Role of immunohistochemistry for hepatitis D and hepatitis B virus in hepatitis delta

V I R A L H E P A T I T I S

Role of immunohistochemistry for hepatitis D and hepatitis B virus in

hepatitis delta

G€okhan Kabacßam1, Heiner Wedemeyer2, Berna Savasß3, Onur Keskin1, George Dalekos4, Fehmi Tabak5, Ramazan Idilman1, Andreas Erhardt6, Kendal Yalcßın7, Mithat A. Bozdayi1, Hakan Bozkaya1, Michael Manns2, Hans Dienes8 and Cihan Yurdaydın1_{for the HIDIT-1 Study Group*}

1 Department of Gastroenterology, University of Ankara Medical School, Ankara, Turkey

2 Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany 3 Department of Pathology, University of Ankara Medical School, Ankara, Turkey

4 Department of Internal Medicine (Academic Liver Unit and Research Laboratory of Internal Medicine), Medical School, University of Thessaly, Larissa, Greece

5 Department of Infectious Diseases, Cerrahpasa Medical School, University of Istanbul, Istanbul, Turkey

6 Department of Gastroenterology, Hepatology and Infectious Diseases, Heinrich Heine University, Dusseldorf, Gemany 7 Department of Gastroenterology, Medical Faculty, Dicle University, Diyarbakir, Turkey

8 Institute of Pathology, University of Cologne, Cologne, Germany

Keywords

Chronic Delta Hepatitis – Delta Antigen – HBcAg – HBsAg – Immunohistochemistry – Quantitative Surface Antigen

Abbreviations

AD, adefovir dipivoxil; ALT, alanine aminotransferase; AST, aspartate transaminase; cccDNA, covalently closed circular deoxyribonucleic acid; CDH, chronic delta hepatitis; CHB, chronic hepatitis B; GGT, gamma-glutamyl transferase; HAI, hepatic activity index; HBcAg, hepatitis B core antigen; HBeAg, hepatitis B early antigen; HBsAg, hepatitis B surface antigen; HBV DNA, hepatitis B virus deoxyribonucleic acid; HBV, hepatitis B virus; HDAg, hepatitis D antigen; HDV RNA, hepatitis delta virus ribonucleic acid; HDV, hepatitis delta virus; HIDIT 1, hep-net/international delta hepatitis intervention trial 1; IHC,

immunohistochemical; Peg-IFN, pegylated interferon; SVR, sustained virologic response. Correspondence

Professor Dr Cihan Yurdaydin, Department of Gastroenterology, University of Ankara

Abstract

Background & Aims: Immunohistochemical assessment of liver tissue in chronic delta hepatitis (CDH) is underinvestigated. Aim of the study was (i) to assess variables associated with hepatitis D antigen (HDAg), hepatitis B surface antigen (HBsAg) and hepatitis B core antigen (HBcAg) staining in the liver. Methods: Demographic, biochemical and virologic data collected from the HIDIT 1 study were used. HBsAg, HBcAg and HDAg immunohis-tochemical (IHC) staining was semiquantitatively assessed. Results: Hepati-tis D antigen immunohistochemical staining displayed positive correlations with age and alanine aminotransferase (ALT) and negative correlations with serum HBsAg (P = 0.01 for all). HBsAg IHC displayed a negative correlation with gamma glutamyl transferase and positive correlations with serum HBV DNA, serum HBsAg levels and HBeAg serology (P< 0.001, P = 0.02 and P = 0.007 respectively). HBcAg staining was mainly nuclear and displayed negative correlations with serum HBsAg and histologic activity (P= 0.002 and P = 0.02 respectively). Pegylated IFN based treatment led to a decline of all IHC markers, however, these markers had no impact on treatment out-come. Conclusions: These data suggest an association of liver injury with HDAg expression in CDH whereas the negative correlation between HBcAg expression and liver injury and the overall nuclear localization of HBcAg sug-gest that HBcAg does not contribute to liver injury in CDH. HDV cases with high level of HBV replication, high serum HBsAg levels, HBeAg positivity, that are probably in the earlier stages of disease (low gamma-glutamyl trans-ferase), had a more intense HBsAg staining profile. Overall, the data enforce the importance of HDAg and HBsAg in different phases of CDH infection.

Medical School, Cebeci Tip Fak€ultesi Hastanesi, Dikimevi, 06100 Ankara, Turkey Tel: +90 312 595 6110 Fax: +90 312 363 6213 e-mail: cihan.yurdaydin@medicine.ankara. edu.tr Received 18 August 2013 Accepted 31 October 2013 DOI:10.1111/liv.12376 Liver Int. 2014: 34: 1207–1215

*HIDIT-1 Study Group: T. Berg, W. B€ocher, H. Bohle, P. Buggisch, M. Cornberg, G. Ers€oz, S. Feyerabend, H. Hinrichsen, M. Manok, D. H€aussinger, W. Herzog, W.P. Hofmann, K. Plein,, H. Hardt, H. Porst, I. Mederacke, E.I. Rigopoulou, H. Sent€urk, E. K. v. Wagner.

Introduction

Chronic delta hepatitis (CDH) is the less prevalent but the most progressive form of chronic viral hepatitis. It continues to be a significant health problem not only in areas of the world where the disease is endemic but also in the western world where the once observed decline in CDH prevalence has been ceased largely because of migration (1, 2). The hepatitis delta virus (HDV) is an approximately 36 nm large particle containing HDV RNA and delta antigens, coated with the envelope pro-tein derived from the pre-S and S antigens of the Hepati-tis B virus (HBV). HepatiHepati-tis delta antigen (HDAg) is the only translational product of HDV. Immunohistochemi-cal (IHC) detection of HDAg has led to the discovery of HDV (3) and since then HDAg in liver tissue has been considered as a sensitive marker for the diagnosis of CDH (4, 5). HDV can propogate in humans only in the presence of the HBV. Thus, any study exploring delta hepatitis needs to take the HBV also into account. HDV infection is in general associated with suppression of HBV replication. This has been shown by a decrease in serum HBV DNA and hepatitis B surface antigen (HBsAg) levels in early chimpanzee experiments and also by IHC studies which showed a decline or disappearance of hepatitis B core antigen (HBcAg) in liver tissue (1, 6). Analysis of HBV and HDV IHC may potentially give clues to the pathogenesis of HDV infection but appears to be relatively underinvestigated (7). A recent study suggests that HDAg in liver tissue may have potential prognostic implications after liver transplantation (8) and availability of quantitative HBsAg determinations may further our understanding in viral pathogenesis as studied for chronic hepatitis B (CHB) (9). Currently, the only established evidence-based treatment of CDH is with conventional or pegylated interferons, effective in roughly a quarter of patients, although optimal duration, standardization of HDV RNA measurement and reliabil-ity of post-treatment viral response are issues which still need to be addressed in future studies (10, 11). In this study, variables associated with HDAg, HBsAg and HBcAg staining in the liver were assessed in CDH. Any potential clinical implications of these associations were investigated and finally, the effect of treatment on these parameters were validated.

Material and methods Patients

Data of patients who participated in the recently pub-lished HIDIT-1 Study were used (12). This study was conducted at Hannover Medical School, and at centres connected to Hannover Medical School through Hep-Net in Germany, at six university hospitals in Turkey and one hospital in Greece. Ethical approval is obtained through the local ethics committees of the participating centres. For detailed inclusion and

exclu-sion criteria of the HIDIT-1 Study, the reader is referred to the original article (12). Briefly, in the study, patients were randomized to receive pegylated interferon a 2a (Peg-IFN), 180 lg, qw + placebo (n = 29), Peg-IFN + adefovir dipivoxil (AD), 10 mg, qd (n= 32) or AD alone (n = 30) for 48 weeks. Per protocol, all patients had a baseline liver biopsy assess-ment. A second liver biopsy was performed at the end of the one year treatment period. Liver biopsy was assessed according to Ishak et al. (13). Baseline liver biopsies were used for assessment of the effects of several baseline demographic, haematological, bio-chemical, serologic and virologic parameters on HBsAg, HBcAg and HDAg expression in liver tissue. In 79 patients, liver tissue was available for performing these analyses. Baseline demographic, hematologic, biochemical, serologic and viral factors assessed for their potential affect on hepatitis viral protein expres-sion include patient age, sex, platelet count, prothrom-bin time, alanine aminotransferase (ALT), aspartate transaminase (AST), alkaline phosphatase, gamma-glutamyl transferase (GGT), total bilirubin, albumin levels, HBV DNA and HDV RNA viral loads, hepatitis e antigen (HBeAg) status, liver biopsy fibrosis and activity scores (HAI). The effect of treatment on HBsAg, HBcAg and HDAg expressions could be evaluated in 51 patients with available paired liver biopsies.

Determination of viral markers

Serum HBV DNA was quantified by Cobas TaqMan HBV test (Roche Molecular Systems, Inc., Branchburg, NJ, USA), and HDV RNA was determined by real-time PCR as described previously (14). Detection limit of the HDV RNA assay was 120 copies/ml. HBsAg was quanti-fied by Architect HBsAg assay (Abbott Diagnostics, Wiesbaden, Germany) according to the manufacturers’ instructions. This was a chemiluminescent microparticle immunoassay with sensitivity less than or equal to 0.05 IU/ml. Qualitative hepatitis serologies including HBsAg, HBsAb, HBeAg and HBeAb were determined by a microparticle enzyme immunoassay method (Abbott Laboratories, Abbott Park, IL, USA) and anti-HDV was determined by an enzyme immunoassay (Abbott Laboratories).

Definition of the viral dominant pattern

Chronic delta hepatitis was defined as HDV dominant when HDV RNA was less than or equal to four log10

copies/ml and HBV DNA was less than 2000 IU/ml. When HDV RNA was greater than or equal to four log10 copies/ml and HBV DNA was greater than

2000 IU/ml, this was considered as HDV-HBV co-dom-inant. HBV dominance was defined as HBV DNA greater than or equal to 2000 IU/ml and HDV RNA less than four log10copies/ml.

Immunohistochemical staining of HBsAg, HBcAg and HDAg in liver tissues

Immunohistochemical staining of liver tissues for hepa-titis B and delta antigens were performed in paraffin-embedded liver biopsies archived in the pathology departments of the participating centres of the HIDIT study. HBsAg and HBcAg staining of liver biopsies were performed as previously described (14). For the deter-mination of HBcAg and HBsAg expressions in liver tissue, a semi-quantitative scale based on percentage of HBcAg or HBsAg positive hepatocytes was applied: 0, less than 5%, 1+: 6–33%, 2+: 34–66%, 3+: >66%. For semiquantitative HDAg expression determination in liver tissue, HDAg IHC staining was performed with polyclonal rabbit antibodies to delta, kindly provided by John M. Cullen, VMD, Ph.D., from College of Veteri-nary Medicine at North Carolina State University, NC, USA. Polyclonal rabbit antibodies consisted of LP-2 antibody and LP-3 antibody. LP-3 antibody identifies only the large form of HDAg whereas LP-2 antibody identifies both the large and small forms of HDAg (15) and was therefore used to identify the presence of HDAg in liver tissue. HDAg positive hepatocytes among 500 adjacent hepatocytes was counted as previously described (15). Briefly, a 5-grade scale based on the pro-portion of HDAg positive hepatocytes was applied: 0: none; 1+: 1–10%; 2+: 11–25%; 3+: 26–50% and 4+: if ≥51% of hepatocytes had IHC staining for HDAg. Statistics

Continuous variables were compared using Student’s t test or Mann–Whitney U test where appropriate. For categoric variables, the Chi square/Fischer’s exact tests were used. Pearson’s correlation coefficient was used to describe the correlation between two continuous, nor-mally distributed variables. A Spearman’s correlation was used where variables were not normally distributed. Variables with P< 0.15 in the univariate models were included in the linear regression analysis. For paired comparisons paired Student’s t-test or Wilcoxon’s signed rank test were used where appropriate. Statistical analysis was performed bySPSSsoftware version 18 (SPSS

Inc., Chicago, IL, USA). A p-value less than 0.05 was considered as significant.

Results Baseline features

Of the 91 patients who participated in the HIDIT-1 Study, semi-quantitative IHC staining for HBsAg, HBcAg and HDAg could be performed in 79 patients. Baseline features of these patients are depicted in Table 1. All patients had genotype 1 HDV and the geno-type for HBV in patients where HBV genotyping was possible was D as reported elsewhere (12). Forty nine of

the patients were male and 30 were female with median age of 39. Eighteen per cent of patients were HBeAg positive. There was HDV dominance in 69%, HDV-HBV codominance in 12% and HDV-HBV dominance in 3% of patients. The rest of the patients had HDV RNA levels which were less than 4 log10copies/ml and HBV DNA

<2000 IU/ml, and were considered low viremic where none of the viruses appeared dominant or where HDV dominance was observed at a lower level. This group of patients was not further analyzed as a subgroup.

Weak negative correlations existed between HDV-dominance and HBeAg positive serology and sustained virologic response (SVR; virologic response 6 months post-treatment) (r = 0.274, P = 0.017 and r = 0.343, P = 0.024, respectively, Table 2). In patients with HDV-HBV codominance, there was a positive cor-relation with HBeAg positivity (r = 0.377, P = 0.001) and HBsAg immunostaining frequency (r = 0.305, P = 0.011). No correlation analyses were performed for HBV-dominant cases because only two patients fell into this category.

Immunostaining for HDAg, HBsAg and HBcAg

Liver samples of CDH patients stained frequently posi-tive for HBsAg (Fig. 1A) and to a lesser extent for HBcAg (Fig. 1B). Strong HBsAg staining (2–3+ immu-nostaining) was observed in 32.9% of patients whereas

Table 1. Baseline demographic, serologic, biochemical and histo-logic features of patients (Data are given as median [range] if not otherwise indicated

Variable Result

Age 39 (17–62)

Sex (male/ female) 49/30

ALT (IU/ml)n = 79 91 (11–641)

AST (IU/ml)n = 79 66 (19–324)

Alkaline Phosphatase (IU/ml)n = 79 97 (47–364)

GGTn = 74 43.5 (13–497)

Albuminn = 67 4.1 (3.2–6.1)

Platelets (9109

/ml)n = 79 165 (74–344)

Log10HBV DNA (copies/ml)n = 75 2.08 (u.d.–8.04) Log10HDV RNA (copies/ml)n = 74 5.78 (u.d.–7.92) Log10HBsAg (IU/ml)n = 75 4.12 (1.83–4.90)

HBeAg (+) (%) 14 (17.7)

HBeAg ( ) (%) 65 (82.3)

Fibrosis score (according to Ishak et al11_{)n = 77 3 (0–6)} Histologic activity indexn = 77 7 (3–11)

HDV dominant* (%) 51 (69)

HDV-HBV codominant** (%) 9 (12)

HBV dominant#_{(%) 2 (3)}

ALT= alanine aminotransferase; AST= aspartate transaminase; GGT= gamma-glutamyl transferase; HDV dominant = HDV RNA ≥104 log, copies/ml and HBV DNA<104_{log copies/ml; HDV-HBV} co-domi-nant= HDV RNA ≥104_{log copies/ml and HBV DNA≥10}4_{log copies/ml;} HBV dominant= HDV RNA <104_{log, copies/ml and HBV DNA≥10}4_log copies/ml).

strong HBcAg staining was seen in only 3.8% (HBsAg vs. HBcAg, P < 0.001). Importantly, HDAg staining was observed in all samples where HDV immunochemical staining was applied (Fig. 1C). It was always associated with HBsAg staining but HBcAg staining was absent in 11 cases.

Hepatitis D antigen expression was seen mainly in hepatocyte nuclei (Fig. 1C) whereas HBsAg expression was observed in the cytoplasm and plasma membrane.

HBcAg staining was mostly in the nuclei of hepatocytes, sometimes in cases of high replication also in the cyto-plasm. However, out of 14 patients with baseline high HBV DNA load (≥4 log10 copies/ml), the majority (10

patients) had no HBcAg staining. Thus HBcAg staining did not correlate with HBV replication in hepatitis delta.

Factors affecting HDAg expression in liver tissue

Factors assessed for their potential effect on HDAg expression in liver tissue are given in Table 3. HDAg IHC displayed positive correlations with age (r= 0.289, P = 0.01) (Fig. 2A) and ALT (r = 0.304, P = 0.01) (Fig. 2B) and negative correlations with bilirubin (r = 0.298, P = 0.01) (Fig. 2C), and serum HBsAg (r = 0.315, P = 0.01) (Fig. 2D). Importantly no correlation existed between serum HDV RNA and HDAg expression frequency in liver tissue. By multivariate regression analysis, only serum HBsAg remained as an independent parameter affect-ing HDAg expression (95% CI: ( )25.3 to ( )0.53, P = 0.041). Except bilirubin, all other parameters (age, ALT and serum HBsAg) displayed significant correlations with HDAg expression in liver tissue also in HDV-dominant cases (r= 0.32, P= 0.037, r = 0.308, P = 0.044, r = 0.393, P = 0.009 respec-tively). In the co-dominant cases lower platelet levels were associated with HDAg staining (r = 0.847, P = 0.016).

Factors affecting HBsAg expression in liver tissue

Univariate correlation analyses of factors assessed for their potential effect on HBsAg expression in liver tissue are given in Table 3. A negative correlation existed between GGT and HBsAg expression in liver tissue (r= 0.303, P = 0,009) (Table 3). A positive correla-tion was evident between HBV DNA (r= 0.379, P < 0.001), serum quantitative HBsAg level (r = 0.266, P = 0.02), HBeAg positivity (r = 0.304, P = 0.007) and

Table 2. Correlation of the hepatitis delta virus dominant pattern and hepatitis delta virus hepatitis B virus codominant pattern (see text for explanation) with several baseline parameters and treat-ment outcome Variable Hepatitis D virus dominant pattern Hepatitis D&B codominant pattern Age r = 0.158 p = 0.171 p = 0.059r = 0.216 Sex r = 0.215 p = 0.060 p = 0.331r = 0.112 Platelets r = 0.032 p = 0.786 p = 0.368r = 0.104 ALT r = 0.098 p = 0.395 p = 0.917r = 0.012 GGT r = 0.071 p = 0.566 p = 0.680r = 0.051 HBsAg (log) r = 0.068 p = 0.557 p = 0.329r = 0.113 HBeAg r = 0.274 p = 0.017 p = 0.001r = 0.377 HAI r = 0.032 p = 0.790 p = 0.151r = 0.173 Fibrosis Score r = 0.034 p = 0.782 p = 0.407r = 0.101 IHC HBsAg r = 0.158 p = 0.198 p = 0.011r = 0.305 IHC HBcAg r = 0.064 p = 0.603 p = 0.254r = 0.140 IHC HDAg r = 0.034 p = 0.801 p = 0.889r = 0.017 SVR 72 r = 0.343 p = 0.024 p = 0.662r = 0.069 (A) (B) (C)

Fig. 1. Representative immunohistochemical staining of HBsAg (Fig. 1A), HBcAg (Fig. 1B) and HDAg (Fig. 1C) of hepatocytes of patients with delta hepatitis. HDAg was detected in the nuclei of hepatocytes in every patient; HBsAg staining was mainly cytoplasmic and observed in the majority of patients; HBcAg staining was mainly observed in the nuclei and there was rarely intense staining as can be seen on this representative illustration.

HBsAg immunostaining. By multivariate linear logistic regression analysis, only serum HBV DNA was an independent factor of HBsAg expression (95% CI= 0.049–0.166, P = 0.001). In addition serum HBsAg level displayed borderline significance (95% CI= ( )0.006 to 0.387, P = 0.057). In purely HDV-dominant cases, the negative correlation between GGT and HBsAg expression in liver tissue persisted (r= 0.331, P = 0.04). In addition, there was a nega-tive correlation also with AST (r= 0.304, P = 0.045). Factors affecting HBcAg expression in liver tissue: Univariate correlation analyses of factors assessed for their potential effect on HBcAg expression in liver tissue are given in Table 3. There were negative corre-lations with serum HBsAg (r= 0.351, P = 0.002) and the level of inflammation (HAI) (r = 0.257, P = 0.02). By linear logistic regression analysis, serum HBsAg remained significant (95% CI = ( )0.422 to

( )0.04, P:0.019). In addition, HBsAg staining fre-quency of hepatocytes also became significant (95% CI = 0.067–0.491, P = 0.011). In HDV-dominant cases, only the negative correlation with serum HBsAg persisted (r = 0.319, P = 0.035). No signifi-cant correlations existed for HDV-HBV co-infected patients.

Serum HBV DNA vs serum HBsAg and HBsAg IHC: Patients with high HBV DNA (>2000 IU/ml) displayed a trend for higher serum HBsAg levels compared to patients with low HBV DNA (<2000 IU/ml) (4.18 log10± 0.34 [±SD] IU/ml vs. 3.85 ± 0.65,

P = 0.07, Fig. 3A). In line with this, patients with high viral load had more frequent HBsAg IHC staining when compared with patients with low viral load (2 ± 0.71 [±SD] vs. 1.23 ± 0.42, P < 0.0001). However, when serum HBsAg levels were adjusted for viral load, the relationship was reversed (median 0.87 vs. 2.50, P < 0.0001, Fig. 3B). The same happened when HBsAg IHC staining frequency was also adjusted for viral load (2.39 ± 1.04 [±SD] vs. 3.40 ± 0.94 (P = 0.0009, Fig. 3C) suggesting that HBsAg production was pre-served in the typical CDH patient with suppressed HBV viral load.

Effect of treatment on hepatitis delta and hepatitis B protein expression in liver tissue:

Sustained virologic response rates were not different between treatment with Peg-IFN-AD and Peg-IFN monotherapy as previously reported (12). Treatment with Peg-IFN based regimens decreased HDAg in liver tissue from 45.7 ± 33 to 16.32 ± 27(P < 0.0001). HBsAg and HBcAg staining semi-quantitative scoring decreased from 1.4 ± 0.6 to 0.9 ± 0.6 (P = 0.004), and from 0.9 ± 0.3 to 0.5 ± 0.7(P = 0.001) respectively (Table 4). There was no relationship between the SVR in 72nd week with the decline of the three IHC parame-ters. When Peg-IFN-AD treated patients where com-pared with Peg-IFN monotherapy patients, no difference in the delta decline of HDAg-IHC, HBsAg-IHC and HBcAg-HBsAg-IHC between baseline and end-of treatment liver biopsy samples was observed (30.8 ± 33.6 [n = 16] vs. 28.0 ± 41.7 [n = 15], P:0.84; 0.54 ± 0.88 [n = 13] vs. 0.44 ± 0.81 [n = 16], P:0.85; 0.20 ± 0.77 [n = 15] vs. 0.19 ± 0.81 [n = 21], P:0.97 respectively). None of the patients treated with AD had developed a sustained virologic response (12). AD had no effect on HBsAg and HBcAg expression in liver tissue, although HDV Ag expression decline achieved borderline significance (P = 0.041).

Of the 17 patients who had a virologic response at 6 months post-treatment and were considered sustained virologic responders, nine patients had a virologic relapse on long-term follow-up lasting up to 5 years (11), which was documented both in the local

labora-Table 3. Factors affecting HBsAg, HBcAg, and HDAg expression in liver tissue Variables Factors affecting HBsAg expression Factors affecting HBcAg expression Factors affecting HDAg expression Age r = 0.160 p = 0.158 p = 0.650r = 0.052 p = 0.012r = 0.289 Sex r = 0.049 p = 0.669 p = 0.459r = 0.085 p = 0.954r = 0.007 Platelets r = 0.024 p = 0.833 p = 0.756r = 0.035 p = 0.975r = 0.004 ALT r = 0.153 p = 0.177 p = 0.889r = 0.016 p = 0.010r = 0.304 GGT r = 0.303 p = 0.009 p = 0.958r = 0.006 p = 0.886r = 0.018 Bilirubin (mg/ dL) r = 0.094 p = 0.412 p = 0.760r = 0.035 p = 0.012r = 0.298 HBsAg (log) r = 0.266 p = 0.021 p = 0.002r = 0.351 p = 0.010r = 0.315 HBeAg r = 0.304 p = 0.007 p = 0.182r = 0.153 p = 0.112r = 0.190 HAI r = 0.045 p = 0.700 p = 0.024r = 0.257 p = 0.165r = 0.168 Fibrosis Score r = 0.060 p = 0.605 p = 0.927r = 0.011 p = 0.681r = 0.050 HBV DNA r = 0.379 p < 0.001 p = 0.235r = 0.139 p = 0.986r = -0.002 HDV RNA r = 0.094 p = 0.426 p = 0.938r = 0.009 p = 0.585r = 0.069 IHC HBsAg – r = 0.190 p = 0.094 p = 0.932r = 0.011 IHC HBcAg r = 0.190 p = 0.094 – p = 0.932r = 0.011 IHC HDAg r = 0.011 p = 0.932 p = 0.980r = 0.003 – SVR 72 r = 0.227 p = 0.144 p = 0.326r = 0.153 p = 0.112r = 0.246

tory and the central laboratory where HDV RNA deter-minations in the original study were performed. These nine patients were compared with the eight patients who continued to have a virologic response. Seven of nine patients who had a virologic relapse had an end of treatment liver biopsy with IHC analysis. In six of them, there was positive HDAg staining in liver tissue. End of treatment liver tissue for immunochemical staining was available in five of eight patients with continued virologic response and HDAg IHC positive staining was evident in 4.

Discussion

This study provides the first detailed description of HBsAg and HBcAg and HDAg IHC staining in liver tis-sue in a large cohort of patients with CDH. Expression patterns of HBV and HDV proteins in the liver have been studied before (6, 7, 16) but not in a large patient cohort, and in those studies factors affecting HBV and HDV protein expression in liver tissue had not been

explored in detail. This analysis has confirmed that HDAg expression was mainly nuclear in location (1, 2, 16), that HBsAg staining was mainly cytoplasmic and almost always associated with HDAg expression and that HBcAg expression was less frequently observed than HDAg and HBsAg staining in patients with CDH (16). Further, we have provided evidence that: (i) HDAg IHC displayed positive correlations with ALT and HBsAg; (ii) HBcAg does not contribute to liver injury in CDH; (iii) HBsAg-IHC staining is more prominent in the early stages of liver disease with high HBV replication and finally (iv) HBsAg production is pre-served in CDH where most cases are known to have low levels of HBV DNA .

This study explored the effect of several baseline fac-tors on HBsAg, HBcAg and HDAg expression in liver tissue. HDAg IHC displayed positive correlations with age, ALT and serum HBsAg. Further, there was a negative correlation between bilirubin levels and HDAg immunochemical staining. Age, ALT and serum HBsAg displayed significant correlations with HDAg expression

(A) (B)

(D) (C)

Fig. 2. Correlations between HDAg IHC with age (Fig. 2A) (r = 0.289, P = 0.01), ALT (Fig. 2B) (r = 0.304, P = 0.01), serum bilirubin (Fig. 2C) (r = 0.298, P = 0.01) and serum quantitative HBsAg levels (Fig. 2D) (r = 0.315, P = 0.01).

in liver tissue also in HDV-dominant cases. These data suggest an association of liver injury with HDAg expres-sion in CDH, especially in HDV-dominant CDH. We have previously reported a positive correlation between serum HBsAg and serum HDV RNA (14). In this analy-sis, serum HBsAg did correlate with HDAg but there was no correlation between serum HDV RNA and HDAg expression frequency in liver tissue. HDAg in

liver tissue has long been considered as a sensitive mar-ker of HDV infection; HBsAg on the other hand, repre-sents the only product of HBV needed by HDV to be infective in humans. The positive correlation of HDAg, ALT and serum HBsAg emphasizes the importance of two essential factors for liver injury in CDH infection in humans, namely, HBsAg and HDAg. The negative cor-relation of bilirubin may point to less advanced disease.

2

Low HBV DNA High HBV DNA

+ 1.5 2 2.5 3 3.5 4 4.5 5 2.5 3.5 4.5 P: 0.07 3

Log HBsAg (IU/ml)

HBsAg IHC score / HBV DNA (IU/ml)

Log HBsAg titre (IU/ml) / Log HBV DNA (IU/ml)

4 5 1 2 3 4 5 6 7 8 +

Low HBV DNA High HBV DNA

Low HBV DNA High HBV DNA

P < 0.0001

P: 0.0009

(A) (B)

(C)

Fig. 3. Serum HBsAg levels and semi-quantitative HBsAg IHC staining scores in CDH patients with low level serum HBV DNA (<2000 IU/ml, n:14) versus high level serum HBV DNA (≥2000 IU/ml, n:72). In Fig. 3A serum HBsAg levels in patients with low versus high level serum HBV DNA is depicted. In Fig. 3B serum HBsAg levels in patients with low versus high level serum HBV DNA after adjusting for HBV viral load is depicted and in Fig. 3C semi-quantitative HBsAg IHC staining scores in patients with low versus high level serum HBV DNA after adjusting for HBV viral load is shown.

Table 4. Change of the immunohistochemical variables with PegIFN based treatment

IHC variable

Pegylated IFN-based treatment Adefovir treatment

Baseline End of treatment P-value Baseline End of treatment p-value

HDAg IHC* 45.7± 33 16.3± 27 <0.0001 27.2± 30.8 14.9± 20.9 0.041

HBsAg IHC† 1.4± 0.6 0.9± 0.6 0.004 1.40± 0.60 1.50± 0.69 0.61

HBcAg IHC† 0.9± 0.3 0.5± 0.7 0.001 0.68± 0.72 0.91± 0.61 0.17

*HDAg expression is given as the mean [± SD] number of cells stained within the investigated 500 cells during semi-quantitative assessment. †HBsAg and HBcAg expressions are given as the mean semi-quantitative scale scores [±SD].

Early studies have suggested that HDAg staining frequency would diminish with more advanced disease (17). However, we were unable to observe a change in the frequency of HDAg staining in patients with advanced disease or documented cirrhosis (data not shown). The lack of a correlation of serum HDV RNA with HDAg staining is in line with the suggestion that genotype 1 HDV is not cytopathic but rather leads to immune-mediated liver injury (18).

With regard to HBsAg expression in liver tissue, a negative correlation existed between GGT and HBsAg expression and a positive correlation existed between serum HBV DNA, serum quantitative HBsAg level, HBeAg positivity and HBsAg immunostaining. Thus, HDV cases with high level of HBV replication, high serum HBsAg levels, HBeAg positivity, that are probably in the earlier stages of disease (low GGT), had a more intense HBsAg staining profile. We have previously reported that HBeAg-positive patients are characterized with higher HBV DNA, ALT and HBsAg levels and are younger when compared with HBeAg-negative CDH patients (19) and our current observation is in line with those data. Further, in an analysis of the HIDIT-I Study, HBeAg-positive CDH cases were also found to have sig-nificantly higher HBV DNA levels when compared with HBeAg-negative CDH cases (20). Our data to some extent resemble data from patients with CHB mono-infection, where HBV replication correlated with HBsAg immunochemical staining in HBeAg-positive CHB patients but not in HBeAg-negative CHB (9). However, in HBeAg-negative CHB despite low level of HBV repli-cation HBsAg production appeared to be preserved (9). Similarly, in this study, patients with high HBV DNA appeared to have higher HBsAg levels and more fre-quent HBsAg staining in hepatocytes when compared with CDH patients with low level HBV viremia. How-ever, when HBsAg levels were adjusted for HBV viremia the above mentioned relationship was reversed which suggested preserved HBsAg production in CDH patients with low level viremia. As this group represent the majority of patients with CDH, this information is of importance. In the CHB mono-infection study, Thomp-son et al. suggested integrated HBV as a likely source of preserved HBsAg production in HBeAg-negative CHB (9). This type of reasoning was questioned because inte-gration resulting in independent HBsAg production would be expected to occur through clonal expansion of the HBV region coding for the envelop proteins which is expected to occur in tumoural rather than non-tumo-ural tissue (21). In the setting of CDH, HBV integration into the host genome appears not to have been studied so far. However, Pollicino et al. have elegantly docu-mented a dissociation between pregenomic (pg)RNA and pre-S/S RNA production in patients with CDH in the context that per cccDNA molecule more pre-S/S RNA was produced in HDV when compared with HBV patients which may account for preserved HBsAg pro-duction in the setting of CDH although the molecular

mechanism(s) remain unknown (21). But at least such a mechanism appears to be biologically reasonable as the transcription of pg RNA is under the control of two different HBV genomic regulatory regions (21, 22).

Hepatitis B core antigen expression displayed a sig-nificant albeit weak negative correlation with liver injury. In hepatitis B mono-infection, most studies have linked nuclear HBcAg expression to viral replication and cytoplasmic HBcAg expression to liver injury (23, 24). Our finding of a negative correlation between HBcAg expression and liver injury and the overall nuclear localization of HBcAg suggest that HBcAg does not contribute to liver injury in CDH. On the other hand, in HDV-HBV co-dominant CDH, there was a trend for a significant negative correlation between HBcAg expression frequency and GGT and AST levels. This suggests that even in the setting of HDV-HBV co-dominant CDH, HBcAg expression is not linked to liver injury. It would have been interesting to explore ef-fectors of HBcAg expression in patients with HBV-dom-inant CDH cases. However, with only two such patients this analysis could not be performed in this study.

In this study, we explored differences between patients who according to our arbitrary chosen definition had HDV-dominant versus HDV-HBV-co-dominant CDH. HDV-dominant CDH cases were more likely to have HBeAg-negative serology and were less likely to have an SVR, defined by HDV RNA negativity at week 24 post-treatment. On the other hand, HDV-HBV codominant CDH patients were more likely to have HBeAg-positive serology and to be younger. They also showed a positive correlation with HBsAg staining frequency of hepato-cytes. These data were in line with our previously reported data that HBeAg-positive CDH patients were younger when compared with HBeAg-negative CDH (19) and suggested that in the early stages of CDH dis-ease HBV contributed to the disdis-ease process. HDV-dom-inant CDH were more likely to have HBeAg-negative serology and thus may be at a later stage of CDH (20) which may account for its negative impact on SVR.

Treatment with Peg-IFN led to significant decreases in HDAg, HBsAg and HBcAg staining frequency of hepato-cytes, the major decrease being observed with HDAg immunochemical staining. Treatment with adefovir had no effect on HBsAg and HBcAg expression, however, HDAg expression showed a decrease with borderline sig-nificance. Nucleoside analog-treatment-induced decline of HDAg staining frequency in hepatocytes has been reported before (15), however, in that study as well the nucleoside analog used was overall ineffective and the issue of sampling error cannot be ruled out. The data are in line with interferon being the only effective drug in the management of CDH. Nine patients who originally had a virologic response six months post-treatment had a viro-logic relapse on longer follow-up. This raises the issue of the reliability of the six months post-treatment virologic response as ‘sustained response’ in CDH (10). IHC HDAg staining did not contribute for prediction of these

‘late relapsers.’ The vast majority of relapsers as well as non-relapsers continued to have positive HDAg staining at end of treatment liver biopsy. This is probably because of the sensitivity of HDAg expression in liver tissues, as we have observed recently in the liver transplant setting (8).

In conclusion, in this study a detailed description of factors affecting HBsAg, HBcAg and HDAg IHC has been provided and discussed. Overall, the data provided enforce the importance of HDAg and HBsAg in differ-ent phases of CDH infection. It is important to note that most correlations were weak which suggests a complex interplay between HDV as well as HBV epitopes in the immune pathogenesis of this disease. Further, we have provided evidence for preserved HBsAg production in the typical CDH patient with low level HBV viremia. Acknowledgements

Financial support: The original trial received financial support from F. Hoffmann-La Roche and Gilead Sciences who approved the study design as proposed by the investigators. Current Controlled Trials number: ISRCTN83587695.

Conflict of interest: Dr Cihan Yurdaydin has been in the advisory boards of Merck Pharma, Janssen Pharma and Gilead Pharma and in the Speakers bureau of Roche Pharma and Gilead Pharma. George Dalekos is in the speakers bureau of Gilead. Michael Manns has been in the advisory boards of Schering Plough, Roche, BMS, Gilead, Valeant, Boehringer Ingelheim, Novartis, Idenix, Tibotec, Vertex, GlaxoSmith Kline, and in the speakers bureau of Roche, BMS, GlaxoSmithKline, Gilead. Hein-er WedemeyHein-er has been in the advisory boards of Roche, Gilead, and in the Speakers bureau of Roche and Gilead. Other authors have nothing to declare.

References

1. Yurdaydin C, Idilman R, Bozkaya H, Bozdayi AM. Natural history and treatment of chronic delta hepatitis. J Viral Hepat 2010;17: 749–56.

2. Wedemeyer H, Manns MP. Epidemiology, pathogenesis and management of hepatitis D: update and challenges ahead. Nat Rev Gastroenterol Hepatol 2010;7: 31–40. 3. Rizzetto M, Canese MG, Arico S, Crivelli O, Trepo C,

Bonino F, Verme G. Immunofluorescence detection of new antigen-antibody system (d/anti-d) associated to hepatitis B virus in liver and in serum of HBsAg carriers. Gut 1977;18: 997–1003.

4. Negro F, Bonino F, Di Bisceglie A, Hoofnagle JH, Gerin JL. Intrahepatic markers of hepatitis delta virus infection: a study by in situ hybridization. Hepatology 1989;10: 916– 20.

5. Cullen JM, David C, Wang JG, Becherer P, Lemon SM. Subcellular distribution of large and small hepatitis delta antigen in hepatocytes of hepatitis delta virus superinfect-ed woodchucks. Hepatology 1995;22: 1090–100.

6. Rizzetto M, Canese MG, Gerin JL, London WT, Sly DL, Purcell RH. Transmission of hepatitis B virus-associated

delta antigen in chimpanzees. J Infect Dis 1980;141: 590– 602.

7. Ryley NG, Heryet AR, Goldin R, Monjardino J, Saldanha J, Fleming KA. Co-expression of markers for hepatitis delta and hepatitis B viruses in human liver. Histopathol-ogy 1992;20: 331–7.

8. Mederacke I, Filmann N, Yurdaydin C, et al. Rapid early HDV RNA decline in the peripheral blood but prolonged intrahepatic hepatitis delta antigen persistence after liver transplantation. J Hepatol 2012;56: 115–22.

9. Thompson AJV, Nguyen T, Iser D, et al. Serum hepatitis B surface antigen and hepatitis B e antigen titers: disease phase influences correlation with viral load and intrahe-patic hepatitis B virus markers. Hepatology 2010; 51: 1933–44.

10. Yurdaydin C. Treatment of chronic delta hepatitis. Sem Liver Dis 2012;32: 237–44.

11. Heidrich B, Yurdaydin C, Kabacam G, et al. Long-term follow-up after Peg-IFNa2a based therapy of chronic hepa-titis delta (abstr.). J Hepatol 2013;58: S20.

12. Wedemeyer H, Yurdaydin C, Dalekos GN, et al. Peginter-feron plus adefovir versus either drug alone for hepatitis delta. N Engl J Med 2011;364: 322–331.

13. Ishak K, Baptista A, Bianchi I, et al. Histological grading and staging of chronic hepatitis. J Hepatol 1995;22: 696–9. 14. Zacahou K, Yurdaydin C, Drebber U, et al. Quantitative HBsAg and HDV RNA levels in chronic delta hepatitis. Liver Int 2010;30: 430–7.

15. Yurdaydin C, Bozkaya H, G€urel S, et al. Famciclovir treat-ment of chronic delta hepatitis. J Hepatol 2002;37: 266–71. 16. Chu CM, Liaw YF. Intrahepatic expression of hepatitis B core and surface antigens in chronic hepatitis delta-virus infection. J Hepatol 1992;16: 153–8.

17. Rizzetto M, Verme G, Recchia S, et al. Chronic hepatitis in carriers of hepatitis B surface antigen with intrahepatic expression of the delta antigen. An active and progressive disease unresponsive to immunosuppressive treatment. Ann Intern Med 1983;98: 437–41.

18. Grabowski J, Wedemeyer H. Hepatitis delta: immuno-pathogenesis and clinical challenges. Dig Dis 2010; 28: 133–8.

19. Heidrich B, Serrano BC, Idilman R, et al. HBeAg-positive hepatitis delta: virological patterns and clinical long-term outcome. Liver Int 2012;32: 1415–25.

20. Kabacßam G, Dalekos GN, Cßakaloglu Y, et al. Pegylated interferon-based treatment in patients with advanced liver disease due to chronic delta hepatitis. Turk J Gastroenterol 2012;23: 560–8.

21. Pollicino T, Raffa G, Santantonio T, et al. Replicative and transcriptional activities of hepatitis B virus in patients coinfected with hepatitis B and hepatitis delta viruses. J Virol 2011;85: 432–9.

22. Seeger C, Mason WS. Hepatitis B virus biology. Microbiol Mol Biol Rev 2000;64: 51–68.

23. Chu CM, Yeh CT, Chien RN, Sheen LS, Liaw YF. The degrees of hepatocyte nuclear but not cytoplasmic expression of hepatitis B core antigen reflect the level of viral replication in chronic hepatitis B infection. J Clin Microbiol 1997;35: 102–5.

24. Serin€oz E, Varlı M, Erden E, et al. Nuclear localization of hepatitis core antigen and its relations to liver injury, hepatocyte proliferation and viral load. J Clin Gastroenterol 2003;36: 269–72.