Potential conflict of interest: Dr. Fried has received research grants for clinical research and honoraria for speaking at symposia. He has no stock ownership, equity interest, or patent-licensing arrangement with Roche. Dr. Fried is a consultant for Roche. Dr. Piratvisuth advises, is on the speakers' bureau of, and received grants from Roche, Novartis, Schering-Plough, and GlaxoSmithKline. He also received grants from Bristol-Myers Squibb. Drs. Paik and Cooksley received grants from Roche. Dr. Lau received grants from Roche, Gilead, and Novartis. Dr. Marcellin is a consultant for, advises, is on the speakers' bureau of, and received grants from Roche, Schering-Plough, Gilead, Bristol-Myers Squibb, GlaxoSmithKline, Vertex, Idenix-Novartis, Valeant, Human Genome Sciences, Coley Pharma, Cytheris, Intermune, Wyeth, and Tibotec.
HBeAg and hepatitis B virus DNA as outcome predictors during therapy with peginterferon alfa-2a for HBeAg-positive chronic hepatitis B
Article first published online: 7 JAN 2008
Copyright © 2008 American Association for the Study of Liver Diseases
Volume 47, Issue 2, pages 428–434, February 2008
How to Cite
Fried, M. W., Piratvisuth, T., Lau, G. K. K., Marcellin, P., Chow, W.-C., Cooksley, G., Luo, K.-X., Paik, S. W., Liaw, Y.-F., Button, P. and Popescu, M. (2008), HBeAg and hepatitis B virus DNA as outcome predictors during therapy with peginterferon alfa-2a for HBeAg-positive chronic hepatitis B. Hepatology, 47: 428–434. doi: 10.1002/hep.22065
- Issue published online: 26 JAN 2008
- Article first published online: 7 JAN 2008
- Accepted manuscript online: 7 JAN 2008 12:00AM EST
- Manuscript Accepted: 26 SEP 2007
- Manuscript Received: 28 FEB 2007
- Roche, Basel, Switzerland
- Unknown funding agency. Grant Numbers: K24 DK06614, RR000046
The aims of this study were to evaluate the usefulness of quantitative hepatitis B e antigen (HBeAg) values for predicting HBeAg seroconversion in patients treated with peginterferon alfa-2a and to assess the dynamic changes in quantitative HBeAg during therapy, compared with conventional measures of serum hepatitis B virus DNA. Data were analyzed from a large, randomized, multinational phase III registration trial involving 271 HBV-infected HBeAg-positive patients who received peginterferon alfa-2a plus oral placebo for 48 weeks. HBeAg levels were measured serially during therapy using a microparticle enzyme immunoassay validated with in-house reference standards obtained from the Paul Ehrlich Institute (PEIU/mL). In patients who achieved HBeAg seroconversion, levels of HBeAg consistently decreased during treatment and remained at their lowest level during the 24 weeks of posttreatment follow-up. After 24 weeks of treatment, 4% of patients with the highest levels of HBeAg (≥100 PEIU/mL) achieved HBeAg seroconversion, yielding a negative predictive value of 96%, which was greater than that obtained for levels of HBV DNA (86%). Late responders to peginterferon alfa-2a could also be differentiated from nonresponders by continued decrease in HBeAg values, which were not evident by changes in HBV DNA. Conclusion: These analyses suggest quantitative HBeAg is a useful adjunctive measurement for predicting HBeAg seroconversion in patients treated with peginterferon when considering both sensitivity and specificity compared with serum HBV DNA. (HEPATOLOGY 2008;47:428–434.)
The goal of therapy in patients with chronic hepatitis B is the prevention of cirrhosis, hepatocellular carcinoma, and hepatitis B virus (HBV)-related mortality.1 However, it is difficult to assess these outcomes during short-term clinical trials because of the prolonged natural history of chronic hepatitis B. Surrogate endpoints are used for assessing the efficacy of anti-HBV drugs. Reduction in hepatitis B virus deoxyribonucleic acid (HBV DNA) is an excellent measure of antiviral efficacy. However, a decrease in HBV DNA alone may be transient, particularly once antiviral agents are discontinued.2, 3 Seroconversion of hepatitis B e antigen (HBeAg) is another surrogate endpoint that is considered to be a marker for durable therapeutic response and improved clinical outcome in HBeAg-positive patients with chronic hepatitis B.4, 5 Patients treated with interferon-based therapies who achieved HBeAg seroconversion had decreased liver-related morbidity and mortality.6 Early studies on interferon-based therapies for chronic hepatitis B identified several factors that were associated with a higher likelihood of HBeAg seroconversion, including elevated serum alanine aminotransferase activity, lower levels of hepatitis B virus deoxyribonucleic acid (HBV DNA), and increased histologic activity in biopsy specimens.7–12
Although it is important to identify those patients for whom antiviral treatment will be beneficial, it is also worthwhile to discontinue or consider alternative treatments for patients who have no possibility of a favorable response. An example of the usefulness of such a “stopping rule” has been found in the treatment of hepatitis C virus. The use of on-treatment hepatitis C virus RNA measurements was found to be valuable for identifying infected patients who were unlikely to respond to therapy and has been incorporated into treatment guidelines.13, 14 However, in HBV-infected patients, the relationship between HBV DNA and HBeAg is not as straightforward, because seroconversion does not necessarily occur even though HBV DNA is markedly suppressed.15 This was emphasized in a recent study that involved 715 HBeAg-positive patients who received either entecavir or lamivudine for a minimum of 52 weeks.16 HBV DNA levels after 48 weeks were undetectable in 67% of entecavir-treated patients and 36% of patients given lamivudine. However, despite this profound viral suppression, HBeAg seroconversion occurred in only 21% of those treated with entecavir and a similar rate of 18% of those receiving lamivudine. This equivocal relationship has also been noted in trials involving interferon, as well as nucleoside analogs.17, 18 A recent study has demonstrated that quantitative HBV DNA levels may be an early predictor of nonresponse in patients with chronic HBV.19 An analysis of a phase III trial of telbivudine versus lamivudine showed that clinical efficacy outcomes after 52 weeks of treatment were proportional to HBV DNA levels after 24 weeks of treatment.20
Results from several small studies using conventional interferon have suggested that dynamic monitoring of quantitative HBeAg may help to predict the likelihood of subsequent HBeAg seroconversion.15, 18, 21 Responders to interferon treatment tended to have lower pretreatment HBeAg levels than nonresponders while decreasing HBeAg levels during the first 8 weeks of therapy was also somewhat predictive of subsequent seroconversion. The aims of the current study were to compare the usefulness of measuring quantitative HBeAg and HBV DNA during treatment as a means of predicting HBeAg seroconversion among patients treated with peginterferon alfa-2a.
Patients and Methods
Data for the present analysis of on-treatment predictors of outcome were generated from a large, randomized, multinational phase III registration study.22 A total of 814 HBV-infected HBeAg-positive patients were given either 180 μg/week peginterferon alfa-2a plus oral placebo, 100 mg/day peginterferon alfa-2a plus lamivudine, or 100 mg/day lamivudine alone.22 Patients received treatment for 48 weeks, and follow-up responses were assessed 24 weeks after therapy had been completed. The majority of participants were Asian (87%).
Quantification of HBeAg was performed retrospectively on frozen sera from patients who received peginterferon alfa-2a plus oral placebo daily for 48 weeks. Because no commercial assay was available for measuring HBeAg concentrations, a microparticle enzyme immunoassay (AXSYM HBe 2.0, Abbott Laboratories, Abbott Park, IL) was used and validated in-house using reference standards obtained from the Paul Ehrlich Institute (PEIU/mL). The dynamic range of the assay is 0.15-200 PEIU/mL, with samples of concentrations beyond this range being diluted with fetal bovine serum to ensure linearity. A linear correlation was observed with the reference standards (data not shown).
HBV DNA was measured using in vitro nucleic acid amplification using the COBAS AMPLICOR HBV Monitor (Roche Diagnostics, Branchburg, NJ), with a working range of 4.0 × 102 (lower limit of detection) to 4.0 × 107 copies/mL. The performance of this assay has been evaluated in a routine diagnostic laboratory and was found to be a sensitive means of detecting HBV DNA.23
Pretreatment levels of HBeAg were analyzed within quartiles in order to determine their relationship with subsequent HBeAg seroconversion. Patients who exhibited HBeAg seroconversion at the end of 72 weeks (48 weeks of treatment and 24 weeks of follow-up) were regarded as responders, whereas those who did not achieve seroconversion at this time point were considered nonresponders. A late responder was defined as a patient who did not achieve HBeAg seroconversion by the end of therapy but did achieve seroconversion by the end of the 24-week follow-up. Patients with a viral load of <100,000 copies/mL after 48 weeks of treatment were considered to have achieved a virologic response; those with viral loads >100,000 copies/mL did not attain a virologic response. Receiver operating characteristic curves were generated to compare the relative sensitivity and specificity of HBeAg and HBV DNA as a predictor of HBeAg seroconversion.
HBeAg Serum Levels.
In the originally reported study,22 the highest rate of HBeAg seroconversion occurred in patients treated with peginterferon alfa-2a monotherapy (32%), followed by those treated with peginterferon alfa-2a plus lamivudine (27%) and those treated with lamivudine alone (19%). However, the degree of viral suppression was greatest in those patients treated with combination therapy (−6.9 to −7.5 log copies/mL) at the end of treatment.
Pretreatment levels of HBeAg were analyzed within quartiles in order to determine their relationship with subsequent seroconversion (Table 1). Among patients within the lowest quartile of HBeAg, 54% achieved HBeAg seroconversion [P < 0.001; risk ratio = 1.79 (95% confidence interval 1.33-2.4)]. As pretreatment level of HBeAg increased, the rate of subsequent seroconversion diminished.
|HBeAg at Baseline||Quartile||Total No. of Patients||No. of Patients with Seroconversion||Percentage of Patients with Seroconversion|
Mean HBeAg concentrations over time in patients receiving peginterferon alfa-2a therapy are displayed in Fig. 1. Among 87 patients who achieved HBeAg seroconversion, the concentrations of HBeAg decreased consistently during treatment and remained at the lowest levels during the posttreatment follow-up. Conversely, patients who did not seroconvert (n = 184) showed a modest decrease in HBeAg levels during therapy, but experienced a prompt rebound after treatment was discontinued.
The relationship between levels of HBeAg and subsequent HBeAg seroconversion was evaluated at week 12 and week 24 during therapy (Table 2). More than half of the patients (53%) with HBeAg levels of ≤10 PEIU/mL at week 12 of therapy achieved HBeAg seroconversion at 24 weeks posttreatment. Conversely, increasing HBeAg values at week 12 were associated with a greatly diminished likelihood of seroconversion at 24 weeks posttreatment, so that only 14% of patients with HBeAg levels of ≥100 PEIU/mL at week 12 achieved seroconversion. A similar relationship was observed at week 24 of therapy, where lower levels of HBeAg were associated with a greater likelihood of HBeAg seroconversion. Among those patients with HBeAg ≥100 PEIU/mL at week 24, only 4% subsequently achieved seroconversion.
|HBeAg (PEIU/mL)||Percent of Patients with HbeAg Seroconversion at week 72|
|Week 12||Week 24|
HBV DNA Levels.
As was observed for HBeAg levels, HBV DNA decreased consistently during treatment, and remained at the lowest levels during the posttreatment period in responders (that is, for those patients who exhibited HBeAg seroconversion at 24 weeks posttreatment) (Fig. 2). Patients who did not achieve an HBeAg response also demonstrated decreased HBV DNA levels during treatment, but then experienced a rapid rebound effect following treatment cessation.
Relationship Between HBV DNA and HBeAg Seroconversion.
Stratification of patients who achieved HBeAg seroconversion according to their levels of HBV DNA at week 12 and week 24 of treatment revealed that 64% of patients with HBV DNA of <3 log10 copies/mL by week 12 achieved HBeAg seroconversion, as did 69% of patients at week 24 (Table 3). In addition, increasing HBV DNA values at weeks 12 and 24, signifying less viral suppression, were associated with a greatly diminished likelihood of HBeAg seroconversion. Seroconversion was observed in only 21% of patients whose HBV DNA levels remained >7 log10 copies/mL after 12 weeks of treatment. A similar relationship was noted with HBV DNA levels after 24 weeks of treatment; among patients with HBV DNA >7 log10 copies/mL, only 14% subsequently achieved HBeAg seroconversion.
|HBV DNA (log10 copies/mL)||Patients with HBeAg Seroconversion at Week 72|
|Week 12||Week 24|
Comparison of HBeAg and HBV DNA as Predictors of HBeAg Seroconversion.
Response rates and negative predictive value (NPV) were calculated for HBeAg after 24 weeks of treatment with peginterferon alfa-2a (Fig. 3A). At week 24 of treatment, 52% of patients had a decrease in HBeAg levels to <10 PEIU/mL, 21% had levels between 10 PEIU/mL and 100 PEIU/mL, and 27% had levels >100 PEIU/mL. In all, 52% of patients with the lowest HBeAg levels at week 24 attained seroconversion, as did 20% of patients with intermediate levels of HBeAg and 4% of patients with the highest levels of HBeAg. Analysis yielded an NPV of 96% for HBeAg >100 PEIU/mL at week 24 of treatment (Fig. 3A). In contrast, analyses of data derived after 4 weeks, 8 weeks, and 12 weeks of treatment yielded lower NPVs (data not shown).
The NPVs for HBV DNA levels after 24 weeks of treatment were also calculated (Fig. 3B). It was observed that 21% of patients had HBV DNA >9 log10 copies/mL after 24 weeks of treatment, and of these, 14% achieved HBeAg seroconversion. This was unexpected in view of the persistently high levels of viremia, and yielded a NPV of 86%.
Receiver Operating Characteristics Curves: Prediction of HBeAg Seroconversion.
Receiver operating characteristic curves, developed to aid the prediction of HBeAg seroconversion, were derived from serum values after 24 weeks of treatment for HBeAg and HBV DNA (Fig. 4). Receiver operating curves are useful in evaluating predictive models or other tests that produce output values over a continuous range, because they capture the trade-off between sensitivity and specificity over the range used.24 HBeAg had greater power (P = 0.014) to predict HBeAg seroconversion than HBV DNA.
HBeAg and HBV DNA Relationship Among Late Responders Treated with Peginterferon.
The relationship between HBV DNA and HBeAg levels among late responders was explored and compared with nonresponders (Fig. 5). As might be expected, HBV DNA levels were lower throughout therapy for late responders compared with nonresponders. However, levels reached a plateau during the later few months of therapy, where no further decline in HBV DNA levels occurred despite subsequent HBeAg seroconversion after discontinuation of therapy. HBeAg levels were also consistently lower during therapy in late responders compared with nonresponders. Interestingly, there was a divergence between HBeAg and HBV DNA dynamics among late responders, with a persistent decrease in HBeAg levels, whereas HBV DNA levels remained relatively flat. This reinforces the view that quantitative HBeAg measurements are more predictive of HBeAg seroconversion than HBV DNA levels.
Measurement of HBV DNA in sera remains an important tool for monitoring outcomes in chronic hepatitis B infection. Recent studies have demonstrated an association between level of HBV DNA and the likelihood of development of cirrhosis and hepatocellular carcinoma.25, 26 Similarly, HBV DNA levels in serum have been recommended as a means to stratify patients being considered for treatment and for assessing the efficacy of antiviral therapy, particularly oral nucleoside analogues.27, 28 However, it is also recognized that measurement of HBV DNA may be technically challenging, costly, and subject to variability.29–31 The present study suggests that quantitation of HBeAg in serum may be a useful adjunct to serum HBV DNA assays to fully evaluate the response to antiviral therapy in patients treated with peginterferon.
Measurement of HBeAg has proved useful for gauging the level of HBV DNA replication and, by implication, the activity of liver disease.32–36 Seroconversion of HBeAg to anti-HBe has been associated with a reduction in the viral load, which improves long-term clinical outcome of chronic hepatitis B patients.37–40 Thus, HBeAg seroconversion is considered one of the most important surrogate markers for assessing the durability and efficacy of antiviral therapy in patients with HBeAg-positive chronic hepatitis B. Because numerous clinical studies have confirmed the imperfect association between HBV DNA suppression and HBeAg seroconversion,27, 41–43 it is logical to carefully evaluate HBeAg as a more direct measure and predictor of HBeAg seroconversion.
In the present study, among patients who ultimately achieved HBeAg seroconversion, levels of HBeAg decreased consistently and remained at the lowest levels during the follow-up period. In contrast, a rebound was observed in patients who failed to achieve seroconversion after treatment was discontinued. In addition, analysis of HBeAg levels after 24 weeks of treatment provided a better indicator of nonresponse to peginterferon alfa-2a than did a similar analysis of HBV DNA levels.
Quantifying HBeAg during therapy also allowed demarcation of late responders from nonresponders, a subgroup of patients who were not differentiated by changes in HBV DNA levels. Among late HBeAg responders, HBV DNA levels were lower than for non-HBeAg responders. However, a plateau was noted toward the end of treatment in which HBV DNA levels did not decrease despite subsequent HBeAg seroconversion after discontinuation of treatment. This suggests that HBV DNA levels are less reflective of HBeAg seroconversion than HBeAg levels. Indeed, 14% of participants who had maintained more than 7log10 of HBV DNA in sera at week 24, still achieved HBeAg seroconversion, indicating that stopping rules based on HBV DNA for peginterferon treatment could not be established. These findings from the current study are consistent with the disparity seen between HBV DNA suppression achieved by potent direct antivirals and the relatively low rate of HBeAg seroconversion during clinical trials of these agents.16, 43–45 In clinical practice, therefore, changes to HBV DNA during therapy with peginterferon should be interpreted cautiously and not used as the sole basis for early treatment termination.
Consequently, we sought to compare quantitative HBV DNA and HBeAg measurements in terms of their predictive value for therapeutic outcome. We obtained an NPV of 86% for HBV DNA response after 24 weeks of treatment, which compared unfavorably with an NPV of 96% for HBeAg measurements. This indicates that quantification of HBeAg after 24 weeks of treatment would provide sufficient information for deciding whether premature discontinuation of peginterferon alfa-2a was justifiable. The analysis we conducted in this study suggests that quantification of HBeAg may be a useful clinical tool for predicting the absence of response to peginterferon alfa-2a in an individual patient. Our results suggest that a critical level of reduction of HBeAg may be definable which, if not achieved by patients undergoing treatment, may indicate that a discontinuation of peginterferon monotherapy should be considered.
The main practical difficulty in this approach is the absence of a commercially available assay for quantifying HBeAg, and lends support to the reintroduction of a standardized assay method for HBeAg measurement to augment measures of HBV DNA. Whereas qualitative assays for HBeAg are routinely available, the usefulness of quantitative assays has not been embraced. Although this test is not licensed for commercial use in the United States or Europe, it has some attractive features that could make it useful as an adjunct to current assays. In addition, future studies evaluating the kinetics of HBeAg and the relationship to anti-HBe during antiviral therapy, which can be measured with this platform,21 may provide additional insights into the complex relationship between HBV DNA inhibition and mechanisms of HBeAg seroconversion and further refine treatment algorithms.
- 13NIH Consensus Statement on Management of Hepatitis C: 2002. NIH Consens State Sci Statements 2002; 19: 1–46.
- 17Treatment of chronic hepatitis B infection using interferon. Med J Malaysia 2005; 60: 28–33..
- 20Optimal virologic and clinical efficacy at one year is associated with maximal early HBV suppression in nucleoside-treated hepatitis B patients. In: 41st Annual Meeting of the European Association for the Study of the Liver. Vienna, Austria: April 26-30, 2006., , , , , , et al.
- 34Correlation between the prevalence of serum HBV DNA and immunoserologic HBV markers in the subjects with or without hepatitis. Immunol Invest 1992; 21275–21281., , , .
- 39Hepatitis B e antigen and antibody (HBeAg/anti-HBe). In: GeretyRJ, ed. Hepatitis B. Orlando, FL: Academic Press; 1985: 47–76., .