Hepatitis B viral kinetics: A dynamic puzzle still to be resolved

Authors

  • Avidan U. Neumann

    Corresponding author
    1. Faculty of Life Sciences, University of Bar-Ilan, Ramat-Gan, Israel
    Current affiliation:
    1. National Institute of Diabetes & Digestive & Kidney Diseases, Bethesda, MD
    • University of Bar-Ilan, Ramat-Gan 52900, Israel
    Search for more papers by this author
    • fax: 972-3-635-6353


  • Potential conflict of interest: A.U.N. received research grants from Roche Pharmaceuticals and Gilead and is a consultant for Roche, Gilead, Schering-Plough, Idenix, and other pharmaceutical companies.

The study of viral kinetics (VK) during treatment of chronic infection with HIV,1–4 hepatitis C virus (HCV),5 and hepatitis B virus (HBV)6–8 has resulted in improved understanding of viral and immunological dynamics and has led to several clinical implications. In vivo antiviral effects of interferon α (IFN) against HCV and of various reverse-transcriptase inhibitors and protease inhibitors against HIV have been identified and quantified using models of viral dynamics. However, although chronic HBV infection has long been treated with IFN, hepatitis B VK has been reported only during treatment of this infection with various polymerase inhibitors.8–16 In the case of HBV infection, for which treatment is still far from optimal, it is of special importance to understand viral dynamics and antiviral mechanisms associated with all possible therapies. A first step toward achieving this objective is the comprehensive study reported by Sypsa et al.17 comparing HBV DNA kinetics during the first month of treatment of HBV infection with pegylated interferon α2b (Peg-IFN-α2b), lamivudine, and their combination.

Abbreviations

VK, viral kinetics; HCV, hepatitis C virus; HBV, hepatitis B virus; IFN, interferon α; Peg-IFN, pegylated interferon α; HBeAg, hepatits B e antigen; ALT, alanine aminotransferase.

Hepatitis B VK During the Beginning of Treatment

VK during the first month of treatment is influenced mainly by three dynamical parameters. The first is the effectiveness (ϵ) of blocking virion production, which is reflected by the first phase of HBV DNA decline; the HBV DNA log decrease after 1 week of treatment is approximately log(1−ϵ). Second, the half-life of the free virions (LN(2)/c), where c is the clearance rate constant of free virus, can be estimated via nonlinear fitting of the HBV DNA first phase decline slope. Third, the half-life of productively infected cells (LN(2)/δ), where δ is the rate constant of loss of infected cells, can be estimated approximately from the HBV DNA second phase decline slope. Table 1 summarizes the available reported values for HBV kinetics during treatment with various therapies.

Table 1. Summary of Early VK Parameters (Means and Ranges) With Different Therapies
Drug/Dose*Log HBV DNA Decline at 1 wk (Log cp/mL) (First Phase Decline)Half-Life of Free Virions (hrs) (From First Phase Slope)Half-Life of Infected Cells (days) (From Second Phase Slope)Correlations/Comments
HBeAg+HBeAg−HBeAg+HBeAg−HBeAg+HBeAg−
  • Abbreviations: NA, not available; LAM, lamivudine.

  • *

    These are indicative values summarized from different studies and should not be used for comparison between the different drugs, which would necessitate carefully planned randomized studies.

  • The half-life of HBV free virions can be accurately estimated only in studies with at least four samples in the first week.

  • M. R. Brunetto, personal communication.

  • §

    Preliminary study with only 10 patients combined from four dose groups. No early VK reported for larger studies of Entecavir.

  • Unpublished data.

  • Significant difference between Peg-IFN and Lamivudine 100 mg (or combination of Peg + LAM) within the same study.

Lamivudine10, 13, 17, 18      Correlation of second slope with baseline ALT in HBeAg-positive studies
 100–150 mg1.1–1.31.4–1.915 (9–22)9 (3–16)2.4 to >704.0–56
600 mg1.4NA12 (9–18)NA3.3–12.3NA
Adefovir8, 15      Correlation of second slope with baseline ALT but no dose dependence
 5 mg1.0NANA   
 10 mg1.5NA14 (9–27)NA5.3 to >70NA
 30 mg1.7NANA   
 60 mg1.7NANA    
Entecavir12§      N = 10 only§
 0.05–1.0 mg1.4§NA16 (12–29)NA4.7–32NA 
Telbivudine16      Second phase slope significantly dose dependent
 25–100 mg1.9NANANA8.4 (4–27)NA
 200–800 mg2.2NANANA5.1 (2–20)NA
Peg-IFN-α2a18       
 180 μg qwNA0.8NA9 (3–30)NA3.0–47 
180 + LAM 100 mgNA2.2     
Peg-IFN-α2b17      Combination not significantly different from LAM alone
 100–200 μg qwNA0.8NA13 (3–70)NA2.7 to >70
 100/200 + LAM 100 mgNA1.2    

The first step in modeling VK is to determine the major in vivo antiviral mechanism of the drug being used. For HIV, both protease- and reverse-transcriptase inhibitors partially block de novo infection, as expected from in vitro studies. On the other hand, modeling of VK predicted that IFN partially blocks production of hepatitis C virions by cells that are already infected5; this prediction was later corroborated using the replicon assay. Initial studies of HBV kinetics demonstrated rapid viral dynamics, but the early models that were proposed6, 7 are not adequate for fitting HBV kinetics. Subsequent studies demonstrated that HBV DNA kinetics is compatible with the assumption that the major in vivo effects of both polymerase inhibitors8, 10, 13, 15, 16 and IFN17, 18 (M. R. Brunetto, personal communication) are attributable to partial blocking of the production of virions.

Whether de novo infection with HBV is also blocked by polymerase inhibitors and, if so, to what extent (effectiveness η) is an unsettled issue that cannot be resolved by current VK data. In various studies, it has been assumed that de novo infection is completely blocked (η = 1.0),8 not blocked (η = 0),15 or partially blocked (η = 0.5).10 Alternatively, a best-fit of HBV DNA kinetics has been used to determine η.11, 12 None of these approaches is appropriate, because the current resolution of HBV DNA kinetics during the first month of treatment does not permit a valid determination of η (see Neumann et al.5 for more details). Thus, the methodology used by Sypsa et al.17 as well as Wolters et al.13—in which VK data were fitted for both η = 0 and η = 1.0 and the range of results generated was provided—constitutes the most correct approach. Note that η has a substantial effect on VK (particularly on δ) only when blocking of viral production is low (ϵ < 0.9), and hence the impossibility of determining η. In fact, when blocking of viral production is very low (less than 0.5 log reduction in the first week), a flat second phase slope should not necessarily be attributed to a long half-life of infected cells; it could also be due to poor blocking of viral production. Thus, when ϵ < 67%, δ cannot be accurately determined, although the implications of a flat second slope may still be clinically relevant, as correctly pointed out by Sypsa et al.17

An interesting result is that the effectiveness of Peg-IFN, both Peg-IFN-α2b17 and Peg-IFN-α2a18 (M. R. Brunetto, personal communication), in blocking virion production is significantly lower (0.8 log HBV DNA decrease after 1 week) than that of lamivudine in the same studies (>1.2 log). Moreover, the administration of weekly injections of Peg-IFN gives rise to a decrease in serum IFN levels at the end of each week due to the pharmacokinetic properties of Peg-IFN. Indeed, Sypsa et al.17 report that 50% of patients treated with Peg-IFN-α2b have increases in the levels of HBV DNA at the end of each week. Such weekly increases in HBV DNA are not as great as those for HCV during Peg-IFN-α2b therapy, possibly because of slower viral clearance or less sensitivity to changes in IFN levels. No information is available regarding weekly increases in HBV DNA during Peg-IFN-α2a treatment. Results of studies of hepatitis B VK during treatment with daily doses of standard IFN are not available.

Based on various studies of VK during polymerase inhibitor therapy,8, 10, 12, 13, 15 the mean half-life of free virions of HBV is 12 to 16 hours (range 9–29); whereas that of HIV and HCV is shorter at 1 to 4 hours.3–5 It is interesting to note that Sypsa et al.17 and Colombatto et al.18 (M. R. Brunetto, personal communication) found that during Peg-IFN treatment the mean half-life was 9 to 13 hours, and half-lives of 3 to 9 hours occurred in 30% to 50% of the patients. However, it is possible that the more rapid clearances are due to patients being hepatits B e antigen (HBeAg)-negative in these studies, since no significant differences between treatment arms were observed. In previous studies, in which sampling had been frequent (at least 4 time points during the first week), so that the slope of the first phase of VK could be accurately determined, the slope was not dependent on the drug or the dose administered.

More information on the host and viral correlates of the large observed range of hepatitis B VK is not available. The slower observed virions half-life may imply that HBV clearance is biologically different compared with that of HIV and HCV (e.g., since HBV is a smaller virus) or that the underlying assumptions of the model being used for HBV dynamics are not accurate. We have no data to support slower clearance of HBV from observations in other nontreatment settings, such as we have for HCV and HIV during aphaeresis19 or for HCV during liver transplantation.20

An important issue to clarify during Peg-IFN treatment is the half-life of productively infected cells, which hypothetically may be shorter due to activation of the immune system by IFN. Interestingly, both Sypsa et al.17 and Columbato et al.18 (M. R. Brunetto, personal communication) report that the half-life of infected cells is not significantly shorter during Peg-IFN than lamivudine therapy in the same study; the values are in the same range (3–70 days) as those reported in previous studies of therapy with polymerase inhibitors.8, 10, 12, 13, 15, 16 However, a limiting factor that precludes a definite conclusion in this context is the wide range of values reported in each study, which may be related to host factors, such as immunological status. Large randomized studies are necessary to resolve this issue.

Moreover, there are no VK data available during Peg-IFN treatment of HBeAg-positive patients, in whom the effect of IFN on the immune response may play an important role. Indeed, in most previous studies of HBeAg-positive patients, the half-life of infected cells correlated with alanine aminotransferase (ALT) levels at baseline; in the study conducted by Sypsa et al.,17 it did not. Furthermore, the HBV DNA second phase decline slope is, in general, not dose-dependent, although in a study of treatment with telbivudine, the slope during high-dose therapy was significantly different from that during low-dose therapy.16 Such dose-dependence may be expected when the effectiveness of blocking viral production is low, thus giving rise to a flat second phase slope, but not when this effectiveness is high as was found in the study of telbivudine. Additional studies of this phenomenon during treatment with highly effective polymerase inhibitors are needed to further clarify this important issue.

Overall, the data of Sypsa et al.17 seem to indicate that early hepatitis B VK during Peg-IFN therapy is disappointing. As expected, its direct antiviral effect seems less potent than that of any polymerase inhibitor; furthermore, an immune-enhancing effect of IFN was not observed during the first month of treatment. Nevertheless, it is too early to dismiss a beneficial effect of IFN therapy; indeed, such an effect becomes more evident later into treatment. Although nonlinear fitting of VK to a model of viral dynamics is appropriate only when frequently obtained data are available, usually only during the first 4 weeks, it is, nevertheless, important to understand long-term VK as well. Analysis of long-term HBV kinetics during treatment with Peg-IFN and with polymerase inhibitors is needed to determine the effects of Peg-IFN. Furthermore, analysis of the kinetics of ALT, which may be a marker of immune activation, is still not available.

Effect of Baseline Factors on VK

Unfortunately, currently available data on early hepatitis B VK during Peg-IFN therapy17, 18 (M. R. Brunetto, personal communication) were obtained from HBeAg-negative patients only; it is not clear whether similar results would be obtained in HBeAg-positive patients. Indeed, the first phase of HBV DNA decline during lamivudine treatment seems to be greater and somewhat faster in HBeAg-negative patients than in HBeAg-positive patients (Table 1). More studies are needed to close gaps in the data in Table 1 so that the role of HBeAg status on HBV dynamics can be better understood.

There are additional important gaps in our knowledge of the effects of baseline parameters on early HBV kinetics during therapy that need to be elucidated. Notably, whereas the differences in VK associated with different genotypes of HCV are well documented,21 the effect of HBV genotype on hepatitis B VK has not yet been studied. HBV genotype usually correlates with the race and/or the region of origin of patients; the region correlates strongly with the mode and time of infection (e.g., infancy vs. adulthood). An improved understanding of the effects of these various factors on HBV dynamics may help in better tailoring of treatment to individual patients.

Long-Term Hepatitis B VK During Treatment

Of clinical importance is the correlation between VK early during treatment and the long-term response to treatment. A simple analysis of VK, using the population mean, shows that: (1) patients receiving placebo have little or no decrease in HBV DNA levels over a period of a year22, 23; (2) patients undergoing polymerase inhibitor treatment exhibit a rapid decrease in HBV DNA to levels 4 to 6 logs below baseline after 48 weeks of treatment, and most patients show a relapse in these levels once treatment is stopped22–24; and (3) patients receiving Peg-IFN exhibit a slower decrease in HBV DNA levels, but the decrease is sustained in a larger proportion of patients after the end of treatment24–26 (Fig. 1A).

Figure 1.

HBV DNA kinetics during 1 year of treatment. Schematic representation of the data for illustration only. VK patterns are given as forward branching patterns—that is, each branching point encountered as one moves forward in time depicts different possibilities for that VK pattern to take. (A) Mean decline in HBV DNA during treatment with placebo, polymerase inhibitors, and Peg-IFN (based on references 22–28). (B) During placebo treatment, there are actually two main VK patterns: a fraction of patients have a fixed level of HBV DNA, while a significant fraction of patients shows extensive (1–5 log) oscillations in HBV DNA, which in some of the patients becomes undetectable.29 (C) During treatment with polymerase inhibitors, practically all patients have an immediate rapid decline in viremia. Following that, patients have either a flat second phase throughout treatment or a continuous biphasic or multiphasic decline to low or undetectable HBV DNA levels. Some patients show a staircase viral kinetic pattern with accelerations of decline. At the end of treatment, most patients have a relapse.24, 29 (D) During Peg-IFN treatment some patients have an immediate rapid decline, but a significant fraction of patients have only a delayed response after 8 to 52 weeks or no decline at all. Some patients have a viral rebound during treatment, while others relapse at the end of treatment, but a significant fraction has sustained lower viremia after the end of treatment.30 The patterns presented assume compliance to therapy and no development of viral resistance. The dashed horizontal line represents the limit of detection at about 400 cp/mL. HBV, hepatitis B virus; Peg-IFN, pegylated interferon α; UD, undetectable; BL, baseline.

In addition to HBV DNA, the loss of HBeAg (or, more accurately, its decrease to below its detection level) may be used as another VK marker in patients that were initially HBeAg-positive. In general, patients treated with Peg-IFN have a higher probability of losing HBeAg during treatment than those treated with polymerase inhibitors.22–26 Moreover, when treatment with an enzyme inhibitor is interrupted, a large proportion of patients become HBeAg-positive again,22–24 while mean fraction of patients that lose HBeAg remains approximately the same 24 weeks after the end of treatment with Peg-IFN.25, 26 Furthermore, therapies that give rise to larger decreases in HBV DNA (e.g., more potent enzyme inhibitors) do not usually result in a larger fraction of patients that lose HBeAg.27, 28 A potential hypothesis for the apparent discrepancy between the extent of decrease of HBV DNA and loss of HBeAg is that HBeAg is a marker of the number of infected cells. A more potent antiviral drug may block production of virions more effectively, thereby giving rise to lower viremia, but the same threshold number of infected cells, at the time of HBeAg loss. On the other hand, the immunogenic properties of IFN may result in an increased rate of decrease of infected cells, and hence increase the loss of HBeAg. HBeAg seroconversion (i.e., the appearance of anti-HBeAg antibodies above detection level), which is more frequently observed during IFN-based therapy, may be a marker of a real change in immune status, but it could also be the consequence of less masking by HBeAg. Quantitative measurements of HBeAg, as well as HBsAg, should be obtained frequently during treatment to allow adequate kinetic analysis of these important markers.

More importantly, recent findings have shown the existence of distinct VK patterns for different specific therapies (Fig. 1). A proportion of patients receiving placebo for 48 weeks undergoes no change in HBV DNA levels, a predictable finding if there is an equilibrium between production and clearance of the virus. However, another large proportion of patients receiving placebo was found to exhibit substantial (1–5 logs) spontaneous decreases in HBV DNA levels, which correlated with elevated levels of ALT and loss of HBeAg29 (unpublished data) (Fig. 1B). These spontaneous declines are usually followed by increases in the levels of HBV DNA, ALT, and HBeAg within 4 to 32 weeks. A potential explanation for this oscillatory behavior is that the evolution of HBV to escape the effects of the immune response and the evolution of the specific adaptive immune response to control the virus are in a “red-queen race” with similar time scales. This situation differs substantially from the steady state in viral load of untreated chronic hepatitis C patients, which may be explained either by the evolution of HCV being faster than that of the adaptive immune pressure on this virus or by an inadequate immune response to HCV.

Furthermore, also during treatment a number of distinct patterns of HBV DNA kinetics have been observed. During treatment with polymerase inhibitors (Fig. 1C), some patients have a flat partial response curve, some exhibit either a rapid biphasic or a slow multiphasic decrease of HBV DNA, and some exhibit a “staircase” pattern of decrease in HBV DNA10, 29 (unpublished data). These different patterns of response are still found after excluding patients in whom drug levels were different, those that discontinued treatment, and those that developed resistance to treatment. When therapy with Peg-IFN is initiated (Fig. 1D), a proportion of patients undergoes an immediate rapid decrease of HBV DNA, but a substantial proportion has either a delayed response or no decrease in HBV DNA30 (H. L. A. Janssen, personal communication). In addition, a proportion of patients treated with Peg-IFN exhibit a rebound in viral levels during treatment; other patients undergo a relapse of viral infection at the end of treatment; while some have a sustained virological response30 (H. L. A. Janssen, personal communication). The different patterns of long-term decrease in HBV levels during treatment with Peg-IFN still need to be related to drug exposure, but are nevertheless striking.

Thus, analysis of HBV kinetics based only on the mean response to each therapy may be misleading. Nevertheless, one can recognize characteristic responses to each therapy (Fig. 1). All patients treated with polymerase inhibitors exhibit an immediate decrease in HBV DNA and, unless viral resistance develops or therapy is interrupted, no rebounds in HBV DNA are observed. On the other hand, during therapy with Peg-IFN, some patients exhibit no response or a delayed response, and rebound in HBV DNA may occur. In compensation, after the end of IFN-based treatment, a proportion of patients exhibit no increase in HBV DNA at all. These findings need to be investigated further to understand better the different effects of IFN-based and enzyme inhibitor–based therapies. Additional factors (e.g., spontaneous immune responses, the existence of more than one compartment of infected cells, and resistance to treatment) may need to be incorporated into the models used.

Last, analysis of the long-term kinetics of HBV has shown a disassociation between early and long-term kinetics of HBV during treatment in a large proportion of patients. Thus, unlike HCV, early kinetics of HBV during therapy may not be a good predictor of the outcome of treatment, especially IFN-based therapy. During treatment of HCV infection, a rapid second phase decrease of HCV (after 4 weeks of treatment) is a good predictor of a sustained virological response and is consistent with eradication of productively infected cells in patients with a sustained virological response. During treatment of HBV infection, although a rapid second phase decrease of HBV may be a good positive predictor of HBV DNA decreasing below its detection level, such a finding cannot yet be used to predict which patients are going to relapse and which are not. Furthermore, even patients who do not exhibit a decrease in HBV during the first month of therapy may still exhibit a significant decrease later; thus the possible negative predictive value of early VK during treatment needs further investigation. A determination of the possible existence of additional virus-infected compartments—as occurs, for example, in HIV infection—is important in the assessment of whether HBV replication can only be controlled or whether it can be eradicated.

VK and the Benefit of Combination Treatment

The combination of Peg-IFN with lamivudine does not seem to result in significantly better first- or second-phase hepatitis B VK during the first month of treatment than those associated with treatment with lamivudine alone17, 18 (M. R. Brunetto, personal communication). Moreover, in the long term, compared with monotherapy, this combination has been associated with only a modest increase in the mean decrease in HBV DNA levels and mean loss of HBeAg after 48 to 52 weeks of treatment.24 On the other hand, after 24 weeks of follow-up after a course of combination treatment, the mean decrease in HBV DNA and mean loss of HBeAg were similar to those associated with Peg-IFN monotherapy.24, 26 Information on the effect of combination treatment on long-term therapy with lamivudine, as well as the effects of a combination of Peg-IFN with other enzyme inhibitors, may provide further understanding of the potential benefits of combination therapy.

Currently, interpreting the available mean VK data combination therapy with Peg-IFN and an enzyme inhibitor such as lamivudine does not appear to be useful. If polymerase inhibitor therapy is administered long-term, addition of therapy with IFN for a limited period does not seem to sufficiently increase the average response during treatment to be cost-effective. On the other hand, if IFN-based therapy is given for a limited period, the addition of lamivudine to the therapy does not increase the mean rate of sustained virological response after a year of treatment.

However, when extremely different individual patterns of HBV VK are observed for the same treatment, but there are nevertheless some qualitative differences in responses to IFN- versus enzyme inhibitor–based therapies, assessment of the mean VK is a mediocre approach to interpretation of the data. A more rigorous, individual VK based analysis of the kinetics of the different pieces of the puzzle: HBV DNA, HBeAg, HBsAg, and ALT, as function of different baseline factors (e.g., HBV genotype) is needed for each of the various therapies available. A better understanding of the underlying therapeutic mechanisms of Peg-IFN and enzyme inhibitors in relation to individual host and viral factors may then be achieved. Such insights may lead to more innovative approaches to combination therapy, dynamically using the favorable properties of each drug, and make the treatment for chronic HBV infection more successful.

Ancillary