Background/Purpose: It is known that radiographic progression of sacroiliitis in axial spondyloarthritis (axSpA) is quite slow, with only few predictors of such progression identified. An analysis of data from the Assessment of SpondyloArthritis international Society (ASAS) Cohort, based on local assessment of radiographs, revealed surprisingly high rate of regression from radiographic axSpA (r-axSpA) to non-radiographic axSpA (nr-axSpA). The objective of the study was to analyze the rates and predictors for radiographic progression of sacroiliitis in patients with axSpA from the ASAS Cohort, based on the central reading of radiographs.

Methods: A total of 205 patients included in the ASAS Cohort and diagnosed with axSpA by local rheumatologists, with baseline pelvic radiographs available for central reading, were included in the current study. Among them, 106 patients had a pelvic radiographs available at follow-up (mean follow-up time 4.4±0.8 years). Images were independently assessed by 2 central readers (MP, FP), blinded for the chronology of the radiographs, according to the grading system of modified New York criteria (grade 0-4). In case of discrepancy in classification (nr-axSpA or r-axSpA), the final classification was defined by adjudicator (DP). Sacroiliitis sum score (0-8) was calculated based on scoring results of all readers as the sum of the mean scores of all readers for both joints. The primary outcome was the proportion of patients progressing from nr-axSpA to r-axSpA at follow-up. Predictors of progression were investigated in univariable and multivariable logistic regression analyses.

Results: Among 106 patients, 49 (46.2%) were classified as nr-axSpA, 57 (53.7%) as r-axSpA at baseline. The agreement between primary readers in classification (nr-axSpA or r-axSpA) was moderate to substantial (κ=0.54 – baseline, κ=0.63 – follow-up); between local and central readers – poor to moderate (κ=0.18 – baseline, κ=0.58 – follow-up). At follow-up, 8 (7.5%) patients progressed from nr-axSpA to AS, 6 (5.7%) were reclassified from AS to nr-axSpA. The sacroiliitis sum score increased in 43 (40.6%) patients, decreased in 21 (19.8%) and did not change in 42 (39.6%). Logistic regression analysis showed an association of active and chronic changes on baseline MRI, existing structural damage in sacroiliac joints at baseline and younger age with higher odds for progression from nr-axSpA to r-axSpA.

Conclusion: There was a low but still detectable progression from nr-axSpA to r-axSpA in the ASAS cohort over 4.4 years of follow-up. Active and chronic changes on MRI, initial structural damage on radiographs, and younger age at baseline were associated with higher odds for progression from nr-axSpA to r-axSpA.