Background/Purpose: There is growing awareness of the importance of fat metaplasia in the pathogenesis of SpA and its potential link with inflammation and new bone formation. recent data has demonstrated fat metaplasia at vertebral corners after resolution of inflammation in patients receiving anti-TNF therapies. There is no data that assesses this link in patients receiving non-biological (non-B) therapies. These processes can be reliably measured using STIR and T1-weighted (T1W) MRI sequences, respectively, and we aimed to compare these findings in patients with SpA according to treatment.

Methods: Two readers assessed MRI scans from 103 patients with SpA, males n = 81, mean age = 39.3 years, mean disease duration = 16.2 years, in a prospective cohort. MRI was conducted at intervals up to 4 years (mean 18.1 months) and 56 patients received anti-TNF while 47 received non-B therapies. Fat metaplasia was scored on T1W scans using a new scoring method, the CanDen FAt SpA Spine Score (FASSS) which scores six different types of fat lesions defined according to anatomical location. Lesions are recorded dichotomously (present/absent) at each vertebral endplate from C2 lower to S1 upper (scoring range per disco-vertebral unit (DVU) in C spine: 0-8, and in T and L spine: 0-18). Inflammation was scored on STIR scans using the SPARCC MRI Spine 23DVU method with lesions being scored at each DVU (scoring range per DVU 0-18). We calculated SPARCC 23DVU and FASSS change scores, responsiveness by standardized response mean (SRM), and associations by Pearson c² and regression.

Results: Change (a decrease) in SPARCC 23-DVU score from baseline was significant in anti-TNF treated patients (p = 0.01) but not non-B patients while a significant increase in FASSS was evident in both anti-TNF (p = 0.005, SRM = 0.38) and non-B (p = 0.003, SRM = 0.45) treated patients indicating new fat lesions irrespective of treatment. For patients with at least 2 year follow up, 14(56%) and 16 (61.5%) of anti-TNF and non-B treated patients, respectively, had an increase in FASSS score while 7 (28%) and 6 (23.1%), respectively, had a decrease in FASSS score indicating resolution of fat lesions in a minority of patients. However, change in FASSS and SPARCC 23-DVU were highly correlated in anti-TNF treated patients (c² = -0.58, p = 3 x 10^-6) but not in non-B treated patients. Change in SPARCC 23-DVU was also independently associated with change in FASSS only in anti-TNF treated patients (β = -0.61, p <0.0001) (adjusted for age, sex, disease duration, BASDAI, CRP).

Conclusion: Long-term follow up reveals a strong link between inflammation and fat metaplasia in patients on anti-TNF while a disconnect is evident in patients receiving non-B therapies. This may have implications for pathways leading to new bone formation.