Background/Purpose: Inflammatory lesions on MRI of the sacroiliac joints show volatility over a short period of 3 months in patients with recent onset axial spondyloarthritis (axSpA)¹. We investigated whether the site and/or type of lesions in the spine change over a 3-month period in patients with recent onset axSpA and possible axSpA.

Methods: 158 patients with back pain (≥3 months, ≤2 years, onset <45 years) included in the SPondyloArthritis Caught Early (SPACE)-cohort from the 5 participating centers underwent MRI of the spine (MRI-spine) at baseline and after 3 months. MRIs-spine were scored by 2 readers independently, blinded for the time sequence. Presence of corner inflammatory lesions (CIL; type A and B) and fatty lesions (FL) were scored^2,3. Scores of both time points were compared on the vertebral unit level (VU; 23 per patient) for inflammatory lesions and on the quadrant level (4 per VU) for all lesions. Quadrants were summed (14536 in total) to study (changes in) the number, site and type of lesions. All possible changes were studied for each reader separately.

Results: On the VU level, reader 1 scored no CIL at both time points in 98/158 patients (62.0%). In 16.5% of the patients the number of inflamed VUs increased over 3 months (mean 1.7 SD 1.3; range 1-4); in 21.5% the number decreased (mean -2.1 SD 1.6; range -7 to -1). The upper 9 VUs were almost never affected. On the quadrant level, no CIL and FL were scored at both time points in 68/158 (43.7%) patients. In 18 patients, lesions (in 9 patients CIL; in 6 patients FL; in 3 patients both CIL and FL) did not change over time (mean 1.9 SD 1.7; range 1-8). In 72/158 patients (45.6%) site and type of lesions changed over time; in 18/72 purely occurrence of lesions (any type) was seen (mean 1.7 SD 0.8; range 1-3) and in 19/72 purely resolving of lesions (any type) was seen (mean -2.2 SD 1.8; range -8 to -1). Within the remaining 35/72 patients, various types of changes were seen (e.g. occurrence of CIL in a quadrant and disappearance of FL in another quadrant etc; mean number of changed quadrants 6.7 SD 5.1; range 1-22) as well as lesions that remained stable. Remarkably, more FL occurred in quadrants without previous CIL (46 FL; in 23 patients) than in quadrants with previous CIL (14 FL; in 7 patients), and 118 FL (whether or not surrounded by CIL) resolved over time (in 48 patients) (table). The results of reader 2 are very similar to reader 1 (table).

Conclusion: Almost half of the patients (45.6%) showed changes in site and/or type of lesions in the spine over a 3-month period only. Noteworthy, FL occurred more frequently de novo than in quadrants with previous CIL. Furthermore, FL can resolve over time, also when surrouneded by inflammation. The value of FLs in the spine need to be re-evaluated.

References: ¹de Hooge ARD 2012;71(Suppl3):301 ²Lambert J Rheumatol 2009;36 Suppl 84:3-17 ³Østergaard J Rheumatol 2009;36 Suppl 84:18-34