Construct validity of pixel-by-pixel DCE-MRI: Correlation with conventional MRI scores in juvenile idiopathic arthritis

Abstract

ObjectivesTo assess the capability of the pixel-by-pixel DCE-MRI time intensity curve (TIC)-shape analysis method in the evaluation of juvenile idiopathic arthritis (JIA) disease activity by correlating DCE-MRI parameters with semi-quantitative conventional-MRI scores of synovitis. MethodsClinical, laboratory, and (DCE)-MRI datasets of 85 JIA patients were prospectively obtained. TIC-shapes of each voxel were classified into one of seven predefined color-coded TIC shape categories. Spatial information on the relative amount of TIC-shapes, maximal enhancement (ME), maximal initial slope (MIS), initial area under the curve (iAUC), time-to-peak (TTP), enhancing volume (EV) was calculated of the synovial membrane. The grade of synovitis was scored on conventional MR images by two readers using the validated JAMRIS system. The Bonferroni method was used to correct for multiple testing, therefore, a P value of <0.0056 is considered significant (0.05/9=0.0056). ResultsThe semi-quantitative JAMRIS synovitis score correlated substantially with the ME, EV, and iAUC (Rs=0.658, P<0.001; Rs=0.618, P<0.001; Rs=0.639, P<0.001), and moderately with MIS (Rs=0.453, P<0.001). A poor correlation was observed between the relative number of TIC-shapes 2–5 and the JAMRIS synovitis score (Rs=0.209, P=0.054; Rs=0.328, P=0.002; Rs=0.241, P=0.023; Rs=−0.241, P=0.026). ConclusionIn this explorative study, both TIC shape and semi-quantitative DCE-MRI analysis methods showed moderate to substantial correlations with conventional MRI scores of disease activity, indicating that this methods are feasible. Further research is warranted whether DCE-MRI holds potential to become an objective and quantitative method for the evaluation of disease activity in JIA.