Tumour necrosis factor α (TNFα) antagonist therapies have been successfully used to treat patients with established and early rheumatoid arthritis (RA) and have extended the therapeutic goal. The current aim of treatment is to prevent radiographic damage and preserve function and quality of life, with early combination treatment, tight disease control and early use of biological agents.1^–6 Perhaps the most ambitious goal is to prevent RA by treating patients with evolving disease—that is, those with poor prognosis undifferentiated inflammatory arthritis (UA). These high-risk patients can be identified by symptom duration >12 weeks and relapse after intramuscular steroid.7

It has previously been shown that a course of methotrexate (MTX) delays, but does not “reset”, the course of disease in probable RA.8 The beneficial outcomes of the “PROMPT” trial were most pronounced in anti-cyclic citrullinated peptide CCP (anti-CCP) antibody-positive (and rheumatoid factor (RF)-positive) patients.

Infliximab has been shown to produce drug-free remission in patients with early RA.5 However, the impact of TNF antagonist therapy on patients with poor prognosis UA before the diagnosis of RA, is unknown.

The aim of this pilot study was to assess the short- and long-term benefits of induction treatment with infliximab in poor prognosis UA.

PATIENTS AND METHODS

This study was a randomised, placebo-controlled trial of disease-modifying antirheumatic drug (DMARD) naïve patients with UA affecting more than one joint, with <12 months of symptoms. Patients had to have had a recurrence of synovitis of affected joints after a single dose of intramuscular corticosteroid. Patients were also required to have a C-reactive protein (CRP) >10 mg/l within the past 3 months and a Health Assessment Questionnaire (HAQ) raw score >4 (HAQ 1.3). The protocol was approved by the local ethics committee in 2003 and all patients provided written informed consent.

Patients received infliximab monotherapy (3 mg/kg) or placebo at 0, 2, 6 and 14 weeks. When clinical inflammation (joint synovitis with raised CRP) was persistent, disease-modifying treatment (MTX) was started after week 14. Primary outcome was the number of patients in clinical remission (CRP <10 mg/l and no clinical synovitis) at 26 weeks. Secondary outcomes included change (compared with week 0) in clinical, laboratory and quality of life measurements. Persistent inflammatory symmetrical arthritis (PISA) scores were calculated for patients at baseline. This combined five variables, including serology, gender, genetics and baseline function, as indicators of poorer prognosis in early arthritis. A score of 3 or greater out of 6 was an indicator of disease severity.

Anti-CCP antibody results were determined from blood samples obtained during follow-up. x-Ray examinations of hands and feet were performed at baseline, at the time of diagnosis of RA and yearly thereafter.

Statistical analysis

This study was initiated as a pilot study with 20 patients in each arm. Because of the poor outcome of all subjects, recruitment was halted ahead of time by the independent safety monitors. An intention-to-treat analysis was performed using the last observation carried forward method. For categorical outcomes non-responder imputation was used. Nominal data were compared between groups using continuity-corrected χ² tests. Continuous and ordinal data were compared between groups using Mann–Whitney U tests. Data are presented as median (interquartile range), unless otherwise stated.

RESULTS

Seventeen patients with UA were recruited; on randomisation 10 received infliximab and seven placebo. There was no difference between the two groups at baseline (table 1). Fifteen patients had started MTX by week 26. Four patients failed to complete the study—two patients from the infliximab group; one with repeated chest infections and the other with a drug-induced antineutrophil cytoplasmic antibody (ANCA) positive vasculitis (both withdrawn at week 6 and before the start of MTX). Two patients were withdrawn from the placebo group owing to non-response.

At week 14, the infliximab group had greater improvements in CRP and HAQ (table 2). The mean 28 joint count Disease Activity Score (DAS28) was 4.46 with no significant difference between the two groups.

Only three patients fulfilled the primary end-point of remission at week 26. Five patients had withdrawn by week 26; three from placebo group and two from the infliximab group. Assuming non-response, 1/7 (14%) of patients in the placebo group and 2/10 (20%) in the infliximab group achieved clinical remission. Of the three patients in clinical remission, two were positive for both anti-CCP antibody and RF and one was negative for both antibodies and received placebo. The former two patients were diagnosed with RA before study completion and the latter patient continues to be in DAS28/clinical remission without DMARDs 6 months after completion of the study.

The infliximab group had a median improvement in EMS, TJC, SJC and HAQ that exceeded the placebo group at week 26; however, a rebound of fatigue was observed on a visual analogue scale (p = 0.043). This was not seen when non-responder imputation was used. The mean DAS28 was 4.31 with no significant difference between the two groups at week 26.

All patients in the infliximab group and five patients in the placebo group fulfilled criteria for RA by week 52 with a difference in the time to develop RA between the two groups (infliximab group median 26 weeks; placebo group median 14 weeks). However this difference was not significant as illustrated by the Kaplan–Meier graph (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Comparison of time to diagnosis of rheumatoid arthritis (RA) between the infliximab and placebo groups.

All the patients who developed RA had baseline PISA scores ⩾3, whereas the two patients who did not develop RA had PISA scores ⩽2. Eighty per cent of patients who developed RA were anti-CCP antibody positive. At week 52 the patients who received infliximab had lower levels of CRP (p = 0.043).

For radiographic progression, two patients in the infliximab group and two patients in the placebo group developed new erosions by week 52.

DISCUSSION

As far as we know, this is the first placebo-controlled trial of TNF blockade in patients with UA. In a group of patients with UA with poor prognosis who had relapsed after a single corticosteroid injection, a short course of TNF antagonist therapy provided modest short-term relief but importantly did not prevent progression to RA.

These results are of particular interest in light of recent findings by van Dongen et al,8 who found that in anti-CCP positive patients with probable RA, progression to RA was delayed but not prevented by a course of MTX. The high rate of development of RA seen in the current study was not predicted, although only patients with a poor prognosis who had relapsed after corticosteroid were included. This latter feature, together with a disease duration >12 weeks and possession of the shared epitope are among the best markers for persistent disease.7 9 Various prediction scores for poor outcome and persistence have been designed. This study used the PISA10 11 score, which is a simple numerical score that predicts poor prognosis. The calculation of PISA scores was adapted in light of recent evidence to include anti-CCP antibody instead of shared epitope.12 A score of ⩾3 has been associated with poor outcome10 and in this study, the development of RA. PISA scores have not previously been used to predict progression to RA.

Recently, van der Helm-Van Mil et al described a prediction rule for patients with early UA such that a score ⩾9 was associated with progression to RA in 84% of patients.13 This rule was applied to our patients with a slight modification (converting minutes of EMS to a score; 1 point for 60−120 minutes and 2 points for greater than 120 minutes). The mean score for the patients who developed RA was 9.6 (range 6.9–14.38) versus 5.71 (range 5.2–6.22) for those who did not.

There are limitations to this study; with the most obvious being the small numbers, making it difficult to extrapolate the results on a wider scale. However, the results were quite emphatic as shown by the fact that the independent monitors considered it unethical to continue the study. Furthermore, a previous pilot study of 20 patients5 accurately predicted the effect size of early TNF blocking therapy in RA.4 The withdrawal rate due to lack of efficacy may have been reduced by use of DMARDs in combination with infliximab or higher doses of infliximab (5 mg/kg). As patients with UA have milder disease with lower CRP and TNF levels, it was postulated that lower levels of TNF blocker would be required for effect.

In conclusion, this study suggests that the outcome of poor prognosis UA is not influenced by short-term treatment with TNF blockade as measured either by remission or by preventing progression to RA. The importance of markers of poor prognosis, especially flare following corticosteroid injection, in patients presenting with UA was confirmed. Thus it is clear that patients can be identified who will progress to RA at a stage before damage. This finding is crucial in attempts to induce remission and prevent disability.