Perivascular spaces are annular gaps that exist between cerebral blood vessels and brain parenchyma¹. They are generally composed of cerebrospinal fluid (CSF), brain interstitial fluid, macrophages, and microglia among other components. In the absence of lymphatic vessels in the brain, several studies have established that metabolic wastes in the brain extracellular space are transported along these spaces propelled by cardiac pulsations^2–5. Abnormalities in the perivascular pathway have been implicated in neurodegenerative disorders such as Alzheimer’s² and syringomyelia⁶. In this study, we have obtained a high resolution 3D reconstruction of the perivascular network in the rat brain for the first time. The obtained network can be used to simulate transport along these spaces, and further the understanding of perivascular neuropathologies.

The vasculature was identified with Microfil silicone polymer (Flow Tech, Carver, MA), and the perivascular spaces using gadolinium-DTPA tagged with albumin (Gd-albumin, MW ~ 86 kDa with ~35 molecules of Gd-DTPA per albumin molecule, R. Brasch Laboratory, University of California, San Francisco, CA). 10 mL of 0.2 mM Gd-albumin mixed with Evans blue (1 mg Evans blue/ 50 mg Gd-Albumin) was infused into the lateral ventricle of Sprague-Dawley rats (n = 2) at 0.5 mL/min in-vivo, and the tracer was allowed to circulate for an additional 25 mins before cardiac perfusion with 1x PBS and formalin. Following extravasation, 15 mL of low viscosity formulation of Microfil⁷ was infused trans-cardially at a rate of 1.5 mL/min. The carcass was allowed to sit overnight for the polymer to set before the head was removed for imaging. The sample was rinsed with 1x PBS every 12 hours for 24 hours to remove formalin from the extracellular space, and then transferred to Fluorinert 12 hours before imaging.

The rat head was imaged at 17.6T Bruker spectrometer in a 25 mm NMR tube using a linear birdcage RF coil. Microfil was imaged using a T2* weighted 3D gradient echo sequence with the following parameters: TR = 150 ms, TE = 12 ms, flip angle = 22 degrees, NEX = 1, field of view = 24 mm x 24 mm x 24 mm and matrix size = 512 x 512 x 512 resulting in an isotropic resolution of approximately 47 mm. Fixing the field of view and matrix size, Gd-albumin was visualized using a fat suppressed T1 weighted 3D spin echo sequence with the following parameters: TR = 300 ms, TE = 13.5287 ms and NEX = 2.

Microfil with a long T1 (~ 3 seconds at 17.6T) and short T2* (~ 8 ms at 17.6T) appears dark on both T1 and T2* weighted images, while the Gd-albumin appears bright on T1 weighted images. The acquired images were processed in FSL using the rBET⁸ plugin to extract the brain, and Fiji⁹ to perform 3D reconstruction and analysis. The vasculature was reconstructed using the procedure outlined in another paper¹⁰. The perivascular network was reconstructed by thresholding the T1 weighted images to highlight the bright regions in the image. The results were then overlaid to demonstrate the perivascular uptake of the tracer along the vessels (Figure 1).The results clearly show the perivascular uptake of Gd-albumin mostly on the ventral surface of the brain following infusion into the lateral ventricles. Such a distribution was previously observed following intrathecal injection of Gd-DTPA, however the cause was unknown¹¹. Combining the reconstructed vascular and perivascular networks using the current method with physical models may shed light into mechanisms underlying perivascular transport in normal and pathological states.