Background The ability of magnetic resonance imaging (MRI) to detect structural changes of the sacroiliac joint (SIJ) is subject to discussion, whereas radiography as established method provides poor reliability.

Objectives The aim of this study was to compare the ability of MRI and radiography in detecting erosions, sclerosis, pseudowidening and ankylosis of the SIJ using low dose computed tomography (ldCT) as standard of reference.

Methods We prospectively included 110 patients in the study. All patients underwent a ldCT (Somatom Definition Flash, Siemens, Germany) and an MRI (3.0T Magnetom Skyra, Siemens, Germany) including a paracoronal T1 sequence of the SI-joints. Imaging data were compared with X-ray of the pelvis. Two readers independently scored the datasets for the presence of erosions, sclerosis and changes of the joint space in a blinded fashion using a comprehensive scoring system. Contingency table analysis was performed using ldCT as standard of reference. Inter-rater reliability was calculated using Cohen's kappa. The agreement of all modalities with the expert rheumatologists diagnosis was calculated for three diagnostic categories (axSpA, mechanical back pain, normal).

Results We found following sensitivity and specificity for the detection of structural lesions on patient's level: erosions (0.42/0.86 for radiography, 0.79/0.93 for MRI), sclerosis (0.7/0.9 radiography, 0.3/0.96 MRI) and joint space alterations (0.41/0.95 radiography, 0.76/0.99 MRI), respectively. Inter-rater reliability was fair for radiography (0.32 to 0.4) and substantial for MRI (0.35 to 0.62) and ldCT (0.5 to 0.7) with the poorest reliability for sclerosis each. Diagnostic accuracy ranged from 0.41 to 0.52 for radiography, 0.62 to 0.72 for MRI and 0.63 to 0.75 for ldCT.

Conclusions MRI is superior to radiography in detecting erosions and alterations of the joint space, however, inferior in detecting sclerosis. Most reliable diagnostic parameters amongst all imaging modalities were erosions and joint space alterations. Cross-sectional imaging modalities, including MRI, provide a superior diagnostic accuracy compared to conventional radiography.

Acknowledgement The authors thank the Assessment of Spondyloarthritis international Society (ASAS) for supporting the study with a research grant.

Disclosure of Interest None declared