Independent Role of White Matter Hyperintensity Volume and Location in Alzheimer’s Disease Risk Beyond Hippocampal Atrophy
Article information
Abstract
Objective
Increases in white matter hyperintensities (WMH) observed on brain MRI are associated with the onset of Alzheimer’s disease (AD) and cognitive decline. Recent hypotheses suggest that the impact of WMH on cognition may differ by their distance from the ventricular surface. This study aimed to investigate the effects of WMH volume and location, classified by distance from the ventricular surface, on cognitive function in individuals with AD.
Methods
A total of 112 normal cognition (NC) individuals and 171 patients with AD underwent clinical evaluation, volumetric MRI, and neuropsychological testing using the Korean version of the Consortium to Establish a Registry for Alzheimer’s Disease. WMH volume was categorized as juxtaventricular (JVWMH, <3 mm from ventricle), periventricular (PVWMH, 3–13 mm), and deep (DWMH, >13 mm).
Results
The mean WMH volume was significantly higher in AD group (20.7±18.2 mL) than in the NC group (6.8±8.1 mL, p<0.001). A tenfold increase in WMH volume led to a 5.967-fold increased risk of AD (95% confidence interval [CI]=1.550–22.986). A similar risk association was observed for PVWMH (OR=4.021, 95% CI=1.592–10.156), and DWMH showed a significant risk association (OR= 2.873, 95% CI=1.227–6.731). Total WMH, JVWMH, and PVWMH were associated with poorer performance in verbal fluency and memory tasks, while DWMH showed no significant cognitive association.
Conclusion
WMH volume and location independently contribute to AD risk and cognitive decline, with PVWMH and JVWMH particularly affecting executive and memory functions, regardless of hippocampal atrophy.
INTRODUCTION
Globally, the rapid growth of the aging population has intensified focus on age-related neurodegenerative disorders, particular Alzheimer’s disease (AD), which accounts for 60%–80% of dementia cases [1]. AD imposes substantial social and economic burdens while severely compromising quality of life for patients and caregivers, underscoring the urgency of understanding its pathophysiology [1,2]. Among neuroimaging biomarkers, white matter hyperintensities (WMH)-bright lesions on fluid-attenuated inversion recovery (FLAIR) MRI-have emerged as critical markers of cerebral small vessel disease and neurodegeneration. Prevalence estimated for WMH vary widely, from 39% in individuals aged 55%–85% to 95% in those age 60–90, with longitudinal studies demonstrating progressive increases over time [3-5]. These lesions are linked not only to cognitive decline [6,7] but also to psychiatric symptoms [8,9], motor impairments [7,10], and cerebrovascular disorders [11], highlighting their multifactorial impact on brain health.
The pathogenesis of WMH involves diverse mechanisms, traditionally categorized into ischemic (e.g., atherosclerosis, hypoperfusion) and non-ischemic (e.g., blood-brain barrier dysfunction, neuroinflammation) processes [12]. Kim et al.’s [13] classification system further refines this understanding by correlating WMH location with etiology. Juxtaventricular WMH (JVWMH; >3 mm from vernticles) is associated with non-ischemic subependymal gliosis and disrupted cerebrospinal fluid (CSF) dynamics [13,14], while periventricular WMH (PVWMH; 3–13 mm) and deep WMH (DWMH; >13 mm) reflect ischemic damage from small vessel disease. JVWMH, near the corticomedullary junction, involves distinct U-fiber pathways [15]. Despite these advances, the role of WMH in AD remains contentious. While vascular mechanisms dominate WMH research, recent evidence suggests AD-specific pathways, such as amyloid-β-mediated neuroinflammation or tau-driven axonal degeneration, may contribute to WMH formation independently of vascular pathology [16,17].
Critically, WMH topography may influence their functional impact. Posterior regions, particularly the splenium of the corpus callosum, exhibit stronger associations with cognitive deficits in AD, while PVWMH demonstrate greater links to executive dysfunction than DWMH [18,19]. However, studies often conflate AD with mixed dementia or lack biomarker confirmation, obscuring etiology specific relationships. For instance, amyloid-positive AD patients show elevated WMH volumes in strategic white matter tracts compared to vascular risk factor-matched controls, independent of cortical atrophy [19]. These findings underscore the need to disentangle AD-related WMH from vascular contributions.
This study investigates the impact of WMH volume and location-classified by distance from the ventricular surface (JVWMH, PVWMH, and DWMH)-on AD risk and domainspecific cognitive decline. By integrating etiological stratification with spatial analysis, we aim to clarify whether WMH effects in AD arise from vascular insults, AD-specific neurodegeneration, or synergistic interactions. Our findings could refine diagnostic criteria and therapeutic strategies by elucidating spatially distinct WMH pathways in AD progression.
METHODS
Subjects
This study recruited 283 participants aged 60 years or older, including 112 normal cognition (NC) individuals and 171 patients diagnosed with AD, from the dementia clinic at Jeju National University Hospital (JNUH) in Jeju-do, South Korea. The diagnosis of AD was established based on the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria [20], incorporating clinical, and neuropsychological evidence, and we included subjects with probable AD and possible AD. All participants underwent comprehensive clinical evaluation, including physical and neurological examinations, laboratory tests to exclude reversible cause of dementia and MRI scans. Neuropsychological assessments were conducted by four trained neuropsychologists using the Korean version of the Consortium to Establish a Registry for Alzheimer’s Disease Assessment Packet (CERAD-K) [21]. The CERAD-K included tests for verbal fluency, the 15-item Boston Naming Test, the Korean the Korean version of the Mini-Mental State Examination (MMSE-KC), Word List Memory, Constructional Praxis, Word List Recall, Word List Recognition, Constructional Recall, and the Trail Making Test. Exclusion criteria were applied to ensure diagnostic accuracy. Participants with major psychiatric disorders such as schizophrenia or bipolar disorder were excluded, as were those with a history of stoke, focal neurological signs, or other types of dementia. Final diagnoses were reviewed by two dementia-specialized neuropsychiatrists blinded to WMH data. Written informed consent was obtained from all participants or their legal representatives, and the study protocol was approved by the institutional review board of JNUH (JEJUNUH 2020-06-008).
MRI acquisition
MRI scans were performed using a 3.0 T Philips Intera scanner (Philips). The imaging protocol included acquiring three-dimensional (3D) T1-weighted anatomical images with an acquisition voxel size of 1.0×0.5×0.5 mm, a sagittal slice thickness of 1.0 mm with no inter-slice gaps, a repetition time of 4.61 ms, an echo time of 8.15 ms, one excitation, a flip angle of 8°, a field of view of 240×240 mm, and an acquisition matrix size of 175×256×256 mm in the x-, y-, and z-dimensions. Additionally, 3D FLAIR images were obtained with a voxel dimension of 1×1×3 mm3, a repetition time of 9,900 ms, an echo time of 125 ms, an inversion time of 2,800 ms, one excitation, a flip angle of 90°, a field of view of 240 mm, an axial plane matrix of 256×256 mm, a thickness of 3 mm, and no interslice gap. To ensure high-quality imaging results free from motion artifacts that could compromise diagnostic value or analysis reliability, participants received detailed instructions on proper positioning during scans from an experienced imaging technician. After scanning completion, quality control assessments were performed by the radiology specialist.
MRI processing and analysis
Structural volumetric were analyzed using FreeSurfer version 7.0.0 software (Athinoula A. Martinos Center for Biomedical Imaging) [22] to segment cortical and subcortical regions basedon T1-weighted images. Outputs included total brain volume (TBV), estimated intracranial volume (eICV), and hippocampal volume (HV). WMH segmentation was performed by co-registering FLAIR images with T1-weighted images using Statistical Parametric Mapping version 12 (SPM12; Wellcome Institute of Neurology, University College London) for MATLAB (MathWorks Inc.). Lesion mapping utilized the Lesion Sementation Toolbox for SPM12 to identify WMH regions automatically [23]. WMH spatial classification followed Kim et al.’s distance-based approach using an in-house MATLAB script. Initially, the ventricle was segmented from the T1-weighted image of each subject. A distance map was then generated for each WMH voxel relative to the ventricle. Based on their proximity to the ventricle, each WMH voxel was classified into one of three categories: JVWMH were defined as areas within 4 voxels of the ventricle, PVWMH as those between 4 and 13 voxels, and DWMH as those more than 13 voxels away. The Euclidean distance from each WMH voxel to the nearest ventricle voxel was calculated, resulting in a distance map that encodes these distances.
Additional measures
The apolipoprotein E genotype (APOE) for each participant was identified using a single-stage polymerase chain reaction from venous blood samples, following the procedure outlined by Wenham et al. [24]. Depressive symptoms were assessed using the Korean version of the Short Form of the Geriatric Depression Scale (SGDS-K), wihich ranges from scores of 0–15; higher scores indicate greater symptoms severity [25]. Comorbidity burden was quantified using Charlson comorbidity index (CCI, increasingly worse). CCI was used to quantify comorbid condition. Each condition is assigned a weight from 1 to 6, which are summed to calculated a total CCI score for each participant. Its widespread use and validation across multiple studies and settings to its utility and realibility [26].
Statistical analysis
To account for variations in head size among participants, WMH volumes were normalized by dividing total WMH volume by eICV. The ratio variables were log-transformed due to its non-normal distribution. Group comparisons between NC and AD participants utilized analysis of covariance for continuous variables adjusted for age/sex and chi-squared tests for categorical variables here appropriate. Multiple linear regression models assessed associations between logtransformed WMH volumes and covariates including age, sex, years of education, APOE ε4 allele status, HV, SGDS-K, CCI and Clinical Dementia Rating Scale Sum of Boxes. To evaluate cognitive associations with WMH subtypes classified by proximity to ventricles (JVWMH/PVWMH/DWMH), CERAD-K subset scores served as dependent variables in regression models adjusted for demographic/clinical covariates. All analyses were conducted using STATA version 15.1 software (Stata Corp.) with statistical significance defined at p<0.05.
RESULTS
Sociodemographic and clinical characteristics of participants
In the study, 112 participants were categorized as having NC and 171 as having AD. The mean age of the AD group significantly higher at 80.7 years compared to 73.4 years in the NC group (p<0.001). Sex distribution revealed a higher proportion of females in the AD (70.8%) compared to the NC group (51.8%, p=0.001). Educational attainment was lower in the AD group, with an average of 4.6 years of education versus 10.2 years in the NC group (p<0.001) (Table 1). WMH volumes were log-transformed for analysis, revealing higher values in the AD group across all measures. The total WMH volume in the AD group had an average log value of 4.1±0.4, compared to 3.6±0.4 in the NC group (p<0.001). Similarly, JVWMH had a mean log value of 3.8±0.4 in the AD group versus 3.4±0.4 in the NC group (p<0.001). PVWMH showed an average log value of 3.7±0.6 for the AD group and 2.9±0.7 for the NC group (p<0.001). DWMH also demonstrated higher values in the AD group (2.7±0.7) compared to the NC group (2.1±0.7, p<0.001, t-test) (Table 1 and Supplementary Figure 1).
Comparison of demographic and clinical characteristics between NC and AD group
Factors associated with WMH
To identify factors associated with WMH, a linear regression analysis was conducted. In the NC group, WMH volume was positively associated with age (p=0.002) and eICV (p=0.044), suggesting that older age and larger eICV contribute to increased WMH volume. In contrast, WMH volume in the AD group was associated with age (p<0.001), eICV (p<0.001), and TBV (p=0.049), indicating that older age, larger eICV and smaller TBV were significant predictors of increased WMH volume (Table 2).
Multivariate linear regression analysis of various factors associated with WMH volume in the groups
Factors associated with AD
To identify factors associated with AD, logistic regression analysis was conducted using age, sex, years of education, presence of the ApoE ε4 status, family history of dementia, HV, WMH, CCI, and the SGDS-K as covariates (Table 3).
Multivariate logistic regression analysis of various factors including WMH and sub-classified WMH volume associated with Alzheimer’s disease
Higher educational attainment was associated with reduced risk of AD (odds ratio [OR]=0.796; 95% confidence interval [CI]=0.699–1.135), while decreased HV significantly increased the risk (OR=0.998; 95% CI=0.997–0.999). WMH volume was strongly associated with AD risk, with a tenfold increase in WMH volume corresponding to a 5.967-fold increase in AD likelihood (95% CI=1.550–22.986), as the data were logtransformed. Sub-classified WMHs showed distinct associations: JVWMH was not significantly linked to AD risk (p=0.058), but PVWMH and DWMH were significant predictors. A tenfold increase in PVWMH volume resulted in a fourfold increase in AD risk (OR=4.021, p=0.003), while a tenfold increase in DWMH volume led to a nearly threefold increase in risk (OR=2.783, p=0.015).
Impact of white matter hyperintensity on cognitive function
The impact of WMHs on cognitive function was assessed using linear regression models adjusted for demographic and clinical variables across all participants and within NC and AD groups (Tables 4 and 5) (Additionally, the relationship between WMH volume and neuropsychological tests in the all participants is presented in Supplementary Table 1). Total WMH volumes were associated with poorer performance on the Word List Memory Test (p=0.013) and Trail Making Test B (p=0.017) across all participants, while in the NC group, total WMH correlated only with Trail Making Test B performance (p=0.022). In the AD group, total WMH volumes were linked to impaired Verbal Fluency Test scores (p=0.008) and Word List Memory Test scores (p=0.023). For JVWMH, poorer performance on the Boston Naming Test (p=0.021), Word List Memory Test (p=0.023), and Trail Making Test B (p=0.005) was observed across all participants, while in the NC group, JVWMH correlated only with Trail Making Test B performance (p=0.015). In the AD group, JVWMH was associated with Verbal Fluency Test scores (p=0.013) and Forward Digit Span Test scores (p=0.037). PVWMH volumes were associated with Word List Memory Test scores across all participants (p=0.020) but showed no associations within the NC group; however, in the AD group, PVWMH correlated with Verbal Fluency Test scores (p=0.011) and Word List Memory Test scores (p=0.021). DWMH volumes showed no significant associations across all participants except for Trail Making Test B performance in the NC group alone (p=0.039), while no significant associations were found within the AD group.
The association between WMH volume and neuropsychological tests in normal cognition group
The association between WMH volume and neuropsychological tests in Alzheimer`s disease group
DISCUSSION
In this study, the average age of participants in the AD group was 80.7 years, significantly higher than the average age of 73.4 years in the NC group (p<0.001). Additionally, the proportion of females was higher in the AD group (70.8%) compared to the NC group (51.8%, p=0.001). The AD group also had a lower average educational attainment of 4.6 years compared to 10.2 years in the NC group (p<0.001). These findings are consistent with previous studies that have identified advanced age, female gender, and lower educational attainment as risk factors for AD [27,28]. Furthermore, WMH volume was significantly larger in the AD group (20.7±18.2 mL) compared to the NC group (6.8±8.1 mL, p<0.001), supporting prior research linking increased WMH volume to AD risk [29,30]. Using automated quantification methods for brain MRI analysis, we confirmed that WMH volume increases with age and is larger in individuals with AD, consistent with previous findings that associated WMH with aging [31] and higher AD risk [3,27,32].
Logistic regression analysis revealed that larger volumes of WMH, PVWMH, and DWMH were associated with an increased risk of AD. Specifically, WMH in individuals with AD was related to dysfunctions in frontal lobe functions such as verbal fluency, immediate memory, and working memory. These findings align with recent studies demonstrating associations between WMH and executive function [33-36], psychomotor speed [33,34,37,38], memory [35,39], language ability [40], attention [41], and global cognitive function [29,42,43]. The relationship between WMH and frontal lobe dysfunction may be explained by its impact on glucose metabolism; ischemic vascular damage predominantly located in the frontal lobe could lead to reduced cortical metabolism and diminished function [44].
This study divided WMH into JVWMH, PVWMH, and DWMH based on etiological and functional considerations to investigate their differ rential impacts on AD risk and cognitive function. Our findings suggest that greater PVWMH and DWMH volumes may be important risk factors for AD. In terms of cognitive function, JVWMH and PVWMH were associated with declines in frontal lobe functions such as executive function, immediate memory, and working memory in individuals with AD, while DWMH did not significantly affect cognitive function. Pathophysiologically, JVWMH is a non-ischemic lesion located within 3 mm of the ventricular surface and is related to CSF outflow, whereas PVWMH is an ischemic lesion located between 3 mmg and 13 mm from the ventricular surface caused by sustained hypoperfusion [13]. These findings suggest that JVWMH and PVWMH have similar effects on cognitive function due to shared microstructural characteristics such as CSF-like free water content; however, DWMH differs structurally as it comprises WM-like material [45].
Although DWMH did not significantly affect cognitive decline in individuals with AD, it was identified as a risk factor for the disease onset. In our study, DWMH volume was found to have a relatively small volume and limited cognitive impact, as confirmed through CERAD-K in AD. However, our multivariate logistic regression analysis, which adjusted for variables such as age, sex, education level, and hippocampal atrophy, confirmed that increased DWMH volume is independently associated with AD risk.
This study has several limitations. Firstly, it was conducted at a single institution within one region, limiting the generalizability; multi-institutional studies across diverse regions are needed to validate these findings. Secondly, as a cross-sectional study, this research examined association between WMH volume and cognitive function at a single point in time; longitudinal studies could provide insights into changes in WMH volume over time and their impact on AD progression and cognitive decline. Thirdly, while this study contributes by classifying WMH based on distance from the ventricular surface into three regions and examining their association with cognitive function, it did not analyze cerebral cortex regions by individual lobes. Lastly, although we adjusted for comorbidity burden using the CCI, this composite measure may not fully capture vascular risk. Sensitivity analyses including hypertension and diabetes as separate covariates yielded results consistent with our primary findings, but reliance on the CCI remains a limitation. Future research could benefit from categorizing WMH according to brain lobes—such as the frontal, temporal, parietal, and occipital lobes—and further subclassifying them into JVWMH, PVWMH, and DWMH within each lobe to elucidate nuanced differences based on both lobe location and proximity to ventricles.
Supplementary Materials
The Supplement is available with this article at https://doi.org/10.30773/pi.2025.0127.
The association between WMH volume and neuropsychological tests in all participants
Boxplots comparing WMH and sub-classfied WMH volume between NC group and AD group. WMH, white matter hyperintensities; log WMH, logarithms of WMH; Log JVWMH, logarithms of juxtaventricular WMH; Log PVWMH, logarithms of periventricular WMH; Log DWMH, logarithms of deep WMH; NC, normal control; AD, Alzheimer’s disease.
Notes
Availability of Data and Material
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Hyun Ju Yang, Jae Min Song. Data acquisition: Hyun Ju Yang, Jae Min Song. Formal analysis: Joon Hyuk Park. Funding acquisition: Hyun Ju Yang, Joon Hyuk Park. Writing—original draft: Hyun Ju Yang, Jae Min Song. Writing—review & editing: Hyun Ju Yang, Joon Hyuk Park.
Funding Statement
This work was supported by a research grant from Jeju National University Hospital in 2020.
Acknowledgments
None