The diversity of MRI lesions in the sacroiliac joints of patients with axial spondyloarthritis (axSpA) has only recently been appreciated and consistent terminology, descriptions, and definitions have not yet been internationally accepted. The ASAS MRI group has generated updated consensus lesion definitions (ASAS_MRI_defⁿ) and these now require validation to support widespread adoption for clinical practice and research. We aimed to assess the distribution by diagnosis, reliability of detection, and construct validity of active and structural lesions as defined by the ASAS-MRI group (ASAS_MRI_defⁿ) on MRI scans from the ECHOSPA cohort. Methods:

Consecutive outpatients with age <50 years and symptoms >3 months suggestive of SpA were enrolled in the French ECHOSPA cohort study. MRI scans from 412 of the 470 recruited cases were available for evaluation by 2 readers and an adjudicator. ASAS_MRI_defⁿ were recorded in an ASAS consensus-derived eCRF that comprises global assessment (active and/or structural lesion typical of axSpA present/absent) and detailed scoring of individual lesions (SPARCC SIJ inflammation, SPARCC SIJ structural). Definite lesions were defined according to confidence ≥3 (0-4 scale). Reliability of detection of lesions assessed as present/absent by global assessment was analyzed using kappa and detailed scoring of SIJ quadrants by intra-class correlation coefficient (ICC). For construct validity we calculated optimal cut-offs for bone marrow edema (BME) and erosion that defined active and structural lesion typical of axSpA, respectively. Results:

At baseline, mean age of the 412 cases with MRI scans was 39.3 years, mean duration of symptoms was 2.5 years, 41.3% were HLA-B27 positive, and 63.2% were female. Active and structural lesions typical of axSpA were present in 9.7% and 10.8%, respectively, and ASAS positive MRI in 9.3%. Subchondral BME (13.6%) and erosion (9.4%) were the most frequent active and structural lesions, respectively. Active but not structural lesions were present in 3.0% while the converse was evident in 4.0%. Both active and structural lesions were present in 6.9% while either active or structural lesions were present in 13.8%. AxSpA was diagnosed at baseline in 88.1% and all categories of active and structural lesions were higher in those with axSpA. Substantial κ values (95%CI) were evident for detection of these lesions with comparable reliability for active and structural lesions: active lesion (0.76 (0.65-0.88)), ASAS positive MRI (0.78 (0.66-0.89), structural lesion (0.76 (0.65-0.87). Detailed scoring per SIJ quadrant that reflect expert opinion as to what constitutes an active or structural lesion typical of axSpA are provided in the Table. Conclusion:

SPARCC BME score of ≥3 and Erosion Score ≥2 may optimally reflect active and structural lesions typical of axSpA, respectively. MRI lesions defined by the ASAS-MRI group can be reliably detected.

Table.