The impact of MRI slice thickness on the detection of spinal syndesmophytes in axial spondyloarthritis

Background Conventional radiographs (CR) are the gold standard for detecting syndesmophytes in radiographic axial Spondyloarthritis (r-axSpA), while the ability of magnetic resonance imaging (MRI) to detect such bony structures is questionable due to its slicing technique. We aimed to assess the ability and performance for detection of syndesmophytes on MRI using different slice thicknesses and compare them with CR in r-axSpA. Methods MRI (T1-weighted (T1W) sequences) with slice thicknesses of 1-6mm of the lower thoracic and lumbar spine were prospectively performed in patients with available CR. Each vertebral corner (VC) (anterior and posterior) from thoracic (Th11) to lumbar (L5) was assessed for presence/absence of syndesmophytes and/or fat lesions (FL, MRI only) by two experienced readers in independent MRI and CR sessions and agreement was then reached in consensus. Results A total of 1.204 VCs were assessed from 43 r-axSpA patients. Syndesmophytes were recorded in 19.3% VCs on CR and in 38.3%, 37.5%, 34.8%, 33.7%, 31.4%, 28.7% VCs on MRI slice thicknesses of 1-6mm, respectively (all p ≤ 0.001 vs. CR). Although more syndesmophytes were recorded on MRI than CR, MRI also missed 21%-31.3% syndesmophytes detected in CR. Agreement with CR was found in 72.6%, 73.8%, 75.9%, 76%, 77.3% and 78.5% on MRI slice thicknesses of 1-6mm, respectively. FL were detected in 38.2%-39.2% in slice thicknesses 1-6mm. Occurrence of FL was associated with better agreement between MRI and CR findings. Conclusion The thinner the MRI slices, the more syndesmophytes were detected compared to CR, but the best agreement with CR was found in the thicker slices. The presence of fat lesions on MRI was associated with better agreement with CR for syndesmophyte detection.