Tenofovir disoproxil for the treatment of chronic hepatitis B

The Appraisal Committee considered evidence submitted by the manufacturer of tenofovir disoproxil and a review of this submission by the Evidence Review Group (ERG).

The manufacturer approached the decision problem by comparing tenofovir disoproxil monotherapy with lamivudine, adefovir dipivoxil and entecavir in adults with compensated liver disease and active chronic hepatitis B (that is, evidence of viral replication and active liver inflammation). The manufacturer did not compare tenofovir disoproxil with any of the interferons. The primary outcome measures outlined in the decision problem were virological response (hepatitis B virus [HBV] DNA), histological improvement (inflammation and fibrosis), biochemical response (for example, ALT levels), and hepatitis B surface antigen (HBsAg) and hepatitis B 'e' antigen (HBeAg) seroconversion rate.

The manufacturer's submission presented evidence on the clinical effectiveness of tenofovir disoproxil from two randomised controlled trials (RCTs) that compared tenofovir disoproxil with adefovir dipivoxil. The protocol for both studies specified that the populations would be people who had not previously received nucleotide analogue therapy, although prior experience of the nucleoside analogues lamivudine or emtricitabine was allowed in the study of people with HBeAg-negative chronic hepatitis B. One trial compared tenofovir disoproxil with adefovir dipivoxil in people with HBeAg-positive hepatitis B (176 participants received tenofovir disoproxil and 90 adefovir dipivoxil); the other made the same comparison in people with HBeAg-negative disease (250 participants received tenofovir disoproxil and 125 adefovir dipivoxil).

The results of the two RCTs showed that at 48 weeks tenofovir disoproxil gave a greater proportion of complete responses (that is, histological response and HBV DNA below 400 copies per ml) than adefovir dipivoxil in people with HBeAg-positive and HBeAg-negative disease. The difference was statistically significant (p0.001, in both trials). A similar proportion of people with HBeAg-positive disease had seroconversion or HBeAg loss with tenofovir disoproxil and adefovir dipivoxil, but significantly more people treated with tenofovir disoproxil had hepatitis B surface antigen (HBsAg) loss at 48 weeks (3.2% versus 0.0%, p=0.018). No people with HBeAg-negative disease in either treatment group had experienced HBsAg loss or seroconverted to anti-HBs at 48 weeks.

Both RCTs reported a smaller proportion of people with HBV mutation-conserved site changes (suggestive of a potential for future resistance) with tenofovir disoproxil than with adefovir dipivoxil at 48 weeks. There were no cases of substitution in the HBV polymerase/reverse transcriptase associated with resistance to tenofovir disoproxil in either study. There were no cases of viral resistance.

The incidence of severe, life-threatening or disabling adverse events was similar between treatment groups, with no deaths reported in either study. However, more participants had at least one treatment-related adverse event in the tenofovir disoproxil treatment group in one study (30.7% versus 16.7%, p=0.018); the manufacturer attributed this to a higher incidence of 'mild nausea'. The incidence of arthralgia was higher for the group receiving tenofovir disoproxil in the other study (6.0% versus 0.0%, p=0.003).

The manufacturer pointed out that there were no trials that included all treatment options in any of the patient populations and therefore a series of mixed-treatment comparison meta-analyses were carried out to assess the relative efficacy of adefovir dipivoxil, entecavir, lamivudine, tenofovir disoproxil and placebo in nucleoside-naive and lamivudine-refractory patients. For HBeAg-positive disease, the included outcomes were the probability of achieving HBV DNA suppression (less than 300 copies per ml), and the probability of HBeAg seroconversion over 1 year of treatment. All analyses were conducted using random-effects models unless the between-studies standard deviation was close to zero. For HBeAg-positive nucleoside- and nucleotide-naive participants, the mixed-treatment comparison showed that tenofovir disoproxil had a statistically significantly higher predicted probability of HBV DNA suppression than all comparators. There was no statistically significant difference between the antiviral drugs for the probability of seroconversion.

For HBeAg-negative disease, the manufacturer explained that no meaningful analysis could be undertaken because of the small number of trials identified. The manufacturer undertook an additional analysis combining trials investigating patients with HbeAg-positive disease and those with HbeAg-negative disease. In this additional analysis, the proportion of participants who were HBeAg positive was considered as a covariate. The results for HbeAg-negative participants were similar to those seen in the HBeAg-positive subgroup in terms of the probability of achieving HBV DNA suppression (less than 300 copies per ml). The manufacturer pooled data from the RCTs and the observational studies identified while undertaking the systematic review in order to obtain and compare estimates of resistance for available treatments for HBeAg-positive treatment-naive and lamivudine-refractory patients. The results suggested a low risk of viral resistance with tenofovir disoproxil in both treatment-naive and lamivudine-refractory patients and there were no cases of resistance with up to 2 years of use.

The manufacturer submitted a cost-effectiveness analysis using a Markov model that could be applied either to a cohort of people with HBeAg-positive or HBeAg-negative disease at the start of treatment. The model had 11 main states defined as: active chronic hepatitis B (HBV DNA less than or equal to 300 copies per ml), viral suppression (HBV DNA less than 300 copies per ml), HBeAg seroconverted (not applicable to HBeAg-negative disease), HBsAg seroconverted, compensated cirrhosis with detectable HBV DNA, compensated cirrhosis with undetectable HBV DNA, decompensated cirrhosis, hepatocellular carcinoma, liver transplantation (year in which transplantation occurs), post liver transplantation and death. These were based on health states used in previous economic evaluations. The model was designed to compare tenofovir disoproxil, adefovir dipivoxil, lamivudine and entecavir. It incorporated sequences of first-, second- and third-line treatments and people were assumed to move on to the next treatment regimen if they developed resistance to their current treatment. For people with HBeAg-positive disease and those with HBeAg-negative disease the model has a lifetime horizon, a cycle length of 1 year, and patients are assumed to continue to receive an antiviral regimen until they die, undergo HBeAg seroconversion, undergo HBsAg seroconversion, or develop resistance, at which stage they would switch to an alternative regimen.

In the base-case manufacturer's economic analysis, after treatment sequences that were dominated or extendedly dominated were excluded, the incremental cost-effectiveness ratios (ICERs) of interest in HBeAg-positive chronic hepatitis B were as follows:

• lamivudine as first-line treatment followed by tenofovir disoproxil had an ICER of £6,014 per quality-adjusted life year (QALY) relative to lamivudine followed by best supportive care

• tenofovir disoproxil as first-line treatment followed by lamivudine had an ICER of £9,940 per QALY relative to lamivudine followed by tenofovir disoproxil

• tenofovir disoproxil as first-line treatment followed by the combination of tenofovir disoproxil and lamivudine had an ICER of £13,619 per QALY relative to tenofovir disoproxil followed by lamivudine

• tenofovir disoproxil as first-line treatment followed by the combination of tenofovir and lamivudine followed by entecavir had an ICER of £36,583 per QALY relative to tenofovir disoproxil followed by tenofovir disoproxil plus lamivudine.

In the base-case manufacturer's economic analysis, after treatment sequences that were dominated or extendedly dominated were excluded, ICERs of interest in HBeAg-negative chronic hepatitis B were as follows:

• tenofovir disoproxil as first-line treatment followed by lamivudine had an ICER of £9,811 per QALY relative to best supportive care

• tenofovir disoproxil as first-line treatment followed by the combination of tenofovir and lamivudine had an ICER of £13,854 per QALY relative to tenofovir disoproxil followed by lamivudine alone

• tenofovir disoproxil as first-line treatment followed by the combination of tenofovir and lamivudine followed by entecavir had an ICER of £20,781 per QALY relative to tenofovir disoproxil followed by tenofovir disoproxil plus lamivudine.

The manufacturer also presented results for an analysis in a cohort in which lamivudine resistance had developed:

• tenofovir disoproxil alone as treatment for lamivudine-refractory HBeAg-positive disease had an ICER of £7,707 per QALY relative to best supportive care

• tenofovir disoproxil alone as treatment for lamivudine-refractory HBeAg-negative disease had an ICER of £11,078 per QALY relative to best supportive care.

The ERG viewed the mixed-treatment comparison methodology to be generally sound, but pointed out that it was weakened by the small number of studies (as low as 1 in some networks), a lack of quality assessment of included studies, no discussion of potential clinical heterogeneity and limited discussion of statistical heterogeneity. Therefore the ERG concluded that the results should be treated with caution.

The ERG viewed the pooled analysis of resistance in the manufacturer's submission as appropriate, but pointed out that data for long-term resistance (more than 2 years) are currently unavailable.

The ERG pointed out that there were a number of analytical errors in the manufacturer's electronic model and therefore re-ran the model with discount factors for future health effects applied to all of the model cycles, amendments to transition matrices and a once-only application of a reduction of excess mortality for patients with compensated cirrhosis achieving viral suppression. The results for people with HBeAg-positive disease gave an ICER for first-line tenofovir disoproxil followed by lamivudine relative to lamivudine followed by tenofovir disoproxil of £17,590. The ICER for tenofovir disoproxil followed by lamivudine plus tenofovir disoproxil relative to first-line tenofovir disoproxil followed by lamivudine was £27,479. The results for people with HBeAg-negative disease gave an ICER for first-line tenofovir disoproxil followed by lamivudine relative to lamivudine followed by tenofovir disoproxil of £17,640. The ICER for tenofovir disoproxil followed by lamivudine plus tenofovir disoproxil relative to first-line tenofovir disoproxil followed by lamivudine was £28,324.

Full details of all the evidence are in the manufacturer's submission and the ERG report.