Abstract
Depiction of inflammatory lesions by magnetic resonance imaging (MRI) in ankylosing spondylitis (AS) is possible both by short-tau inversion recovery (STIR) imaging and by gadolinium-enhanced T1-weighted imaging with fat saturation (T1/Gd). The aim of this prospective study was to investigate whether Gd-enhanced sequences add relevant information compared to STIR imaging alone in the detection of active spinal lesions. MRI of the spine was performed in 48 patients with AS, who participated in a clinical trial of tumor necrosis factor blocking drugs, by STIR and T1/Gd at baseline and after 6 months. Images were evaluated separately for the 2 techniques by 2 readers blinded for true time sequence and treatment. The ASspiMRI-a scoring method was used, in which 23 vertebral units are graded for inflammation from 0 to 6 (total score 0 to 138). Mean scorings of both techniques within readers were in the same range (reader 1: STIR 7.8, T1/Gd 7.7; reader 2: STIR 4.4, T1/Gd 4.7). Intraclass correlation coefficients comparing STIR and T1/Gd where high for both status scores (reader 1: 0.88; reader 2: 0.90) and change scores (both readers: 0.88). Bland and Altman analysis for both sequences showed homogeneous interreader variability along the entire spectrum of scorings, for both status scores and change scores. Smallest detectable change for status scores was 6.2 for STIR and 6.7 for T1/Gd, and for change scores 6.5 and 6.3, respectively. Standardized response means were comparable for both methods (range: 0.80-1.09). In conclusion, both STIR and T1/Gd sequences measure inflammation of the spine, as well as change of inflammation, with a high level of agreement between the 2 sequences. For future clinical randomized trials with MRI of the spine as outcome measure, STIR could be considered for use as the sole imaging technique.