Objectives To investigate the longitudinal relationship between inflammatory lesions in sacroiliac joints on MRI (MRI-SI) and clinical disease activity measures (DA) in patients with axial spondyloarthritis (axSpA).
Methods Two-year follow-up data from 167 patients (50% males, mean (SD) age 33 (9) years) fulfilling the Assessment of SpondyloArthritis international Society axSpA criteria in the DEvenir des Spondylarthopathies Indifférenciées Récentes cohort with MRI-SI at baseline, 1 year and 2 years were analysed. The relationship between MRI-SI (as dependent variable) and DA (Ankylosing Spondylitis Disease Activity Score (ASDAS), Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), patient's global DA, night pain, C reactive protein and erythrocyte sedimentation rate, as independent variables) was investigated using two types of generalised estimating equations (GEE) models: model of absolute scores and model of change scores.
Results In the model of absolute scores, the relationship between DA and MRI-SI was different for males and females: in males, but not in females, a statistically significant relationship with MRI-SI was found for all DA except BASDAI. In the model of changes, only ASDAS (beta (95% CI): 2.79 (0.85 to 4.73) and pain at night (0.97 (0.04 to 1.90)) were significantly associated in males while again in females no significant relationship was found. ASDAS fitted the data best.
Conclusions In male patients, but not in female patients, with axSpA, clinical DA, especially if measured by ASDAS, is longitudinally associated with MRI-SI inflammatory lesions.
Statistics from Altmetric.com
The diagnostic value of MRI of sacroiliac joints (MRI-SI) in axial spondyloarthritis (axSpA) is nowadays clearly established.1 For reasons of understanding the disease axSpA better, it is important to know whether or not clinical signs and symptoms are related to the degree of inflammation that is occurring in the sacroiliac joints, which can be observed on MRI. Data evaluating this are scarce. Furthermore, most of them stem from clinical trials including patients with active and mostly established disease, but almost no data are available for patients in early stages of the disease or with low disease activity.2–5 Additionally, all these studies only provided cross-sectional correlations, but in none of these a longitudinal analysis was conducted. Based on this, the current study aims to investigate the longitudinal relationship between MRI-SI inflammatory lesions and clinical measures (DA) in patients with early axSpA; that is, whether or not changes on MRI-SI inflammation are related to changes on DA over time.
Two-year follow-up data from the DEvenir des Spondylarthopathies Indifférenciées Récentes (DESIR) cohort were analysed. Design and inclusion criteria of this cohort have been reported.6 In summary, 708 patients with inflammatory back pain (>3 months but <3 years), age 18–50 years old and probability of SpA >50% based on the physician's assessment were recruited. Database used for this analysis was locked on 30 June 2014.
For this study, patients fulfilling the Assessment of SpondyloArthritis international Society (ASAS) criteria for axSpA at baseline with MRI-SI and DA available for at least two consecutive visits were included.
Clinical disease activity measures
DA were collected every six months and included Ankylosing Spondylitis Disease Activity Score with C reactive protein (ASDAS-CRP), Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), patient’s visual analogue scale of night pain and global assessment of DA, CRP and erythrocyte sedimentation rate (ESR). If serum CRP was below the limit of detection, a value of 2 was used for the ASDAS calculation.7
MRI of sacroiliac joints
MRI-SI (T1 and short tau inversion recovery sequences) was performed at baseline. Additionally, MRI-SI was repeated at 12 and 24 months, but due to budget restrictions, only in 9 out of the 25 participating centres. Two trained readers, blinded to clinical data and timepoint, independently scored all MRI-SI images for the presence of inflammatory lesions according to the Spondyloarthritis Research Consortium of Canada (SPARCC) scoring method (range 0–72).8 The mean of the two readers was used for this analysis. The intraclass correlation coefficient was 0.94 for absolute baseline scores and 0.93 for change scores (baseline 12 months). The smallest detectable difference was 6.1 (absolute baseline scores), and the smallest detectable change was 3.6 (mean of two change intervals).
Where dichotomous MRI data (positive vs negative MRI) were required, a cut-off level of 2 SPARCC units was assumed to appropriately reflect the distinction between an ASAS-positive and ASAS-negative MRI.9
The longitudinal relationship between MRI-SI (SPARCC) and DA was analysed using generalised estimating equations (GEE) models, with MRI-SI (SPARCC) as dependent variable and the separate clinical measures as independent variable. Two types of models were run: (1) a standard model, for which the absolute scores of SPARCC and DA were used; and (2) a model of changes, for which the change scores between two consecutive measurements of both SPARCC and DA were employed. While the standard model includes a pooled analysis of longitudinal (within-subject) and cross-sectional (between-subjects) relationships, the model of changes ‘removes’ the cross-sectional part of the analysis enabling a real longitudinal interpretation.10
Interactions with age, gender, human leucocyte antigen B27 and symptom duration were tested. If relevant interactions were found, analyses were stratified. If not, variables were entered as covariates. Model fit was estimated with the quasi-likelihood under the independence model criterion (QIC): the lower the QIC, the better the data fit the model.
Additional sensitivity analyses were performed to examine the potential influence of the following characteristics on the association between DA and MRI-SI: (1) reader variation, (2) variation in MRI-SI status at baseline (positive vs negative according to ASAS definition) and (3) antitumour necrosis factor (TNF)- α therapy during the study. Data were analysed using SPSS V.20.0.
A total of 486 patients fulfilled ASAS criteria for axSpA in the DESIR cohort. Out of these patients, 167 met the inclusion criteria for this study. Overall, baseline data of included patients were similar to the entire cohort (table 1).
The distribution of SPARCC scores, stratified for males and females, is shown in online supplementary figure S1. The results show that both the frequency (55 vs 26%) and the amplitude (42 vs 23 units) of positive MRI scores are higher in males than in females.
Relationship between clinical disease activity measures and MRI-SI
Standard model (absolute scores)
A significant interaction was found between gender and several DA (ASDAS, BASDAI, pain at night and patient’s global assessment) with regard to the relationship with MRI-SI. No other relevant interactions were found. In males, a significant relationship between all DA, except BASDAI, and MRI was found in the standard models (table 2), with the lowest QIC values for the models with ASDAS and acute phase reactants. In contrast, in females none of the clinical measures were significantly related to MRI-SI inflammation.
Model of changes (change scores)
In males, changes in ASDAS, BASDAI and night pain between two consecutive visits were significantly associated with changes in SPARCC over the same time interval (table 3; no longitudinal data for ESR available). The model with the ASDAS had the lowest QIC. Consistent with the models with absolute scores, in females the change in SPARCC was not significantly related with the change in any of the clinical measures.
All analyses were repeated for both readers separately and showed identical results for both types of models. Analyses restricted to patients (1) with a positive MRI at baseline (see online supplementary tables S1 and S2), (2) with an ASAS classification according to the ‘imaging arm’ (see online supplementary tables S3 and S4) and (3) who did not use anti-TNF therapy (not shown) yielded similar results.
This study shows that DA is longitudinally associated with MRI-SI inflammatory lesions over time, with ASDAS as the measure most closely related to MRI: an increase of one unit in ASDAS coincides with an increase of 2.8 units in SPARCC score. Further, this association is specific for males.
Among the investigated measures, ASDAS and pain at night were significantly related to MRI-SI inflammatory lesions in both types of models, thus fulfilling a requirement for a longitudinal interpretation.
Unlike ASDAS, BASDAI was not significantly associated with MRI-SI in the standard model but were associated in the model of changes. Most likely this means that BASDAI, which is fully patient-reported, is not only reflecting inflammation (such as in SI joints) but also patient-specific factors independent of inflammation, while ASDAS is likely a better reflection of inflammation than BASDAI, as has been shown by previous cross-sectional data too.4 ,11–14
These longitudinal observations couple typical imaging findings such as bone marrow oedema in the SI joints to clinical signs and symptoms (DA) in patients with axSpA. But the question of whether or not to use of MRI-SI in monitoring patients with axSpA has not been solved yet and requires an in-depth analysis of MRI-SI as a biomarker.15 Several studies have partially assessed this, but no study has evaluated both of these in the same cohort.11 ,16–18
The gender difference found in this study is remarkable and has not been addressed previously in detail. Our recent study has shown that the association between ASDAS and radiographic progression is stronger in males than in females.19 In that AS study, it could not be entirely excluded that a low female patient number has precluded a robust conclusion. But in DESIR, gender distribution is equal. The question is what the gender disparity here means: many will claim that in females a positive MRI is infrequent and of lower intensity. In this regard, we cannot entirely exclude that a true relationship does exist in females but that we have not been able to demonstrate it because of statistical reasons. Others argue that many females classified as axial SpA do not truly have axial SpA and that MRI in females often is false positive. While misclassification in females cannot be completely ruled out here, it cannot be an explanation for the gender disparity: Even in the imaging positives the association between DA and MRI-SI scores is gender-dependent. Differences in anti-TNF treatment between males and females can also not explain the gender disparity: analyses in those that did not use anti-TNF treatment (70% of patients) yielded similar results.
So the most likely explanation for the gender disparity is, therefore, that axSpA has a different expression in males than in females, as has been suggested previously.20 While in male patients clinical signs and symptoms coincide with MRI positivity and with subsequent structural damage,19 in female patients symptoms attributed to SpA (and measured by patient-reported outcomes) occur independently of MRI inflammation and subsequent structural damage. The reasons for such gender-related uncoupling are unknown and deserve further research.
This study has limitations. First, symptom duration was short while clinical measures and MRI-SI correlate better in patients with long disease duration.12 Second, MRI spine was not analysed here.
Pertinent strengths of this study are the longitudinal analysis, the inclusion of patients with axSpA (and not only those fulfilling the modified New York criteria) and the analytical rigour: results obtained in all patients were similar when reanalysed in patients that were imaging positive and when separate MRI reader scores instead of average reader scores were used.
In conclusion, we can state that in male patients, but not in female patients, with axSpA, clinical disease activity measured by ASDAS is longitudinally associated with MRI-SI inflammatory lesions.
This study was conducted under the umbrella of the French Society of Rheumatology and Institut National de la Santé et de la Recherche Médicale. The database management was performed within the Department of Epidemiology and Biostatistics (Professor Paul Landais, D.I.M., Nîmes, France). The authors thank the different regional participating centres: Pr Maxime Dougados (Paris—Cochin B), Pr André Kahan (Paris—Cochin A), Pr Olivier Meyer (Paris—Bichat), Pr Pierre Bourgeois (Paris—La Pitié Salpetrière), Pr Francis Berenbaum (Paris—Saint Antoine), Pr Pascal Claudepierre (Créteil), Pr Maxime Breban (Boulogne Billancourt), Dr Bernadette Saint-Marcoux (Aulnay-sous-Bois), Pr Philippe Goupille (Tours), Pr Jean-Francis Maillefert (Dijon), Dr Xavier Puéchal, Dr Emmanuel Dernis (Le Mans), Pr Daniel Wendling (Besançon), Pr Bernard Combe (Montpellier), Pr Liana Euller-Ziegler (Nice), Pr Philippe Orcel, Dr Pascal Richette (Paris—Lariboisière), Pr Pierre Lafforgue (Marseille), Dr Patrick Boumier (Amiens), Pr Jean-Michel Ristori, Pr Martin Soubrier (Clermont-Ferrand), Dr Nadia Mehsen (Bordeaux), Pr Damien Loeuille (Nancy), Pr René-Marc Flipo (Lille), Pr Alain Saraux (Brest), Pr Corinne Miceli (Le Kremlin Bicêtre), Pr Alain Cantagrel (Toulouse), Pr Olivier Vittecoq (Rouen). The authors also thank URC-CIC Paris Centre (Jean Marc Tréluyer) for the implementation, monitoring of the study.
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.
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors VN-C, SR, RL and DvdH designed the study. VN-C and SR performed the statistical analyses. RL, MD and DvdH critically interpreted the results. VNC wrote the first draft of the manuscript. All authors reviewed the draft version and approved the final manuscript.
Funding An unrestricted grant from Pfizer was allocated for the 10 years of the follow-up of the recruited patients..
Competing interests None declared.
Patient consent Obtained.
Ethics approval Local ethical committee of each of the 25 centres participating in DESIR.
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.