Background High inter-reader variability of radiographic sacroiliitis assessment has been reported in a number of previous studies, suggesting its low reliability for the diagnosing and classification of axial spondyloarthritis (axSpA).
Objectives To compare the results of local versus central scoring of radiographic sacroiliitis in a large multinational cohort of patients (pts) with recently diagnosed axSpA.
Methods PROOF is a prospective observational study evaluating clinical and radiographic outcomes in axSpA pts in rheumatology clinical practice in 29 countries. Pts with axSpA fulfilling ASAS classification criteria were eligible if diagnosed ≤1 year prior to study enrolment. Radiographs of sacroiliac joints (SIJ) collected at baseline were graded according to the modified New York (mNY) criteria (0–4 for each SIJ). Pts with sacroiliitis of grade ≥2 bilaterally or grade ≥3 unilaterally were classified as ankylosing spondylitis (AS); otherwise pts were classified as non-radiographic axSpA (nr-axSpA). All available radiographs were assessed first by a local reader (LR) and then by a central reader (CR1), who was blinded to the results of the LR. In the case of a disagreement in the classification (AS or nr-axSpA), the radiograph was evaluated by the 2nd central reader (CR2), who was blinded to the previous assessments and the final classification was made based on the decision of 2 out of 3 readers.
Results Of the 2126 pts enrolled in PROOF, 1583 were included in this analysis based on evaluable radiographs of the SIJ. Based on the LR judgment, 987 pts were classified as AS and 596 as nr-axSpA, while 1158 were classified as AS and 425 as nr-axSpA according to CR1. Following CR1 assessment, 1146 (72.4%) pts retained their LR classification, while 437 (27.6%) pts were classified differently. Of the 437 pts with discrepant classification assessed by CR2, 175 (40%) retained their initial LR classification and 265 (60%) were re-classified. The agreement between the CR1 and CR2 (kappa=0.24 [95% CI: 0.17–0.32]) was lower than between LR and CR1 (kappa=0.38 [95% CI: 0.33–0.42]). Finally, 1039 pts were classified as AS and 544 as axSpA; 1321 (83.5%) pts retained their initial classification and 262 (16.5%) were re-classified (157 from nr-axSpA to AS and 105 from AS to nr-axSpA). There was a substantial agreement between local and final central classification (kappa=0.64 [95% CI: 0.60–0.68], Figure). Importantly, pts initially classified by a LR as nr-axSpA (157/596, 26.3%) had significantly higher odds (odds ratio=3.0 (95% CI: 2.3–3.9) of being re-classified compared with pts classified as AS (105/987, 10.6%).
Conclusions In the PROOF study, the agreement between local and central classification of pts with axSpA (nr-axSpA vs AS) based on the grading of SIJ radiographs by mNY criteria was reasonably good. Pts locally classified as nr-axSpA were three times more likely to be re-classified compared with AS pts, which may be related to difficulty in the assessment of less advanced structural changes.
Acknowledgements AbbVie funded the PROOF study, contributed to its design and participated in data collection, analysis and interpretation of the data, and in writing, review, and approval of the publication. Medical writing support was provided by Deepa Venkitaramani, PhD, of AbbVie.
Disclosure of Interest D. Poddubnyy Grant/research support from: AbbVie, Janssen, MSD, Novartis, Pfizer, Consultant for: AbbVie, BMS, Boehringer, MSD, Novartis, Pfizer, and UCB, Speakers bureau: AbbVie, BMS, Janssen, MSD, Novartis, Pfizer, Roche, and UCB, R. Inman Grant/research support from: AbbVie, Amgen, and Janssen, Consultant for: AbbVie, Amgen, Janssen, Lilly, Novartis, and Pfizer, J. Sieper Grant/research support from: AbbVie, Merck, and Pfizer, Consultant for: AbbVie, Janssen, Lilly, Merck, Novartis, Pfizer, and UCB, Speakers bureau: AbbVie, Janssen, Merck, Novartis, Pfizer, Roche, and UCB, H. Haibel Consultant for: Boehringer, MSD, and Novartis, Speakers bureau: AbbVie, MSD, and Pfizer., M. Hojnik Shareholder of: AbbVie, Employee of: AbbVie