Abstract
Brain glymphatic dysfunction is critical in neurodegenerative processes. While animal studies have provided substantial insights, understandings in humans remains limited. Recent attention has focused on the non‐invasive evaluation of brain glymphatic function. However, its association with brain parenchymal lesions in large‐scale population remains under‐investigated. In this cross‐sectional analysis of 1030 participants (57.14 ± 9.34 years, 37.18% males) from the Shunyi cohort, we developed an automated pipeline to calculate diffusion‐weighted image analysis along the perivascular space (ALPS), with a lower ALPS value indicating worse glymphatic function. The automated ALPS showed high consistency with the manual calculation of this index (ICC = 0.81, 95% CI: 0.662–0.898). We found that those with older age and male sex had lower automated ALPS values (β = −0.051, SE = 0.004, p < .001, per 10 years, and β = −0.036, SE = 0.008, p < .001, respectively). White matter hyperintensity (β = −2.458, SE = 0.175, p < .001) and presence of lacunes (OR = 0.004, 95% CI < 0.002–0.016, p < .001) were significantly correlated with decreased ALPS. The brain parenchymal and hippocampal fractions were significantly associated with decreased ALPS (β = 0.067, SE = 0.007, p < .001 and β = 0.040, SE = 0.014, p = .006, respectively) independent of white matter hyperintensity. Our research implies that the automated ALPS index is potentially a valuable imaging marker for the glymphatic system, deepening our understanding of glymphatic dysfunction.
Keywords: brain atrophy, diffusion tensor imaging, diffusion‐weighted image analysis along the perivascular space, glymphatic system, vascular brain damage
We evaluated the glymphatic system function in the brain using a non‐invasive method, calculation of automated diffusion‐weighted image analysis along the perivascular space (ALPS) index, and found that dysfunction of the glymphatic system potentially influences both vascular and degenerative brain parenchymal lesions in a community‐dwelling cohort.
1. INTRODUCTION
The brain glymphatic system, first identified in mice as an influx and efflux pathway through the paravascular spaces (PVS) using fluorescent tracers (Iliff et al., 2012), has been proposed as a critical mechanism in the brain waste clearance system (Iliff et al., 2012; Nedergaard, 1979) and is related to neurodegenerative processes. However, its structural features, pathological mechanisms, and clinical relevance have not been fully elucidated in vivo. Tracer‐based methods have been used in magnetic resonance imaging (MRI) to assess brain glymphatic function in animals (Jiang et al., 2017; Mortensen et al., 2019) and humans (Zhang, Zhou, et al., 2021; Zhou et al., 2020). Alternatively, the diffusion along the perivascular space (ALPS) index measures diffusivity along the PVS taking advantage of the perpendicular relationship between veins and major fiber bundles in the periventricular regions. With a growing knowledge of the anatomical basis of the brain lymphatic system, this index has been applied in several researches and acknowledged as an imaging marker for brain glymphatic function (Tian et al., 2023).
Previous studies have found a link between the ALPS index and MRI markers of cerebral small vessel disease (CSVD) (Tian et al., 2023; Zhang, Zhou, et al., 2021), implying a potential relationship between small brain vessels and the glymphatic system. Moreover, an impaired glymphatic system is associated with the accumulation of tau (Harrison et al., 2020) and amyloid β‐proteins (Aβ) (Iliff et al., 2012) in mice; hence, investigating the association between neurodegeneration markers and ALPS in humans is important. Taoka et al. (2017) identified a relationship between cognitive function and ALPS index in 31 patients with Alzheimer's disease (AD) (Taoka et al., 2017). However, whether the ALPS index is associated with brain parenchymal atrophy was less discussed (Hsiao et al., 2023; Zhang, Zhang, et al., 2021).
In this study, we aimed to develop an automated method for the ALPS calculation and validate its consistency using the ALPS index obtained manually. Additionally, we investigated the relationships of the ALPS index and choroid plexus volume with brain parenchymal damage, and vascular‐origin lesions in a community‐based population.
2. MATERIALS AND METHODS
2.1. Participants
In this cross‐sectional investigation, we evaluated a cohort of 1030 non‐demented participants from the Shunyi Study (Han et al., 2020), a community‐based initiative in Beijing's Shunyi District (Figure 1). This study was conducted between June 2013 and April 2017, initially enrolling 1586 individuals aged ≥35 years. A subset of 1222 participants underwent MRI scans suitable for structural segmentation, diffusion MRI (dMRI) analysis, and automated ALPS calculation. Additionally, 1152 participants consented to the cognitive evaluations, from which 9 individuals with a dementia diagnosis were excluded. A total of 113 participants were excluded because of stroke (n = 50) or tumor history (n = 63), resulting in a final cohort of 1030 participants for the analysis. The study protocol was approved by the Ethics Committee at Peking Union Medical College Hospital (reference no. B‐160), and all participants provided written informed consent.
FIGURE 1.
Flow chart of the study population.
2.2. Demographic information and risk factors assessments
Demographic information was collected through structured interviews conducted by trained interviewers. Vascular risk factors such as hypertension, dyslipidemia, diabetes, smoking status, and alcohol consumption were assessed based on self‐reported history and laboratory tests. Detailed information can be found in Appendix S1 and our previous publication (Zhai et al., 2018).
2.3. MRI acquisition and processing
All participants underwent brain MRI using a 3 T MRI scanner. The MRI protocol included a T1‐weighted, a T2‐weighted fluid‐attenuated inversion recovery (FLAIR) sequence, susceptibility‐weighted images (SWI), and a single‐shell diffusion tensor image (DTI). The structural and dMRI data were processed using the UK Biobank processing pipeline (Alfaro‐Almagro et al., 2018) (https://git.fmrib.ox.ac.uk/falmagro/UK_biobank_pipeline_v_1). Further details regarding MRI acquisition and processing are available in our previous publication (Han et al., 2020).
2.4. DTI processing
The original DTI data were first corrected for eddy currents, head motion, and gradient distortions. The outputs were then inputted into the DTI fitting tool (DTIFIT, part of the FSL (Jenkinson et al., 2012)) to create fractional anisotropy (FA) maps and tensor images. The tensor images were divided into six directions. Nonlinear registrations of FMRIB58_FA_1mm, a white matter (WM) skeleton map in the MNI152 space, to the FA images, were additionally conducted to obtain the transformation matrices from the standard space to the FA images for each participant.
2.5. Evaluations of imaging markers
The definitions of CSVD imaging markers have been previously described by Zhai et al. (2018). Briefly, lacunes were defined as focal fluid‐intensity cavities, 3–15 mm in diameter, located in the basal ganglia (BG), subcortical WM, or brainstem. Cerebral microbleeds (CMBs) were defined as small, round, hypointense lesions on SWI (Wardlaw et al., 2013). The severity of enlarged PVS (EPVS) in the BG (EPVS‐BG) and WM (EPVS‐WM) was evaluated using a previously established 4‐level severity score on the T1‐weighted images (Zhu et al., 2010). The number of deep medullary veins (DMVs) was assessed by visual quantification of the SWI images in the region of interest (ROI) placed in the periventricular WM between the frontal and occipital horns of each cerebral hemisphere, in accordance with our previous study (Ao et al., 2021). The intra‐rater agreements for the lacunes, CMBs, EPVS‐BG, EPVS‐WM, and number of DMVs were 0.95, 0.90, 0.75, 0.67, and 0.68, respectively.
White matter hypersensitivity (WMH) was automatically segmented using a Brain Intensity Abnormality Classification Algorithm (BIANCA) (Griffanti et al., 2016). The area of the WMH was first manually labeled on the FLAIR sequence of 25 participants as training data, and then a probability map thresholded at 0.95 as a binary map of lesions was produced using BIANCA. The WMH volume was acquired from the WMH probability map. In our dataset, the Dice coefficient was 0.87.
The gray matter (GM), WM, cerebrospinal fluid (CSF), total intracranial volume (TIV), and choroid plexus volume were automatically acquired from structural T1‐weighted images using FreeSurfer (http://surfer.nmr.mgh.harvard.edu/) (Fischl, 2012). The brain parenchymal fraction was defined as the ratio of brain tissue volume (GM and WM volume) to the TIV. The hippocampal fraction was calculated as the ratio of the hippocampal volume to TIV.
2.6. ALPS and automated ALPS indices calculation
The ALPS index was calculated in 40 participants based on dMRI data using the previously established method (Taoka et al., 2017). ROIs with a 3 mm radius were manually placed in the area of projection and association fibers on both hemispheres according to a color‐coded FA map of the plane at the level of the lateral ventricle body. The ALPS index was calculated as the ratio of the mean x‐axis diffusivity in the ROI of projection fibers (Dxxproj) and the ROI of association fibers (Dxxassoc) to the mean y‐axis diffusivity in the ROI of projection fibers (Dyyproj) and z‐axis diffusivity in the ROI of association fibers (Dzzassoc). The ALPS indices of both hemispheres were calculated separately and averaged in this study.
The details of the automated ALPS calculations are shown in Figure 2. First, spherical ROIs (sROIs) with a radius of 3 mm were placed on the MNI152 space centered at (37, −12, 24), (26, −12, 24), (−26, −12, 24), and (−37, −12, 24), selected according to manual observations of the bilateral projection and association fibers. Projection fiber (superior longitudinal fasciculus) and association fiber (corticofugal corona radiata tract fibers) masks were selected based on the JHU‐ICBM‐labels‐1 mm template in the MNI152 space. The sROI mask, projection and association fiber masks were transformed into the FA space for each participant. For each hemisphere, the regions of the transformed association and projection fibers in the maximum range of the y‐ and z‐axis coordinates of the sROI were separately extracted for automated ALPS calculations in each participant's space. The mean Dxxproj, Dxxassoc, Dyyproj, and Dzzassoc values were used to calculate the automated ALPS index using the same formula as the ALPS calculation.
FIGURE 2.
Calculation of automated ALPS. This flowchart illustrates the automated APLS calculation process for the left hemisphere of a random participant. First, the projection fibers, association fibers, and sROIs were determined on the JHU‐ICBM template in the standard space and transformed into the participant's FA space. Subsequently, the sROI planes and rectangular cuboid areas restricted by the maximum ranges of the y‐ and z‐axis coordinates of the sROI were extracted for each hemisphere. Finally, the intersection of the sROI planes and the transformed fiber tract area were selected as ROIs for automated ALPS calculation. ALPS, along the perivascular space; FA, fractional anisotropy; ROI, region of interest; sROI, spherical ROI.
The intraclass correlation coefficient (ICC) was used to ascertain the degree of conformity between the two methods.
2.7. Statistical analysis
Demographic, clinical, and imaging data were summarized using descriptive statistics. Associations between demographic characteristics, vascular risk factors, continuous CSVD imaging marker variables, automated ALPS, and choroid plexus volume were evaluated using linear regression models. The association between incontinuous CSVD imaging marker variables, ALPS, and choroid plexus volume was analyzed using logistic regression models. Logarithmic transformations were applied for variables with skewed distributions. All models were adjusted for age and sex. In analyses involving brain structure variables, vascular risk factors, including hypertension, diabetes, smoking, a history of alcohol consumption, and APOE ε4 carriership, were further adjusted for. The assumptions of the above models were verified using residual diagnosis plots. T‐tests were used to examine whether there were any demographic differences between the excluded participants or participants with a manually calculated ALPS index and those included in the analysis. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC), with a two‐tailed significance level of p < .05.
2.8. Code accessibility
All code used in the data processing and analysis will be fully accessible upon acceptance and publication of the manuscript.
3. RESULTS
3.1. Population characteristics
The demographic characteristics, vascular risk factors, brain structural measurements, and automated ALPS indices of the 1030 non‐demented participants are presented in Table 1. The average age of the participants was 57.14 ± 9.34 years (males, 37.18%). The mean automated ALPS index was 1.41 ± 0.13. The characteristics of the included and excluded populations were balanced (Table S1).
TABLE 1.
Characteristics of participants.
| Variables | Total (n = 1030) |
|---|---|
| Demographic characteristics | |
| Age, mean (SD), year | 57.14 (9.34) |
| Male, no. (%) | 383 (37.18) |
| Automated ALPS | 1.41 (0.13) |
| Vascular risk factors | |
| BMI, mean (SD), kg/m2 | 26.44 (3.78) |
| Systolic BP, mean (SD), mmHg | 133 (19.10) |
| Diastolic BP, mean (SD), mmHg | 78.54 (10.66) |
| Hypertension, no. (%) | 533 (51.75) |
| Diabetes mellitus, no. (%) | 164 (15.92) |
| Hyperlipidemia, no. (%) | 499 (48.45) |
| Smoking history, no. (%) | 295 (29.15) |
| Alcohol consumption history, no. (%) | 275 (27.17) |
| ApoE ε4 carrier, no. (%) | 146 (15.35) |
| Brain structural measurements | |
| Choroid plexus volume, median (interquartile range), mL | 18.35 (13.09, 24.99) |
| Brain parenchymal fraction, mean (SD) | 0.70 (0.04) |
| Hippocampus fraction, mean (SD), % | 0.74 (0.06) |
| WMH volume, median (interquartile range), mL | 3.01 (1.94, 5.36) |
| Existence of CMBs, no. (%) | 96 (9.32) |
| Number of CMB (in participants with CMB), median (IQR) | 1.0 (1.0, 2.5) |
| Existence of lacunes, no. (%) | 151 (14.66) |
| DMVs number, mean (SD) | 55.10 (1.70) |
| Severe EPVS in BG, a no. (%) | 142 (13.83) |
| Severe EPVS in WM, a no. (%) | 152 (14.80) |
Abbreviations: ALPS, assessment of diffusion tensor imaging along the perivascular space; BG, basal ganglia; BMI, body mass index; BP, blood pressure; CMBs, cerebral microbleeds; DMVs, deep medullary veins; EPVS, enlarged perivascular space; no., number; SD, standard deviation; WM, white matter; WMH, white matter hyperintensity.
Severe EPVS in the BG or WM was defined as an EPVS of level 3 or 4.
3.2. Reliability of automated ALPS index
To examine the consistency and comparability of the ALPS and automated ALPS indices, we utilized MRI scans from a randomly chosen subset of 40 participants for manual computation of the ALPS index. The ICC was 0.81 (95%CI 0.662–0.898). A scatter plot of ALPS and automated ALPS indices is shown in Figure S1, indicating a high level of consistency between the ALPS and automated ALPS indices.
T‐tests revealed no statistically significant differences in the demographic information between the 40 selected participants and the entire population included in the analysis (Table S2).
3.3. Risk factors of reduced ALPS
The associations between demographic characteristics, vascular risk factors, and automated ALPS are shown in Table 2. A lower ALPS level was significantly associated with increased age (per 10 years, β = −0.051, p < .001), male sex (β = −0.036, p < .001), hypertension (β = −0.019, p = .022), diabetes mellitus (β = −0.033, p = .004), smoking (β = −0.033, p = .001), and alcohol consumption history (β = −0.029, p = .004) (Model 1). After adjusting for age and sex (Model 2), the associations between ALPS and all traditional vascular risk factors became insignificant.
TABLE 2.
Associations between demographic characteristics, vascular risk factors, and automated ALPS.
| Variables | Model 1 | Model 2 | ||
|---|---|---|---|---|
| β (SE) × 10−2 | p | β (SE) × 10−2 | p | |
| Demographic characteristics | ||||
| Age (per 10‐year increase) | −5.074 (0.410) | <.001 | −4.996 (0.408) | <.001 |
| Male sex | −3.565 (0.842) | <.001 | −3.096 (0.788) | <.001 |
| Vascular risk factors | ||||
| BMI | 0.180 (0.110) | .103 | 0.020 (0.103) | .847 |
| Hypertension | −1.880 (0.819) | .022 | 0.601 (0.785) | .444 |
| Diabetes mellitus | −3.251 (1.117) | .004 | −1.229 (1.052) | .243 |
| Hyperlipidemia | −0.523 (0.821) | .524 | 0.178 (0.779) | .820 |
| Smoking history | −3.315 (0.909) | .001 | −0.268 (1.301) | .837 |
| Alcohol consumption history | −2.891 (0.930) | .004 | −0.468 (1.161) | .687 |
Note: Model 1 is univariate. Model 2 was adjusted for age and sex. Age and sex were adjusted for other variables. Bold values indicate p < .05.
Abbreviations: ALPS, assessment of diffusion tensor imaging along the perivascular space; BMI, body mass index; SE, standard error.
3.4. Associations between brain imaging markers and automated ALPS
The associations between brain atrophy and automated ALPS are shown in Table 3 and Figure 3. The results indicated that brain (β = 0.130, p < .001) and hippocampal atrophy (β = 0.082, p < .001) were related to a diminished automated ALPS value. These associations remained stable in all models, indicating a strong relationship between brain atrophy and impaired glymphatic function independent of vascular‐origin lesions.
TABLE 3.
Associations between automated ALPS and brain atrophy.
| Variable | Model 1 | Model 2 | Model 3 | Model 4 | ||||
|---|---|---|---|---|---|---|---|---|
| β (SE) | p | β (SE) | p | β (SE) | p | β (SE) | p | |
| Brain parenchymal fraction | 0.130 (0.008) | <.001 | 0.073 (0.007) | <.001 | 0.069 (0.007) | <.001 | 0.067 (0.007) | <.001 |
| Hippocampus fraction | 0.082 (0.014) | <.001 | 0.046 (0.014) | .001 | 0.046 (0.014) | .001 | 0.040 (0.014) | .006 |
Note: Model 1 is univariate. Model 2 was adjusted for age and sex. Model 3 was adjusted for age, sex, and vascular risk factors, including hypertension, diabetes, hyperlipidemia, smoking, alcohol consumption history, and ApoE ε4 carriership. Model 4 was adjusted for age, sex, and logWMH volume. Bold values indicate p < .05.
Abbreviations: ALPS, along the perivascular space; CI, confidence interval.
FIGURE 3.
Associations between brain and hippocampal atrophy with values and residuals of automated ALPS index. Relationships between (a) the brain parenchymal fraction and automated ALPS index; (b) the brain parenchymal fraction and residuals from the linear regression of the automated ALPS index when adjusted for age, sex, and WMH volume; (c) the hippocampal fraction and automated ALPS index; and (d) the hippocampal fraction and residuals from the linear regression of the automated ALPS index when adjusted for age, sex, and WMH volume. Each plot shows the fitted regression line, the 95% confidence limit, and the 95% prediction limit. ALPS, along the perivascular space; SE, standard deviation.
As shown in Table 4, CSVD imaging markers, including WMH volume (β = −2.458, p < .001), existence and number of CMB (β = −0.236, p = .001, and β = −2.750, p < .001), presence of lacunes (β = −0.644, p < .001), DMV number (β = 1.777, p < .001), and EPVS in the BG (β = −2.093, p = .011), were associated with lower automated ALPS in the crude model. After adjusting for demographic characteristics and vascular risk factors (Model 3), associations between larger WMH volume (β = −1.325, p = .167), CMB number (β = −2.466, p = .001), existence of lacunes (β = −0.365, p < .001), and lower automated ALPS values remained significant. With the adjustment of age, sex, and WMH volume, only the association between DMV (β = 1.472, p = .003) and CMB number (β = −1.632, p = .043) remained significant.
TABLE 4.
Associations between automated ALPS and CSVD imaging markers.
| Variable | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| β/OR | p | β/OR | p | β/OR | p | |
| (SE)/(95%CI) | (SE)/(95%CI) | (SE)/(95%CI) | ||||
| WMH volume | −2.458 (0.175) | <.001 | −1.241 (0.161) | <.001 | −1.325 (0.167) | <.001 |
| With vs. without lacunes | 0.004 (<0.001, 0.016) | <.001 | 0.035 (0.007, 0.181) | <.001 | 0.033 (0.006, 0.187) | .001 |
| With vs. without CMBs | 0.057 (0.011, 0.297) | .001 | 0.616 (0.097, 3.908) | .607 | 0.563 (0.084, 3.767) | .553 |
| Number of CMBs | −2.750 (0.662) | <.001 | −2.440 (0.710) | .001 | −2.481 (0.747) | .001 |
| With vs. without severe EPVS in BG a | 0.169 (0.043, 0.665) | .011 | 0.736 (0.162, 3.346) | .692 | 0.786 (0.167, 3.709) | .761 |
| With vs. without severe EPVS in WM a | 0.500 (0.134, 1.866) | .302 | 1.463 (0.344, 6.271) | .607 | 1.800 (0.390, 8.313) | .452 |
| Number of DMVs | 1.777 (0.437) | <.001 | 1.508 (0.473) | .002 | 1.497 (0.494) | .003 |
Note: Model 1 is univariate. Model 2 was adjusted for age and sex. Model 3 was adjusted for age, sex, and vascular risk factors, including hypertension, diabetes, hyperlipidemia, smoking, alcohol consumption history, and ApoE ε4 carriership. Bold values indicate p < .05.
Abbreviations: BG, basal ganglia; CI, confidence interval; CMBs, cerebral microbleeds; DMVs, deep medullary veins; EPVS, enlarged perivascular space; OR, odds ratio; vs., versus; SE, standard error; WM, white matter; WMH, white matter hyperintensity.
Severe EPVS in the BG or WM was defined as an EPVS of level 3 or 4.
3.5. Using the choroid plexus volume to evaluate glymphatic system function
In addition to the ALPS index, we also calculated the choroid plexus volume, a useful indicator of the glymphatic system, and assessed its relationship with the ALPS index, age, sex, vascular risk factors, brain atrophy, and CSVD imaging markers (Tables S3–S5). The results showed a significant correlation between choroid plexus volume and the ALPS index (r = 0.557, p < .001). Consistent with the ALPS index results, higher choroid plexus volume was associated with older age (β = 0.456, p < .001) and male sex (β = 6.192, p < .001), and was widely associated with brain atrophy and CSVD imaging markers.
4. DISCUSSION
In this community‐based population of over 1000 individuals, we developed an automatic ALPS index calculation method and found that a diminished ALPS index was related to increased age, male sex, and heavier burden of CSVD markers. Notably, lower ALPS levels were associated with widespread brain atrophy after adjusting for age, sex, vascular risk factors, and CSVD burden, suggesting a potential relationship between the cerebral glymphatic system and brain parenchymal destruction, independent of vascular risk factors and vascular‐origin brain lesions.
4.1. Automated method for ALPS calculation
The value of the ALPS index largely depends on the placement of the ROI, as the physiological implications are built on the assumed efflux pathways along the perivascular space perpendicular to the projection and association fibers. Thus, precise labeling of these fibers is crucial for minimizing the confounding effects of fiber bundles on the diffusion indices. In the process of manual ROI selection, which relies on the visual judgment and anatomical knowledge of a single neurologist, inconsistencies within and between individual annotators may inevitably occur, posing challenges when applying this method in larger populations. In our study, we proposed a standardized and practicable method for ROI selection, which restricts the ROI to fiber bundles segmented in the atlas to avoid potential deviations, thereby ensuring that the results are consistent and free from individual bias. Previous studies have developed automated calculation techniques for ALPS (Saito et al., 2023; Zhang, Zhang, et al., 2021), which register diffusion maps in the participant's space to a standard FA template, and perform the ROI placement and ALPS index calculation in the standard space. However, the deformation introduced during the transformation process may affect the diffusivity of the tensor images, leading to potential inaccuracies. To avoid this problem, we acquired the automated ALPS index in each participant's space. Only the shape and size of the ROI underwent transformations in this process, which do not affect the ALPS index calculation (Taoka, Ito, Nakamichi, Kamagata, et al., 2022), ensuring the reliability of the automated method.
4.2. Risk factors of lower ALPS
In our study, the ALPS index was strongly associated with older age and male sex. For a 10‐year increase in age, there was a decline in the ALPS value of 0.05, approximately 3% of the mean ALPS value in our population, and males possessed 2.5% lower ALPS values than females. The similar relationship between increased age and a lower ALPS index has also been reported in previous studies (Taoka, Ito, Nakamichi, Nakane, et al., 2022; Zhang, Zhang, et al., 2021).
Accumulating evidence from animal studies suggests that the brain glymphatic system is affected by aging through various mechanisms (Kress et al., 2014), including a decrease in polarized aquaporin‐4 channels on the endfeet of astrocytes, decreased CSF production (Liu et al., 2020) and CSF pressure, and diminished arterial pulsatility (Valenza et al., 2020). Although a lower ALPS index is consistently associated with male sex (Hsiao et al., 2023; Tian et al., 2023; Zhang, Zhang, et al., 2021), the role of sex in glymphatic system function remains unclear. Animal studies (Han et al., 2023; Liu et al., 2020) have reported conflicting results regarding the existence of sex differences in glymphatic system function. The relationship between sex and the glymphatic system remains unclear and needs to be explained via further research.
In the present study, we also observed correlations between lower ALPS values and vascular risk factors, including hypertension, diabetes, smoking, and alcohol consumption history. However, these associations did not remain significant after adjusting for age and sex. Most previous studies have also presented similar results, in that most associations became unstable and lacked consistency between studies after adjusting for different variables (Tang et al., 2022; Zhang, Zhang, et al., 2021). This result indicates that the impact of vascular risk factors on the brain's glymphatic system might be lacking or at least weak compared to that of age and sex. Therefore, therapeutic or preventative methods targeting vascular risk factors might be ineffective for brain glymphatic system dysfunction.
4.3. Relationship between ALPS and brain parenchymal lesions
Our analysis revealed a prevalent association between reduced automated ALPS levels and brain atrophy at the whole brain level. Our results align with those of previous studies that demonstrated associations between gray matter atrophy and ALPS in a cohort of 84 healthy older individuals, as well as in a diverse cohort of 228 participants ranging in age from 13 to 88 years (Hsiao et al., 2023; Siow et al., 2022). Moreover, we discovered that the association between the brain and hippocampal atrophy was independent of age, sex, vascular risk factors, and vascular‐origin brain damage, indicating an intrinsic link between brain parenchymal lesions and glymphatic function independent of vascular factors. This finding likely supports the hypothesis that the glymphatic system of the brain is involved in neurodegenerative pathophysiological processes. The possible mechanisms include the accumulation of harmful substances within the brain in the context of impaired glymphatic function, such as soluble Aβ and tau oligomers, which can cause neuronal lesions and subsequently lead to brain atrophy and pathological changes in neurodegenerative diseases (Tarasoff‐Conway et al., 2015). Notably, we noticed that a lower automated ALPS index was related to hippocampal atrophy, a characteristic early change in AD, further supporting the association between neurodegeneration and glymphatic dysfunction, as well as the potential of ALPS for the early detection of neurodegenerative diseases.
Investigating the association between CSVD imaging markers and ALPS is important for understanding the relationship between small cerebral vessels and the glymphatic system and its crosstalk. In our study, we validated the significant association between diminished ALPS and the severity of CSVD‐related brain damage, such as WMH, lacunes, and CMBs, which has been consistently reported in previous studies (Liu et al., 2022; Tian et al., 2023; Zhang, Zhou, et al., 2021). Additionally, we directly observed periventricular venules using SWI and found that a decreased number of DMVs was significantly associated with decreased ALPS. Because parenchymal venules have been identified as components of the glymphatic system efflux pathway (Iliff et al., 2012), this result further supports the potential relationship between brain venules and the glymphatic system. Notably, EPVS, recognized as a sign of glymphatic fluid stasis, has been hypothesized to be linked to the functioning of the glymphatic system (Yu et al., 2022). However, direct evidence for this is lacking. In our study, after further adjusting for age and sex, we did not observe a significant correlation between the severity of visible PVS enlargement and ALPS index values. This finding challenges the previous hypothesis that posits a direct physiological and anatomical correlation between EPVS and glymphatic dysfunction. Overall, our results confirmed the functional involvement of cerebral small vessels, especially venules, in the glymphatic system; however, we were unable to prove that PVS enlargement is a sign of glymphatic dysfunction.
The potential cause of brain atrophy in patients with CSVD has been a topic of speculation for several years. In this study, we found a notable correlation among CSVD, glymphatic system dysfunction, and brain atrophy, which warrants further investigations to elucidate the underlying causal relationship.
The strengths of our study include the use of an automatic method for the efficiency and stability of ALPS calculation and the large sample size of the general population.
This study has several limitations. First, the ALPS index is an indirect assessment of the brain glymphatic system function and only reflects changes in the periventricular space instead of the whole brain. Other methods, such as choroid plexus volume and perivascular space volume, should to be used together to validate the ability of this index to explain glymphatic system function and further support the proposed associations. Second, although the ALPS index is highly correlated with gadolinium‐enhanced clearance rates and choroid plexus volume, indicating the ability of ALPS index to reflect glymphatic system function, the results reported by Wright et al. (2023) suggest that the calculation of the ALPS index dose not completely exclude the influence of white matter tracts. Future studies are needed to refine the methodology of the ALPS index to eliminate this influence. Third, DMVs tend to become tortuous during brain aging, owing to blood pressure changes and collagenosis‐related stenosis, and the perpendicular condition between the perivenous pathways and major fiber bundles could be altered, which should be considered in future studies. Finally, further studies are required to confirm our findings in other populations.
5. CONCLUSION
Our findings indicate that dysfunction of the glymphatic system may play an important role in both vascular and degenerative brain parenchymal lesions. The ALPS index is potentially a valuable imaging marker for the internal brain environment and may help understand the interaction between CSVD, glymphatic system dysfunction, and cerebral degeneration.
FUNDING INFORMATION
This study was supported by the CAMS Innovation Fund for Medical Sciences (CIFMS 2022‐I2M‐1‐001), National High Level Hospital Clinical Research Funding (grant number: 2022‐PUMCH‐D‐007) and the Strategic Priority Research Program (Biological Basis of Aging and Therapeutic Strategies) of the Chinese Academy of Sciences (grant XDB39040300).
CONFLICT OF INTEREST STATEMENT
The authors declare that this study was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest.
PATIENT CONSENT STATEMENT
All participants included in the study have provided written informed consent.
Supporting information
Appendix S1: Supporting information.
ACKNOWLEDGMENTS
Anonymized data will be made available to eligible investigators upon request for sole purpose of verifying the methods and findings in the article after ethics clearance and consent from the entire project team.
Li, W.‐X. , Liu, Z.‐Y. , Zhai, F.‐F. , Han, F. , Li, M.‐L. , Zhou, L.‐X. , Ni, J. , Yao, M. , Zhang, S.‐Y. , Cui, L.‐Y. , Jin, Z.‐Y. , & Zhu, Y.‐C. (2024). Automated diffusion‐weighted image analysis along the perivascular space index reveals glymphatic dysfunction in association with brain parenchymal lesions. Human Brain Mapping, 45(11), e26790. 10.1002/hbm.26790
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
REFERENCES
- Alfaro‐Almagro, F. , Jenkinson, M. , Bangerter, N. K. , Andersson, J. L. R. , Griffanti, L. , Douaud, G. , Sotiropoulos, S. N. , Jbabdi, S. , Hernandez‐Fernandez, M. , Vallee, E. , Vidaurre, D. , Webster, M. , McCarthy, P. , Rorden, C. , Daducci, A. , Alexander, D. C. , Zhang, H. , Dragonu, I. , Matthews, P. M. , … Smith, S. M. (2018). Image processing and quality control for the first 10,000 brain imaging datasets from UK Biobank. NeuroImage, 166, 400–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ao, D.‐H. , Zhang, D.‐D. , Zhai, F.‐F. , Zhang, J. T. , Han, F. , Li, M. L. , Ni, J. , Yao, M. , Zhang, S. Y. , Cui, L. Y. , Jin, Z. Y. , Zhou, L. X. , & Zhu, Y. C. (2021). Brain deep medullary veins on 3‐T MRI in a population‐based cohort. Journal of Cerebral Blood Flow & Metabolism, 41, 561–568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fischl, B. (2012). FreeSurfer. Neuroimage, 62, 774–781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Griffanti, L. , Zamboni, G. , Khan, A. , Li, L. , Bonifacio, G. , Sundaresan, V. , Schulz, U. G. , Kuker, W. , Battaglini, M. , Rothwell, P. M. , & Jenkinson, M. (2016). BIANCA (Brain Intensity AbNormality Classification Algorithm): A new tool for automated segmentation of white matter hyperintensities. NeuroImage, 141, 191–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Han, F. , Liu, X. , Yang, Y. , & Liu, X. (2023). Sex‐specific age‐related changes in glymphatic function assessed by resting‐state functional magnetic resonance imaging. bioRxiv. 10.1101/2023.04.02.535258 [DOI] [Google Scholar]
- Han, F. , Zhou, L.‐X. , Ni, J. , Yao, M. , Zhai, F. F. , Liu, Y. T. , Wu, W. , Xue, H. D. , Li, M. L. , Yang, M. , Dai, Q. , Cui, L. Y. , Jin, Z. Y. , Zhu, Y. C. , & Zhang, S. Y. (2020). Design of the Shunyi study on cardiovascular disease and age‐related brain changes: A community‐based, prospective, cohort study. Annals of Translational Medicine, 8, 1579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harrison, I. F. , Ismail, O. , Machhada, A. , Colgan, N. , Ohene, Y. , Nahavandi, P. , Ahmed, Z. , Fisher, A. , Meftah, S. , Murray, T. K. , Ottersen, O. P. , Nagelhus, E. A. , O'Neill, M. J. , Wells, J. A. , & Lythgoe, M. F. (2020). Impaired glymphatic function and clearance of tau in an Alzheimer's disease model. Brain, 143, 2576–2593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hsiao, W.‐C. , Chang, H.‐I. , Hsu, S.‐W. , Lee, C. C. , Huang, S. H. , Cheng, C. H. , Huang, C. W. , & Chang, C. C. (2023). Association of Cognition and Brain Reserve in aging and Glymphatic function using diffusion tensor image‐along the perivascular space (DTI‐ALPS). Neuroscience, 524, 11–20. [DOI] [PubMed] [Google Scholar]
- Iliff, J. J. , Wang, M. , Liao, Y. , Plogg, B. A. , Peng, W. , Gundersen, G. A. , Benveniste, H. , Vates, G. E. , Deane, R. , Goldman, S. A. , Nagelhus, E. A. , & Nedergaard, M. (2012). A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid β. Science Translational Medicine, 4, 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jenkinson, M. , Beckmann, C. F. , Behrens, T. E. J. , Woolrich, M. W. , & Smith, S. M. (2012). Fsl. Neuroimage, 62, 782–790. [DOI] [PubMed] [Google Scholar]
- Jiang, Q. , Zhang, L. , Ding, G. , Davoodi‐Bojd, E. , Li, Q. , Li, L. , Sadry, N. , Nedergaard, M. , Chopp, M. , & Zhang, Z. (2017). Impairment of the glymphatic system after diabetes. Journal of Cerebral Blood Flow and Metabolism, 37, 1326–1337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kress, B. T. , Iliff, J. J. , Xia, M. , Wang, M. , Wei, H. S. , Zeppenfeld, D. , Xie, L. , Kang, H. , Xu, Q. , Liew, J. A. , Plog, B. A. , Ding, F. , Deane, R. , & Nedergaard, M. (2014). Impairment of paravascular clearance pathways in the aging brain. Annals of Neurology, 76, 845–861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liu, G. , Mestre, H. , Sweeney, A. M. , Sun, Q. , Weikop, P. , Du, T. , & Nedergaard, M. (2020). Direct measurement of cerebrospinal fluid production in mice. Cell Reports, 33, 108524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liu, H. , Yang, S. , He, W. , Liu, X. , Sun, S. , Wang, S. , Wang, Y. , Zhou, X. , Tang, T. , Xia, J. , Liu, Y. , & Huang, Q. (2022). Associations Among Diffusion Tensor Image Along the Perivascular Space (DTI‐ALPS), enlarged perivascular space (ePVS), and cognitive functions in asymptomatic patients with carotid plaque. Frontiers in Neurology, 12, 789918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mortensen, K. N. , Sanggaard, S. , Mestre, H. , Lee, H. , Kostrikov, S. , Xavier, A. L. R. , Gjedde, A. , Benveniste, H. , & Nedergaard, M. (2019). Impaired glymphatic transport in spontaneously hypertensive rats. Journal of Neuroscience, 39, 6365–6377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nedergaard, M. (1979). Garbage truck of the brain. Science, 2013(340), 1529–1530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Saito, Y. , Kamagata, K. , Andica, C. , Uchida, W. , Takabayashi, K. , Yoshida, S. , Nakaya, M. , Tanaka, Y. , Kamiyo, S. , Sato, K. , Nishizawa, M. , Akashi, T. , Shimoji, K. , Wada, A. , & Aoki, S. (2023). Reproducibility of automated calculation technique for diffusion tensor image analysis along the perivascular space. Japanese Journal of Radiology, 41, 1–8. [DOI] [PubMed] [Google Scholar]
- Siow, T. Y. , Toh, C. H. , Hsu, J.‐L. , Liu, G. H. , Lee, S. H. , Chen, N. H. , Fu, C. J. , Castillo, M. , & Fang, J. T. (2022). Association of sleep, neuropsychological performance, and gray matter volume with glymphatic function in community‐dwelling older adults. Neurology, 98, e829–e838. [DOI] [PubMed] [Google Scholar]
- Tang, J. , Zhang, M. , Liu, N. , Xue, Y. , Ren, X. , Huang, Q. , Shi, L. , & Fu, J. (2022). The association between glymphatic system dysfunction and cognitive impairment in cerebral small vessel disease. Frontiers in Aging Neuroscience, 14, 916633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Taoka, T. , Ito, R. , Nakamichi, R. , Kamagata, K. , Sakai, M. , Kawai, H. , Nakane, T. , Abe, T. , Ichikawa, K. , Kikuta, J. , Aoki, S. , & Naganawa, S. (2022). Reproducibility of diffusion tensor image analysis along the perivascular space (DTI‐ALPS) for evaluating interstitial fluid diffusivity and glymphatic function: Changes in alps index on multiple condition acquisition experiment (CHAMONIX) study. Japanese Journal of Radiology, 40, 147–158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Taoka, T. , Ito, R. , Nakamichi, R. , Nakane, T. , Sakai, M. , Ichikawa, K. , Kawai, H. , & Naganawa, S. (2022). Diffusion‐weighted image analysis along the perivascular space (DWI–ALPS) for evaluating interstitial fluid status: Age dependence in normal subjects. Japanese Journal of Radiology, 40, 894–902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Taoka, T. , Masutani, Y. , Kawai, H. , Nakane, T. , Matsuoka, K. , Yasuno, F. , Kishimoto, T. , & Naganawa, S. (2017). Evaluation of glymphatic system activity with the diffusion MR technique: Diffusion tensor image analysis along the perivascular space (DTI‐ALPS) in Alzheimer's disease cases. Japanese Journal of Radiology, 35, 172–178. [DOI] [PubMed] [Google Scholar]
- Tarasoff‐Conway, J. M. , Carare, R. O. , Osorio, R. S. , Glodzik, L. , Butler, T. , Fieremans, E. , Axel, L. , Rusinek, H. , Nicholson, C. , Zlokovic, B. V. , Frangione, B. , Blennow, K. , Ménard, J. , Zetterberg, H. , Wisniewski, T. , & de Leon, M. J. (2015). Clearance systems in the brain—Implications for Alzheimer disease. Nature Reviews Neurology, 11, 457–470. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tian, Y. , Cai, X. , Zhou, Y. , Jin, A. , Wang, S. , Yang, Y. , Mei, L. , Jing, J. , Li, S. , Meng, X. , Wei, T. , Liu, T. , Wang, Y. , Pan, Y. , & Wang, Y. (2023). Impaired glymphatic system as evidenced by low diffusivity along perivascular spaces is associated with cerebral small vessel disease: A population‐based study. Stroke and Vascular Neurology, 8, e002191–e002423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Valenza, M. , Facchinetti, R. , Steardo, L. , & Scuderi, C. (2020). Altered waste disposal system in aging and Alzheimer's disease: Focus on astrocytic aquaporin‐4. Frontiers in Pharmacology, 10, 1656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wardlaw, J. M. , Smith, E. E. , Biessels, G. J. , Cordonnier, C. , Fazekas, F. , Frayne, R. , Lindley, R. I. , O'Brien, J. T. , Barkhof, F. , Benavente, O. R. , Black, S. E. , Brayne, C. , Breteler, M. , Chabriat, H. , Decarli, C. , de Leeuw, F. E. , Doubal, F. , Duering, M. , Fox, N. C. , … STandards for ReportIng Vascular changes on nEuroimaging (STRIVE v1) . (2013). Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurology, 12, 822–838. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wright, A. M. , Wu, Y.‐C. , Chen, N.‐K. , & Wen, Q. (2023). Exploring radial asymmetry in MR diffusion tensor imaging and its impact on the interpretation of glymphatic mechanisms. Journal of Magnetic Resonance Imaging. 10.1002/jmri.29203 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yu, L. , Hu, X. , Li, H. , & Zhao, Y. (2022). Perivascular spaces, glymphatic system and MR. Frontiers in Neurology, 13, 844938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhai, F.‐F. , Yan, S. , Li, M.‐L. , Han, F. , Wang, Q. , Zhou, L. X. , Ni, J. , Yao, M. , Zhang, S. Y. , Cui, L. Y. , Jin, Z. Y. , & Zhu, Y. C. (2018). Intracranial arterial dolichoectasia and stenosis: Risk factors and relation to cerebral small vessel disease. Stroke, 49, 1135–1140. [DOI] [PubMed] [Google Scholar]
- Zhang, W. , Zhou, Y. , Wang, J. , Gong, X. , Chen, Z. , Zhang, X. , Cai, J. , Chen, S. , Fang, L. , Sun, J. , & Lou, M. (2021). Glymphatic clearance function in patients with cerebral small vessel disease. NeuroImage, 238, 118257. [DOI] [PubMed] [Google Scholar]
- Zhang, Y. , Zhang, R. , Ye, Y. , Wang, S. , Jiaerken, Y. , Hong, H. , Li, K. , Zeng, Q. , Luo, X. , Xu, X. , Yu, X. , Wu, X. , Yu, W. , Zhang, M. , & Huang, P. (2021). The influence of demographics and vascular risk factors on Glymphatic function measured by diffusion along perivascular space. Frontiers in Aging Neuroscience, 13, 693787. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhou, Y. , Cai, J. , Zhang, W. , Gong, X. , Yan, S. , Zhang, K. , Luo, Z. , Sun, J. , Jiang, Q. , & Lou, M. (2020). Impairment of the glymphatic pathway and putative meningeal lymphatic vessels in the aging human. Annals of Neurology, 87, 357–369. [DOI] [PubMed] [Google Scholar]
- Zhu, Y.‐C. , Tzourio, C. , Soumaré, A. , Mazoyer, B. , Dufouil, C. , & Chabriat, H. (2010). Severity of dilated Virchow‐Robin spaces is associated with age, blood pressure, and MRI markers of small vessel disease: A population‐based study. Stroke, 41, 2483–2490. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1: Supporting information.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.