Abstract Text: Introduction The active transport of cerebrospinal fluid (CSF) in the brain may be important for waste clearance and a site of dysfunction following traumatic brain injury(1–3). Pre-clinical MRI studies in rodents have been used to better understand the basic physiologic aspects of this system using both contrast (4,5) and non-contrast methods, including by controlling CSF transport states using isoflurane and dexmedetomidine(6,7). While these studies have begun to elucidate fundamental aspects of CSF transport, considerable gaps in knowledge remain about basic physiology of CSF transport and translational MRI markers for this phenomenon are quite nascent. The objective of the present study was to identify quantitative MRI techniques able to detect CSF transport enhancement in anesthetized rats. Methods A total of 16 rats were imaged in this study and each rat was imaged twice – 1) CSF enhancement paradigm with the scan battery performed under isoflurane anesthesia and then again under dexmedetomidine for direct comparison of brain changes due to CSF enhancement within the same scan session and 2) control paradigm in which the scan batter was performed twice sequentially without changing the anesthesia. Acquisition: A Bruker 7T pre-clinical MRI scanner was used to collect whole brain multi-echo T2 MRI scans. Single slice imaging within the cortex included time-of-flight (TOF) angiography and multi-gradient echo (MGE) MRI to generate vascular maps of the penetrating arteriole and venules(8). In the same cortical plane, diffusion MRI was performed using 2D (n=8) or 3D (n=8) EPI with b=400,600 and 800 s/mm2 with 32 directions each. MRI Processing: For diffusion MRI, fitting of the diffusion tensor for calculation of Trace and also dual-tensor fitting for calculation of CSF signal fraction was performed using TORTOISE (9). For tensor-based morphometry (TBM) to assess morphometric alterations, ANTs Affine and SyN registration(10,11) was used to register within-session brain volumes for the generation of the TBM LogJ maps(12) – which indicate local volume changes between anesthesia conditions. For T2 mapping, an exponential fit was performed at each voxel location in the brain of the multi-echo data using custom Matlab tools. Voxelwise analysis: ANTs tools were also used to register all DTI, T2 and LogJ maps into a common space and ratiometric difference maps between within-session outcomes were evaluated by group. Results TBM showed striking local volume increases in the CSF spaces for the CSF enhanced group and T2 differences were greatest in these regions as well, especially at the boundaries between CSF and the brain. There were no changes in parenchymal T2, although LogJ increases and decreases were evident. The most prominent observation in the diffusion cortical slice mas was the loss of midline diffusivity when anesthesia was switched from isoflurane to dexmedetomidine and ROI analyses in these maps revealed a statistically significant (p=0.02) decrease CSF signal fraction for the enhanced CSF condition compared with controls, but no significant difference for Trace. Discussion Several key findings in the present study advance MRI measurements of CSF transport in the brain. Voxelwise mapping of LogJ values are consistent with segmentation-based findings(6) and with voxel-based morphometry analysis (7) although appear to be more expansive in mapping regions with increased volume and resemble the maps generated from contrast-based MRI methods. This improvement is likely from the direct nature of the TBM methods to characterize non-linear deformations on a voxelwise basis. T2 values were notably increased in ventricular regions of known CSF flow and completely unchanged in the brain itself, although the source for increased T2 is not immediately clear. If this pattern is associated with CSF transport alterations however, it could be a potentially useful non-contrast method to assess the health of this system. Diffusion MRI is an attractive technique for characterizing CSF transport, but it is quite challenging to sensitize the diffusion experiment to the assumed flow rates for CSF transport. The decrease in free water volume fraction in the cortical parenchyma with dexmedetomidine was somewhat unexpected given the increase in extracellular space that has been observed with activation of this system, but additional studies will be necessary to know the physiologic source for the volume fraction finding.
Keywords: Diffusion MRI, Relaxometry, Morphometry, Glymphatics, Rat model, Cerebrospinal Fluid
