Abstracts

Rationale: As the knowledge surrounding the neurobiological basis for Conversion Disorders (CDs) increases, psychogenic nonepileptic seizures (PNES) have been proposed as a network brain disorder. Patient profiles following traumatic brain injury (TBI) may depend on whether and how the limbic and motor regions that comprise the specific PNES network are affected. Recent advances in diffusion imaging data collection and analyses, known as Neurite Orientation Dispersion and Density Indices (NODDI), can extend the PNES network model by assessing white matter regions for both axonal integrity and neurite morphology. Based on prior literature, we hypothesized that differences in structure within limbic and sensorimotor PNES networks would correspond with mental health symptom severity. Methods: Fifteen participants with a history of TBI and PNES (TBI+PNES) and 24 participants with a history of TBI only (TBI) completed high angular resolution diffusion imaging on two 3T Siemens Prisma MRI scanners (UAB and Brown University). Patient profiles were indexed for mood (BDI-II), anxiety (BAI), post-traumatic stress (PCL-5), dissociative (PCL-90 PSY), and somatization (PCL-90 SOM) symptoms (for all, higher score means worsened symptoms). Voxel-wise group differences (TBI vs. TBI+PNES) in diffusion indices were assessed using AFNI's 3dLME and 3dClustSim. Diffusion indices from significant clusters were extracted using 3dROIstats compared to psychiatric and behavioral measures using Pearson’s correlation, regardless of group (TBI and TBI+PNES combined). Results: Each dataset was first corrected for motion, eddy currents and susceptibility artifacts using NIH’s TORTOISE. After preprocessing, diffusion tensor and NODDI metrics were estimated (NODDI toolbox, NITRC.org) for white matter axonal integrity [fractional anisotropy (FA) and mean diffusivity (MD)], as well as neurite dispersion [orientation dispersion index (ODI)] and density [intracellular volume fraction (ICVF), and extracellular free water (isotropic) volume fraction (V-ISO)]. After warping to MNI space was performed (3dQwarp), whole-brain cluster volume extent thresholds (corrected p<0.05) were calculated to correct for multiple comparisons using 3dClustSim, producing thresholds for FA (461 mm3), MD (899 mm3), ODI (743mm3), ICVF (1152 mm3), and V-ISO (756 mm3). The 3dLME analyses revealed that the TBI+PNES group exhibited clusters of decreased FA within the corpus callosum, superior longitudinal fasciculus, and stria terminalis; an increased V-ISO cluster within the anterior cingulum bundle; and decreased ICVF clusters within the longitudinal fasciculus and cingulum bundle. Correlation analysis revealed that decreases in FA were associated with increases (worsened symptoms) in BDI-II, BAI, PCL-5, SCL-90 PSY and SCL-90 SOM values (all rs<-0.33, ps<0.05). Likewise, decreases in ICVF values were associated with increases in BDI-II and BAI values (all rs<-0.32, ps<0.05), while increases in V-ISO values were associated with increased BAI and PCL-5 values (both rs>0.32, ps<0.05). Conclusions: The current study found relationships between structural white matter alterations of stria terminalis and cingulum bundle (pathways connecting limbic brain regions that comprise an emotion regulation network) and the superior longitudinal fasciculus (pathway connecting brain regions that comprise a sensorimotor network), and mental health symptom severity. These findings may suggest specific neuropathophysiological mechanisms linking patient profiles to network disruptions in PNES following TBI. Funding: This work was supported by the US Department of Defense (W81XH-17-0619).