Assessment of MRI abnormalities of the sacroiliac joints and their ability to predict axial spondyloarthritis: a retrospective pilot study on 110 patients

Abstract

To assess sacroiliac joint (SIJ) modifications on MRI and their ability to predict axial spondyloarthritis (SpA) with the purpose of identifying parameters for future prospective studies. Retrospective study was carried out of 110 consecutive patients referred for SIJ MRI with coronal, axial short TI inversion recovery (STIR), and axial T1 sequences over 6 months. Factors associated with SpA, including MRI SIJ modifications (fat deposition, structural abnormalities on T1-weighted images, and bone marrow edema [BME] on STIR sequences) and age were explored using multivariate logistic regression. The reference diagnosis was made 1-1.5 years later based on clinical, radiological, and biological findings, according to Assessment of SpondyloArthritis International Society (ASAS) criteria. Twenty-eight patients were diagnosed with SpA (female/male: 19/9, age 41 ± 13 years). Abnormal findings were found in up to 21 % of patients without SpA (including 11 % with BME), versus 64 % of SpA patients (50 % with BME). A threshold age of 42.6 years was found to discriminate SpA patients (ROC AUC: 0.71, 95 % CI: 0.59-0.81). BME location in the sacral (OR: 7.07 [1.05, 47.6], p = 0.044) and both sacral and iliac areas (OR: 36.0 [5.61, 231], p = 0.0002), as well as age (OR: 0.95 [0.92, 0.98], p = 0.0019) were found to be independent predictors of SpA. 83.6 % of patients were effectively diagnosed using BME location and patient age in a classification and regression tree (CART) algorithm (sensitivity: 61 %, specificity: 91 %, PPV: 71 %, NPV: 87 %). The BME location combined with the patient's age (threshold 42.6 years) could help predict SpA. Further studies are required before these features can be used by radiologists to boost their confidence in reporting SIJ MRI.