This site is intended for health professionals only

TNF-alpha-targeted therapy in inflammatory arthritides: a valuable option?


The extra costs of TNF inhibitors must be weighed against their extra benefits, and research is warranted to identify patients who will respond significantly better to biologicals than other strategies


Annelies Boonen


Department of Internal Medicine
Division of Rheumatology and Caphri Research Institute
University Hospital Maastricht

According to the Collins Dictionary of the English Language, the value of something refers to (1) the usefulness, appreciation or importance of something, (2) the price, (3) the balance between the ­importance and its price, and (4) the moral principle that it is important in life.(1) When discussing the “value of TNF-α inhibition” in patients with a chronic inflammatory rheumatological condition, it is important to consider all these aspects. In this overview, these issues will be addressed, but first a brief description of the burden of the conditions and the role of TNF-α inhibition (TNFi) will be given.

Rheumatoid arthritis (RA), psoriatic arthritis (PsA) and ankylosing spondylitis (AS) are the three main inflammatory arthritides. Although there are large differences in pathophysiology, epidemiology and clinical manifestations, all three diseases are characterised by persistent joint or spinal inflammation and in most patients increasing damage over time.

In addition to the articular manifestations, most ­patients with AS have one or more associated extra-articular manifestations, including uveitis, inflammatory bowel disease and psoriasis, and in PsA the psoriasis adds to the burden of the disease. Further, patients with inflammatory arthritis are at increased risk of several comorbidities, such as peptic ulcer disease, osteoporosis, cardiovascular disease and in RA, increased mortality.(2) As a consequence of all these manifestations, patients have decreased health-related quality of life (QoL) and increased healthcare consumption and experience restrictions in several aspects of life involvement, including participation in paid and unpaid work or engaging in social and leisure activities.

Medical treatment aims to suppress inflammation as well as retard or arrest joint or spinal damage. In RA and especially AS it is now clear that damage is not (only) the consequence of inflammation but caused by (partly) separate (“disconnected”) physiological pathways. The classic nonsteroidal anti-inflammatory drugs (NSAIDs) can reduce stiffness and pain, but have not been shown to have disease-modifying effects in RA and PsA and have only been shown to have doubtful influence on reduction of spinal damage in AS. Several disease-modifying antirheumatic drugs (DMARDs) are available. In RA, tight disease control using DMARDs and early induction of remission (low disease activity) result in better long-term prognosis. Combination therapy with DMARDS, including corticosteroids, is more successful in achieving this treatment goal than (sequential) monotherapy. In PsA, classic DMARDs showed less beneficial effects and in AS classic DMARDs have no effects on axial manifestations of the disease. With the advent of biologicals, new treatment options emerged.

In RA, strategies that include TNF-α inhibition (TNFi) more often induced long-term remission and resulted in better suppression of radiographic damage in hands and feet. Some studies even showed repair of radiographic damage. Similarly, in PsA, better effects on inflammation and reduction of radiographic progression were shown. In AS, unprecedented suppression of disease activity was seen. It is still controversial whether TNFi will arrest syndesmophyte formation and stop radiographic damage.

Biologically developed drugs are typically expensive and decision-makers are only willing to pay the price if there is substantial additional gain in health compared with classic treatment strategies. A cost-effectiveness (CE) ­ratio represents differences in costs of the new compared to the usual intervention, against differences in effects.

cost (b) – cost (a)
effect (b) – effect (a)

It is important to realise that several perspectives can be taken when computing the ratio. The perspective of payers, such as hospitals or insurance companies, is ­essentially different from that of national authorities. For specific payers only the costs incurred by the organisation for a given effect matter, while for national authorities societal perspective should prevail, which includes all possible costs and all health effects for patient and society. In this overview, we will consider the societal perspective.

Main discussion
Utilities and effect of TNF inhibition
Although disease-specific clinical effect measures are easier to understand when interpreting the CE ratio, decision-makers prefer utility as effect measure. Utilities aims to represent all aspects of health-related QoL and have the advantage of being comparable across diseases. They have the additional characteristic of assessing the “value of health” in the presence of a “choice” between health states. From a health-economic theoretical viewpoint a choice-derived estimate provides a better representation of “true” utility. Various approaches and instruments exist, but may provide different estimates. However, there is so far no consensus on the preferred approach. The most practical and most frequently applied is indirect assessment, in which patients complete their own health profile using a questionnaire. Each profile can be converted into a “societal” utility based on functions derived from choice experiments with healthy persons. The most widely ­applied indirect utility instruments are the EuroQol-5 dimensions (EQ-5D), the Short Form-6 dimensions (SF-6D) and the Health Utility Index (HUI).

Table 1 summarises the literature on utility in cross-sectional assessments in the three inflammatory arthritic conditions and in populations eligible for TNFi.(3−11) Clearly, utility is reduced compared with the general population, which is roughly 0.8 (not taking into account age and gender differences). When assessing utilities in patients eligible for TNFi, utility is much lower than in unselected groups.(12,13,15,16) Note that the main application of the utility is in calculating the QALY – the time-integrated utility:

QALY = time × utility

Considering that AS and PsA start at an earlier age, the loss in QALYs over life for AS and PsA  might well be higher than in RA.


Increasing data support the beneficial effect of TNFi on utility. In active RA, the gain QALY (HUI) was 0.104 to 0.145 a year when adding adalimumab compared with continuing care as usual.(13) In observational cohorts of patients starting on TNFi, a Swedish register reported a gain in QALY (EQ-5D) of 0.28 in the first year(14) and the UK register reported a gain in utility after six months in the responders of 0.25 (EQ-5D) and 0.10 (SF-6D).(12) All these studies included patients with DMARD-resistant RA. In active AS the QALY gain per person per year after starting etanercept was 0.17 (EQ-5D) and 0.07 (SF-6D).(15) In active PsA an increase of 0.066 in SF-6D was reported in an observational cohort after six months’ treatment with TNFi, compared with 0.04 in patients who started methotrexate.(16) An important recent observation is the potential of TNFi to reduce mortality in RA. Clearly, saving life-years will add importantly to the lifetime gain in QALY.

Cost of illness and TNF-α inhibition effects on costs
As yet, no cost-of-illness studies specific for PsA are available. When summarising the literature on the cost of illness of RA and AS and after excluding studies comprising patients on biologicals, the total average costs amounted to €14,369 and €9,475/patient/year for RA and AS respectively. Of these, productivity costs were 59% and 66% in RA and AS respectively (systematic review, unpublished).

In turn, there is evidence that patients with RA or AS who start on TNFi have lower hospitalisation costs (including reduction in incidence of orthopaedic surgery) in the year after introduction of the biological.(14,17) Reductions in sick-leave in the first years on TNFi have been shown in AS.(17) In RA, randomised controlled studies showed no ­improvements in employment compared to placebo,(18,19) but long-term observational data suggest employment might be better in patients receiving biologicals.(20,21) One should note that in the available trials patients with high levels of work disability were included, leaving less opportunity to show beneficial effect on work participation.

Cost-utility models of TNF-α inhibition
In an economic environment with limited financial resources, it is relevant to know whether the additional benefits of costly treatments can be reached at acceptable additional costs compared with the best available alternative treatment. Since the three conditions are chronic and need chronic treatment, long-term health-economic models are more relevant estimates than short-term analyses.

There is a booming literature on models of cost-­utility (CU) of TNFi in inflammatory arthritis. Summarising the studies is a challenge since meta-analytic methods cannot easily be applied in economic evaluations. Among other reasons, published studies are heterogeneous with ­regard to the time horizon over which effects are modelled, the type of patients selected, criteria for starting and continuing TNFi, the comparator treatment or strategy, and technical aspects of the model. It should be realised that several of these issues are country-specific.

We found ten incremental CU models in RA, five in AS and two in PsA. In DMARD-resistant RA, the CU of starting a TNFi was reported to be below €40,000 in five of the nine studies.(12,22–26) Some studies suggest trying another DMARD initially, only starting TNFi in case of failure. In early RA, one study modelled TNFi compared with methotrexate and showed ratios above $60,000/QALY.(25) No model has yet included possible reduction in long-term mortality. Although the cost-effectiveness of switching to another TNFi is much debated, no formal publications are available, despite ever more publications on the effectiveness of switching. In active AS the long-term (25 years or longer) CU of starting a biological showed ratios below €40,000/QALY.(10,27–29) In short-term (two- to five-year) studies the data are conflicting.(10,27–30) This points to the importance of gaining more insight into the natural course of the disease and long-term effectiveness.

In active PsA, etanercept was cost-effective from the UK perspective compared with no change in treatment or with change in DMARD strategy.(31,32) Infliximab resulted in CU ratios above €200,000 per QALY.(32) None of these models included the effect of TNFi of the skin disease.

When comparing the three commercialised biologicals, the initial loading dose of infliximab, the more frequent need for dose increase and in AS the higher dosing regimens likely favours the subcutaneous products from a cost-effectiveness point of view, but some models ­contradict this expectation.

Cost–utility ratio and decision-making
The main aim of the incremental CU ratio is to inform the decision maker how society’s health can be maximised with the lowest additional costs, when decisions on ­reimbursement must be made for different interventions across different diseases. Ideally, ranked listings (league tables) of incremental CU ratios could be used to facilitate comparisons. Some administrations use specific thresholds to guide decisions on whether a ratio is considered cost-effective and the technology should be adopted.

There is ongoing debate on whether incremental CU ratios can be truly compared across diseases and whether the full value of the treatment is represented. Further, the ratio cannot inform on the initial burden of the disease under consideration or on budget implications, and cannot posses relevant equity questions. Finally, there is no agreement on which threshold for a CU ratio is acceptable. Therefore, most administrations also take into account budget implications and considerations on disease burden and equity when making reimbursement decisions.

An increasingly difficult issue is the role of clinicians in reimbursement negotiations. Clinicians face conflict between their role as healthcare providers for individual patients and their societal responsibility to keep healthcare affordable to all those who need it. Maximal improvement of health should always be the first priority. When including cost considerations clinicians should act from the societal perspective, including all aspects of the economic impact of the disease. The primary responsibility of and challenge for clinical researchers will be to provide further insight into which treatment strategy is the best for a given type of patient at a given point in time.

There are substantial indications that the large clinical benefits reported in randomised clinical trials with biologicals in AS and PsA can be a cost-effective option. For AS, this was mainly true for the long-term models, pointing to the relevance of valid information on the course of the disease and the long-term effect of TNFi. In PsA, data are scarce and did not include the effect on skin disease. In RA the data conflict. In early RA, TNFi as first-line treatment seems not cost-effective and in DMARD-resistant RA some models suggest it might be preferable to first try an alternative conventional treatment (if available). In all models, cost-effectiveness improved significantly when productivity costs, an important burden to patients and society, were included. Several issues should be considered when interpreting cost-effectiveness ratios. First, the results of these models depend heavily on the strategies considered and the quality of data used. In this vein, improvements are required (1) to better assess utility and calculate QALYs, (2) to compare the effectiveness of biologicals alone or together with other treatment strategies and (3) to clarify the long-term effects of disease and treatment, including effects on mortality. Second, it should be realised that the full value of TNFi is not necessarily captured in the CU ratio. The initial burden of the disease, the intangible effects on improved societal participation and effect on the burden for the immediate caregiver are likely to be insufficiently reflected. For clinicians, maximising health should remain the primary responsibility. More specific selection of patients who will benefit from “targeted” treatment will likely improve cost-effectiveness. ■

1. Collins Cobuild English Language Dictionary. London: HarperCollins; 1991.
2. Hochberg M, et al, editors. Rheumatology. Edinburgh: Mosby; 2003.
3. Health Qual Life Outcomes 2006;4:25.
4. Arthritis Rheum 2006;55:751-6.
5. Arthritis Rheum 2003;49:483-7.
6. Ann Rheum Dis 2007;66:771-7.
7. Br J Rheumatol 1997;36:551-9.
8. Med Care 2004;42:1125-31.
9. Rheumatology 2002;41:1380-7.
10. Rheumatology 2004;43:1158-66.
11. J Rheumatol 2006;33:289-95.
12. Rheumatology 2007;46:1345-54.
13. Rheumatology 2004;43:712-18.
14. Ann Rheum Dis 2004;63:4-10.
15. J Rheumatol 2007;accepted.
16. Ann Rheum Dis 2007;66:1038-42.
17. Ann Rheum Dis 2004;63:1670-1672.
18. Arthritis Rheum 2006;54:716-22.
19. J Rheumatol 2004;31:849-55.
20. Arthritis Rheum 2003;48:3046-54.
21. J Rheumatol 2007;34:2211-17.
22. Drugs 2005;65:473-96.
23. Rheumatology 2004;43:62-72.
24. Arthritis Rheum 2004;51:964-73.
25. Pharmacoeconomics 2006;24:1221-32.
26. Jobanputra P, et al. The clinical effectiveness and cost-effectiveness of new drug treatments for rheumatoid arthritis: etanercept and infliximab. London: NICE; 2001.
27. Rheumatology 2007;46:1338-44.
28. Rheumatology 2007;46:1320-8.
29. J Rheumatol 2006;33:732-40.
30. Ann Rheum Dis 2006;65:201-8.
31. Rheumatology 2006;45:1029-38.
32. Rheumatology 2007;46:1729-35.

Latest Issue

Be in the know
Subscribe to Hospital Pharmacy Europe newsletter and magazine