Background Joint destruction in rheumatoid arthritis (RA) was until recently seen as an irreversible state. Lately, it was found that repair of bone erosions occurs; however, little is known about its prevalence.
Objective To investigate the frequency of repair and patients' characteristics associated with repair in an inception cohort.
Patients and methods 250 patients with RA, included in the Leiden Early Arthritis Clinic between 1993 and 2000 and treated with conventional disease-modifying antirheumatic drugs, were studied (mean follow-up 10.1 years). Radiographs obtained annually were scored using the Sharp–van der Heijde method, initially aware of the chronology. Patients with a negative change in erosion scores on subsequent radiographs were selected and their series of radiographs were rescored with concealed time sequence by three readers. Repair was defined as agreement between two readers of a negative change in erosion scores that persisted for at least 2 years.
Results Repair was identified in 32 joints in 18 patients (7.2%). Patients with repair had a greater prevalence of autoantibodies (rheumatoid arthritis, anti-citrullinated protein antibody) and a higher level of joint destruction. In the joints with repair, arthritis was absent in the 2 years preceding repair.
Conclusions Repair occurred in 7.2% of the patients with RA, particularly in clinically inactive joints in patients with severe destructive disease.
Statistics from Altmetric.com
Rheumatoid arthritis (RA) often results in destruction of bone and cartilage, visualised on radiographs as erosions and joint space narrowing, respectively. For a long time the bone damage was considered to be permanent.1 Recently some studies sustained the possibility of radiological repair.2,–,6 Dedicated research in the context of Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT),7 8 together with literature reviews,9 10 led to the conclusion that ‘repair does exist’. This is of utmost clinical relevance because it demonstrated that the ‘repair machinery’ can take away, at least partly, the consequences of damage by RA. If the biological basis of this phenomenon could be understood, it would allow the development of treatments specifically targeted at stimulating these repair mechanisms. This study aims to assess the frequency of repair in a large inception cohort of patients with RA treated with conventional disease-modifying antirheumatic drugs (DMARDs), and to characterise the patients showing repair.
Patients and methods
Two hundred and fifty patients with RA, consecutively included in the Leiden Early Arthritis Clinic between 1993 and 2000, were studied. The era 1993–2000 was chosen as it has the longest duration of follow-up (mean 10.1 years, SD 2.3), resulting in a real opportunity for repair to occur. Clinical and laboratory characteristics were measured each year and radiographs of hands and feet were also taken each year.11 Treatment strategies differed for each inclusion period. Patients included between 1993 and 1995 received delayed treatment (initially analgesics, subsequently chloroquine or sulfasalazine (Salazopyrin) and between 1996 and 2000 with prompt initiation of chloroquine, sulfasalazine or methotrexate. Biological agents or aggressive combination therapies were not used.
The radiographs were scored using the Sharp–van der Heijde method12 by one reader, blinded to the clinical data, initially aware of the chronology. Patients with a negative change in erosion scores on subsequent radiographs were selected. Their series of radiographs were mixed with a series of patients with stable or positive change in erosion scores, so that the readers were unaware of the scores that had been assigned previously. These radiographs were rescored with concealed time sequence by three trained readers. The intrareader, intraclass correlation coefficient for the status scores was 0.91. The intraclass correlation coefficient between reader 1 and 2 was 0.94, between reader 1 and 3, 0.95 and between reader 2 and 3, 0.93.
Definition of repair
Repair was defined as fulfilling all of the following three criteria: (a) presence of a negative change in erosion score on a joint level on two subsequent time points both when scored with known and concealed time order; (b) persistence of the lower erosion score for ≥2 subsequent years; (c) agreement on the negative change in erosion score between at least two of the three readers. When data on two subsequent years were not available, all three readers had to agree with the negative change in erosion score.
Patients with and without repair were compared for several baseline characteristics and for the total Sharp–van der Heijde scores during follow-up. Achievement of sustained DMARD-free remission (ie, persistent (≥1 year) absence of arthritis after cessation of treatment with DMARDs)13 was evaluated in both groups. The annually assessed swollen joint count was studied to investigate whether the joints that showed repair had clinically detectable arthritis in the 2 years preceding the development of radiologically visible repair.
Differences in means were analysed with the Mann–Whitney test. Proportions were compared using the χ2 test. The Statistical program for Social Sciences (SPSS) version 14 was used. p Values <0.05 were considered significant.
Prevalence of repair
Seventy of 250 patients with RA had a decrease in the erosion score in any of the joints on at least one occasion, when all series of radiographs were evaluated with known time order. After rescoring with concealed time order, 32 joints with repair were identified in 18 (7.2%) patients. Of these, 26 were small joints of the hands (eight metacarpophalangeal joints, nine proximal interphalangeal joints and nine radiocarpal joints) and six were metatarsophalangeal joints. Thirty joints showed a persistently negative change in erosion score for ≥2 years and for two joints no data on two additional years were available but there was agreement of all three readers on the identification of repair. Sixty-one per cent of the patients showed repair in one joint; 11%, 17% and 11% expressed repair in two, three and four joints, respectively. The highest frequency of repair occurred after 4–6 years' follow-up (figure 1). The frequency of repair was 13.0% for inclusion between 1993 and 1995 and 5.2% for inclusion between 1996 and 2000.
Baseline characteristics of patients expressing repair
Patients with and without repair did not differ in age, gender, Ritchie score, swollen joint count, C-reactive protein (CRP) level and total Sharp–van der Heijde score at baseline (table 1). In contrast, patients with repair were more often rheumatoid factor-IgM positive (OR=3.7, 95% CI 1.2 to 11.5, p=0.025) and anti-cyclic citrullinated peptide positive (OR 7.9, 95% CI 1.8 to 35.2, p=0.007) than the non-repair group.
Disease course of patients expressing repair
Seventeen patients with repair (94%) had an increase in total Sharp score at the same time as showing repair in individual joints; only one patient showed a decrease in total Sharp score, indicating that, next to repair, simultaneous progression was present in other joints.
During the disease course patients with repair had a significantly higher Sharp–van der Heijde scores than patients without repair (figure 2A). A similar observation was made for the total erosion score (figure 2B).
The achievement of sustained DMARD-free remission was compared for patients with and without repair. One patient of the repair group had clinical remission (5.6%), compared with 24.1% (56/232 patients) in the non-repair group (OR=0.15, 95% CI 0.01 to 1.37, p=0.07).
The presence of joint swelling for the 23 joints showing repair in the metacarpophalangeal, proximal interphalangeal or metatarsophalangeal joints was evaluated at the two previous years. This showed that joint swelling was absent in 22 joints in 2 years preceding repair and in one joint, swelling was absent 1 year preceding repair.
This study investigated repair in an inception cohort with a long duration of follow-up. Previous studies dealt with data from clinical trials or evaluated a selected set of patients with RA.2 5 6 14 Importantly, these studies formed the basis for the acceptance of the existence of repair. We now report the prevalence in a large longitudinal cohort of patients with RA treated with conventional treatment strategies. Our results show that despite the absence of aggressive or biological antirheumatic treatment, repair occurs in part of the general RA population.
The prevalence of repair observed here (7.2%) is somewhat lower than reported previously (10.7%).14 We chose a strict definition of repair to reduce the chance of false-positive findings; this may mean that our prevalence is an underestimate. In addition, the comparison of erosion scores of individual joints between two consecutive time points may have introduced misclassification; in some cases repair would have been more easily detected if a larger interval between the radiographs had been compared. Third, the finding of a lower prevalence may be because we studied a general RA population and not a selection of patients with RA.
Interestingly, repair occurred preferentially in patients with severe joint destruction. This might seem surprising as it might be expected that repair will predominantly be present in the patients with a low total level of joint destruction. Several possibilities may explain this observation. First, it may be a methodological factor based on the presumption that a refill is more easily detected in large erosions. If this is true, repair should predominantly be present in joints with a high erosion score. Our data do not support this suggestion. The erosion score for individual joints ranges between 0 (no erosion) and 5 (maximum score). The majority of patients with repair showed a decrease in the erosion score from 2 to 1 or from 1 to 0, and thus did not show repair in joints that were particularly severely damaged. A second possibility is again methodological. In patients with considerable damage, many joints show erosions and therefore in these patients more joints are ‘at risk’ for showing repair. A third possible explanation is biological. In general, the human body tends to heal destruction and aims for homoeostasis. It can be hypothesised that the more destruction is present, the more regenerating processes are activated. Then after the inflammation or the processes that drive the destruction of bone have disappeared, the enhanced regenerating mechanisms may result in repair.
At the same time as repair occurred in some joints, the total Sharp–van der Heijde score increased, indicating progression in other joints. This is in concordance with a study performed by the OMERACT group,4 and implies that repair is a localised process. The observed absence of joint swelling in the 2 years preceding repair is in line with similar findings in the TEMPO trial.15
In conclusion, repair occurred in 7.2% of conventionally treated patients with RA, particularly in clinically inactive joints in patients with severe destructive disease. Further studies on the biological basis of repair are needed as they may allow the development of treatments specifically targeted at stimulating these repair mechanisms.
The work of AHMH is supported by The Netherlands Organisation for Health Research and Development and the Dutch Arthritis Association.
MPML and RB contributed equally.
Competing interests None.
Ethics approval This study was conducted with the approval of the LUMC medical ethical committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.