Objectives To explore whether changes of MRI-defined synovitis and bone marrow lesions (BMLs) are related to changes in joint tenderness in a 5-year longitudinal study of the Oslo hand osteoarthritis (OA) cohort.
Methods We included 70 patients (63 women, mean (SD) age 67.9 (5.5) years). BMLs and contrast-enhanced synovitis in the distal and proximal interphalangeal joints were evaluated on 0–3 scales in n=69 and n=48 patients, respectively. Among joints without tenderness at baseline, we explored whether increasing/incident synovitis and BMLs were associated with incident joint tenderness using generalised estimating equations. Among joints with tenderness at baseline, we explored whether decreasing or resolution of synovitis and BMLs were associated with loss of joint tenderness. We adjusted for age, sex, body mass index, follow-up time and changes in radiographic OA.
Results Among joints without tenderness at baseline, increasing/incident synovitis and BMLs were seen in 45 of 220 (20.5%) and 47 of 312 (15.1%) joints, respectively. Statistically significant associations to incident joint tenderness were found for increasing/incident synovitis (OR=2.66, 95% CI 1.38 to 5.11) and BMLs (OR=2.85, 95% CI 1.23 to 6.58) independent of structural progression. We found a trend that resolution of synovitis (OR=1.72, 95% CI 0.80 to 3.68) and moderate/large decreases of BMLs (OR=1.90, 95% CI 0.57 to 6.33) were associated with loss of joint tenderness, but these associations were non-significant.
Conclusions The Oslo hand OA cohort is the first study with longitudinal hand MRIs. Increasing synovitis and BMLs were significantly associated with incident joint tenderness, whereas no significant associations were found for decreasing or loss of synovitis and BMLs.
- Hand Osteoarthritis
- Magnetic Resonance Imaging
- Outcomes research
Statistics from Altmetric.com
Hand osteoarthritis (OA) may lead to substantial pain, in similar degree as in rheumatoid arthritis.1 Currently, there is limited knowledge about the aetiology of pain in hand OA. The determinants of pain in OA are not well understood, but are believed to involve multiple interactive pathways that can be best understood in a biopsychosocial model.2 From a biological perspective, pain may arise from the multitude of affected joint tissues with nociceptive fibres, and these are likely sources of pain in OA. Whereas the cartilage is aneural, the bone, synovium, articular capsule and ligaments are richly innervated.
Several cross-sectional imaging studies have suggested that radiographic findings are associated with pain.3 In recent years, more attention has been given to ultrasound and MRI.4 The advantage of ultrasound and MRI is the ability to demonstrate joint inflammation,5 whereas MRI is the only imaging modality that can demonstrate bone marrow lesions (BMLs).6 BMLs are often occurring in joint areas with increased loading,7 ,8 and correspond to areas with fibrosis, necrosis, oedema, bleeding of fatty marrow and abnormal trabeculae on histopathology.9 ,10 Several cross-sectional ultrasound studies have indicated an association between synovitis and hand OA pain,11 ,12 and we have previously shown that MRI-defined synovitis and BMLs are cross-sectionally associated with joint tenderness.13 In general, significant associations have been found between imaging features and pain/tenderness in the same joint, whereas the associations between sum scores of imaging features and patient-reported hand pain are much weaker or absent.11 ,13
Due to the cross-sectional design of previously performed hand OA studies on imaging and pain, conclusions about the causality cannot be drawn. In general, the number of longitudinal studies exploring causes of pain in hand OA is limited. Using longitudinal data from the Oslo hand OA cohort, we have previously shown that radiographic progression, and especially incident erosions, is associated with de novo pain in the same joint.14 In an open-label trial of methylprednisolone in hand OA, no significant association was found between decrease of ultrasound-detected synovitis and decreased pain.15 As hand OA studies with repeated MRI are lacking, longitudinal associations between synovitis and BMLs and joint tenderness are unknown.
Hence, the aim of this study was to explore whether changes of synovitis and BMLs are related to changes in tenderness on an individual joint level in a 5-year longitudinal hand OA study.
Participants of the Oslo hand OA cohort have attended three scheduled examinations (figure 1). In 2001–2003 (examination 1), 209 patients with hand OA were examined with joint assessments, conventional radiography and questionnaires. Details of the patient recruitment have been published previously.1 The same examinations were performed in 2008–2009 (examination 2)13 ,16 and 2013 (examination 3). At examinations 2 and 3, the participants also underwent MRI examination of the dominant hand. All participants signed informed consent at all three examinations.
In the current analyses, we included 70 participants with available data on joint tenderness and MRI of the dominant hand at examination 2 (hereafter referred to as ‘baseline’) and examination 3 (hereafter referred to as ‘follow-up’). Among the 209 participants who were included in the Oslo hand OA cohort, there were no significant differences in sex distribution (p=0.75), mean age (p=0.12), mean Australian/Canadian (AUSCAN) pain score (0.90), mean number of tender joints (p=0.69) and mean amount of radiographic OA (p=0.93) at the first examination in 2001–2003 between those who were included (n=70) and not included (n=139) in the current analyses (figure 1).
MRI of the dominant hand was performed at baseline (2008–2009) and follow-up (2013) with the same MRI unit and protocol. The second to fifth distal interphalangeal (DIP) and proximal interphalangeal (PIP) joints were imaged using a high-field extremity 1.0 T MRI unit (ONI, GE Healthcare, Waukesha, Wisconsin, USA) and a cylindrical coil (diameter 10 cm). Coronal, sagittal and axial T1-weighted fat-suppressed preintravenous and postintravenous gadolinium (0.1 mmol gadolinium/kg body weight; Magnevist, Bayer Schering Pharma AG, Leverkusen, Germany) images were acquired from a three-dimensional dual-echo Dixon technique (repetition time (TR) 20 ms, echo time (TE) 5 ms, 1 mm slice thickness), in addition to coronal and axial Short tau inversion recovery (STIR) images (TR 2850 and 3150 ms, TE 16.3 and 21 ms, 2 and 3 mm slice thickness, respectively).17
One reader (IKH) scored the paired MRIs with known time sequence according to OMERACT (Outcome Measures in Rheumatology) hand OA MRI score for the presence of synovitis (grade 0–3) and BMLs (grade 0–3). Detailed description of the scoring system and its reliability has been published.18 Synovitis was defined as thickened synovium with enhancement of gadolinium and was scored as 0=normal, 1=mild, 2=moderate and 3=severe synovitis according to a proposed atlas.18 BMLs were defined as lesions with signal characteristics consistent with high signal intensity on STIR images and ill-defined margins. Grade 1, 2 and 3 BMLs represented lesions with size 1%–33% (mild), 34%–66% (moderate) and 67%–100% (severe) of the assessed bone volume, respectively. The proximal and distal parts of the joint were assessed combined and the assessed bone volume extended from the articular surface to a depth of 0.5 cm for the DIP joints and 1.0 cm for the PIP joints.18 We allowed 0.5 increments for change scores of synovitis and BMLs, in case of definite changes within the same grade.18 In total, MRI scans of 40 joints from five patients at baseline and follow-up were re-evaluated. The intrareader reliability for change scores (increase vs decrease vs no change) was good for both synovitis (κ=0.77) and BMLs (κ=0.76).
Bilateral hand radiographs (posteroanterior view) were obtained at baseline (2008–2009) and follow-up (2013). One reader (IKH) scored the paired hand radiographs with known time sequence. The bilateral second to fifth DIP, second to fifth PIP, thumb interphalangeal, first to fifth metacarpophalangeal and thumb base joints were scored for OA severity according to a modified Kellgren–Lawrence Scale (grade 0–4): KL grade 0=no hand OA; 1=minimal hand OA, that is, questionable osteophyte and/or joint space narrowing (JSN); 2=mild hand OA, that is, small osteophyte(s) and/or mild JSN, sclerosis may be present; 3=moderate hand OA, that is, moderate osteophyte(s) and/or moderate JSN, sclerosis and erosions may be present; 4=severe hand OA, that is, large osteophyte(s) and/or severe JSN, sclerosis and erosions may be present.19 ,20 Scoring of the thumb base joints included assessment of the first carpometacarpal joints and triscaphoid joints in a combined score. The reader has shown moderate to good intrareader reliability (weighted κ=0.82 for status scores, κ=0.83 for change scores).4
Readings of MRIs and conventional radiographs were performed blinded for clinical information and other imaging features.
Clinical joint examinations were performed at baseline (2008–2009) and follow-up (2013). One rheumatologist (BSC) examined all finger joints (including the DIP and PIP joints imaged by MRI) for absence/presence of joint tenderness upon palpation and/or movement.21
The participants completed questionnaires at baseline (2008–2009) and follow-up (2013). The AUSCAN pain subscale includes five questions about hand pain during rest and daily activities, each scored on a 0–4 scale (total range 0–20).22 ,23
The primary aim was to explore the longitudinal associations between MRI features (independent variables) and tenderness in the same joint (dependent variable). Among joints without tenderness at baseline, we explored whether incident/increasing synovitis and BML scores during follow-up were associated with de novo tenderness in the same joint using generalised estimating equations (GEE). Joints with no change or decreasing synovitis and BMLs served as references.
Among joints with tenderness at baseline, we explored whether decrease of synovitis and BMLs were associated with loss of joint tenderness. In these analyses, joints with no change or increasing synovitis and BMLs served as reference. We repeated the analyses using resolution of synovitis or BMLs as the independent variables, and joints with decreased but no resolution of synovitis or BMLs were included as reference joints. Separate models were performed for synovitis and BMLs.
The secondary aim was to explore whether changes in the sum scores of synovitis and BMLs were associated with changes in AUSCAN pain using linear regression analyses. Separate models were performed for each sum score. The sum scores at baseline and follow-up were calculated by summing the individual scores for the eight DIP and PIP joints imaged by MRI, and the change scores from baseline to follow-up were calculated subsequently.
The analyses were adjusted for age, sex, body mass index, follow-up time and changes in radiographic OA (ie, change in Kellgren–Lawrence grade in individual joints for analyses at joint level and change in Kellgren–Lawrence sum score for the 30 hand joints for analyses on patient level).
Baseline characteristics are shown in table 1. Among the five participants who did not fulfil the American College of Rheumatology criteria for hand OA, all patients had hand pain and radiographic evidence of hand OA. In total, 70 participants were included in the analyses; 69 participants (n=551 joints included, one joint missing) with longitudinal STIR images were included in the analyses on BMLs and 48 participants (n=373 joints included, 11 joints missing) with longitudinal pre- and post-gadolinium T1-weighted MRI and post-gadolinium T1-weighted MRI were included in the analyses on synovitis (see online supplementary table S1).
Longitudinal associations between MRI features and tenderness in the same joint
The mean (SD) follow-up time was 4.7 (0.4) years. Table 2 shows the changes of joint features during follow-up. The majority of joints with incident/increasing synovitis (70/96 joints) and BMLs (64/88 joints) had small changes (0.5–1 grade). Similarly, small changes were observed in the majority of joints with decreasing synovitis (54/63 joints) and BMLs (37/47 joints). In joints with decreasing scores, complete resolution of synovitis and BMLs occurred in 39 of 63 and 30 of 47 joints, respectively. Radiographic progression was rather high and occurred in 20.9% of the joints.
Among joints without tenderness at baseline, increasing/incident synovitis and BMLs were significantly associated with incident tenderness in the same joint (table 3). The associations were independent of each other with significant associations for increasing/incident synovitis (OR=2.50, 95% CI 1.25 to 5.00) and BMLs (OR=3.43, 95% CI 1.23 to 9.56) when both features were included in the same model with additional adjustment for radiographic progression (OR=1.24, 95% CI 0.58 to 2.63). For BMLs, there was a stronger association with incident joint tenderness for joints with a moderate/large increase (2–3 grades) than for joints with a small increase (0.5–1 grade; OR=6.67, 95% CI 1.40 to 31.79 vs OR=1.83, 95% CI 0.74 to 4.57). We found no dose–response relationship for synovitis (data not shown).
Among joints with tenderness at baseline, no significant associations were found between decreasing synovitis and BMLs and loss of joint tenderness during follow-up (table 4). For BMLs, we found a trend towards a dose–response relationship with a stronger, although statistically non-significant, association for joints with moderate/large decrease than for joints with small decrease (OR=1.90, 95% CI 0.57 to 6.33 vs OR=0.64, 95% CI 0.33 to 1.26). No dose–response relationship was observed for synovitis (data not shown). However, there was a non-significant trend that complete resolution of synovitis was associated with loss of joint tenderness (table 4).
Longitudinal associations between MRI sum scores and patient-reported hand pain
In general, the patients reported less hand pain at follow-up than at baseline. The median (IQR) change of AUSCAN pain was −4.5 (−6 to −2) on a 0–20 scale. Changes in the sum scores of BMLs and synovitis were also small with median (IQR) of 1 (0–2) and 1 (−1 to 3), respectively (both on 0–24 scales). No significant associations were found between changes of synovitis and BML sum scores and changes in AUSCAN pain (table 5).
The Oslo hand OA cohort is the first hand OA study with longitudinal hand MRIs. Increasing synovitis and BMLs were significantly associated with incident joint tenderness, and a non-significant trend towards associations between resolution of synovitis and moderate/large decrease of BMLs and loss of tenderness was observed.
Our results are in line with previous cross-sectional hand OA studies, supporting that synovitis and BMLs are causes of pain in hand OA.11–13 A systematic review of knee OA studies, mainly consisting of cross-sectional studies, also concluded that synovitis and BMLs were associated with pain.24 Cross-sectional studies are hampered by the inability to draw conclusions about causality. Due to involvement of all joint structures in OA, including both inflammatory and structural abnormalities, confounding by other joint abnormalities that are co-occurring is difficult to avoid.
Even if structural OA progresses over time, synovitis and BMLs may fluctuate, as shown in the current study. In our study, we observed only small changes in the total amount of synovitis and BMLs. However at joint level, a large proportion of joints showed either progression or decrease of the relevant MRI features. In fact, almost half of the joints (42.6%) showed either increase or decrease of synovitis, whereas BMLs changes were less common (24.5%). These observations are in line with recently published results by Kortekaas et al12 showing that ultrasound-detected synovitis remains stable at patient level over a period of 3 months, whereas 20.5% of the joints had inflammatory features at one time point only. Initially, it was hypothesised that BMLs could only increase with time.25 However, recent knee OA studies have shown that they fluctuate, also in a short-term period of 6–12 weeks.26–29 In the current study, we observed relatively few BMLs compared with the study by Wittoek et al.30 The rather low prevalence in this study, even in joints with severe OA,16 may be explained by the slice thickness of 2–3 mm for the STIR images making it difficult to detect small BMLs. Similar low prevalence was observed by Vlychou et al31 using MRIs with slice thickness of 3–4 mm.
Similar fluctuations can also be observed for OA-related pain.12 In the current study, the patients reported less pain at follow-up as measured by the AUSCAN Hand Index. A mean change of −4.5 points on a 0–20 scale is clearly above the minimal clinically important improvement (MCII) of 1.49.32 ,33 As much as 79.4% of the patients had an improvement in AUSCAN pain above the MCII value. There are several possible explanations, such as regression to the mean. Patients may also develop strategies for how to effectively cope with the disease, how to avoid pain by not performing provoking activities and may be provided with better pain treatment and use of technical aids. We found no evidence for the disease ‘burning out’ with time, as the total amount of synovitis slightly increased during follow-up. However, this hypothesis should be explored at joint level with longitudinal analyses of the changes of synovitis in relation to structural progression.
Although the total amount of hand pain decreased, we observed an increase in the number of tender joints. This discordance may be due to measurement errors related to both patient-reported and physician-based outcomes. Alternatively, these two pain outcomes may reflect different pathogenetic processes. Whereas joint tenderness may reflect allodynia as a consequence of peripheral and/or central sensitisation in OA joints, activity-related pain (constituting four of five items in AUSCAN) may be more related to the subjects´ experiences and coping strategies. Furthermore, the discordance could potentially be due to differences in the assessed joints. However, when we included the number of tender metacarpophalangeal and thumb base joints in the analyses, we still found an increase in tender joint count (data not shown).
Due to the fluctuating nature of synovitis, BMLs and pain in hand OA, the causal associations between them can be more accurately explored in longitudinal studies. Studying improvement or complete resolution of synovitis and BMLs and their relation to decreased hand OA circumvents the problem of confounding by structural OA worsening. A strength of our study is the adjustment for changes in radiographic OA, supporting that our findings are independent of changes in structural abnormalities.
Our results are also in line with previous longitudinal studies on both BMLs25 ,27–28 ,34 and synovitis in knee OA,35 although the findings are not consistent across all studies.29 ,36 ,37 In line with our results, Zhang et al27 showed that worsening of synovitis and effusion was associated with an increased risk of pain. They were also able to demonstrate an association between improvement of BMLs and decreased risk of pain. The causal association between BMLs and pain was also confirmed in a randomised trial on bisphosphonate (Zoledronic acid) in knee OA.38 In the current study, no significant associations could be observed between decrease of synovitis and BMLs and less pain. In line with our results, Keen et al15 found no significant association between decrease of ultrasound-detected synovitis and decreased pain in a trial of methylprednisolone in hand OA. However, a placebo-effect on pain could not be ruled out due to the open-label study design, and ultrasound was possibly not sensitive enough to detect synovial changes in this cohort of patients with established OA.15 First of all, the lack of associations in the current study could be due to the fact that joint tenderness was not assessed on a semiquantitative scale, but rather as ‘absent/present’. Hence, we were not able to detect decreases in the severity of joint tenderness. There was a tendency that resolution of synovitis and moderate/large decrease of BMLs were associated with loss of joint tenderness. However, the numbers were small, and future studies should explore these possible associations in larger patient samples.
Some study limitations should be acknowledged. Although the sample size of 70 persons was rather small, we were able to study as much as 560 joints due to the polyarticular nature of hand OA, while taking the within-patient dependency into account by using GEE. However, future studies with larger patient-samples are needed in order to confirm the results from the current study. We did not have any joint-specific measure of pain except from tenderness by palpation/movement. The correlations between physician-reported tenderness on palpation/movement and patient-reported pain in the same joints have not yet been explored. The MRIs and radiographs were scored by one reader only. Using a second reader, a more accurate scoring of OA pathology could have been obtained. Intraperson reliability for joint examination was not assessed.
In conclusion, using longitudinal MRI and clinical data from the Oslo hand OA cohort, we demonstrated significant associations between increasing/incident synovitis and BMLs and incident joint tenderness, supporting the validity of synovitis and BMLs as sources of pain in hand OA. Both synovitis and BMLs may potentially be targets for pharmacological interventions. Future observational and intervention studies should explore whether reducing synovitis and BMLs lead to less pain, which we were not able to prove in this study.
We thank the patients of the Oslo hand OA cohort for participating in this study, project coordinators Ingvild Karlsen Eeg and Silje Røysen Salvador, study nurse Anne-Katrine Kongtorp, medical students serving as research assistants and the technicians in the Department of Radiology for helping us to collect the data.
Handling editor Francis Berenbaum
Contributors IKH: Study design, data collection, analysing the data, interpretation of results, drafting the work and final approval of the paper. BSC: Data collection, interpretation of results, revising the work critically and final approval of the paper. PB: Data collection, interpretation of results, revising the work critically and final approval of the paper. DvdH: Interpretation of results, revising the work critically and final approval of the paper. TKK: Study design, interpretation of results, revising the work critically and final approval of the paper.
Funding The study was supported by grants from ‘Extrastiftelsen’/‘the Norwegian Rheumatism Association’, ‘Grethe Harbitz’ legacy for combating rheumatic diseases’ and ‘Dr Trygve Gythfeldt og wife's research fund’.
Competing interests None.
Patient consent Obtained.
Ethics approval Regional 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.