Serum hepatitis B surface antigen and hepatitis B e antigen titers: Disease phase influences...

Serum Hepatitis B Surface Antigen and HepatitisB e Antigen Titers: Disease Phase InfluencesCorrelation with Viral Load and Intrahepatic

Hepatitis B Virus MarkersAlexander J.V. Thompson,1,2 Tin Nguyen,1,2 David Iser,1,2,3 Anna Ayres,1 Kathy Jackson,1 Margaret Littlejohn,1

John Slavin,2 Scott Bowden,1 Edward J. Gane,4 William Abbott,4 George K.K. Lau,5 Sharon R. Lewin,3,6,7

Kumar Visvanathan,6,8 Paul V. Desmond,2 and Stephen A. Locarnini1

Although threshold levels for hepatitis B surface antigen (HBsAg) and hepatitis B e anti-gen (HBeAg) titers have recently been proposed to guide therapy for chronic hepatitis B(CHB), their relationship to circulating hepatitis B virus (HBV) DNA and intrahepaticHBV replicative intermediates, and the significance of emerging viral variants, remainsunclear. We therefore tested the hypothesis that HBsAg and HBeAg titers may vary inde-pendently of viral replication in vivo. In all, 149 treatment-naı̈ve CHB patients wererecruited (HBeAg-positive, n 5 71; HBeAg-negative, n 5 78). Quantification of HBeAgand HBsAg was performed by enzyme immunoassay. Virological characterization includedserum HBV DNA load, HBV genotype, basal core promoter (BCP)/precore (PC) sequence,and, in a subset (n 5 44), measurement of intrahepatic covalently closed circular DNA(cccDNA) and total HBV DNA, as well as quantitative immunohistochemical (IHC) stain-ing for HBsAg. In HBeAg-positive CHB, HBsAg was positively correlated with serumHBV DNA and intrahepatic cccDNA and total HBV DNA (r 5 0.69, 0.71, 0.76, P <0.01). HBeAg correlated with serum HBV DNA (r 5 0.60, P < 0.0001), although emerg-ing BCP/PC variants reduced HBeAg titer independent of viral replication. In HBeAg-negative CHB, HBsAg correlated poorly with serum HBV DNA (r 5 0.28, P 5 0.01) anddid not correlate with intrahepatic cccDNA nor total HBV DNA. Quantitative IHC for he-patocyte HBsAg confirmed a relationship with viral replication only in HBeAg-positivepatients. Conclusion: The correlation between quantitative HBsAg titer and serum and in-trahepatic markers of HBV replication differs between patients with HBeAg-positive andHBeAg-negative CHB. HBeAg titers may fall independent of viral replication as HBeAg-defective variants emerge prior to HBeAg seroconversion. These findings provide newinsights into viral pathogenesis and have practical implications for the use of quantitativeserology as a clinical biomarker. (HEPATOLOGY 2010;51:1933-1944)

Abbreviations: ALT, alanine aminotransferase; BCP, basal core promoter; cccDNA, covalently closed circular DNA; CHB, chronic hepatitis B; EIA, enzymeimmunoassay; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; IH, intrahepatic; NA, and nucleos(t)ide analogs; PC,precore; PE, Paul Ehrlich; PEGIFN, pegylated-interferon; WT, wildtype.From the 1Department of Research and Molecular Development, Victorian Infectious Diseases Reference Laboratory, North Melbourne, Victoria, Australia;

2Department of Pathology, St. Vincent’s Hospital, Melbourne, Australia; 3Infectious Diseases Unit, Alfred Hospital, Melbourne, Victoria, Australia; 4New Zealand LiverUnit, Auckland City Hospital, Auckland, New Zealand; 5Queen Mary Hospital, University of Hong Kong, Hong Kong Special Administrative Region of China;6Monash University, Clayton, Victoria, Australia; 7Burnet Institute, Melbourne, Australia; and 8Infectious Diseases Department, Monash Medical Centre, Clayton,Victoria, Australia.Received May 12, 2009; accepted January 19, 2010.Additional Supporting Information may be found in the online version of this article.Supported by a Postgraduate Scholarship from the National Health and Medical Research Council (NHMRC) of Australia (to A.J.V.T).Address reprint requests to: Professor Stephen A. Locarnini, Professor?Victorian Infectious Diseases Reference Laboratory (VIDRL), 10 Wreckyn St, North

Melbourne, Victoria, 3051, Australia. E-mail: [email protected]; fax: þ61 3 9342 2666.Copyright VC 2010 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.23571Potential conflict of interest: Abbott Laboratories provided an ARCHITECT machine to VIDRL. Stephen Locarnini has consulted for Abbott Laboratories.

Dr. Lau is a consultant for Roche and Novartis. Dr. Locarnini is a consultant for and advises Abbott.

1933

Recent developments in the treatment ofchronic hepatitis B (CHB) have led to inter-est in novel biomarkers for predicting treat-

ment response. Quantitative serology for hepatitis Bsurface antigen (HBsAg) and hepatitis B e antigen(HBeAg) are two promising candidates. Preliminarydata have supported the clinical utility in treatmentprotocols for pegylated-interferon a (PEGIFN) andnucleos(t)ide analogs (NA), respectively.1,2

HBsAg seroconversion is the ultimate laboratorymarker of successful therapy for patients with CHB.Data from clinical trials suggest a role for quantifica-tion of HBsAg at baseline and/or on-treatment inidentifying patients most likely to achieve HBsAg sero-conversion.3,4 Mathematical modeling of HBsAgdecline on treatment has also been used to predict theduration of therapy required for HBsAg seroconversionfor both IFN- and NA-based therapies.5 There are anumber of assays that have been developed to measurequantitative HBsAg titer.5-8 Importantly, these tests areinexpensive (<10% of the cost of a serum hepatitis Bvirus [HBV] DNA assay) and use high-throughputplatforms.HBeAg seroconversion is an established therapeutic

endpoint for the management of HBeAg-positiveCHB and is associated with reduced morbidity andmortality.9 Quantification of HBeAg has recently beenevaluated as a biomarker during PEGIFN therapy.1 Ina large study of patients with HBeAg-positive CHB, abaseline HBeAg titer of �31 Paul Ehrlich (PE) IU/mLwas associated with a positive predictive value of 54%for HBeAg seroconversion following 48 weeks of ther-apy. Furthermore, a failure of HBeAg titer to declineto <100 PE IU/mL after 24 weeks of therapy wasassociated with a negative predictive value for HBeAgseroconversion of 96%, greater than that obtainedfrom monitoring levels of serum HBV DNA. Quanti-tative enzyme immunoassays (EIAs) for HBeAg arenot widely commercially licensed, and this has contrib-uted to these assays remaining research tools.HBeAg and HBsAg titer have both been proposed

as surrogates for infected liver cell mass, or HBV cova-lently closed circular DNA (cccDNA), the hepatocytenuclear reservoir that is responsible for viral persist-ence.10 This concept underpins their use as bio-markers. If the hepatocyte is considered in isolation,HBsAg, HBeAg, and serum HBV DNA would beexpected to directly correlate with each other, and withliver cccDNA, as all are translated from separate tran-scripts (Pre-S1, Pre-S2/S, precore/core, and prege-nomic messenger RNA [mRNA], respectively) directlyderived from cccDNA. However, the published data

that describes these relationships are limited and con-flicting. Positive correlations have been observedbetween HBeAg or HBsAg titer and serum HBVDNA in some studies,6 but not all.2,5,11 Similarly, thepositive correlation between HBsAg titer and HBVcccDNA/total intrahepatic (IH) HBV DNA that haspreviously been described,8,12 was not seen in tworecent studies.13,14 Importantly, most of the studies ofHBsAg and HBV cccDNA involved small cohorts andwere limited by the sensitivity of the available assaysfor serum HBV DNA. In addition, studies examiningHBsAg quantification have not compared HBeAg-posi-tive and HBeAg-negative populations, known to becharacterized by distinct immunological milieu anddifferent levels of virion productivity.12,15 Studies ofHBeAg have not previously considered the emergenceof A1726T/G1764A basal core promoter (BCP) orG1896A precore (PC) variants coexisting with wild-type (WT) HBV in the viral quasispecies pool. Suchvariants are known to emerge in the years leading upto HBeAg seroconversion.16 Although the BCP andthe PC variants have been shown to reduce or abolishHBeAg production, respectively, in vitro,17 theirimpact on HBeAg titer in vivo, in the context of theviral quasispecies pool and an active immune system,has not been defined. The relationship betweenHBeAg titer and IH markers of HBV has not beeninvestigated.Here we provide the first detailed and comprehen-

sive analysis of serum HBsAg and HBeAg titers withina treatment-naı̈ve patient cohort, with a focus on theirrelationship to serum and liver markers of HBV repli-cation, as well as determining the significance ofemerging viral variants. We tested the hypothesis thatHBsAg and HBeAg titers may vary independently ofviral replication in vivo.

Patients and MethodsPatientsThe patient cohort was collected from three separate

clinical sites: 1) St. Vincent’s Hospital, Melbourne,Australia (n ¼ 110)18; 2) Auckland City Hospital,Auckland, New Zealand (n ¼ 15); and 3) QueenMary Hospital, University of Hong Kong, Hong KongSpecial Administrative Region of China (n ¼ 30). Allpatients were positive for HBsAg for at least 6 monthsprior to enrolment and had detectable HBV DNA inthe serum. All HBeAg-positive patients had docu-mented elevation of serum alanine aminotransferase(ALT) for 3-6 months prior to liver biopsy (ALT >30

1934 THOMPSON ET AL. HEPATOLOGY, June 2010

IU/mL for men, and >19 IU/mL for women19). Nopatients had evidence of decompensated liver diseaseor hepatocellular carcinoma. All patients tested nega-tive for hepatitis C virus (HCV), hepatitis D virus,and human immunodeficiency virus (HIV); markersfor coexistent autoimmune or metabolic liver diseasewere negative. Liver biopsy for routine histology wasavailable in 73 patients from the Australian center; allresults were reported using the METAVIR scoring sys-tem.20 Fresh-frozen liver tissue was collected for analy-sis of intrahepatic HBV replicative intermediates in asubset of patients from the Australian site (n ¼ 44/73)(see below). Serum samples for biochemistry and HBVvirological studies were collected on the same day asliver biopsy. Institutional ethics approval for theresearch protocol was obtained and all patients pro-vided informed consent according to the Declarationof Helsinki.

Quantitative HBsAg AssaySerum HBsAg titer was measured by EIA using the

ARCHITECT platform (Abbott Laboratories, Chi-cago, IL), as per the manufacturer’s instructions, which

expresses HBsAg against an internal World Health Or-ganization reference standard in IU/mL.

Quantitative HBeAg AssayThe development of the quantitative HBeAg assay

was based on the method of Perrillo et al.21 SerumHBeAg was measured by EIA (ARCHITECT plat-form; Abbott Laboratories). Serial dilutions of the PEHBeAg reference standard (Paul-Ehrlich Institute, Lan-gen, Germany) were used to define the linear range ofthe assay and generate a reference curve for linearregression (described in detail in Supporting Methods).

HBV Viral Load AnalysisPatients recruited from St. Vincent’s Hospital and

Auckland Hospital had HBV viral load testing per-formed using the Versant HBV DNA 3.0 assay(bDNA) (Siemens Healthcare Diagnostics, Tarrytown,NY) according to the manufacturer’s instructions.Patients recruited from the Queen Mary Hospital inHong Kong were tested using the Digene Hybrid cap-ture II assay with the ultrasensitive detection protocol(Digene Diagnostics, Beltsville, MD). The lower limitof detection (LLD) for both assays is �102-3 IU/mL.

Table 1. Baseline Patient Data

Overall HBeAg1 HBeAg2 P Value

N 149 71 78

Age (years) Median 40 33 48 <.0001

25th-75th centile 28-48 26-40 37-52

Asian N 133 63 70 0.8422

% 89% 89% 90%

M N 96 42 54 0.1994

% 64% 59% 69%

ALT (IU/mL) Median 65 107 45 <.0001

25th-75th centile 38-124 65-178 28-71

Viral load Median 6.2 7.3 4.6 <.0001

(log10 IU/mL) 25th-75th centile 4.4-7.5 6.8-8.3 3.2-5.8

Viral load <2000 IU/mL N 21 0 21 <.0001

% 14% 0% 27%

METAVIR F3-4 N 13/73 2/22 11/51 0.32

% 18% 9% 22%

Genotype

A 1% 2% 1% 1.0000

B 47% 35% 58% 0.0073

C 40% 58% 24% <.0001

D 11% 6% 15% 0.1019

G 1% 0% 1% 1.0000

Intrahepatic markers (n ¼ 44) (n ¼ 17) (n ¼ 27)

cccDNA (c/Geq) Median 0.43 3.65 0.2 <0.0001

25th-75th centile 0.15-2.30 1.43-8.55 0.10-0.43

Total IH HBV DNA (c/Geq) Median 3.09 207.66 1.19 <0.0001

25th-75th centile 0.64-204.71 39.17-420.96 0.40-3.09

Statistical comparison between HBeAg-positive and HBeAg-negative patients was performed using the Mann-Whitney U test (MWUT) for continuous data and the

chi-squared test for categorical data. c/Geq ¼ copies per genome equivalent.

HEPATOLOGY, Vol. 51, No. 6, 2010 THOMPSON ET AL. 1935

The Digene assay viral load results were converted tothe international reference unit for analysis by approxi-mating 1 pg ¼ 280,000 copies and 1 IU/mL ¼ 5.7copies/mL, allowing comparison between the assays.

Serum HBV DNA Polymerase Chain Reaction(PCR) and Sequence Analysis

Specimen Extraction. HBV DNA was extractedfrom 200 lL of patient serum using the QIAampDNA MiniKit (Qiagen, Hilden, Germany) accordingto the manufacturer’s instructions. DNA was eluted ina final volume of 50 lL of supplied elution buffer.HBV DNA Amplification and Sequence Analysis.

Two different PCR reactions were used to amplify thepolymerase and BCP/PC regions of the HBV genome,as previously described.22 HBV genes were analyzedusing the online HBV genome analysis program (Seq-HepB: www.seqvirology.com). This program comparesinput sequence data with known reference sequencesto determine HBV genotype, variants, and mutationsassociated with antiviral resistance.23 The PC variantwas defined by the presence of the G1896A mutation

(PC mutation) and the BCP variant by the A1762T/G1764A mutation (BCP mutation).

Quantification of HBV cccDNA and Total HBVDNA from Liver Tissue by Real-Time PCRIntrahepatic HBV cccDNA and total IH HBV

DNA levels were quantified by real-time PCR (RT-PCR) using a LightCycler instrument (Roche Diagnos-tics, Mannheim, Germany) as described.24 b-Globinwas used as the housekeeping gene (LightCycler Con-trol Kit DNA, Roche Diagnostics) to allow standardi-zation of the extracted DNA and expression of HBVcccDNA and total IH HBV DNA as copies per ge-nome equivalent (c/Geq).

Quantitative Immunohistochemistry(IHC) for HBsAgImmunostaining of HBV antigens was performed on

sections from paraffin-embedded liver biopsies. The pri-mary antibody used was a mouse monoclonal againstHBsAg (M3506, Dako, Carpinteria, CA). The chromo-gen used was diaminobenzidine (DAB) and staining wasperformed using BenchMark XT automated staining

Fig. 1. (A) Population distribution of HBsAg titer (n ¼ 149), comparing patients with HBeAg-positive versus HBeAg-negative CHB in the overallcohort. (B) Comparison for patients with HBsAg titers <10,000 IU/mL (n ¼ 99). (C) Plot of HBsAg titer (log IU/mL) according to HBeAg status(n ¼ 149); P < 0.0001, Mann-Whitney U test. (D) Plot of HBsAg titer (log IU/mL) adjusted for viral load, according to HBeAg status; P <0.0001, Mann-Whitney U test.


system (Ventana, Tucson, AZ). Liver sections werescored by microscopic examination by two investigators(D.I., J.S.) blinded to clinical data, using a semiquanti-tative scale of 0-4, corresponding to positivity in 0, <5,5-9, 10-29, �30% of hepatocytes, respectively (modi-fied from Ref. 25). Intensity of staining was judgedsemiquantitatively as absent, minimal, moderate, orstrong.25 Staining pattern was reported as cytoplasmic,membranous, or absent, as described.26,27

Statistical AnalysisContinuous and categorical variables were compared

between groups using, respectively, the Mann-Whitneytest and the chi-square/Fisher’s exact test. Pearson’s cor-relation coefficient (r) was used to describe the correla-tion between two continuous, normally distributedvariables. A Spearman’s correlation was used where var-iables were not normally distributed. All statisticalanalyses were performed using SAS v. 9.1 (SAS Insti-tute, Cary, NC).

Results

One hundred and forty-nine treatment-naı̈vepatients with CHB were recruited into the study (Ta-

ble 1). The majority of patients were Asian andinfected with genotype B or C HBV. All 71 HBeAg-positive patients were in the immuno-eliminationphase of disease, defined by persistently elevated serumALT. Of the HBeAg-negative patients, 57/78 (73%)were in the reactivation phase of disease, defined by se-rum HBV DNA persistently >2,000 IU/mL andraised ALT.28 Twenty-one (27%) were in the low repli-cative phase of disease (serum HBV DNA <2,000 IU/mL). Patients with HBeAg-positive CHB wereyounger, had higher serum ALT and HBV DNA levels,and were more likely to be infected with genotype CHBV. In patients with HBeAg-negative CHB, medianALT was higher in patients with viral load >2,000 IU/mL (median 54 [25th-25th centile 38-84] versus 25[22-36], P < 0.0001). Liver histology was available in73 patients (HBeAg-positive n ¼ 22, HBeAg-negativen ¼ 51). More patients with HBeAg-negative CHBhad METAVIR F3-4 fibrosis. In the subset of patientsin whom IH HBV replicative markers were measured,patients with HBeAg-positive CHB had higher levelsof HBV cccDNA and total IH DNA.Distribution of HBsAg Titers. The distribution of

serum HBsAg titer across the study population is illus-trated in Fig. 1A,B. The distribution was skewed

Table 2. Correlation Between HBsAg and HBeAg Titer and Serum and Intrahepatic HBV Replicative Markers

R * P

HBeAg-positive patients

N ¼ 71 HBsAg Serum HBV DNA 0.6863 <.0001

ALT 0.06989 0.5654

cccDNA (N ¼ 17) 0.7110 0.0014

total IH HBV DNA (N ¼ 17) 0.75815 0.0004

Serum HBV DNA ALT 0.0872 0.4728

cccDNA (N ¼ 17) 0.80997 <.0001

Total IH HBV DNA (N ¼ 17) 0.97471 <.0001

HBeAg Serum HBV DNA 0.6023 <.0001

HBsAg 0.5947 <.0001

ALT 0.0142 0.9071

cccDNA (N ¼ 17) 0.4001 0.1115

Total IH HBV DNA (N ¼ 17) 0.5990 0.0111

HBeAg-negative patients

Overall HBsAg Serum HBV DNA 0.2818 0.0124

N ¼ 78 ALT 0.1522 0.1836

cccDNA (N ¼ 27) 0.1504 0.4539

Total IH HBV DNA (N ¼ 27) 0.2525 0.2038

Serum HBV DNA ALT 0.4591 <.0001

cccDNA (N ¼ 27) 0.2979 0.1312

Total IH HBV DNA (N ¼ 27) 0.8371 <.0001

Low replicative HBsAg Serum HBV DNA 0.1464 0.5267

Serum HBV DNA < 2,000 IU/mL ALT �0.1125 0.6272

N ¼ 21 cccDNA (N ¼ 1) * *

Total IH HBV DNA (N ¼ 1) * *

Reactivation HBsAg Serum HBV DNA 0.3172 0.0162

Serum HBV DNA > 2,000 IU/mL ALT 0.1682 0.2112

N ¼ 57 cccDNA (N ¼ 26) 0.1474 0.4725

Total IH HBV DNA (N ¼ 26) 0.2509 0.2164

*Correlation coefficient for comparisons of logarithmically transformed data.


(median 3,940 IU/mL, range 3-794,000 IU/mL). Themedian HBsAg titer was significantly higher inpatients with HBeAg-positive CHB compared toHBeAg-negative CHB (median 9,459 versus 1,642IU/mL, P < 0.0001) (Fig. 1C). Interestingly, whenHBsAg was adjusted for viral load the relationship wasreversed, suggesting that HBsAg production was rela-tively preserved at the lower viral load ‘‘set-point’’observed in HBeAg-negative CHB (Fig. 1D). We didnot identify any HBsAg thresholds that could mean-ingfully differentiate patients according to viral loadstrata, in contrast to a previous report.29 There was nostatistical association between serum ALT levels and se-rum HBsAg titer (Table 2). Serum HBsAg titers werealso analyzed according to age, gender, viral genotype,and METAVIR fibrosis score. None of these variableswere associated with HBsAg titer after correction forviral load and HBeAg status (data not shown).Correlation of HBsAg with Markers of HBV Rep-

lication. HBsAg titer was positively correlated with se-rum HBV DNA (r ¼ 0.58, P < 0.0001) and HBeAg(r ¼ 0.60, P < 0.0001) (Fig. 2; Table 2). The correla-tion between HBsAg titer and serum HBV DNA wasstronger in patients with HBeAg-positive CHB (r ¼0.69, P < 0.0001). In contrast, the correlationbetween HBsAg-titer and viral load were weakly corre-lated in the overall HBeAg-negative cohort (r ¼ 0.28,P ¼ 0.012). Among HBeAg-negative patients in theinactive phase (<2,000 IU/mL), HBsAg titer did notcorrelate with HBV DNA (r ¼ 0.15, P ¼ 0.53).In 44 patients (17 HBeAg-positive and 27 HBeAg-

negative), serum HBsAg titer was then correlated withthe IH markers of HBV replication (the relationshipbetween HBsAg titer and serum HBV DNA in thesmaller subset was similar to that in the larger cohort,r ¼ 0.63, P < 0.001). Again, HBsAg titer was alsoonly correlated with IH replicative intermediates in theHBeAg-positive patients, and not in the HBeAg-nega-tive patients (Fig. 2; Table 2).Quantitative IHC for HBsAg. In these 44 patients

we also performed quantitative immunohistochemistryfor HBsAg on liver specimens (Fig. 3; Table 3). Specif-ically, we judged the liver specimens according to thepercentage of hepatocytes staining positive (Score), theintensity of the scoring (Intensity) as well as the cellu-lar distribution of staining (Pattern ¼ no staining,cytoplasmic or membranous for HBsAg26,27). InHBeAg-positive and HBeAg-negative patients, serumHBsAg titer was positively correlated with the numberand intensity of HBsAg-positive hepatocytes (r ¼0.50, P ¼ 0.04, and r ¼ 0.61, P ¼ 0.001, respectively,Table 3). High serum HBsAg titers were associated

with a membranous staining pattern (P < 0.001;Fig. 3).In patients with HBeAg-positive CHB, serum HBV

DNA and total IH HBV DNA were positively corre-lated with the Intensity of HBsAg staining and astrong trend was noted for a positive correlation with

Fig. 2. (A) Correlation of serum HBsAg titer and HBV DNA level,according to HBeAg status (HBeAg-positive, r ¼ 0.69, P < 0.0001,HBeAg-negative, r ¼ 0.28, P ¼ 0.01). (B) Correlation of serum HBsAgtiter and total IH HBV DNA level, according to HBeAg status (HBeAg-positive, r ¼ 0.76, P ¼ 0.0004, HBeAg-negative, r ¼ 0.25, P ¼0.20). (C) Correlation of serum HBsAg titer and HBV cccDNA levelaccording to HBeAg status (HBeAg-positive, r ¼ 0.71, P¼ 0.001,HBeAg-negative, r ¼ 0.15, P ¼ 0.45). Data were not normally distrib-uted and were therefore log transformed.


the HBsAg Score (Fig. 3; Table 3). HBeAg positivityand high level viral replication were associated with amembranous pattern of HBsAg staining (Fig. 3). A neg-ative correlation was noted between serum ALT andboth hepatocyte HBsAg Score and Intensity (r ¼�0.53, P ¼ 0.029, and 0.51, P ¼ 0.036, respectively).In contrast, no correlation was observed between vi-

ral replication (serum HBV DNA/total IH HBVDNA) and hepatocyte HBsAg Score/Intensity inpatients with HBeAg-negative CHB. The pattern ofstaining was also distinct, with no patient having amembranous pattern of HBsAg (Table 3; Fig. 3). Of

note, in two patients with serum HBV DNA levels<5,000 IU/mL, more than 30% of hepatocytes dis-played HBsAg staining of at least moderate intensity.No relationship between serum ALT and HBsAg IHCwas noted in HBeAg-negative patients.Distribution of HBeAg Titers. Seventy-one patients

were HBeAg-positive (Table 1). The distribution of se-rum HBeAg titer is illustrated in Fig. 4A,B. The distri-bution was skewed (median 1,147 PE IU/mL, range¼ 1.5-14,236 PE IU/mL), with 35/71 (49%) patientshaving HBeAg titer <1000 PE IU/mL, and 13/71(18%) an HBeAg titer of �31 PE IU/mL.1

Fig. 3. In patients with HBeAg-negative CHB, hepatocyte HBsAg expression, as measured by IHC, may remain relatively preserved despite low-level viral replication. The plots present correlation of (A) serum HBV DNA level and HBsAg score (percent hepatocyte staining) according toHBeAg status (HBeAg-positive, r ¼ 0.43, P ¼ 0.08, HBeAg-negative, r ¼ 0.05, P ¼ 0.77). (B) Total IH HBV DNA level and HBsAg score (per-cent hepatocyte staining) according to HBeAg status (HBeAg-positive, r ¼ 0.45, P ¼ 0.07, HBeAg-negative, r ¼ 0.32, P ¼ 0.10). (C) SerumHBV DNA level and HBsAg intensity according to HBeAg status (HBeAg-positive, r ¼ 0.54, P ¼ 0.02, HBeAg-negative, r ¼ 0.01, P ¼ 0.96). (D)Total IH HBV DNA level and HBsAg intensity according to HBeAg status (HBeAg-positive, r ¼ 0.53, P ¼ 0.03, HBeAg-negative, r ¼ 0.27, P ¼0.17). (E,F) Comparisons of serum HBV DNA level and total IH HBV DNA according to the pattern of hepatocyte HBsAg staining according toHBeAg status.


Correlation of HBeAg with Markers of HBV Rep-lication. HBeAg titer correlated positively with HBVDNA (r ¼ 0.60, P < 0.0001) (Table 2). In a subset(n ¼ 17), serum HBeAg titer was also correlated withthe IH markers of HBV replication. HBeAg correlatedwith total IH DNA (r ¼ 0.60, P ¼ 0.01) but notHBV cccDNA was (r ¼ 0.40, P ¼ 0.11) (Table 2).The relationship between HBeAg and HBV DNA inthis subset (n ¼ 17) was significant and identical tothe larger cohort (r ¼ 0.64, P ¼ 0.005).

Analysis of the Impact of BCP/PC Variants onHBeAg Titer. Sequencing of the BCP/PC region ofthe HBV genome was performed in 61 HBeAg-posi-tive patients (sequencing was not possible in 10patients for technical reasons). In 27 patients HBVDNA sequence was WT at both the BCP and PC sites,in 22 patients the BCP mutation alone was present,and in 10 patients PC mutation alone. Both weredetected in two patients. Median HBeAg titer washigher in the WT group compared to the patientswith either BCP or PC variants (3675 versus 80 PEIU/mL, P < 0.001) (Fig. 4C). It was possible that thelower HBeAg titer was secondary to reduced viral rep-lication. However, the effect of the BCP/PC variantspersisted after adjustment for viral load (Fig. 4D). Anonsignificant trend was also noted for lower HBeAgtiter in the setting of PC variants compared to BCPvariants, although this analysis was limited by thesmall numbers of HBeAg-positive patients with thePC variant as the dominant virus (data not shown).Serum HBeAg titers did not differ according to gen-der, serum ALT levels, viral genotype (B versus C), orMETAVIR fibrosis score (data not shown).Correlation of Serum HBV DNA with IH

Markers of HBV Replication. The relationship of se-rum HBV DNA with the IH markers of replicationwas analyzed (n ¼ 44). This relationship was found tobe strong (serum HBV DNA versus cccDNA, r ¼0.75, P < 0.001, and serum HBV DNA versus totalIH HBV DNA, r ¼ 0.94, P < 0.0001), asreported.7,15 Correlation of serum HBV DNA withtotal IH DNA was significant in both HBeAg-positiveand HBeAg-negative CHB (r ¼ 0.97, P < 0.0001,versus r ¼ 0.84, P < 0.0001). However, serum HBVDNA correlated significantly with cccDNA only inHBeAg-positive patients (r ¼ 0.81, P < 0.001, versusr ¼ 0.30, P ¼ 0.13 in HBeAg-negative CHB).

Discussion

This study provides the first detailed description ofserum HBsAg and HBeAg titers across a large cohortof treatment-naı̈ve patients with CHB. Both HBVantigens have recently been proposed as biomarkers fortreatment response, as well as clinical surrogates for IHHBV cccDNA levels, the intranuclear HBV reservoirresponsible for persistence. However, our data indicatethat the relationships between HBsAg, HBeAg, serumHBV DNA, and HBV replicative intermediates arecomplex. HBsAg titer correlated with serum HBVDNA, IH HBV cccDNA, and total IH DNA only inpatients with HBeAg-positive CHB, but not in

Table 3. (A) Correlation Between HBsAg, HBeAg Titer,Serum/Intrahepatic HBV Replicative Markers, and

Quantitative IHC Staining of Liver Biopsies for HBsAg. (B)Kruskal-Wallis Tests for Trend in HBsAg Titer, HBeAg Titer,Serum HBV DNA, cccDNA, and Total IH DNA According toStaining Pattern for HBsAg in Patients with HBeAg-Positive

and HBeAg-Negative CHB, Respectively

Quantitative

Immunohistochemistry Spearman R* P

A.HBeAg-positive HBsAg S-score 0.4936 0.0440

N ¼ 17 S- intensity 0.6062 0.0099

HBeAg S-score 0.5734 0.016

S- intensity 0.5893 0.0128

Serum HBV DNA S-score 0.4316 0.0837

S- intensity 0.5411 0.0249

cccDNA S-score 0.1683 0.5184

S- intensity 0.2829 0.2712

Total IH HBV DNA S-score 0.4506 0.0695

S- intensity 0.5293 0.0289

HBeAg-negative HBsAg S-score 0.5394 0.0037

N ¼ 27 S- intensity 0.4392 0.0219

Serum HBV DNA S-score 0.0583 0.7726

S- intensity 0.0107 0.9577

cccDNA S-score �0.0825 0.6825

S- intensity �0.2206 0.2689

Total IH HBV DNA S-score 0.3235 0.0997

S- intensity 0.2723 0.1695

P

Qualitative

Immunohistochemistry, S-pattern K - W test

B.HBeAg-positive HBsAg 0.0464

N ¼ 17 HBeAg 0.0212

Serum HBV DNA 0.06

cccDNA 0.589

Total IH HBV DNA 0.0725

HBeAg-negative HBsAg 0.0147

N ¼ 27 Serum HBV DNA 0.8836

cccDNA 0.1877

Total IH HBV DNA 0.2831

S/C score ¼ 0, no hepatocytes stained positive for HBsAg/HBcAg; 1,

<5% hepatocytes; 2-5, 9% hepatocytes; 3-10, 29% hepatocytes; 4, >30%

hepatocytes. S/C intensity ¼ 0, none; 1, mild; 2, moderate; 3, strong.

S-pattern ¼ no staining, cytoplasmic or membranous staining.


patients with HBeAg-negative CHB, where HBsAgtiters were preserved relative to HBV replication.HBeAg correlated with viral load, but emerging BCP/PC variants were observed to independently reduceHBeAg titer. Clinical interpretation of HBsAg andHBeAg titers might therefore be refined by consideringthe phase of disease, as well as quasispecies diversity.16

IHC studies showed for the first time that serumHBsAg titers correlated with the number and intensityof HBsAg-positive cells. However, the number and in-tensity of hepatocytes staining positive for HBsAg, aswell as the cellular pattern of distribution, were relatedto HBV replication only in patients with HBeAg-posi-tive CHB, confirming that the association betweenHBsAg production and HBV replication breaks downin the HBeAg-negative phase of disease. This mightoccur if HBsAg was produced from a source otherthan the intranuclear cccDNA (Fig. 5). In HBV infec-tion, viral integration into the host genome has beenobserved to begin early in infection.30,31 The observa-tion that the number and intensity of HBsAg-positive

cells in the liver may be preserved despite low-levelHBV replication is consistent with the occurrence ofintegration. Although integration is believed to be arandom event, a high preference for integration hasbeen observed at the DR1 and DR2 sequences on theHBV genome,32 and sequences of the S genes are of-ten present in integrated segments.33 Although inte-grated sequences cannot provide a template for pro-ductive viral replication, HBsAg may be produced.33

Progressive integration might therefore provide a tem-plate for persistent HBsAg production independent ofviral load. To date, no method for reliably differentiat-ing truncated HBs peptides in serum has been devel-oped to test this hypothesis. Another explanationmight involve preferential control of the viral replica-tion pathway at the posttranscriptional level, sparingHBV cccDNA and HBsAg transcription/secretion.This has been demonstrated to occur in vitro in thesetting of cytokine effects (targeting the encapsidationstep,34), and is consistent with the more profoundimmune pressure that is present in the HBeAg-

Fig. 4. Population distribution of HBeAg titer (n ¼ 71), for (A) the overall cohort and (B) patients with HBeAg titer <1000 PE IU/mL at right(n ¼ 35) (ii). The maximum titer observed was 14,236 PE IU/mL. (C) Plot of HBeAg titer according to BCP/PC sequence, comparing patientswith WT/WT sequence to patients with BCP and/or PC variant sequence (n ¼ 61; in 10 patients BCP/PC sequence was not available); P <0.0001, Mann-Whitney U test. (D) Plot of HBeAg titer adjusted for viral load, comparing patients with WT/WT sequence to patients with BCPand/or PC variant sequence; P < 0.0001, Mann-Whitney U test.


negative phase of disease. These two hypotheses arenot mutually exclusive.Serum ALT levels did not correlate with HBsAg ti-

ter in patients, regardless of HBeAg status. In contrast,a negative correlation was noted between serum ALTand the number and intensity of HBsAg-positive hepa-tocytes by IHC in the HBeAg-positive cohort. In thissituation it is likely that the serum ALT level is amarker for more active anti-HBV immunity, suppress-ing hepatocyte HBV replication and capsid assembly.The fact that a similar correlation is not seen with se-rum HBsAg titer may reflect that subviral particles areproduced far in excess of HBV capsids, confoundingthe relationship in the peripheral circulation. In

HBeAg-negative disease, where serum ALT levels tendto be uniformly lower, and HBsAg expression may beindependent of viral replication, no association wasobserved.The correlation between HBeAg and HBV replica-

tive markers was modest and the HBeAg titer did notcorrelate with liver HBV cccDNA (r ¼ 0.40, P ¼0.11). The data therefore implied another ‘‘discon-nect,’’ this time between the HBeAg production path-way and the viral replication pathway (Fig. 5). Themost likely explanation was that viral variants defectivefor HBeAg production were emerging before serocon-version, as we observed.16 The BCP/PC variants arethe most common to be associated with HBeAg-

Fig. 5. Diagram of the lifecycle ofHBV, highlighting the separate intracel-lular pathways for viral replication,HBsAg production and HBeAg produc-tion. HBeAg production may bereduced independent of viral load bythe emergence of viral variants defec-tive for HBeAg production (e.g., theG1896A PC variant). HBsAg productionfrom integrated sequence may preservetiters independent of viral load.Immune-mediated inhibition of HBVreplication (e.g., at the level of encap-sidation) may selectively decrease HBVvirion production. PC RNA ¼ precoremRNA, pgRNA ¼ pregenomic RNA.


negative CHB. The mechanisms of reduced or abol-ished HBeAg production have been defined in vitro.The ntA1762T and G1764A BCP mutations arethought to alter the binding of transcriptional proteinsto reduce CP-directed transcription, decreasing pre-CmRNA levels and causing a relative reduction inHBeAg.35 The G1896A PC mutation encodes for astop codon that prevents translation of the PC proteinfrom PC mRNA.36 The emergence of these variantswould therefore be expected to be associated with adecline in HBeAg titer independent of viralreplication.In this context, failure to consider quasispecies com-

position is the most likely explanation for previousconflicting descriptions of correlations between HBeAgand serum HBV DNA. Sequence analysis of the HBVquasispecies therefore should be considered whendeveloping clinical algorithms that include HBeAg ti-ter. That this might improve predictive power is anattractive hypothesis that requires further testing. Inthe setting of a homogenous WT viral population, atiter of �31 PE IU/mL may reflect an active immunesystem that will be responsive to antiviral therapy, par-ticularly with an immune modulator such as PEGIFN.Such a group might achieve seroconversion rates of>50%. In contrast, an HBeAg titer of �31 PE IU/mL in the setting of emerging PC variant virus, andimmune escape, may indicate a more treatment-refrac-tory group.In conclusion, quantitative HBeAg and HBsAg

assays are inexpensive, can be used to screen largenumbers of samples, and are compatible with automa-tion. It is clear that they can offer much more relevantdata than can be obtained by standard qualitative se-rology. They have the potential to aid decisions regard-ing treatment initiation and may open the way for aresponse-guided therapy approach in CHB. However,the relationship between HBsAg and HBeAg titers, IHreplicative intermediates, and serum HBV DNA iscomplex, and is likely to reflect an interplay betweenvirological and host immunological factors. We haveobserved key differences in HBsAg expression compar-ing patients with HBeAg-positive and HBeAg-negativeCHB. An important role for the emergence of BCP/PC variant quasispecies influencing HBeAg titer inde-pendent of viral load was also identified. These find-ings have practical implications for the use of quantita-tive serology as clinical biomarkers.

Acknowledgment: The authors thank Abbott Labo-ratories for provision of reagents and equipment; Pro-fessor Hans Tillmann for critically evaluating the data

and providing useful comments; and Ms. Alison Boydfor assistance in performing the liver immunohisto-chemical studies.

References1. Fried MW, Piratvisuth T, Lau GK, Marcellin P, Chow WC, Cooksley

G, et al. HBeAg and hepatitis B virus DNA as outcome predictors dur-ing therapy with peginterferon alfa-2a for HBeAg-positive chronic hep-atitis B. HEPATOLOGY 2008;47:428-434.

2. Wiegand J, Wedemeyer H, Finger A, Heidrich B, Rosenau J, MichelG, et al. A decline in hepatitis B virus surface antigen (hbsag) predictsclearance, but does not correlate with quantitative hbeag or HBV DNAlevels. Antivir Ther 2008;13:547-554.

3. Marcellin P, Brunetto MR, Bonino F, Hadziyannis S, Kapprell H,McCloud P, et al. In patients with HBeAg-negative chronic hepatitis BHBsAg serum levels early during treatment with peginterferon alfa-2apredict HBsAg clearance 4 years post-treatment. HEPATOLOGY 2008;48(Suppl 1):718A (919).

4. Brunetto MR, Moriconi F, Bonino F, Lau GK, Farci P, Yurdaydin C,et al. Hepatitis B virus surface antigen levels: a guide to sustainedresponse to peginterferon alfa-2a in HBeAg-negative chronic hepatitisB. HEPATOLOGY 2009;49:1141-1150.

5. Manesis EK, Hadziyannis E, Angelopoulou OP, Hadziyannis SJ. Pre-diction of treatment-related HBsAg loss in HBeAg-negative chronichepatitis B: a clue from serum HBsAg levels. Antivir Ther 2007;12:73-82.

6. Deguchi M, Yamashita N, Kagita M, Asari S, Iwatani Y, Tsuchida T,et al. Quantitation of hepatitis B surface antigen by an automatedchemiluminescent microparticle immunoassay. J Virol Methods 2004;115:217-222.

7. Werle-Lapostolle B, Bowden S, Locarnini S, Wursthorn K, Petersen J,Lau G, et al. Persistence of cccDNA during the natural history ofchronic hepatitis B and decline during adefovir dipivoxil therapy. Gas-troenterology 2004;126:1750-1758.

8. Wursthorn K, Lutgehetmann M, Dandri M, Volz T, Buggisch P, Zoll-ner B, et al. Peginterferon alpha-2b plus adefovir induce strongcccDNA decline and HBsAg reduction in patients with chronic hepati-tis B. HEPATOLOGY 2006;44:675-684.

9. Liaw YF, Sung JJ, Chow WC, Farrell G, Lee CZ, Yuen H, et al. Lami-vudine for patients with chronic hepatitis B and advanced liver disease.N Engl J Med 2004;351:1521-1531.

10. Nowak MA, Bonhoeffer S, Hill AM, Boehme R, Thomas HC,McDade H. Viral dynamics in hepatitis B virus infection. Proc NatlAcad Sci U S A 1996;93:4398-4402.

11. Kuhns MC, Kleinman SH, McNamara AL, Rawal B, Glynn S, BuschMP. Lack of correlation between HBsAg and HBV DNA levels inblood donors who test positive for HBsAg and anti-HBc: implicationsfor future HBV screening policy. Transfusion 2004;44:1332-1339.

12. Volz T, Lutgehetmann M, Wachtler P, Jacob A, Quaas A, Murray JM,et al. Impaired intrahepatic hepatitis B virus productivity contributes tolow viremia in most HBeAg-negative patients. Gastroenterology 2007;133:843-852.

13. Lu L, Ye D, Wang Y, Kwok A, Wong A, Yueng Y, et al. Correlationbetween HBV cccDNA and HBsAg levels and their reduction bypeginterferon alfa-2a based therapy in patients with chronic hepatitis B[Abstract]. HEPATOLOGY 2008;48(Suppl 1):746A (979).

14. Manesis EK, Papatheodoridis GV, Hadziyannis E, Nastos T, KarayannisP. HBsAg serum levels correlate with total liver HBV DNA but notwith cccDNA [Abstract]. HEPATOLOGY 2008;48(Suppl 1):371A (147).

15. Laras A, Koskinas J, Dimou E, Kostamena A, Hadziyannis SJ. Intrahepaticlevels and replicative activity of covalently closed circular hepatitis B virusDNA in chronically infected patients. HEPATOLOGY 2006;44:694-702.

16. Lim S, Cheng Y, Guindon S, Seet B, Lee L, Hu P, et al. Viral quasispe-cies evolution during hepatitis B ‘‘E’’ antigen seroconversion. Gastroen-terology 2007;133:951-958.


17. Tacke F, Gehrke C, Luedde T, Heim A, Manns MP, Trautwein C. Basalcore promoter and precore mutations in the hepatitis B virus genomeenhance replication efficacy of Lamivudine-resistant mutants. J Virol2004;78:8524-8535.

18. Nguyen T, Thompson A, Desmond P, Bowden DS, Levy M, LocarniniS. Serum HBeAg and HBsAg concentrations: changing levels duringthe natural history of chronic hepatitis B. J Hepatol 2009. In press.

19. Keeffe EB, Dieterich DT, Han SH, Jacobson IM, Martin P, Schiff ER,et al. A treatment algorithm for the management of chronic hepatitis Bvirus infection in the United States: 2008 update. Clin GastroenterolHepatol 2008;6:1315-1341.

20. Bedossa P, Poynard T. An algorithm for the grading of activity inchronic hepatitis C. The METAVIR Cooperative Study Group.HEPATOLOGY 1996;24:289-293.

21. Perrillo R, Mimms L, Schechtman K, Robbins D, Campbell C. Moni-toring of antiviral therapy with quantitative evaluation of HBeAg: acomparison with HBV DNA testing. HEPATOLOGY 1993;18:1306-1312.

22. Ayres A, Locarnini S, Bartholomeusz A. HBV genotyping and analysisfor unique mutations. Methods Mol Med 2004;95:125-149.

23. Yuen LK, Ayres A, Littlejohn M, Colledge D, Edgely A, Maskill WJ,et al. SeqHepB: a sequence analysis program and relational databasesystem for chronic hepatitis B. Antiviral Res 2007;75:64-74.

24. Bowden S, Jackson K, Littlejohn M, Locarnini S. Quantification ofHBV covalently closed circular DNA from liver tissue by real-timePCR. Methods Mol Med 2004;95:41-50.

25. Chu C, Yeh C, Chien R, Sheen I, Liaw Y. The degrees of hepatocytenuclear but not cytoplasmic expression of hepatitis B core antigenreflect the level of viral replication in chronic hepatitis B virus infec-tion. J Clin Microbiol 1997;35:102-105.

26. Chu C, Liaw Y. Natural history of chronic hepatitis B virus infection: animmunopathological study. J Gastroenterol Hepatol 1997;12:S218-222.

27. Kim CW, Yoon SK, Jung ES, Jung CK, Jang JW, Kim MS, et al. Cor-relation of hepatitis B core antigen and beta-catenin expression on he-patocytes in chronic hepatitis B virus infection: relevance to the severityof liver damage and viral replication. J Gastroenterol Hepatol 2007;22:1534-1542.

28. Lok AS, McMahon BJ. Chronic hepatitis B: update 2009. HEPATOLOGY

2009;50:661-662.29. Chen CH, Lee CM, Wang JH, Tung HD, Hung CH, Lu SN. Correla-

tion of quantitative assay of hepatitis B surface antigen and HBV DNAlevels in asymptomatic hepatitis B virus carriers. Eur J GastroenterolHepatol 2004;16:1213-1218.

30. Kimbi GC, Kramvis A, Kew MC. Integration of hepatitis B virusDNA into chromosomal DNA during acute hepatitis B. World J Gas-troenterol 2005;11:6416-6421.

31. Bill CA, Summers J. Genomic DNA double-strand breaks are targetsfor hepadnaviral DNA integration. Proc Natl Acad Sci U S A 2004;101:11135-11140.

32. Nagaya T, Nakamura T, Tokino T, Tsurimoto T, Imai M, Mayumi T,et al. The mode of hepatitis B virus DNA integration in chromosomesof human hepatocellular carcinoma. Genes Dev 1987;1:773-782.

33. Thomas HC, Lemon SM (eds.). Viral Hepatitis. 3rd ed. Boston, MA:Blackwell; 2005.

34. Biermer M, Puro R, Schneider RJ. Tumor necrosis factor alpha inhibi-tion of hepatitis B virus replication involves disruption of capsid integ-rity through activation of NF-kappaB. J Virol 2003;77:4033-4042.

35. Laras A, Koskinas J, Hadziyannis SJ. In vivo suppression of precoremRNA synthesis is associated with mutations in the hepatitis B viruscore promoter. Virology 2002;295:86-96.

36. Carman WF, Jacyna MR, Hadziyannis S, Karayiannis P, McGarvey MJ,Makris A, et al. Mutation preventing formation of hepatitis B e antigenin patients with chronic hepatitis B infection. Lancet 1989;2:588-591.


Serum hepatitis B surface antigen and hepatitis B e antigen titers: Disease phase influences...

Documents

Transcript of Serum hepatitis B surface antigen and hepatitis B e antigen titers: Disease phase influences...