Diffusion along perivascular spaces provides evidence interlinking compromised glymphatic function with aging in Parkinson's disease

Abstract Aims The aim of the study was to evaluate the glymphatic function and its related factors in patients with Parkinson's disease (PD) and patients with PD of different ages using the diffusion tensor image analysis along the perivascular space (DTI‐ALPS) method. Methods Medical records and imaging data of 93 patients with idiopathic PD and 42 age‐ and sex‐matched healthy controls (HCs) were retrospectively reviewed and analyzed. The diffusivity along the perivascular spaces, projection fibers, and association fibers were calculated on diffusion tensor imaging (DTI) to acquire the analysis along the perivascular space (ALPS) index. Results PD patients exhibited a reduced ALPS index compared with the HCs. Negative correlations between the ALPS index and clinical information including age, age at disease onset, Parkinson's disease sleep scale 2nd version (PDSS‐2) scores, and history of diabetes mellitus were revealed in the PD group. Besides, a negative correlation between the ALPS index and the severity of motor symptoms was identified in the subgroup aged 65 and above, rather than in the younger ones. Conclusions The results demonstrate that reduced ALPS index, a potential noninvasive measure of compromised glymphatic activity, is involved in the pathophysiology of PD, especially in the aged ones and those with sleep disorders.


| INTRODUC TI ON
Parkinson's disease (PD) is the second most common neurodegenerative disease. 1 It is heterogeneous in clinical manifestations and progression, and related pathophysiology has been explored only to a limited extent. 2 Major neuropathological changes in PD patients involve aggregation and propagation of misfolded αsynuclein. 3 Age is widely accepted as a major risk of PD. 4 And it has been reported that the accumulation of α-syn in the brain is greatly related to age. 5 However, the underlying mechanism between aging and the pathogenesis and progression of PD is poorly understood. [6][7][8] The glymphatic (glial-lymphatic) pathway was first identified in rodent brains. 9 In the glymphatic system, cerebrospinal fluid (CSF) exchanges with interstitial fluid (ISF) and then drains extracellular soluble proteins and metabolites along the perivascular spaces into the cervical lymphatic vessels. [10][11][12] This process is mainly driven by arterial pulsation and mass flow facilitated by astroglial water channel aquaporin 4 (AQP4), 9,11 and is with robust activity during sleep. 9,13,14 Commonly shared by PD patients and the aged population, sleep disorders contribute to decreased glymphatic clearance.
In addition, aging has been reported to be associated with a steep fall in glymphatic flow in the brains of both rodents and humans. 15,16 Although α-synuclein mainly deposits intracellularly, samples of CSF and extracellular fluid have documented α-synuclein outside the cytosol. 17,18 Moreover, the transmission of α-synuclein involves the extracellular pathway. 19 Therefore, glymphatic dysfunction may contribute to decreased clearance of α-synuclein and provide the construct with the involvement of aging and sleep disorders in the pathogenesis and progression of PD. 20 Glymphatic magnetic resonance imaging (MRI) is the classic method for evaluating glymphatic function directly. However, its application is limited due to the invasive intrathecal injection of gadolinium and repeated MRI scans with fixed time intervals. 21 A noninvasive method, the index for diffusion tensor image analysis along the perivascular space (DTI-ALPS), has been proposed as a metric with possible sensitivity to glymphatic function. 22 DTI-ALPS method has been modified by Zhang et al. in 2021. They simplified the calculation method on only DTI, generated an ALPS index by this modified method ( Figure 1), and verified the correlation between this ALPS index and the classical glymphatic function on glymphatic MRI. 23 In recent years, the DTI-ALPS method has been widely used in studies of cerebral small vascular disease (CSVD), 24 Alzheimer's disease, 22,25 idiopathic hydrocephalus, 26 diabetes mellitus, 27 and aging. 25,28 Most studies focused on the relationship between ALPS index and cognition. ALPS index has also been revealed to be correlated with CSVD imaging markers, especially white matter hyperintensities (WMHs). Three studies used the DTI-ALPS method to explore glymphatic activity in PD patients. [29][30][31] McKnight et al. 31 reported reduced ALPS index in PD patients compared to that in patients with essential tremor. Chen et al. 29 and Ma et al. 30 found that PD patients exhibited a reduced ALPS index than the healthy controls (HCs). However, the results of the correlations between the ALPS index and its related factors in different PD cohorts were controversial. Moreover, how the ALPS index interacts with sleep disorders and how it is involved in the disease severity of PD of different ages have not been reported yet.
F I G U R E 1 Calculation of index for diffusion tensor image analysis along the perivascular space (ALPS index). ( Figure 1A). shows colorcoded fractional map exhibiting the direction of the projection fibers (blue; z-axis), association fibers (green; y-axis), and the subcortical fibers (red; x-axis). Four 5-mm-diameter regions of interest (ROIs) were placed on bilateral projection fibers (blue area) and association fibers (green area) along the body of lateral ventricle, respectively. We recorded the diffusivities in the directions of the x-axis (Dx, Figure 1B), y-axis (Dy, Figure 1C), and z-axis (Dz, Figure 1D) of ROIs on projection fibers and association fibers as Dxproj, Dyproj, Dzproj, Dxassoc, Dyassoc, Dzassoc, respectively. Then we calculated the unilateral ALPS index as [(Dxproj + Dxassoc) / (Dyproj + Dzassoc)] and obtained the average value of bilateral ALPS index The coupling of glymphatic function, sleep, age, and clearance of misfolded protein has prompted speculation that the pathophysiology of PD may relate to compromised glymphatic circulation. Therefore, the current study aimed to use the modified DTI-ALPS method to explore the glymphatic activity and its related factors, including the severity of motor and non-motor symptoms in PD patients. Age has been reported to be associated with different PD phenotypes by previous studies, 6,7 underpins which we hypothesized that glymphatic function plays different roles in the pathogenesis and progression of PD among patients of different age groups. So, associations between ALPS index and disease severity in PD patients of different age groups were also explored.
Besides, we speculated there may be a link between risk factors of cardiovascular disease (CVD) and glymphatic dysfunction in PD for reasons as follows: CVD and its risk factors are common in the aged population 32 ; arterial pulsation drives the glymphatic circulation 11 ; and although CVD and PD share concordant risk factors including hypertension and diabetes mellitus, the relationship between PD and CVD and its risk factors are still undetermined. 32 Hence, CVD risk factors and WMHs were also examined in this study.

| Participants
We retrospectively reviewed and analyzed the medical records and neuroimaging data of patients with idiopathic PD, recruited

| Magnetic resonance imaging acquisition
All MRI examinations were performed after withdrawing from anti-Parkinsonian medications for at least 12 h by using GE 3.0 T slice thickness = 2.5 mm; voxel size = 2 × 2 × 3 mm 3 ; and NEX =1.
Images of 25 different nonlinear diffusion-weighted gradient directions (b = 1000 s/mm 2 ) and non-diffusion-weighted gradient direction (b = 0) were collected, each gradient was scanned for 60 layers, and a total of 1560 files were obtained. All data were saved in DICOM format.

| Hypothesis and MRI analysis
At superior levels of the ventricular body, the deep medullary veins run perpendicularly to the ventricle wall. 35 Thus, the direction of the perivascular space (PVS) is mostly in the right-left direction (x-axis) on the axial plane, which is also vertical to the direction of the projection fibers (mostly in the z-axis) as well as the association fibers (mostly in the y-axis). 23 At moderate-to-high b-values typically used in DTI (i.e., b = 1000 s/mm 2 ), diffusivity from flowing venous blood is suppressed. Therefore, the diffusivity along the x-axis at regions with projection/association fibers will at least partly represent the diffusivity along the PVS, contributed by the glymphatic fluid transport. In the DTI-ALPS model, Taoka et al. 22 generated the ALPS index, quantified as the amount of diffusion along the x-axis over the diffusion along the y-axis and z-axis, on DTI and susceptibility-weighted imaging (SWI). Of note, the direction of medullary veins is homologous in most people, especially at the superior layer of the lateral ventricle body, 35 making it possible to measure the diffusivity without the visualization of veins on SWI ( Figure 1).
All DTI data were preprocessed as follows. The raw imaging data saved as DICOM format were first converted to NIFTI format by using the dcm2nii.exe toolkit in the MRIcroN software (https:// www.nitrc.org/proje cts/mricron). At the same time, a b-vector file and a b-value file were created, which would be used to process and calculate the DTI data later. Then the converted data were denoised, head motion-corrected, and eddy current-corrected by using Automatic Image Registration toolkits on the DTI Studio index represents a good diffusivity along PVS, indicating a good glymphatic function. We tried to co-register the diffusivity maps into a common FA map template and placed ROIs on the registered diffusivity maps, as proposed by Zhang et al .23 However, the majority of the ROIs covered the CSF area of the ventricle or the subarachnoid space on the diffusivity maps, whose diffusivities in all 3 directions were greatly impacted by those of free water. Hence, registration into a common template was abandoned. One trained neurologist blind to clinical data independently placed ROIs on each FA map and calculated ALPS index for each patient and re-placed ROIs and re-calculated ALPS index after 3 days interval for intraobserver reliability analysis.
A trained neurologist blind to the clinical information reviewed T2WI and FLAIR images. Grading of deep white matter hyperintensities (DWMH) and periventricular white matter hyperintensities (PVWMH) was each assessed on a 0-3 point scale according to Fazekas, et al. 36 ; DWMH was graded as 0 = absence, 1 = punctate foci, 2 = beginning confluence of foci, 3 = large confluent areas; PVWMH was graded as 0 = absence, 1 = cap, 2 = smooth halo, 3 = irregular and extending into the subcortical white matter. White matter hyperintensities (WMH) were a combination of DWMH and PVWMH.

| Statistical analysis
Data were analyzed with SPSS Statistics version 23.0 (IBM cooperation). Internal consistency with Cronbach's α was conducted between intra-observer analyses of ALPS index. The Kolmogorov-Smirnov test was performed to assess normality for the distribution of the continuous variables. Two-sample t-test, Mann-Whitney test, and chi-square test were conducted to examine the clinical differences of normally distributed continuous variables, non-normally distributed continuous variables, and categorical variables, between HC and PD groups, and between PD patients aged below 65 years old and PD patients aged 65 years old and above, respectively. In addition, with the ALPS index as the dependent variables, Pearson's correlation, Mann-Whitney test, and Spearman's correlation were performed to determine its correlations with the normally distributed continuous variables, dichotomous variables, non-normally distributed continuous variables, and discrete variables, respectively, within all subjects, the PD group and the PD subgroups divided by the age of 65. And then, a stepwise multivariate linear regression analysis was carried out between the ALPS index and covariates with a p-value below 0.1. A p-value less than 0.05 is considered statistically significant.

Two-sample t-test between HCs and PD. c Spearman's correlation between HCs and PD.
d Mann-Whitney test between PD group below the age of 65 and PD group aged 65 and above.
e Chi-square test between PD group below the age of 65 and PD group aged 65 and above.
f Two-sample t-test between PD group below the age of 65 and PD group aged 65 and above.  Table 1).

| DISCUSS ION
The current study exhibited reduced ALPS index in PD patients compared with the HCs and revealed inverse correlations between the ALPS index and clinical information including age at disease onset, age at MRI scanning, sleep disorders indicated by high PDSS-2 scores, and history of diabetes mellitus in the PD group. In addition, with disease duration as the covariate, subgroup analysis revealed a negative correlation between ALPS index and the severity of motor symptoms measured by MDS-UPDRS part III in the older PD group, rather than in the younger one. Using the non-invasive DTI-ALPS method to measure glymphatic activity, this study contributed to accumulating evidence of the involvement of glymphatic dysfunction in Parkinson's disease, especially in those with advanced age and indicated that compromised glymphatic function may be the reasonable mechanism that interlinks aging and PD. With the measure of comorbidities, including CVD risk factors and WMHs, and multivariate linear regression test with the comorbidities and demographic data as covariates, results related to age and age subgroups should be interpreted as relatively directly related to ALPS index.
The identification of glymphatic system has promoted an understanding of how the brain maintains its homeostasis by clearing and draining proteins and toxic metabolites from brain parenchyma through the directionally polarized perivascular pathway. 12 Data of the current study support the hypothesis that compromised glymphatic circulation is an underlying mechanism of α-synuclein accumulation, contributing to the pathogenesis and progression of PD.
PD is typically considered age-related. Previous studies demonstrated that an up to 85% reduction of CSF inflow of large tracers in aged wild-type mice 15 and that contrast clearance in human brain tissue was inversely correlated to age in all subjects with intrathecal injection of gadolinium. 16 In the current PD group, a correlation was identified between age and decreased glymphatic circulation, indicating that compromised glymphatic function and the resulted decreased clearance of metabolites and toxic proteins underlie the mechanism from the natural aging process to PD. Age and disease duration are the two important timescales of disease progression. 5 However, the age at disease onset of different PD patients

| CON CLUS ION
This study demonstrated reduced ALPS index, a potential noninvasive measure of compromised glymphatic function, in PD patients and provided evidence supporting the association between glymphatic dysfunction and aging, sleep disorders, and diabetes mellitus.
In addition, the different findings related to age subgroups denote glymphatic activity may be differently involved in the pathophysiology of PD patients of different ages. These results support the conclusion that compromised glymphatic transport may be the mechanism underlying the pathogenesis and progression of PD, especially in those aged 65 and above. Special attention should be paid to glymphatic function in the aged ones, especially those with sleep disorders and diabetes mellitus.

CO N FLI C T O F I NTE R E S T
All authors declare no actual or potential conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Origin data of the current study are available from the corresponding author on reasonable request.