Background Clinical trials in juvenile spondyloarthritis (JSpA) and axial disease are lacking. To assess the effectiveness of medications, we need measures to evaluate structural progression in the pediatric sacroiliac joint (SIJ).
Objectives To evaluate the reliability of the SPARCC sacroiliac joint structural score (SSS) in children with suspected or confirmed JSpA.
Methods The SSS assesses a spectrum of structural lesions of the SIJ on MRI including fat metaplasia, erosion, backfill, and ankylosis on 5 consecutive slices through the cartilaginous part of the joint. These components are scored 0–20 (backfill and ankylosis) or 0–40 (fat metaplasia, erosion). We developed a pediatric training module that included a detailed description of each SSS component plus sclerosis (0–40), scoring methodology, and numerous examples. After reviewing the module, 6 readers (mix of adult and pediatric radiologists and rheumatologists) scored 30 studies (Exercise 1). All readers then reviewed a second training module based on DICOM images and scored an additional 29 studies (Exercise 2). Inter-observer reliability was assessed using intraclass correlation (ICC).
Results The SSS had face validity and was feasible to score in the 59 pediatric cases. 35 (59%) were male and median age was 15 years (IQR 12–16). The ICCs for the SSS components from both calibration exercises are shown in the Table. In the 1st exercise, fat metaplasia and sclerosis had good reliability (≥0.4) while reliability for erosion, backfill, and ankylosis were low. During the first exercise, 31 (17%), 131 (73%), 56 (31%), 100 (56%), and 21 (12%) of the 180 ratings from 6 readers had a score>0 for fat metaplasia, erosion, backfill, sclerosis, and ankylosis, respectively. In the second exercise, the ICC for erosion, backfill, and sclerosis improved and had good reliability. Of the 174 ratings from 6 readers, 91 (52%), 18 (10%), and 49 (28%) of scores were >0 for erosion, backfill, or sclerosis in exercise 2, respectively. Reliability for fat metaplasia and ankylosis were not calculated due to low frequency of lesions (8 (5%) and 3 (2%), respectively).
Conclusions The SSS was feasible to score and had acceptable reliability for pediatric SIJ MRI evaluation. ICC improved with additional calibration exercises based on DICOM, even for readers with limited experience.
Disclosure of Interest None declared