SGM Journals
Animal: DNA Viruses

High genetic variability of the group-specific a-determinant of hepatitis B virus surface antigen (HBsAg) and the corresponding fragment of the viral polymerase in chronic virus carriers lacking detectable HBsAg in serum

Klaus M. Weinberger, Tanja Bauer, Stephan Böhm, Wolfgang Jilg

  • Institute for Medical Microbiology and Hygiene, University of Regensburg, Franz-Josef-Strauß-Allee 11, D-93053 Regensburg, Germany1
  • Author for correspondence: Klaus Weinberger. Fax +49 941 944 6402. e-mail klaus-michael.weinberger{at}klinik.uni-regensburg.de

    • Journal of General Virology 2000; 81(5):1165–1174

    PubMed

    Abstract

    Chronic carriers of hepatitis B virus (HBV) usually show hepatitis B surface antigen (HBsAg) in their sera, which is considered the best marker for acute and chronic HBV infection. In some individuals, however, this antigen cannot be detected by routine serological assays despite the presence of virus in liver and peripheral blood. One reason for this lack of HBsAg might be mutations in the part of the molecule recognized by specific antibodies. To test this hypothesis, the HBV S gene sequences were determined of isolates from 33 virus carriers who were negative for HBsAg but showed antibodies against the virus core (anti-HBc) as the only serological marker of hepatitis B. Isolates from 36 HBsAg-positive patients served as controls. In both groups, a considerable number of novel mutations were found. In isolates from individuals with anti-HBc reactivity only, the variability of the major hydrophilic loop of HBsAg, the main target for neutralizing and diagnostic antibodies, was raised significantly when compared with the residual protein (22·6 vs 9·4 mutations per 1000 amino acids; P<0·001) and with the corresponding region in the controls (22·6 vs 7·5 exchanges per 1000 residues; P<0·001). A similar hypervariable spot was identified in the reverse transcriptase domain of the viral polymerase, encoded by the same nucleotide sequence in an overlapping reading frame. These findings suggest that at least some of the chronic low-level carriers of HBV, where surface antigen is not detected, could be infected by diagnostic escape mutants and/or by variants with impaired replication.

    Introduction

    The serological diagnosis of hepatitis B virus (HBV) infection is based mainly on assays for the small surface antigen (sHBsAg), the major glycoprotein of the virus envelope. The presence of HBsAg in the peripheral blood is a very sensitive marker for viraemia, especially since non-infectious spherical or tubular forms of HBsAg (the so-called 22 nm particles) are found in up to 1000-fold excess over intact virions in the sera of viraemic individuals (Kann & Gerlich, 1998⇓ ). On the other hand, the absence of HBsAg is believed to exclude infectivity; HBsAg testing is therefore widely used as a screening method for blood and organ donors.

    There have been repeated reports of cases of post-transfusion hepatitis B, however, despite these measures (Hoofnagle et al., 1978⇓ ; Norder et al., 1992⇓ ), and HBsAg-negative chronic carriers of HBV have been identified in completely different groups of individuals (Carman et al., 1991⇓ ; Coursaget et al., 1991⇓ ; Jilg et al., 1995⇓ ; Thijssen et al., 1993⇓ ). The most prevalent serological pattern among these carriers is an isolated anti-HBV core (HBc) reactivity, which has been interpreted mostly as resolved hepatitis B with anti-HBs having disappeared with time. A recent study that included a representative portion of the German population revealed that individuals with this pattern are found even more frequently than HBsAg carriers (1·4 vs 0·6%; n=5377) (W. Jilg, B. Hottenträger, K. Schlottmann, E. Frick, A. Holstege, J. Schölmerich and K. D. Palitzsch, unpublished results).

    Since highly sensitive PCR methods have been introduced for the detection of HBV (Sumazaki et al., 1989⇓ ), it has been shown that at least 10% of these cases (Kroes et al., 1991⇓ ; Weinberger et al., 1997⇓ ), but up to 40% in groups with a high risk of infection (Jilg et al., 1995⇓ ; Joller-Jemelka et al., 1994⇓ ; Sánchez-Quijano et al., 1993⇓ ), are viraemic, although generally at a very low level. This means that a minimum of one out of 1000 individuals is a potentially infectious carrier of HBV, without being detectable by standard screening measures.

    Various attempts have been made to clarify the mechanism(s) responsible for this serological pattern. Two possibilities are discussed: either the antigen is indeed absent from the peripheral blood or it is present but not detectable. The first alternative, absence of HBsAg, could be due to mutations that block the export of the antigen, as described for certain pre-S deletions (Melegari et al., 1994⇓ ). Another explanation is the frequently observed co-infection with hepatitis C virus (HCV) (Jilg et al., 1995⇓ ; Lee et al., 1997⇓ ), the core protein of which was shown to down-regulate HBV replication and protein synthesis in a hepatoma cell line (Shih et al., 1993⇓ ). The second alternative, i.e. HBsAg is present but not detectable by standard enzyme immunoassay techniques, could be ascribed to the presence of circulating immune complexes between HBsAg and anti-HBs, which can worsen (Ackerman et al., 1994⇓ ) or even inhibit completely (Joller-Jemelka et al., 1994⇓ ) the detection of both the antigen and the antibody.

    There is an additional possible explanation for the lack of detectable HBsAg, however. Mutations in the major hydrophilic loop (MHL, amino acids 98–156), the main target for antibodies used in diagnostic tests, could lead to an escape from recognition by routinely used assays. Since the frequency and clinical significance of immune-escape mutants of HBV with structural alterations in this part of HBsAg have been discussed intensely over the last few years (Bahn et al., 1997⇓ ; Carman, 1997⇓ ; Ghany et al., 1998⇓ ; Protzer-Knolle et al., 1998⇓ ; Schätzl et al., 1997⇓ ; Waters et al., 1992⇓ ), we address here the question of whether similar variations of the target structures for diagnostic antibodies could lead to a diagnostic escape.

    Surprisingly, apart from case reports (Grethe et al., 1998⇓ ), very few sequences derived from solely anti-HBc-positive carriers have been published to date. We therefore determined the genomic sequences of the gene encoding the HBsAg of isolates from 33 virus carriers who were serologically negative for HBsAg and showed anti-HBc reactivity as the only marker of HBV infection. In addition, to get an impression of the general variability of this part of the viral genome, we also analysed all HBsAg sequences published in the GenBank database and sequenced the HBV S genes of virus isolates from 36 ‘normal’ HBsAg-positive virus carriers as controls. The nature and frequency of mutations found in both groups were compared and differences were analysed statistically.

    Genetic alterations in the HBV S gene have an effect on both the HBsAg and the reverse transcriptase (RT) domain of the polymerase. Since the RT is an important target for antiviral therapeutics, mutations that confer resistance to lamivudine (Bartholomew et al., 1997⇓ ; Honkoop et al., 1997⇓ ; Tipples et al., 1996⇓ ) or famciclovir (Melegari et al., 1998⇓ ; Pichoud et al., 1999⇓ ; Zoulim & Trépo, 1998⇓ ) yet impair the enzymatic activity have been characterized in detail. A lower replication activity of the RT, resulting from the same mutations that cause exchanges in the immunogenic parts of HBsAg, could act synergistically in HBsAg-negative low-level carriers. Therefore, the influence of the variations observed on the viral polymerase was also considered.

    Methods

    ▪ Sera.

    The solely anti-HBc-positive sera tested in this study (n=357) were collected from different sources. The majority of the samples (n=193; 54·1%) was taken from studies concerning the prevalence of HBV in a random portion of the German population (n=75), the occupational risk of HBV infection for dentists (n=4), the prevalence of HBV markers in a strictly selected collection of blood donors (n=50) (W. Jilg, K. Körner and B. Kubanek, unpublished results) and among prisoners (n=64) (Neifer et al., 1997⇓ ). The remaining samples were sent to our institute for diagnostic purposes (n=164; 45·9%), as well as the HBsAg-positive sera (n=36) used as controls. Cases of acute hepatitis B with positive tests for anti-HBc IgM were excluded. Both groups were matched for their age and sex distribution. The sera serving as known negative controls were collected from members of the institute after vaccination against HBV. All sera were either processed immediately after arrival or stored at −20 °C.

    ▪ Serology.

    Qualitative serological tests for HBsAg, anti-HBc, anti-HBc IgM, HBV e antigen (HBeAg), anti-HBe and anti-HCV as well as the quantification of anti-HBs were performed by using standard, commercially available, microparticle enzyme immunoassays (AxSym, Abbott Laboratories). Positive results for anti-HBc in individuals that were negative for HBsAg and anti-HBs were always confirmed by using a second test system (ImX core, Abbott Laboratories) (Hughes et al., 1995⇓ ; Turner et al., 1997⇓ ).

    ▪ DNA isolation and amplification.

    Viral nucleic acid was prepared from serum samples by using the QIAamp blood and tissue kit (Qiagen) following a slightly modified protocol (initial sample volume of 400 μl, proteinase K concentration of 1 μg/ml in the lysis reaction and elution volume of 100 μl). Viral particles were enriched from 4 ml serum, if enough material was available, by ultracentrifugation overnight through a 20% (w/v) sucrose cushion at 70000 g. The resulting pellet was resuspended in 400 μl PBS prior to DNA isolation (Weinberger et al., 1999⇓ ). DNA preparations were stored at −20 °C in 10 mM Tris–HCl, pH 8·0, and thawed immediately before amplification. The primers, probes and reaction conditions used for first-round and nested amplification of the HBV S gene, as well as for identification of the amplified products in Southern blot analyses, have been described previously (Weinberger et al., 1997⇓ ). Appropriate measures were taken to minimize the risk of cross-contamination (Kwok & Higuchi, 1989⇓ ). In addition, 102 sera from healthy individuals who were actively immunized against HBV were distributed randomly among the samples for DNA isolation and served as negative controls. Moreover, a critical comparison of all the genomic sequences from the same batch of samples was performed in order to minimize further the risk of false positives. Semi-quantification of the viral DNA was achieved by comparison with triplicate samples of plasmid pHBV991 (genotype A, serotype adw2, GenBank accession number X51970, cloned in the unique BamHI site of pBR322) in a serial 10-fold dilution, starting at 106 copies down to one copy per reaction. For comparison, selected samples were also analysed by using a novel quantitative TaqMan assay, yielding a linear range of proportionality (threshold cycle vs log10 template concentration) that covers more than seven orders of magnitude (Weinberger et al., 2000⇓ ).

    ▪ Sequence analysis.

    The sequence of the HBV S gene was determined from three widely overlapping partial sequences, analysed on both strands of at least three independently amplified PCR products (Weinberger et al., 1997⇓ ) by using the PRISM ready reaction dye deoxy terminator cycle sequencing kit (Perkin Elmer) according to the manufacturer’s instructions. The fluorescence signals were detected on gel (ABI 373A) or capillary (ABI 310) systems (Applied Biosystems). The ABI sequence editor software served as a tool for comparing the different sequencing reactions and for constructing the entire gene. All further sequence analysis was performed by modules of the GCG software package (version 10.0; Genetics Computer Group, Madison, WI, USA). Briefly, the genomic sequence of the entire HBV S gene was compared with all available HBV sequences in the GenBank and EMBL databases. The predicted translations of the surface and polymerase reading frames were compared with the PIR and SWISS-PROT resources. Because of the genetic diversity between HBV genotypes A to F (Norder et al., 1993⇓ ), there is no useful consensus sequence that could serve as a standard for all comparisons. Therefore, deviations from only the closest related entries were considered to be mutations and taken into account for the statistical analyses. The biochemical significance of amino acid substitutions was either estimated from the empirical Swiss2 homology matrix, based on the natural occurrence of different residues at homologous positions in functionally related proteins (Gonnet et al., 1992⇓ ), or predicted by knowledge-based molecular-modelling studies of the MHL with the SYBYL software package (version 6.4, Tripos Inc.).

    Statistical significance was determined by the χ2-test in 2×2 contingency tables.

    Results

    Semi-quantitative PCR of HBsAg-positive and solely anti-HBc-positive sera

    Sera from 357 individuals with anti-HBc as the only marker for hepatitis B infection were tested for HBV DNA. Thirty-three (9·2%) were found to be positive by nested PCR and Southern blot, with a generally low virus load that did not exceed 104 viral genomes per ml serum (median, 5×102 per ml).

    Thirty-six HBsAg-positive virus carriers served as controls. Their sera showed moderate to high DNA levels, ranging from 103 to 109 genomes per ml (median, 107 per ml). All known negative sera (n=102) from vaccinated individuals were repeatedly negative by nested PCR.

    Sequencing of the HBV S gene and statistical comparison

    Among the HBsAg-negative virus carriers, sequencing of the entire S gene showed novel DNA sequences in 26 of 33 isolates (78%). For 22 of the isolates (67%), the corresponding amino acid sequences have not yet been published (sequence data shown in Table 1⇓). In order to measure the degree of variability in these new sequences, we compared the differences from the most-closely related published S gene sequences: a mean deviation of 4·97 nucleotides per isolate (0·73%) was found for all 33 isolates, with a maximum of 16 nucleotides (2·3%). At the polypeptide level, 2·94 exchanges per isolate were observed on average (range, 0 to 9 residues substituted). Within the HBsAg-negative group, anti-HCV-positive sera did not differ significantly from anti-HCV-negatives when comparing the frequency of mutations.

    Table 1. Amino acid exchanges in the HBsAg of 33 solely anti-HBc-positive sera

    Surprisingly, the isolates from the control group also showed a high variability in their S gene sequences. Only one sample had a genomic sequence that had already been described; the others differed by up to 25 nucleotides (3·7%) from their most related correlates, with an average of 5·04 nucleotides (0·74%) per isolate. In 27 cases (75%), the corresponding HBsAg sequences were new (sequence data shown in Table 2⇓), showing a mean deviation of 2·49 exchanged amino acids (1·10%) with a maximum of 11 substitutions (4·9%). Moreover, two sera even contained genomes with a premature stop codon in the HBsAg reading frame, after 181 and 215 residues.

    Table 2. Amino acid exchanges in the HBsAg of 36 HBsAg-positive sera

    Despite the comparably high variability of the HBsAg sequences in both groups, significant differences were found in the distribution of amino acid exchanges within the HBsAg molecules (Fig. 1⇓). In isolates of the HBsAg-negative virus carriers, there were twice as many exchanges in the MHL region as in the residual protein (22·6 vs 9·4 substitutions per 1000 amino acids; P<0·001) and almost exactly three times more than in the corresponding segment of the controls (22·6 vs 7·5 per 1000 amino acids; P<0·001). On the other hand, the isolates from HBsAg-positive carriers showed the opposite distribution: here, the frequency of exchanges within the MHL was even lower than in the residual polypeptide (7·5 vs 12·0 per 1000 amino acids) (Fig. 2⇓).

    Figure image not available in archive

    Fig. 1. Mutation frequencies of HBsAg in solely anti-HBc-positive isolates (filled bars, n=33) and HBsAg-positive isolates (shaded bars, n=36), analysed in intervals of 10 amino acids each. The structural correlations are given at the top of the diagram. Note that, in five out of six decades within the MHL, the variability of the solely anti-HBc-positive isolates is significantly higher than in the controls (marked by asterisks).

    Figure image not available in archive

    Fig. 2. Comparison of the mutation frequencies in functionally defined regions of HBsAg. The variability of the MHL in solely anti-HBc-positive samples (filled bars, n=33) is significantly higher than in the residual protein, as well as in the corresponding region of the controls (shaded bars, n=36). Levels of significance were calculated by χ2-tests and are indicated by * (P<0·001), † (P<0·001) and ‡ (P<0·1).

    In addition to the distribution, we also found differences in the nature of the mutations observed in the MHL. As assessed by the Swiss2 homology matrix, the more radical exchanges clustered within the two loops considered crucial for antibody binding (aa 121/124–147/149). Within the complete MHL, 28 amino acid exchanges were unique for isolates from solely anti-HBc-positive subjects (Fig. 3⇓).

    Figure image not available in archive

    Fig. 3. Amino acid exchanges in the MHL of HBsAg unique for isolates from HBsAg-negative virus carriers. The schematic drawing of the MHL (aa 98–156) is based on the most plausible alternatives for the disulphide bridging pattern. Twenty-eight substitutions that were observed only in HBsAg-negative samples are shown. •, Residues with mutations;

    Figure image not available in archive
    , cysteines (disulphide bridges are shown); ○, other residues. The complex N-linked glycan is bound to 146N.

    When genotypes from both groups were compared, further differences were also observed. Twenty of 33 isolates (60·6%) of HBsAg-negative carriers belonged to genogroup D and only 10 (30·3%) to genogroup A, whereas the reverse was seen in the controls: here, a majority of 52·8% (19 of 36) could be placed in genogroup A and exactly one third (12 of 36) in group D (Fig. 4⇓).

    Figure image not available in archive

    Fig. 4. Genotyping of HBV S gene sequences. Grouping was performed by standard genetic algorithms by using a panel of 24 prototype HBV genomes. The predominance of genotypes A and D is observed in both groups, but the distribution between these two types differs markedly: in HBsAg-negative samples (filled bars, n=33), type D is found most frequently, while type A is predominant in the controls (shaded bars, n=36).

    Analysis of the polymerase reading frame

    The sequence of a 226 amino acid fragment of the viral polymerase was deduced from each DNA sequence, representing a main part of the RT domain, including regions B and C, which are associated with therapy-resistant variants. Only six of the solely anti-HBc-positive samples (18·2%) and no HBsAg-positive isolates encoded polypeptides with a published sequence. The mean deviations from the most-closely related sequences were 2·1% (maximum 7·1% or 16 amino acids) in the HBsAg-negative group and 1·7% (maximum 5·8% or 13 exchanges) in the controls (the deviations for both reading frames are summarized in Table 3⇓). In both groups, one serum each was found with a stop mutation, after residues 405 and 502. Yet, with the exception of the two truncated sequences, the YMDD motif, which is essential for the enzymatic activity of many RTs, was conserved in all samples.

    Table 3. Deviation from the most-closely related GenBank entries given for isolates from both groups analysed

    When the frequency of mutations was analysed as had been done previously for the HBsAg (Fig. 5⇓), the fragment encoded by the same region as the MHL (residues 440–490) showed a similar hypervariability (40·9 vs 27·3 exchanges per 1000 amino acids; P=0·016) when compared with the controls. Additionally, in isolates from the solely anti-HBc-positive individuals, a second region with a significantly raised frequency of mutations could be identified, between residues 510 and 550, covering the catalytic domains B and C (14·4 vs 3·5 substitutions per 1000 amino acids; P=0·002).

    Figure image not available in archive

    Fig. 5. Variability of the polymerase reading frame in solely anti-HBc-positive isolates (filled bars, n=33) and HBsAg-positive isolates (shaded bars, n=36), analysed in intervals of 10 amino acids each (as in Fig. 1⇑ for HBsAg). The two regions with significantly increased variability (aa 440–500 and 510–550) are close to the catalytic centres of the RT domain.

    Discussion

    The group of HBsAg-negative, solely anti-HBc-positive carriers of HBV represents a challenge for serological diagnosis and for the safety measures employed to prevent HBV transmission by blood, blood products or donor organs. The present study addressed the question of how often and which virus variants emerge that could be responsible for the lack of detectable HBsAg in the sera of these individuals.

    So far, only a few sequences of HBV isolates derived from solely anti-HBc-positive sera have been published. In our study, we analysed a fairly large group of 33 HBsAg-negative virus carriers and compared the results obtained with those derived from the analysis of 36 ‘normal’, i.e. HBsAg-positive, carriers. Two interesting phenomena were observed. Firstly, a surprisingly large degree of variability of HBsAg was seen in isolates from HBsAg-negative virus carriers and from controls, as well as striking differences in the frequency and distribution of mutations in different functionally defined regions of the molecule. Secondly, the polymerase reading frame was affected in a similar way, showing two regions with significantly higher frequencies of amino acid substitutions.

    Evidence from many indirect approaches has led to a structural model of the HBsAg, which includes four membrane-spanning α-helices (A–D) and the MHL, between helices B and C, exposed on the surface of the viral and subviral particles. This protein loop carries the most important target structures for neutralizing, as well as for diagnostic, antibodies and seems to be defined structurally by multiple disulphide bridges, of which one plausible conformation is chosen for Fig. 3⇑.

    In virus strains found in our group with isolated anti-HBc-reactivity, this part of HBsAg turned out to be significantly more variable than the residual protein as well as the same region in the controls. There are no single characteristic mutations present in the majority of this group and no previously described immune escape variant emerged, such as 144A or 145R. Yet, the increased number of exchanges that we observed may mirror structural alterations that could affect the binding affinity to certain antibodies.

    This high level of variability could not be found in the controls; on the contrary, in HBsAg-carriers, the MHL was even slightly conserved. This is not only deduced from our sequence data, but is in agreement with a statistical analysis of 117 published HBsAg sequences, where the frequency of DNA mutations that affect the resulting protein sequence is significantly lower within the MHL (350 substituted amino acids caused by 1033 exchanged nucleotides or 33·9% in the MHL vs 912 of 1611 or 56·6% in the residual molecule; P<0·001).

    The prediction of structural and biochemical effects from amino acid substitutions is always difficult, especially when the three-dimensional structure of the protein is still unknown. However, considering the properties and the position of the respective residues, one can estimate the significance of some of the exchanges. Of the mutations that we found only in the solely anti-HBc-positive sera, the substitutions C121W and C147W almost certainly affect the immunological properties of the protein by impeding the correct formation of essential disulphide bridges. Three additional sites of substitution were found within the second loop, immediately adjacent to the site of N-glycosylation (146N) and the known escape mutants 145R and 144A. A site of hypervariability was located at position 134, where three novel residues could be identified. Interestingly, one isolate also contained a mutation of the d/y-serotype determination residue 122, where the neutral I replaces the basic residues K or R and should extinguish the d/y binding property completely. Altogether, 28 exchanges within the MHL were observed only in the solely anti-HBc-positive sera and these exchanges need to be characterized immunologically in further studies.

    Differences were also found in the distribution of the six HBV genotypes: the data from both groups of individuals confirm that genogroups A and D are the most prevalent types in central Europe (Norder et al., 1993⇓ ). Nevertheless, there is a marked contrast between the predominance of genotype A in the controls and of genotype D among the solely anti-HBc-positive samples. This predominance may well influence the biological significance of certain naturally occurring variants, especially considering the fact that the immune-escape mutant 145R is closely associated with type D genomes (Carman, 1997⇓ ). Additionally, genotype D has been described to emerge preferentially during seroconversion from HBsAg to anti-HBs (Bahn et al., 1997⇓ ) and from HBeAg to anti-HBe (Friedt et al., 1999⇓ ; Gerner et al., 1998⇓ ), indicating a correlation with the selective pressure of the host’s immune response.

    The interpretation of the increased variability of the polymerase reading frame poses similar difficulties, although the structural knowledge of homologous RTs is much more complete (Kohlstaedt et al., 1992⇓ ). However, certain mutations have been identified that impair virus replication to different extents (Melegari et al., 1998⇓ ; Pichoud et al., 1999⇓ ). Therefore, in solely anti-HBc-positive individuals, one cannot neglect the possibility that the typically low level of HBV DNA observed may be due to defects in the catalytic RT domains, particularly regions B and C (Zoulim & Trépo, 1998⇓ ), which probably also result in a low level of protein synthesis.

    As all our sequence data were gained by direct sequencing of PCR products, we cannot exclude the possibility that viral subpopulations with other genomic sequences were also present in the sera studied, particularly in cases with truncated polymerase proteins. Yet, as each sequence was determined from at least three independent amplifications, the resulting data (in the case of identity of all three experiments) certainly represent the highly predominant species of HBV in the respective sample.

    To summarize, there are several significant differences in the genetic variability of HBV between chronic carriers with anti-HBc as the only serological marker and HBsAg-positives. These differences affect functionally important regions of both the HBsAg and the polymerase. We suggest, therefore, that impaired recognition through diagnostic assays or reduced replicative activity of the RT or both of these factors could cause HBsAg-negative carriership in at least some of these cases.

    Acknowledgments

    This work was supported by a grant from the ‘Structure, function and clinical relevance of HBV variants’ project of the Deutsches Bundesministerium für Bildung und Forschung. The authors wish to thank Barbara Hottenträger and Elke Kreuzpaintner for excellent technical assistance. Plasmid pHBV991 was kindly provided by Professor R. Thomssen, University of Göttingen, Germany.

    References

    1. Ackerman, Z., Wands, J. R., Gazitt, Y., Bréchot, C., Kew, M. C. & Shouval, D. (1994). Enhancement of HBsAg detection in serum of patients with chronic liver disease following removal of circulating immune complexes.Journal of Hepatology 20, 398-404.
    2. Bahn, A., Gerner, P., Martine, U., Bortolotti, F. & Wirth, S. (1997). Detection of different viral strains of hepatitis B virus in chronically infected children after seroconversion from HBsAg to anti-HBs indicating viral persistence.Journal of Hepatology 27, 973-978.
    3. Bartholomew, M. M., Jansen, R. W., Jeffers, L. J., Reddy, K. R., Johnson, L. C., Bunzendahl, H., Condreay, L. D., Tzakis, A. G., Schiff, E. R. & Brown, N. A. (1997). Hepatitis-B-virus resistance to lamivudine given for recurrent infection after orthotopic liver transplantation.Lancet 349, 20-22.
    4. Carman, W. F. (1997). The clinical significance of surface antigen variants of hepatitis B virus. Journal of Viral Hepatitis 4 (Suppl. 1), 11–20.
    5. Carman, W. F., Dourakis, S., Karayiannis, P., Crossey, M., Drobner, R. & Thomas, H. C. (1991). Incidence of hepatitis B viraemia, detected using the polymerase chain reaction, after successful therapy of hepatitis B virus carriers with interferon-alpha.Journal of Medical Virology 34, 114-118.
    6. Coursaget, P., Le Cann, P., Leboulleux, D., Diop, M. T., Bao, O. & Coll, A. M. (1991). Detection of hepatitis B virus DNA by polymerase chain reaction in HBsAg negative Senegalese patients suffering from cirrhosis or primary liver cancer.FEMS Microbiology Letters 67, 35-38.
    7. Friedt, M., Gerner, P., Lausch, E., Trübel, H., Zabel, B. & Wirth, S. (1999). Mutations in the basic core promotor and the precore region of hepatitis B virus and their selection in children with fulminant and chronic hepatitis B.Hepatology 29, 1252-1258.
    8. Gerner, P. R., Friedt, M., Oettinger, R., Lausch, E. & Wirth, S. (1998). The hepatitis B virus seroconversion to anti-HBe is frequently associated with HBV genotype changes and selection of pre-S2-defective particles in chronically infected children.Virology 245, 163-172.
    9. Ghany, M. G., Ayola, B., Villamil, F. G., Gish, R. G., Rojter, S., Vierling, J. M. & Lok, A. S. F. (1998). Hepatitis B virus S mutants in liver transplant recipients who were reinfected despite hepatitis B immune globulin prophylaxis.Hepatology 27, 213-222.
    10. Gonnet, G. H., Cohen, M. A. & Benner, S. A. (1992). Exhaustive matching of the entire protein sequence database.Science 256, 1443-1445.
    11. Grethe, S., Monazahian, M., Böhme, I. & Thomssen, R. (1998). Characterization of unusual escape variants of hepatitis B virus isolated from a hepatitis B surface antigen-negative subject.Journal of Virology 72, 7692-7696.
    12. Honkoop, P., Niesters, H. G., de Man, R. A., Osterhaus, A. D. & Schalm, S. W. (1997). Lamivudine resistance in immunocompetent chronic hepatitis B. Incidence and patterns.Journal of Hepatology 26, 1393-1395.
    13. Hoofnagle, J. H., Seefe, L. B., Bales, Z. B. & Zimmerman, H. J. (1978). Type B hepatitis after transfusion with blood containing antibody to hepatitis B core antigen. The Veterans Administration Hepatitis Cooperative Study Group.New England Journal of Medicine 298, 1379-1383.
    14. Hughes, W., Barr, A., Dow, B. C., Follett, E. A. & Barbara, J. A. (1995). A multicentre assessment of the specificity of ten anti-HBc screening tests.Transfusion Medicine 5, 225-230.
    15. Jilg, W., Sieger, E., Zachoval, R. & Schätzl, H. (1995). Individuals with antibodies against hepatitis B core antigen as the only serological marker for hepatitis B infection: high percentage of carriers of hepatitis B and C virus.Journal of Hepatology 23, 14-20.
    16. Joller-Jemelka, H. I., Wicki, A. N. & Grob, P. J. (1994). Detection of HBs antigen in ‘anti-HBc alone’ positive sera.Journal of Hepatology 21, 269-272.
    17. Kann, M. & Gerlich, W. H. (1998). Hepatitis B. In Microbiology and Microbial Infections, vol. 1, Virology, pp. 745-773. Edited by L. Collier, A. Balows & M. Sussman. London: Topley & Wilson.
    18. Kohlstaedt, L. A., Wang, J., Friedman, J. M., Rice, P. A. & Steitz, T. A. (1992). Crystal structure at 3·5 Å resolution of HIV-1 reverse transcriptase complexed with an inhibitor.Science 256, 1783-1790.
    19. Kroes, A. C. M., Quint, W. G. V. & Heijtink, R. A. (1991). Significance of isolated hepatitis B core antibodies detected by enzyme immunoassay in a high risk population.Journal of Medical Virology 35, 96-100.
    20. Kwok, S. & Higuchi, R. (1989). Avoiding false positives with PCR.Nature 339, 237-238.
    21. Lee, D. S., Huh, K., Lee, E. H., Lee, D. H., Hong, K. S. & Sung, Y. C. (1997). HCV and HBV coexist in HBsAg-negative patients with HCV viraemia: possibility of coinfection in these patients must be considered in HBV-high endemic area.Journal of Gastroenterology and Hepatology 12, 855-861.
    22. Melegari, M., Bruno, S. & Wands, J. R. (1994). Properties of hepatitis B virus pre-S1 deletion mutants.Virology 199, 292-300.
    23. Melegari, M., Scaglioni, P. P. & Wands, J. R. (1998). Hepatitis B virus mutants associated with 3TC and famciclovir administration are replication defective.Hepatology 27, 628-633.
    24. Neifer, S., Molz, B., Sucker, U., Kreuzpaintner, E., Weinberger, K. & Jilg, W. (1997). Hoher Prozentsatz isoliert anti-HBc-positiver Personen unter Strafgefangenen.Gesundheitswesen 59, 409-412.
    25. Norder, H., Hammas, B., Larsen, J., Skaug, K. & Magnius, L. O. (1992). Detection of HBV DNA by PCR in serum from an HBsAg negative blood donor implicated in cases of post-transfusion hepatitis B.Archives of Virology Supplementum 4, 116-118.
    26. Norder, H., Hammas, B., Lee, S.-D., Bile, K., Couroucé, A.-M., Mushahwar, I. K. & Magnius, L. O. (1993). Genetic relatedness of hepatitis B viral strains of diverse geographical origin and natural variations in the primary structure of the surface antigen.Journal of General Virology 74, 1341-1348.
    27. Pichoud, C., Seignères, B., Wang, Z., Trépo, C. & Zoulim, F. (1999). Transient selection of a hepatitis B virus polymerase gene mutant associated with a decreased replication capacity and famciclovir resistance.Hepatology 29, 230-237.
    28. Protzer-Knolle, U., Naumann, U., Bartenschlager, R., Berg, T., Hopf, U., Meyer zum Büschenfelde, K. H., Neuhaus, P. & Gerken, G. (1998). Hepatitis B virus with antigenically altered hepatitis B surface antigen is selected by high-dose hepatitis B immune globulin after liver transplantation.Hepatology 27, 254-263.
    29. Sánchez-Quijano, A., Jauregui, J. I., Leal, M., Pineda, J. A., Castilla, A., Abad, M. A., Civeira, M. P., Garcia de Pesquera, F., Prieto, J. & Lissen, E. (1993). Hepatitis B virus occult infection in subjects with persistent isolated anti-HBc reactivity.Journal of Hepatology 17, 288-293.
    30. Schätzl, H. M., Sieger, E., Jilg, W., Nitschko, H. & Zachoval, R. (1997). Variability of the hepatitis B surface protein in HBV-infected liver transplant recipients.Journal of Biomedical Science 4, 146-154.
    31. Shih, C. M., Lo, S. J., Miyamura, T., Chen, S. Y. & Lee, Y. H. W. (1993). Suppression of hepatitis B virus expression and replication by hepatitis C virus core protein in HuH-7 cells.Journal of Virology 67, 5823-5832.
    32. Sumazaki, R., Motz, M., Wolf, H., Heinig, J., Jilg, W. & Deinhardt, F. (1989). Detection of hepatitis B virus in serum using amplification of viral DNA by means of the polymerase chain reaction.Journal of Medical Virology 27, 304-308.
    33. Thijssen, E. J., Kroes, A. C., Bos, E., Persijn, G. G. & Rothbarth, P. H. (1993). The significance of complete serological testing for hepatitis B in heart valve banking.Transplantation 56, 82-84.
    34. Tipples, G. A., Ma, M. M., Fischer, K. P., Bain, V. G., Kneteman, N. M. & Tyrrell, D. L. (1996). Mutation in HBV RNA-dependent DNA polymerase confers resistance to lamivudine in vivo.Hepatology 24, 714-717.
    35. Turner, D. P., Zuckerman, M., Alexander, G. J., Waite, J. & Wreghitt, T. (1997). Risk of inappropriate exclusion of organ donors by introduction of hepatitis B core antibody testing.Transplantation 63, 775-777.
    36. Waters, J. A., Kennedy, M., Voet, P., Hauser, P., Petre, J., Carman, W. & Thomas, H. C. (1992). Loss of the common ‘A’ determinant of hepatitis B surface antigen by a vaccine-induced escape mutant.Journal of Clinical Investigation 90, 2543-2547.
    37. Weinberger, K. M., Kreuzpaintner, E. A., Hottenträger, B., Neifer, S. & Jilg, W. (1997). Mutations in the S-gene of hepatitis B virus (HBV) isolates from chronic carriers with anti-HBc as the only serological marker of HBV infection. In Viral Hepatitis and Liver Disease, pp. 138-143. Edited by M. Rizzetto, R. H. Purcell, J. L. Gerin & G. Verme. Turin: Edizioni Minerva Medica.
    38. Weinberger, K. M., Zoulek, G., Bauer, T., Böhm, S. & Jilg, W. (1999). A novel deletion mutant of hepatitis B virus surface antigen.Journal of Medical Virology 58, 105-110.
    39. Weinberger, K. M., Wiedenmann, E., Böhm, S. & Jilg, W. (2000). Sensitive and accurate quantitation of hepatitis B virus DNA using a kinetic fluorescence detection system (TaqMan PCR).Journal of Virological Methods 85, 75-82.
    40. Zoulim, F. & Trépo, C. (1998). Drug therapy for chronic hepatitis B: antiviral efficacy and influence of hepatitis B virus polymerase mutations on the outcome of therapy.Journal of Hepatology 29, 151-168.