In various studies of patients with recent onset arthritis, 6–55% of the patients who presented with undifferentiated arthritis (UA) progressed into rheumatoid arthritis (RA) according to the American College of Rheumatology (ACR) classification criteria or according to the rheumatologist.¹ Multivariate analysis of characteristics of these patients showed that the presence of antibodies to cyclic citrullinated peptide (anti-CCP), polyarthritis, symmetric arthritis, and erosions on radiographs predict the development of RA. Among these characteristics, anti-CCP has the highest specificity and positive predictive value for RA.² Thus, either by clinical or by serological data, clinicians will be confronted with patients with UA who are likely to develop RA. The natural course of patients who present with UA and subsequently develop RA is unknown compared with patients who present with RA. In patients with RA many studies have demonstrated the benefit of the early initiation of disease modifying antirheumatic drugs (DMARDs), resulting in less disease activity and reduction of radiographic joint damage.^3–,⁵ Patients included in these studies fulfilled the 1987 ACR classification criteria for RA.⁶ However, the ACR classification criteria were not developed for the purpose of early diagnosis as is illustrated by the fact that a minority of patients with inflammatory polyarthritis fulfilled the ACR criteria at the first visit.⁷

Although patients with RA benefit from early aggressive treatment,⁸ it is unknown whether patients with UA whose disease will evolve into RA will also benefit from such a treatment strategy. Hypothetically, if at presentation the duration of complaints is similar, patients who present with UA might have a disease course that evolves more slowly than that of patients who present with RA (fig 1⇓). In the worst situation, the “window of opportunity” for early treatment to induce long term disease modification in the patients who presented with UA has been passed once the diagnosis RA is made, and long term outcome will be comparable or worse than patients who were diagnosed with RA at presentation and treated immediately.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Hypothetical disease course of UA and RA.

Here, we describe the 1 year diagnosis and outcome of patients who presented with UA, and compare the disease course of the patients with UA who had developed RA within 1 year after presentation (UA-RA group) with that of patients who presented with RA (RA-RA group). Radiographic progression was the major outcome measure; disease activity and functional capacity were minor outcome measures.

PATIENTS AND METHODS

Study design

From the EAC database (fig 2⇓), patients diagnosed with UA at the second visit were selected and their subsequent diagnosis was recorded at 1 year. UA comprised arthritis of unknown cause and “probable RA” according to the 1958 ACR criteria.¹⁰ The group of patients diagnosed with UA who evolved into RA fulfilling the 1987 ACR criteria within 1 year (UA-RA group) was followed up for 4 years. Because these patients had the opportunity to be treated with DMARDs as soon as the diagnosis was made, the long term outcome was compared with that of a group of patients in whom RA was diagnosed at the first and/or second visit and who still had RA after 1 year, and who were treated with DMARDs as soon as RA was diagnosed (RA-RA group).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

 Flow diagram for definition of the groups.

Statistical analysis

Differences between patient characteristics were tested with the Mann-Whitney U test or the χ² test. For each patient separately, simple linear regression was used to fit the course of the radiographic progression in time. Progression rates were calculated for varying time intervals, assuming that radiographic damage progresses at a constant rate.^13,14 The slopes of these regression lines estimate the average increase in Sharp/van der Heijde score per year. Differences between slopes of the groups of interest were tested with the Mann-Whitney U test.

To construct the radiographic progression as depicted in fig 3⇓, the geometric mean of the Sharp/van der Heijde scores was estimated using a linear mixed model, in which account was taken of missing measurements.¹⁵ Differences in disease activity and functional capacity were compared with a linear mixed model. All tests were two tailed and p values ⩽0.05 were considered significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

 Estimated geometric mean Sharp/van der Heijde Score (95% confidence interval) of the UA-RA and RA-RA groups from baseline to 4 years.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

 Secondary outcome values of the UA-RA and RA-RA groups from baseline to 4 years. (A) Modified Disease Activity score; (B) Health Assessment Questionnaire score; (C) erythrocyte sedimentation rate in mm/1st h; (D) C reactive protein concentration. Mean values (SEM).

RESULTS

DISCUSSION

Here, the hypothesis was tested whether patients with UA who developed RA within 1 year have a different disease course than those who presented to the Leiden EAC with RA fulfilling the ACR criteria.⁶ Of the 330 patients who initially presented with UA, 91 (28%) had developed RA within 1 year. Although the patients with UA might hypothetically have evolved more slowly towards RA, joint damage, disease activity, and functional capacity were the same in both groups over a follow up period of 4 years.

UA is a commonly encountered problem in daily practice. In one third of patients with recent onset arthritis it is not possible to come to a definite diagnosis at presentation. Overall, 63% of the patients at the Leiden EAC had a self limiting disease course and 28% progressed from UA to RA. Other studies show that 6–55% of patients with UA will eventually evolve into RA.¹ These percentages are dependent on the different inclusion criteria, recruitment procedures, and disease criteria. For example, low percentages are found in RF negative patients with monarthritis,¹⁶ and higher percentages are found in inception cohorts where patients with polyarthritis are diagnosed with RA, regardless of fulfilment of the ACR criteria.¹⁷

Despite the fact that many studies have tried to describe the natural history of UA, the long term outcome of the subgroup of patients with UA who develop RA has never been described. As can be expected, the UA-RA and RA-RA groups differed at presentation in duration of morning stiffness, number of swollen joints, modified DAS, and percentage of RF positive patients, because these variables are directly or indirectly part of the 1987 ACR classification criteria for RA. The presence of anti-CCP was equal in both groups and is similar to previous studies.¹⁸ The group of patients with UA who did not develop RA within 1 year differed significantly in all patient characteristics at baseline from the other two groups. This is a very heterogeneous group considering the diagnosis after 1 year, but all these patients were diagnosed with UA at baseline.

Median symptom duration is about the same in both groups. However, given the large variation in symptom duration some misclassification of patients might have occurred. Theoretically, patients with UA who presented with short symptom duration might have belonged to the RA-RA group, and patients with RA with very long symptom duration might have had the chance of being included in the UA-RA group. However, analysis of subgroups with different intervals of symptom duration did not show an effect on joint damage.

In the dropout analysis, only CRP in the UA-RA group was significantly higher in the dropout group than for the completers. These data argue against the possibility that the UA-RA group equals the RA-RA group because the dropouts in the UA-RA group were the patients who presented with a milder disease, and only the patients with more aggressive disease completed the follow up.

At all time points, the UA-RA group received fewer DMARDs than the RA-RA group. In this observational study, this might reflect a historical development in DMARD prescription. It also might be a confounding by indication, as clinicians are inclined to prescribe DMARDs to patients who they expect to have a more destructive disease course. Then, patients in the RA-RA group might have a more aggressive disease course that is modified by DMARDs. However, the power to detect an effect of the presence or absence of DMARD use on radiographic progression was too low given the small number of patients with RA who did not use DMARDs (n = 16). In the UA-RA group, 16/91 patients did not receive DMARDs during the 4 year follow up. Although 11 of these patients were RF negative, 10 had erosive disease. However, with a median radiographic progression of one point a year, this subgroup had less progressive disease than the group treated with DMARDs. A randomised clinical trial with DMARDs in patients with UA would be the most appropriate study design to overcome this issue.

The overall outcome of UA poses challenging treatment goals. Patients with UA who will go into remission will need treatment aimed at reducing symptoms, whereas patients prone to progress to RA will need treatment to improve their long term outcome as well. Furthermore, there is a need for new criteria to identify patients with UA that will evolve into RA, as the current ACR criteria for RA are not suitable for this purpose. This study implies that the long term outcome of patients who present with UA which evolves into RA is the same as that of patients who present with RA. New serological tests combined with clinical judgment will help to determine which patients with UA are most likely to develop RA. Randomised clinical trials in such patients with UA are currently under way to test whether this group of patients may benefit from very early DMARD treatment.