BRIEF REPORT

The time-averaged inflammatory disease activity estimatesthe progression of erosions in MRI of the sacroiliac jointsin ankylosing spondylitis

Marius C. Wick & Cecilia Grundtman &

Rüdiger J. Weiss & Johann Gruber &

Martin Kastlunger & Werner Jaschke &

Andrea S. Klauser

Received: 29 December 2011 /Revised: 2 February 2012 /Accepted: 5 March 2012 /Published online: 16 March 2012# Clinical Rheumatology 2012

Abstract A method to estimate the individual ankylosingspondylitis (AS) patient radiological progression of semi-quantitative magnetic resonance imaging (MRI) changes inthe sacroiliac joints has not been described yet, which thisstudy examines. Inflammatory disease activity and MRIs ofthe sacroiliac joints of 38 patients with recent onset estab-lished AS were analyzed at baseline and during follow-up.Sacroiliac MRIs were semi-quantitatively assessed using amodification of the “Spondylarthritis Research Consortiumof Canada” (SPARCC) method. In each patient, the annualinflammatory disease activity was estimated by the time-averaged C-reactive protein (CRP; mg/l), calculated as thearea under the curve. The mean (SD) CRP decreased from

1.3 (1.8) at baseline to 0.5 (0.6) at follow-up MRI (p<0.04),which has been performed after a mean (SD) disease course of2.8 (1.5) years. The mean (SD) annual increase (Δ) ofSPARCC score from baseline to follow-up MRI was 0.4(0.4). Baseline individual SPARCC sub-score for bonemarrowedema did not statistically significantly correlate with individ-ualΔSPARCC sub-score for erosions (p0N.S.). The individualAS patient correlation between annual time-averaged inflam-matory disease activity and each annual ΔSPARCC sub-scoreswas only statistically significant for erosions (p<0.01; r00.71). Our results show that bone marrow edema andcontrast-medium enhancement at baseline do not relate to theprogression of erosions but the calculation of the individualpatient annual time-averaged inflammatory disease activityallows to estimate the annual progression of erosions in sacro-iliac MRIs of patients with AS.

Keywords Ankylosing spondylitis . Erosions .Magneticresonance imaging . Radiology . SPARCC score

Introduction

Magnetic resonance imaging (MRI) of the sacroiliac joints inpatients suspicious for ankylosing spondylitis (AS) has yet notbeen implemented in the diagnosis classification criteria[1–5]. However, sacroiliac MRI is frequently used in dailyclinical practice to support a rapid and reliable clinical diag-nosis. The Outcome Measures in Rheumatology (OMER-ACT), in collaboration with the Assessments in AnkylosingSpondylitis (ASAS) international MRI in AS working groups,have proposed the “Spondylarthritis Research Consortium ofCanada” (SPARCC) index for assessment of sacroiliac joints

M. C. Wick (*) :M. Kastlunger :W. Jaschke :A. S. KlauserDepartment of Radiology, Innsbruck Medical University,Anichstrasse 35,6020 Innsbruck, Austriae-mail: marius.wick@i-med.ac.at

C. GrundtmanLaboratory of Autoimmunity, Division of ExperimentalPathophysiology and Immunology, Biocentre,Innsbruck Medical University,Innsbruck, Austria

R. J. WeissDepartment of Molecular Medicine and Surgery,Section of Orthopaedics and Sports Medicine,Karolinska University Hospital (Solna), Karolinska Institutet,17176 Stockholm, Sweden

J. GruberDepartment of Internal Medicine I, Innsbruck Medical University,Anichstrasse 35,6020 Innsbruck, Austria

Clin Rheumatol (2012) 31:1117–1121DOI 10.1007/s10067-012-1973-9

in AS a good method of semi-quantitative measuring sacroil-iac inflammation [6, 7]. It has previously been determined thatonly erosions on sacroiliac MRIs were the radiological diag-nostic feature unique for AS in patients with unspecific AS-suspicious sacroiliac pain [8]. While the SPARCC and otherMRI quantification methods give equal scoring value to, e.g.,bone marrow edema and erosions MRI changes in their sum-scores, it has been suggested that erosions, because of theirradiological specificity for AS, should be a major consider-ation in semi-quantitativeMRI assessment scores of sacroiliacjoints [8].

The progression of radiological assessable sacroiliac jointdamage in AS is still unpredictable, showing a slow or stablecourse in some patients and rapid progression in others, andMRI joint damage sometimes may continue to accrue despiteeffective suppression of clinical signs by treatment. In addition,inflammatory disease activity may show vast variations overtime, which cannot be seen on single values of serial measure-ments but could better be judged by using time-averagedcalculations.

Therefore, this study investigated if there is a link betweenthe individual time-averaged inflammatory disease activityand the progression of radiological changes in patients withestablished but short duration AS. Such findings could allowus better to estimate the individual effects of treatment on jointdestruction and to test whether there are differences betweeninflammatory and destructiveMRI changes in sacroiliac jointsin AS.

Material and methods

Patients We retrospectively analyzed sacroiliac MRIs in ASpatients referred to the Radiology Department for pre-diagnosis MRI and follow-up evaluation. Of note, thechoice of follow-up MRI for each patient was induced bythe referring physicians based on individual patient clinicalconsiderations. Baseline and follow-up MRIs were notperformed in all patients with AS or whose radiographswere inconclusive but only in some AS patients duringthe observation period. Thus, these patients representeda “real life” sample of patients with AS at the Rheuma-tology Department and who were referred to repeated MRIexaminations.

Eligibility for inclusion in this study required that patientshad (a) a definite diagnosis of AS [9] at follow-upMRI, (b) C-reactive protein (CRP; mg/l) values and MRIs determinedwithin 1 month after their first clinical presentation, (c) CRPat baseline and at 3, 6, 9, 12, 15, 18, 24, 30, and 36 months,and (d) complete sets of baseline and follow-up sacroiliacMRIs following a standardized protocol at least between 2and 4 years after their first sacroiliac MRI [8]. All patients inthis study were HLA-B27+ [9].

Calculations The individual patient annual inflammatory dis-ease activity was estimated by the time-averaged CRP, calcu-lated by the area under the curve (AUC) method [10]. Thechange (Δ) in SPARCC scores and SPARCC sub-scores wascalculated by subtracting baseline from follow-up values.Annual ΔSPARCC scores and ΔSPARCC sub-scores wereestimated by dividing the total values by the exact elapsedtime between baseline and follow-up MRI.

Statistical analysis All data are given as means ± standarddeviation (SD). Statistical comparisons were performed us-ing analyses of variance (ANOVA) followed by post-hocBonferroni’s test. The intra-observer reproducibility of mod-ified SPARCC sum-scores and of sub-scores was calculatedusing ANOVA to provide the intraclass correlation coeffi-cient (ICC) and kappa (k) statistics. Linear correlations arePearson product moment products. All statistical analyseswere performed using SPSS 11.5 for Windows (SPSS Inc.,Chicago, IL).

MRI and scoring of MRI lesions In each patient, MRIswere performed with a 1.5 Tesla (T) (Siemens, Magne-tom Vision, Erlangen, Germany) system using appropri-ate surface coils. Sequences of this protocol (in-housecalled “Bechterew-protocol”) have previously been described[8].

MRIs were semi-quantitatively scored chronologically by1 experienced Radiology consultant (MCW), blinded toclinical and laboratory information, after achieving highperformance intra-observer reproducibility (ICC00.83; k00.84−0.92). Scoring of the sacroiliac joints was only con-fined to the synovial portion of the joint [7]. Sacroiliac jointswere semi-quantitatively assessed for joint-space width,erosions, bone marrow edema, sub-cortical cysts, andcontrast-medium enhancement. The aggregate SPARCCmethod was slightly modified to assess all slices on whichthe sacroiliac joints within the iliac bone and within thesacrum up to the sacral foramina were assessable to scorethe presence or absence of the radiological features on anyslices [7]. The modification of the SPARCC method haspreviously been described [8]. The maximum possible mod-ified SPARCC radiological sum-score on sacroiliac MRIswas 58. Individual patients annual and total SPARCC andΔSPARCC sub-scores were calculated for each SPARCCelement.

Ethics and consent

This study was approved by the regional ethics committeeunder a general approval waiver for retrospective studies basedon data from local quality registries. Informed consent wasobtained from all patients.

1118 Clin Rheumatol (2012) 31:1117–1121

Results

Table 1 summarizes the demographic and radiological base-line and follow-up values of the 38 AS patients (25 males)included in this analysis.

The mean (SD) duration from onset of sacroiliac pain untilbaseline MRI examination was 8 (4.0) months. None of thepatients received any systemic anti-rheumatic or local anesthet-ic treatment prior to baseline MRI. None of the patients hadbiological treatment during the study period but all were onindividually different combinations of NSAIDs andDMARDs.Follow-up MRI has been performed after a mean (SD) diseasecourse of 2.8 (1.5) years.

The mean (SD) CRP decreased significantly from 1.3(1.8) at baseline MRI to 0.5 (0.6) at follow-up MRI (p<0.04), indicating overall decreasing inflammatory diseaseactivity (Table 1). The mean (SD) time-averaged inflamma-tory disease activity, calculated as the AUC of CRP, frombaseline to follow-up MRI was 906 (61) mg/l. The meanannual individual patient inflammatory disease activity de-creased significantly between the 1st and last year of follow-up (p<0.05). The mean (SD) 1st, 2nd, and 3rd-year inflam-matory disease activity was 529 (86) mg/l, 302 (73) mg/l,and 106 (29) mg/l, respectively.

Besides the significant decrease in overall inflamma-tory disease activity, the mean (SD) total SPARCC scorechanged from 2.2 (1.3) at baseline to 3.2 (2.2) at follow-upMRI (p0N.S.), corresponding to a mean (SD) annual totalΔSPARCC score of 0.4 (0.4). Both the mean SPARCCsub-scores at baseline and follow-up MRI as well as the

mean ΔSPARCC sub-scores during follow-up are shown inTable 1.

Of note, the mean (SD) baseline SPARCC sub-score forerosions was 0.6 (0.6) and significantly increased to 1.4 (1.2)at follow-up MRI (p<0.04).

When the individual patient baseline SPARCC sub-scorefor bone marrow edema and contrast-medium enhancementwere correlated to the individual ΔSPARCC sub-score forerosions, no statistically significant correlations could befound (p0N.S.).

When each patient SPARCC sub-score at baseline andfollow-up MRI was correlated to the respective CRP, nostatistically significant correlation could be found (p0N.S.).

However, when each patients annual ΔSPARCC sub-scores were correlated with his/her individual annual time-averaged inflammatory disease activity, a statistically signifi-cant correlation could only be found for the annual change oferosions (p<0.01, r00.71; Fig. 1) but not for bone marrowedema (p0N.S.), contrast-medium enhancement (p0N.S.),subcortical cysts (p0N.S.), or joint space irregularities (p0N.S.). In other words, in each patient the progression of theSPARCC sub-score for erosions significantly correlated withher/his individual time-averaged inflammatory diseaseactivity.

Discussion

Sacroiliac MRI can identify active inflammatory and chron-ic structural changes and is being increasingly used for early

Table 1 Demographic and magnetic resonance imaging (MRI) characteristics of 38 AS patients included in this study

Demographic characteristics

n 38

Age 39 (7)

CRP (mg/l) Baseline Follow-up* p§

1.3 (1.8) 0.5 (0.6) <0.04MRI characteristics

Baseline Follow-up* Δ p§

Total SPARCC MRI Score 2.2 (1.3) 3.2 (2.2) 1.0 (1.1) N.S.

JSI Score 0.2 (0.4) 0.3 (0.4) 0.1 (0.4) N.S.

BME Score 0.9 (0.8) 0.8 (0.7) −0.1 (0.6) N.S.

EROS Score 0.6 (0.6) 1.4 (1.2) 0.8 (0.9) <0.04

SCC Score 0.07 (0.3) 0.1 (0.3) 0.1 (0.3) N.S.

CM Score 0.5 (0.6) 0.6 (0.9) 0.1 (0.7) N.S.

Values are given as means (±SD). *After a mean disease course of 2.8 (1.5) years. § ANOVA followed by post-hoc Bonferroni’s test. The meanchange (Δ) in the SPARCC scores was calculated by subtracting baseline SPARCC scores values from the respective follow-up SPARCC scores.BME bone marrow edema; CM contrast-medium enhancement; CRP C-reactive protein; EROS erosions; JSI joint space irregularities; SCCsubcortical cysts.

Clin Rheumatol (2012) 31:1117–1121 1119

imaging assessment in AS patients. Bone marrow edema orinflammation can be visualized using TIRM or sequencesafter administration of contrast-medium [11], revealing acutelesions brightly in contrast to dark normal bone marrow [12].Chronic changes, such as erosions or ankylosis, can be visu-alized on high spatial resolution T1-sequences with or withoutfat saturation. However, there have been no internationalagreements to define/score active and chronic changes in thesacroiliac joints because of the difficulty to interpret thesechanges, even when utilizing experienced readers. There isno understanding as yet as to what is the relationship betweensacroiliac MRI and X-ray erosions and whether the former areindeed a forerunner of the latter. Furthermore, it is not clearhow clinically significant the development of sacroiliac MRIerosions is.

Baraliakos et al. have studied the relationship betweenchronic radiographic X-ray changes of the spine of patientswith AS and the degree of spinal inflammation, as assessed bysensitive MRI sequences, and found both a link and a disso-ciation of inflammation and radiographic damage [13]. An-other, cross-sectional, study of patients with AS described therelationship between CRP and disease activity onMRI, with amean CRP significantly lower in patients without enhance-ment or subchondral bone marrow edema compared to thoseshowing MRI changes of active inflammation [14]. Theauthors speculated that inflammation and joint damage maynot be completely co-dependent. However, it still stands toreason that inflammation and radiological damage are notdissociated from each other and that bone damage may occuronly in the presence of inflammation, but not in its absence as

some SPARCC features do also in other diseases than in AS[8].

Utilizing a slight modification of the SPARCC method tocalculate the individual patients total ΔSPARCC andΔSPARCC sub-scores and the individual patients time-averaged burden of inflammatory disease activity over time,we were able to define the interrelationship between inflam-matory disease activity and joint destruction in AS. Interest-ingly, only the correlations between destructive radiologicalfeatures, i.e. erosions, but not bonemarrow edema or contrast-medium enhancement, and the individual annual time-averaged inflammatory disease activity during the course ofthe disease was markedly statistically significant. We areaware that statistical correlation does not necessarily meandirect predication, but our results suggest that erosions arethe result of the time-averaged inflammatory disease activityand the calculation of the individual patient annual inflamma-tory disease activity allows an estimation of the progression oferosions in sacroiliac MRIs of patients with AS. Put in otherterms, in approximately 50 % of patients, the individualannual degree of the change of erosions is estimated by thetime-averaged inflammatory disease activity, calculated as theAUC of CRP. In contrast, the radiological feature of bonemarrow edema and contrast-medium enhancement at baselinehad no predictive value on our patient’s risk of developingbony erosions.

It may be of significant interest to use such calculations inprospective studies of anti-rheumatic treatments in AS. Itneeds to be mentioned, however, that our patients have notbeen treated with biological therapy that may have an evenstronger impact on the reduction of the overall inflammatoryburden, but only with different NSAIDs and DMARDs. Nev-ertheless, the potential of NSAIDs and DMARDs to reducethe overall MRI progression can be seen in our study whencomparing the amount of pre-diagnosis ΔSPARCC score (2.2units in 8 months) with the progression rate after diagnosisand follow-up MRI (1.0 units in 2.8 years). In addition, therelatively small numbers of patients might also reflect achance occurrence due to the highly significant correlationdescribed herein [13].

A limitation of this observational Radiology study was thatthe choice of MRI for each patient was based on individualpatient clinical considerations. Therefore, a possible selectionbias due to their clinical activity cannot be fully ruled out.However, maybe only in presumably “more active” patientsthe development of erosions can be studied. Moreover, it alsoneeds to bementioned that all includedAS patients in our studywere HLA-B27+, which might increase the specificity of ourfindings for definite AS but have some limiting impact on thegeneralizability of our results. Another limitation of this studyis that no control group of patients with consecutive sacroiliacMRIs was available for comparisons. In addition, clinicaldisease parameters and activity indices were not available

Fig. 1 Correlation of the individual AS patients annual time-averagedinflammatory disease activity and the individual annual ΔSPARCCsub-scores for erosions indicates a highly significant correlation(p<0.01; r00.71)

1120 Clin Rheumatol (2012) 31:1117–1121

for all of these patients and have not been included in ourcorrelation with the radiological analyses.

Conclusions

Our results show that erosions, because of their radiologicalspecificity for AS and, despite the rather crude estimates in thismodel presented herein, their approximately 50 % estimatedinterrelationship to the degree of inflammatory disease activityover time, should be a major consideration in semi-quantitativeMRI assessment scores of sacroiliac joints. In contrast, scoringof bone marrow edema and contrast-medium enhancementmight only be of description radiological interest but do notprovide value for the future development of sacroiliac erosionsin AS. Additional work to reconcile our model on largerprospective unselected patient cohorts further elucidating thisissue using the method presented herein would be of interestand warrant further research.

Disclosures None.

References

1. Baraliakos X, Landewe R, Hermann KG et al (2005) Inflammationin ankylosing spondylitis: a systematic description of the extentand frequency of acute spinal changes using magnetic resonanceimaging. Ann Rheum Dis 64(5):730–734

2. BollowM, Braun J, HammB et al (1995) Early sacroiliitis in patientswith spondyloarthropathy: evaluation with dynamic gadolinium-enhanced MR imaging. Radiology 194(2):529–536

3. Bollow M, Hermann KG, Biedermann T, Sieper J, Schontube M,Braun J (2005) Very early spondyloarthritis: where the inflammation

in the sacroiliac joints starts. Ann Rheum Dis 64(11):1644–1646

4. Bollow M, Loreck D, Banzer D et al (1999) Imaging diagnosis insuspected inflammatory rheumatoid axial skeleton diseases (sac-roilitis). Z Rheumatol 58(2):61–70

5. Braun J, Bollow M, Sieper J (1998) Radiologic diagnosis andpathology of the spondyloarthropathies. Rheum Dis Clin NorthAm 24(4):697–735

6. van der Heijde DM, Landewe RB, Hermann KG et al (2005) Appli-cation of the OMERACT filter to scoring methods for magneticresonance imaging of the sacroiliac joints and the spine. Recommen-dations for a research agenda at OMERACT 7. J Rheumatol 32(10):2042–2047

7. Maksymowych WP, Inman RD, Salonen D et al (2005) Spondy-loarthritis research Consortium of Canada magnetic resonanceimaging index for assessment of sacroiliac joint inflammation inankylosing spondylitis. Arthritis Rheum 53(5):703–709

8. Wick MC, Weiss RJ, Jaschke W, Klauser AS (2010) Erosions arethe most relevant magnetic resonance imaging features in quanti-fication of sacroiliac joints in ankylosing spondylitis. J Rheumatol37(3):622–627

9. van der Linden S, Valkenburg HA, Cats A (1984) Evaluationof diagnostic criteria for ankylosing spondylitis. A proposal formodification of the New York criteria. Arthritis Rheum 27(4):361–368

10. Matthews JN, AltmanDG, Campbell MJ, Royston P (1990) Analysisof serial measurements in medical research. BMJ 300(6719):230–235

11. Hermann KG, Althoff CE, Schneider U et al (2005) Spinal changesin patients with spondyloarthritis: comparison of MR imaging andradiographic appearances. Radiographics 25(3):559–569, discussion569–70

12. Braun J, Golder W, Bollow M, Sieper J, van der Heijde D (2002)Imaging and scoring in ankylosing spondylitis. Clin Exp Rheumatol20(6 Suppl 28):S178–S184

13. Baraliakos X, Listing J, Rudwaleit M, Sieper J, Braun J (2008) Therelationship between inflammation and new bone formation inpatients with ankylosing spondylitis. Arthritis Res Ther 10(5):R104

14. Bredella MA, Steinbach LS, Morgan S, Ward M, Davis JC (2006)MRI of the sacroiliac joints in patients with moderate to severeankylosing spondylitis. AJR Am J Roentgenol 187(6):1420–1426

Clin Rheumatol (2012) 31:1117–1121 1121