Article Text

Concise report
MRI-detected subclinical joint inflammation is associated with radiographic progression
  1. A Krabben1,
  2. W Stomp2,
  3. J A B van Nies1,
  4. T W J Huizinga1,
  5. D van der Heijde1,
  6. J L Bloem2,
  7. M Reijnierse2,
  8. A H M van der Helm-van Mil1
  1. 1Department of Rheumatology, Leiden University Medical Center, The Netherlands
  2. 2Department of Radiology, Leiden University Medical Center, The Netherlands
  1. Correspondence to A Krabben, Department of Rheumatology, Leiden University Medical Center, P.O. Box 9600, Leiden 2300 RC, The Netherlands; A.Krabben{at}


Background We recently demonstrated that MRI inflammation is prevalent in clinically non-swollen joints of early arthritis patients. In this study, we assessed the relevance of this subclinical inflammation with regard to radiographic progression.

Methods 1130 joints (unilateral metacarpophalangeal 2–5, wrist and metatarsophalangeal 1–5) of 113 early arthritis patients underwent clinical examination and 1.5 T MRI at baseline, and radiographs at baseline and 1 year. Two readers scored the MRIs for synovitis, bone marrow oedema (BME) and tenosynovitis according to Rheumatoid Arthritis (RA) Magnetic Resonance Imaging (MRI) Scoring System (RAMRIS). Radiographic progression over 1 year was determined using the Sharp–van der Heijde scoring method.

Results On patient level, BME, synovitis and tenosynovitis were associated with radiographic progression, independent of known risk factors (p=0.003, 0.001 and 0.011, respectively). Of all non-swollen joints (n=932), 232 joints (26%) had subclinical inflammation (≥1 MRI-inflammation feature present). These joints were distributed among 91% of patients. Radiographic progression was present in 4% of non-swollen joints with subclinical inflammation compared to 1% of non-swollen joints without subclinical inflammation (relative risks (RR) 3.5, 95% CI 1.3 to 9.6). Similar observations were done for BME (RR5.3, 95% CI 2.0 to 14.0), synovitis (RR3.4, 95% CI 1.2 to 9.3) and tenosynovitis (RR3.0, 95% CI 0.7 to 12.7) separately.

Conclusions Radiographic progression was infrequent, but joints with subclinical inflammation had an increased risk of radiographic progression within year 1. This demonstrates the relevance of MRI-detected subclinical inflammation.

  • Arthritis
  • Inflammation
  • Magnetic Resonance Imaging
  • Rheumatoid Arthritis
  • Outcomes research

Statistics from


The severity of radiographic joint destruction varies between patients with rheumatoid arthritis (RA). Much research has been focussed on identifying risk factors for radiographic progression, with the ultimate aim of achieving individualised medicine at disease onset. Well-known risk factors are presence of erosions, presence of RA-related autoantibodies (rheumatoid factor (RF), anticitrullinated-peptide antibodies (ACPA)) and measures of inflammation (C-reactive protein (CRP), number of swollen joints (SJC)).

MRI measures inflammation sensitively; it allows assessment of bone marrow oedema (BME), tenosynovitis and synovitis. Previous studies in wrist and MCP joints showed that inflammation detected with MRI, especially BME, is a strong predictor for radiographic progression.1–8 (see online supplementary table S1) However, none of these studies investigated MRI findings in relation to clinical joint inflammation. We recently studied MRI inflammation in small joints that were not swollen at physical examination, and observed MRI inflammation in 27%, 66% and 13% of non-swollen metacarpophalangeal (MCP), wrist and metatarsophalangeal (MTP) joints of early arthritis patients.9 The identification of these inflamed joints may indicate the added value of MRI. However, the relevance of such subclinical inflammation is supported when it associates with radiographic progression. In this study, we therefore investigated whether MRI-detected subclinical inflammation is associated with radiographic progression.

Patients and methods


Between August 2010 and February 2012, MRI was performed in 179 early arthritis patients of the Leiden Early Arthritis Clinic, a population-based inception cohort including patients with confirmed clinical arthritis and symptoms for <2 years. At baseline, questionnaires and 66-SJCs were performed and serum obtained.10 All patients underwent extremity MRI at baseline; of these 113 had 1-year follow-up including radiographs. Patients without 1-year follow-up were less often diagnosed with RA (see online supplementary table S2). Of the 113 patients studied, 53 fulfilled the 2010-criteria for RA at baseline. During the first year, three-quarters of the patients were treated with conventional disease-modifying antirheumatic drugs (table 1). All participants provided written informed consent. The study was approved by the local medical ethics committee.

Table 1

Baseline characteristics 113 early arthritis patients

MRI and radiographs

The MCP2-5, wrist and MTP1-5-joints at the most painful side (or dominant side in case of equally severe symptoms at both sides) were scanned with a 1.5T-extremity MRI scanner, according to the Outcome Measures in Rheumatology (OMERACT)-Rheumatoid Arthritis (RA) Magnetic Resonance Imaging (MRI) Scoring System (RAMRIS) protocol. A detailed scan protocol is provided in the online supplementary methods. Synovitis and BME were scored according to RAMRIS11; tenosynovitis in MCP joints and wrists was assessed as described elsewhere.12 Two readers scored the MRIs independently, blinded to clinical data; the mean scores were studied. Within-reader ICCs for the total MRI inflammation score were 0.99 and 0.93; the between-reader ICC was 0.87. The total MRI inflammation was the sum of synovitis, BME and tenosynovitis scores. Subclinical inflammation was defined as inflammation on MRI in clinically non-swollen joints. Radiographs were scored according to the Sharp–van der Heijde scoring (SHS) method by one reader in chronological order (ICC baseline SHS 0.86). Radiographic progression was defined as the difference in SHS between year 1 and baseline.


First, to replicate previous associations between MRI inflammation and radiographic progression,1–7 analyses were done on patient level using a linear regression model with radiographic progression (continuous variable) as outcome. Univariable and multivariable analyses (adjusting for age, gender, ACPA, RF, CRP level and 66SJC) for total inflammation and each MRI feature separately, were performed. Joints that could not be completely scored on MRI due to insufficient image quality (1.1% of all individual scores) were imputed with the median value for that feature across all joints or bones within the same patient.

Next, analyses were performed on joint level. For MRI, the sum scores per joint and feature were determined according to RAMRIS. Similarly, the sum scores per joint on radiographs were determined according to SHS (see online supplementary methods). Missing MRI scores were not imputed at joint level. To compare frequencies of subclinical inflammation the continuous MRI data was dichotomised; joints with a score ≥1 were considered positive. In sensitivity analyses a cut-off ≥2 was evaluated. Joints that progressed ≥1 SHS point in 1 year were considered to have radiographic progression. SPSS V.20.0 was used; p values <0.05 were considered significant.


Total MRI inflammation on patient level

Table 1 presents baseline characteristics of early arthritis and patients with RA. First, we evaluated the total MRI inflammation scores in relation to radiographic progression and observed that both the total score and the individual MRI features were associated with radiographic progression in early arthritis and RA, independent of known risk factors (p<0.05 except for tenosynovitis in RA, table 2). We herewith replicated previous observations.1–7

Table 2

Associations between baseline MRI inflammation and radiographic SHS progression over the first year

Subclinical MRI inflammation on patient level

When separating patients’ total MRI inflammation scores (mean 13.9±12.0) into the scores obtained in swollen joints, and the scores obtained in non-swollen joints, the mean score in swollen joints was 7.2±9.4 and in non-swollen joints 6.6±7.8. This suggests that the cumulative amount of MRI-inflammation in non-swollen joints was comparable with that in swollen joints. We subsequently studied non-swollen joints only.

Online supplementary figure S1 showing the number of joints with subclinical inflammation reveals that most patients had 1–3 subclinical inflamed joints, and only 9% of patient had no joints with subclinical inflammation. Comparing the number of subclinical inflamed joints between patients with RA and other diagnoses revealed that patients with RA tended to have more joints with subclinical inflammation (2.3 vs 1.9, p=0.11). Evaluating the different inflammation features separately showed that patients with RA mainly had more joints with BME in non-swollen joints (1.6 vs 1.1, p=0.03).

Subclinical inflammation on joint level

Next, we investigated subclinical inflammation at joint level. Of all 1130 joints studied 932 joints were clinically non-swollen. Of these, 232 (26%) had any subclinical MRI inflammation:17% of non-swollen joints had BME, 16% had synovitis and 21% tenosynovitis (figure 1). Two percent of the 932 non-swollen joints had radiographic progression during year 1 (compared with 8% of the swollen joints). The non-swollen joints with and without subclinical MRI inflammation were compared to determine the relative risks (RR) of radiographic progression:4% of non-swollen joints with any MRI inflammation had radiographic progression versus 1% of the joints without subclinical inflammation (RR 3.5, 95% CI 1.3 to 9.6). Similar analyses for the individual MRI inflammation features revealed that BME in non-swollen joints had the highest RR of radiographic progression (RR 5.3, 95% CI 2.0 to 14.0; 7% vs 1%). For synovitis the RR was 3.4 (95% CI 1.2 to 9.3; 5% vs 1%). For tenosynovitis the RR was 3.0 (95% CI 0.7 to 12.7). When repeating these analyses in joints that were non-swollen and also non-tender, the RRs obtained were comparable, although 95% CIs were broader (see online supplementary figure S2). When analysing wrist joints only, a higher percentage of non-swollen joints with subclinical inflammation was observed (86% vs 26%) and radiographic progression occurred more frequently (11% vs 3%, online supplementary figure S3). When analysing radiographic progression in non-swollen hand and foot joints of patients with RA (n=53), the RRs went into the same direction for synovitis and BME, but the 95% CIs were broad (see online supplementary figure S4). Furthermore, sensitivity analyses with a higher cut-off to define MRI inflammation (≥2 instead of ≥1) revealed similar results (see online supplementary figure S5).

Figure 1

Distribution of MRI inflammation among clinically non-swollen joints and the percentages of joints with radiographic progression during 1-year follow-up.


This study is the first investigating the radiographic outcome of MRI-detected inflammation in clinically non-inflamed joints at disease presentation. We observed that joints with subclinical inflammation, especially with BME, had an increased risk of radiographic progression during the first year.

Our observation that the total level of MRI inflammation (in both swollen and non-swollen joints) was an independent predictor for radiographic progression fits with previous findings.1–7 Notably, the effect sizes of BME and synovitis in our study were similar to those of Boyesen et al1 (β 0.04 and 0.06 for BME and 0.11 and 0.12 for synovitis, respectively). Furthermore, our finding that subclinical inflammation at disease onset is associated with radiographic progression is in line with previous findings on subclinical inflammation in patients with RA in clinical remission.13–17

This study has several limitations. First, the sample size, particularly of the RA group, was moderate. Although the effect sizes obtained in the total group and RA group were similar, the 95% CIs of the estimates in the RA group were broad. Second, radiographic progression was infrequent, in all early arthritis patients (3% of all joints) but also in RA, a group of patients that is more likely to erode (4%). Third, radiographic progression was defined as ΔSHS on joint level of ≥1, this concerned a small increase but the radiographs were scored in chronological order which reduced the chance on measurement errors. Up-to-date treatment strategies will have contributed to the low prevalence of radiographic progression. Furthermore, the frequency of progression was determined on joint level, which is relatively uncommon. To obtain a reference, we also analysed previously SHS-scored unilateral MCP2–5, wrist and MTP1–5 joints of EAC RA patients included in 1993–1999 and 2000–2006 when different treatment strategies were applied.10 Here, using the same definition, radiographic progression during year 1 was present in 15% and 9% of joints, suggesting that in the currently studied patients treatment had effectively reduced radiographic progression. Nonetheless, despite radiographic progression being infrequent nowadays, progression was significantly more frequent in joints with subclinical inflammation. This suggests that MRI may be valuable to identify joints with increased risk of progression, despite normal physical examination and current treatment strategies.

This study mainly increases the comprehension of the connection between inflammation and structural damage early in the disease. Whether subclinical MRI inflammation is relevant to clinical practice remains a question, as rheumatologists treat patients and not joints. Information on subclinical inflamed joints would affect treatment decisions most when patients have few clinically swollen joints. A subanalysis in patients with ≤2 swollen joints showed a slight tendency towards more progression in the presence of more subclinically inflamed joints (ΔSHS1.4 in case of ≥3 subclinically inflamed joints vs ΔSHS 1.0 in case of ≤2 subclinically inflamed joints). Larger studies are required to ascertain whether information on MRI inflammation is relevant for clinical practice.

MRI is a sensitive tool, and MRI inflammation has been reported in symptom-free persons.18–20 Nevertheless, present data indicate that MRI-detected subclinical inflammation in early arthritis negatively affects radiological outcome.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • Contributors Clinical and MRI data were collected by AHMvdHvM, MR, WS and AK. MRIs were scored by WS and AK. Radiogrpahs were scored by JABvN. Analyses were performed by AK. The manuscript was written by AHMvdHvM and AK. Critical and valuable comments on the article were made by WS, JABvN, TWJH, DvdH, JLB and MR.

  • Funding Dutch arthritis Foundation, Dutch Organisation of Health Research and Development, CTMM, TRACER, Masterswitch. The work of A Krabben is supported by a grant of the Dutch Arthritis Foundation. This work is supported by a Vidi grant of the Dutch organisation of Health Research and Development. The work of W Stomp and M Reijnierse is supported by the CTMM, the Center for Translational Molecular Medicine (, and the Dutch Arthritis Foundation, project TRACER (Grant 04I-202). The research has been funded by The European Community Seventh Framework Programme FP7 Health-F2-2008-223404 (Masterswitch) and by a core grant of the Dutch Arthritis Foundation.

  • Competing interests None.

  • Ethics approval Local medical ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.