There are now two licensed therapies for chronic hepatitis B: interferon α and lamivudine. Interferon α was first shown to have activity against hepatitis B in 1976,1 but was not formally approved for use in chronic hepatitis B until 1992. The currently recommended regimen for interferon is 5 million units (mu) given daily or 10 mu given three times a week by subcutaneous (sc) injection for four to six months. This regimen results in long term beneficial responses in roughly 33% of patients.2 3

Lamivudine was first shown to have activity against hepatitis B virus (HBV) in 19924 and was approved for use in chronic hepatitis B in 1998. The currently recommended regimen for lamivudine is 100 mg given daily by mouth for one year. This regimen results in beneficial responses in 16–20% of patients with typical chronic hepatitis B.5-7

Thus, there are now two choices for therapy. Which should be used? Which should be used first? Should the second be tried if the first fails? What about combining the two? These are simple questions, but they do not have simple answers.

In this issue (see page 562), Schalm and collaborators report results of a large, multinational, randomised, double blind, placebo controlled trial of lamivudine, interferon α and the combination of both in 230 patients with typical chronic hepatitis B. This trial was one of the largest studies ever conducted in hepatitis B and was large enough to have answered the questions posed above, at least in part. In the study, 18% of patients receiving lamivudine alone, 19% receiving interferon alone and 29% receiving the combination had a beneficial response, the higher response rate with combination therapy being of borderline statistical significance. Side effects from the combination were no greater than those with interferon alone. These results suggest that the combination of lamivudine and interferon α is the optimal initial therapy of this disease.

Before such recommendations are embraced, however, a closer look at this study is needed. Unfortunately, the study suffered in several regards: in design, conduct and analysis.

The design of this three arm study was complex. The first two arms were standard. One group received the recommended regimen of lamivudine and the second, the recommended regimen of interferon. The combination group, however, received an unusual and irregular regimen of both: lamivudine was given for six months only and interferon was started late, two months after initiating lamivudine therapy. The “lead in” phase of lamivudine before combination therapy was done because of previous studies suggesting that interferon is more effective in patients with lower levels of viral DNA.2 3The difficulty is that interferon is also more effective in patients with higher serum aminotransferase activities, and lamivudine therapy often lowers serum enzyme activities. Furthermore, the complex regimen made the blinding of placebo treatment difficult.

Another major problem with the design was the differences in the timing of the end point evaluation in relation to therapy. Thus, the end points were measured in patients receiving lamivudine while they were still on treatment, but in those receiving interferon or the combination at a point six months after stopping therapy. This variability complicated the comparison of end points among groups. Thus, at the one year point, HBV DNA levels, aminotransferases and liver histology may well have been affected by the continuation of lamivudine therapy. Furthermore, scant information was given about follow up of treated patients after stopping lamivudine. Of course, it is difficult to compare therapies if the durations are different between groups (and this problem has plagued other studies of therapy of viral hepatitis).8 This discrepancy can only be overcome by adequate follow up, which would be at least 12 months after stopping lamivudine and, therefore, 18 months after stopping interferon with or without lamivudine.

The conduct of the trial was also problematic. The number of protocol violations was high, occurring in 50 of the 230 randomised patients. These violations consisted of dropouts, errors in randomisation, errors in assignment of inclusion and exclusion criteria, use of other antiviral agents during the study, and other problems. The number of protocol violations may have been because of the complexity of the design, lack of acceptance of a placebo control, the diversity of the therapies, and the large number of geographically dispersed centres involved.

The analysis and presentation of results of this trial were also complex and difficult. Results were presented using five different populations of patients enrolled: the total “as treated” population of 230 patients, the “intention to treat” population of 226 patients, the “per protocol” population of 180 patients (minus the violations), the population of 212 patients who reached the 12 month point for analysis, and the population of 165 patients reaching the 15 month point. The shifting denominator of number of patients makes it difficult to state what the response rates were in terms that are understandable clinically. It is also difficult to say which group should have been used. For instance, in the “per protocol” analysis, the response rate to combination therapy was 36%, quite a bit better and supportive of combination therapy than the 22% for interferon alone and 19% for lamivudine alone.

A final critical issue in analysis was the end point used to define benefit. A “response” was defined as seroconversion from HBeAg to anti-HBe (and absence of HBV DNA by hybridisation testing) by the 12 month point. This end point differs from most trials of antiviral therapy of hepatitis B which used the loss of HBeAg (with or without development of anti-HBe) as an end point.2 3Using loss of HBeAg as a definition of response (from table 2 and the population of 214 patients), 29% of lamivudine treated, 36% of interferon treated, and 51% of combination treated patients responded. Loss of HBeAg is a surrogate marker that has been found to be reasonably reliable in predicting a long term remission after interferon therapy of chronic hepatitis B as shown in several long term follow up studies.9 10 Whether loss of HBeAg or seroconversion to anti-HBe is a reliable surrogate marker for a sustained remission after nucleoside analogue therapy is unknown. In this study, nine of 11 patients who lost HBeAg during lamivudine therapy remained HBeAg negative three months later (data from panel A of figure 2 are, however, incompatible with this statement from the text). This number represents a 18% relapse rate within three months of stopping therapy, quite high when one considers that this is a chronic, often life-long disease.

Despite these shortcomings, this trial provides a large amount of important information about therapy of hepatitis B. Additional information would have been helpful, especially with regard to the effect of the lead in phase on HBV DNA and alanine aminotransferase activities and whether responses were sustained long term. Additionally, use of a more sensitive assay for HBV DNA levels (such as PCR) might have helped to resolve differences in the three groups.

What can be recommended from this study? Another study, certainly. But if a clinician plans to treat patients with hepatitis B, a combination approach is reasonable and should probably employ a conventional regimen of both agents, starting both interferon and lamivudine at the same time, continuing interferon for four months and lamivudine for 12 months. In clinical practice, as in clinical research studies, long term follow up is needed of all treated patients to document absence of reactivation and continued resolution of disease activity.

Antiviral therapy of hepatitis B has entered adolescence, the awkward age—restive and resistant to management. The availability of two agents to treat this disease has provided more opportunities, but with it goes more responsibilities and more difficult choices. A safe and secure maturity will probably require a third or fourth agent active against this disease.