THU0240 The size and frequency of bone marrow oedema on sacroiliac joint mri differs in a cluster-wise comparison of patients with findings suggestive of axial spondyloarthritis

Background Multiple correspondence analysis (MCA) is a statistical method that allows for the translation of data from multiple categorical variables into single coordinates per individual. When combined with cluster analysis, individuals that share phenotypic characteristics can be grouped together. The Assessment of Axial Spondyloarthritis international Society (ASAS) published classification criteria for axial spondyloarthritis (axSpA) in 20091 that included active sacroiliitis on magnet resonance imaging (MRI) of the sacroiliac joints (SIJ) defined as bone marrow oedema (BME) present in either as 2 lesions on one slice, or 1 lesion on 2 consecutive slices.2 It has been shown that extent of BME is associated with risk of radiographic progression.3 Objectives To identify phenotypes in low back pain (LBP) patients recruited from primary care with findings suggestive of early axSpA and to assess differences in size and frequency of BME lesions on SIJ MRI. Methods Age, gender and the ASAS SpA features (SIJ BME on MRI, HLA-B27, inflammatory back pain, arthritis, heel enthesitis, uveitis, psoriasis, inflammatory bowel disease, good response to NSAIDs, family history of SpA, but not dactylitis (not observed) or radiographic sacroiliitis) of 134 LBP patients who either met of were one feature short of meeting the ASAS classification criteria for axSpA were analysed by MCA and subsequent k-means cluster analysis in order to identify various clinical phenotypes. The above listed SpA features, and if present, the size and frequency of SIJ BME as defined by the Aarhus scoring module,4 were compared across clusters. Results MCA and cluster analysis revealed 3 clusters. Cluster 1 was predominantly HLA-B27 positive (96.7%) with SIJ BME in half of the cases. Cluster 2 and 3 had SIJ BME in all cases. Cluster 3 had fewer features suggestive of axSpA than clusters 1 and 2. There were significant differences in frequency (3.0 vs 2.0 vs 1.0) and extent of BME (4.0 vs 2.0 vs 1.0) across clusters 1, 2 and 3. Conclusions Three clusters were identified among LBP patients with findings suggestive of axSpA: 1 predominantly HLA-B27 positive cluster with SIJ BME in half of the subjects, and 2 clusters having less features suggestive of SpA and with SIJ BME in all subjects. The predominantly HLA-B27 positive cluster had more and larger BME lesions than the other 2, which may indicate individuals at risk for progression. References [1] Ann Rheum Dis2009;68(6):777–83; [2] Ann Rheum Dis2009;68(10):1520–7. [3] Arthritis Rheum2008;58(11):3413–3418. [4] ACR62(1):11–8. Disclosure of Interest None declared