Premium
The regional and local relationships between amyloid and MRI‐visible perivascular spaces in white matter differ according to amyloid load: An ADNI study
Author(s) -
Schwartz Daniel L,
Boespflug Erin L,
Lahna David,
Dodge Hiroko H,
Roese Natalie E,
Baker Suzanne L,
Woltjer Randall,
Jagust William J,
Kaye Jeffrey,
Silbert Lisa C
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.046186
Subject(s) - white matter , amyloid (mycology) , cerebrospinal fluid , pathology , perivascular space , neuroimaging , alzheimer's disease neuroimaging initiative , parenchyma , fluid attenuated inversion recovery , magnetic resonance imaging , medicine , nuclear medicine , alzheimer's disease , neuroscience , psychology , disease , radiology
Abstract Background The relationship between amyloid (Aβ) production and clearance is disturbed in Alzheimer’s disease (AD). Evidence suggests that a decrease in clearance efficiency from cortical gray matter (GM) is relevant in late‐onset cases, a framework supported by findings of concomitant decreases in cerebrospinal fluid Aβ. A peri/paravascular bulk flow pathway for CSF and interstitial fluid in parenchyma, which transports solutes like Aβ, was identified in mammals more than 30 years ago; perivascular spaces (PVS) in white matter (WM) are part of this clearance system, and PVS dysfunction may be both a cause and effect of Aβ buildup in GM. Method Florbetapir PET, T 1 , and FLAIR MRI datasets of 95 ADNI subjects were used. Amyloid positive subjects (Aβ+) were defined by ADNI PET thresholds. MRI‐visible PVS were segmented using an automated algorithm and each was manually inspected, resulting in 8,392 PVS; GM was divided into 34 parcels per hemisphere and underlying WM was assigned to a parcel (Figure 1A). After Muller‐Gartner partial volume correction, amyloid SUVR in each GM parcel and PVS density in its associated WM, as well as the amyloid load in juxta‐WM GM (jGM) nearest each confirmed PVS (Figure 1B), were used as variables to describe the regional and local relationships between amyloid and PVS, respectively. Result Cohort‐wide amyloid load and PVS density varied widely across brain regions (Figure 2, middle row). Regional measurements of PVS and Aβ significantly correlated in Aβ+ subjects but not Aβ‐ (Figure 2, bottom). A pattern of decreasing amyloid load 1‐4mm away from the nearest jGM voxel was found in both Aβ+ and Aβ‐; amyloid burden in jGM near PVS was higher than non PVS‐associated jGM in Aβ+ but lower in Aβ‐ (Figure 3). Conclusion The complicated relationship between WM PVS and amyloid in GM differs according to amyloid load. Regional results suggest that PVS density is higher in regions with higher amyloid load, but only in Aβ+ individuals. Locally, WM PVS can be conceptualized as the clogged drain of a catchment system in Aβ+ subjects but as a functioning clearance system in those with sub‐threshold amyloid burden.