• Potential conflict of interest: Nothing to report.


We thank Drs Chen and Tang for their interest in our article on quantitative hepatitis B e antigen (HbeAg). Using multivariate logistic regression analysis on baseline characteristics, HBeAg and hepatitis B virus (HBV) DNA were both independent predictors of sustained HBeAg seroconversion in our study (Table 1), in addition to alanine aminotransferase levels (ALT). If we look at the HBV DNA and HBeAg levels during treatment, as the authors request, we do see a correlation between the HBeAg and HBV DNA levels (Table 2). The highest response rates are seen in the patients with lowest HBeAg levels, while within patients with lower or higher HBeAg levels, the level of HBV DNA had little effect (Fig. 1). This finding suggests that both factors could be taken into consideration to monitor on-treatment response to therapy. However, the data presented in our article, including the negative predictive value of HBeAg (compared to HBV DNA), the receiver operating characteristic curves, and the differentiation of late versus nonresponders, suggest that if one had to choose, then HBeAg would be superior to HBV DNA in predicting HBeAg seroconversion after treatment.

Table 1. Multivariate Logistic Regression Analysis on Baseline Variables and Sustained HBeAg Seroconversion 6 Months After Treatment
VariableComparisonOdds Ratio (95% CI)P Value
  1. Abbreviations: Peg-IFN α-2a, peginterferon alfa-2a; LAM, lamivudine

Age10 year increase0.97 (0.81–1.16)0.716
SexFemale versus Male1.49 (0.96–2.30)0.077
 A versus B1.42 (0.41–4.94)0.580
 A versus C1.80 (0.54–6.01)0.342
 A versus D2.05 (0.67–6.26)0.210
 B versus C1.26 (0.86–1.86)0.235
 B versus D1.44 (0.37–5.58)0.598
 C versus D1.14 (0.30–4.28)0.848
RaceAsian versus Caucasian/Other1.20 (0.38–3.82)0.757
Body weight10 kg increase0.98 (0.84–1.14)0.790
ALT1 log10 unit (IU/L) increase2.71 (1.57–4.68)<0.001
HBV DNA1 log10 unit (copies/mL) decrease1.20 (1.08–1.33)0.001
HBeAg1 log10 unit (IU/mL) decrease1.52 (1.25–1.85)<0.001
TreatmentPeg-IFN α-2a versus LAM2.05 (1.34–3.12)<0.001
 Peg-IFN α-2a + LAM versus LAM1.71 (1.11–2.63)0.015
 Peg-IFN α-2a versus Peg-IFN α-2a + LAM0.83 (0.56–1.24)0.366
Table 2. Relationship Between HBeAg and HBV DNA at Baseline and During Treatment
Time pointHBeAg PEIU/mL
Number of patientsN = 39N = 60N = 168
Mean ± SD (log10 copies/mL)7.7 ± 2.089.4 ± 1.3410.6 ± 1.81
Median (log10 copies/mL)8.19.510.0
Range (log10 copies/mL)4.4–12.74.9–13.14.9–16.1
Week 12   
Number of patientsN = 112N = 62N = 90
Mean ± SD (log10 copies/mL)4.6 ± 1.937.4 ± 2.179.2 ± 1.82
Median (log10 copies/mL)
Range (log10 copies/mL)2.3–12.53.1–13.54.9–15.7
Week 24   
Number of patientsN = 138N = 54N = 71
Mean ± SD (log10 copies/mL)4.0 ± 1.717.3 ± 2.489.0 ± 1.64
Median (log10 copies/mL)
Range (log10 copies/mL)2.3–9.63.4–15.15.3–13.1
Figure 1.

Sustained HBeAg seroconversion by HBeAg and HBV DNA levels in patients treated with peginterferon alfa-2a monotherapy. Values expressed for HBV DNA and HBeAg (6.9 log10 copies/mL and 18.6 Paul Ehrlich Institute units [PEIU]/mL, respectively) correspond to median levels at week 12.

The authors also requested clarification on how HBV DNA was quantified given the upper working range of 4.0 × 107 copies/mL. The HBV DNA quantification was performed in a central laboratory according to the manufacturer's instructions.1 The assay had a linear range up to at least 200,000 copies/mL (5.3 log10). If the calculated copies per milliliter were above the linear range of the assay, the original sample was diluted with HBV-negative human plasma, and the test was repeated with the result multiplied by the dilution factor. In the interim, newer improved tests with an extended linear range have become available.2 It remains to be determined whether the aforementioned relationships between HBeAg and HBV DNA are similar with newer assays.

It is also worthy to note that since publication of our manuscript there have been several reports on the potential of hepatitis B surface antigen (HbsAg) quantification as an on-treatment marker of sustained response to peginterferon alfa-2a therapy.3–5 Retrospective analysis of HBsAg levels in the patients treated with peginterferon alfa-2a therapy (with or without lamivudine) in the same study that formed the basis of the analysis of HBeAg referred to here, demonstrated that HBsAg levels ≤1500 IU/mL at week 12 and 24 were associated with high rates of HBeAg seroconversion 6 months after treatment of around 50% which were four-fold to five-fold higher than those in patients with HBsAg levels >20,000 IU/mL.6 Thus, as an alternative to HBeAg quantification (for which there is currently no commercially available assay) HBsAg level may prove to be a more widely available adjunct to help predict response to peginterferon alfa-2a therapy. Future studies will incorporate quantitative assessment of HBsAg and HBeAg during treatment to further validate these findings.

Michael W. Fried*, Peter Button*, Teerha Piratvisuth*, George K. K. Lau*, Patrick Marcellin*, Wan-Cheng Chou*, Graham Cooksley*, Kang-Xian Luo*, Seung Woon Paik*, Yun-Fan Liaw*, Matei Popescu*, * Liver Program, University of North Carolina, Chapel Hill, NC.