Background

Children with spondyloarthritis (SpA) and axial arthritis are at risk of progression to ankylosing spondylitis, an inflammatory disease that causes joint fusion and leads to permanent functional impairment. While there are similarities between juvenile and adult SpA, there are distinct phenotypic differences that warrant specific focus and clinical trials in juvenile disease [1,2,3,11] that have not been validated in children. We did not see correlation with disease activity measures, perhaps secondary to the relatively small sample size or the lack of validated specific measures of axial disease activity to use as reference standards. One measure specifically designed to assess clinical disease activity in juvenile SpA is the JSpADA index [15], but this is an assessment of overall SpA disease activity and axial disease activity accounts for only two of eight variables contributing to the score. No other validated juvenile arthritis disease activity tools assess axial disease activity. The SPARCC SIS did discriminate between those with and without elevated C-reactive protein. This is not entirely surprising given the results from the aforementioned adult studies [11] and the pediatric studies demonstrating increased predictive probability of sacroiliitis with elevated C-reactive protein [4]. Discrimination was not uniform across multiple clinical attributes as it did not discriminate those with and without back pain. This latter finding may be a problem unique to pediatric SpA in which back pain has been shown to be poorly correlated with the presence of axial arthritis in multiple studies [4, 16].

There are several limitations to our study that should be considered. First, the number of pediatric studies was limited, largely due to missing semi-coronal sequences which are essential for SPARCC scoring. The limited number of studies necessitated re-evaluation during the second exercise of some cases used in the first reading exercise. We do not believe using the same studies more than once impacted our results because the cases were not discussed amongst the readers after the first reading exercise and the two reading exercises were separated by approximately 15 months. The limited number of studies also means we could not assess the minimal detectable difference or change in scores; this will be a focus of future work. Second, the studies utilized in the reading and calibration exercises were primarily from only two North American hospitals. Data collection from only two hospitals frequently leads to overrepresentation of regional demographics and a cohort that may not be typical of the entire disease population; however, this limitation is mitigated by the fact that both of these institutions are large referral centers, so the studies included are likely to be representative of cases evaluated in a wide variety of geographic areas. Third, some physician and patient-reported assessments were missing, which is expected in a retrospective study. Fourth, specific clinical measures of axial disease activity were not available as they are neither validated nor routinely collected in pediatric clinical care. Although these latter two issues limited the assessment of construct and discriminative validity, they do not affect the feasibility or reliability assessment. These relatively minor limitations are to be expected in the first systematic assessment of the feasibility and reliability of a new tool. Fifth, there was no gold standard by which to assess construct validity of the tool. The closest constructs to measure clinical (not radiographic) disease activity were the juvenile SpA disease activity index and the physician global assessment of disease activity. At most, we hypothesized there would be modest correlation. The absence a current gold standard should not preclude assessment of the feasibility and reliability utility of the tool.

Conclusions

We have demonstrated the feasibility and reliability of the SPARCC sacroiliac joint inflammation scoring methodology in children with established or suspected spondyloarthritis. This scoring system is based upon dichotomous scoring of lesions on consecutive slices through the cartilaginous part of the sacroiliac joint. We have not only established feasibility but also demonstrated that inexperienced readers can have their reading calibrated using standardized definitions, DICOM reference cases and an interactive calibration module. Further work is needed to assess the responsiveness and prognostic significance of the SPARCC SIS in evaluating MRI lesions in children.