The impaired hemodynamics of choroid plexus and its interactions within the glymphatic system in Parkinson’s disease

Abstract

The purpose of this study was to identify the perfusion pattern of choroid plexus (CP) impairment in patients with Parkinson’s disease (PD) and determine its relationship with the volume of the CP and its clinical relevance within the glymphatic system. Data from multidelayed arterial spin labeling (m-ASL), T1-weighted imaging, and diffusion tensor imaging were obtained from our consecutively enrolled patients with PD and matched healthy controls (HCs), followed by calculations of perfusion metrics, CP volume (CPV), and analysis along the perivascular space (ALPS) index and further statistical analyses. The pathological pattern of CP perfusion in PD patients included reduced cerebral blood flow (CBF) and arterial blood volume (aCBV) as well as increased CBF-arterial transit time (CBF–ATT) coupling (p < 0.05), which were significantly correlated with the CPV and ALPS indices (p < 0.05). Among the correlations with altered metrics, the linkages between CPV and perfusion metrics (CBF, aCBV, and CBF–ATT coupling) were specific to the PD group. Furthermore, the interactions between CBF and CPV and their clinical relevance to disease severity were mediated by the ALPS index (p < 0.05). These findings elucidate the diverse alterations in hemodynamics of CP and its mediating role within the glymphatic system in PD, underscoring further insights into glymphatic dynamics from the perspective of CP perfusion.

Introduction

Parkinson’s disease (PD) is characterized by heterogeneous neurodegeneration with clinical presentation of movement disorders. In addition to its cardinal pathology of dopaminergic neuron depletion in nigrostriatal pathways, misfolded α-synuclein is deposited as Lewy bodies and progressively aggregates from the brainstem to neocortices, confirming the proteinopathic nature of PD [1, 2]. The decreased clearance of metabolites has recently been identified as another pivotal pathogenesis that exacerbates the accumulation of neurotoxic α-synuclein, as evidenced by experimental findings of glymphatic dysfunction [3, 4]. Notably, an increasing number of neuroimaging studies have proposed that diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) and the choroid plexus volume (CPV) are robust and sensitive metrics for the noninvasive assessment of glymphatic activity while elaborating the inefficient drainage of interstitial fluid (ISF) across perivascular channels (reduced ALPS index) and metabolic disturbance of cerebral spinal fluid (CSF, enlarged CPV) in patients with PD [5–8]. However, the inherent interactions that underlie these glymphatic disturbances remain to be further elucidated for a better understanding of the neurodegenerative processes in PD.

As a highly vascularized structure that comprises the blood–CSF barrier (BCSFB), the choroid plexus (CP) is responsible for the production and secretion of CSF via filtration of plasma from capillaries to the ventricle, which is crucial for the modulation of glymphatic dynamics [9]. Emerging studies have shown that reduced vascularization in the CP caused by capillary loss impairs the BCSFB, manifesting as hypoperfusion of the CP with CSF underproduction, in addition to documented morphological alterations [10, 11]. Accordingly, an integrative investigation of functional and morphological alterations in the CP could provide more insight into glymphatic dysfunction in patients with PD. Although several imaging approaches using the injection of radioactive tracers or gadolinium-based contrast agents are utilized for functional assessment [12–14], a major issue is the inaccurate and invasive quantification of net secretion across the BCSFB, since the CP is a small structure with rapid exchange from the blood to the CSF.

Given that rapid exchange can be traced by arterial spin labeling (ASL) noninvasively by labeling endogenous arterial blood, recent studies have highlighted that cerebral blood flow (CBF) in the CP can be treated as a surrogate for the secretory rate of the BCSFB [15–17]. Emerging modifications of ASL hold promise for cerebral perfusion with more accurate quantification, especially multidelayed ASL (m-ASL). Benefiting from a labeling strategy using multiple postlabel delays (PLDs), m-ASL could correct the underestimation of CBF that occurs in single delayed ASL (s-ASL) and provide more perfusion metrics, including arterial transit time (ATT) and arterial cerebral blood volume (aCBV) [18]. Liu et al. [19] further extended the clinical application of m-ASL to CP, achieving convincing findings of the improved accuracy and reliability of CBF assessment by m-ASL compared with s-ASL at the voxel level, as well as the linkage between decreased CBF and increased volume with aging.

Given that there is still a lack of investigations of glymphatic dysfunction in PD in light of BCSFB dysfunction, the aim of the present study was to explore the perfusion pattern of the CP and its relationship with an enlarged CPV and glymphatic impairment. We compared the intergroup differences in CP perfusion by obtaining more metrics on the basis of ASL data from PD patients and matched healthy controls (HCs), including CBF, aCBV, ATT, the CBF–ATT coupling coefficient, total blood flow (TBF), and total blood volume (TBV), which should characterize impaired hemodynamics across the BCSFB comprehensively and have clinical relevance for movement disorders. In addition to perfusion status, the CPV and ALPS indices calculated from T1-weighted and DTI data were also acquired from the same cohort for further analysis, attempting to gain more understanding of the relationship between perfusion and volume within the CP and potential interactions between the glymphatic dynamics of the CSF and ISF, which corresponds to neurodegeneration in PD patients.

Materials and methods

Participants

This retrospective study consecutively recruited patients with idiopathic PD who visited our Department of Neurology between July 2024 and June 2025, applying the inclusion criteria of the United Kingdom PD Society Brain Bank Clinical Diagnostic Criteria, Hoehn and Yahr (H-Y) stage I–II, stable on medications for more than 4 weeks, and right handedness. During the same interval, matched (age, sex, education degree, and handedness) HC subjects from the local community were also consecutively enrolled via online advertisements. Clinical assessments of disease stage and severity were conducted in the PD group using the H-Y scale and the Unified PD Rating Scale part III (UPDRS-III), respectively, whereas global cognition was evaluated using the Mini-Mental State Examination (MMSE) for all participants. The exclusion criteria for all the subjects were as follows: (1) other Parkinsonism or family history of PD; (2) other neurological or psychiatric disorders (e.g., Alzheimer’s disease, epilepsy, schizophrenia, or stroke); (3) presence of cerebral lesions (e.g., tumor, cerebrovascular disease, or hemorrhage); (4) other diseases that can cause neural dysfunction (e.g., diabetes mellitus, hypertension, anemia, or sleep disorders); (5) subjective complaint of cognitive decline or MMSE score < 27; (6) history of surgery, trauma, drug addiction, smoking, or alcoholism; and (7) contradictions to magnetic resonance imaging (MRI) scanning or poor image quality. The study flowchart of the selection of all participants is depicted in Fig. 1. This study was approved by the Ethics Committee of our hospital, and in accordance with the Declaration of Helsinki. The clinical assessments and enrollments were conducted by two neurologists with more than 15 years of experience. Additionally, the levodopa equivalent dose was recorded from each patient with PD, while they were instructed to not take antiparkinsonian medications at least 12 h before the clinical assessments and MRI scanning. The sample size was evaluated using G Power software based on the statistical assumptions of effect size (f = 0.4, α = 0.05, and power of 80%), indicating the minimum sample size of 86 cases. To ensure the research power, a total of 97 cases were finally included in this study.

Fig. 1.

Flowchart of the participant inclusion and exclusion. PD, Parkinson’s disease; HC, healthy control; MMSE, mini-mental state examination; MRI, magnetic resonance imagingFlowchart of the participant inclusion and exclusion. PD, Parkinson’s disease; HC, healthy control; MMSE, mini-mental state examination; MRI, magnetic resonance imaging

Data acquisition

To control for the effects of circadian rhythms on CSF production activity, MRI acquisition for the participants was consistently performed between 8 a.m. and 11 a.m. A 3.0 Tesla MRI scanner (Discovery 750, GE Healthcare, Milwaukee, WI, USA) equipped with an eight-channel head coil was utilized for data acquisition. All the subjects were instructed to wear an eye mask and earplugs and to lie in a supine position with their head fixed in place by a sponge cushion. During the entire scanning time, each participant was required to keep their eyes closed and to stay awake without thinking, and the individual respiratory rate and heart rate were monitored. m-ASL data were obtained using a three-dimensional pseudocontinuous ASL (3D-pCASL) sequence with the Hadamard-encoded labeling strategy: repetition time (TR) = 6019 ms; echo time (TE) = 11.4 ms; flip angle (FA) = 111°; slice thickness = 4.5 mm without a gap; field of view (FOV) = 220 × 220 mm²; matrix size = 128 × 128; labeling duration = 220, 260, 300, 370, 480, 680, 1180 ms; PLD = 1000, 1220, 1480, 1780, 2150, 2630, 3320 ms; number of slices = 36; 4 arms with 640 points; and total time = 4 min and 3 s. s-ASL data were obtained using a 3D-pCASL sequence: TR = 4836 ms, TE = 4.9 ms, FA = 111°, slice thickness = 4.5 mm without a gap, FOV = 220 × 220 mm², matrix size = 128 × 128, labeling duration = 1800 ms, PLD = 2025 ms, number of slices = 36, 8 arms with 512 points, and total time = 4 min and 12 s. In terms of perfusion quantization and registration, these ASL sequences also include proton density acquisition with matched parameters. Morphological data were obtained to conduct CPV calculations and spatial normalization via a 3D high-resolution brain volume imaging sequence: TR = 12 ms, TE = 5.1 ms, inversion time = 450 ms, FA = 15°, slice thickness = 1 mm without a gap, FOV = 240 × 240 mm², matrix size = 256 × 256, number of slices = 172, and total time = 5 min and 20 s. Diffusion data were obtained to calculate the ALPS via a single-shot echo planar imaging sequence: TR = 8000 ms, TE = 80 ms, FA 90°, b value = 0, 1000 s/mm², slice thickness = 2 mm without a gap, FOV = 256 × 256 mm², matrix size = 128 × 128, number of slices = 67, and total time = 5 min and 30 s.

Data preprocessing and calculation

Referring to equations proposed previously [20–22], we utilized the automatic software “CereFlow” (AnImage, Beijing, China) to calculate the perfusion metrics from raw ASL data, including m-ASL metrics of CBF (ml/100 g/min), aCBV (ml/100 g), and ATT (s), as well as the s-ASL metric of CBF. All perfusion and morphological data were preprocessed by statistical parametric mapping (SPM, version 12, https://www.fil.ion.ucl.ac.uk/spm) with a toolkit running in MATLAB software (version 2018a, MathWorks, Natick, MA, USA). The detailed procedures have been described in previous studies [23, 24]. The procedures included bias-field inhomogeneity correction and head motion correction (displacement > 2 mm or rotation > 2.0° [25]), segmentation of morphological data, spatial normalization to Montreal Neurological Institute (MNI) space (1 mm isotropic) through segmented morphological maps, and spatial smoothing by a Gaussian kernel with a full width at half-maximum (FWHM) of 6 × 6 × 6 mm³. The total intracranial volume (TIV) was obtained during the segmentation step, followed by spatial normalization, standardization, and spatial smoothing according to the above parameters.

All diffusion data were preprocessed using the FMRIB Software Library (FSL, Version 6.05, http://www.fmrib.ox.ac.uk/fsl/). The procedures included brain extraction, eddy-current correction, head motion correction, and tensor fitting. The generated diffusivity maps in all three dimensions (Dxx, Dyy and Dzz) were spatially normalized to MNI space (1 mm isotropic) through the segmented morphological maps using the SPM12 toolkit.

The segmentation of CP in this study was performed using an automatic pipeline [26], including spatial normalization to MNI space (1 mm isotropic) using the Advanced Normalization Tools nonlinear registration tools and segmentation of CP using a fully convolutional neural network model with 3D U-Net architecture. The total volume of the bilateral CP was calculated as the CPV for an individual.

On the basis of the generated CP, we quantified individual BCSFB function by measuring the hemodynamics of the CP. The metrics included the average values of the CBF, ATT, and aCBV of all voxels in the CP; the CBF–ATT correlation coefficient between the CBF and ATT of all voxels in the CP; the TBF of the CP; and the TBV of the CP. Specifically, the TBF (ml/min) and TBV (ml) were calculated using the following formulas:

$$TBF=\rho \times CPV\times CBF$$

$$TBV=\rho \times CPV\times aCBV$$

where ρ indicates the density (g/ml) of the CP, which was set to the mean density of the brain parenchyma (1.08 g/ml) [10].

In accordance with previous studies [27, 28], two regions of interest (ROIs) in each hemisphere were placed on the projection and association fibers in MNI space. The ROIs with radius 5 mm were automatically defined in the areas of bilateral superior corona radiate (center coordinates: left, 116, 110, 99; right, 64, 110, 99) and superior longitudinal fasciculus (center coordinates: left, 128, 110, 99; right, 51, 110, 99). The average diffusivity values of each diffusivity map were extracted on the basis of the ROIs and utilized for the calculation of the ALPS using the following formula:

$$ALPSindex=\frac{mean(Dxx\_projection,Dxx\_association)}{mean(Dyy\_projection,Dzz\_association)}$$

The average value of the bilateral ALPS indices was used to quantify glymphatic function for individuals in the present study.

Statistical analysis

The normality of our data was analyzed using the Kolmogorov‒Smirnov test. The chi-square test was used to compare the intergroup differences in sex. Intergroup differences in education level and MMSE score were analyzed using the Mann‒Whitney U test. We applied independent sample t tests to compare age, perfusion metrics, and the CPV and ALPS indices between the HC and PD groups. For the imaging data in both groups, the Pearson correlation test was used to determine the potential relationships between continuous variables with a normal distribution, whereas the Spearman rank correlation test was used to identify the potential relationships between non-normally distributed variables. The above analyses were conducted by SPSS software (version 25.0; IBM, Inc.) with a statistical significance threshold of p < 0.05. The intergroup differences in the potentially overlapping correlations were estimated using a permutation test (5000 random permutations) by the Gretna toolkit (http://www.nitrc.org/projects/gretna/), with the threshold for significance set at p < 0.05. According to the significant correlations, mediation analyses were further performed to explore whether the ALPS index mediates the relationship between perfusion metrics of CP and CPV and the relationship between the metrics of CP and clinical scales. These analyses were performed using PROCESS in SPSS software with 5000 bootstrap samples, and the statistical significance was assessed by p < 0.05 with a bootstrap 95% confidence interval (CI) that does not include zero. Age, sex, education level, and TIV were used as covariates for the above analyses.

Results

Participants

We enrolled 47 patients with PD and 50 HC subjects. The HC group was matched to the PD group in terms of age, sex, education level, MMSE score, and TIV (p > 0.05 for all). The demographic and clinical information is listed in Table 1.

Table 1.

Demographic data of enrolled subjects and statistical comparisons

	HC (n = 50)	PD (n = 47)	Statistic	p
Age (year)	66.880 ± 6.742	67.957 ± 7.945	− 0.722	0.472
Sex (M/F)	24/26	23/24	0.009	0.927
Education degree (year)	10.000 ± 3.865	10.000 ± 3.464	− 0.202	0.84
MMSE (score)	28.260 ± 0.986	28.021 ± 0.159	− 1.292	0.196
UPDRS-III (score)	N/A	24.660 ± 10.921	N/A	N/A
Disease course (year)	N/A	3.840 ± 2.329	N/A	N/A
H-Y (stage)	N/A	1.596 ± 0.496	N/A	N/A
LEED (mg)	N/A	457.513 ± 216.759	N/A	N/A
TIV (ml)	1493.193 ± 150.038	1483.199 ± 146.942	0.526	0.741

N/A not applicable,

HC healthy control, PD Parkinson’s disease, MMSE Mini-Mental State Examination, UPDRS-III Unified PD Rating Scale part III, H-Y Hoehn and Yahr, LEED levodopa equivalent dose, TIV total intracranial volume

Glymphatic alterations

With respect to the BCSFB function quantified using m-ASL, we observed that patients with PD presented diversely altered hemodynamics of the CP compared with HCs (Fig. 2a–f), including decreased CBF (p = 0.007), decreased aCBV (p = 0.011), and increased CBF–ATT coupling (p = 0.011), whereas the ATT, TBF, and TBV were not significantly altered (p > 0.05). When s-ASL was used (Fig. 3a–b), the TBF was significantly greater (p < 0.001) in the PD group than in the HC group, whereas a significant intergroup difference in CBF was not detected (p > 0.05). Compared with HCs, patients with PD (Fig. 3c–d) presented significantly greater CPV (p < 0.001) and a lower ALPS index (p = 0.002).

Fig. 2.

Intergroup differences of perfusion metrics from m-ASL. Comparison of perfusion metrics between HC subjects and patients with PD using independent sample t tests (a–f). HC, healthy control; PD, Parkinson’s disease; m-ASL, multidelayed arterial spin labeling; CBF, cerebral blood flow; aCBV, arterial cerebral blood volume; ATT, arterial transit time; TBF, total blood flow; TBV, total blood volume

Fig. 3.

Intergroup differences of perfusion metrics from s-ASL and glymphatic metrics. Comparison of perfusion metrics (a–b) and glymphatic metrics (c–d) between HC subjects and patients with PD using independent sample t tests. HC, healthy control; PD, Parkinson’s disease; s-ASL, single-delayed arterial spin labeling; CBF, cerebral blood flow; TBF, total blood flow; CPV, choroid plexus volume; ALPS, analysis along the perivascular space

Metrical correlations and clinical relevance

Several significant correlations were identified between HCs and patients with PD (p < 0.05 for all), which are presented in Fig. 4 and Table 2. In terms of the altered metrics, we observed significant correlations in both groups, including correlations between ALPS and metrics of CBF, aCBV, CBF–ATT coupling and CPV, and correlations between CBF and metrics of aCBV and CPV. However, several correlations were observed only in patients with PD, including correlations between CPV and metrics of aCBV and CBF–ATT coupling, and correlations between CBF and CBF–ATT coupling.

Fig. 4.

Correlation analysis of imaging metrics and their linkages to age. The relationships between paired imaging metrics and their relationships to age were conducted in HC and PD group using Pearson correlation test (a–b). * indicates statistical significance with p < 0.05. HC, healthy control; PD, Parkinson’s disease; m-ASL, multidelayed arterial spin labeling; CBF, cerebral blood flow; aCBV, arterial cerebral blood volume; ATT, arterial transit time; TBF, total blood flow; TBV, total blood volume; r, correlation coefficient

Table 2.

Significant correlations in each group and statistical comparisons

		HC			PD		pPermutation
		r	pPearson		r	pPearson
CBF	TBV	0.681	< 0.001		0.729	< 0.001	0.326
CBF	ALPS	0.360	0.010		0.382	0.008	0.450
CBF	aCBV	0.933	< 0.001		0.941	< 0.001	0.388
CBF	TBF	0.291	0.040		0.293	0.046	0.503
CBF	CPV	− 0.360	0.010		− 0.520	< 0.001	0.087
CBF	ATT	0.371	0.008		N/A	N/A	N/A
CBF	Age	N/A	N/A		− 0.313	0.032	N/A
CBF	CBF-ATT coupling	N/A	N/A		− 0.318	0.029	N/A
aCBV	TBV	0.796	< 0.001		0.795	< 0.001	0.490
aCBV	ALPS	0.347	0.014		0.306	0.037	0.420
aCBV	TBF	0.373	0.008		0.314	0.032	0.374
aCBV	ATT	0.674	< 0.001		0.416	0.004	0.028
aCBV	CPV	N/A	N/A		− 0.477	0.001	N/A
ATT	TBV	0.652	< 0.001		0.320	0.028	0.032
ATT	TBF	0.356	0.011		N/A	N/A	N/A
CBF-ATT coupling	ALPS	− 0.439	0.001		− 0.506	< 0.001	0.324
CBF-ATT coupling	CPV	N/A	N/A		0.317	0.030	N/A
ALPS	Age	− 0.309	0.029		− 0.415	0.004	0.279
TBF	TBV	0.844	< 0.001		0.731	< 0.001	0.151
TBF	CPV	0.781	< 0.001		0.643	< 0.001	0.153
TBF	Age	0.431	0.002		N/A	N/A	N/A
TBV	CPV	0.369	0.008		N/A	N/A	N/A
CPV	Age	0.447	0.001		0.485	0.001	0.382
CPV	ALPS	− 0.415	0.003		− 0.463	0.001	0.400

N/A not applicable,

HC healthy control, PD Parkinson’s disease, CBF cerebral blood flow, aCBV arterial cerebral blood volume, ATT arterial transit time, TBF total blood flow, TBV total blood volume, CPV choroid plexus volume, ALPS analysis along the perivascular space, r correlation coefficient

In terms of clinical relevance (Fig. 5a–e), we observed that increased UPDRS-III scores were correlated with decreased CBF (r =  − 0.441, p = 0.002), decreased aCBV (r =  − 0.424, p = 0.003), increased CBF‒ATT coupling coefficient (r = 0.364, p = 0.012), enlarged CPV (r = 0.551, p < 0.001), and decreased ALPS (r =  − 0.521, p < 0.001). However, there was no significant relationship between the altered perfusion metrics and the H-Y and MMSE scores (p > 0.05 for both).

Fig. 5.

Clinical relevance of altered imaging metrics in patients with PD. Significant correlations between altered imaging metrics (X axis) and disease severity scored by UPDRS-III scale (Y axis) using Pearson correlation test (a–e). PD, Parkinson’s disease; CBF, cerebral blood flow; aCBV, arterial cerebral blood volume; ATT, arterial transit time; CPV, choroid plexus volume; ALPS, analysis along the perivascular space; UPDRS-III, Unified PD Rating Scale part III; r, correlation coefficient

Altered linkages and mediation effects

With respect to the overlapping correlations, the permutation test revealed significant intergroup differences in the correlations between ATT and the aCBV and TBV metrics and between CBF and CPV (p < 0.05). The statistical details of the correlations are summarized in Table 2.

Furthermore, we found a mediating effect of ALPS on the interaction between CBF and CPV (indirect effect, − 0.119; 95% CI = [− 0.329, − 0.003]), as well as an interaction between CP metrics (CBF and CPV) and UPDRS-III scores (indirect effect, − 0.156, 95% CI = [− 0.356; − 0.028]; indirect effect, 0.158, 95% CI = [0.031, 0.389]) (Fig. 6).

Fig. 6.

Mediation analysis within CP’s metrics and their relevance in patients with PD. The interaction between CBF and CPV as well as their linkages to UPDRS-III score were mediated by ALPS index. PD, Parkinson’s disease; CBF, cerebral blood flow; CPV, choroid plexus volume; ALPS, analysis along the perivascular space; UPDRS-III, Unified PD Rating Scale part III

Discussion

An increasing number of investigations focused on glymphatic clearance have provided novel insights into the complex etiologies of PD [29]. In the present study, we explored glymphatic disturbances from the perspective of CP perfusion, benefiting from the abundant metrics of m-ASL. Our findings revealed that hypoperfusion of the CP in patients with PD was characterized by decreased CBF and aCBV as well as increased CBF‒ATT coupling, which were associated with glymphatic metrics (enlarged CPV and a reduced ALPS index) and disease severity (UPDRS-III). We further noted that the interactions within the CP between perfusion and volume and the clinical linkage of CP alterations (CBF, aCBV, and CPV) were all partially mediated by the ALPS index and specific to neurodegenerative status. This study identified the perfusion pattern associated with CP impairment and addressed its relationship with morphological enlargement and movement disorders, emphasizing the clinical implications of BCSFB impairment for understanding glymphatic dysfunction in PD.

Given that the CP is an intraventricular architecture with abundant vascularity, proper selection among neuroimaging approaches is a prerequisite for accurate assessment of its hemodynamics in vivo. The ASL protocol utilized in the present study has been recommended for the noninvasive estimation of BCSFB function with the following technical features [16, 17, 19, 30]: (1) the major constituent of hydrogen protons in the CP is arterial blood supplied by anterior and posterior choroidal arteries, which can be efficiently labeled by ASL; (2) the rapid exchange of labeled endogenous arterial blood with the CSF across the extensive epithelial surface, which can be quantitatively measured by ASL; and (3) CP perfusion could be an effective surrogate of CSF secretion due to its tight coupling, which can be more accurately detected with more information when m-ASL is used. Although pilot trials have been conducted in HC subjects, this is the first study to employ the m-ASL protocol to explore the aberrant perfusion of the CP in patients with PD while making a comparison with s-ASL. We specified the intergroup differences in perfusion metrics from m-ASL and their linkages to CPV, ALPS, and UPDRS-III. These findings assisted in our elucidation of inherent interactions among the glymphatic system that correspond to neurodegeneration in PD patients, which could be misled by underestimations (CBF and its linkages) or false positives (TBFs) when using s-ASL.

In contrast to the conventional microcirculation in the cerebral parenchyma, the capillaries in the CP principally engage in rapid filtration of arterial blood into the stroma using the fenestrated endothelial cells [15, 31]. The biological meaning of altered perfusion metrics from m-ASL should be more inclined to the function of the CP on the basis of its special vascularity. Previous studies in humans and animal models have shown that the CBF of the CP represents the secretory rate of CSF, which is supported by the free passage of water molecules in artery blood across the BCSFB [32, 33]. Accordingly, we inferred that the perfusion metrics of the aCBV, TBV, and TBF, which are calculated on the basis of the CBF, could also highly correspond to the average volume, total volume, and total rate of CSF production in the CP, respectively. Recent studies have reported that the CP, during aging or neurodegeneration, is vulnerable to vascular degradation because of its highly vascularized property, which manifests as vascular tortuosity caused by arterial stiffening [17, 34]. The decreased CBF and aCBV in patients with PD were in accordance with these pathological features and illustrate the consequent inefficiency of the BSCFSB for CSF production. Moreover, we observed that the PD patients who were not affected by cerebrovascular diseases (stenosis or occlusion) had greater CBF–ATT coupling. Although its linkage to glymphatic metrics and clinical relevance were also detected in this study, the meaning of this metric is rather ambiguous. The correlation between CBF and ATT is theoretically in accordance with the central volume principle and is sensitive to vessel tortuosity associated with a reduced stiffness of vessel walls [35, 36]. One possible interpretation of this alteration might be the sparse vascular density caused by capillary loss in the CP [37], which pathologically decouples the balance between CBF and ATT.

The existing studies on PD have focused mainly on morphological alterations in the CP, which can be attributed to the implementation of high-resolution sequences for anatomy and automatic methodology for segmentation [38–40]. Our observation of the enlarged CPV and its identified correlations are consistent with previous findings, collectively supporting the chronic hyperplasia of CP (stromal fibrosis, lipofuscin deposition, and thickening of the basement membrane) that induces drainage impairment and motor dysfunction [41, 42]. In addition to the inverse alterations between CBF and CPV reported previously [17, 43], the interaction between hyperplasia and hypoperfusion was emphasized by our detection of the correlations between other altered perfusion metrics (aCBV and CBF–ATT coupling) and enlarged volume in patients with PD. Importantly, this study further suggested a pathway of these pathological alterations within the CP rather than the controversial conjectures. In the same cohort of PD patients who were at an early stage, the maintained amount of total perfusion (TBF and TBV) and the mediating effect of the ALPS index were noted, which convinced us that an enlarged CPV is primarily due to vascular remodeling and is probably a response (epithelial hypertrophy) to its functional impairment. As hypoxia or ischemia occurs in the CP, the secretion rate of CSF decreases, which disturbs the glymphatic dynamics in the presence of drainage dysfunction, followed by compensatory maintenance of cerebral homeostasis using the gradual enlargement of the CP for sufficient CSF production.

With the evolving conception of the clearance system in the brain, many studies have elucidated the vicious cycle formed between impaired glymphatic dynamics (ISF and CSF) and neurodegenerative signatures of PD (α-synuclein aggregation and dopaminergic denervation), subsequently leading to clinical features of movement disorders [3, 44, 45]. The ALPS index utilized in this study measures water diffusivity along the perivascular space and is firmly representative of the drainage status of the ISF [27, 46], which is also impaired in PD patients. We additionally found that the aberrant ALPS index mediated the pathological process within the CP and its clinical relevance to motor dysfunction. These observations support the proposed glymphatic physiopathology in the pathogenesis of PD [47, 48] and further reveal the sequential mechanism of BCSFB dysfunction along the glymphatic pathway. Although the explored correlations in the HC group were in agreement with the reported findings with aging [49, 50], it should be noted that the coupling between the dynamics of the ISF and CSF is independent of PD, since no intergroup differences were detected. However, the relationships between the perfusion metrics of CP and CPV and the mediating effect of ALPS were specific to patients with PD. Accordingly, we suggest that the CP, which drives glymphatic dynamics, is more vulnerable to pathological involvement in PD, and its functional status should receive more attention from multiple aspects of hemodynamics.

There are several limitations that should be recognized in this study. First, the statistical reliability of our results might be influenced by the relatively small sample size, which needs validation via further longitudinal and cross-sectional studies with multicenter data and detailed clinical scales. Second, although the neuroimaging metrics employed in this study correspond to certain glymphatic properties, their biological definitions and causality warrant confirmation by integrating other pathological markers (e.g., α-synuclein content in CSF and cerebral parenchyma) in subsequent studies. Third, the MRI sequences and preprocessing methods were utilized in consideration of their documented robustness and superiority. However, more advanced neuroimaging protocols (e.g., large-model-driven clinical information integration and vascular morphology-based quantitative prediction [51, 52]) in the future could provide better evaluations of the glymphatic system. Fourth, our focus on glymphatic dynamics is still not enough for a comprehensive understanding of the complicated neurodegeneration in PD, which requires complementary evidence from other aspects (e.g., iron deposition, dopaminergic denervation, synaptic deficiency, and asymmetric effects). Finally, we attempted to minimize the effects of medication by withdrawing medications prior to the clinical evaluations and data acquisition. Nevertheless, this limitation remains a concern in this preliminary study due to the residual medication effects and could be addressed by recruiting a medication-naïve cohort of PD patients.

Conclusion

In conclusion, this study focused on BCSFB dysfunction in PD patients by detecting aberrant hemodynamics and their relationships via the integration of perfusion metrics from m-ASL and glymphatic data from the CPV and ALPS indices. We revealed that diverse impairments in BCSFB perfusion occurred in patients with PD, which initiated the pathological processes of a compensatory enlargement of the CPV and disruption of ICS dynamics, leading to a reduced ALPS index and, consequently, clinical symptoms of movement disorders. Our findings offer potentially novel perceptions of CSF dynamics for deeper insights into the glymphatic mechanism underlying motor dysfunction in PD patients and affirmed the clinical implementation of m-ASL for comprehensive investigations of glymphatic disturbances in neurodegenerative diseases.

Abbreviations

HC

Healthy control

PD

Parkinson’s disease

aCBV

Arterial cerebral blood volume

ALPS

Analysis along the perivascular space

ATT

Arterial transit time

BCSFB

Blood–cerebral spinal fluid barrier

CBF

Cerebral blood flow

CPV

Choroid plexus volume

CSF

Cerebral spinal fluid

ISF

Interstitial fluid

m-ASL

Multi-delayed arterial spin labeling

r

Correlation coefficient

s-ASL

Single-delayed arterial spin labeling

TBF

Total blood flow

TBV

Total blood volume

UPDRS-III

Unified PD Rating Scale part III

Author contribution

Song’an Shang: writing—review and editing; writing—original draft; visualization; validation; methodology; investigation; data curation; conceptualization. Shaohua Ding: visualization, validation, methodology, investigation, data curation, conceptualization. Hongying Zhang: supervision, formal analysis. Xiang Lv: data curation. Beilei Chen: data curation. Daming Shen: methodology. Jing Ye: supervision, project administration, funding acquisition, formal analysis, data curation, conceptualization. Yu-Chen Chen: supervision, methodology, funding acquisition, data curation.

Funding

This work was supported by the National Natural Science Foundation of China (82202120), Science and Technology Project of Yangzhou (YZ2024066, YZ2024093), Special Scientific Research Found of Yangzhou Health Commission (2023–2-08), Medical Research Program of Jiangsu Health Commission (M2022059), and Clinical research project of Clinical Medical College, Yangzhou University (SBLC22004).

Data availability

Data generated or analyzed during the study are available from the corresponding author on reasonable request.

Declarations

Ethics approval

The study was conducted in accordance with the Helsinki Declaration. All the data were from clinical tests, and there was no clinical intervention for the participants in the study. Informed consent for routine clinical tests was obtained from all participants.

Competing interests

The authors declare no competing interests.