Limitations of the Review

We restricted our review to publications in the English language but conducted additional searches in MEDLINE^® for RCTs of eligible antiviral drugs. We identified ten publications in the Chinese language, including one study of adefovir, ¹⁸⁴ two studies of entecavir,^{185, 186} two studies of interferon alfa-2b or pegylated interferon alfa-2b,^{187, 188} and five publications of lamivudine^189–193 We reviewed the abstracts and concluded that language bias, if present, could not change overall conclusions about efficacy of the tested antiviral drugs in adults with CHB. Tenovir has recently been approved by the FDA for treatment of CHB. However, we were unable to find any published data regarding efficacy and safety in these patients.

We did not review the effects of antiviral drugs that have not been approved by the FDA for CHB. Several new medications have been tested in the published and ongoing clinical trials, including emtricitabine, clevudine, pradefovir, valtorcitabine, thymosin alpha1, and anti viral vaccine.¹¹ The drugs did not show significant prevention of liver cancer or decompensation. The studies of emtricitabine reported improvement in laboratory measures of normalized ALT and loss of HBV DNA,¹⁹⁴ HBeAg seroconversion and improved histology,¹⁹⁵ and antiviral mutations.¹⁹⁶ One study addressed the question of which biological markers specific for the disease can predict better response to emtricitabine. The authors concluded that HBV core-specific clusters of differentiation (CD) CD4+T-cells were associated with viral clearance with no changes in HBV-specific CD8+T-cells.¹⁹⁶ Clevudine sustained viral clearance and ALT normalization 6 months off the treatments in HBeAg-positive^{197, 198} but not in HBeAg-negative patients.¹⁹⁹ Two drugs, pradefovir and valtorcitabine, are being tested in phase I and II clinical trials. Tenofavir showed promising results in patients coinfected with HBV and HIV.^{200, 201}

Ongoing clinical trials do not aim to examine clinical outcomes but rather evaluate safety and effectiveness of antiviral drugs in specific patient populations. Effects of entecavir are being tested in Blacks/African Americans and Hispanics with HBV infection (Clinical Trials Database-number NCT00371150) (Appendix E ^* Tables 15 and 16). Sustained HBV DNA loss is expected after combined therapy with entecavir and tenofovir versus adefovir in adults with lamivudine-resistant HBV infection (NCT00605384). Patients experiencing virologic breakthrough after lamivudine therapy are being recruited to the RCTs of telbivudine, adefovir, or their combinations (NCT00376259). Adefovir plus entecavir or adefovir plus lamivudine compared to entecavir alone is being examined on several criteria of resolved hepatitis B in lamivudine-resistant adults with HBV infection (NCT00410202). Peginterferon alfa-2a is being tested in phase IV clinical trial to resolve hepatitis B in HBeAg-positive patients that would experience sustained loss of HBeAg and HBV DNA, HBsAg loss and seroconversion, and ALT normalization 6 months off the treatment (NCT00435825). Ongoing trails will not show the long-term effects of antiviral drugs on all cause and liver related mortality, liver cancer, and decompensation.

Ongoing (Appendix E Table 15) and completed (Appendix E Table 16) RCTs in patients with CHB did not aim to examine clinical outcomes but rather intermediate viral, biochemical, histological measures, and nonspecified safety outcomes. Only one (NCT00096785) RCT was completed less than a year ago; 10 RCTs (Appendix E Table 16) were completed more than year ago without identified publications in PubMed. We could not explore the reasons for nonpublications or pending status of the submitted publications since such information is not available on either the www.clinicaltrials.gov or the FDA web site. Reporting of outcomes during trial registration was not consistent with different levels of details (Appendix E Tables 15 and 16). However, variability in definitions of HBV DNA clearance is substantial, as well as the length of sustained outcomes assessment. Some studies did not provide any information about examined outcomes. Safety outcomes were not defined; few studies intend to evaluate rates of discontinuation of drug administration due to adverse events of laboratory toxicity. One study (NCT00412750) that evaluated the effects of telbivudine and peginterferon alpha-2a mono was terminated for safety issues with no further details about severity and frequency of adverse events. The study was designed to evaluate HBV DNA loss. Safety outcomes were not specified and require larger sample size. We can only assume that the rates of adverse events were unexpected and substantial to stop the study due to safety issues.

Few ongoing RCTs plan to examine the effects of drugs in patient subpopulations; therefore, individualized treatment recommendations based on RCTs would not be possible during the next decade. Few nonrandomized phase IV clinical trials (not shown) aim to investigate the role of baseline viral load on the effects of peginterferon alfa-2a plus ribavirin (NCT00154869) in patients with hepatitis B and C or liver function after telbivudine versus adefovir administration among patients with different viral genotype (NCT00640588). Ongoing observational studies did not intend to examine clinical outcomes or intermediate laboratory measured in patient subpopulations; therefore, upcoming publications would not clarify which subgroups may experience the greatest benefit from the treatments. The fact that several registered RCTs were terminated due to poor recruiting may serve as additional justification for creating a national registry of the patients with CHB.

Gaps in Evidence and Recommendations for Future Research

The greatest gap in knowledge in the management of CHB derives from the lack of large, long-term RCTs assessing the effect of antiviral agents, alone or in combination, on clinical outcomes such as all-cause mortality, liver-related mortality, hepatocellular carcinoma, and hepatic decompensation. Additional valid clinical outcomes could include quality of life and hospitalizations. Cirrhosis is frequently described as a clinical outcome, but in most cases this is determined by liver biopsy performed in the absence of clinical symptoms. While predictive of future clinical events, such as liver-related mortality and all-cause mortality, cirrhosis may be more appropriately defined as a prognostic and potential surrogate measure. There is a moderate level of evidence that therapies can improve combined biochemical and virological outcomes used to define resolved hepatitis. However, randomized trials did not reliably demonstrate sustained HBsAg clearance off therapy. Therefore, there is insufficient evidence that any of these agents can reduce long-term infectivity or that they will improve clinical outcomes. Until randomized trials demonstrate that antiviral drugs improve clinical outcomes or provide sustained reduction in hepatitis B virus transmissibility, the accurate assessment of clinical effectiveness, the validity of putatitve surrogate measures, and decisions on whom to treat remain unknown.

Studies were not designed to detect significant effects of the drugs on clinical outcomes. Only one trial reported significant protective effect on clinical outcomes. Lamivudine reduced hepatocellular carcinoma, but only after post hoc adjusted analyses that excluded five individuals who developed hepatocellular cancer within the first year of the study.¹³² Because the incidence of clinical outcomes is generally low among patients with CHB, randomized trials will require large sample sizes and long duration to have power to accurately assess clinical effects. Alternatively, enrollment of patients at high-risk of disease outcomes (e.g., patients with cirrhosis) would provide an opportunity to more quickly examine the effects of antiviral drugs in this group. Until such studies are completed, a multinational registry combining individual patient data may provide sufficient estimations of drug benefits and harms in patient subgroups.

We recommend that future research focus on clinically important outcomes (mortality, HCC, hepatic decompensation) or sustained criteria of resolved hepatitis B (s and e antigen seroconversion and loss of HBV DNA). More than 75 percent of 3,188 abstracted hypotheses from 92 publications resulted in random differences in outcomes. Available studies examined selected outcomes at the end of the treatments and at different times of followup off the treatments. Therefore, any positive effects could be at least partly due to statistical chance. Additionally, selective reporting of outcomes to emphasize positive is a real possibility.

Studies were not designed to test treatment differences on clinical outcomes and resolved hepatitis in patient subpopulations. Additionally, these studies involved relatively short-term treatment and followup evaluation off treatment. Many were designed to test treatment efficacy related to selected intermediate biochemical or virological measures rather than clinical outcomes. This is of particular concern due to the long natural history of the disease, including long subclinical phase prior to initiation of any treatment and the long followup required prior to development of any clinical events. The reported studies may not reflect current practice that is initiating longer courses (including indefinite length) of treatment. Further research is needed to determine whether current treatment strategies will improve long-term clinical outcomes. Studies should be sufficiently large to assess outcomes in patients with multiple clinical and disease characteristics currently used by clinicians and guideline groups to make treatment decisions (e.g., according to eAg and HBV DNA status).

There have been several very large prospective studies on patients with chronic HBV infection. These studies have shown that various patient characteristics and clinical markers are predictive of important chronic HBV-related outcomes such as cirrhosis, HCC, and death. What remains to be addressed is the extent to which these predictors of disease progression represent clinically useful therapeutic targets or disease surrogates. Observational studies that report longitudinal measurements of these predictors and collect outcome data could better identify whether change in predictor status leads to change in outcomes, instead of the currently more common approach of whether a one-time measurement predicts outcomes. While there was strong evidence that cirrhosis was associated with significantly poorer clinical outcomes, there was very little evidence available that provided information on the predictive ability of other indicators of liver histology. Large studies with baseline histology measurements would help to fill this gap. The vast majority of research on the natural history of chronic hepatitis, even within the United States, is comprised primarily of people with perinatally acquired HBV infections. Therefore the evidence base for patients with HBV infection acquired later in life is much weaker and basically involves extrapolation other populations. Since recent clinical guidelines classify patients into diagnostic groups based on HBeAg status, serum HBV DNA, ALT/AST levels and biopsy results, it is important that future observational studies at a minimum measure these factors and analyze data controlling or stratifying for these variables. Future studies would benefit from creating cohorts of people within existing diagnostic groups: inactive carrier, chronic hepatitis HBeAg-positive, chronic hepatitis HBeAg-negative, and chronic hepatitis with cirrhosis, and presenting key findings separately for these groups.

Additional research needs include:

• Develop valid surrogates and demonstrate the effect of a treatment agent on the surrogate as well as the clinical endpoints.
• Clarify candidate surrogate markers. The change in surrogate due to treatment should predict and explain the change in the outcome.
• Differentiate a surrogate from an outcome. Cirrhosis is frequently described as a clinical outcome. However, many studies included patients with baseline cirrhosis to predict difference in future clinical events, such as liver-related mortality and all-cause mortality. The role of viral mutations and drug resistance during treatment with reverse transcriptase inhibitors as surrogate or endpoint measures had not been defined yet.
• Develop and standardize definitions of surrogate markers. Adopting a uniform scoring system for liver biopsies and deciding on a single definition of what constitutes clinically meaningful change in score, such as ‘fibrosis progression’ (is it a one, two, or four point change in fibrosis or necroinflammatory activity or both). This requires an international effort, particularly since different continents seem to use different scoring systems. Definitions of ALT elevation should be standardized (any abnormal ALT, two times or four times the upper limit of normal). For HBV DNA, we found the greatest heterogeneity in assay used limit of detection and definition of what constituted ‘high’ versus ‘low’ viral load. In the absence of uniform definitions, clinical significance is often lost, as data are grouped and categorized to achieve a statistical significance.
• Develop, standardize, and disseminate the laboratory assay and methods used to quantify surrogate markers of interest. In the case of HBV DNA, we need to have a uniform requirement of real-time PCR with a standard cutoff for upper and lower detection limit.
• Develop standard timing of measurement of intermediary measures to demonstrate a change. For example, should we measure ALT and HBV DNA at baseline and every 6 months on treatment and after treatment for at least 5 years? What magnitude of change between which of these values constitutes a ‘change’ that can be predictive of outcome?
• In the absence of hard clinical endpoints, additional valid clinical outcomes could be considered as endpoints, such as quality of life, cost effectiveness, and hospitalizations. These may be additional benefits favoring treatment, but it needs to be made clear to the patient and treating physician for clinical decisionmaking.

Conclusion

Adults with CHB are at increased risk for poorer health outcomes, though the absolute risk generally is small and requires many years to manifest. Available drugs have not been demonstrated to improve clinical outcomes or resolve hepatitis B. Presence of cirrhosis is the greatest risk factor leading to poor clinical outcomes. Interferons, reverse transcriptase inhibitors, and their combinations provided mid-duration sustained off treatment improvements in selected intermediate outcomes. Baseline patient and disease characteristics may change the natural history of the disease and response in intermediate outcomes. Most drugs are relatively well tolerated with few adverse effects that are generally mild. Reliable surrogate measures to assess treatment effectiveness do not exist. Long-term RCTs are needed to assess long-term effects of antiviral agents on clinical outcomes and among patient subpopulations.

Footnotes

*

Appendixes and evidence tables cited in this report are provided electronically at http://www.ahrq.gov/downloads/pub/evidence/ pdf/hepb/hepb.pdf