Whole-joint histogram analysis of different models of diffusion weighted imaging in evaluating disease activity of axial spondyloarthritis

Abstract

Objectives To assess whole-joint histogram analysis of mono-exponential and bi-exponential diffusion weighted and diffusion kurtosis imaging in evaluating disease activity of axial spondyloarthritis (axSpA). Methods A total of 82 patients with axSpA who underwent both DKI and multi b diffusion weighted imaging of the sacroiliac joints were divided into active and inactive disease groups based on clinical activity indices. Another 17 patients with nonspecific low back pain were included as a control group. The SPARCC scoring system was used to assess the level of sacroiliac joint bone marrow edema. Histogram parameters of apparent diffusion coefficient (ADC), true diffusion coefficient (D), perfusion fraction (f), pseudodiffusion coefficient (D*), mean kurtosis (MK), and mean diffusivity (MD) were calculated. Regions of interest were placed covering the entire sacroiliac joint. Receiver operating characteristic (ROC) analysis was performed to assess the diagnostic performance of imaging parameters in distinguishing different groups. Regression analysis was applied to determine the correlation between imaging parameters and clinical activity indices. Results All of the histogram parameters distinguished the active from inactive groups with a low area under the curve (AUC) (max AUCs≤0.71), while the SPARCC score failed to differentiate the two groups (p = 0.08). MD, MK, D, D*, and ADC showed good performance in differentiating active and inactive from control groups (max AUC = 0. 81 ~ 0.98). f50 differentiated the active from control groups with an AUC of 0.72, significantly lower than the maximum AUC for MD, MK, D, ADC, and SPARCC score (all p < 0.05). The max AUC of MD in differentiating inactive from control groups was significantly higher than that of D* and the SPARCC score. MD, D, D*, f, and ADC were positively correlated with BASDAI, while MK was negatively correlated with BASDAI. Only MD was positively correlated with hsCRP. Conclusions Whole-joint histogram analysis of mono-exponential, bi-exponential diffusion weighted, and diffusion kurtosis imaging showed good diagnostic performance in differentiating active and inactive axSpA from patients with non-specific back pain. All the imaging parameters were correlated with BASDAI except for SPARCC score. Only DKI-derived MD was correlated with an increase in hsCRP, suggesting its potential use as an imaging biomarker for disease activity in axSpA. Advances in knowledge 1. No significant difference was found between the three models of diffusion weighted imaging in evaluating disease activity of axial spondyloarthritis. 2. Only DKI-derived MD was correlated with an increase in hsCRP, suggesting its potential use as an imaging biomarker for disease activity in axSpA.