AN INTERNATIONAL MULTICENTRE STUDY OF THE DIAGNOSTIC UTILITY OF MRI IN EARLY SPONDYLOARTHRITIS: IS IT ESSENTIAL TO INCLUDE RADIOLOGISTS?

Background: There is growing recognition of the diagnostic utility of magnetic resonance imaging (MRI) for early spondyloarthritis (SpA) through assessment of the sacroiliac joints. Diagnostic conclusions from systematic evaluation by MRI require the integration of complex information from several distinct MRI sequences in the assessment of an anatomically highly complex joint so that optimal evaluation requires specific training. Despite this complexity, no prior studies have addressed the optimal selection of readers, specifically the contribution of radiologists, in the assessment of the diagnostic utility of MRI in early SpA. Objectives: To compare radiologist and rheumatologist evaluation of the diagnostic utility of MRI in early SpA. Methods: Five experienced readers (3 rheumatologists, 2 radiologists) from 3 international centres, blinded to patient and diagnosis, independently assessed T1-weighted and STIR MRI scans from the following 187 subjects: 77 patients with ankylosing spondylitis (AS) and 26 patients with mechanical causes of low back pain (mLBP) that were <45 years of age and had symptom duration < 10 years, 25 patients with inflammatory back pain (IBP) and mean (SD)(range) symptom duration of 26 (22.2)(4-96) months but who did not meet radiographic modified New York criteria for AS, and 59 healthy volunteers <45 years of age. Semi-coronal MRI slices through the synovial portion of the sacroiliac joints (SIJ) were read systematically from anterior to posterior as described in a standardized online training module developed by the Spondyloarthritis Research Consortium of Canada (SPARCC). We recorded bone marrow edema, marrow fat replacement, joint erosions, and ankylosis according to standardized definitions using an online data entry system with schematics of the SIJ divided into quadrants allowing the recording of acute and chronic lesions in each quadrant of each coronal slice. Readers also addressed the following questions: 1.This SIJ scan confirms the presence of SpA? (yes/no). 2. Your diagnostic conclusion is primarily based on abnormalities observed on the STIR, T1, or both sequences. Sensitivity, specificity, and likelihood ratios were calculated for concordant data by reader pair according to clinical diagnosis. Results: Diagnostic utility of MRI was only slightly better for concordant data from radiologist as compared to rheumatologist reader pairs. Moreover, minimal improvement in diagnostic utility was noted when radiologist data was paired with rheumatologist data. The STIR sequence was regarded as significantly more important in formulating a diagnostic conclusion by rheumatologists compared to radiologists (35% vs 7.1%, p<0.0001). Conversely, radiologists considered the T1 sequence as significantly more important than rheumatologists (27.4% vs 11.3%, p <0.0001). Diagnostic utility of MRI by reader category Category of reader pairDisease categoryNo of reader pairs*SensitivitySpecificityLR+LR- RheumatologistAS379.694.916.40.21 RadiologistsAS185.7100nc0.14 Rheumatologist/RadiologistAS682.995.319.90.18 RheumatologistIBP34894.910.00.55 RadiologistsIBP148100nc0.52 Rheumatologist/RadiologistIBP647.395.311.70.55 *Mean data for reader pairs nc not calculable. Conclusion: Despite focusing on different abnormalities on MRI to formulate diagnostic conclusions rheumatologists and radiologists achieve similar degrees of diagnostic performance. This emphasizes the relatively greater importance of standardization of reading methodology. Disclosure of Interest: none declaredCitation: Annals of the Rheumatic Diseases, volume 68, supplement 3, year 2009, page 652Session: Spondylarthropathies Clinical aspects (other than treatment) (Poster Presentations )