Comparison of Serum Hepatitis B Virus RNA Levels and Quasispecies Evolution Patterns between Entecavir and Pegylated-Interferon Mono-treatment in Chronic Hepatitis B Patients

Hepatitis B virus (HBV) RNA may independently predict virological and serological response. This study aimed to compare dynamic changes in serum HBV RNA levels and HBV quasispecies evolution patterns between entecavir and pegylated-interferon mono-treatment in chronic hepatitis B patients and to determine the clinical significance during treatment. TaqMan real-time PCR was used for quantitative analysis. HBV RNA levels were retrospectively determined in serial serum samples from 178 chronic hepatitis B patients who received either entecavir or pegylated-interferon treatment.

number of mutations that distinguished any two sequences from the population, was evaluated by the mean genetic distance (d; also called Hamming distance). Estimation of mean genetic distance was conducted using the Tamura 3-parameter model. Before downstream analysis, error correction was carried out by using previously published software (13,14). All parameters were calculated using MEGA 7.0 software.
Statistical analyses. HBV DNA, HBsAg, HBeAg, and HBV RNA data were log 10 transformed prior to analyses and expressed as mean (standard deviation). The values were compared by Student's t test or the Mann-Whitney test as appropriate. To assess the distribution of HBV genotypes in different patient groups, a chi-square test ( 2 ) was applied. Logistic regression was used to identify factors associated with HBeAg seroconversion. Receiver operating characteristics (ROC) analysis was performed to evaluate the prediction value in HBeAg seroconversion. The area under the ROC curve (AUROC) was calculated to compare different variables. Optimal cutoff values were obtained from Youden's index (J ϭ sensitivity ϩ specificity Ϫ 1). The sensitivity, specificity, positive predictive value, and negative predictive value were calculated at the optimal cutoff value. Graphs were plotted using GraphPad Prism 7.0. Statistical software used included SPSS 24.0 and MedCalc 15.2.2. A two-sided P value of less than 0.05 was considered statistically significant.
Data availability. The data that support the findings of this study are available from the corresponding author upon reasonable request.

RESULTS
Demographic, clinical, and laboratory data. Study population characteristics are shown in Table 1. Overall, 178 CHB patients were enrolled in this study. The mean age was 37.8 years, and 127 patients were male and 51 female. The HBV genotype distribution was 59 (35%) for genotype B and 110 (65%) for genotype C. Among the patients, 122 were treated with ETV (72 were HBeAg positive) and 56 were treated with Peg-IFN (44 were HBeAg positive). No differences between the two treatment groups were found in HBV genotype or the mean levels of ALT, HBsAg, HBeAg, HBV DNA, or HBV RNA at baseline.
Sequential changes in serum HBV RNA levels during ETV versus Peg-IFN treatment. Before treatment, serum HBV RNA levels, as well as HBV DNA and HBsAg levels, showed no difference between ETV-and Peg-IFN-treated groups. During 48 weeks of treatment, serum HBV RNA could still be detected even when HBV DNA levels were below the LLD in some patients in both treatment groups (Fig. 1A). In the Peg-IFN treatment group, 65.5% of patients had undetectable HBV DNA levels, and the proportion of patients with HBV RNA levels below the LLD was 31.0% at week 48. Meanwhile, in the ETV treatment group, 90.0% of patients had undetectable HBV DNA levels, but only 18.0% of patients had HBV RNA levels below the LLD at week 48.
ETV showed more powerful suppression of HBV DNA levels than Peg-IFN at all time points during treatment. However, the inhibition of HBV RNA levels was less efficient in ETV-treated patients than in Peg-IFN-treated patients, and thus, HBV RNA levels remained higher than HBV DNA levels at all time points during ETV treatment, whereas HBV DNA levels could be either higher or lower than HBV RNA levels in Peg-IFN-treated patients (Fig. 1B). Peg-IFN induced stronger decreases in HBV RNA levels from baseline to week 4 (1.28 versus 0.54 copies/ml), week 24 (2.42 versus 1.73 copies/ml), and week 48 (2.92 versus 2.11 copies/ml) than did ETV mono-treatment ( Fig. S1A in the supplemental material). A significant difference was also found between the reductions of HBsAg levels in the two treatment groups at week 48 posttreatment. HBsAg levels decreased from baseline by only 0.42 log 10 copies/ml in the ETV treatment group and by 0.89 log 10 copies/ml in the Peg-IFN treatment group at week 48 ( Fig. S1C). Serum HBV RNA levels in relation to HBeAg seroconversion. The mean serum HBV RNA level was 6.41 (1.86) copies/ml at baseline; it was higher in HBeAg-positive patients, at 7.28 (1.48) copies/ml, than in HBeAg-negative patients, where it was 4.75 (1.32) copies/ml (P Ͻ 0.01). HBV RNA levels declined continuously, though they remained higher in HBeAg-positive patients than in HBeAg-negative patients at all subsequent time points in both treatment groups (P Ͻ 0.05). The proportions of patients who achieved HBeAg seroconversion were 11/72 (15.3%) in the ETV treatment group and 16/44 (36.3%) in the Peg-IFN treatment group.
ETV-treated patients with subsequent HBeAg seroconversion had significantly lower HBV RNA levels at baseline than those who did not achieve HBeAg seroconversion ( Fig.  2A), and stronger declines were observed in the HBeAg seroconversion subgroup than in the no-seroconversion subgroup at all time points. In the Peg-IFN treatment group, patients with subsequent HBeAg seroconversion also had lower HBV RNA levels at baseline and during treatment than those without subsequent seroconversion, but a significant difference was only observed at week 24 (Fig. 2B). In HBeAg-negative patients, a marked decline was also observed in HBV RNA levels during treatment.
Prediction of HBeAg seroconversion by serum HBV RNA levels. Lower levels of HBsAg, HBV DNA, and HBV RNA at baseline were correlated with HBeAg seroconversion in all 178 patients by univariate logistic regression, while age, gender, ALT level, and HBV genotype were not associated with HBeAg seroconversion. In multivariate analysis, only the baseline HBV RNA level was an independent predictor of HBeAg seroconversion (Table 2).
We further performed ROC analysis to evaluate the prediction value of factors associated with HBeAg seroconversion. At baseline, the AUROCs of HBV RNA levels were highest in predicting HBeAg seroconversion in both ETV (  cutoff values of serum HBV RNA levels and the corresponding sensitivities, specificities, positive predictive values, and negative predictive values were calculated ( Table 3). The predictive value of serum HBV RNA levels was better in the ETV treatment group than in the Peg-IFN treatment group. Serum HBV RNA and DNA QS complexity and diversity. Because of the low viral load and other technical reasons, we finally obtained high-quality sequencing data for 122 paired HBV RNA and DNA samples at baseline (ETV, 87 paired samples, and Peg-IFN, 35 paired samples). To compare HBV RNA and HBV DNA QS characteristics, both QS complexity and viral genetic distance were calculated. HBV RNA QS complexity and diversity were significantly correlated (r ϭ 0.4183, P Ͻ 0.01), but neither complexity nor diversity was correlated with quantitative HBV RNA levels (Fig. S3). Overall, baseline quasispecies were different between HBV RNA and HBV DNA. (i) Regarding genetic complexity, no significant difference was found between HBV RNA and HBV DNA QS complexity at baseline (Fig. 3A and C). (ii) Regarding genetic diversity, the mean genetic distance of baseline HBV RNA was significantly greater than that of HBV DNA in both treatment groups ( Fig. 3B and D).
HBV RNA QS evolution patterns in ETV versus Peg-IFN treatment groups. At baseline, HBV RNA QS complexities and mean distances showed no differences between ETV and Peg-IFN treatment groups. To investigate the evolution patterns of HBV QS, both QS complexity and viral genetic distance were also calculated at week 4. Overall, the virus quasispecies were successfully amplified, and matched QS data were obtained in 48 patients at week 4 (ETV, 36 patients, and Peg-IFN, 12 patients). (i) Regarding genetic complexity, the QS complexity of HBV RNA trended to a significant decrease after 4 weeks of treatment in both the ETV and the Peg-IFN treatment group (Fig. 3E). (ii) Regarding genetic diversity, a significant decline in HBV RNA diversity was only observed in the ETV treatment group at week 4 (Fig. 3F). The diversity of HBV RNA was significantly lower than the diversity of HBV DNA at week 4 in the ETV treatment group (Fig. 3B). However, HBV DNA diversity was lower than HBV RNA diversity at week 4 in the Peg-IFN treatment group (Fig. 3D).
Patients in the Peg-IFN treatment group with positive HBV RNA QS amplification results at week 4 had significantly lower baseline mean genetic distances (Fig. 4A); meanwhile, they had significantly higher week 4 HBV RNA levels (Fig. 4B). Therefore, we analyzed the relationship between baseline mean genetic distances and declines in HBV RNA levels. Patients with HBV RNA level declines greater than 1 log 10 copies/ml after 4 weeks of treatment had greater mean genetic distances, especially in the Peg-IFN treatment group (Fig. 4C). Thus, a greater baseline mean genetic distance may indicate a rapid HBV RNA level decline to Ͼ1 log 10 copies/ml at week 4 posttreatment in the Peg-IFN treatment group (Fig. 4D).

DISCUSSION
There are five different types of viral RNAs, including the 3.5-kb pregenomic RNA (pgRNA) and precore RNA, 2.4-kb and 2.1-kb HBsAg RNAs, and 0.7-kb HBx RNA (15). However, the origin and the form of serum HBV RNA are still unclear. It is generally accepted that HBV RNA can be released into serum and the major form is enveloped 3.5-kb pregenomic-RNA-containing virions (11,16). Due to the diversity of RNA forms, different methods have been developed for HBV RNA quantification; hence, the sequence regions used for HBV RNA amplification are also inconsistent (17). Some previous studies used rapid amplification of cDNA ends with PCR (RACE-PCR), which targets the 3= end of the HBx gene (18)(19)(20). In this study, we used a TaqMan real-time technique that is also applied by some other research groups (11,(21)(22)(23), with specific primers that target the unique sequence of the 3.5-kb-long HBV RNA. Different amplification regions for full-length or truncated HBV RNA fragments may result in different quantification results, so further evidence is needed to prove consistency among these methods.
Peg-IFN has both immunomodulatory and weak direct antiviral actions to control the virus, while NAs are direct inhibitors of HBV polymerase at the reverse transcription step of HBV replication (24). Previous studies showed the kinetics of changes in HBV RNA levels in NA-based (16) or Peg-IFN-based treatment (20). In the present study, the kinetics of HBV RNA were compared between the two kinds of mono-treatment with different antiviral mechanisms. Because HBV RNA levels are related to the antiviral potency of NAs (25), we only selected patients receiving ETV mono-treatment in the NA cohort to eliminate the potential confounders. Our results showed that serum HBV RNA could still be detected even when HBV DNA levels were below the LLD, and this phenomenon was particularly pronounced in the ETV group. ETV treatment induced weaker declines in HBV RNA levels than Peg-IFN treatment, indicating that, in contrast to its strong suppression of HBV DNA levels, the effect of ETV treatment on HBV RNA levels seems to be limited, since the generation of HBV RNA cannot be inhibited by NA directly.
In accordance with recent findings, we observed a stronger decline in HBV RNA levels in HBeAg seroconversion patients (18), and the decline mode in HBeAg-negative patients was similar to that in HBeAg seroconversion patients, suggesting that serum HBV RNA may reflect the virological and immunological properties of the host. Serum HBV pgRNA may also represent a marker for the persistence and transcription activity of HBV cccDNA (26,27), and due to the persistence of cccDNA, HBV pgRNA-containing virus can be continuously produced. NAs have no effect on the transcription of HBV cccDNA, and therefore, in HBeAg seroconversion patients treated with ETV, serum HBV RNA levels were significantly lower at baseline and during treatment, indicating better virus control. In contrast, serum HBV RNA levels were not significantly different in patients receiving Peg-IFN with or without subsequent HBeAg seroconversion. Since Peg-IFN can activate the immune system and have a long-term antiviral effect, which leads to the degradation of the viral RNA and the exhaustion of HBV cccDNA, Peg-IFN treatment may also reduce HBV RNA levels in nonresponders to a certain extent; thus, differences between the two subgroups were not sufficient for statistical analysis.
The proportion of patients achieving HBeAg seroconversion was limited even with Peg-IFN treatment (28); therefore, specific indicators or models to predict HBeAg seroconversion are desirable during antiviral treatment. By using antiviral drugs, espe- cially NAs with high antiviral activity, HBV replication can be significantly inhibited and serum HBV DNA levels can be rapidly decreased to below the LLD in the majority of patients. Serum HBsAg levels often remain unchanged, and its production can be derived from either cccDNA or integrated DNA, so it could persist even after prolonged therapy (29); thus, the usefulness of serum HBV DNA and HBsAg levels for prediction is diminished. Some recent findings revealed the utility of serum HBV RNA levels as a novel biomarker to monitor infection and antiviral response. Luo et al. reported that serum HBV RNA levels could predict HBeAg seroconversion during ETV treatment, but their sample size was relatively small (22). Jia et al. found that serum HBV RNA levels might be helpful for predicting HBeAg seroconversion in patients treated with Peg-IFN alone or in combination with adefovir (23). In the present study, we expanded the sample size and compared the prediction value of serum HBV RNA between ETV and Peg-IFN mono-treatment. The results revealed that HBV RNA levels were better predictors for HBeAg seroconversion than baseline HBV DNA and HBsAg levels. Corroborating the previous observations, the absolute HBV RNA levels were superior to the decline from baseline for predicting response (20). The AUROC scores were better in the entecavir treatment group. In this study, a high negative predictive value (Ͼ90%) at the optimal cutoff value indicated that HBV RNA levels could help identify nonresponders and prompt the consideration of a combination treatment. Therefore, serum HBV RNA during treatment may serve as a potential new biomarker for monitoring antiviral treatment.
Next-generation sequencing technologies allow massive parallel amplification and detection of individual molecules, which makes them useful in HBV QS studies (30). Previous studies have investigated the early changes in HBV DNA QS during lamivudine treatment (31) and ETV treatment (32) and found that the HBV DNA QS heterogeneity pretreatment may also predict virological outcomes (33). However, the evolution pattern of HBV RNA quasispecies during treatment has not been well characterized, particularly in the early stage. Since the virus reservoir changes dynamically over time and might be constantly replenished by a small amount of ongoing viral replication even under antiviral pressure, it would be interesting to observe the changes in QS patterns of HBV virions from the peripheral blood.
Our results showed that baseline HBV RNA QS diversity was significantly higher than that of HBV DNA, which leads us to speculate that the virus replicates and evolves actively, but only a portion of packaged pgRNA is reversed transcribed into HBV DNA and released into the peripheral blood. When under antiviral pressure, Peg-IFN-treated patients showed QS genetic characteristics quite different from those of ETV-treated patients. It appeared that HBV RNA QS heterogeneity was more likely to be reduced during ETV treatment, and viral QS preserved better replicative fitness in the Peg-IFN group. Higher baseline QS diversity may result in better outcomes in Peg-IFN-treated patients than in ETV-treated patients. The amplified fragment encodes terminal protein (TP), which serves as a primer during reverse transcription and guides the synthesis of minus-strand DNA, and it can also prevent the activation of interferon-induced genes in host cells and inhibit the effect on interferon (34). This may be related to the phenomenon in which the QS diversity of this fragment is more correlated with Peg-IFN treatment response.
In conclusion, we compared the different kinetics of serum HBV RNA levels in the two treatment groups and observed a stronger correlation between HBV RNA levels and subsequent HBeAg seroconversion during treatment. Baseline HBV RNA quasispecies diversity is more relevant to the Peg-IFN treatment response, which leads to the hypothesis that higher HBV RNA QS diversity indicates a relatively efficient host immune response. Overall, these findings will help us to better understand the clinical significance of serum HBV RNA and expand our knowledge of HBV RNA quasispecies evolution patterns in the early stage of ETV and Peg-IFN treatment. We propose that serum HBV RNA may serve as a potential biomarker for predicting HBeAg seroconversion during two different kinds of antiviral treatment.