Longitudinal Diffusion Tensor Imaging after a Controlled Cortical Impact Model of Traumatic Brain Injury (TBI) in Male and Female Sprague-Dawley Rats from Four Sites in the Translational Outcomes Project for Neurotrauma (TOP-NT)

STUDY PURPOSE: This study aimed to implement and test reproducibility across sites, utilize centralized data analyses, demonstrate the feasibility of conducting a multi-center preclinical research trial, and validate a set of diffusion tensor imaging (DTI) neuroimaging biomarkers useful for predicting outcomes after controlled cortical impact (CCI) traumatic brain injury (TBI) in rats. The study was conducted across four sites: University of California, Los Angeles (UCLA); Uniformed Services University of Health Sciences (USUHS)/Georgetown University (GU) Medical Center; University of Florida (UF); and Johns Hopkins University (JHU) as part of the Translational Outcomes Project in Neurotrauma (TOP-NT). DATA COLLECTED: Diffusion-weighted imaging (DWI) data were collected at 3 and 30 days after CCI or sham surgery in male and female Sprague-Dawley rats (n=16/group, injured vs. sham, equal sexes, age at injury ~2 months).
CCI was performed by extending a rigid impactor tip through a parasagittal craniectomy onto the exposed dura (Leica Impact One, 5 mm beveled-edge tip, speed 5 m/s, dwell time 200 ms). The skull coordinates were set to the left hemisphere (3.5 mm AP; 3 mm laterally, over the left dorsal hippocampus). Two levels of injury were used, with impact depths ranging from 1.2–1.7 mm (low) and 2.1–2.8 mm (high), accounting for inter-machine variability, which was tested, validated, and set at each performance site [CCI SOP #4]. All referenced SOPs are available in the Supplementary Materials.

The MRI scanners, software, and coils used were:
UCLA: Bruker 7T, ParaVision 5.1.0, 72-mm quadrature volume transmit coil, receive-only surface coil.
USUHS/GU: Bruker 7T, ParaVision 6.0.1, 86-mm quadrature volume transmit coil, receive-only 4-channel array coil.
JHU: Bruker 11.7T, ParaVision 6.0.1, 72-mm quadrature volume transmit coil, receive-only 4-channel array coil.
UF: Bruker 11.1T, ParaVision 6.0.1, in-house quadrature surface coil/transceive mode.

Animals were anesthetized with a mixture of medical air and 3% isoflurane (ISO) at a flow rate of 3 L/min. Animals then underwent a bolus intraperitoneal (I.P.) injection of Dexmedetomidine (DEX, 0.1 mg/kg). Respiration rate, peripheral oxygen saturation (SpO2), heart rate, and rectal temperature were continuously monitored throughout experiments. Anesthesia was prior to and maintained during DWI acquisition and lasted about 1 to 1.5hrs, a subcutaneous (S.C.) infusion of DEX (0.05 mg/kg/hour) was used, and flow was reduced to 0.5 L/min with the ISO concentration set to 0.5%. A few minutes before DWI, the S.C. infusion of DEX was stopped, and anesthesia for the remainder of the DWI MR session was maintained at 1 L/min MA and 1.25-1.5% ISO.

All data were acquired using standard operating procedures harmonized across sites. Harmonized DWI acquisition parameters were: 3D single-shot Spin-Echo Echo-Planar Imaging (3D SE-EPI), repetition/echo time 1000/28 ms, field of view (FOV) 24x18x12.25 mm³, matrix 96x72x49, partial-FT (96x60x49), isotropic resolution 250 µm³, 4 B0 and 42 non-collinear diffusion directions, b = 1000, 3000 s/mm², small delta 5 ms, big delta 12 ms, fat suppression, bandwidth 357 kHz, total scan time 1hr 12min. An additional acquisition of 4 B0 with reverse phase was conducted for image corrections (4 min). [MRI SOPs #15,#16,#17]

Harmonized, post-acquisition image processing was conducted. Raw data were converted to NIFTI format using BrkRaw (https://brkraw.github.io/). BET (FSL, FMRIB Software Library) and joint label fusion (ANTs, https://stnava.github.io/ANTs/) with manual correction were used for skull stripping. Eddy current distortion correction (FSL), denoising (MRtrix3, https://www.mrtrix.org/), and degibbs algorithms (MRtrix3) were applied. The dwi2tensor tool from MRtrix3 was used to fit the pre-processed DWI data and generate maps of fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). All maps were registered to a study-specific template computed using the ANTs library. Z-scores for each map were calculated for each site using the site-specific maps from sham rats. The volumes of abnormal values (P<0.001) were derived and reported using Z = -3.1 (low) and Z = 3.1 (high).
NOTE: Raw images are available for download in FITBIR, see relevant links: DOI: 10.23718/meta_study/260. DATA USAGE NOTES: