Glymphatic Function in Prodromal Parkinson's Disease: Associations With Symptoms, Gray Matter Volume, and Phenoconversion Risk

ABSTRACT

Background and Objectives

Glymphatic dysfunction occurs in Parkinson's disease (PD), but its status in prodromal PD (pPD) is unclear. Using the diffusion tensor imaging along perivascular spaces (DTI‐ALPS) index as an indirect proxy, this study aims to evaluate glymphatic‐related changes in a pPD cohort and to explore their relationship with symptoms, gray matter volume, and risk of phenoconversion.

Methods

We analyzed data from the Parkinson's Progression Marker Initiative, including 51 healthy controls (HC), 83 individuals with pPD, and 202 with de novo PD (dnPD). The pPD cohort underwent ≥ 4‐year follow‐up. Cross‐sectional analyses compared the DTI‐ALPS index across three groups and examined associations between the DTI‐ALPS index and clinical features/gray matter volume in the pPD cohort. For longitudinal analysis, the relationship of DTI‐ALPS index with the risk of phenoconversion was assessed via Kaplan–Meier and Cox regression.

Results

DTI‐ALPS index was significantly reduced in pPD and dnPD vs. HC (p < 0.001). In pPD, lower DTI‐ALPS index correlated with: higher anxiety (STAI: r = −0.46, p = 0.02), elevated CSF p‐tau (r = 0.52, p = 0.04) and t‐tau (r = 0.42, p = 0.03), lower volumes of right temporal pole (p = 0.03), left thalamus (p = 0.01), right superior occipital gyrus (p = 0.04), and higher volume of left posterior central gyrus (p = 0.03). Among 83 pPD subjects, 10 phenoconverted to PD. Each standard deviation decrease in DTI‐ALPS increased conversion risk by 13% (adjusted HR = 0.87, 95% CI 0.83–0.92; p = 0.018).

Conclusion

Reduced DTI‐ALPS index in pPD—which may reflect altered glymphatic function—is associated with accelerated phenoconversion and correlates with clinical symptoms and cortico‐thalamic structural changes.

1. Introduction

Parkinson's disease (PD) is a progressive neurodegenerative disorder marked by the pathological accumulation of alpha‐synuclein and the depletion of dopaminergic neurons in the substantia nigra [1, 2]. While the clinical diagnosis of PD primarily relies on the presence of motor symptoms, the underlying pathological changes and neurodegenerative processes actually initiate years before the onset of these cardinal motor manifestations, during what is known as the prodromal phase of Parkinson's disease (pPD) [3, 4]. Characterized predominantly by subtle non‐motor symptoms, pPD often goes unnoticed in its early stages, posing challenges for early detection [5]. There is a critical need for robust biomarkers that can facilitate the early and precise diagnosis of pPD and identify those at risk of progressing to PD. The identification of such individuals is paramount for the timely implementation of neuroprotective strategies, which are essential for maximizing therapeutic efficacy.

The glymphatic system of the brain, discovered in recent years, is a highly organized waste removal pathway that removes soluble proteins, including amyloid‐beta (Aβ) and tau, from the brain [6, 7]. Emerging evidence suggests that dysfunctions in the glymphatic system's ability to clear these waste products may contribute to the onset and progression of various neurodegenerative disorders, including Alzheimer's disease (AD) [8]. In a seminal study, Zou and colleagues demonstrated through animal models that the abnormal accumulation of α‐synuclein resulting from impaired glymphatic function is a pivotal factor in the pathogenesis of PD [9].

A previous study investigated brain lymphoid system dysfunction in patients with idiopathic rapid eye movement sleep behavior disorder (iRBD), which is considered a prodromal Parkinson's disease [10]. Longitudinal research has further revealed that diminished glymphatic system function in iRBD is indicative of a heightened risk for the development of Parkinson's disease [11, 12]. However, iRBD is only one specific type of pPD, and its pathologic development pattern may differ from other types [13, 14]. Therefore, the value of altered brain glymphatic system function should be validated in a mixed pPD population.

The initial techniques for assessing the glymphatic system involved the intrathecal injection of gadolinium‐based contrast agents (GBCA) as tracers [15, 16]. An alternative method involved intravenous injection of GBCA to monitor the enhancement of the interstitial space surrounding blood vessels as an indicator of glymphatic system function [17]. But all of these methods are invasive. Recently, diffusion tensor image analysis along the perivascular space (DTI‐ALPS) has emerged as a non‐invasive, real‐time technique for evaluating the glymphatic system [18]. DTI‐ALPS has demonstrated a significant correlation with glymphatic clearance function as assessed by intrathecal vascular tracing methods [19] and has proven to be a reliable approach, consistent across different MRI scanners [20, 21]. Currently, ALPS is the most extensively utilized MRI‐based technique for evaluating human glymphatic function and has been applied in the study of various diseases, including Alzheimer's disease, type 2 diabetes, ischemic stroke, multiple sclerosis, and Parkinson's disease [22, 23, 24, 25, 26, 27].

In this study, the DTI‐ALPS index was used to compare the glymphatic function in pPD, PD, and healthy controls, and the relationship between DTI‐ALPS and clinical features and cerebrospinal fluid biological markers in the pPD group was analyzed. In addition, we tried to explore the relationship between the function of the glymphatic system and the volume of gray matter in pPD. Then, in a follow‐up study, we assessed the association of the DTI‐ALPS index with the probability of conversion from pPD to clinical PD.

2. Methods

2.1. Participants

This study analyzed data from participants with pPD, dnPD patients, and HCs, all of whom were part of the Parkinson's Disease Progression Markers Initiative (PPMI). PPMI is a prospective, longitudinal, observational multicenter study aimed at validating biomarkers of PD progression.

PD patients were identified as individuals at risk for developing PD based on clinical characteristics, genetic variations, or other biomarkers, such as RBD, genetic risk variants (LRRK2, GBA), UPSIT‐based hyposmia, and positive dopamine transporter (DAT) SPECT on visual inspection. Detailed criteria for participant inclusion are available online at ppmi‐info.org/study‐design.

Study visits were scheduled at baseline, then every 3 months for the first year, and subsequently every 6 months. The data for this study were downloaded in May 2023. In our study, we included participants who met the following criteria: (1) All participants were required to have complete 3D T1‐weighted and DTI data at baseline. (2) To ensure consistency between image data, only participants with MRI images obtained on 3 Tesla Siemens MRI scanners were included. (3) The pPD group was followed up for at least 4 years. Participants with poor‐quality MRI images were excluded.

PD converters are defined as individuals who have received at least two consecutive diagnoses of PD in accordance with the Movement Disorder Society (MDS) Clinical Diagnostic Criteria for Parkinson's Disease 2015 [28] (2 patients who were initially diagnosed with PD but later diagnosed with other synaptoprotein diseases were excluded from this study).

2.2. Ethics Statement

The PPMI study is registered at Clinical Trials.gov (NCT01141023). Each participating PPMI site received approval from an ethical standards committee on human experimentation before the start of the study. Written informed consent for the study was obtained from all participating individuals.

2.3. Clinical Evaluation

All participants were clinically assessed at baseline and these scales were used to assess motor and non‐motor symptoms, including the Unified Parkinson's Disease Rating Scale, Part III (UPDRS III), University of Pennsylvania Odor Recognition Test (UPSIT), Rapid Eye Movement Sleep Behavior Disorder Questionnaire (RBDSQ), Epworth Sleepiness Scale (ESS), autonomic symptoms of Parkinson's disease (UPSIT), Montreal Cognitive Assessment (MoCA), State Trait Anxiety List (STAI), Geriatric Depression Scale (GDS).

2.4. Image Acquisition

Non‐contrast enhanced 3D volumetric T1‐weighted MRI and DTI scans of the total subjects were acquired on 3T Siemens MRI scanners (Erlangen, Germany) using an MPRAGE sequence at different centers. Acquisition parameters and detailed protocols are available on the PPMI website. The following indexes were included in the protocol: (1) DTI: 72 axial slices, echo time (TE) = 88 ms, repetition time (TR) = 500–9000 ms, voxel size: 2.0 × 2.0 × 2.0 mm³, acquisition matrix = 1044 × 1044; one diffusion‐unweighted (b0) image and 64 diffusion‐sensitive gradient directions at b = 1000 s/mm² images; (2) 3D‐T1WI: TE = 2.98 ms, TR = 2300 ms, voxel size: 1.0 × 1.0 × 1.0 mm³, acquisition matrix = 240 × 256. Staff at each center were trained to ensure that the data were collected in a standardized manner. In subsequent data analysis, subjects with missing data were excluded.

2.5. ALPS Index Processing

We adopted the method for DTI‐ALPS processing and measurement from the previous publication [18]. First, the diffusion‐weighted images were processed by FMRIB software (FSL, FSL/http://www.fmrib.ox.ac.uk/). The process includes the following steps: (1) We converted DTI images in raw data into 4D NIfTI format files using the dcm2nii software. The head motion and eddy current of the DTI image were corrected. (3) The skull of scalp was removed to obtain brain tissue. (4) The dtift command of FSL was used to reconstruct the tensor from the DTI image; the color fraction anisotropy (FA) image was obtained.

At the level of the lateral ventricle, the direction of the subcortical fibers and the perivascular space is almost the same, both perpendicular to the lateral ventricle, mainly in the left and right direction, that is, the X‐axis. The associated fibers are mainly in the forward and backward direction (Y‐axis), while the projection fibers maintain the up and down direction (Z‐axis). The directions of the perivascular space, associated fibers, and projection fibers are perpendicular to each other.

Two trained neuroradiologists evaluated all images without knowledge of clinical data, independently selecting the ROI. A spherical region of interest (ROI) with a diameter of 4 mm was placed on the projective fibers, associated fibers, and subcortical fibers in the left and right hemispheres. Fsleyes software was used to extract x, y, z direction fusion fibers Dxx, Dyy, and Dzz for projection and correlation fibers within the ROI. The ALPS index for the left and right hemispheres was calculated separately, and then the average ALPS index for both sides was used for the main analysis. Intra‐class correlation coefficients (ICCs) are used to determine the consistency between estimates of the ROI segmentation program (Figure 1).

FIGURE 1.

Schematic drawing of the DTI‐ALPS methodology. (a) colored FA map shows the ROIs on the projection fibers (blue), association fibers (green), and subcortical fibers (red). (b) Schematic drawing of the spatial relationships between the perivascular space and subcortical fibers (red; x‐axis), association fibers (green; y‐axis), and projection fibers (blue; z‐axis).

The formula for calculating DTI‐ALPS index is as follows:

$$\text{ALPS index}=\frac{\text{mean}\left(\text{Dxxproj},\text{Dxxassoc}\right)}{\text{mean}\left(\text{Dyyproj},\text{Dzzassoc}\right)}$$ (1)

Dxxproj: diffusivity along the x‐axis in the projection fiber, Dxxassoci: diffusivity along the x‐axis in the association fiber, Dyyproj: diffusivity along the y‐axis in the projection fiber, Dzzassoci: diffusivity along the z‐axis in the association fiber.

2.6. Gray Matter Volume Measurement

Baseline brain MRI sequences were derived from the PPMI database. We applied a voxel‐based morphometry method using high‐resolution 3‐D T1‐weighted images. By affine and high‐dimensional nonlinear registration, the individual image was spatially normalized to the standard stereotactic space. Using the CAT12 toolbox (dbm.neuro.uni‐jena.de/cat/) and MATLAB SPM12 (fil.ion.ucl.ac.uk/spm/software/spm12/), we segmented MRI images into gray matter and white matter in the standard space of the Montreal Neurological Institute (MNI) in high dimension, and then segmented gray matter into different brain regions using the AAL3 template (Table S1).

2.7. Statistics

Statistical analysis was performed by SPSS 26.0 and R statistical software (version 3.4.3; http://www.r‐project.org/). Statistical plots were generated using GraphPad Prism 8.0a (GraphPad Inc., San Diego, CA, USA).

2.7.1. Demographic and Clinical Differences

Continuous variables of normal distribution and skew distribution were represented by mean ± standard deviation (SD) and median (1st quartile, 3rd quartile), respectively. Categorical variables were represented by frequency. ANOVA was used for continuous variables with normal distribution, and Mann‐Whitney U test and Kruskal‐Wallis test were used for skewed continuous variables. The comparison of categorical variables was performed by chi‐square test. In addition, we performed a Bonferroni post hoc analysis of the ALPS Index. We proposed a K‐Nearest Neighbor (KNN) weighted imputation method for missing data. p < 0.05 (bilateral) was considered statistically significant.

2.7.2. Correlations Between the ALPS Index, Gray Matter Volume, and Clinical Characters

In pPD group, we performed a partial correlation analysis to identify the relationship between ALPS index and gray matter volume, as well as the relationship between ALPS index and various clinical features after adjusting for age, sex, and years of education. To account for multiple comparisons across all correlation analyses, the False Discovery Rate (FDR) correction was applied, with a significance threshold set at FDR‐adjusted p < 0.05.

2.7.3. Connections Between the ALPS Index and Progression of pPD in the Follow‐Up Cohort

Univariate COX regression analysis was used to screen out possible predictors, and then factors with p < 0.1 in univariate analysis were included in multivariate Cox regression analysis. The influence of DTI‐ALPS index levels on pPD to PD conversion was examined by Kaplan–Meier survival analysis, with differences between groups evaluated through the log‐rank test. Double‐tailed test was used, and p < 0.05 was considered statistically significant.

2.7.4. Sensitivity Analysis

Sensitivity analysis was performed by multi‐model Cox regression analysis to examine the robustness of our results. Model 1 is crude. Model 2 includes three confounding factors: sex, age, and education. Model 3 was adjusted for sex, age, education, RBDSQ, and STAI_anxiety_total scores. Model 4 was further adjusted for sex, age, education, RBDSQ, STAI_anxiety_total scores, GBA_PATHVAR, and LRRK2_PATHVAR. Hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated to determine the risk of PD conversion.

3. Results

3.1. Demographics and Clinical Characteristics

A total of 51 cases of HCs, 83 cases of pPD, and 202 cases of dnPD were included in this study. The demographic and clinical characteristics of these participants at baseline were shown in Table 1. No significant differences were observed in age and sex among the three groups. However, there were statistically significant differences in years of education, UPSIT scores, RBDSQ scores, SCOPA‐AUT scores, UPDRS III scores, and in the distribution of GBA and LRRK2 genotypes.

TABLE 1.

Demographic and clinical characteristics of subjects at baseline.

Variables	HC (n = 51)	pPD (n = 83)	dnPD (n = 202)	Statistic	p *
Male Sex, n (%)	30 (58.82)	41 (49.40)	125 (61.85)	χ2 = 3.78	0.169
Age (y), Mean ± SD	61.98 ± 9.71	62.60 ± 7.23	62.12 ± 9.57	F = 0.10	0.408
Education (y), Mean ± SD	16.00 ± 5.26	17.00 ± 6.25	15.69 ± 5.87	F = 0.16	0.262
UPSIT, M (Q1, Q3)	39.50 (36.00, 40.00)	32.28 (23.00, 38.00)	30.12 (21.00, 39.00)	χ2 = 24.13#	< 0.001
RBDSQ, M (Q1, Q3)	1.45 (1.00, 3.00)	8.23 (2.00, 17.00)	9.12 (4.00, 18.00)	χ2 = 45.27#	< 0.001
ESS, Mean ± SD	6.21 ± 3.73	8.32 ± 4.36	8.68 ± 3.97	F = 0.45	0.351
SCOPA AUT, M (Q1, Q3)	—	6.73 (4.00, 11.00)	8.21 (6.00, 15.00)	χ2 = 12.19#	0.026
MoCA, Mean ± SD	28.73 ± 3.18	26.82 ± 3.89	25.76 ± 3.71	F = 0.81	0.065
STAI, M (Q1, Q3)	25.24 (21.00, 29.00)	27.31 (23.00, 37.00)	28.39 (22.00, 41.00)	χ2 = 12.62#	0.172
GDS, M (Q1, Q3)	2.27 (0, 4.00)	3.82 (2.00, 6.00)	4.36 (2.00, 7.00)	χ2 = 0.18#	0.821
UPDRS III, M (Q1, Q3)	—	10.00 (3.00, 14.00)	28.00 (21.00, 36.00)	χ2 = 18.46#	< 0.001
CSF α‐syn (pg/mL), Mean ± SD	1522.00 ± 296.34	1421.40 ± 309.86	1373.90 ± 341.25	F = 0.95	0.372
CSF t‐Tau (pg/mL), Mean ± SD	166.85 ± 27.98	177.30 ± 29.51	151.80 ± 24.81	F = 0.72	0.233
CSF p‐Tau (pg/mL), Mean ± SD	15.26 ± 4.81	14.77 ± 4.28	13.17 ± 3.96	F = 0.37	0.265
GBA, n (%)	—	26 (33.77)	18 (9.63)	χ2 = 35.55#	< 0.001
LRRK2, n (%)	—	38 (49.35)	26 (13.90)	χ2 = 56.67#	< 0.001

Note: F: ANOVA, #: Kruskal‐wallis test, χ²: Chi‐square test; SD: standard deviation; M: Median; Q₁: 1st Quartile; Q₃: 3rd Quartile.

Abbreviations: α‐syn, alpha‐synuclein; CSF, cerebrospinal fluid; dnPD, de novo Parkinson's disease; ESS, Epworth sleepiness scale; GDS, the Geriatric Depression Scale; HC, healthy control; MoCA, Montreal Cognitive Assessment; p‐tau, tau phosphorylated; pPD, prodromal Parkinson's disease; RBDSQ, Rapid Eye Movement Sleep Behavior Disorder Questionnaire; SCOPA‐AUT, autonomic symptoms in Parkinson's disease; STAI, State–trait anxiety inventory; UPDRS III, Unified Parkinson's Disease Rating Scale, Part III; UPSIT, University of Pennsylvania Smell Identification Test.

^* p was adjusted for multiple comparisons using the Bonferroni correction. Statistically significant p < 0.05.

3.2. Inter‐Observer Consistency in the DTI‐ALPS Index

For the DTI‐ALPS index (intra‐group correlation coefficient 0.85 [95% CI: 0.75, 0.91]), inter‐observer agreement was excellent.

3.3. Difference of DTI‐ALPS Index Among Groups

The mean, left, and right DTI‐ALPS indices were all significantly different among the three groups (p < 0.001). Post hoc tests, with FDR correction applied for multiple comparisons. The subsequent tests showed that the mean, left, and right DTI‐ALPS index in the pPD group and the PD group were significantly lower than that in the HC group, but the difference in the mean, left, and right DTI‐ALPS index between the pPD group and the dnPD group was not statistically significant (Figure 2).

FIGURE 2.

Raincloud plot of the DTI‐ALPS (a), DTI‐ALPS_L (b), DTI‐ALPS_R (c) among the HC, pPD, and dnPD. HC: Healthy controls, pPD: Prodromal Parkinson's disease, dnPD: Newly diagnosed Parkinson's disease. APLS: Diffusion tensor image analysis along the perivascular space. Error bars represent the standard error. **p < 0.05.

3.4. Correlation Between ALPS Index and Clinical Features

In the pPD group, after adjusting for age, sex, and years of education, the ALPS index was significantly negatively correlated with STAI score (r = 0.46, FDR‐adjusted p = 0.02); and it was positively correlated with p‐tau concentrations (r = 0.52, FDR‐adjusted p = 0.04) and t‐tau concentrations (r = 0.42, FDR‐adjusted p = 0.03) (Figure 3).

FIGURE 3.

Correlations between the ALPS index and Clinical features. There was a significant negative correlation between ALPS index and STAI‐anxiety total scores (r = 0.46, p = 0.02) (a), There was a significant positive correlation between ALPS index and p‐Tau (r = 0.52, p = 0.04) (b) and t‐tau (r = 0.42, p = 0.03) (c).

3.5. Association Between DTI‐ALPS Index and Gray Matter Volume

After adjusting for age, sex, and total intracranial volume, significant positive associations were observed between the DTI‐ALPS index and gray matter volumes of the right temporal pole (r = 0.36, FDR‐adjusted p = 0.03), left thalamus (r = 0.27, FDR‐adjusted p = 0.01), right superior occipital gyrus (r = 0.41, FDR‐adjusted p = 0.04), and significant negative associations were observed between the DTI‐ALPS index and the gray matter volume of the left posterior central gyrus (r = −0.33, FDR‐adjusted p = 0.03) (Figure 4).

FIGURE 4.

Associations between DTI‐ALPS index and regional gray matter volume. Illustration of the regions‐of‐interest (defined in AAL3) in which the regional gray matter volumes are significantly associated with the DTI‐ALPS index. Areas marked in red represent positive correlations, including the right temporal pole, left thalamus, and right superior occipital gyrus. Areas marked in blue represent positive correlations, including the left posterior central gyrus. DTI‐ALPS indicates diffusion tensor imaging analysis along the perivascular space.

3.6. Relationship Between DTI‐ALPS and Risk of PD Phenoconversion

Among the 83 pPD patients, 10 cases were converted to PD (conversion rate = 12.1%). The median follow‐up time was 5.48 years. Table 2 shows the baseline characteristics of risk factors in pPD patients with PD conversion versus those without PD conversion during the follow‐up period, and the results revealed that the frequency is higher in RBD patients and patients with anxiety (Due to the substantial missing data in DAT‐SPECT, it was not included in the statistics).

TABLE 2.

Baseline variables in the pPD cohort of PD converters and nonconverters.

Risk markers	Total (N = 83)	PD nonconverters (N = 73)	PD converters (N = 10)	p
Age, Mean ± SD	62.60 ± 7.23	62.86 ± 7.32	60.70 ± 6.52	0.378
Sex, male (%)	41 (49.40)	34 (46.58)	7 (70.00)	0.029
RBD	35 (42.17%)	30 (41.10%)	5 (50%)	0.041
Depression	15 (18.07%)	13 (17.81%)	2 (20%)	0.326
Anxiety	23 (27.71%)	19 (26.03%)	4 (40%)	0.012
Cognitive deficit	16 (19.28%)	14 (19.18%)	2 (20%)	0.143
Dysolfactory	54 (65.06%)	47 (64.38%)	7 (70%)	0.875
Constipation	17 (20.48%)	15 (20.55%)	2 (20%)	0.734
GBA	26 (33.77)	24 (34.78)	2 (25.00)	0.874
LRRK2	39 (50.65)	34 (49.28)	5 (62.50)	0.738

Note: Pearson Chi‐square was used to compare the difference in baseline variables between cases with a PD converter at follow‐up (N = 10) or without a new PD converter (N = 73). For categorical variables, the reported statistic is N (percentage), for numeric variables it is mean (standard deviation). Categorical variables were compared using Fisher's exact test unless otherwise noted. Numeric variables were compared using the Mann–Whitney U test. Statistically significant p < 0.05.

In multivariate Cox regression analysis, we included factors with p < 0.1 in univariate Cox regression, including ALPS, STAI score, RBDQ score, GBA, and LRRK2. The results showed that ALPS (HR = 0.87, p = 0.004), RBDQ score (HR = 1.83, p = 0.016), and STAI score (HR = 1.29, p = 0.027) were independent predictors of PD conversion (Table 3). Kaplan–Meier survival curves revealed that a higher DTI‐ALPS index was associated with a lower progression rate from pPD to clinical PD (p = 0.0079) (Figure 5).

TABLE 3.

Result of Cox regression analyses for potential predictors of PD.

Variables	Univariable analysis					Multivariable analysis
	β	SE	Z	p	HR (95% CI)	β	SE	Z	p	HR (95% CI)
Sex (male)	−0.88	0.69	−1.28	0.201	0.41 (0.11~1.60)
Age	−0.04	0.04	−0.90	0.366	0.96 (0.88~1.05)
Education	−0.11	0.09	−1.24	0.214	0.89 (0.75~1.07)
MoCA	−0.07	0.12	−0.57	0.570	0.94 (0.75~1.18)
UPSIT	0.03	0.06	0.48	0.630	1.03 (0.91~1.17)
RBDSQ	−0.21	0.13	−1.60	0.029	1.81 (1.63~2.05)	−0.18	0.13	−1.40	0.016	1.83 (1.64~2.08)
ESS	−0.01	0.08	−0.15	0.880	0.99 (0.84~1.16)
SCOPA AUT	−0.07	0.06	−1.22	0.223	0.93 (0.83~1.05)
STAI	0.11	0.07	1.46	0.045	1.11 (1.06~1.29)	0.08	0.08	1.11	0.027	1.29 (1.04~1.46)
GDS	0.12	0.27	0.45	0.652	1.13 (0.66~1.93)
UPDRS‐III	0.00	0.03	0.09	0.930	1.00 (0.95~1.06)
CSF α‐syn	0.00	0.00	0.24	0.807	1.00 (1.00~1.00)
CSF t‐Tau	−0.00	0.01	−0.24	0.813	1.00 (0.99~1.01)
CSF p‐Tau	−0.02	0.06	−0.35	0.726	0.98 (0.87~1.10)
GBA	−0.60	0.79	−0.75	0.152	0.55 (0.12~2.60)
LRRK2	0.70	0.69	1.02	0.108	2.02 (0.52~7.82)
DTI‐ALPS	−3.61	1.35	−2.67	0.008	0.85 (0.58~0.92)	−26.43	9.19	−2.88	0.004	0.87 (0.66~0.94)

Note: Statistically significant p < 0.05.

Abbreviations: CI, Confidence Interval; CSF, cerebrospinal fluid; DTI‐ALPS, diffusion tensor image analysis along the perivascular space; ESS, Epworth Sleepiness Scale; GDS, Geriatric Depression Scale; HR, Hazard Ratio; MoCA, Montreal Cognitive Assessment; PD, Parkinson's disease; p‐tau, tau phosphorylated; RBDSQ, REM Sleep Behavior Disorder Questionnaire Score; SCOPA‐AUT, Scale for Outcomes in Parkinson's Disease‐Autonomic; STAI, State–Trait Anxiety Inventory; UPDRS III, Unified Parkinson's Disease Rating Scale, Part III; UPSIT, University of Pennsylvania Smell Identification Test; α‐syn, alpha‐synuclein.

FIGURE 5.

Kaplan–Meier survival curves revealed higher DTI‐ALPS index levels showing a lower rate of progression from pPD to PD.

Multi‐model COX regression analysis showed that ALPS is an independent predictor of conversion from pPD to PD (Model 1: HR = 0.84, 95% CI: 0.81–0.90, p = 0.012; Model 2: HR = 0.84, 95% CI: 0.82–0.90, p = 0.006; Model 3: HR = 0.87, 95% CI: 0.82–0.92, p = 0.013; Model 4: HR = 0.87, 95% CI: 0.83–0.92, p = 0.018) (Table 4).

TABLE 4.

Association between ALPS and PD conversion.

Variables	Model 1		Model 2		Model 3		Model 4
	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p	HR (95% CI)	p
DTI‐ALPS	0.84 (0.81~0.90)	0.012	0.84 (0.82~0.90)	0.006	0.87 (0.82~0.92)	0.013	0.87 (0.83~0.92)	0.018

Note: Model 1: Crude; Model 2: Adjust: sex, age, education; Model 3: Adjust: sex, age, education, RBDSQ, STAI_anxiety_total; Model 4: Adjust: sex, age, education, RBDSQ, STAI_anxiety_total, GBA_PATHVAR, LRRK2_PATHVAR.

Abbreviations: CI, confidence interval; OR, odds ratio.

4. Discussion

This study assessed glymphatic integrity in pPD and its associations with clinical symptoms, cortical volume, and phenoconversion risk, utilizing combined cross‐sectional and longitudinal analyses The DTI‐ALPS index was applied as a non‐invasive proxy for glymphatic function evaluation, and the findings indicate that: (1) the DTI‐ALPS index was significantly reduced in the pPD group compared to healthy controls; (2) the DTI‐ALPS index was negatively associated with the total anxiety score of the State–Trait Anxiety Inventory (STAI) and positively correlated with cerebrospinal fluid (CSF) p‐tau and t‐tau levels in the pPD group; (3) the DTI‐ALPS index correlated with the cortical volume of the right temporal pole, left thalamus, right superior occipital gyrus and left posterior central gyrus; (4) Reduced DTI‐ALPS index was significantly associated with an elevated risk of phenoconversion to PD.

DTI‐ALPS is a recently proposed non‐invasive measurement method that quantitatively reflects the function of the glymphatic system. DTI‐ALPS has been validated in a previous human‐based study, which demonstrated a significant correlation between the ALPS index and the glymphatic function calculated by classical intrathecal injection of a contrast agent, indicating that the DTI‐ALPS index can reflect the clearing function of the glymphatic system [19]. Additionally, the measurement of DTI‐ALPS is simple and can yield results within a few minutes, allowing for the real‐time reflection of glymphatic system function [29]. DTI‐ALPS exhibits good stability, which has been confirmed in previous studies, and our research has also demonstrated good inter‐observer consistency. Therefore, DTI‐ALPS has been applied in the detection and monitoring of many clinical conditions, including Alzheimer's disease, hydrocephalus, diabetes, and Parkinson's disease, and has shown a correlation with the severity of clinical diseases [20, 30, 31, 32]. To the best of our knowledge, DTI‐ALPS has not yet been applied to the detection of pPD, especially for longitudinal follow‐up to predict the conversion of pPD to PD.

We observed a significantly reduced DTI‐ALPS index in the pPD cohort compared to HC. As the preclinical stage of PD, pPD typically spans decades and often evades early detection due to subtle non‐motor manifestations. This aligns with established evidence that neuropathological changes precede classical motor symptoms by years [33], supporting our inference that glymphatic dysfunction emerges during this prodromal phase. A previous study on the iRBD population also obtained similar results, showing that the ALPS index in the iRBD group was significantly lower than that in the HC group [23]. iRBD is a representative group of the general pPD population, and our PPD population contains 61.6% of iRBD patients, but our cohort may be more representative of the general pPD population. However, we also noticed that the ALPS index showed no significant difference between the PPD and PD groups. We analyzed the reasons that might be that most of the included PD population were early PD patients with H‐Y grade 1–2. The ALPS index did not have sufficient efficacy to show the difference in a certain sample size. Future investigations with expanded cohorts spanning the full disease continuum are warranted to clarify the temporal dynamics of glymphatic decline.

Our results showed that the DTI‐ALPS index was negatively correlated with the STAI score in the pPD group, after controlling for age, sex, and education. Previous surveys of non‐motor symptoms in people with pPD have shown that anxiety is a very common neuropsychiatric symptom of pPD [34]. Another animal study on mice after sleep deprivation showed that mice with chronic sleep restriction had reduced glymphatic function and showed significant anxiety‐like behavior. It suggests that dysfunction of the glymphatic system may play a role in the development of anxiety‐like behaviors in mice after chronic sleep restriction [35].

Additionally, with regard to the CSF biomarkers, the DTI‐ALPS index was positively correlated with p‐Tau and t‐Tau levels. As we know, Parkinson's disease is related to the abnormal aggregation and clearance of neurotoxic proteins, including Tau [36]. Tau is a microtubule‐associated protein, but it is also released physiologically into the extracellular fluid. The presence of Tau protein in cerebrospinal fluid indicates that it is ultimately cleared from the brain to the periphery. Recent animal studies have confirmed that extracellular Tau protein is eliminated from the brain to cerebrospinal fluid via an AQP4‐dependent mechanism, which suggests the involvement of the glymphatic system [37]. Our results also confirm this, as glymphatic dysfunction begins in the pPD period prior to PD, leading to impairment in the clearance of neurotoxic proteins such as Tau, thereby participating in the pathogenesis of pPD.

Notably, this study reveals novel associations between DTI‐ALPS index and regional gray matter atrophy in prodromal PD. Following adjustment for age, sex, education, and total intracranial volume, diminished DTI‐ALPS index correlated with volumetric changes in key regions: the left postcentral gyrus, right superior occipital gyrus, right temporal pole, and left thalamus. These territories collectively constitute fronto‐temporo‐occipital cortical networks and thalamic nuclei critically engaged in higher cognition and affective regulation—a neuroanatomical profile corroborated by extant literature. Community‐based studies demonstrate glymphatic dysfunction correlates with frontotemporal and thalamic gray matter loss in aging populations [38, 39], while human PET imaging reveals acute sleep deprivation induces amyloid‐β accumulation in thalamic and temporo‐occipital regions, mechanistically attributed to transient glymphatic suppression [40]. Our convergence with these independent observations establishes the fronto‐temporo‐occipital cortex and thalamus as selectively vulnerable territories to glymphatic compromise, suggesting conserved pathophysiological pathways across neurological states.

Our longitudinal findings demonstrate that a reduced DTI‐ALPS index independently predicts accelerated phenoconversion in prodromal PD, with each standard deviation decrease conferring a 13% increased risk (adjusted HR = 0.87 per SD increase, 95% CI 0.83–0.92, p = 0.018). This aligns with emerging evidence linking the ALPS index to PD progression rates [41] and extends its potential biomarker utility to the pre‐diagnostic phase. The temporal emergence of altered water diffusivity along perivascular spaces (as measured by the ALPS index) during the prodromal PD phase may suggest a pivotal pathophysiological mechanism, potentially involving impaired clearance of neurotoxic proteins (notably α‐synuclein), which could potentiate abnormal protein aggregation, ultimately accelerating phenoconversion to overt Parkinson's disease. The DTI‐ALPS index, as a rapid and non‐invasive neuroimaging technique, enables prospective identification of high‐risk individuals and could provide a potential mechanistic foundation for early therapeutic intervention. These insights illuminate two critical translational avenues: (1) development of novel therapies targeting brain clearance pathways and (2) optimization of interventional timing within biologically active windows before irreversible neuronal loss.

Our study had several limitations. First, the correlation between the ALPS index and human glymphatic function has not yet been substantially and rigorously verified through pathophysiological studies. Therefore, we should be cautious in interpreting the relationship between the ALPS index and glymphatic clearance, and our findings should be interpreted as showing an association with altered water diffusivity rather than a direct measure of glymphatic function. Although Zhang conducted a validation study on the ALPS method, they found a strong correlation between the ALPS index and the intrathecal contrast agent injection method for evaluating glymphatic function [19]. Second, as an indirect measure, the ALPS index can be influenced by other factors unrelated to glymphatic function, including microstructural changes, white matter hyperintensities, vascular alterations, and general atrophy patterns. Although we adjusted for key covariates, residual confounding by these or other unmeasured factors (e.g., vascular risk factors) remains a possibility and represents an important limitation of both our study and the DTI‐ALPS method itself. Third, it is currently unclear whether these pPD participants will ultimately develop PD as a result of short‐term follow‐up. This is a trap that has existed in all studies to date. Fourth, the follow‐up analysis of pPD subjects had a small sample size and a relatively short follow‐up time, and future studies will need a larger sample size and longer follow‐up time. Fifth, the pPD subjects in this study were a pPD population defined by PPMI, which was slightly different from the International Parkinson's Disease and Movement Disorders Association (MDS) research criteria for pPD (MDS‐pPD) [42], so the results need to be verified by the subjects who meet the MDS‐pPD criteria in the future. Finally, given these limitations and the observational nature of our study, the proposed mechanistic link between the ALPS index and phenoconversion risk must be considered speculative and hypothesis‐generating, warranting further investigation.

In summary, reduced ALPS index, which may reflect impaired glymphatic function, is present in pPD, correlates with symptoms and regional gray matter atrophy, and predicts phenoconversion to PD, supporting its potential as an early biomarker.

Author Contributions

Xinhua Wei: conceived and designed the experiments. Zhanyu Kuang and Junxiang Huang: contributed significantly to the experiments. Jinyu Wen and Yongzhou Xu: arranging data and performing data analyses. Amei Chen: wrote the draft manuscript. Pek‐Lan Khong and Yoon Seong Choi: revised the manuscript. All authors read and approved the final manuscript.

Ethics Statement

Data used in the preparation of this article were obtained on [2023‐08‐19] from the PPMI database (https://www.ppmi‐info.org/access‐data‐specimens/download‐data), RRID:SCR_006431. For up‐to‐date information on the study, visit http://www.ppmi‐info.org. All participating institutions' regional ethical committees approved the PPMI, and written informed consent was obtained from all study participants.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Names and abbreviations of 170 brain regions using the AAL3 template.

Chen A., Kuang Z., Khong P.‐L., et al., “Glymphatic Function in Prodromal Parkinson's Disease: Associations With Symptoms, Gray Matter Volume, and Phenoconversion Risk,” European Journal of Neurology 32, no. 10 (2025): e70389, 10.1111/ene.70389.

Data Availability Statement

Data used in the preparation of this study were obtained from the Parkinson's Progression Markers Initiative database. All data is full access and available at www.ppmi‐info.org.