Assessment of clinical effectiveness

We searched for studies of the clinical effectiveness of adefovir dipivoxil, pegylated interferon alpha-2a (PEG-α-2a) and pegylated interferon alpha-2b (PEG-α-2b) (note that the latter was not included in the original report). Searches were run from the beginning of 2005 to September 2007. Thirteen bibliographic databases were searched, including MEDLINE, EMBASE and the Cochrane Library.

All studies were screened against a set of pre-specified inclusion criteria. For the clinical effectiveness review, we included randomised controlled trials (RCTs) which compared ADV, PEG-α-2a, and PEG-α-2b with currently licensed treatments for CHB, including the immunomodulatory drug non-pegylated interferon alpha (IFN-α) and the nucleoside analogue lamivudine (LAM).

Outcomes included biochemical (alanine aminotransferase, ALT), histological (liver fibrosis and necroinflammation) and virological [HBV deoxyribonucleic acid (DNA)] response to treatment, drug resistance and adverse effects. The trials were reviewed in a narrative synthesis but meta-analysis was not undertaken because of heterogeneity in the interventions and comparators evaluated.

Assessment of cost-effectiveness

A systematic review of economic evaluations of antiviral treatments for CHB was conducted. In addition, the economic model devised for our previous report was updated using utility values based on a recent study eliciting health-state valuations from CHB-infected patients. The model was also updated to account for changes in methodological guidance on discount rates for costs and outcomes. Health-state and treatment costs were inflated to 2006–7 prices. Evidence for the clinical effectiveness of PEG-α-2b was used in the model to estimate the cost-effectiveness of PEG-α-2b compared with IFN-α.

Clinical effectiveness

Literature searches yielded a total of 735 articles. Of these, 653 were excluded on the basis of title and, where available, abstract. Eighty-two papers were retrieved for detailed screening and eight randomised controlled trials (RCTs) were included in the systematic review:

• Three evaluated ADV, one of which was a long-term follow-up of a trial included in our original assessment report. In two trials ADV was compared with placebo, and in a third ADV was compared with ADV added to LAM in patients with LAM resistance.
• Four evaluated PEG-α-2b. In two of these PEG-α-2b was combined with LAM and compared with either PEG-α-2b monotherapy or LAM monotherapy. Another compared three staggered regimens of PEG-α-2b combined with LAM. The fourth trial compared PEG-α-2b monotherapy with IFN-α.
• A further PEG-α-2b RCT was included from our original literature search database (but not included in the original assessment report as it was not in the scope of the review at that time). This RCT compared PEG-α-2b combined with LAM with PEG-α-2b monotherapy.
• No RCTs of PEG-α-2a were identified.

The trials varied in terms of aims, size and design characteristics. Five included only HBeAg-positive patients, with the remaining three including only HBeAg-negative patients.

Methodological quality also varied. Some trials reported adequate blinding, allocation concealment and randomisation methods, while other trials either failed to report such details or were judged inadequate.

Cost-effectiveness

The systematic review of cost-effectiveness studies identified four relevant full economic evaluations, in addition to one full economic evaluation identified and partially reviewed in our original assessment report. Two of the evaluations assessed PEG-α-2a; the remainder assessed ADV. Four of the five economic evaluations used Markov models, with lifetime horizons, while the other study used a decision tree with a 4-year time horizon. State-transition diagrams in the evaluations were similar, identifying the treatment aim as inducing HBeAg seroconversion for patients with HBeAg-positive CHB and viral suppression for patients with either HBeAg-positive or HBeAg-negative CHB.

Economic evaluations of PEG-α-2a found that it was associated with increased treatment costs but also gains in quality-adjusted life expectancy. In a UK study, the incremental cost-effectiveness ratio (ICER) for PEG-α-2a was £10,444 per QALY gained compared with LAM. Evaluations of ADV found that LAM monotherapy was dominated, while the ICER for ADV monotherapy compared with 'doing nothing' was $19,731($14,342–$24,224) at 2005 prices.

A review of health-state utility values used in economic evaluations of antiviral treatments for CHB showed that widely varying values were used, many of which were not specific to CHB patients. A recently published study reporting health-state utilities for patients with CHB infection and for non-infected general population samples, derived using the standard gamble technique, was identified and reviewed.

The ICERs generated by the update of our economic model were generally less favourable than those reported in the original assessment report. However, it appears that much of the difference arises from recent changes to methodological guidance (i.e. discounting costs and outcomes at 3.5% rather than 6% and 1.5% respectively) rather than from changes in costs or health-state utilities.

The sequential treatment strategies identified as optimal in our original report remained optimal in the updated model, i.e. interferon (pegylated or non-pegylated) followed by LAM, with ADV as salvage for patients who develop LAM resistance.

The results of the updated analysis were generally robust to changes in deterministic sensitivity analysis. The most notable changes were in the ICER for the strategy including ADV as salvage therapy for patients who develop resistance to LAM, in some cases increasing the ICER beyond the threshold conventionally used to indicate cost-effectiveness in the context of NHS decision making.

• The most influential structural assumption was excluding the possibility of HBeAg seroconversion (in HBeAg-positive CHB) in patients with compensated cirrhosis, which increased the ICER to £40,833 per QALY gained.
• In terms of the baseline characteristics of the treated cohort, decreasing the proportion with HBeAg-positive CHB and increasing age were associated with less favourable ICERs.
• The most influential parameter values related to the gain in utility associated with HBeAg seroconversion and loss of the surface antigen (HBsAg). This affected the ICERs for all strategies, but was most notable for the strategy including ADV as salvage for patients who develop resistance to LAM. If there is no utility gain for HBeAg seroconversion or loss of HBsAg, the ICER increases to £31,114.

In a probabilistic sensitivity analysis the same sequence of treatments was identified as optimal. However, the strategy including ADV as salvage becomes optimal only above a willingness-to-pay threshold of £27,000 per QALY. This is at the upper limit of the range of ICERs regarded as cost-effective from an NHS decision-making perspective. Interferon (conventional or pegylated) followed by LAM is optimal for a willingness to pay of £9000–£26,000, compared with a range of £5000–£11,500 in our previous report. As discussed, much of this difference arises from changes in the practice of discounting rather than changes to input values in the model.

The ICER for PEG-α-2b, compared with IFN-α-2b, in patients with HBeAg-positive CHB was £9169, based on the results of a clinical trial of 24 weeks of interferon treatment. The trial did not include a placebo arm, so no ICER for PEG-α-2b compared with best supportive care was estimated. Results were generally robust to changes in deterministic sensitivity analysis.

• Increasing age of the cohort and lower utility gains from HBeAg seroconversion or loss of HBsAg were associated with less favourable ICERs.
• Alternative discount rates (6% for costs and 1.5% for outcomes, as in our previous report, or 0% for both costs and outcomes) and a reduction in cost for PEG-α-2b were associated with more favourable ICERs.
• All ICERs in the one-way sensitivity analyses were below the threshold conventionally deemed as cost-effective.

In a probabilistic sensitivity analysis, PEG-α-2b had a probability of being cost-effective (compared with IFN-α-2b) of 79% at a willingness-to-pay threshold of £20,000 per QALY, and 86% at a willingness-to-pay threshold of £30,000 per QALY.