Background The structural damage of axial spondyloarthritis can be evaluated either at the spine or at the Sacroiliac Joint (SIJ) level and by using either plain X-Rays or MRI. So far, the evaluation of the changes in structural damage at the SIJ level has referred mainly to plain X-Rays and to a binary variable (e.g. fulfillment of the modified New York (mNY) criteria yes/no).
Objectives To evaluate the reliability and the sensitivity to change of different outcome measures of SIJ structural damage observed at pelvic X-Rays.
Methods Study design: Prospective longitudinal (5 years) follow-up of patients with recent onset (≤3 years) axial SpA (according to the treating rheumatologist) enrolled in the DESIR cohort. Data collected: Pelvic X-Rays at baseline and after 2 and 5 years of follow-up. Reading procedure: The films were evaluated by 3 trained readers unaware of their chronology according to the 0–4 mNY grading scale (from 0 = normal to 4 = fusion) for each SIJ (left and right). Outcome measures: One continuous variable: change in the total score (from 0 to 8) and 3 dichotomous variables: A/ Switch from non-radiographic (nr) to radiographic (r) axial SpA according to the mNY definition (e.g. at least unilateral grade III or bilateral grade II), B/ Change of at least one grade in at least one SIJ C/ Change of at least one grade in at least one SIJ and an absolute final value of the worsened joint of at least 2. Statistical analysis: a) inter-reader reliability of the changes in the outcome measures (intra-class correlation coefficient for the continuous outcome and Kappa statistics for the dichotomous outcomes) b) sensitivity to change by evaluating the % of net progressors (e.g. worsened minus improved) on both the completer population and the patients with at least one post baseline radiological evaluation using different missing data handling technics (Last Observation Carried Forward, Linear Extrapolation).
Results The number of patients with available readings from all 3 readers were: 416 (baseline and 5 years), 378 (all time points) and 557 (baseline and at least one post baseline). The results were similar whatever the technic of data missing handling.
In the completer population, (a) The inter-reader reliability was low to modest: 0.21 (0.15–0.28) for the continuous variable and 0.23 (0.10–0.41), 0.24 (0.16–0.34) and 0.23 (0.13–0.35) for the dichotomous variables A, B and C, respectively.
(b) The changes in the total continuous score (from 1.41±1.68 to 1.60±1.83) was modest but highly significant (e.g. 0. 19±0.55 p<0.0001). The percentage of net progressors are summarized in the figure. These were 5.1%, 13.6% and 10.2% for the dichotomus variables A, B and C respectively.
Conclusions These data suggest that the structural progression at the SIJ level in recent onset SpA does exist but is modest. Different definitions of changes (e.g. at least one grade in at least one SIJ) seem to be more sensitive than the conventional definition (e.g. switch from nr to r) while reliability was similar. Therefore, these definitions might be useful to evaluate the natural history of the disease.
Disclosure of Interest None declared