4 Evidence and interpretation

The Appraisal Committee (appendix A) considered evidence from a number of sources (appendix B).

4.1 Clinical effectiveness

Peginterferon alfa-2a

4.1.1 The Assessment Group found three randomised controlled trials investigating peginterferon alfa-2a in chronic hepatitis B. Two trials were in people with HBeAg-positive chronic hepatitis B, and the third was in people with HBeAg-negative chronic hepatitis B.

4.1.2 In HBeAg-positive chronic hepatitis B, the most useful measure of response is HBeAg loss or seroconversion because this signals transition to the inactive carrier state, which is associated with a relatively benign outcome. The rate of HBeAg seroconversion was reported in both studies of peginterferon alfa-2a in HBeAg-positive chronic hepatitis B.

4.1.3 One trial was an open-label study in which 194 participants were randomised to one of three doses of peginterferon alfa-2a or interferon alfa-2a. Treatment was for 24 weeks, and was followed by 24-weeks' treatment-free follow-up. The rate of HBeAg seroconversion was higher in the patients treated with peginterferon alfa-2a (32% for all three doses combined) than in patients treated with interferon alfa-2a (25%), but the difference was not statistically significant. ALT levels were also more likely to return to the normal range in the people treated with peginterferon alfa-2a (36% for all three doses combined compared with 25% for interferon alfa-2a). This difference was also not statistically significant. The only statistically significant difference between treatment and control in this trial was for the rate of 'combined response', defined as HBeAg loss, HBV DNA normalisation and ALT normalisation (24% for all three treatment doses combined compared with 12% for control, p = 0.036).

4.1.4 The second trial in patients with HBeAg-positive chronic hepatitis B was a comparison of peginterferon alfa-2a with lamivudine, in which 814 people were randomised to peginterferon alfa-2a plus placebo, peginterferon alfa-2a plus lamivudine, or lamivudine alone. The study was partially double blind in that those who were receiving peginterferon alfa-2a were blinded as to whether they were receiving lamivudine. Treatment was for 48 weeks, and was followed by 24-weeks' treatment-free follow-up. The rate of HBeAg seroconversion at week 72 was significantly higher in the patients treated with peginterferon alfa-2a plus placebo (32%) than in those treated with lamivudine alone (19%), p < 0.001. Addition of lamivudine to peginterferon alfa-2a did not improve the HBeAg seroconversion rate over peginterferon alfa-2a alone: the rate in the group that received both lamivudine and peginterferon alfa-2a was 27%. Normalisation of ALT occurred in a higher proportion of the group taking peginterferon alfa-2a than of the group taking lamivudine alone (41% versus 28%, p = 0.002). Again, adding lamivudine to peginterferon alfa-2a did not improve the response rate for this endpoint (ALT normalisation 39%).

4.1.5 In HBeAg-negative chronic hepatitis B, a major objective of treatment is suppressing viral replication and preventing progressive liver disease. In clinical trials these outcomes have usually been expressed in terms of decrease in the levels of HBV DNA found in the serum and ALT normalisation.

4.1.6 The only study of peginterferon alfa-2a in HBeAg-negative chronic hepatitis B was a three-way comparison of peginterferon alfa-2a plus placebo, peginterferon alfa-2a plus lamivudine, and lamivudine alone (n = 537). Treatment was for 48 weeks and was followed by 24-weeks' treatment-free follow up (72 weeks in total). The study had two predetermined primary measures of efficacy: the normalisation of ALT levels and the suppression of HBV DNA levels to less than 20,000 copies/ml. At week 72, the percentage with normalised ALT was significantly higher in the groups treated with peginterferon alfa-2a than in the lamivudine group (59% with peginterferon alfa-2a monotherapy and 60% for peginterferon alfa-2a plus lamivudine, versus 44% for lamivudine; p = 0.004 and p = 0.003 respectively). For the outcome of virological response (HBV DNA < 20,000 copies/ml) at week 72, the percentage of patients who had a response was significantly higher in the groups treated with peginterferon alfa-2a than in the lamivudine group (43% for peginterferon alfa-2a monotherapy and 44% for peginterferon alfa-2a plus lamivudine, versus 29% for lamivudine; p = 0.007 and p = 0.003 respectively).

4.1.7 Treatment-resistance mutations were detected in 32 people (18%) in the lamivudine group and one person (< 1%) in the peginterferon alfa-2a plus lamivudine group (p < 0.001).

Adefovir dipivoxil

4.1.8 The assessment report included five randomised controlled trials of adefovir dipivoxil in chronic hepatitis B. Four of the studies were conducted in people with HBeAg-positive chronic hepatitis B and one was in people with HBeAg-negative chronic hepatitis B.

4.1.9 The largest study (n = 515) was a comparison of two doses of adefovir dipivoxil (10 mg and 30 mg) with placebo in people with HBeAg-positive chronic hepatitis B and compensated liver disease. The primary endpoint was histological improvement (defined in terms of a reduction in Knodell necroinflammatory score). Histological improvement was seen in 59% of patients in the adefovir dipivoxil 30 mg group, 53% of patients in the adefovir dipivoxil 10 mg group and 25% of patients in the placebo group (p < 0.001 for both comparisons with placebo). The HBeAg seroconversion rates were 14% for the adefovir dipivoxil 30 mg group, 12% for the adefovir dipivoxil 10 mg group and 6%, for the placebo group. The rates of ALT normalisation were 55% for the 30 mg, 48% for the 10 mg and 16% for the placebo group.

4.1.10 Another study (n = 59) investigated adefovir dipivoxil in people with HBeAg-positive chronic hepatitis B, genotypic evidence of lamivudine-resistance, raised ALT and a serum HBV DNA level of at least 106 copies/ml despite ongoing treatment with lamivudine. The patients were randomised to adefovir dipivoxil 10 mg, lamivudine 100 mg, or the addition of adefovir dipivoxil to ongoing lamivudine treatment. The primary endpoint was time-weighted change from baseline in serum HBV DNA level. Reductions in this endpoint were seen in all recipients of adefovir dipivoxil. HBV DNA levels were 'undetectable' (< 1000 copies/ml) in 26% of patients receiving adefovir dipivoxil plus placebo, in 35% of patients receiving adefovir dipivoxil plus lamivudine, and in none of the patients receiving lamivudine plus placebo (p = 0.005). ALT was normalised in 53% of the adefovir dipivoxil plus lamivudine group and 47% of adefovir dipivoxil group, compared with 5% (1/19) of the lamivudine group (p = 0.001 and p = 0.004, respectively)

4.1.11 The remaining two studies in HBeAg-positive chronic hepatitis B investigated the combination of adefovir dipivoxil with lamivudine compared with lamivudine alone. One was in treatment-naive patients (n = 112) and the other was in patients with lamivudine resistance (n = 94). In the study in treatment-naive patients (which was still ongoing at the time of this appraisal) there was no advantage in adding adefovir dipivoxil to lamivudine at 1 year in terms of virological, serological or biochemical outcome. However, there was a higher incidence of lamivudine-resistance mutations and viral breakthrough in the group that received lamivudine alone. In the other study of patients with lamivudine resistance, adefovir dipivoxil plus lamivudine was more effective than lamivudine alone in terms of both virological and ALT responses. HBV DNA level fell to 105 copies/ml or less in 85% of those taking adefovir dipivoxil plus lamivudine, compared with 11% of patients taking lamivudine alone (p < 0.001). ALT was normalised in 37% of those taking the combined treatment, compared with 9% of those taking lamivudine (p = 0.003).

4.1.12 One study compared adefovir dipivoxil with placebo in people with HBeAg-negative chronic hepatitis B. This was a double-blind study in which 185 people were randomised in a 2:1 ratio to adefovir dipivoxil or placebo for 48 weeks. The primary endpoint was histological improvement defined in terms of a reduction in Knodell necroinflammatory score with no worsening of fibrosis. A total of 167 patients (91%) had assessable pre-treatment and post-treatment liver biopsy specimens. Significantly more patients had histological improvement in the adefovir dipivoxil group than in the placebo group (64% versus 33%, p < 0.001). At week 48, 51% of the adefovir dipivoxil group had undetectable HBV DNA levels (< 400 copies/ml), compared with no one in the placebo group (p < 0.001), and 72% had normalised ALT levels compared with 29% in the placebo group (p < 0.001).

4.1.13 Long-term follow-up data have been published recently. After week 48, 123 people who had been assigned to adefovir dipivoxil in the initial study were randomised to continue adefovir dipivoxil at a dose of 10 mg daily or to switch to placebo. Of the 61 patients who had initially been randomised to placebo, 60 switched to treatment with adefovir dipivoxil 10 mg daily. At week 96, HBV DNA was undetectable in 71% of the group continuing with adefovir dipivoxil, compared with 8% of the placebo group. In the group who received adefovir dipivoxil for 48 weeks having previously received placebo for 48 weeks, HBV DNA was undetectable in 76% of patients. The cumulative incidence of resistance to adefovir dipivoxil among all patients was 3% at 96 weeks and 5.9% at 144 weeks. This trial will continue until all patients have completed 5 years' follow-up.

Evidence from clinical experts

4.1.14 The Committee heard from the clinical experts that the decision to treat chronic hepatitis B was determined by the degree of fibrosis and/or necroinflammation on liver biopsy, combined with evidence of active viral replication (HBV DNA levels) and the persistent elevation of serum ALT. Treatment was not necessarily indicated for people with minimal elevation of ALT (1.5–2.0 times the upper limit of normal) and low necroinflammatory scores on liver biopsy. However, these people should be carefully monitored because the disease can change rapidly from a quiescent to an active state.

4.1.15 The experts expressed concern about the development of viruses with mutations rendering them resistant to antiviral drugs. Lamivudine resistance develops in a high proportion of patients on monotherapy, and could limit the options for future treatment through cross-resistance to related drugs. The experts noted that a strategy of treating chronic hepatitis B with lamivudine followed by adefovir dipivoxil for those in whom lamivudine-resistance developed reflected current practice. However, the experts noted that there was a subgroup of people with highly replicative disease in whom resistance could develop rapidly; in these people, a strategy of using adefovir dipivoxil in combination with lamivudine might be appropriate.

4.1.16 In principle, the use of combination therapies should minimise the risk of developing resistant variants, although long-term follow-up data from studies are lacking. The clinical experts stressed the need for further research on the long-term effectiveness of combination regimens in preventing resistance.

4.2 Cost effectiveness

4.2.1 The Committee considered evidence from four economic models: one by the Assessment Group, one by each of the two manufacturers involved, and one published analysis by Kanwal and colleagues (which was published after the Assessment Report's deadline for inclusion). The models have similar structures and parameters, and their results are in broad agreement.

Assessment Group model

4.2.2 The Assessment Group model presents analyses for people with HBeAg-positive disease and HBeAg-negative disease separately, and also as a single group, with proportions of people with HBeAg-positive disease and HBeAg-negative disease determined from the parameters of the model.

4.2.3 Results are presented for peginterferon alfa-2a (48 weeks' treatment) compared with interferon alfa-2a (24 weeks for HBeAg-positive disease or 48 weeks for HBeAg-negative disease), and for adefovir dipivoxil compared with lamivudine. The incremental cost-effectiveness ratios (ICERs) for these two comparisons in HBeAg-positive and HBeAg-negative groups combined were £6100 and £16,500 per quality-adjusted life year (QALY) respectively. However, this analysis assumes that, if the first therapy does not produce a sustained response, people receive no further treatment other than best supportive care, whereas in practice people may go on to receive treatment with an alternative agent. The Assessment Group therefore produced an analysis that considered more clinically relevant scenarios in which people could receive a sequence of drug treatments as necessary.

4.2.4 It is assumed that all patients are first given a course of treatment with either interferon alfa-2a (for 24 or 48 weeks, depending on HBeAg status) or peginterferon alfa-2a for 48 weeks. If HBeAg or HBsAg seroconversion occurs on one of the interferons, the patients are assumed to take no further antiviral medication; if not, they proceed to one of the following sequences of treatment, which form the basis of the comparison:

  • best supportive care, that is, no further antiviral medication

  • lamivudine, then best supportive care on seroconversion or development of resistance

  • adefovir dipivoxil, then best supportive care on seroconversion or development of resistance

  • lamivudine, then adefovir dipivoxil on development of resistance, then best supportive care on seroconversion or development of resistance.

4.2.5 The Assessment Group analysis also reports ICERs for the use of lamivudine then best supportive care without first using an interferon, and of adefovir dipivoxil then best supportive care without first using an interferon. It did not report on sequences of both lamivudine and adefovir dipivoxil that were not preceded by an interferon. However, because of the underlying assumptions used in the model, these ICERs can be inferred from the equivalent comparisons for sequences including peginterferon alfa-2a or interferon alfa-2a.

4.2.6 Comparing these sets of sequences generated a large number of ICER estimates. However, some of the strategies can be excluded from consideration using the notion of dominance or extended dominance. This means that when strategies A, B and C (in order of increasing cost) are compared, if the ICER for B compared with A is higher than that for C compared with A, and if the benefits of B are less than those of C, then B is excluded.

4.2.7 For HBeAg-positive disease, the treatment sequences that are not excluded by extended dominance are as follows:

  • lamivudine then best supportive care (estimated ICER £3500 per QALY compared with best supportive care)

  • interferon alfa-2a then lamivudine then adefovir dipivoxil then best supportive care (estimated ICER £9900 per QALY compared with lamivudine then best supportive care)

  • peginterferon alfa-2a then lamivudine then adefovir dipivoxil then best supportive care (estimated ICER £18,800 per QALY compared with interferon alfa-2a then lamivudine then adefovir dipivoxil then best supportive care)

  • peginterferon alfa-2a then adefovir dipivoxil then lamivudine then best supportive care (estimated ICER £57,000 per QALY compared with peginterferon alfa-2a then lamivudine then adefovir dipivoxil then best supportive care). Note that this treatment sequence did not appear in the Assessment Report, but was prepared by the Assessment Group for the Appraisal Committee before its first meeting.

4.2.8 For HBeAg-negative disease, the treatment sequences that are not excluded by extended dominance are as follows:

  • peginterferon alfa-2a then best supportive care (estimated ICER £3000 per QALY compared with best supportive care)

  • peginterferon alfa-2a then lamivudine then best supportive care (estimated ICER £4900 per QALY compared with peginterferon alfa-2a then best supportive care)

  • peginterferon alfa-2a then lamivudine then adefovir dipivoxil then best supportive care (estimated ICER £18,000 per QALY compared with peginterferon alfa-2a then lamivudine then best supportive care).

Manufacturer's model (Roche)

4.2.9 The model submitted by Roche reports a number of estimated cost effectiveness results, all involving either interferon alfa-2a or peginterferon alfa-2a, for HBeAg-positive and HBeAg-negative disease separately. Nothing is said about later choices for people for whom the interferon has not been clinically effective. The estimated ICERs for people with HBeAg-positive disease are as follows:

  • peginterferon alfa-2a for 24 weeks compared with interferon alfa-2a for 24 weeks (ICER £2700 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with interferon alfa-2a for 24 weeks (ICER £13,900 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with lamivudine for 48 weeks (ICER £5300 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with lamivudine for 208 weeks (ICER £5900 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with adefovir dipivoxil for 48 weeks (ICER £1400 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with adefovir dipivoxil for 208 weeks (dominant, that is, greater benefit at lower cost than adefovir dipivoxil)

  • peginterferon alfa-2a for 48 weeks compared with no treatment (ICER £2800 per QALY).

4.2.10 The ICERs for people with HBeAg-negative disease are as follows:

  • peginterferon alfa-2a for 48 weeks compared with lamivudine for 48 weeks (ICER £3200 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with lamivudine for 208 weeks (ICER £1900 per QALY)

  • peginterferon alfa-2a for 48 weeks compared with no treatment (ICER £1500 per QALY).

Manufacturer's model (Gilead)

4.2.11 The model submitted by Gilead presents estimated cost-effectiveness ratios for a number of single treatments and treatment sequences. However, the cost effectiveness of adefovir dipivoxil compared with best supportive care is not reported. The interferons are not considered as treatment options, and the analyses are for a combined population of people with HBeAg-positive and HBeAg-negative disease. The estimated ICERs are as follows:

  • lamivudine then adefovir dipivoxil (when lamivudine resistance emerges) compared with best supportive care – ICER £6700 per QALY

  • lamivudine then adefovir dipivoxil (when lamivudine resistance emerges) compared with lamivudine then best supportive care (when lamivudine resistance emerges) – ICER £9200 per QALY

  • adefovir dipivoxil then lamivudine (when adefovir dipivoxil resistance emerges) compared with best supportive care – ICER £8200 per QALY

  • adefovir dipivoxil then lamivudine (when adefovir dipivoxil resistance emerges) compared with lamivudine then best supportive care (when lamivudine resistance emerges) – ICER £11,400 per QALY

  • adefovir dipivoxil then lamivudine (when adefovir dipivoxil resistance emerges) compared with lamivudine then adefovir dipivoxil (when lamivudine resistance emerges) – ICER £29,400 per QALY.

Published model

4.2.12 The published model (Kanwal and colleagues) evaluated a 'do nothing' strategy (equivalent to best supportive care), interferon alfa, lamivudine, adefovir dipivoxil, and lamivudine then adefovir dipivoxil. For HBeAg-positive disease, the estimated ICERs for the non-dominated strategies are as follows:

  • interferon alfa compared with best supportive care – ICER $2300 (£1300 at an exchange rate of $1.75 per £1) per QALY

  • lamivudine then adefovir dipivoxil, compared with interferon alfa – ICER $16,600 (£9500) per QALY

  • adefovir dipivoxil, compared with lamivudine then adefovir dipivoxil) – ICER $91,000 (£52,000) per QALY.

4.2.13 In the analysis for HBeAg-negative disease, lamivudine then adefovir dipivoxil (as salvage therapy after resistance to lamivudine develops) is both cheaper and more effective than all other treatment options.

Overall results of the models

4.2.14 The models show that interferon alfa or peginterferon alfa-2a therapies followed by lamivudine then adefovir dipivoxil, where necessary, appear to be cost effective relative to alternative strategies. In most of the analyses, strategies in which adefovir dipivoxil is used before lamivudine, or without lamivudine, in the sequence are dominated by the alternative strategies. The exceptions are Gilead's estimated ICER of £29,400 per QALY for adefovir dipivoxil then lamivudine, compared with lamivudine then adefovir dipivoxil, and the Assessment Group's estimated ICER of £57,000 per QALY (for HBeAg-positive patients) for peginterferon alfa-2a then adefovir dipivoxil then lamivudine, compared with peginterferon alfa-2a then lamivudine then adefovir dipivoxil.

4.3 Consideration of the evidence

4.3.1 The Committee reviewed the data available on the clinical and cost effectiveness of adefovir dipivoxil and peginterferon alfa-2a for the treatment of chronic hepatitis B, having considered evidence on the nature of the condition and the value placed on the benefits of adefovir dipivoxil and peginterferon alfa-2a by people with chronic hepatitis B, those who represent them, and clinical experts. It was also mindful of the need to take account of the effective use of NHS resources.

4.3.2 The Committee considered that peginterferon alfa-2a was clinically and cost effective compared with interferon alfa therapy for the treatment of chronic hepatitis B on the basis of the cost effectiveness modelling. However, given that the difference in the rates of HBeAg seroconversion and ALT normalisation (HBeAg-positive chronic hepatitis B) compared with standard interferon alfa-2a were not statistically significant, the Committee was not persuaded that the potential advantages of peginterferon alfa-2a over standard interferon had been conclusively proven. The Committee concluded, therefore, that peginterferon alfa-2ashould be considered as an alternative to interferon alfa for the management of chronic hepatitis B during the initial phase of treatment.

4.3.3 The experts stated that the effects of initial treatment with peginterferon alfa-2a in achieving a good response were different for HBeAg-positive and HBeAg-negative disease. The Committee was persuaded, however, that the response to peginterferon alfa-2a as measured by HBeAg seroconversion (in the case of HBeAg-positive disease) or an adequate reduction in viral load (in the case of HBeAg-negative disease) was clinically important in both cases. Thus the Committee concluded, on this basis, that peginterferon alfa-2a should be recommended as an option in first-line therapy for both HBeAg-positive and HBeAg-negative chronic hepatitis B.

4.3.4 The Committee next considered the use of the oral antiviral agents (specifically adefovir dipivoxil) for the long-term treatment of both HBeAg-positive and HBeAg-negative chronic hepatitis B. The Committee considered the cost-effectiveness analysis presented by the assessment team on the use of adefovir dipivoxil compared with lamivudine in the long-term treatment of chronic hepatitis B. In addition the Committee discussed the cost-effectiveness analysis of the various treatment sequences that could be used over the entire disease process. It was apparent that adefovir dipivoxil was cost effective when compared directly with lamivudine for long-term therapy, if it was assumed that a switch could not be made to the alternative treatment if resistance developed. However, when considering the use of adefovir dipivoxil in the various treatment sequences, the Committee heard that the most cost-effective option was for it to be used following the emergence of viral resistance to lamivudine.

4.3.5 The Committee heard the concerns of the clinical experts about the potential impact of the development of resistant viral mutations on future treatment options, based on the current experience of lamivudine use. The experts also stated that adefovir dipivoxil was less likely than lamivudine to result in viral resistance over the short term. They also expressed the view that viral resistance may be encouraged by the use of single agents and discussed with the Committee the possibility of recommending combination therapies for long-term treatment.

4.3.6 The Committee was persuaded that it was likely that drug-resistance might be attenuated by using antiviral drugs in combination. The Committee felt unable to recommend routine use of combination therapies in the absence of trial evidence on the effect of combination therapy on drug resistance and its cost effectiveness. However, the Committee considered that it was not appropriate under these circumstances to recommend the use of adefovir dipivoxil only after resistance had already developed to another antiviral agent. It was persuaded by the expert testimony that it was possible to identify groups of people for whom lamivudine resistance is more likely to occur rapidly (for example, those with highly replicative disease), and those for whom development of lamivudine resistance is likely to have an adverse outcome (for example, those in whom a flare up of the infection may precipitate decompensated liver disease). Consequently it was sympathetic to the experts' view that it was appropriate to use adefovir dipivoxil for patients in whom the development of lamivudine-resistance would be considered particularly hazardous. The Committee was also sympathetic to the experts' view that under these circumstances the use of adefovir dipivoxil alone or in combination with lamivudine might be appropriate.

4.3.7 The Committee concluded that it was appropriate to recommend adefovir dipivoxil as an option for the treatment of chronic hepatitis B for patients in whom prolonged oral antiviral treatment is required. It was also persuaded that this should be only after the use of an interferon as initial treatment unless this was contraindicated. The Committee was mindful that in the Assessment Group's economic model, the sequence of therapy with adefovir dipivoxil before lamivudine was not cost effective compared with the alternative of lamivudine followed by adefovir dipivoxil. The Committee was, however, persuaded that the decision to use adefovir dipivoxil (alone or in combination with lamivudine) would need to be made on a case-by-case basis. This should take into account various factors including HBeAg status, stage of disease process (for example the presence of compensated or decompensated cirrhosis) and the presence of, or likelihood of the emergence of, virus resistance.

4.3.8 The Committee accepted consultee advice that drug treatment should be initiated only by a healthcare professional with expertise in the management of viral hepatitis, but that the task of continuing and monitoring therapy could be undertaken by general practitioners under shared-care arrangements.