Which scoring method depicts spinal radiographic damage in early axial spondyloarthritis best? Five-year results from the DESIR cohort

Objective. To compare the performance of different spinal radiographic damage scoring methods in patients with early axial spondyloarthritis (axSpA). Methods. Five-year spinal radiographs from the DESIR cohort were scored by three readers (averaged) for the calculation of the Stoke AS Spine Score (SASSS), modified Stoke Ankylosing Spondylitis Spine Score (mSASSS), Radiographic AS Spinal Score (RASSS), BASRI-spine and BASRI-total, and following the OMERACT filter, scores were compared according to truth, discrimination (reliability and sensitivity to change) and feasibility. The proportion of patients with a net change > smallest detectable change and > 1 was calculated. The proportion of total variance explained by the patient (true variance) was calculated for the change scores as a measure of reliability, using analysis of variance. Results. In total 699 patients were included. Five-year net changes > smallest detectable change (> 1) were: RASSS 17% (17%), mSASSS 12% (12%), BASRI-spine and BASRI-total 12% (9%), SASSS 11% (11%). The mSASSS and the RASSS performed the best in terms of capturing the signal (positive change) related to noise (negative change). The proportion of variance explained by the patient was highest for the mSASSS and RASSS (85% for both 5-year progression scores vs 50-55% for other methods). The proportion of patient variance in the thoracic segment of the RASSS was unsatisfactory (46% for progression). Conclusion. The existing scoring methods to assess spinal radiographic damage performed well in early phases of axSpA. The mSASSS and RASSS captured most change. There was no clear gain in additionally scoring the thoracic spine for the RASSS. The mSASSS remains the most sensitive and valid scoring method in axSpA, including early phases of the disease.