Early on-treatment prediction of response to peginterferon alfa-2a for HBeAg-negative chronic hepatitis B using HBsAg and HBV DNA levels


  • Potential conflict of interest: Dr. Ferenci is a consultant for, advises, is on the speakers' bureau of, and received grants from Roche. Dr. Janssen is a consultant for and received grants from Roche, Bristol-Myers Squibb, Schering-Plough, Novartis, and Gilead. Dr. Flisiak received grants from Roche.


Peginterferon alfa-2a results in a sustained response (SR) in a minority of patients with hepatitis B e antigen (HBeAg)–negative chronic hepatitis B (CHB). This study investigated the role of early on-treatment serum hepatitis B surface antigen (HBsAg) levels in the prediction of SR in HBeAg-negative patients receiving peginterferon alfa-2a. HBsAg (Architect from Abbott) was quantified at the baseline and during treatment (weeks 4, 8, 12, 24, 36, and 48) and follow-up (weeks 60 and 72) in the sera from 107 patients who participated in an international multicenter trial (peginterferon alfa-2a, n = 53, versus peginterferon alfa-2a and ribavirin, n = 54). Overall, 24 patients (22%) achieved SR [serum hepatitis B virus (HBV) DNA level < 10,000 copies/mL and normal alanine aminotransferase levels at week 72]. Baseline characteristics were comparable between sustained responders and nonresponders. From week 8 onward, serum HBsAg levels markedly decreased in sustained responders, whereas only a modest decline was observed in nonresponders. However, HBsAg declines alone were of limited value in the prediction of SR [area under the receiver operating characteristic curve (AUC) at weeks 4, 8, and 12 = 0.59, 0.56, and 0.69, respectively]. Combining the declines in HBsAg and HBV DNA allowed the best prediction of SR (AUC at week 12 = 0.74). None of the 20 patients (20% of the study population) in whom a decrease in serum HBsAg levels was absent and whose HBV DNA levels declined less than 2 log copies/mL exhibited an SR (negative predictive value = 100%). Conclusion: At week 12 of peginterferon alfa-2a treatment for HBeAg-negative CHB, a solid stopping rule was established with a combination of declines in serum HBV DNA and HBsAg levels from the baseline. Quantitative serum HBsAg in combination with HBV DNA enables on-treatment adjustments of peginterferon therapy for HBeAg-negative CHB. (HEPATOLOGY 2010)

Chronic hepatitis B virus (HBV) infection affects 350 to 400 million people worldwide and is responsible for 1 million deaths every year.1 Hepatitis B e antigen (HBeAg)–negative chronic hepatitis B (CHB) represents a late phase in the course of the infection, which is recognized worldwide with increasing prevalence.2 Therapeutic intervention is often indicated for HBeAg-negative patients because spontaneous remission rarely occurs and patients have more advanced liver disease in comparison with HBeAg-positive patients.3

In the last decade, great strides have been made in the treatment of CHB, but the management of the HBeAg-negative type remains difficult. Nucleos(t)ide analogs are able to maintain suppression of viral replication in the majority of HBeAg-negative patients and are well tolerated,4, 5 but it is highly uncertain whether oral antiviral therapy can be discontinued.6-8 In contrast to nucleos(t)ide analogs, 1 year of peginterferon therapy can result in an off-treatment sustained response (SR) in HBeAg-negative patients.9, 10 However, treatment with peginterferon is often complicated by the occurrence of side effects, and a minority of patients with HBeAg-negative disease achieve SR. It is therefore a major challenge to identify patients who are likely to benefit from peginterferon therapy as early as possible during the treatment course.

HBV DNA quantification is widely used as a marker of viral replication to assess the response to nucleos(t)ide analogs, but the prediction of response to peginterferon by means of serum HBV DNA levels is difficult.11, 12 Advances in technology have enabled the development of a quantitative assay for hepatitis B surface antigen (HBsAg). The serum concentration of HBsAg appears to reflect the amount of covalently closed circular DNA (cccDNA) in the liver, which acts as a template for the transcription of viral genes.13, 14 Recently, several studies have suggested that serum HBsAg levels may be indicative of the likelihood of response to interferon-based therapy.15-17 The aim of this study was to clarify the role of early on-treatment quantitative serum HBsAg in the prediction of SR in patients with HBeAg-negative CHB treated with peginterferon alfa-2a.


AIC, Akaike's information criterion; ALT, alanine aminotransferase; AUC, area under the receiver operating characteristic curve; cccDNA, covalently closed circular DNA; CHB, chronic hepatitis B; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; IQR, interquartile range; SD, standard deviation; SR, sustained response.

Patients and Methods


HBsAg levels were measured in sera from a total of 107 of 133 patients with HBeAg-negative CHB who participated in an investigator-initiated, multicenter, randomized, double-blind, controlled trial.9 Patients were randomly assigned in a one-to-one ratio to receive 180 μg of peginterferon alfa-2a weekly and 1000 (body weight <75 kg) or 1200 mg (body weight ≥75 kg) of ribavirin daily or 180 μg of peginterferon alfa-2a weekly and placebo daily. The duration of therapy was 48 weeks, and this was followed by a 24-week observation period. Patients attended the outpatient clinic every 4 weeks. Results at the end of treatment (week 48) and at the end of follow-up (week 72) have been reported previously.9 Patients who were treated according to the protocol and completed the follow-up phase were selected for the present study.

Patients eligible for the original study had been positive for HBsAg for more than 6 months, were HBeAg-negative and antibody to HBeAg–positive on two occasions within 2 months before randomization, had two episodes of elevated serum alanine aminotransferase (ALT) levels (>1.5 but ≤10 times the upper limit of normal of the normal range) within 2 months before randomization, and had a serum HBV DNA level >100,000 copies/mL (17,143 IU/mL). Exclusion criteria were as follows: antiviral or immunosuppressive therapy within the previous 6 months; coinfection with hepatitis C, hepatitis D, or human immunodeficiency virus; other acquired or inherited causes of liver disease; and preexisting cytopenia or decompensated liver disease. The study was conducted in accordance with the guidelines of the Declaration of Helsinki and the principles of good clinical practice. All patients gave written, informed consent.

Laboratory Measurements.

Serum HBsAg was quantified in samples taken at the baseline, during the treatment period (weeks 4, 8, 12, 24, 36, and 48), and during follow-up (weeks 60 and 72) with the Architect HBsAg assay (Abbott Laboratories; range = 0.05-250 IU/mL).18 Serum HBV DNA was measured at the same time points with the TaqMan polymerase chain reaction assay [TaqMan HBV assay, Roche Diagnostics; lower limit of quantification = 35 copies/mL (6 IU/mL)]. Aminotransferases were measured locally at the time of sampling in accordance with standard procedures. The HBV genotype was assessed with the INNO-LiPA assay (Innogenetics).

Liver Histology.

Liver biopsy was performed in all patients within 1 year before randomization. The necroinflammation grade (range = 0-18) and fibrosis stage (range = 0-6) were assessed with the Ishak scoring system.19

Statistical Analysis.

SR, the predefined primary endpoint in the original study, was defined according to the European Association for the Study of the Liver guidelines as the combined presence of serum HBV DNA levels less than 10,000 copies/mL (1714 IU/mL) and normalization of ALT at the end of follow-up (week 72).20 The association between the baseline factors and SR was assessed by univariate logistic regression analyses. Predictive values of early on-treatment serum HBsAg levels as well as HBV DNA and ALT levels (weeks 4, 8, and 12) were explored with logistic regression analysis techniques. Discrimination, which is the ability to distinguish patients who will develop SR from those who will not, was quantified by the area under the receiver operating characteristic curve (AUC). The best model fit was assessed by a comparison of the AUC and Akaike's information criterion (AIC). Hereafter, the optimal cutoff values for serum HBsAg and HBV DNA levels during treatment were established with the use of explanatory plots and the maximum chi-square approach to find a clinically useful rule for (dis)continuation of therapy.21 SPSS version 15.0 (SPSS, Inc., Chicago, IL) and SAS 9.2 (SAS Institute, Inc., Cary, NC) were used to perform statistical analyses. All statistical tests were two-sided and were evaluated at the 0.05 level of significance.


SR Rate.

Twenty-four of 107 patients (22%) developed SR. The number of sustained responders was comparable between the peginterferon alfa-2a monotherapy group and the peginterferon alfa-2a and ribavirin combination therapy group [14 of 53 (26%) versus 10 of 54 patients (19%), respectively, P = 0.33]. The two treatment groups were therefore pooled for further analysis. Among the 24 sustained responders, one patient cleared of HBsAg from serum and developed antibody to HBsAg.

Baseline Characteristics.

Baseline characteristics of the 107 patients are shown in Table 1. The mean pretreatment serum HBsAg level was 3.8 log IU/mL (range = 1.1-5.0 log IU/mL), and the mean serum HBV DNA level was 6.8 log copies/mL (range = 4.3-9.5 log copies/mL); both were stable during the screening period. There was no significant correlation between serum HBsAg and other factors at the baseline, including serum HBV DNA and ALT levels, HBV genotype, age, gender, body mass index, and liver histology. Baseline characteristics, including age, gender, HBV genotype, serum ALT, HBV DNA, and HBsAg levels, and liver necroinflammatory and fibrosis scores, were comparable for patients with and without SR (Table 1).

Table 1. Baseline Characteristics According to SR
CharacteristicAll Patients (n = 107)SR+ (n = 24)SR− (n = 83)P Value
  • Abbreviations: IQR, interquartile range; SD, standard deviation.

  • *

    Multiples of the upper limit of normal range.

  • Ishak fibrosis score of 5 to 6.

Age, years [mean (SD)]42 (10)41 (11)42 (10)0.59
Male [n (%)]77 (72.0)16 (66.7)61 (73.5)0.51
Ethnicity [n (%)]   0.73
 Caucasian102 (95.3)23 (95.8)79 (95.2) 
 Other5 (4.7)1 (4.2)4 (4.8) 
HBV genotype [n (%)]   0.13
 A15 (14.0)015 (18.1) 
 D85 (79.4)23 (95.8)62 (74.7) 
 Other/mixed7 (6.5)1 (4.2)6 (7.2) 
ALT [median (IQR)]*2.3 (1.6-4.1)2.0 (1.7-3.9)2.3 (1.6-4.1)0.82
HBV DNA, log copies/mL [mean (SD)]6.8 (1.2)6.9 (1.2)6.7 (1.2)0.52
HBsAg, log IU/mL [mean (SD)]3.8 (0.5)3.8 (0.4)3.8 (0.6)0.80
Liver necroinflammation [median (IQR)]5 (4-7)5 (4-6)5 (4-7)0.52
Liver fibrosis [median (IQR)]3 (1-3)2 (1-3)3 (1-3)0.57
Cirrhosis [n (%)]3 (2.8)03 (3.6)1.0

Serum HBsAg and HBV DNA Levels During Treatment and Follow-Up.

Overall, the mean serum HBsAg concentration decreased significantly after 48 weeks of therapy (mean change versus the baseline = −0.47 log IU/mL, P < 0.001). HBsAg remained at end-of-treatment levels during the posttreatment follow-up (mean change at week 72 versus the baseline = −0.52 log IU/mL, P < 0.001). Serum HBV DNA levels declined significantly during the treatment period as well (mean change at week 48 versus the baseline = −3.29 log copies/mL, P < 0.001). In contrast to HBsAg levels, HBV DNA levels relapsed after treatment discontinuation (mean change at week 72 versus the baseline = −1.55 log copies/mL, P = 0.004).

A weak positive correlation was present between serum HBsAg and HBV DNA levels when all available samples were considered (r = 0.35, P < 0.001). From the baseline until week 12, serum HBsAg and HBV DNA levels were not correlated (r < 0.15, P > 0.11). However, the correlation became stronger at the end of the treatment phase (week 48; r = 0.36, P < 0.001) and further increased at the end of follow-up (week 72; r = 0.53, P < 0.001).

Serum HBsAg and HBV DNA Levels According to Response.

The mean HBsAg declines from the baseline for sustained responders and nonresponders are shown in Fig. 1A. During the first 8 weeks of therapy, the mean serum HBsAg levels remained stable in both patient groups (Fig. 1A). From week 8 onward, however, HBsAg levels markedly decreased among the 24 patients who developed SR, whereas only a modest decrease in HBsAg levels was observed in patients who failed to achieve SR (P < 0.05 for the comparison of HBsAg declines between patients with and without SR at all time points from week 8 with correction for multiple testing).

Figure 1.

Mean change versus the baseline for (A) HBsAg and (B) HBV DNA levels in patients who achieved SR and those who did not.

Mean HBV DNA declines from the baseline for patients with and without SR are displayed in Fig. 1B. A significant reduction in the serum HBV DNA level was observed at week 4 in contrast to the later on-treatment decline in the serum HBsAg level. Although the magnitude of the on-treatment HBV DNA decline was larger in patients who eventually developed SR (P < 0.01 for the comparison of HBV DNA declines between patients with and without SR at all time points with correction for multiple testing), HBV DNA levels also decreased substantially in patients who did not achieve SR (Fig. 1B).

Serum ALT levels behaved similarly in sustained responders and nonresponders during the treatment period and were not predictive of SR.

Prediction of SR.

The relationship between serum HBsAg and HBV DNA levels and the subsequent achievement of SR was assessed during weeks 4, 8, and 12 of therapy. The performance of HBsAg and HBV DNA declines from the baseline with respect to SR was superior to absolute values. The AUC for declines in HBsAg and HBV DNA levels is shown in Fig. 2. The reductions in HBsAg levels at weeks 4 and 8 were not associated with SR by logistic regression analysis. The HBsAg decline at week 12 was significantly associated with SR, but the overall discrimination remained unsatisfactory (AUC at weeks 4, 8, and 12 = 0.59, 0.56, and 0.69, respectively).

Figure 2.

AUC for the HBsAg decline from the baseline, the HBV DNA decline from the baseline, and a combination of these two markers for the prediction of SR.

In contrast to HBsAg declines, HBV DNA declines were associated with SR as early as week 4 of treatment. HBV DNA declines performed better with respect to the prediction of SR than HBsAg declines at weeks 4, 8, and 12 (Fig. 2). The best model fit, however, which was based on the AUC and AIC, was achieved through a combination of HBsAg and HBV DNA declines (AUC at week 12 = 0.74). The performance of the model at week 24 did not improve significantly in comparison with the performance at week 12 (P = 0.37). The treatment regimen was not associated with SR when it was added to the logistic regression models (P ≥ 0.35 for all time points).

Treatment Algorithm.

To find a clinically useful guiding rule, optimal cutoff values for a combination of HBsAg and HBV DNA declines at week 12 were established. We aimed to identify a stopping rule enabling discontinuation of therapy in patients who have a very low chance of SR while maintaining more than 95% of sustained responders on treatment. Serum samples for measuring HBsAg and HBV DNA declines at week 12 were available for 102 patients. Figure 3 illustrates the chance of SR within four patient groups defined according to the presence of an HBsAg decline and/or an HBV DNA decline ≥2 log copies/mL at week 12. None of the patients in whom a decline in serum HBsAg levels was absent and whose HBV DNA levels decreased less than 2 log copies/mL (20% of the study population) exhibited an SR (negative predictive value = 100%). In contrast, patients in whom both these virological declines were achieved had the highest probability of SR (39%), which was almost double the overall response rate of 22%. Rates of SR were intermediate in patients with either a ≥2 log copies/mL decline in HBV DNA (24%) or a decline in the HBsAg concentration only (25%). Separate analyses for the two treatment regimens (peginterferon alfa-2a with or without ribavirin) resulted in identical cutoff values for HBsAg and HBV DNA declines at week 12.

Figure 3.

Algorithm showing the chances of SR based on (1) an HBsAg decline and (2) an HBV DNA decline ≥2 log copies/mL at week 12 versus the baseline.


Patients with HBeAg-negative CHB represent a difficult-to-treat population at high risk for liver-related complications.3 All of the major practice guidelines recommend both peginterferon and nucleos(t)ide analogs as initial treatment options,20, 22, 23 but the optimal choice for individual patients remains controversial. Because of the higher chance of disease relapse after treatment discontinuation, peginterferon is less often prescribed to HBeAg-negative patients versus HBeAg-positive patients. A treatment course with peginterferon should, however, be considered for HBeAg-negative patients with a high likelihood of response because a finite treatment course can lead to an off-treatment SR. Otherwise, prolonged or indefinite treatment with a nucleos(t)ide analog is likely. Unfortunately, baseline predictors of response to peginterferon are poorly defined in comparison with HBeAg-positive disease.24, 25 One study reported that the baseline serum HBV DNA and ALT levels, patient age and gender, and infecting HBV genotype were significantly associated with the response to peginterferon alfa-2a with or without lamivudine therapy,26 but this was not confirmed in our patient population. Recent studies on peginterferon in HBeAg-negative patients have focused on the identification of markers allowing on-treatment prediction of response.15-17

We found that the accurate prediction of SR to peginterferon for HBeAg-negative disease in an early treatment phase is not possible on the basis of serum HBsAg levels alone. However, combining on-treatment declines in serum HBsAg and HBV DNA concentrations resulted in a solid stopping rule. At week 12, the absence of a decline in HBsAg levels combined with a decrease in HBV DNA levels of less than 2 log copies/mL identified a substantial proportion of the total study population (20%) in which therapy could be discontinued without a loss of sustained responders. In contrast, patients in whom both declines were present had the highest probability of SR (39%). This group should be encouraged to complete the 48-week treatment phase because these patients are the most likely group to benefit from therapy. Table 2 provides recommendations for (dis)continuation of therapy for patient groups based on the chance of developing SR. Obviously, the final decision to (dis)continue therapy is at the discretion of the treating physician, who should take into account other factors such as drug tolerability as well. Another important finding is that a guiding rule before 12 weeks of therapy could not be established because discrimination of serum HBsAg and HBV DNA levels during the first 8 weeks of treatment did not prove sufficient. Also, the decision to discontinue therapy should not be postponed because the prediction of SR did not improve significantly at week 24 compared to week 12.

Table 2. Recommendations for Continuation of Peginterferon Alfa-2a Therapy for HBeAg-Negative Patients with CHB at Week 12
Week 12 Versus the BaselineChance of SRRecommendation to Continue
HBsAg DeclineHBV DNA Decline ≥2 Log Copies/mL
yesyesHighstrong recommendation for continuation

The kinetics of serum HBsAg and HBV DNA levels clearly differed during the treatment phase. HBV DNA levels decreased throughout the entire treatment period, whereas a later decline was observed in serum HBsAg levels. HBsAg and HBV DNA levels were not correlated at the baseline and early during the treatment phase, and this further underlined the additional value of HBsAg levels in the prediction of SR. The added information that is provided by a quantitative assessment of serum HBsAg may be explained by the dual antiviral and immunomodulatory mode of action of peginterferon. The on-treatment reduction in serum HBV DNA primarily reflects the direct antiviral effect of peginterferon. In contrast, the decline in serum HBsAg may be a marker of its immunomodulatory effects, which result in gradual clearance of infected hepatocytes from the liver through the induction of cytotoxic T cell activity.27 In line with these findings, it has been demonstrated that reductions in serum HBsAg mirror the decline in intrahepatic cccDNA.13, 14

Recently, high predictive values for on-treatment HBsAg declines at weeks 12 and 24 with respect to sustained virological response (HBV DNA <70 copies/mL) were reported in a cohort of 48 patients treated with peginterferon alfa-2a for 48 weeks.17 This finding was not confirmed in our larger study population, which was derived from a randomized controlled trial. This discrepancy may be generated by the substantial difference in response rates between the two studies. In the study by Moucari et al.,17 25% of patients developed a sustained virological response. This response rate is substantially higher than that in any peginterferon study for HBeAg-negative patients and suggests that a selection bias may have affected the results of this retrospective study.

In our study, SR had previously been defined as the combined presence of a serum HBV DNA level <10,000 copies/mL and a normal ALT level at 6 months after treatment discontinuation. One could argue that the HBV DNA threshold should have been set at a lower level. Indeed, the off-treatment undetectability of serum HBV DNA by a sensitive polymerase chain reaction assay is a major virological endpoint and is strongly associated with HBsAg clearance from serum in the years afterward.28 However, these preferred treatment endpoints occur infrequently in HBeAg-negative patients treated with peginterferon. In fact, another important goal of therapy for HBeAg-negative CHB is the induction of the HBsAg inactive carrier phase. Our endpoint of a serum HBV DNA level <10,000 copies/mL combined with a normal ALT level appears to differentiate reliably between inactive carriers and patients with HBeAg-negative CHB.29 In addition, large population studies have shown that HBsAg-positive patients with an HBV DNA concentration below this level of viral replication have a reduced risk of progression to cirrhosis and hepatocellular carcinoma.30-32 Furthermore, this HBV DNA threshold and the duration of follow-up correspond with the definition of response to peginterferon therapy according to the recent European guidelines and the pivotal studies on peginterferon in CHB, respectively.10, 20, 33

The large majority of our patients were of Caucasian origin and were infected with HBV genotypes A and D. Responsiveness to interferon-based therapy appears to be lower in patients with genotype D versus patients with other genotypes, and this may explain the limited efficacy of peginterferon in our study population.9, 10, 26, 34 A recent retrospective analysis of 264 HBeAg-negative patients treated with peginterferon alfa-2a alone or in combination with lamivudine reported that pretreatment HBsAg levels varied according to the genotype. The highest concentrations were found in patients infected with genotypes A and D. Although serum HBsAg levels decreased during the treatment phase for all genotypes, the HBsAg decline was least pronounced in patients with genotype D.35 Therefore, our data on the decline in HBsAg levels need to be confirmed in patients with genotypes B and C.

In summary, the current study shows that a combination of early quantitative serum HBsAg and HBV DNA levels allows the best selection of patients with HBeAg-negative CHB who will not respond to a 48-week course of peginterferon alfa-2a therapy. The discontinuation of peginterferon therapy and a switch to an alternative treatment appear to be indicated in patients without a decline in HBsAg levels combined with a decline in HBV DNA levels of less than 2 log copies/mL at week 12.


In addition to the authors, the study group includes the following members: in Austria, P. Munda, T. M. Scherzer, and K. Staufer (Medical University of Vienna, Vienna) and W. Vogel and I. Graziadei (Innsbruck Medical University, Innsbruck); in Germany, G. Gerken (University Hospital Essen, Essen) and C. Niederau (St. Josef Hospital Oberhausen, Oberhausen); in Greece, G. Germanidis (Papageorgiou General Hospital, Thessaloniki), G. Hatzis (Laikon General Hospital, Athens), G. Kitis and P. Xiarchos (George Papanikolaou Hospital, Thessaloniki), M. Raptopoulou-Gigi, E. Gigi, and E. Sinakos (Aristotle University of Thessaloniki, Thessaloniki), and I. Vafiadis-Zouboulis, P. Nicolaou, and G. Paraskevi (University of Athens Medical School, Athens); in Italy, P. Grima (S. Caterina Novella Hospital, Galatina), G. Montalto (Universita di Palermo, Palermo), M. Russello (Azienda Ospedaliera Garibaldi–Nesima, Catania), G. Scifo (Presidio Ospedaliero Muscatello, Augusta), A. Spadaro (University Hospital Messina, Messina), and S. Tripi (Universita di Palermo, Palermo); in the Netherlands, M. F. C. Beersma, M. L. op den Brouw, S. D. Diepstraten, G. J. van Doornum, C. van der Ent, A. Heijens, A. Keizerwaard, M. Ouwendijk, G. Ramdjan, L. A. van Santen, S. M. J. Scherbeijn, M. Schutten, W. Tielemans, A. M. Woltman, and P. E. Zondervan (Erasmus MC–University Medical Center, Rotterdam); in Poland, A. Kalinowska and T. W. Lapinski (Medical University of Bialystok, Bialystok), W. Halota (Hospital Bydgoszcz, Bydgoszcz), T. Mach (Medical College, Jagiellonian University, Krakow), and M. Pazgan-Simon (Medical University Wroclaw, Wroclaw); and in Turkey, G. Ersoz (Ege University Faculty of Medicine, Izmir), N. Sasmaz (Turkiye Yuksek lhtisas Hospital, Ankara), B. Pinarbasi (Istanbul University Medical School, Istanbul), N. Örmeci and Z. Balik (Ankara University School of Medicine, Ankara), and H. Senturk (Istanbul University Cerrahpasa Medical School, Istanbul).