Hepatitis B surface antigen levels: association with 5-year response to peginterferon alfa-2a in hepatitis B e-antigen-negative patients

Chronic hepatitis B virus (HBV) infection is estimated to affect 350 million people worldwide and is associated with development of cirrhosis and hepatocellular carcinoma [1]. Serum hepatitis B surface antigen (HBsAg) is usually used as a qualitative marker for the diagnosis of HBV infection, and chronic hepatitis B (CHB) is defined as persistence of HBsAg in the circulation for more than 6 months. HBsAg clearance is considered a marker of complete and definitive remission of HBV activity [2], because it is associated with reduced incidence of hepatocellular carcinoma and improved rates of survival, and is the closest outcome to clinical cure of CHB [3‐5].

Previous studies with conventional interferon alfa-2a in patients with hepatitis B e-antigen (HBeAg)-negative CHB demonstrated that HBsAg clearance could be achieved by this population. Approximately, one-third of patients who sustained a biochemical response for a median period of 7 years post-treatment achieved HBsAg clearance [6]. Subsequently, a finite course of peginterferon alfa-2a has been shown to result in HBsAg clearance, which increases in the years following the completion of the course of therapy. Indeed, rates of HBsAg clearance have been shown to reach 9% at 3 years post-treatment in HBeAg-negative patients [7]. Although HBsAg clearance is the ultimate goal in HBeAg-negative CHB [8], rates of response in the short-term are low and, therefore, it is not usually the primary efficacy endpoint in clinical studies. Alternative markers of response have been identified that are associated with subsequent increased rates of HBsAg clearance and long-term clinical benefits. In HBeAg-negative CHB, HBV DNA suppression ≤2,000 IU/mL is used to define sustained immune control with pegylated interferon in the shorter term (e.g., 1 year post-treatment) as this level of HBV DNA is associated with inactive disease and a low risk of hepatocellular carcinoma [2, 9, 10]. Patients who achieve sustained immune control following interferon-based therapy are likely to clear HBsAg during long-term treatment-free follow-up [11, 12].

The pivotal study of peginterferon alfa-2a in HBeAg-negative patients showed that approximately one-third of patients achieve sustained immune control [13]. Identification of responders either before or early during treatment would be of great benefit as it would not only ensure timely initiation of treatment in patients likely to respond but also allow modification of the treatment regimen in those patients unlikely to respond to the standard duration (48 weeks) of peginterferon alfa-2a monotherapy. Recent analyses have shown that serologic (e.g., HBsAg) and virologic (e.g., HBV DNA) markers either before or during treatment with pegylated interferon may help identify those patients most likely to respond post-treatment [14]. In a retrospective study, it was also shown that low pretreatment HBsAg levels and their decline during treatment with conventional interferon alfa predicted subsequent HBsAg clearance [15].

The current analysis describes the response rate at 5 years post-treatment in HBeAg-negative patients treated with peginterferon alfa-2a during the Phase 3 study—this is the final planned efficacy analysis from this pivotal trial. In addition, it investigates the potential of HBsAg quantification during treatment to predict long-term response.

A total of 230 peginterferon alfa-2a ± lamivudine-treated patients were included in the long-term follow-up study. Of these, 120 (52%) had HBsAg values available at all time-points (i.e., baseline and weeks 12, 24, and 48 of therapy and 6 months post-treatment). Of the patients with HBsAg values available at all time-points, the majority (78%) was from the Asia–Pacific region. Baseline characteristics were similar between the two populations (Table 1).

The pivotal trial of peginterferon alfa-2a in HBeAg-negative CHB has provided a wealth of information about the long-term post-treatment effects of a finite course of treatment. A previous analysis of this study demonstrated that rates of HBsAg clearance increased after completion of treatment, with 9% of patients treated with peginterferon alfa-2a ± lamivudine achieving HBsAg clearance 3 years after treatment [7]. In the protocol-defined analysis described, rates of HBsAg clearance were shown to further increase during longer-term follow-up, with 12% of patients achieving HBsAg clearance at 5 years post-treatment. HBsAg clearance is the ultimate treatment goal of patients with HBeAg-negative CHB as it is associated with improved outcomes [3‐5]. Although HBsAg clearance has been observed during treatment of HBeAg-positive CHB with some of the newer nucleos(t)ide analogs [18, 19], the rates of HBsAg clearance in HBeAg-negative patients during nucleos(t)ide analog therapy are negligible [19]. In contrast, the current study clearly demonstrates that rates of HBsAg clearance after peginterferon alfa-2a therapy are substantial and durable. Patients achieving HBV DNA suppression at 1 year post-treatment achieved high rates of HBsAg clearance at 5 years post-treatment. As 28% of patients with sustained immune control (HBV DNA ≤2,000 IU/mL at 1 year post-treatment) achieved HBsAg clearance at 5 years post-treatment, this endpoint appears to be a valuable early indicator of long-term response.

The analyses conducted in patients enrolled in this international, multicenter, randomized trial have considerably increased the knowledge of the potential value of peginterferon alfa-2a therapy in patients with HBeAg-negative CHB. As a result of the heterogeneous population enrolled, this study—the largest and the most extensive of peginterferon alfa-2a in HBeAg-negative CHB to date—closely reflects the global clinical situation. The long-term benefits achieved following a finite course of peginterferon alfa-2a in such a population are clearly encouraging. Early identification of patients who could benefit from this treatment approach would be valuable, as it would allow clinicians to motivate patients likely to achieve a long-term response to complete therapy while also identifying those patients for whom an alternative treatment regimen may be necessary.

Recently, two groups showed that HBsAg levels vary considerably during the natural history of CHB and reflect the disease phase [20, 21], and previous results demonstrated that on-treatment HBsAg kinetics and HBsAg levels at the end of treatment are associated with sustained response to peginterferon alfa-2a [14, 17]. The observation that lower levels of HBsAg are associated with greater immune control has resulted in considerable interest in HBsAg kinetics during peginterferon alfa-2a therapy.

A difference in HBsAg decline between responders and non-responders, as occurred in this study, was described initially by Brunetto et al. [17]. In the current analysis, differences in HBsAg decline patterns were also observed in patients achieving a long-term response to treatment compared with those relapsing after experiencing an on-treatment response. This confirms data from Moucari et al. [14] that showed the association between HBsAg decline and short-term post-treatment response. As variations in HBsAg kinetics between responders, non-responders, and relapsers to peginterferon alfa-2a therapy have been established in both short-term and long-term follow-up, research has focused on whether clinicians can use the knowledge of HBsAg kinetics to make disease-management decisions.

In the current study, receiver operating characteristic analysis identified a decline in HBsAg levels at weeks 12 and 24, which was associated with high rates of post-treatment response. Patients with a ≥10% log₁₀ IU/mL decline in HBsAg from baseline achieved significantly higher rates of HBV DNA ≤2,000 IU/mL and HBsAg clearance up to 5 years post-treatment than patients not achieving this level of decline. It is worth noting that 40–45% of patients with this level of on-treatment HBsAg decline and sustained immune control at 1 year post-treatment achieved HBsAg clearance at 5 years post-treatment. Although PPVs for long-term sustained immune control were slightly lower at week 24 than at week 12, the high rates of HBsAg clearance achieved by patients with a ≥10% decline during treatment and sustained immune control 1 year post-treatment suggest that on-treatment quantification at either time-point provides clinically important information.

The target was to achieve NPV ≥95%, as this would identify patients unlikely to achieve a sustained response and allow physicians to consider stopping treatment. However, as the NPVs generated by this analysis did not reach the target level, using this as a stopping rule would mean that 13–16% of potential responders would have their treatment stopped prematurely. Some smaller studies have identified on-treatment HBsAg cut-off levels that generate higher NPVs. For example, Moucari et al. [14] showed that an HBsAg decline of 0.5 log₁₀ IU/mL (68% decline from baseline) at week 12 generated an NPV of 90% and an HBsAg decline of 1.0 log₁₀ IU/mL (90% decline from baseline) at week 24 gave an NPV of 97%. However, these levels of decline did not produce similar results in the current analysis (data not shown). The lack of consistent findings may be explained by differences in the study design, response parameters, and study populations between the two studies. For example, Marcellin et al. [7] defined post-treatment response as undetectable HBV DNA at 6 months post-treatment, rather than the HBV DNA <2,000 IU/mL and HBsAg clearance at 1 year and 5 years post-treatment employed in the current analysis. Given that relapse is known to occur between 6 months and 1 year post-treatment [7], 1 year may be a more appropriate time-point for assessment of sustained response. In addition, the differences in cut-off levels identified in these two populations may be explained, at least in part, by HBV genotype, which is known to affect post-treatment response [14, 17, 22, 23]. In the current analysis, the patients were infected predominantly with HBV genotype C (46%) and D (22%) with very few infected with HBV genotype A (10%); while in the Moucari et al. [14] analysis, 27% of patients were infected with HBV genotype A. The influence of HBV genotype on HBsAg kinetics in samples from patients who were also included in the current analysis was described initially by Brunetto et al. [17] who showed that patients infected with all the major genotypes achieve some level of HBsAg decline, but this was most pronounced in patients infected with genotypes A and B. Genotype is, therefore, likely to be a major influencing factor on decline of HBsAg levels in HBeAg-negative subjects. Further analyses are needed to elucidate the influence of genotype on HBsAg kinetics and to determine how knowledge of infecting genotype combined with HBsAg quantification can be used to predict response to peginterferon alfa-2a.

While most investigations have examined on-treatment prediction of response, there is also potential value in identifying at baseline those patients likely to respond to peginterferon alfa-2a. In the current analysis, patients with HBsAg ≤5,000 IU/mL at baseline achieved the highest rates of response post-treatment, but the PPVs (approximately 30%) and NPVs (approximately 80%) calculated were lower than those relating to on-treatment HBsAg quantification. Consequently, on-treatment HBsAg quantification appears to be a more appropriate predictor than quantification of HBsAg at baseline.

Unlike on-treatment HBsAg quantification, measurement of HBV DNA levels during treatment does not differentiate between treatment responders and relapsers [14]. This observation was based on response at 6 months post-treatment but was also seen in the current analysis, where a significant difference in HBV DNA decline between patients with a response at 5 years post-treatment and relapsers could not be demonstrated. In addition, there were differences in HBV DNA decline between peginterferon alfa-2a-treated patients and patients receiving combination therapy with lamivudine. Wherever there was a difference in HBV DNA decline between responders/relapsers and non-responders in monotherapy-treated patients, this was not present in the combination therapy group. Consequently, HBV DNA quantification may not be as valuable for identifying long-term responders to therapy as HBsAg quantification.

A recent analysis of another study of peginterferon alfa-2a in HBeAg-negative patients investigated the potential of combining HBsAg and HBV DNA response during treatment to improve NPVs [24]. The NPV of 100% was reported in patients who did not achieve an HBsAg decline or an HBV DNA decline >2 log₁₀ copies/mL at week 12 of treatment. The importance of infecting genotype on HBsAg kinetics was discussed earlier, and as most patients in this analysis were infected with HBV genotype D, further analysis is required to determine whether this rule can be used in patients infected with other genotypes.

It is interesting to speculate how HBsAg quantification could be used in clinical practice to help individualize peginterferon alfa-2a therapy. Preliminary data from a study of extended peginterferon alfa-2a therapy in HBeAg-negative patients have shown that extension therapy improves sustained response rates as a result of a reduction in relapse [25]. An important future consideration will be whether patients likely to benefit from extended therapy can be identified early during the initial phase of treatment. Brunetto et al. [26] have studied in-depth HBsAg kinetics in the HBeAg-negative patients included in the Phase 3 peginterferon alfa-2a study and showed that response rate is linked to decline pattern. Although patients with a continuous HBsAg decline from baseline (≥10% decline from baseline to week 24 and ≥10% from weeks 24 to 48) achieved the highest rates of response, patients with a late decline (≥10% decline from baseline after week 24) also achieved high rates of response when compared with patients with a <10% HBsAg decline during the entire 48-week treatment period. It is possible that patients with a late HBsAg decline will benefit from an extended period of peginterferon alfa-2a therapy; however, this needs to be studied in prospective clinical trials, which also consider in more detail the role of infecting genotype.

The current analysis has limitations. Only a proportion of patients included in the initial or follow-up studies had HBsAg levels determined during treatment and 6 months post-treatment. In addition, only patients with HBsAg data available at all on-treatment and post-treatment time-points were included and, consequently, there is the potential for selection bias. However, baseline characteristics and response rates in the 230 patients in the long-term analysis were similar to those achieved by the 120 patients in the current analysis, and the statistical methods used were conservative as missing samples were taken as non-responders. Wherever LOCF methodology was used, a secondary parameter was included (HBV DNA <71 IU/mL) to reduce the chance of false-positive data.

In conclusion, a finite course of peginterferon alfa-2a resulted in increasing rates of HBsAg clearance up to 5 years post-treatment in HBeAg-negative patients. Sustained immune control at 1 year post-treatment was an early indicator of subsequent HBsAg clearance. Analysis of data from this large, long-term study has shown that HBsAg quantification may be an appropriate on-treatment tool for monitoring response to peginterferon alfa-2a in HBeAg-negative patients, thereby confirming observations in small-scale studies. Further prospective studies are required before clear clinical guidance on use of HBsAg monitoring can be provided for physicians. In the future, increased understanding of the kinetics of HBsAg decline in HBeAg-negative patients treated with peginterferon alfa-2a may help physicians make individualized treatment decisions that should ultimately increase the rate of response in patients with CHB.