein is controlled by a liver-specific yet weaker Pre-S1 promoter (Pre-S1p).4 It is likely that these differences in the various strengths of the promoters help explain why the L HBsAg represents only 2% of the protein in the 22-nm spherical subviral particles.5 The M protein, along with the L protein, are found in higher proportions as components of virions and the filamentous subviral particles. However, they still only represent ∼20% of the total envelope protein in those structures, which remain predominately composed of S HBsAg.5

Figure 1.

(A) Diagrammatic representation of the HBV envelope proteins L (large), M (middle), and S (small). The HBV DNA genome is shown in the top left-hand corner, transcription of which results in the Pre-S1 RNA from the Pre-S1 promoter (Pre-S1p), whereas the Pre-S2/S RNA is made from the S promoter (Sp). The three translation initiation codons (AUG) are shown as well as the glycosylation sites (CHO). (B) Pre-S2 deletion mutations can be quite heterogeneous23-26 and one common variant is shown here that results in complete deletion of the Pre-S2 sequence (red circle), so that translation of the Pre-S1 RNA results in production of a truncated L protein. This Pre-S2 deletion can also be extensive enough to affect the S promoter (Sp), resulting in significantly reduced translation of the S protein as well.


CHB, chronic hepatitis B; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; NA, nucleos(t)ide analog; ORF, open reading frame; SC, seroconversion; Sp, specific promoter; WHO, World Health Organization.

Although viral integration is not required for normal productive hepadnaviral infection,4 integration of HBV DNA occurs illegitimately through recombination mechanisms using host enzymes acting on the double-stranded linear DNA form of HBV.6 In HBV infection, viral integration seems to occur early but the integrated sequences cannot provide a template for productive viral replication, as a complete genome is lacking. However, given that sequences of the S-ORF with enhancer I elements are often present in integrated segments,4 the HBsAg may be produced, often as truncated envelope proteins.

Quantitative testing for serum HBsAg was developed over two decades ago but the lack of standardization restricted its use to the research setting.7 From a diagnostic perspective, HBsAg quantification assays target all forms of circulating HBsAg because the antibodies used in the quantitative enzyme immunoassays identify epitopes in the S protein, and are not capable of distinguishing between virion-associated HBsAg and subviral particles or HBsAg produced from integrated sequences. Currently, there are two commercialized assays that can quantify HBsAg, the Architect QT assay (Abbott Laboratories) and the Elecsys HBsAg II Quant assay (Roche Diagnostic). Both assays express results in international units per mL (IU/mL), based on the WHO reference standard, with 1 IU/mL being equivalent to ≈1-10 ng/mL of HBsAg,8 which is in turn equivalent to 2 × 108 subviral particles of HBsAg or 5 × 107 virions. There is good correlation between the HBsAg measurements by these two assays9 and their wider availability has resulted in a surge of research and publication activity redefining the natural history of chronic hepatitis B (CHB) and effects of antiviral therapy on HBsAg levels. As well as this clinical interest, there has been some enthusiasm for HBsAg quantification in the basic sciences arena as well, with some showing a correlation with the level of intrahepatic HBV covalently closed circular DNA (cccDNA), the template for viral replication inside the nuclei of hepatocytes,10 implying that HBsAg levels may be a surrogate marker of infected cells in the liver. This has led to many studies addressing the use of HBsAg quantification to monitor the natural history and predict treatment responses in CHB.

Almost all natural history studies have consistently shown that the HBsAg level is highest in the immune tolerant phase (4.5-5.0 log10 IU/mL), then declining in the immune clearance phase (3.0-4.5 log10 IU/mL) and decreasing slowly and progressively after HBeAg seroconversion. The HBsAg level is lowest (1.5-3.0 log10 IU/mL) in those individuals who maintain persistently normal serum alanine aminotransferase (ALT; low replicative phase) but HBsAg can be significantly higher (2.5-4.0 log10 IU/mL) in those patients who develop HBeAg-negative CHB.11-13 Longitudinal studies have further shown that HBsAg levels remained stable in HBeAg-positive patients and tended to reduce slowly in HBeAg-negative patients and it has been proposed that a reduction of HBsAg of ≥1.0 log10 IU/mL might reflect improved immune control.11 Several studies have further examined whether the relationship between the HBsAg level and HBV DNA load at a single timepoint would predict HBsAg loss and allow an accurate identification of “true inactive carriers.” It seems that HBsAg <1,000 IU/mL and HBV DNA of <2,000 IU/mL may be sufficient to identify all inactive carriers with HBV genotype D infection,14 and HBsAg <100 IU/mL can predict HBsAg loss over time in genotype B or C HBV-infected patients.15

The goals of antiviral therapy for CHB include the suppression of viral replication to a level that will lead to biochemical remission, histological improvement, and prevention of disease progression.16 Unfortunately, the most potent nucleos(t)ide analog (NA) therapy minimally impacts the HBsAg levels in blood. This is not surprising because NAs block the HBV DNA replication pathway (by way of inhibiting reverse transcription) and have no direct effect on transcription or translation of the Pre-S1/Pre-S2/S, Pre-core, and X pathways.17 In contrast, immune-based therapies such as interferon-α do block these nonreverse transcriptase-dependent replication pathways, as part of their broader antiviral activity. HBsAg seroconversion (SC) is now being considered the new treatment endpoint because HBsAg loss has been associated with successful immunological control of HBV and durable suppression of viral replication, thus suggesting the question: Can patients actually stop NA therapy?

Assessment of on-treatment changes in HBsAg titer has provided new management algorithms to achieve this new endpoint, especially for patients on pegylated interferon-α therapy.18-20 In HBeAg-positive patients, reduction of HBsAg levels to <1,500 IU/mL at week 12 has been shown to be an early favorable sign of subsequent HBsAg SC. More than 50% of patients on pegylated interferon who achieved this level at week 12 have HBeAg SC 6 months posttreatment, and nearly 20% of these patients achieved subsequent HBsAg clearance at 6 months posttreatment. In contrast, an HBsAg level of >20,000 IU/mL at week 12 was associated with a very low rate of HBeAg SC and so may become a potential stopping rule.20, 21 In HBeAg-negative patients treated with pegylated interferon, the decline in HBsAg titer at week 12 has also been shown to be a useful predictor of achieving an undetectable viral load at 24 weeks posttherapy.19 Among patients who achieved HBsAg decline ≥10% from baseline at week 12 of treatment, almost 50% achieved HBV DNA ≤2,000 IU/mL at 1 year posttreatment, and 40% of these individuals achieved HBsAg clearance at 5 years posttreatment. Rijckborst et al.22 proposed a clinically useful algorithm in HBeAg-negative CHB that any HBsAg decline at week 12 with a 2 log10 drop or more in HBV DNA could predict almost 40% of sustained responders in their cohort. Patients not achieving an HBsAg decline or only having a <2 log drop in HBV DNA did not respond. The data for patients on long-term oral NA therapy is not as robust but several Phase 4 studies are in progress attempting to address this matter.

In this issue of HEPATOLOGY an elegant study from the laboratory of Teresa Pollicino and Giovanni Raimondo23 provides a cautionary note to this recent burst of interest in quantitative HBsAg testing. They report on the finding that Pre-S/S HBV variants, which are commonly found in patients with CHB, can influence the levels of circulating HBsAg without significantly impacting serum HBV DNA load. The reduced level of HBsAg found is not due to modification of the circulating HBsAg protein, as the assays are designed to detect the epitopes in the conserved S protein, but rather due to alterations in the intracellular pathway involved in the synthesis of the envelope proteins. The Pre-S/S HBV variants can be found in up to 30% of patients, and this is not surprising because the viral life cycle of HBV employs an error-prone reverse transcription step, and particular selection pressures such as attempted host immune clearance readily select out escape viral variants (Fig. 1B). Pre S/S HBV variants have been associated with the most progressive forms of liver disease, including hepatocellular carcinoma (HCC),24, 25 possibly because the specific mutations associated with them can cause imbalances in the synthesis of the surface proteins with a decrease in S protein production and secretion (Fig. 1B). Furthermore, changes in the S/L protein ratio can lead to their intracellular retention26 but without necessarily affecting virion egress.27

One of the great attractions to the use of quantitative HBsAg testing in clinical algorithms is that the assays are relatively inexpensive, suitable for high-throughput screening, and the platforms used are common to many laboratories. Nearly all studies investigating the possible role of HBsAg quantification in clinical management have been retrospective. Of note is that in many of these, HBsAg thresholds at baseline or on treatment, or the kinetics of HBsAg decline on treatment, have allowed some potentially useful algorithms to be elucidated. Nevertheless, the association of HBsAg titer with treatment outcome has often only provided intermediate positive and/or negative predictive values. Could it be that the presence of Pre-S/S variants is one of the contributing factors to the low predictive values seen? The findings of Pollicino et al. may have practical repercussions for the use of quantitative HBsAg testing. The presence of surface variants can be determined by conventional population-based sequencing and this could be assessed at baseline. This may be another step on the road to tailored therapy. However, in countries with limited resources this may be an obstacle to the unconditional use of HBsAg quantification as a primary biomarker for staging and then monitoring patients on treatment.