Programmed death 1 expression during antiviral treatment of chronic hepatitis B: Impact of hepatitis B e-antigen seroconversion

Authors


  • Potential conflict of interest: Dr. Naoumov received grants from Idenix Pharmaceuticals.

Abstract

Hyperexpression of the programmed death 1 (PD-1) molecule is a hallmark of exhausted T-cells, having a negative impact on T-cell activation and function. We studied longitudinally 18 hepatitis B e antigen (HBeAg)–positive patients undergoing treatment with direct antivirals (telbivudine or lamivudine) to determine the relationship between treatment-induced viremia reduction and HBeAg seroconversion with respect to PD-1 levels and T-cell reactivity. PD-1 expression was assessed by (1) flow cytometry and (2) quantitative real-time polymerase chain reaction; hepatitis B virus (HBV)–specific CD8+ T-cells were quantitated by pentamer staining; T-cell reactivity to HBV antigens was determined by interferon gamma (IFNγ) and interleukin 10 (IL-10) enzyme-linked immunosorbent spot (ELISPOT) assays; and central/effector memory phenotypes were defined by phenotypic markers. PD-1 expression correlated closely with viremia levels. On therapy, PD-1 decreased significantly on total CD8+ T-cells, HBV-specific CD8+ T-cells, and CD3+/CD8− T-cells both as the percentage of positive cells (P < 0.01) and as the mean fluorescent intensity (P < 0.05), and this was paralleled by a marked reduction of PD-1 messenger RNA levels (P = 0.001). HBeAg serocoversion (in 6/18 patients) resulted in a further PD-1 decrease with a 50% reduction in the frequency of PD-1+/CD8+ T-cells, which was not observed in patients remaining HBeAg-positive. The decrease in PD-1 expression was associated with increased frequencies of IFNγ-producing T-cells and decreased frequencies of IL-10 producing T-cells. At baseline, PD-1 expression correlated directly with the frequency of hepatitis B core antigen (HBcAg) central and effector memory phenotypes, whereas an inverse correlation was observed between PD-1 expression and HBcAg-specific effector phenotypes. Conclusion: These results demonstrate that in chronic HBV infection, both viremia levels and HBeAg drive PD-1 expression and resulting T-cell impairment. Treatment-induced suppression of HBV replication reduces PD-1 expression; however, additional immunotherapeutic interventions are needed for restoration of T-cell functions. (HEPATOLOGY 2008.)

Following exposure to the hepatitis B virus (HBV), the adaptive immune response plays a central role in determining whether chronicity of infection will occur.1 CD4+ and CD8+ T-cell responses to HBV are stronger in patients who spontaneously resolve HBV infection, whereas failure to mount vigorous T-cell responses, directed at multiple epitopes, is the dominant cause of chronicity of viral infections in humans in general and of persistent HBV replication.2–9

Recent data have identified the programmed cell death pathway [programmed death 1 (PD-1)/programmed death ligand 1 (PD-L1)/PD-L2] as a possible mechanism by which CD8+ T-cells are inactivated, thus promoting viral persistence. In a mouse model with chronic lymphocytic choriomeningitis virus (LCMV) infection, a significant up-regulation of PD-1 expression was detected on exhausted virus-specific CD8+ T-cells in comparison with functional LCMV-specific CD8+ T-cells.10 Blockade of PD-1 expression on exhausted CD8+ T-cells resulted in restoration of CD8+ T-cell functions, with increased proliferation, cytotoxicity, and cytokine production.10 A similar relationship between PD-1 expression and virus-specific T-cell reactivity was observed in human immunodeficiency virus (HIV) infection.11–13 PD-1 expression was also found to affect T-cell reactivity and consequently the outcome, that is, resolution of infection or viral persistence, in patients with acute hepatitis B or C.14, 15

In chronic HBV infection, virus-specific T-cells are functionally impaired, but the relative role of viremia (HBV-DNA levels) and/or hepatitis B e antigen (HBeAg) in the impairment of T-cell reactivity has not been defined. The biological function of HBeAg is not fully understood, as it is not required for virus assembly, infection, or replication.16 HBeAg and the viral nucleocapsid, hepatitis B core antigen (HBcAg), share >90% of amino acids and have been found to be cross-reactive at the T-cell level.17 As HBeAg is secreted from hepatocytes, it is thought that it has a central role as a tolerogen during HBV infection. This has been demonstrated in vertical HBV transmission, in which secreted HBeAg appears to be central to the induction of immunological tolerance in utero.18 Given the role of PD-1 as an important inhibitory pathway in T-cell function, the relative expression of PD-1 before and after seroconversion in the absence of changes in HBV-DNA may provide useful information concerning the mechanism of tolerance induced by HBeAg.

To gain further understanding of the role of PD-1 in chronic HBV infection, we investigated longitudinally the relationship between PD-1 expression, viral load, and HBeAg in patients with chronic hepatitis B undergoing treatment with oral antiviral agents. We also examined the impact of changes in the viral load and PD-1 levels on the frequency of virus-specific T-cells producing interferon gamma (IFNγ) and interleukin 10 (IL-10). In addition, we characterized the memory phenotype of the T-cells during treatment, particularly before and after HBeAg seroconversion.

Abbreviations

ALT, alanine aminotransferase; anti-HBe, hepatitis B e antibody; APC, allophycocyanin; CMV, cytomegalovirus; Cy7, cyanin 7; ELISPOT, enzyme-linked immunosorbent spot; FBS, fetal bovine serum; HBc, hepatitis B core; HBcAg, hepatitis B core antigen; HBeAg, hepatitis B e antigen; HBs, hepatitis B surface; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCV, hepatitis C virus; HIV, human immunodeficiency virus; HLA, human leukocyte antigen; IFNγ, interferon gamma; IL-10, interleukin 10; LCMV, lymphocytic choriomeningitic virus; MFI, mean fluorescent intensity; mRNA, messenger RNA; nd, not determined; PBMC, peripheral blood mononuclear cell; PBS, phosphate-buffered saline; PD-1, programmed death 1; PD-L1, programmed death ligand 1; PE, phycoerythrin; RPMI, Roswell Park Memorial Institute; RT-PCR, real-time polymerase chain reaction; SFC, spot-forming cell; T1, time point 1; T2, time point 2; T3, time point 3.

Patients and Methods

Patients.

Eighteen treatment-naïve patients with chronic hepatitis B, attending the Hepatitis Clinic at University College Hospital (London, United Kingdom), were enrolled into the study (Table 1). All patients were seropositive for hepatitis B surface antigen (HBsAg) and HBeAg, with HBV-DNA levels > 106copies/mL. All patients were negative for anti–hepatitis D virus, anti–hepatitis C virus (anti-HCV), anti-HIV1/2, and autoantibodies. Twelve of the patients were male, and the mean age of the patients was 38.9 ± 9.9 years. A liver biopsy was performed in all patients as part of a routine diagnostic evaluation, and the inflammation grade and fibrosis stage were scored according to established criteria.19 Patients were followed serially with protocol visits for a median period of 18 months (range 12-25 months) during a course of antiviral treatment with nucleoside analogues (telbivudine or lamivudine). Written informed consent was obtained from each patient, and the study protocol was approved by the Ethics Committee of University College London Hospitals. During treatment, 6 patients seroconverted to hepatitis B e antibody (anti-HBe; group 1) after a median period of 9 months (range 6-10 months), whereas 12 patients (group 2) remained HBeAg-positive throughout the monitoring period (median 18 months, range 12-25 months).

Table 1. Baseline Characteristics of Patients
Patient NumberBaseline (T1)
HBV-DNA (log10 Copies/mL)ALT (IU/mL)Ishak Fibrosis ScoreHBeAg Status% Pentamer(+) on Total CD8+ T-Cells*% PD-1(+) on Pentamer(+) CD8+ T-Cells*% PD-1(+) on Total CD8+ T-Cells
  • Abbreviations: ALT, alanine aminotransferase; HBcAg, hepatitis B core antigen; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; nd, not determined; PD-1, programmed death 1.

  • *

    Combined frequency detected with pentamers for both HBcAg and HBsAg.

19.522061+0.413.729.17
29.782331+1.294.733.34
39.114361+0.912.695.67
48.48814+1.341.472.5
58.593633+0.891.572.95
69.082211+0.604.775.12
78.05481+1.237.199.75
87.76823+0.76nd0.05
99.251134+1.04nd3.95
108.192522+0.45ndnd
117.37591+0.993.216.04
127.93355+0.764.29nd
136.921071+0.59ndnd
1410.071532+0.90nd1.37
157.06991+1.980.07nd
168.78563+1.466.277.75
177.751392+0.63nd1.16
188.561502+0.541.472.25

The investigations were focused on three time points, with heparinized blood used for the separation of peripheral blood mononuclear cells (PBMCs) obtained serially from all patients:

  • Time point 1 (T1): Baseline. Patients had a high viral load and were HBeAg-positive.

  • Time point 2 (T2): Early period of antiviral treatment (treatment weeks 12-16). Patients had reduced viremia, and all patients were HBeAg-positive.

  • Time point 3 (T3): Late period of antiviral treatment (treatment weeks 32-56). Patients had undetectable or very low serum HBV-DNA (<103 copies/mL) with or without HBeAg loss and anti-HBe positivity in groups 1 and 2, respectively.

Hepatitis Serology.

HBsAg, HBeAg, anti-HBe, and anti-HIV1/2 were determined with commercial immunoassays (Abbott Laboratories, Maidenhead, United Kingdom). Anti-HCV was detected with the Ortho HCV 3.0 enzyme-linked immunosorbent assay (OrthoDiagnostics, High Wycombe, United Kingdom). Serum HBV-DNA was quantified by a sensitive real-time polymerase chain reaction (RT-PCR) technique with a lower limit of quantitation of 300 copies/mL.20

Isolation of PMBC.

PBMCs were isolated from heparinized blood by standard density gradient centrifugation with Lymphoprep (Axis-Shield, Oslo, Norway), as previously described.21, 22 Subsequently, the cells were resuspended in supplemented Roswell Park Memorial Institute (RPMI) 1640 (Gibco, Invitrogen, Auckland, NZ). After isolation, PBMCs were cryopreserved in a medium containing 75% fetal bovine serum (FBS), 15% RPMI, and 10% dimethylsulfoxide. PBMCs were thawed with a step-by-step, gradual dilution method.22 The cell viability was over 90%, as assessed by trypan blue exclusion.

Flow Cytometry.

The following antibodies were used for PBMC staining: anti-CD8/allophycocyanin (APC)–cyanin 7 (Cy7), anti-CD3/phycoerythrin (PE)-Cy7, anti–PD-1/PE, isotype control/PE, anti-CD45RA/PE-Cy7, anti-CD127/PE, anti-CD62L/PE, and anti-CD4/APC-Cy7 (BD Biosciences, Oxford, United Kingdom). HBV-specific CD8+ T-cells were identified with human leukocyte antigen A2 (HLA-A2) pentamers containing HBcAg 18-27 peptide (FLPSDFFPSV) and HBsAg 183-191 peptide (FLLTRILTI), which were labeled with the fluorochrome APC (Proimmune, Oxford, United Kingdom). Cytomegalovirus (CMV)-specific CD8+ T-cells were identified with an APC-labeled HLA-A2 pentamer containing CMV pp65 495-504 (NLVPMVATV). According to published guidelines,22 background PBMC staining with pentamers was established with PBMCs from 10 HLA-A2–negative patients, and a cutoff of 0.02% was identified.

For assessment of PD-1 expression, the staining was performed in a 96-well round-bottom plate. PBMCs (3 × 106) were resuspended in 200 μL of phosphate-buffered saline (PBS)/1% FBS and stained with HBV surface and core pentamers. Because of limitations in the number of cells available and the relatively low frequency of virus-specific T-cells, staining was carried out with pentamers to both HBsAg and HBcAg together. The cells were rested for 10 minutes at room temperature and subsequently resuspended in PBS/50% mouse serum to block nonspecific binding. Staining was then carried out with antibodies for 30 minutes at 4°C. After being washed, the cells were resuspended in PBS/1% FBS, acquired on a Becton-Dickinson FACSArray, and analyzed with BD FACSCanto software. Lymphocytes were gated according to their physical parameters, and CD3+/CD8+ lymphocytes were then selected. Virus-specific CD8+ T-cells were examined as a percentage of the total CD8+ T-cell population, as well as the mean fluorescent intensity (MFI) on the pentamer-positive cells.

The memory phenotype of CD4+ T-cells was determined by staining with fluorochrome-labeled CD45RA and CD62L antibodies. Four subsets of memory cells were identified: naive cells (N:CD45RA+/CD62L+), central memory cells (CM:CD45RA−/CD62L+), effector memory cells (EM:CD45RA−/CD62L−), and effector cells (E:CD45RA+/CD62L−).

The memory phenotype of CD8+ T-cells was determined by staining with fluorochrome-labeled CD45RA and CD127 antibodies. Four subsets of memory cells were identified: naive cells (N:CD45RA+/CD127+), central memory cells (CM:CD45RA−/CD127+), effector memory cells (EM:CD45RA−/CD127−), and effector cells (E:CD45RA+/CD127−). The frequency of each memory subset was evaluated within the total CD4+ or CD8+ populations. For detection of HLA-A2–positive cases, PBMCs were labeled with a mouse anti-human HLA-A2 (OneLambda, Inc., San Diego, CA) followed by a fluorescein isothiocyanate–conjugated goat anti-mouse IgG secondary antibody (Sigma, Dorset, United Kingdom). Cells were washed and acquired on a Becton Dickinson FACSArray (BD BioSciences). Subsequent analysis was performed with FACSDiva software.

Quantitation of PD-1 Messenger RNA (mRNA) by RT-PCR.

Total RNA was extracted from 5 × 105 PBMCs obtained from the same time points described previously. PBMCs were resuspended in 500 μL of Trireagent (Ambion, Texas) homogenized with a 21G needle, and the RNA was extracted according to the manufacturer's instructions. RNA was quantified with a NanoDrop spectrophotometer (Labtech International, Sussex, United Kingdom). The extracted RNA was reverse-transcribed with the Quantitect reverse-transcription kit according to the manufacturer's instructions (Qiagen, Hilden, Germany). RT-PCR was performed with SYBR green with Quantitect two-step RT-PCR according to the manufacturer's instructions on an ABI 7500 RT-PCR machine (Applied Biosystems, Foster City). The primers used were commercially available (PDCD1_1_SG with β-actin as the house-keeping gene HB-ACTB_1_SG; Qiagen). The results were analyzed with AB7500 software.

Enumeration of HBV-Specific, IFNγ-Producing and IL-10–Producing T-Cells by ELISPOT Assays.

The ELISPOT assays were performed with freshly isolated PBMCs, as previously described.21, 23 To determine the frequency of CD4+ T-cells, PBMCs were incubated for 24 hours in 96-well round-bottom tissue culture plates in the presence of commercially available recombinant HBV nucleocapsid protein (HBcAg; >95% purity; final concentration 2 μg/mL) or purified surface protein (HBsAg; >99% purity, final concentration 10 μg/mL; both from American Research Products, Belmont, MA) in triplicate wells with 2 × 105 PBMCs per well at 37°C with 5% CO2. PBMCs were also cultured with phytohemagglutinin (Sigma) as a positive control.

In parallel, 96-well Immobilon-P membrane microtiter plates (Millipore, Bedford, MA) were coated with 10 μg/mL anti-cytokine monoclonal antibody for IFNγ clone 1-D1K or for IL-10 clone 9D7 (MabtechAB, Stockholm, Sweden) for 6 hours at 4°C. The plates were washed and blocked with buffered RPMI/10% human AB serum for 1 hour. The cells were then transferred to the Immobilon-P plates and incubated in the presence of 1 μg/mL biotinylated anti-cytokine antibody (for IFNγ clone 7-B6-1 or for IL-10 clone 12G8; MabtechAB) for 2 hours at room temperature. After washing, 100 μL of 1 μg/mL streptavidin/alkaline phosphatase conjugate (MabtechAB) was added to each well and incubated for 1.5 hours in the dark at room temperature. The unbound conjugate was removed, and 100 μL of the substrate nitroblue tetrazolium chloride/bromochloroinolylphosphate toluidine salt (Roche Diagnostics, Lewes, England) was added to each well. The reaction was stopped with extensive washing with distilled water. The plates were dried overnight at room temperature, and the spots were counted with an ELISPOT reader (AID Diagnostika, Strassberg, Germany).24

The frequency of CD8+ T-cells was tested as described previously by an ELISPOT assay with a known HLA-A2 restricted epitope: HBV nucleocapsid amino acids 18-27 (FLPSDFFPSV). PBMCs were cultured in triplicate with the peptides or influenza peptide control (GILGFVFTL; shown later in Fig. 5).

Figure 5.

Frequency of HBV-specific CD4+ and CD8+ T-cells and influenza-specific CD8+ T-cells producing IFNγ and PD-1 expression over time. An increase in the mean frequency of IFNγ-producing, HBV-specific (A) CD4+ and (B) CD8+ T-cells was observed from T1 to T3. (B) In contrast, there was no significant change in the frequency of influenza-specific CD8+ T-cells producing IFNγ from T1 to T3. In parallel, there was a decrease in mean PD-1 expression on total CD8+ T-cells (MFI) in the subgroup of patients analyzed. Error bars represent standard errors of the mean. Abbreviations: HBV, hepatitis B virus; IFNγ, interferon gamma; MFI, mean fluorescent intensity; PD-1, programmed death 1; SFC, spot-forming cell.

The number of specific spot-forming cells was determined as the mean number of spots in the presence of an antigen minus the mean number of spots in the wells with a medium only and was expressed per 1 × 106 PBMCs. The cutoff for significant response was defined as the mean plus two standard deviations from the testing of 10 subjects without HBV exposure. The cutoff value obtained was as follows: mean = 1.6 ± 1.82 spot forming cells. Repeat testing of positive samples, using the IFNγ or IL-10 ELISPOT assays, following immunomagnetic depletion of CD4+ T-cells confirmed that the responses to rHBcAg and rHBsAg represent CD4+ T-cell reactivity (data not shown).

Statistical Analysis.

Changes in PD-1 expression at different time points were analyzed by the Student t test and Mann-Whitney U test. The correlation between HBV-DNA levels over time and PD-1 expression by FACS was assessed by Pearson's correlation analysis. Correlations between HBV-DNA levels, PD-1 expression, memory phenotype, and the frequency of virus-specific, IFNγ-producing and IL-10–producing T-cells were assessed by Pearson's correlation analysis.

Results

Clinical Outcomes.

Six of 18 patients that were enrolled lost HBeAg and became anti-HBe+ (group 1), whereas 12 patients remained HBeAg+ (group 2). At baseline, serum HBV-DNA levels were comparable between the two groups, whereas patients in group 1 had significantly higher alanine aminotransferase (ALT) levels (P < 0.01; Table 2). The profound suppression of HBV replication early after antiviral treatment was started (T1 to T2) was paralleled by a decrease in serum ALT levels, with no significant differences in the magnitude of HBV-DNA or ALT reductions between the two groups (Table 2). Fifteen of 18 patients had undetectable serum HBV-DNA levels by T3.

Table 2. Changes in Serum HBV-DNA and ALT Levels over the Course of the Antiviral Treatment
 HBV-DNA Baseline (log10 Copies/mL)Δ Serum HBV-DNA (log10 Copies/mL)ALT Baseline (IU/mL)Δ Serum ALT Levels
T1-T2T2-T3T1-T2T2-T3
  • The baseline values are reported as mean ± standard deviation. The differences (Δ) in HBV-DNA or ALT levels between the time points (T1-T2 and T2-T3) are shown as mean ± standard deviation.

  • Abbreviations: ALT, alanine aminotransferase; HBV, hepatitis B virus; T1, time point 1; T2, time point 2; T3, time point 3.

  • *

    Significant difference between groups 1 and 2 (P < 0.01).

Group 19.1 ± 0.55.1 ± 1.11.5 ± 0.8256 ± 125*97 ± 128133 ± 91
Group 28.1 ± 0.94.2 ± 1.20.9 ± 1.2107 ± 60.5*23 ± 7846 ± 70

Longitudinal Analysis of PD-1 Expression by Flow Cytometry.

In all patients, the frequency of HBV-specific CD8+ T-cells (combination of pentamers including HBcAg and HBsAg epitopes) ranged from 0.4% to 1.98% of the total CD8+ T-cells and did not change between the three time points of the study (Table 1). In contrast, PD-1 expression significantly decreased on both total CD8+ and HBV-specific CD8+ T-cells in all patients (Figs. 1 and 2). During antiviral treatment between T1 and T3, the proportion of PD-1–positive cells decreased within the total CD8+ T-cells from 4.98 ± 0.94 to 1.71 ± 0.86 (P = 0.002) and within HBV-specific CD8+ T-cells from 3.28 ± 1.05 to 0.95 ± 0.75 (P = 0.01).

Figure 1.

PD-1 expression on total CD8+ T-cells and CMV-specific CD8+ T-cells and cell gating strategy. (A) Representative dot plot illustrating the cell gating strategy for the selection of CD3+/CD8+/PD-1+ cells from PBMCs. (B) Representative dot plots for PBMCs from two patients with chronic HBV infection stained with anti-CD3, anti-CD8, and anti–PD-1 at baseline (T1) and following a course of antiviral therapy (T3). A reduction in PD-1 expression on CD8+ T-cells over time is illustrated in two representative patients. (C) Representative dot plots for PBMCs illustrating staining with anti-CD3, anti-CD8, anti–PD-1, and CMV-specific pentamers. Levels of PD-1+ CMV-specific CD8 T-cells were low and did not significantly change from T1 to T3. Abbreviations: APC, allophycocyanin; CMV, cytomegalovirus; Cy7, cyanin 7; HBV, hepatitis B virus; PBMC, peripheral blood mononuclear cell; PD-1, programmed death 1; T1, time point 1; T3, time point 3.

Figure 2.

PD-1 expression on total CD8+ and HBV-specific CD8+ T-cells during a course of antiviral therapy. A decrease in PD-1 expression can be seen in terms of both (A,B) the percentage of cells staining positive for PD-1 and (C,D) MFI from baseline (T1) through a course of antiviral treatment. This decrease was observed in both (A,C) total CD8+ T-cells and (B,D) virus-specific CD8+ populations. Each line represents changes in an individual patient's PD-1 expression from baseline (T1) to T3. Changes in HBV-DNA levels from T1 to T3 are shown as bar graphs. Abbreviations: HBc, hepatitis B core; HBs, hepatitis B surface; HBV, hepatitis B virus; MFI, mean fluorescent intensity; PD-1, programmed death 1; T1, time point 1; T3, time point 3.

The levels of PD-1 expression, as assessed by MFI, were also reduced between the same time points: for the total CD8 subset, 721 ± 206.59 at T1 and 383.59 ± 71.27 at T3 (P = 0.03), and for the HBV-specific subset, 762.38 ± 280.26 at T1 and 528 ± 170.3 at T3 (P = 0.05).

In contrast to HBV-specific CD8+ T-cells, CMV-specific CD8+ T-cells exhibited low levels of PD-1 positivity that did not change over time (P > 0.1; Fig. 1C).

There was no correlation between PD-1 expression on HBV-specific CD8+ T-cells and baseline serum ALT levels. Although serum ALT levels were significantly higher in patients who seroconverted on treatment, there was no difference in PD-1 expression between groups 1 and 2 at baseline (P > 0.1).

We also assessed PD-1 expression on CD3+/CD8− T-cells over the three time points described. We observed a significant decrease in both the percentage of CD3+/CD8− T-cells staining positive for PD-1 (P = 0.014) and the MFI (P = 0.014) from baseline (T1) to T3.

HBeAg seroconversion, which occurred in patients in group 1 between T2 and T3, was associated with a trend toward a decrease in the frequency of PD-1–expressing total CD8+ T-cells [T2, 3.27 ± 0.86, and T3, 1.41 ± 0.64; P = 0.068], despite no significant decrease in the viral load. It is possible that the sample size (six patients, group 1) precluded statistical significance in this observation. In contrast, there was no decrease in PD-1 expression between T2 and T3 in the absence of seroconversion group 2 (P > 0.1). Furthermore, the magnitude of the PD-1 change from T2 to T3 differed between the two groups. We observed a 51% decrease in the frequency of PD-1–expressing, HBV-specific CD8+ T-cells from T2 to T3 associated with HBeAg seroconversion (group 1) versus only a 3% decrease in the absence of seroconversion (group 2; P = 0.057).

Three of 18 patients did not achieve undetectable HBV-DNA levels at T3 but showed a reduction in PD-1 expression between T1 and T3, which mirrored the decrease in serum HBV-DNA levels. In a representative patient, MFI on total CD8 decreased (T1 to T3) from 1509 to 565 for a 3-log drop in HBV-DNA levels. None of these patients had seroconverted, and no significant difference in PD-1 expression was observed between T2 and T3.

During antiviral treatment, the frequency of PD-1–expressing, virus-specific CD8+ T-cells correlated closely with HBV-DNA levels (r = 0.998, P = 0.036). The strength of PD-1 expression on CD8+ T-cells also decreased in all patients from baseline to T3. This decrease directly correlated with a decrease in HBV-DNA levels (r = 0.996, P = 0.054; Fig. 3).

Figure 3.

Correlations between PD-1 expression and HBV-DNA levels. A reduction in PD-1 expression on HBV-specific and total CD8+ T-cells correlates with a reduction in HBV-DNA levels. The average HBV-DNA and PD-1 expression levels for each time point (time points 1-3) are represented as diamonds. Error bars illustrate standard errors. The four panels demonstrate changes in PD-1 expression on (A,C) total CD8+ and (B,D) virus-specific CD8+ T-cells as both (C,D) changes in MFI and (A,B) changes in the percentage of positive cells. Abbreviations: HBc, hepatitis B core; HBs, hepatitis B surface; HBV, hepatitis B virus; MFI, mean fluorescent intensity; PD-1, programmed death 1.

Longitudinal Analysis of PD-1 mRNA Expression.

PD-1 mRNA levels decreased significantly during antiviral treatment between T1 and T3 (P = 0.001; Fig. 4). There was a correlation between PD-1 mRNA and HBV-DNA levels, but this was not statistically significant (r = 0.977, P = 0.1). We further analyzed whether there was a direct correlation between PD-1 expression on the cell surface (as assessed by flow cytometry) and PD-1 mRNA levels (as quantitated by RT-PCR). We found a direct correlation between PD-1 expression on HBV-specific CD8+ T-cells in terms of MFI and relative gene expression as assessed by RT-PCR (r = 0.994, P = 0.07).

Figure 4.

Relative expression of PD-1 at the mRNA level as assessed by quantitative RT-PCR from baseline to T3. PD-1 expression at the mRNA level is displayed from a high viral load to a low viral load (T1-T3). The y axis represents the relative PD-1 gene expression for each patient from T1 to T3. Each line represents changes in an individual patient's relative PD-1 expression from T1 to T3. A significant decrease in PD-1 expression was seen (P = 0.001). Abbreviations: PD-1, programmed death 1; mRNA, messenger RNA; T1, time point 1; T3, time point 3.

Correlation Between PD-1 Expression, HBV-DNA Levels, and Frequency of IFNγ-Producing and IL-10–Producing T-Cells.

In 14 of 18 patients, we assessed the frequency of HBV-specific T-cells producing IFNγ at the three time points specified. There was a significant increase in the frequency of IFNγ-producing CD4+ T-cells between T1 and T3 in response to both HBV core (P = 0.009) and surface (P = 0.002) antigens (Fig. 5A). This inversely correlated with the decrease in PD-1 expression and HBV-DNA levels (r = −0.994, P = 0.067). In contrast, the frequency of HBV-specific CD4+ T-cells producing IL-10 decreased markedly between T1 and T3 in response to surface antigen (P = 0.02), and a similar trend was observed in response to HBV core antigen (P = 0.08).

In 6 of 18 patients, we assessed the frequency of HBV-specific CD8+ T-cells producing IFNγ at the three time points specified. The frequency of influenza-specific CD8+ T-cells was also assessed as a control (Fig. 5B). There was a significant increase in the frequency of IFNγ-producing, HBV-specific CD8+ T-cells between T1 and T3 (P = 0.03), whereas there was no significant change in the frequency of IFNγ-producing, influenza-specific CD8+ T-cells (P > 0.1).

Correlation Between PD-1 Expression, HBV-DNA Levels, and Memory Phenotypes.

In four of six patients who seroconverted to anti-HBe (group 1), we investigated whether a reduction in serum HBV-DNA levels or HBeAg loss led to changes in the four memory phenotypes. At baseline (T1), there was a significant direct correlation between the frequency of the hepatitis B core–specific central memory phenotype and PD-1 expression as assessed either by MFI on the total CD8+ cell population (r = 0.988, P = 0.012) or by the combined frequency of hepatitis B core–specific and hepatitis B surface–specific pentamer-positive PD-1+ T-cells (r = 0.985, P = 0015; Fig. 6A). There was also a direct correlation, at baseline, between the effector memory phenotype and PD-1 expression as assessed by MFI on the total CD8+ cell population (r = 0.995, P = 0.005; Fig. 6B). In parallel, there was a strong inverse correlation between the HBcAg-specific effector phenotype and PD-1 expression at baseline (r = −0.999, P = 0.0005; Fig. 6C). A similar but not significant relationship was seen in HBsAg-specific cells. There were no significant changes in the frequency of the memory phenotypes in this subgroup of patients over time (Supplementary Fig. 1).

Figure 6.

Correlations between PD-1 expression and memory phenotypes at baseline. (A) A significant direct correlation was observed between the frequency of the hepatitis B core–specific CM phenotype and PD-1 expression as assessed by either MFI on the total CD8+ cell population (r = 0.988, P = 0.012) or the frequency of HBc-specific and HBs-specific pentamer-positive cells (r = 0.985, P = 0015) at baseline. (B) A direct correlation was observed between the EM phenotype and PD-1 expression as assessed by MFI on the total CD8+ cell population (r = 0.995, P = 0.005) at baseline. (C) An inverse correlation between the hepatitis B core antigen–specific E phenotype and PD-1 expression was observed at baseline (r = −0.999, P = 0.0005). Abbreviations: CM, central memory; E, effector; EM, effector memory; HBc, hepatitis B core; HBs, hepatitis B surface; MFI, mean fluorescent intensity; PD-1, programmed death 1.

Discussion

The present study demonstrates that viremia levels directly correlate with PD-1 expression on T-cells in chronic HBV infection and that treatment-induced suppression of viral replication, manifested by a marked reduction in serum HBV-DNA levels, results in a significant decrease in PD-1 expression on the T-cell surface as well as PD-1 mRNA transcription. The significant reduction in PD-1 expression on antiviral treatment is accompanied by improved virus-specific T-cell reactivity with increased IFNγ production. In addition, our results show that serum HBeAg loss is associated with a decrease in PD-1 expression that is independent of viral load and is accompanied by a further improvement in virus-specific T-cell reactivity.

These findings extend recent observations of patients with acute hepatitis B showing decreased PD-1 expression on CD8+ T-cells and spontaneous resolution of HBV infection, whereas in those with persistent HBV replication, both PD-1 and HBV-DNA levels remained high.15, 25 Serial testing of seven patients with anti-HBe–positive chronic hepatitis B who had spontaneous reactivation of the disease demonstrated that the viral load can directly influence the HBV-specific T-cell repertoire.25 The advantage of the present study design is that by monitoring longitudinally the impact of treatment-induced suppression of HBV replication with direct antivirals (telbivudine or lamivudine), we were able to define the cause-effect relationship between the viral load and PD-1 expression. The results demonstrate a close positive correlation between HBV-DNA levels and PD-1 expression on total CD8+ T-cells, virus-specific CD8+ T-cells, and CD3+/CD8− T-cells, that is, CD4+ T-cells. A number of studies have demonstrated in a mouse model with LCMV infection and in humans infected with HIV, HBV, or HCV that the blockade of PD-1 binding to its ligands results in functional restoration of virus-specific T-cells.10, 11, 12, 14, 25–27 The lack of sufficient numbers of PBMCs precluded us from testing for this well-established effect. The blockade of the PD-1/PD L-1 pathway has been suggested as a possible immunomodulatory approach for enhancing immune control of viral replication. However, potential harmful effects, such as autoimmunity or immunopathology, should be considered.28 Selective targeting of PD-1 on virus-specific T-cells may be a prerequisite for achieving an acceptable risk/benefit balance with this approach.

It is recognized that high ALT levels are associated with an increased rate of HBeAg seroconversion, both during the natural history of the disease and with antiviral treatment.29 If ALT is a surrogate marker of immune response to HBV, then there may be a relationship between ALT at baseline and PD-1 expression. However, the present study showed no such relationship. This was in part due to the fact that PD-1 levels were significantly different at baseline between individual patients, and although a relative decrease in PD-1 expression was observed over time, some patients had higher absolute PD-1 levels at T3 than other patients at baseline. It may also reflect that ALT is a marker of the inflammatory response driven by HBV replication rather than a reflection of virus-specific T-cell reactivity. Analyses of HBV-specific CD8+ T-cells in patients with both HBeAg-positive and HBeAg-negative chronic hepatitis B showed a lack of association between disease exacerbations and the frequency of circulating virus-specific T-cells.3, 25 The present study extends the findings by Webster et al.,3 who demonstrated that the repertoire of HBV-specific CD8+ T-cells is inversely proportional to the level of HBV replication, whereas there is no correlation with the degree of liver damage.

Our findings revealed that treatment-induced suppression of HBV replication resulted in a significant decrease in PD-1 expression in all patients: those with HBeAg seroconversion and those remaining HBeAg-positive. We have recently shown that HBeAg loss during treatment with adefovir dipivoxil is associated with both a profound reduction of HBV-DNA levels and an increase in CD4+ T-cell reactivity, whereas patients with a moderate reduction in serum HBV-DNA and no changes in CD4+ T-cell responses remained HBeAg-positive.30 Thus, a reduction of PD-1 expression, as a result of a profound reduction of the HBV viral load, does not appear enough to fully restore CD4+ and CD8+ T-cell response and to achieve HBeAg seroconversion in all treated patients.

The difference in PD-1 expression and the percentage of various memory subsets seen at baseline is an interesting finding. An analysis of the correlations between PD-1 expression and central memory, effector memory, and effector phenotypes in core-specific cells at baseline suggests that elevated PD-1 levels are associated with a decrease in effector cells and an increase in central memory and effector memory phenotypes. This adds support to the concept that PD-1 plays a central role in T-cell exhaustion and viral persistence. There was no significant relationship between the frequency of different memory phenotypes of CD4+ and CD8+ T-cells expressed at different time points and HBV-DNA levels, e antigen status, and PD-1 expression observed in this study.

Nucleoside analogues are known to interfere with viral replication, lowering HBV-DNA levels, but they have not been proven to influence the development of effective memory T-cell differentiation and function; hence, there is the need for long-term therapy to control viral load.31 This study did not show any difference in the frequency of different memory phenotypes on CD4+ or CD8+ T-cells expressed at different time points. It did, however, show a significant increase in the frequency of IFNγ-producing CD4+ and CD8+ T-cells over time as well as a decrease in the frequency of IL-10–producing T-cells. This suggests that there may indeed be a shift in the cytokine profile of T-cells as HBV-DNA and PD-1 expression decreases with antiviral therapy. Given the mechanism of action of telbivudine and lamivudine, which directly block viral replication, it seems likely that therapy results in a fall in HBV-DNA levels and that this in turn leads to a decrease in PD-1 expression, which results in an improvement in cytokine production by HBV-specific T-cells. This is similar to what has been reported for infection with HIV.11, 12

In conclusion, this study highlights the mechanism linking HBV replication and impaired T-cell functions in chronic hepatitis B. The data reveal a strong correlation between HBV viremia and hyperexpression of PD-1 on all T-cells. Treatment-induced suppression of HBV replication results in a significant reduction of PD-1 expression on the T-cell surface and PD-1 transcription, thus reducing its negative impact on T-cell activation and function. As suggested previously, a proper restoration of antiviral immunity would require a combination of potent suppression of HBV replication plus immunotherapy, which would amplify virus-specific T-cell reactivity to achieve sustained control of HBV replication and resolution of liver disease.32

Ancillary