Abstract
Objective Radiography is still used worldwide for the detection of sacroiliitis in juvenile spondyloarthritis (JSpA), despite its low sensitivity and reliability. We aimed to define unequivocal evidence of sacroiliitis on pelvic radiography in skeletally immature youth for use in classification criteria when magnetic resonance imaging (MRI) is unavailable.
Methods Subjects were a retrospective cohort of juvenile patients with spondyloarthritis with a radiograph and MRI as part of a diagnostic evaluation for axial disease. Six musculoskeletal imaging experts underwent an iterative consensus process to define unequivocal sacroiliitis on radiography in skeletally immature youth. Radiographs were graded using the modified New York (mNY) criteria and the unequivocal sacroiliitis criteria. Interrater agreement was assessed with the Fleiss 
statistic. Specificity, area under the receiver operator characteristic curve (AUROC), and sensitivity of the 2 measures were tested using 2 MRI reference standards.
Results A total of 112 subjects, with a median age of 14.9 (range 6.7-20.1) years, were included. The Fleiss 
was fair for the mNY criteria (0.54, 95% CI 0.42-0.67) and the unequivocal sacroiliitis criteria (0.58, 95% CI 0.46-0.69). The unequivocal sacroiliitis criteria achieved > 90% specificity using both MRI reference standards. Sensitivity (59.26 and 57.14 vs 44.83 and 43.33) and AUROC (0.76 and 0.76 vs 0.71 and 0.71) were higher, for both reference standards, for the unequivocal sacroiliitis in youth definition than for the mNY criteria, respectively.
Conclusion In this study, we propose the first consensus-derived definition to our knowledge of unequivocal sacroiliitis by radiography in skeletally immature youth. This definition achieved excellent specificity and had higher AUROC and sensitivity values than the mNY criteria using both MRI reference standards. This definition has applicability to the JSpA axial disease classification imaging criterion when MRI is unavailable.
Juvenile spondyloarthritis (JSpA) is a heterogeneous disease that accounts for approximately 15% to 20% of juvenile arthritis. Up to 20% of children and adolescents have sacroiliitis detectable by magnetic resonance imaging (MRI) within 6 months of diagnosis.1 MRI plays a key role in the evaluation of the sacroiliac joints (SIJs), as physical examination is insensitive and nonspecific in early disease1,2; limited mobility carries specificity but is a relatively rare finding in youth. Radiographs are not a sensitive or reliable imaging tool for detection of early sacroiliitis in JSpA.3 However, radiographs are still commonly performed in some areas of the world given the difficulty in accessing MRI.
MRI is the preferred imaging modality for evaluation of early inflammatory axial arthritis for several reasons. First, given the relatively short disease duration in youth at presentation, structural damage may not be present. Second, the presence of structural lesions on radiographs does not help to inform whether there is ongoing active inflammation or not, which is key information that pediatric rheumatologists need in order to make therapeutic decisions. Third, interrater reliability for the interpretation of radiographs in the juvenile population is low (
0.34).3 Lastly, cortical irregularities are common in skeletally immature children and may result in frequent false-positive findings.3
Despite the preference for MRI to evaluate early disease, radiographs are still commonly used throughout the world to evaluate axial spondyloarthritis (SpA). In some cases, radiographs are done because insurance coverage necessitates it. In other cases, particularly in less developed areas of the world, MRI is exceedingly difficult to access. Over the last several years there has been an international effort underway to develop classification criteria for axial disease in JSpA, and imaging is included among the most relevant domains in these criteria. To make the classification criteria accessible to as many parts of the world as possible, a panel of clinical SpA experts felt that radiographs needed to be considered when there was no MRI available.
As part of a larger study developing classification criteria for axial disease in youth with SpA, the objective of this project phase was to define unequivocal evidence of sacroiliitis on pelvic radiographs in skeletally immature youth. This definition will be used in the imaging domain of the classification criteria and is not intended for routine clinical use or diagnosis. Given the superiority of MRI for determination of sacroiliitis, radiographs are only intended to be leveraged for the classification criteria of axial disease in JSpA when MRI is unavailable.
METHODS
Ethics. This study was reviewed by the Children’s Hospital of Philadelphia Institutional Review Board (IRB). The IRB determined that the procedures met the exemption criteria per the Code of Federal Regulations (CFR) 45 CFR 46.104(d)(4)(iii) exemption (IRB 19-016078).
Subjects. Subjects were a convenience cross-sectional cohort of youth with SpA and symptom onset prior to 18 years of age. All children fulfilled the provisional Pediatric Rheumatology International Trials Organization criteria for enthesitis/spondylitis–related juvenile idiopathic arthritis4 or had a rheumatologist diagnosis of JSpA. All subjects had a radiograph and an MRI as part of a diagnostic evaluation for axial disease. Subjects were from 5 institutions located in North America, Europe, and Asia.
Generation of unequivocal sacroiliitis criteria. A team of 6 internationally recognized experts in imaging of the SIJ comprised the central imaging team. The central imaging team underwent an iterative consensus process through teleconferencing and subsequent electronic correspondence to define unequivocal evidence of sacroiliitis on pelvic radiography in skeletally immature children. Consensus among all experts was achieved on the final wording of the following criteria: “Unequivocal lesion (erosion, sclerosis, or ankylosis [partial or complete]) that must include at least 1 iliac bone. When sclerosis is present in isolation, if measurable, it should extend ≥ 5 mm from the joint surface. The decision may be influenced by the presence of other lesions, which in themselves do not suffice to meet the criteria.”
Central imaging evaluation. Available imaging was reviewed by the central imaging team blinded to clinical details using an online electronic case report form (eCRF; www.carearthritis.com). Radiographs were evaluated prior to MRI. Experts first graded according to the modified New York (mNY) radiologic criteria.5 Using these criteria, SIJ abnormalities were graded from 0 to 4, with grade 0 being normal and grade 4 demonstrating total ankylosis of the joint. According to the mNY criteria, sacroiliitis on radiographs is defined as bilateral grade 2 or higher or unilateral grade 3 or 4. The presence of erosion, sclerosis, ankylosis, joint space narrowing, and joint space widening was also assessed as part of the scoring module. Higher grades are typically associated with higher confidence that the changes seen reflect sacroiliitis. Radiographs were reassessed, blinded to the mNY assessment, for the presence of unequivocal sacroiliitis.
All MRI ratings were completed using an online eCRF (www.carearthritis.com). Digital Imaging and Communications in Medicine MRI files were transferred using a secure file-sharing platform. All MRI studies included views of the SIJs with T2/short-tau inversion recovery and/or T1 sequences with a small or large field of view (FOV) in the coronal, semicoronal, or axial plane(s). The eCRF comprised 2 sections. In the first section, raters recorded the presence/absence of each type of structural lesion in the iliac and sacral portions of each SIJ. For each lesion, a link was provided to a consensus-based reference image with the wording of the lesion definition. Erosion and fat metaplasia in an erosion cavity (ie, backfill) were defined according to the Assessment of SpondyloArthritis international Society (ASAS) criteria,6 whereas fat lesion/metaplasia, sclerosis, and ankylosis were defined according to the Juvenile Idiopathic Arthritis MRI Score–Outcome Measures in Rheumatology Working Group scoring system.7,8 In the second section, raters provided a yes/no response to the following ASAS standardized global assessment question: “Are typical structural chronic lesions present in the SI joints?” Semicoronal slices through the cartilaginous part of the joint on T1-weighted sequences were required for the detailed scoring of structural lesions portion of the eCRF (section 1), and T1-weighted sequences with a small or large FOV sequence in the coronal, semicoronal, or axial plane were necessary for the global evaluation of structural lesions portion of the eCRF (section 2).
Approximately 18 months following the completion of the original case assessments, the central imaging team reevaluated whether the radiographs met the unequivocal sacroiliitis criteria (yes/no).
Analysis. Interrater agreement for the mNY and unequivocal sacroiliitis criteria was assessed using the Fleiss 
statistic, with agreement interpreted as poor (≤ 0.40), fair (0.41-0.59), good (0.60-0.74), and excellent (≥ 0.75).
The specificity, sensitivity, and area under the receiver operating characteristic curve (AUROC) of the unequivocal sacroiliitis definition for skeletally immature youth were compared to those of the mNY criteria used in adults. We used 2 MRI reference standards: a central imaging majority (≥ 4/6 raters) designation of (1) high confidence (≥ +3 on a confidence scale from −5 [“Definitely No”] to +5 [“Definitely Yes”]) in an ASAS global structural lesion assessment, “Are typical structural chronic lesions present in the SI joints?”, and (2) a data-driven assessment of unequivocal structural lesions typical of axial disease in JSpA6 defined as erosion in ≥ 3 quadrants, sclerosis or fat lesions in ≥ 2 SIJ quadrants, or backfill or ankylosis in ≥ 2 joint halves across all SIJ MRI slices. As this radiographic definition is intended for use in the imaging domain of axial disease in JSpA, we aimed for high specificity (> 90%) over sensitivity and AUROC.
RESULTS
Radiographs from 119 subjects were reviewed by the central imaging team, and 7 cases were excluded because of suboptimal quality. The remaining 112 subjects—60.7% males and a median age of 14.9 (range 6.7-20.1) years—met the inclusion criteria. Among these 112 subjects, 102 (91.1%) patients reported umbar, sacral, buttock, hip, and/or groin pain, with 7 (6.3%), 25 (22.3%), and 70 (62.5%) patients out of 112 reporting duration of pain as < 6 weeks, 6 to < 12 weeks, or ≥ 12 weeks, respectively. In total, 65 (58%) patients had morning stiffness (any duration; 9 patients had an unknown morning stiffness status), and 62 (55.4%) patients were HLA-B27 positive (105 patients had HLA-B27 typing available). In total, 55/224 (24.6%) SIJs had detectable erosions (≥ 2 raters); of these 55 SIJs, 8 (14.5%), 0 (0%), and 2 (3.6%) had erosions on radiographs in the ilium, sacrum, or both, respectively, as assessed by ≥ 4 raters. In total, 53/224 (23.7%) SIJs had detectable sclerosis and 3/224 (1.3%) had ankylosis (≥ 2 raters).
After multiple iterations over email and discussion through teleconferencing, the consensus definition of unequivocal sacroiliitis by radiograph in skeletally immature children and adolescents was deemed as follows: “Unequivocal lesion (erosion, sclerosis, or ankylosis [partial or complete]) that must include at least 1 iliac bone. When sclerosis is present in isolation, if measurable, it should extend ≥ 5 mm from the joint surface. The decision may be influenced by the presence of other lesions, which in themselves do not suffice to meet the criterion.” Examples of radiographs meeting this consensus definition are shown in the Figure.
Examples of radiographs with unequivocal sacroiliitis in skeletally immature youth. (A) Unequivocal sacroiliitis is indicated by definite erosions of both iliac bones (solid arrows) and definite iliac sclerosis (dashed arrow). (B) Unequivocal sacroiliitis is indicated by definite erosions of both iliac bones (solid arrows).
All joints (N = 224) were evaluated using mNY criteria grading. With ≥ 2 raters in agreement, there were 51 joints rated as grade 2 or higher, 37 rated as grade 3 or higher, and 2 rated as grade 4. For cases with ≥ 4 raters in agreement, there were 33 joints rated as grade 2 or higher, 13 rated as grade 3 or higher, and 0 rated as grade 4. In total, 15/112 (13.4%) patients fulfilled the definition for radiographic sacroiliitis according to the mNY criteria (ie, unilateral grade 3 or 4 or bilateral grade ≥ 2) by at least 4 raters (Table 1), and a majority decision was not reached on 3 cases. The Fleiss 
statistic for the mNY criteria for sacroiliitis (yes/no) was fair at 0.54 (95% CI 0.42-0.67).
Frequency of mNY criteria and unequivocal sacroiliitis criteria on radiograph.
Out of 112 radiographs, 20 (17.9%) were rated as unequivocal sacroiliitis by a majority (≥ 4) of central raters, and a majority decision was not reached on 5 (4.5%) radiographs. Of the 20 radiographs rated as unequivocal sacroiliitis by a majority, 31/40 (77.5%) SIJs were rated as mNY grade 2 or higher, 13/40 (32.5%) were rated as mNY grade 3 or higher, and 0/40 (0%) were rated as mNY grade 4 by at least 4 raters. Of the 87 radiographs rated as not unequivocal sacroiliitis by a majority of raters, 5/174 (2.9%) SIJs were assigned a grade of 1 or higher by at least 4 raters (range across raters: grades 0-3). In the 5 cases where a majority was not reached regarding unequivocal sacroiliitis, 5/10 (50%) SIJs received a grade of 1 or higher and 2/10 (20%) SIJs were graded as 2 or higher from at least 4 raters (range across raters: grades 0-3). The Fleiss 
statistic was fair at 0.58 (95% CI 0.46-0.69).
Radiographs from 95 subjects had the required MRI sequences available for the global assessment of structural lesions, and 92 subjects had an MRI with the necessary sequences available for the detailed JSpA quantitative assessment reference standard. The specificity, sensitivity, and AUROC of the mNY criteria and unequivocal sacroiliitis in youth on radiograph with the 2 MRI reference standards are shown in Table 2. Of the 95 patients with an MRI evaluable by the global assessment reference standard, 31 (32.6%) had “typical structural chronic lesions present in the SI joints” (MRI reference standard #1) as determined by ≥ 4 raters. There were 92 patients with the necessary MRI sequences available for quantitative scoring, of which 30 (32.6%) patients had structural lesions typical of axial disease in JSpA (MRI reference standard #2) as determined by ≥ 4 raters. The unequivocal sacroiliitis criteria achieved > 90% specificity using both MRI reference standards. Sensitivity (59.26 and 57.14 vs 44.83 and 43.33) and AUROC (0.76 and 0.76 vs 0.71 and 0.71) were higher, for both reference standards, for the unequivocal sacroiliitis in youth definition than for the mNY criteria, respectively.
Test properties of the mNY criteria and unequivocal sacroiliitis on radiograph criteria with 2 MRI reference standards.
DISCUSSION
MRI has replaced radiography as the preferred modality for assessment of inflammatory sacroiliitis. However, MRI is not always accessible. Classification criteria for axial disease in JSpA are under development and imaging domains are being included. Given that these criteria may be used in populations where MRI may not be readily available, consideration of findings for definite sacroiliitis on radiography was important for inclusion in the criteria imaging domains. Surprisingly, the reliability of the mNY criteria in this exercise was higher than previously reported but still lower than that achieved for the unequivocal sacroiliitis criteria in skeletally immature youth. Both the mNY criteria and the unequivocal sacroiliitis criteria in skeletally immature youth achieved excellent specificity. However, the unequivocal sacroiliitis criteria definition in youth had higher AUROC and sensitivity values than the mNY criteria using both MRI reference standards. Although the unequivocal sacroiliitis criteria in skeletally immature youth definition represents an incremental improvement in the assessment of radiographs for this population, MRI remains the strongly preferred and superior imaging modality to assess the presence or absence of inflammatory sacroiliitis in skeletally immature youth for both clinical care and research. However, in the absence of MRI, this definition is incrementally better for youth than the mNY criteria when radiographs need to be used for the application of the classification criteria for axial disease in JSpA.
Our decision to focus on iliac erosion for the novel unequivocal sacroiliitis definition may be subject to criticism. However, although sacral lesions may be observed in the absence of iliac lesions during everyday clinical practice, for the purposes of classification criteria, iliac involvement was required because iliac erosion early in disease is more common than sacral erosion. Additionally, sacral lesions without iliac lesions were considered rare by the central raters and our data from this study supported this stance. Why early disease has a predilection for the iliac side of the SIJ is unclear, but differences in histology and/or cartilage thickness may be contributory.7,8
There are a few limitations of this study that should be acknowledged. First, all radiographs and MRI scans were obtained as part of routine clinical care and not according to a specified protocol. However, all included images had the minimal sequences necessary to perform both the radiographic assessment and the SIJ quadrant-based scoring for structural lesions. Second, radiographs and MRI scans were reviewed together—radiographs first, then MRI scans—during the quantitative scoring/global assessment portion of the project; the radiographs were then rereviewed using the unequivocal sacroiliitis criteria by the same individuals 18 months later. MRI lesion reporting may have been influenced by the radiograph findings, but this would only bias findings toward strengthening the agreement between mNY criteria and MRI findings as those data were collected from reviewers at the same time, whereas the radiograph unequivocal sacroiliitis definition was evaluated separately after significant time had elapsed. Third, this is primarily an imaging consensus study and no correlation to clinical findings or outcomes was considered. We would not expect any correlation between structural changes in the SIJs and clinical outcomes, such as disease activity, mobility, or functional outcomes. Deriving the definition of unequivocal sacroiliitis on radiographs is aimed toward incorporation into the classification criteria of axial disease in JSpA and not diagnosis or clinical use. The definition requires further validation vs MRI in larger datasets, and this could include erosion-specific MRI sequences. Fourth, since no control subjects were included in the cohort, the central raters were not blinded to clinical diagnosis of SpA, but they were blinded to the clinical details pertaining to axial disease. However, there were a sufficient number of normal SIJ MRI scans and radiographs in the cohort that the lack of control subjects probably did not substantially bias reader interpretation. Lastly, the number of radiographs with nonzero mNY scores was limited: 24 out of 120. However, this is not unexpected given that disease duration in this population is generally limited, and radiographs typically only demonstrate findings accrued after years of damage.
In conclusion, we leveraged imaging from an international sample of children with SpA and suspected axial disease in order to develop the following consensus-based definition of unequivocal sacroiliitis in skeletally immature children and adolescents by radiography: “Unequivocal lesion (erosion, sclerosis, or ankylosis [partial or complete]) that must include at least 1 iliac bone. When sclerosis is present in isolation, if measurable, it should extend ≥ 5 mm from the joint surface. The decision may be influenced by the presence of other lesions, which in themselves do not suffice to meet the criterion.” This definition had better interrater reliability than the mNY criteria in the skeletally immature population, had good performance when using MRI as the reference standard, and has applicability to the JSpA axial disease classification criteria imaging domains when MRI is unavailable.
ACKNOWLEDGMENT
The authors thank Joel Paschke (CARE Arthritis) for preparing the imaging review modules and the following collaborators for their contributions of patient clinical data and radiographs: Ozgur Kasapcopur (Istanbul University-Cerrahpasa), Mehmet Yildiz (Istanbul University-Cerrahpasa), Matthew L. Stoll (University of Alabama-Birmingham), Hemalatha Srinivasalu (NIH/NIAMS and Children’s National Hospital), Manuk Manukyan (NIH/NIAMS), Judith A. Smith (University of Wisconsin-Madison), Thomas P. Callaci (University of Wisconsin-Madison), and John W. Garrett (University of Wisconsin-Madison).
Footnotes
Support for the present manuscript was provided by the National Institutes of Health (NIH) and the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS; grant 1R01AR074098 to PFW).
PFW received grants from the Patient-Centered Outcomes Research Institute, NIH, and the Spondylitis Association of America (payment to institution). The remaining authors declare no conflicts of interest relevant to this article.
- Accepted for publication March 14, 2023.
- Copyright © 2023 by the Journal of Rheumatology
