A psychometric evaluation of diffusion basis spectrum imaging indicates white matter inflammation in depression

Abstract An inflammatory subtype of major depressive disorder (MDD) is associated with treatment resistance pointing to an unmet need for adjunctive treatments. To evaluate treatment-related changes in brain inflammation, diffusion basis spectrum imaging (DBSI) is a promising non-radiation-based technique for longitudinal designs which has been verified with histopathology. However, its use as an endpoint in clinical trials is dependent on its individual-level reliability to robustly track changes. Here, we evaluated two DBSI runs acquired in 94 participants (including 43 participants with MDD) on the same day about 1.5 h apart to assess short-term test–retest reliability. Fiber fraction (reflecting axonal/dendrite density) and hindered fraction (reflecting edema) showed moderate to high test–retest reliability in both gray and white matter regions, whereas restricted fraction (reflecting cellularity) showed lower values in gray and white matter. Group-level reliability was similar in participants with MDD, except for lower reliability of hindered fraction in gray matter. Re-identification rates of individual brain maps were higher using voxel-level white matter signatures compared to gray matter regions of interest (ROIs) ( p <.001). Crucially, participants with MDD showed reduced fiber fraction ( t max =4.68, k =38) and elevated hindered fraction ( t max =4.74, k =32) in the cingulate bundle, consistent with increased white matter inflammation, while gray matter ROI-based classification failed to identify cases. We conclude that DBSI is a promising technique to track inflammatory signatures in MDD, particularly in white matter tracts. Since several frontal and subcortical gray matter ROIs showed insufficient reliability, their assessment would require multiple DBSI runs to provide robust estimates.