Background/Purpose: Structural damage is the main disease outcome to be avoided in JIA. The gold standard to assess joint damage in JIA is X-ray, which is however limited by its late detection of erosions. Alternatively, MRI is sensitive, but distinction between normal variants and bone erosions secondary to the disease is a real challenge. We posit that only bony depressions whose size increases over time should be considered true bone erosions. To test this hypothesis, we developed software to co-register and compare baseline and follow-up MRIs. The aim of this study was to assess the performance of this novel method in detecting bone erosion progression.

Methods: 66 JIA patients with active wrist arthritis who participated in a prospective trial, were enrolled. The baseline and 1 year follow-up MRI of the wrists were assessed separately according to the OMERACT RAMRIS score. Erosive progression was defined as a change between the baseline and 1-year follow-up MRI bone erosion score of at least two points. The novel computer-assisted MRI technique allows to visualize simultaneously two consecutive MRIs and to overlap each of the carpal bones, distal radius, distal ulna and metacarpal bones. Two independent readers assigned 0 or 1 points to each bone indicating absence or presence of progression, respectively. Structural damage progression was defined as progression in at least one bone. The baseline and last X-ray available for each patient were scored according to the Sharp-Van der Heijde method with a difference of at least 2 points indicating joint damage progression. This score served as reference standard for the MRI scores. Inter-reader reliability of the new method was assessed by Cohen’s kappa. Of both MRI scores, the diagnostic properties and correlation with clinical variables were evaluated.

Results: Median disease duration at baseline was 4.9 years. The last X-ray was available in 43 patients after a median time of 4.2 years from baseline. 23 (35%) patients showed radiographic progression of joint damage, while 14 (21%) and 32 (48%) patients showed progression of bone erosions using the RAMRIS score and computer-assisted technique, respectively. Inter-reader agreement for detecting erosive progression using the novel software was substantial, with Cohen’s kappa ranging from 0.63-1 for the individual bones and equaling 0.79 for the total score. The novel method was accurate with a sensitivity of 0.83, specificity of 0.78, positive predictive value (PPV) of 0.83 and negative predictive value (NPV) of 0.78, whereas the RAMRIS bone erosion score performed poorly with a sensitivity of 0.17, specificity of 0.74, PPV of 0.44 and NPV of 0.42. As expected, the novel method did not correlate with traditional measurements of disease activity, while it was correlated with validated measurements of damage such as conventional radiography (Spearman’s rho = 0.60).

Conclusion: The novel method outperformed the RAMRIS bone erosion score and allowed a reliable and accurate evaluation of structural damage progression in JIA, with the potential to select patients who are eligible for a more aggressive treatment.

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/new-prospects-for-interpreting-bony-depressions-in-juvenile-idiopathic-arthritis-a-novel-mri-computer-assisted-technique-for-detection-of-erosive-progression/