• treatment discontinuation
• anti-tumour necrosis factor α
• rheumatoid arthritis

Treatment with biological agents is currently inducing dramatic changes in the treatment of the most common inflammatory rheumatic diseases: rheumatoid arthritis (RA), ankylosing spondylitis (AS) and psoriatic arthritis. Several recent papers have established the efficacy of mainly three different agents directed against tumour necrosis factor α (TNFα)—infliximab, etanercept, and adalimumab—in these rheumatic diseases.^1–6 Because these treatments may have rare but severe side effects⁷ and because they are expensive, guidelines and recommendations have been proposed by expert groups, especially for RA⁸ and recently also for AS.⁹

In the AS consensus paper, which was produced by the ASAS working group, recommendations for the discontinuation of anti-TNF therapy in clinical practice have been included.

This is in contrast with the RA guidelines, which in St Martin were discussed and updated only recently in April 2003. Although guidelines for the discontinuation of treatment in RA were proposed in the discussions held there, this issue has not been put forward mainly as a result of the argument that this subject is too difficult to deal with. Thus, the arguments and proposals published in this letter may represent a minority statement (in relation to the group that met to update the RA guidelines^9a).

In our opinion, a discussion on discontinuation of anti-TNF therapy in RA is important in clinical practice and in clinical studies because (a) treatment with these agents is costly and healthcare systems do not have unlimited financial resources in most countries. Therefore, unnecessary treatment with these agents should not be performed because other patients who are in definite need of this treatment may be excluded. (b) In an increasing number of studies patients are switched from one biological agent to another. This is generally done because it is the expert’s opinion that this is the right thing to do. This may lead to highly subjective assessments and unnecessary disqualification of treatments which have at least produced some degree of improvement. Usually, a definition of a non-responder as an end point is well defined in each clinical trial (those not fulfilling the criteria for a responder). However, the problem arises when patients “not responding to drug A” are included in a trial to investigate drug B. These are often patients gathered in clinical practice, and for these patients no clear definition is given.

Therefore, we feel that some basic measures should be agreed upon which might represent a starting point for this discussion. We propose to consider the following for both clinical practice and clinical studies:

Discontinuation of anti-TNF therapy in RA should be considered and possibly performed if the following problems have occurred:

• Intolerable side effects probably or possibly related to the drug applied

• Lack of efficacy.

The second point deserves further clarification. We think that the treatment with either agent should have lasted for at least two months and include the usual saturation phase in the case of infliximab at weeks 0, 2, and 6. We would like to emphasise that this recommendation says “at least”, which does not imply that one cannot wait one or two months longer, because we are aware that some patients with very severe disease may need longer to experience sufficient clinical benefit.

We realise that the issue of increasing the dosage in selected patients has not been properly assessed to date.

Furthermore we think that it should be recorded that the disease activity has not been substantially decreased. The outcome measures that are most frequently used in RA in clinical trials are ACR20%, ACR50%, and ACR70% and the disease activity score (DAS; reviewed by Verhoeven et al¹⁰). In clinical practice, the DAS is more widely used, at least in Europe.

We propose that patients with RA who do not fulfil the ACR20% and/or a DAS improvement of 1.2 after at least two months can be formally classified as non-responders. It seems possible that this relatively short period of time can be extended to three months in certain cases. However, in our clinical experience no major increases of improvement due to anti-TNF agents can be expected if nothing has changed after two months. Most importantly, the definition of non-response as an inclusion criterion should be given in each trial. If the DAS is used the cut off point for non-response is improvement of at least 1.2. Thus, if these levels of response are reached (ACR20% or DAS improvement of at least 1.2), the patient may not be regarded as a non-responder without giving further definitions on this.

There are some differences in these proposals compared with the recommendations given for AS.⁹ The guidelines for AS recommend assessing disease activity by the Bath AS Disease Activity Index (BASDAI),¹¹ with a value of 4 generally considered to indicate clinically significant disease activity. For assessment of improvement in AS, criteria have been proposed¹² on the basis of trials with non-steroidal anti-inflammatory agents (NSAIDs), in which four domains (pain, function, patient global, and morning stiffness) are defined and three of these must improve by 20%. However, these criteria were recently found not to be suitable for trials with anti-TNF agents¹³ because much higher response rates are obtained with these agents. In the consensus statement on anti-TNF therapy in AS, discontinuation of such treatment was proposed to be considered in patients not reaching 50% improvement of BASDAI—an initially rather arbitrary measure,¹⁴ which, however, has been used in most of the later trials performed in AS.^15–20 In addition to the BASDAI assessment, expert opinion to start or discontinue treatment is essential.

Although difficult to compare directly, this threshold seems higher than the one proposed for RA. This may possibly represent another indication that anti-TNF therapy is more efficacious in AS than in RA.

We do hope that these proposals contribute to a more scientific approach in future studies including so-called “non-responders” in RA and propose to discuss these proposals in more detail in the next meeting on “Targeted Therapies”.