Purpose: To compare the diagnostic accuracy of DWI and STIR sequences in Whole body (WB) MRI of SpA patients.
Materials and methods: Twenty consecutive patients with confirmed active SpA and 20 controls were investigated with identical WB MRI protocols, including DWI and STIR images. Two observers recorded 'lesions' (high signal intensity foci on STIR and high b-value DWI) in 17 anatomical areas, making a 17-point 'area score' and a 40-point 'lesion score'. ROC performance, inter-observer agreement, correlation with clinical parameters and spine and sacro-iliac joints (SIJ) MRI scores were assessed.
Results: SpA patients had significantly higher lesion scores on DWI than on STIR (p<0.025). The lesion score area under the curve was significantly higher with DWI (99.9) than with STIR (95.8, p=0.02). DWI lesion score ≥5 had both sensitivity and specificity ≥85 %. With STIR the best threshold ≥3 yielded sensitivity ≥85 % and specificity ≥60 %. DWI area score ≥3 yielded sensitivity ≥85 % and specificity ≥80 %. With STIR the best threshold ≥4 yielded sensitivity ≥70 % and specificity ≥80 %. Inter-observer agreement was strong for both sequences. In patients, the lesion score was positively correlated with ASDAS-CRP, log(CRP), and local MRI scores.
Conclusions: DWI is a promising alternative to STIR in WB MRI to detect active SpA lesions.
Key points: • DWI is a robust alternative to STIR in WBMRI in SpA. • DWI might be superior in discriminating relevant inflammatory and degenerative changes. • Positive correlations exist between WB MRI, clinical, biological, local MRI data. • Distribution and frequency of abnormal MRI findings in SpA are highlighted.
Keywords: Diffusion, magnetic resonance imaging; Magnetic resonance imaging; Spine; Spondylarthritis; Whole body imaging.