Objectives: This report reviews the clinical effectiveness and cost-effectiveness of adalimumab, etanercept and infliximab, agents that inhibit tumour necrosis factor-alpha ( TNF-alpha), when used in the treatment of rheumatoid arthritis ( RA) in adults. Data sources: Electronic databases were searched up to February 2005. Review methods: Systematic reviews of the literature on effectiveness and cost-effectiveness were undertaken and industry submissions to the National Institute for Health and Clinical Excellence ( NICE) were reviewed. Meta-analyses of effectiveness data were also undertaken for each agent. The Birmingham Rheumatoid Arthritis Model ( BRAM), a simulation model, was further developed and used to produce an incremental cost-effectiveness analysis. Results: Twenty-nine randomised controlled trials (RCTs), most of high quality, were included. The only head-to-head comparisons were against methotrexate. For patients with short disease duration (<= 3 years) who were naive to methotrexate, adalimumab was marginally less and etanercept was marginally more effective than methotrexate in reducing symptoms of RA. Etanercept was better tolerated than methotrexate. Both adalimumab and etanercept were more effective than methotrexate in slowing radiographic joint damage. Etanercept was also marginally more effective and better tolerated than methotrexate in patients with longer disease durations who had not failed methotrexate treatment. Infliximab is only licensed for use with methotrexate. All three agents, either alone ( where so licensed) or in combination with ongoing disease-modifying antirheumatic drugs (DMARDs), were effective in reducing the symptoms and signs of RA in patients with established disease. At the licensed dose, the numbers needed to treat (NNTs) (95% CI) required to produce an American College for Rheumatology (ACR) response compared with placebo were: ACR20: adalimumab 3.6 (3.1 to 4.2), etanercept 2.1 (1.9 to 2.4), infliximab 3.2 (2.7 to 4.0); ACR50: adalimumab 4.2 (3.7 to 5.0), etanercept 3.1 ( 2.7 to 3.6), infliximab 5.0 (3.8 to 6.7); and ACR70: adalimumab 7.7 (5.9 to 11.1), etanercept 7.7 (6.3 to 10.0), infliximab 11.1 ( 7.7 to 20.0). In patients who were naive to methotrexate, or who had not previously failed methotrexate treatment, a TNF inhibitor combined with methotrexate was significantly more effective than methotrexate alone. Infliximab combined with methotrexate had an increased risk of serious infections. All ten published economic evaluations met standard criteria for quality, but the incremental cost-effectiveness ratios (ICERs) ranged from being within established thresholds to being very high because of varying assumptions and parameters. All three sponsors who submitted economic models made assumptions favourable to their product. BRAM incorporates improvements in quality of life and mortality, but assumes no effect of TNF inhibitors on joint replacement. For use in accordance with current NICE guidance as the third DMARD in a sequence of DMARDs, the base-case ICER was around 30,000 pound per quality-adjusted life-year ( QALY) in early RA and 50,000 pound per QALY in late RA. Sensitivity analyses showed that the results were sensitive to the estimates of Health Assessment Questionnaire (HAQ) progression while on TNF inhibitors and the effectiveness of DMARDs, but not to changes in mortality ratios per unit HAQ. TNF inhibitors are most cost-effective when used last. The ICER for etanercept used last is 24,000 pound per QALY, substantially lower than for adalimumab (30,000 pound per QALY) or infliximab (38,000 pound per QALY). First line use as monotherapy generates ICERs around 50,000 pound per QALY for adalimumab and etanercept. Using the combination of methotrexate and a TNF inhibitor as first line treatment generates much higher ICERs, as it precludes subsequent use of methotrexate, which is cheap. The ICERs for sequential use are of the same order as using the TNF inhibitor alone. Conclusions: Adalimumab, etanercept and infliximab are effective treatments compared with placebo for RA patients who are not well controlled by conventional DMARDs, improving control of symptoms, improving physical function, and slowing radiographic changes in joints. The combination of a TNF inhibitor with methotrexate was more effective than methotrexate alone in early RA, although the clinical relevance of this additional benefit is yet to be established, particularly in view of the well-established effectiveness of MTX alone. An increased risk of serious infection cannot be ruled out for the combination of methotrexate with adalimumab or infliximab. The results of the economic evaluation based on BRAM are consistent with the observations from the review of clinical effectiveness, including the ranking of treatments. TNF inhibitors are most cost-effective when used as last active therapy. In this analysis, other things being equal, etanercept may be the TNF inhibitor of choice, although this may also depend on patient preference as to route of administration. The next most cost-effective use of TNF inhibitors is third line, as recommended in the 2002 NICE guidance. Direct comparative RCTs of TNF inhibitors against each other and against other DMARDs, and sequential use in patients who have failed a previous TNF inhibitor, are needed. Longer term studies of the quality of life in patients with RA and the impact of DMARDs on this are needed, as are longer studies that directly assess effects on joint replacement, other morbidity and mortality.
