Enlarged perivascular spaces are associated with decreased brain tau deposition

Abstract Aims The aim of this study was to investigate the associations of enlarged perivascular spaces (EPVS) in the basal ganglia (BG) and centrum semiovale (CSO) with beta‐amyloid (Aβ) and tau deposition in older adults with a diverse cognitive spectrum. Methods A total of 163 (68 cognitively normal and 95 cognitively impaired) older participants underwent [11C] Pittsburgh compound B and [18F] AV‐1451 PET, and MRI. EPVS in the BG and CSO and other small vessel disease markers, such as white matter hyperintensities, lacunes, and deep and lobar microbleeds, were assessed. Results Increased EPVS in the BG showed a significant association with lower cerebral tau deposition, even after controlling for other small vessel disease markers. Further exploratory analyses showed that this association was significant in cognitively impaired, Aβ‐positive, or APOE4‐positive individuals, but not significant in the cognitively normal, Aβ‐negative, or APOE4‐negative participants. In contrast to EPVS in the BG, EPVS in the CSO did not have any relationship with cerebral tau deposition. In addition, none of the two types of EPVS were associated with cerebral Aβ deposition. Conclusion Brain tau deposition appears to be reduced with increased EPVS in the BG, especially in individuals with cognitive impairment, pathological amyloid burden, or genetic Alzheimer's disease risk.


| INTRODUC TI ON
Enlarged perivascular spaces (EPVS) are fluid-filled spaces that surround the penetrating small vessels in the brain and accord with extensions of the subarachnoid space. 1 Enlarged perivascular spaces are typically observed in the basal ganglia (BG) and can also be seen in the centrum semiovale (CSO). 1 There are increasing reports that PVS are elements of the "glymphatic system," which plays an important role in the clearance of interstitial fluid (ISF) and waste from the brain through interstitial fluid (ISF)-mediated bulk flow. [2][3][4][5] Increased visibility of EPVS on brain MRI has been associated with various neurological conditions such as degenerative diseases and stroke. 6 Several previous studies have reported associations between EPVS and clinically defined Alzheimer's disease dementia or mild cognitive impairment (MCI). Chen et al. 7 reported significant associations between overall EPVS and the diagnosis of Alzheimer's disease dementia or MCI. Hansen et al. 8 reported that EPVS in the BG (BG-EPVS) were more prevalent in Alzheimer's disease dementia than in normal controls, while other studies demonstrated that the EPVS in the CSO (CSO-EPVS), rather than the BG-EPVS, were more closely associated with Alzheimer's disease dementia compared with normal control 9 or subcortical vascular cognitive impairment. 10 Nevertheless, the relationship between EPVS and the neuropathological hallmarks of Alzheimer's disease, that is, cerebral beta-amyloid (Aβ) and tau deposition, has not been clearly understood. Some neuroimaging studies reported a positive association between CSO-EPVS and Aβ deposition in individuals with various cognitive spectrums, 11,12 while others found no such association. 10,13,14 With regard to the relationship between EPVS and tau pathology, only limited and controversial information is yet available. One recent study on cognitively normal (CN) individuals showed that CSO-EPVS, but not BG-EPVS, was related to tau PET positivity, 14 while another study on cognitively diverse individuals did not find an association between EPVS in any location and CSF tau levels in individuals without dementia. 13 In this context, we aimed to investigate the association of regional EPVS, that is, CSO-EPVS and BG-EPVS, with in vivo cerebral Aβ and tau deposition measured by PET in older adults with a diverse cognitive spectrum including CN and cognitively impaired (CI) individuals (i.e., MCI and Alzheimer's disease dementia).

| Participants
A total of 163 older adults (68 CN and 95 CI [45 MCI and 50 Alzheimer's disease dementia]), who participated in the Korean Brain Aging Study for Early Diagnosis and Prediction of Alzheimer's Disease (KBASE) study, 15 were included. The CN participants had no diagnosis of MCI or dementia, and a Clinical Dementia Rating score (CDR) of 0. Participants with MCI had a global CDR of 0.5, and fulfilled the core clinical criteria for diagnosis of MCI according to the recommendations of the National Institute on Aging-Alzheimer's Association guidelines (NIA-AA). 16 Participants with Alzheimer's disease dementia had a global CDR score of 0.5 or 1 and met the criteria for dementia in accordance with the Diagnostic and Statistical Manual of Mental Disorders 4th Edition (DSM-IV-TR), and the criteria for probable Alzheimer's disease dementia in accordance with the NIA-AA. 17 Individuals with a major psychiatric illness, significant neurological or medical condition, or comorbidities that could affect mental function were excluded from the study. Details of participant recruitment and the inclusion or exclusion criteria were described in our previous report. 15 The study protocol was approved by the Institutional Review Boards of Seoul National University Hospital and SNU-SMG Boramae Center in Seoul, South Korea; the participants or their legal representatives provided written informed consent.

| Clinical assessment
All participants underwent comprehensive clinical and neuropsychological assessments by trained psychiatrists and neuropsychologists based on the KBASE assessment protocol, which incorporates the Consortium to Establish a Registry for Alzheimer's Disease (CERAD-K). 15 19,20 After assessing all relevant slices for the anatomical area, the highest number of EPVS was recorded (above the anterior commissure for BG-EPVS). 13 Due to the fact that both BG-EPVS and CSO-EPVS were not normally distributed, the degree of each EPVS was dichotomized into low (degree 0-1) and high (degree 2-4), as also used previously. 13,14,21,22 Cerebral microbleeds were defined and counted on the susceptibility-weighted images and divided into lobar or deep cerebral microbleeds based on location. 23  Atlases using SPM (IBMSPM) software in MATLAB. A gray matter mask, which is a binary probabilistic gray matter map generated by a preprocessing step using SPM8, was applied to each individual to extract the gray matter and exclude the non-gray matter portions of the atlas (i.e., white matter and CSF space). The mean regional PiB uptake values from the cerebral regions were extracted using the individual AAL116 atlas from the T1-co-registered PiB-PET images.

| Assessment of enlarged perivascular spaces and other small vessel diseases
Cerebellar gray matter was used as the reference region because of its relatively low Aβ deposition, 27  PET SUVR images were created based on the mean uptake over 80-100 min post-injection, normalized by the mean inferior cerebellar gray matter uptake, and then co-registered and resliced into structural MRIs. The Geometric Transfer Matrix approach for partial volume correction was applied based on FreeSurfer-derived ROIs from the T1-weighted images taken at the follow-up visit, including corrections for extracerebral tissue as described previously. 31,32 For the ROI analyses, the data were extracted from native space according to the method published by Baker et al. 31 We quantified [ 18 F] AV-1451 PET uptake by grouping together ROIs that corresponded to the pathological stages of tau protein tangle deposition in Alzheimer's disease, as described by Braak and Braak. 33 The regions grouped in each Braak stage region of interest have been published previously. 31,34 For the current analysis, a cerebral global tau ROI mean was used; the global ROI consists of the regions corresponding to anatomical definitions of Alzheimer's disease Braak stages I/II, III/IV, and V/VI.

| Statistical analysis
Baseline characteristics are expressed as mean ± standard deviation or median (range) for continuous variables and as numbers

| Characteristics of the participants
The demographic and clinical characteristics of the participants are presented in Table 1. High-degree BG-EPVS was more prevalent in the CI group than in the CN group, while CSO-EPVS did not differ between the two groups (Table 1 and Figure S1).

| Association of EPVS with cerebral Aβ and tau deposition
For all participants, there was a significant negative association between BG-EPVS and cerebral tau deposition (p = 0.009 for model 1), even after additionally controlling for the deep cerebral microbleeds, lobar cerebral microbleeds, number of lacunes, and white matter hyperintensities volume (p = 0.01 for model 2; Table 2 and Figure 1). By contrast, no significant association was observed between CSO-EPVS and cerebral tau deposition. Neither BG-EPVS nor CSO-EPVS had a significant association with cerebral Aβ deposition, regardless of the model ( Table 2).

| Moderation of age, gender, APOE4, cognitive status, and Aβ positivity on the relationship between BG-EPVS and cerebral tau deposition
Considering the relationship between BG-EPVS and cerebral tau deposition, which showed statistical significance, we further explored the moderating effect of age, sex, cognitive status, APOE4 carrier status, and Aβ positivity using the general linear models, including each of them × BG-EPVS interaction term. As shown in Table 3 The findings of this study on the relationship between increased BG-EPVS and lower cerebral tau deposition is a novel one, in contrast to the finding of no association between BG-EPVS and Aβ level. The contrast between tau and Aβ deposition in relation to BG-EPVS may be explained by the different clearance systems of the brain for the two proteins. Aβ is cleared from the brain via blood-brain barrier (BBB) transport, degradation, ISF bulk flow to the perivascular space, and CSF absorption. 4 In particular, BBB transport and the perivascular ISF pathway are responsible for the clearance of Alzheimer's disease-related Aβ from the brain, covering 80-85% and 15-20%, respectively. 4,35 By contrast, tau is removed mainly via degradation, the perivascular ISF pathway, and CSF absorption because it cannot be transported across the BBB. 4 The role of the glymphatic system, which includes drainage of ISF to the perivascular space, is increasingly recognized for the clearance of these proteins. 4,36 Accumulating evidence suggests that BBB impairment associated with cerebrovascular dysfunction and neuroinflammation occurs before the development of cognitive impairment and before detectable increases in Aβ and tau deposition. [37][38][39] According to the two-hit vascular hypothesis of Alzheimer's disease, damage to blood vessels is the initial insult through BBB dysfunction and diminished brain perfusion, which promote Aβ accumulation. BBB breakdown was also reported to be pronounced in APOE4 carriers even before cognitive decline. 40 When BBB permeability or transport is impaired, the compensatory role of the perivascular ISF pathway might increase with the enlargement of perivascular spaces to improve fluid flow. As BBB transport is not related to tau removal, compensatory facilitation of perivascular ISF drainage accompanying EPVS may increase tau clearance, which results in lower tau deposition levels in the brain. By contrast, given that Aβ proteins are cleared via both BBB transport and the perivascular ISF pathway, the compensatory facilitation of the perivascular ISF pathway in response to BBB damage is not likely to significantly change the overall brain Aβ clearance and eventually Aβ deposition in the brain. We observed a significant negative association between EPVS and tau deposition only in BG-EPVS, but not in CSO-EPVS. A recent neuroimaging study provides a possible clue for the explanation of our results. 41 The study showed that compromised BBB integrity, measured using dynamic contrast-enhanced-MRI, was associated with the severity of BG-EPVS but not with that of CSO-EPVS. These findings suggest that BBB dysfunction may be involved in the pathogenesis of BG-EPVS, but not CSO-EPVS.
Only a couple of studies have previously investigated the association between EPVS and cerebral tau pathology. In contrast to our results, Gertje et al. 13  Our finding on the relationship between high BG-EPVS and reduced cerebral tau burden is a novel observation. Nevertheless, some limitations of the current study need to be mentioned. First, as this was a cross-sectional study, it was not possible to infer a causal relationship between EVPS and in vivo AD pathology. Further investigations are needed to clarify the mechanism underlying the association between increased BG-EPVS and low tau deposition.
Although a compensatory increase in the perivascular ISF pathway for tau elimination in response to compromised BBB permeability might be a possible explanation for the mechanism, we did not evaluate BBB permeability. Large longitudinal studies including dynamic contrast-enhanced MRI for the measurement of BBB permeability can be helpful in overcoming the limitations of the present study.
In conclusion, our findings suggest that brain tau pathology appears to be reduced with increased BG-EPVS, especially in individuals with cognitive impairment, pathological amyloid burden, or genetic Alzheimer's disease risk. Further investigations on the mechanism underlying the relationship between high BG-EPVS and lower tau pathology is needed.

AUTH O R CO NTR I B UTI O N S
KMK and DYL contributed to the conception and design of the study.
KMK and DYL contributed to drafting the text and preparing figures. funding source had no role in the study design, data collection, data analysis, data interpretation, writing of the manuscript, or decision to submit it for publication.

CO N FLI C T O F I NTE R E S T
None.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated and analyzed during the present study are not publicly available, owing to ethics considerations and privacy restriction. Data may be available from the corresponding author once approval from the Institutional Review Board of the Seoul National University Hospital, South Korea has been sought.