Background Magnetic resonance imaging (MRI) is a key tool for the assessment of inflammatory lesions used for diagnosis and the monitoring of treatment effects in patients (pts) with ankylosing spondylitis (AS).
Objectives Using data from the anti-tumor-necrosis-factor (TNF) agent golimumab (GLM) in AS study (GO-RAISE), we analysed the distribution and course of inflammatory spinal lesions in different parts of the axial skeleton in detail before and after treatment with GLM in pts with active AS.
Methods Complete MR images at baseline, 3 months (week 14 of the placebo [PBO]-controlled phase) and 2 years (end of open-label extension) of the study were available from 98 AS patients. Among all pts with inflammatory MRI activity at baseline (BL), the number (nr.) of spinal lesions at different time points was assessed. Both, general (total nr. of inflammatory lesions in the entire spine) and detailed (nr. of inflammatory lesions in the cervical [CS], thoracic [TS] and lumbar [LS] spine, in single vertebral units [VUs], in the upper and lower edges for anterior and posterior site of each vertebra, and in the zygoapophyseal joints [ZAJ] were assessed. Improvement in inflammation was defined as any decrease in the MRI score from baseline to year 2 of the study.
Results Overall, evaluable inflamed VU and ZAJ lesions were seen in 81.6% and 31.6% of pts, respectively, while 22.4% of VUs/ZAJs had inf at BL. In patients showing inflammatory activity, the mean number of lesions/patient was 6.7 for VUs and 3.6 for ZAJs, with no difference between pts randomized to GLM or PBO. Inflammation present in ZAJ without concomitant VU inflammation was present in 43/4426 (0.97%) of all VUs, and this combination occurred in 19/91 (20.88%) patients. In detail, 7.2% of VU+ZAJ lesions were found in the CS, 27.9% in the TS and 27.9% in the LS. VU inflammation was detected more frequently in the anterior (23.5%) than in the posterior (8.5%) part of the LS. This difference was not observed in the CS and TS. The most frequently inflamed region in the CS was C7/T1, in the TS it was T8/9-T11/12 and in the LS, there was a fairly evenly distribution across VUs. After 3 months of GLM treatment, the percentage of VUs/ZAJs with inflammation decreased by 2.7%/0.7% in the CS, 17.3%/3.6% in the TS and 17.6%/2.7% in the LS, while almost no change was observed in the PBO pts. The decreased VU/ZAJ involvement afforded by GLM treatment was sustained through 2 years. Similarly, after up to 2 years of GLM treatment, the mean number of lesions/pt showing inflammatory activity decreased to 2.7 for VUs and 2.0 for ZAJs.
Conclusions This analysis confirms the predominance of inflammatory spinal lesions at the lower TS and the LS. While ZAJ inflammation was evident in a substantial nr. of patients it was uncommon for it to occur in isolation in a non-inflamed VU. Inflammatory lesions of VUs and ZAJs are supplementary information to MRI scoring systems and may contribute to a better understanding of clinical trial data. The already reported significance of the lower TS in the inflammatory process in AS clearly needs further study.
Disclosure of Interest X. Baraliakos Grant/research support: Janssen R & D, LLC, K.-G. Hermann Grant/research support: Janssen R & D, LLC, S. Xu Employee of: Janssen R & D, LLC, B. Hsu Employee of: Janssen R & D, LLC, J. Braun Grant/research support: Janssen R & D, LLC