Global and regional white matter hyperintensities in Alzheimer's disease: Exploring etiologies and cognitive correlates

Abstract

Background

White matter hyperintensities (WMH) are commonly observed in Alzheimer's disease (AD), but their underlying pathophysiology remains poorly understood.

Objective

This cross-sectional study aims to explore the multifactorial etiologies and cognitive correlates of global and regional WMH in a cohort of biologically diagnosed AD patients.

Methods

We included 170 AD patients who underwent brain MRI, neuropsychological testing, and cerebrospinal fluid (CSF) biomarker evaluation within three months. Linear and logostic regressions were used to assess associations between global/regional WMH and age, cerebral microbleeds (CMB), AD biomarkers, cortical atrophy, and vascular risk scores. Sensitivity analyses included replacing vascular score with hypertension, p-tau181 with t-tau, and inclusion of APOE ε4 status. Associations between Mini-Mental State Exam scores and WMH and atrophy burden were also evaluated, adjusting for age, sex, and education.

Results

Greater global WMH burden was significantly associated with older age (β = 0.03, p < 0.001), lower CSF Aβ₄₂ levels (β = −0.0014, p = 0.01), and more severe CMB grade (β = 0.48, p = 0.003), but not with vascular risk scores. Regionally, WMH in the paraventricular, frontal, and parietal lobes were linked to CSF Aβ₄₂ and CMB. Cognitive impairment was independently associated with increased paraventricular WMH burden (β = −0.22, p < 0.001) and medial temporal atrophy (β = −0.82, p < 0.001).

Conclusions

Our findings suggest that WMH in AD—particularly in the paraventricular, frontal, and parietal areas—are driven more by AD-specific pathology, including amyloid deposition and cerebral amyloid angiopathy, than by conventional vascular risk. Paraventricular WMH contribute to cognitive decline independent of cortical atrophy, underscoring their relevance as potential imaging biomarkers in AD.

Keywords

Alzheimer's disease amyloidosis biomarkers magnetic resonance imaging white matter

Introduction

White matter hyperintensities (WMH), initially described as leukoaraiosis in the 1980s, are radiologically identified as hyperintense signals on T2-weighted and fluid-attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI), indicative of visible white matter damage. These lesions can range from focal to confluent and are pathologically characterized by demyelination, axonal loss, and reactive gliosis.^1,2

The severity of WMH increases with age and is closely linked to cardiovascular risk factors, particularly hypertension.³ Traditionally, WMH have been attributed to chronic ischemia resulting from cerebral small vessel disease (CSVD).^4,5 WMH affect up to 90% of individuals aged ≥60 years. WMH are even more frequent and severe in Alzheimer's disease (AD) patients compared to non-demented older adults.^6,7 Recent evidence has suggested a more complex relationship between WMH and AD, with studies highlighting diverse mechanisms including cerebral amyloid angiopathy (CAA), AD pathology, and neurodegenerative changes.^8–12 This challenges the traditional vascular-centric view of WMH.

Despite these advancements, most studies on AD-related WMH have taken a fragmented approach, focusing on isolated factors or limited combinations of factors, without simultaneously considering their combined effects. These studies have produced inconsistent findings: some suggest that WMH are associated with amyloid pathology, while others link them to tau pathology.¹³ This gap leaves important questions unanswered, such as whether these factors independently contribute to WMH development, and whether amyloid-β (Aβ) pathology exerts an influence on WMH beyond the effects of CAA.

Although pathological evidence indicates that frontal WMH in AD are associated with small vessel disease, emerging studies have shown no consistent correlation between WMH and vascular risk factors in AD.¹⁴ A recent study published in JAMA reported no correlation between WMH and vascular risk factors in AD within a large longitudinal cohort, although it did not classify WMH by brain regions.¹⁵ This underscores the need for further investigation into whether regional WMH, particularly in the frontal lobes, are related to vascular risk factors in AD.

Here, we address these gaps by adopting a multivariate regression approach to simultaneously evaluate the contributions of age, CAA (via cerebral microbleeds, CMB), AD biomarkers (CSF Aβ₄₂, p-tau181 or t-tau), regional cortical atrophy, and systemic vascular risk factors to both global and regional WMH burden in a cohort of biologically confirmed AD patients. By integrating these multifactorial determinants, our research aims to investigate the potential underlying pathologies of WMH in the AD context. Additionally, we sought to determine whether regional WMH affect cognitive performance independently of cortical atrophy. These findings are expected to provide new insights into the heterogeneity of WMH etiology and its implications for cognitive decline in AD.

Methods

Participants

We enrolled participants from the Peking Union Medical College Hospital (PUMCH) dementia cohort between September 2022 and January 2024. A total of 170 individuals met the 2011 NIA-AA criteria for probable AD and confirmed amyloid positivity via cerebrospinal fluid (CSF) biomarker. All subjects completed brain MRI scanning, neuropsychological testing, and CSF collection within a three-month interval. Most participants also underwent APOE genotyping. Exclusion criteria included: macrohemorrhage (>10 mm in diameter), multiple lacunar infarcts or major-territory stroke, encephalomalacia, space-occupying lesions, or neoplasm on MRI.

Clinical data encompassing medical history, blood biochemical tests, APOE genotyping, MRI, neuropsychological screening results, and CSF biomarkers were systematically collected. To address missing data, we applied listwise deletion for each analysis: the primary WMH models included 140 participants (6 excluded for incomplete SWI/T2*SWAN sequences; 24 for missing vascular risk data), the cognitive analysis comprised 167 participants (3 missing Mini-Mental State Exam (MMSE) scores), and the APOE ε4–sensitivity analysis involved 129 participants (11 missing APOE genotypes).

CSF biomarker measurement

Concentration levels of CSF Aβ₄₂, Aβ₄₀, phosphorylated tau 181 (p-tau181), and total tau (t-tau) were quantified using enzyme-linked immunosorbent assay kits (ELISA; Fujirebio, Ghent, Belgium) per manufacturer protocols. Amyloid (A+) and tau (T+) status was defined by CSF t-tau/Aβ42 > 0.5 and p-tau181 > 50 pg/mL, respectively, using established thresholds.^16–18

Assessment of vascular risk score

Electronic medical history databases were reviewed for all patients, in which pertinent information about vascular health was systematically recorded. The presence or absence of five vascular risk factors, including hypertension, hyperlipemia, diabetes, smoking and hyperhomocysteinemia, were summated to a composite score (max = 5) providing a comprehensive representation of an individual's vascular risk condition.

MRI acquisition

MRIs were acquired with a 3 T MRI scanner (Discovery MR750, GE, Milwaukee, Wisconsin). Sequences included T1-weighted, T2-weighted, FLAIR, diffusionweighted imaging (DWI), Apparent diffusion coefficient (ADC), and susceptibility weighted imaging (SWI)/T2 star weighted angiography (SWAN) sequences.

Assessment of WMH

WMH were evaluated using T2 FLAIR images, supplemented with T1 and T2 weighted MR images for additional detail. Lacunar infarcts were excluded from the WMH assessment.

The Modified Scheltens Rating Scale was employed to assess WMH.¹⁹ This scale differentiates paraventricular white matter hyperintensities (PWMH) from deep white matter hyperintensities (DWMH) and further categorizes DWMH by cerebral lobes. It also considers hyperintensities in the basal ganglia and infratentorial regions. Compared to Fazekas’ scoring method, the Modified Scheltens Rating Scale offers more detailed information and show good intraobserver and interobserver agreement for measuring WMH at a single time point.

The scale provides 4 sum scores in a semiquantitative way: PWMH (0–6), lobar white matter hyperintensities (0–24), basal ganglia hyperintensities (0–30) and infratentorial foci of hyperintensity (0–30). PWMH were identified as continuous, confluent areas of high signal intensity adjacent to anterior or posterior horns of the lateral ventricles (“caps”) and along the lateral ventricles (“bands”). Lobar white matter hyperintensities, located in the deep and subcortical white matter, were separately rated in the frontal, temporal, parietal and occipital regions; WMH directly adjacent to the ventricles were rated separately as PWMH. Periventricular hyperintensities exceeding 10 mm in extension were, by definition, scored as lobar WMH.

Assessment of atrophy

The scores of global and regional atrophy were assessed on axial and sagittal sections of the T1-weighted sequence.

Global atrophy severity was assessed using the global cortical atrophy (GCA) scale (grades 0–3), defined as follows: Grade 0, with no cortical atrophy, no lateral ventricle enlargement; Grade 1, mild cortical atrophy, with widening of the sulci, or with mild lateral ventricle enlargement; Grade 2, moderate cortical atrophy, with reduced gyrus volume, or with moderate lateral ventricle enlargement; Grade 3, severe cortical atrophy with reduced gyrus volume, “razor blade atrophy,” or severe lateral ventricle enlargement.²⁰

Frontal lobe atrophy was evaluated using the same criteria as the GCA scale due to the absence of a dedicated frontal atrophy scale. Grade 0 indicated no atrophy; Grade 1, mild atrophy with sulcal widening; Grade 2, moderate atrophy with sulcal widening and reduced gyrus volume; and Grade 3, severe end-stage atrophy.

Parietal lobe atrophy was assessed using the Koedam score (grades 0–3), where Grade 0 represents no atrophy; Grade 1, mild sulcal widening without evident gyrus volume loss; Grade 2, substantial sulcal widening and gyrus volume loss; and Grade 3, severe end-stage atrophy.²¹

Medial temporal atrophy (MTA) was visually rated on a 5-point scale (grades 0–4): Grade 0, no MTA with normal perimesencephalic cistern and temporal horn morphology and normal medial temporal cortex; Grade 1, questionable atrophy with slight widening of the PC or slit-like TH; Grade 2, mild but definite atrophy with mild widening of the PC and TH; Grade 3, moderate atrophy with moderate widening of the PC and TH, along with hippocampal deformation; and Grade 4, severe atrophy with pronounced widening of the PC and TH and marked angulation of the medial temporal cortex (“knife-edge” change).^22,23

Assessment of CMB

Lobar CMB burden was assessed on SWI or T2*SWAN images. Small lesions (≤10 mm) that were dissociable from small vessels were counted as definite microbleeds. Hemorrhagic lesion in cerebellum or deep hemorrhagic lesion not counted.

We categorized CMB burden into three levels based on the Boston criteria—no CMB, 1 recognized CMB (consistent with possible CAA), and 2 or more recognized CMBs (consistent with probable CAA).²⁴

APOE genotyping

Genomic DNA for APOE genotyping was extracted from peripheral blood mononuclear cells using the Genomic DNA Purification System (Promega, Madison, WI, USA). APOE alleles were determined by targeting the rs429358 and rs7412 polymorphisms that define the ε2, ε3, and ε4 alleles. Participants were classified as APOE ε4 carriers if they had at least one ε4 allele (i.e., ε3/ε4 or ε4/ε4), and as non-carriers otherwise.

Statistical analysis

All statistical testing was conducted in R, version 4.4.1.

Normality of continuous variables was assessed by the Shapiro–Wilk test. Mean and standard deviation (SD) were used if the variable was normally distributed. Median and interquartile range were used for non-normally distributed variables as well as ranked variables. To compare the difference in continuous variables between AD patients with WMH and those without, the independent t-test was used for normally distributed variables and the Mann-Whitney U test was for non-normally distributed variables. The Wilcoxon test was performed for ordered rank variables.

For the primary WMH analysis, 140 participants were included after excluding 30 individuals with missing data (6 lacking SWI/T2*SWAN sequences and 26 missing vascular risk scores). Missing values were handled by listwise deletion. Multivariable linear regression models assessed the associations of age, sex, vascular risk score, CMB grade, atrophy grade, CSF Aβ₄₂, and p-tau181 with log-transformed global and regional WMH volumes. Due to low prevalence, basal ganglia and infratentorial WMH were analyzed using logistic regression. Multicollinearity was evaluated via variance inflation factors (all VIF <2), and p-values were corrected for multiple testing using the false discovery rate (FDR) method.

Sensitivity analyses replaced vascular score with hypertension (n = 140), p-tau181 with t-tau (n = 140), and included APOE ε4 status (n = 129 with complete APOE genotyping).

Additionally, in a separate cohort of 167 subjects (3 excluded for missing MMSE scores), linear regression examined relationships between log-MMSE scores and WMH/atrophy measures, adjusting for age, sex, and education; subsequent models also included vascular risk score or hypertension.

For all the above linear regression analyses, global and regional WMH scores, as well as MMSE scores, were log-transformed to reduce skewness when used as dependent variables.

Results

Participant characteristics

Data from 170 participants were included, with the demographic and clinical characteristics provided in Table 1. Out of 170 participants, 6 participants did not complete SWI/T2*SWAN, 3 did not have a MMSE score, 1 did not have a Clinical Dementia Rating (CDR) score, and 26 did not have a vascular risk factor score due to missing data.

Table 1.

Demographic, clinical, and biomarker characteristics of AD patients with and without WMH.

	Total (n = 170)	WMH-(n = 32)	WMH + (n = 138)	p
Age				0 . 04 ^a
mean (SD)	61.76 (8.40)	58.97 (8.29)	62.41 (8.33)
Sex (%)				0.29^b
F	100 (58.8)	22 (68.8)	78 (56.5)
M	70 (41.2)	10 (31.2)	60 (43.5)
Years of education				0.63^c
median [IQR]	11.00 [8.25, 13.75]	10.50 [8.00, 13.00]	11.00 [9.00, 14.00]
CDR (%)				0.33^d
0.5	35 (20.7)	7 (22.6)	28 (20.3)
1	44 (26.0)	10 (32.3)	34 (24.6)
2	39 (23.1)	7 (22.6)	32 (23.2)
3	51 (30.2)	7 (22.6)	44 (31.9)
MMSE				0.15^c
median [IQR]	13.00 [5.75, 19.00]	15.00 [9.50, 21.50]	13.00 [5.00, 18.00]
≥27	6 (3.6)	2 (6.5)	4 (2.9)
21–26	29 (17.3)	7 (22.6)	22 (16.1)
≤20	133 (79.2)	22 (71.0)	111 (81.0)
CMB grade (%)				0.19^d
0	130 (79.3)	27 (87.1)	103 (77.4)
1	15 (9.1)	3 (9.7)	12 (9.0)
2	19 (11.6)	1 (3.2)	18 (13.5)
Global atrophy (%)				0.05^d
0	9 (5.3)	4 (12.5)	5 (3.6)
1	92 (54.1)	19 (59.4)	73 (52.9)
2	63 (37.1)	8 (25.0)	55 (39.9)
3	6 (3.5)	1 (3.1)	5 (3.6)
Medial temporal atrophy (%)				0 . 003 ^d
0	12 (7.1)	7 (21.9)	5 (3.6)
1	35 (20.6)	10 (31.2)	25 (18.1)
2	62 (36.5)	6 (18.8)	56 (40.6)
3	43 (25.3)	8 (25.0)	35 (25.4)
4	18 (10.6)	1 (3.1)	17 (12.3)
Parietal atrophy (%)				0.14^d
0	19 (11.2)	4 (12.5)	15 (10.9)
1	75 (44.1)	17 (53.1)	58 (42.0)
2	62 (36.5)	11 (34.4)	51 (37.0)
3	14 (8.2)	0 (0.0)	14 (10.1)
Frontal atrophy (%)				0 . 005 ^d
0	38 (22.4)	14 (43.8)	24 (17.4)
1	95 (55.9)	14 (43.8)	81 (58.7)
2	35 (20.6)	3 (9.4)	32 (23.2)
3	2 (1.2)	1 (3.1)	1 (0.7)
Aβ₄₂, pg/ml				0 . 02 ^a
mean (SD)	473.92 (142.10)	532.97 (156.56)	460.23 (135.49)
p-tau181, pg/ml				0 . 001 ^c
median [IQR]	65.12 [51.92, 82.58]	83.48 [67.34, 90.57]	59.96 [48.90, 78.35]
t-tau, pg/ml				0 . 03 ^c
median [IQR]	551.04 [401.74, 821.84]	661.63 [497.22, 956.61]	533.21 [376.39, 771.17]
vascular risk score				0.19^d
median [IQR]	1.00 [0.00, 2.00]	1.00 [0.00, 1.00]	1.00 [0.00, 2.00]
APOE ε4 (%)				0.95^b
carrier	61 (40.1)	11 (37.9)	50 (40.7)
non-carrier	91 (59.9)	18 (62.1)	73 (59.3)

Independent t-test.

Chi-square test.

Mann-Whitney U test.

Wilcoxon test; Bold: P < 0.05.

Data are presented as median (interquartile range) unless otherwise indicated.

WMH: white matter hyperintensities; CDR: Clinical Dementia Rating; MMSE: Mini-Mental State Examination; CMB: cerebral microbleeds; Aβ₄₂: amyloid beta 42; p-tau181: phosphorylated tau 181; t-tau: total tau.

Overall, the average age of participants was 61.8 years, 58.8% were female, A + T + comprised 77.1% of this sample and A + T- comprised 22.9%. There were no significant differences for sex, years of education, CDR, MMSE, CMB grade, global atrophy grade, parietal atrophy grade, and vascular risk factor score (p > 0.05). The WMH + group showed older age, progressive MTA, progressive frontal lobe atrophy, lower CSF Aβ₄₂ levels, but also lower p-tau181 and T-tau levels (Table 1).

Factors associated with global WMH burden

To examine the factors associated with global WMH burden, a total of 140 participants with complete SWI/T2*SWAN and vascular risk data were included in the analysis.

In the primary multivariable linear regression, older age (β = 0.03, SE = 0.01, p < 0.001), lower Aβ₄₂ level of CSF (β = −0.0014, SE = 0.0005, p = 0.01) and more severe CMB grade (β = 0.48, SE = 0.16, p = 0.003) were each independently associated with greater log-transformed global WMH burden (Figure 1 and Table 2).

Figure 1.

Associations between global WMH and CSF Aβ₄₂ level. The black line represents the linear regression model, with the shaded area indicating the 95% confidence interval, illustrating the relationship between CSF Aβ₄₂ levels and global WMH burden (log-transformed). Each point represents an individual participant, with the size of the points corresponding to CMB grade (0, 1, or 2), the shape of the points indicating CSF-based AD biological diagnosis (circles for A + T+, triangles for A + T-), and the color indicating age (blue for younger, brown for older participants). WMH: white matter hyperintensities; CSF: cerebrospinal fluid; Aβ₄₂: amyloid-beta 42; CMB: cerebral microbleeds.

Table 2.

Sensitivity analyses for global WMH burden.

Log global WMH score ∼	Primary Model (r2 = .23)	Sensitivity analysis 1 (r2 = .23)	Sensitivity analysis 2 (r2 = .23)	Sensitivity analysis 3 (r2 = .23)
Age	β = 0.03, SE = 0.01, p = 0.00051*	β = 0.03, SE = 0.01, p = 0.00065*	β = 0.03, SE = 0.01, p = 0.00063*	β = 0.03, SE = 0.0094, p = 0.002**
Sex (Male)	β = 0.14, SE = 0.15, p = 0.36	β = 0.1, SE = 0.14, p = 0.48	β = 0.1, SE = 0.15, p = 0.49	β = 0.1139, SE = 0.1574, p = 0.47
CMB grade	β = 0.48, SE = 0.16, p = 0.00297*	β = 0.46, SE = 0.16, p = 0.00425*	β = 0.48, SE = 0.16, p = 0.00285*	β = 0.41, SE = 0.18, p = 0.02*
Aβ₄₂	β = −0.0014, SE = 0.0005, p = 0.01*	β = −0.0014, SE = 0.0005, p = 0.01*	β = −0.0013, SE = 0.0005, p = 0.01*	β = −0.0017, SE = 0.0005, p = 0.002**
p-tau181	β = −0.00445, SE = 0.00252, p = 0.08	β = −0.00384, SE = 0.00241, p = 0.11	NA	β = −0.0038, SE = 0.0026, p = 0.15
t-tau	NA	NA	β = −0.00035, SE = 0.00019, p = 0.07	NA
Global atrophy	β = 0.43, SE = 0.31, p = 0.16	β = 0.42, SE = 0.31, p = 0.18	β = 0.38, SE = 0.31, p = 0.22	β = 0.4057, SE = 0.3228, p = 0.21
Vascular risk score	β = −0.06, SE = 0.07, p = 0.38	NA	β = −0.04, SE = 0.07, p = 0.53	β = −0.0112, SE = 0.0762, p = 0.88
Hypertension1	NA	β = −0.07, SE = 0.17, p = 0.67	NA	NA
APOE ε4 carrier	NA	NA	NA	β = −0.15, SE = 0.15, p = 0.33

Sensitivity analysis 1: Replaced vascular risk score with the presence/absence of hypertension.

Sensitivity analysis 2: Replaced p-tau181 with t-tau.

Sensitivity analysis 3: Adding APOE status.

Sensitivity analysis 1, in which we substituted hypertension for the composite vascular score, produced nearly identical estimates for age (β = 0.03, SE = 0.01, p < 0.001), Aβ₄₂ (β = –0.0014, SE = 0.0005, p = 0.01), and CMB grade (β = 0.46, SE = 0.16, p = 0.004) (Table 2).

Sensitivity analysis 2, replacing p-tau181 with t-tau, likewise did not materially alter these associations (age: β = 0.03, SE = 0.01, p = 0.00063; Aβ₄₂: β = –0.00126, SE = 0.0005, p = 0.01; CMB: β = 0.48, SE = 0.16, p = 0.003) (Table 2).

Finally, sensitivity analysis 3, which added APOE ε4 carrier status to the model, again confirmed the robustness of our findings: age (β = 0.03, SE = 0.0094, p = 0.002), Aβ₄₂ (β = –0.0017, SE = 0.0005, p = 0.002), and CMB grade (β = 0.41, SE = 0.18, p = 0.02) remained significant, while APOE ε4 carrier status itself was not (β = –0.15, SE = 0.15, p = 0.33) (Table 2).

Across all models, global atrophy, vascular risk (or hypertension), and sex were non-significant, multicollinearity was low (VIF < 2).

Factors associated with regional WMH burden

In regional analyses, PWMH, frontal and parietal WMH burden each was significantly associated with age (paraventricular: β = 0.02, SE = 0.01, p = 0.0003, FDR.p = 0.01; frontal: β = 0.01, SE = 0.01, p = 0.03 FDR.p = 0.27; parietal: β = 0.02, SE = 0.01, p = 0.00008, FDR.p = 0.01), CMB grade (paraventricular: β = 0.35, SE = 0.11, p = 0.002, FDR.p = 0.05; frontal: β = 0.3, SE = 0.11, p = 0.01, FDR.p = 0.09; parietal: β = 0.33, SE = 0.11, p = 0.004, FDR.p = 0.06), and CSF Aβ₄₂ level (paraventricular: β = −0.0008, SE = 0.00035, p = 0.03, FDR.p = 0.27; frontal: β = −0.0007, SE = 0.00036, p = 0.04, FDR.p = 0.3; parietal: β = −0.0009, SE = 0.00035, p = 0.01, FDR.p = 0.1) (Supplemental Table 1). After FDR correction, only PWMH associations with age and CMB grade, as well as the parietal associations with age remained robust (Supplemental Table 1).

Other regions revealed distinct predictors: occipital WMH burden was higher in men (male, β = 0.21, SE = 0.07, p = 0.003, FDR.p = 0.06), while basal ganglia WMH correlated with p-tau181 before correction (β = −0.04, SE = 0.02, p = 0.01, FDR.p = 0.1). Vascular risk score was not linked to WMH in any region (p > 0.05) (Supplemental Table 1).

In sensitivity analyses—replacing vascular risk score with hypertension, p-tau181 with t-tau, or adding APOE ε4 status—core findings persisted. CMB grade remained significant for frontal and parietal WMH; CSF Aβ₄₂ remained significant for PWMH and parietal WMH (though frontal Aβ₄₂ lost significance when p-tau181 was replaced by t-tau, and PWMH CMB effects lost significance after adding APOE ε4). Associations of male sex with occipital WMH and of tau markers with basal ganglia WMH also held. APOE ε4 showed no independent effect, and neither vascular risk nor hypertension predicted WMH burden in any region (Supplemental Tables 2–8).

Contribution of WMH, atrophy, and vascular risk factors to cognitive performance

Lastly, building on the observation that increased WMH are likely to have no association with a progressive gray matter atrophy, we examined whether a higher WMH burden was, independently of atrophy, associated with poorer cognitive performance in AD patients.

167 patients were included in this analysis. After adjusting for age, sex and year of education, decreased cognitive functioning, represented by the MMSE score, was associated with progressive MTA (β = −0.82, SE = 0.23, p = 0.00056) and increased PWMH burden (β = −0.22, SE = 0.05, p = 0.00005) (Figure 2). In addition, an interesting finding was a positive correlation between occipital WMH burden and MMSE score (β = 0.18, SE = 0.09, p = 0.04) in AD patients (Figure 2C).

Figure 2.

Associations between regional atrophy and WMH and log-transformed MMSE score. Lower MMSE score was associated with progressive MTA (A), increased paraventricular WMH burden (B), and lower occipital WMH burden (C). Data points represent individual patients, and the lines represent the fitted regression line with shaded areas indicating the 95% confidence intervals. WMH: white matter hyperintensities; MMSE: Mini-Mental State Examination; MTA: medial temporal atrophy.

Sensitivity analyses that include a vascular risk score or presence of hypertension as a factor did not alter these findings (Table 3).

Table 3.

Multivariate linear regression of WMH, atrophy, and vascular risk factors on cognition.

Log MMSE score∼	Primary Model (r2 = .27)	Sensitivity analysis 1 (r2 = .32)	Sensitivity analysis 2 (r2 = .32)
Age	β = 0.01, SE = 0.01, p = 0.15	β = 0.01, SE = 0.01, p = 0.21	β = 0.01, SE = 0.01, p = 0.13
Sex (Male)	β = 0.14, SE = 0.14, p = 0.3	β = 0.14, SE = 0.15, p = 0.37	β = 0.18, SE = 0.14, p = 0.21
Years of education	β = 0.05, SE = 0.02, p = 0.00372*	β = 0.05, SE = 0.02, p = 0.00149*	β = 0.05, SE = 0.02, p = 0.00223*
Frontal atrophy	β = −0.56, SE = 0.41, p = 0.17	β = 0.23, SE = 0.55, p = 0.67	β = 0.25, SE = 0.54, p = 0.64
Parietal atrophy	β = −0.12, SE = 0.22, p = 0.58	β = −0.15, SE = 0.23, p = 0.51	β = −0.15, SE = 0.23, p = 0.53
Medial temporal atrophy	β = −0.82, SE = 0.23, p = 0.00056*	β = −0.92, SE = 0.23, p = 0.00014*	β = −0.86, SE = 0.23, p = 0.00029*
Paraventricular WMH	β = −0.22, SE = 0.05, p = 0.00005*	β = −0.22, SE = 0.05, p = 0.00006*	β = −0.24, SE = 0.05, p = 0.00001*
Frontal WMH	β = 0.07, SE = 0.05, p = 0.19	β = 0.06, SE = 0.06, p = 0.33	β = 0.06, SE = 0.06, p = 0.27
Parietal WMH	β = 0.03, SE = 0.05, p = 0.61	β = −0.01, SE = 0.05, p = 0.92	β = 0.01, SE = 0.05, p = 0.91
Temporal WMH	β = 0.01, SE = 0.1, p = 0.9	β = 0.05, SE = 0.11, p = 0.64	β = 0.01, SE = 0.12, p = 0.91
Occipital WMH	β = 0.18, SE = 0.09, p = 0.04*	β = 0.18, SE = 0.09, p = 0.05*	β = 0.19, SE = 0.09, p = 0.03*
Basal ganglia WMH	β = 0.05, SE = 0.13, p = 0.71	β = 0.02, SE = 0.13, p = 0.89	β = 0.02, SE = 0.13, p = 0.91
Infratentorial WMH	β = 0.18, SE = 0.24, p = 0.46	β = 0.16, SE = 0.24, p = 0.51	β = 0.17, SE = 0.24, p = 0.49
Vascular risk score	NA	β = 0.09, SE = 0.06, p = 0.18	NA
Hypertension1	NA	NA	β = 0.26, SE = 0.16, p = 0.1

Sensitivity analysis 1: Adding vascular risk score.

Sensitivity analysis 2: Adding presence/absence of hypertension.

Discussion

In the current study, our results indicated that the severity of WMH were positively correlated with higher CMB grades and progressive Aβ42 burden in AD. These effects were confined to the paraventricular, frontal, and parietal regions, suggesting a distinct AD-related pathophysiology localized to these areas. Notably, no significant correlation was found between WMH and CSF tau levels, indicating that CSF Aβ₄₂ is more closely linked to WMH, while tau has long been considered to be associated with cortical atrophy. Additionally, no significant correlation was observed between WMH and vascular risk factors. Furthermore, we identified MTA and PWMH as two core features independently associated with cognitive impairment in AD patients. These findings underscore the potential significance of regional WMH in the context of AD pathological processes.

WMH in AD: Beyond vascular hypotheses

Consistent with recent findings, our study demonstrated that systemic vascular risk factors were not significantly associated with global or regional WMH in AD, even in frontal WMH, where a previous study identified an association with arteriolosclerosis, or in PWMH, which are particularly vulnerable to hypoperfusion and small vessel injury.^14,25,26 Notably, our cohort excluded AD patients with lacunar infarcts. This suggests that those “non-specific” WMH, unrelated to lacunar infarcts, can be driven by AD processes rather than traditional vascular risk factors. While WMHs in the general aging population closely correlate with systemic vascular risk—especially hypertension—and are widely regarded as markers of cerebral small vessel disease, this vascular-WMH relationship appears attenuated or absent in biologically confirmed AD. In this context, CAA-related microvascular dysfunction, along with neuroinflammatory processes driven by Aβ pathology, may play a more direct role in WMH genesis than traditional systemic vascular risks.^26–29

Nevertheless, it remains unclear whether these AD-associated WMHs reflect intrinsic vascular alterations secondary to AD. This highlights a key knowledge gap and underscores the need for further mechanistic studies to delineate the interplay between amyloid pathology, microvascular integrity, and white matter damage in AD. Future research should also prioritize strategies to distinguish WMHs attributable to AD pathology from those of purely vascular origin.

Topographic specificity of AD-related WMH

We observed that greater global WMH correlates with both higher CMB grade and reduced CSF Aβ₄₂, consistent with positron emission tomography (PET)-based findings in autosomal dominant AD and older adults cohorts.¹⁵ While some researchers propose that CAA mediates the relationship between WMH and Aβ, our multivariable models demonstrate that parenchymal Aβ₄₂ exerts a statistically independent effect on WMH severity beyond that of CMB-defined CAA.¹² We acknowledge that statistical independence reflects partitioned variance rather than absolute biological independence. Nonetheless, this result suggests that parenchymal and vascular amyloid may each disrupt white matter integrity via complementary mechanisms—a hypothesis supported by the clinical observation that demyelination is commonly seen in AD patients without CMBs. Future mechanistic experiments are warranted to validate and further delineate the distinct contributions of CAA and parenchymal Aβ to WMH formation.

Moreover, we observed that these correlations are primarily localized to the frontal, parietal, and paraventricular regions, rather than the occipital lobe. These contrasts with previous views suggesting that posterior WMH are associated with AD pathology, while anterior WMH are linked to vascular factors.^6,14,30,31 Our findings propose a more extensive AD-specific WMH pattern that includes anterior regions, suggesting that the impact of CAA and parenchymal amyloid pathology on WMH is more widespread than previously recognized.

Dissociation between tau pathology and WMH

We found no evidence linking increased tau burden—whether measured as CSF p-tau181 or t-tau—with global or regional WMH. However, in the basal ganglia, WMH were unexpectedly associated with lower levels of CSF p-tau181 and t-tau. While most studies report no significant association between WMH and tau biomarkers, some pathological studies have suggested that hyperphosphorylated tau pathology is an independent predictor of WMH but not Aβ.^10,32–34 This discrepancy between biomarker and pathological findings warrants further investigation.

The absence of a correlation between WMH and tau in most brain regions suggests that tau accumulation may not directly contribute to white matter damage as previously hypothesized. Instead, the observed associations in some studies may reflect a relationship between WMH and overall disease severity rather than a direct causative effect.

Limited association between brain atrophy and WMH

In contrast to previous literature, our analysis found only a weak correlation between temporal WMH and MTA, with no significant associations observed in other regions. This suggests that WMH in AD may not primarily result from neurodegeneration. While numerous studies have reported associations between WMH and atrophy, our findings align with an earlier pathological study showing that gray and white matter changes do not coincide regionally, except in the temporal lobe.^{8,9,12,15,35,36}

It is possible that the lack of association observed in our study could be attributed to the use of a visual rating scale, which may reduce sensitivity. The absence of a relationship between WMH and atrophy in our cohort at least suggests that, in AD, the contribution of Wallerian degeneration to WMH may not be the predominant factor, as the two do not align in visual rating scale.

Independent effects of atrophy and WMH on cognition

Our analysis revealed that MTA and PWMH are two core neuroimaging features linked to cognitive decline in AD. This finding is consistent with previous studies highlighting the cognitive relevance of PWMH.^37,38 Importantly, vascular risk factors did not independently affect cognitive performance or modify the association between cognitive decline and MTA or PWMH. This further supports the idea that WMH in AD reflects AD-specific pathology rather than vascular contributions.

Limitations

This study has several limitations. First, WMH and atrophy were assessed using visual rating systems, which are less sensitive than volumetric measures. Second, while we categorized WMH into specific regions, we did not further subdivide PWMH, which may have obscured more nuanced findings. Given the significance of PWMH in relation to cognitive performance in AD patients, further refinement in the assessment of PWMH are needed in future studies. Additionally, some unexpected results were observed in occipital WMH, where male sex and older age were independent risk factors, and occipital WMH appeared to have a protective effect on cognition. Though a previous study also mentioned the specificity of occipital WMH, these findings are difficult to explain and may reflect the unique neurodevelopmental and histological characteristics of the occipital lobe.³²

Conclusions

Our findings highlight that, despite the heterogeneity of WMH, those observed in biologically diagnosed AD reflect underlying AD-associated pathological changes, including amyloid accumulation and CAA. These AD-related WMH are distributed across a wide range of brain regions, including the frontal, parietal, and paraventricular regions, but notably not the occipital lobe.

The lack of association between WMH and CSF tau supports the notion that amyloid pathology is more closely linked to WMH development, while tau pathology contributes primarily to cortical atrophy. Furthermore, PWMH and MTA emerge as key neuroimaging features linked to cognitive impairment in AD.

Taken together, our findings support considering WMH—particularly those not related to lacunar infarcts—as potential supportive biomarker within the “A” (amyloid) domain of the ATNIVS diagnostic framework.³⁹

Supplemental Material

sj-docx-1-alz-10.1177_13872877251360239 - Supplemental material for Global and regional white matter hyperintensities in Alzheimer's disease: Exploring etiologies and cognitive correlates

Supplemental material, sj-docx-1-alz-10.1177_13872877251360239 for Global and regional white matter hyperintensities in Alzheimer's disease: Exploring etiologies and cognitive correlates by Yuyue Qiu, Jialu Bao, Tianyi Wang, Li Shang, Shanshan Chu, Wei Jin, Wenjun Wang, Yuhan Jiang, Bo Li, Yixuan Huang, Bo Hou, Longze Sha, Yunfan You, Yuanheng Li, Ling Qiu, Qi Xu, Feng Feng, Liling Dong, Chenhui Mao and Jing Gao in Journal of Alzheimer's Disease

Footnotes

Acknowledgments

We are grateful to the patients, caregivers, and staffs for providing the clinical information. We acknowledge that all clinical and biological samples were provided by the Clinical Biobank (ISO 20387), Peking Union Medical College Hospital, Chinese Academy of Medical Sciences. We thank the Biobank team for their support in sample management and data access.

ORCID iDs

Yuyue Qiu

Chenhui Mao

Ethical considerations

The study was conducted in accordance with the principles outlined in the Declaration of Helsinki (1975). Ethical approval was granted by the Ethics Committee of Peking Union Medical College Hospital (No. JS2810) and written informed consent was obtained from all participants and their surrogates.

Consent to participate

Written informed consent to participate in the study was obtained from all participants or their legal guardians. The consent process was approved by the Ethics Committee of Peking Union Medical College Hospital (No. JS2810).

Author contributions

Yuyue Qiu: Conceptualization; Data curation; Formal analysis; Methodology; Writing—original draft.

Jialu Bao: Data curation; Investigation; Validation.

Tianyi Wang: Data curation; Investigation.

Li Shang: Data curation; Investigation.

Shanshan Chu: Data curation; Investigation.

Wei Jin: Data curation; Investigation.

Wenjun Wang: Investigation.

Yuhan Jiang: Investigation.

Bo Li: Investigation.

Yixuan Huang: Investigation.

Bo Hou: Data curation; Investigation.

Longze Sha: Data curation; Investigation.

Yunfan You: Data curation; Investigation.

Yuanheng Li: Data curation; Investigation.

Ling Qiu: Project administration; Supervision.

Qi Xu: Project administration; Supervision.

Feng Feng: Project administration; Supervision.

Liling Dong: Project administration; Supervision

Chenhui Mao: Conceptualizat; Methodology; Resources; Supervision; Writing—review & editing.

Jing Gao: Project administration; Resources; Supervision; Writing—review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the National Key Research and Development Program of China (No. 2020YFA0804501, 2020YFA0804500); National High Level Hospital Clinical Research Funding (No. 2022-PUMCH-A-254, 2022-PUMCH-D-007); CAMS Innovation Fund for Medical Sciences (CIFMS) (No. 2021-I2M-1-020).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy or ethical restrictions.

Supplemental material

Supplemental material for this article is available online.

References

Wahlund

Barkhof

Fazekas

, et al. A new rating scale for age-related white matter changes applicable to MRI and CT. Stroke 2001; 32: 1318–1322.

Hachinski

Potter

Merskey

. Leuko-araiosis: an ancient term for a new problem. Can J Neurol Sci 1986; 13: 533–534.

Prins

Scheltens

. White matter hyperintensities, cognitive impairment and dementia: an update. Nat Rev Neurol 2015; 11: 157–165.

Wardlaw

Smith

Biessels

, et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013; 12: 822–838.

Pantoni

. Cerebral small vessel disease: from pathogenesis and clinical characteristics to therapeutic challenges. Lancet Neurol 2010; 9: 689–701.

Yoshita

Fletcher

Harvey

, et al. Extent and distribution of white matter hyperintensities in normal aging, MCI, and AD. Neurology 2006; 67: 2192–2198.

Garnier-Crussard

Bougacha

Wirth

, et al. White matter hyperintensity topography in Alzheimer’s disease and links to cognition. Alzheimers Dement 2022; 18: 422–433.

Morcillo-Nieto

Zsadanyi

Arriola-Infante

, et al. Characterization of white matter hyperintensities in down syndrome. Alzheimers Dement 2024; 20: 6527–6541.

Jiménez-Balado

Habeck

Stern

, et al. The relationship between cortical thickness and white matter hyperintensities in mid to late life. Neurobiol Aging 2024; 141: 129–139.

10.

Cha

W-J

Ahn

, et al. Association between brain amyloid deposition and longitudinal changes of white matter hyperintensities. Alzheimers Res Ther 2024; 16: 50.

11.

Wang

Lin

, et al. Correspondence between white matter hyperintensities and regional grey matter volumes in Alzheimer's disease. Front Aging Neurosci 2024; 16: 1429098.

12.

Zhang

Chen

Wang

, et al. Linking white matter hyperintensities to regional cortical thinning, amyloid deposition, and synaptic density loss in Alzheimer’s disease. Alzheimers Dement 2024; 20: 3931–3942.

13.

Garnier-Crussard

Cotton

Krolak-Salmon

, et al. White matter hyperintensities in Alzheimer’s disease: beyond vascular contribution. Alzheimers Dement 2023; 19: 3738–3748.

14.

McAleese

Miah

Graham

, et al. Frontal white matter lesions in Alzheimer’s disease are associated with both small vessel disease and AD-associated cortical pathology. Acta Neuropathol 2021; 142: 937–950.

15.

Shirzadi

Schultz

Yau

W-YW

, et al. Etiology of white matter hyperintensities in autosomal dominant and sporadic Alzheimer disease. JAMA Neurol 2023; 80: 1353–1363.

16.

Lei

Mao

, et al. CSF Biomarkers for early-onset Alzheimer’s disease in Chinese population from PUMCH dementia cohort. Front Neurol 2023; 13: 1030019.

17.

Dong

Mao

Liu

, et al. Association between common variants of APOE, ABCA7, A2M, BACE1, and cerebrospinal fluid biomarkers in Alzheimer’s disease: data from the PUMCH dementia cohort. J Alzheimers Dis 2022; 85: 1511–1518.

18.

Shang

Dong

Huang

, et al. Association of APOE ε4/ε4 with fluid biomarkers in patients from the PUMCH dementia cohort. Front Aging Neurosci 2023; 15: 1119070.

19.

Scheltens

Barkhof

Leys

, et al. A semiquantative rating scale for the assessment of signal hyperintensities on magnetic resonance imaging. J Neurol Sci 1993; 114: 7–12.

20.

Scheltens

Pasquier

Weerts

, et al. Qualitative assessment of cerebral atrophy on MRI: inter- and intra-observer reproducibility in dementia and normal aging. Eur Neurol 1997; 37: 95–99.

21.

Koedam

ELGE

Lehmann

van der Flier

, et al. Visual assessment of posterior atrophy development of a MRI rating scale. Eur Radiol 2011; 21: 2618–2625.

22.

Scheltens

Leys

Barkhof

, et al. Atrophy of medial temporal lobes on MRI in ‘probable’ Alzheimer’s disease and normal ageing: diagnostic value and neuropsychological correlates. J Neurol Neurosurg Psychiatry 1992; 55: 967–972.

23.

Kim

J-E

Choi

K-G

, et al. T1-weighted axial visual rating scale for an assessment of medial temporal atrophy in Alzheimer’s disease. J Alzheimers Dis 2014; 41: 169–178.

24.

Charidimou

Boulouis

Frosch

, et al. The Boston criteria version 2.0 for cerebral amyloid angiopathy: a multicentre, retrospective, MRI-neuropathology diagnostic accuracy study. Lancet Neurol 2022; 21: 714–725.

25.

Gronewold

Jokisch

Schramm

, et al. Association of regional white matter hyperintensities with hypertension and cognition in the population-based 1000BRAINS study. Eur J Neurol 2023; 30: 1174–1190.

26.

Brickman

Rizvi

. White matter hyperintensities and Alzheimer's disease: an alternative view of an alternative hypothesis. Alzheimers Dement 2023; 19: 4260–4261.

27.

Nasrabady

Rizvi

Goldman

, et al. White matter changes in Alzheimer’s disease: a focus on myelin and oligodendrocytes. Acta Neuropathol Commun 2018; 6: 22.

28.

Keith

Gao

F-Q

Noor

, et al.

Collagenosis of the deep medullary veins: an underrecognized pathologic correlate of white matter hyperintensities and periventricular infarction?

J Neuropathol Exp Neurol 2017; 76: 299–312.

29.

Gouw

Seewann

Vrenken

, et al. Heterogeneity of white matter hyperintensities in Alzheimer’s disease: post-mortem quantitative MRI and neuropathology. Brain 2008; 131: 3286–3298.

30.

Weaver

Doeven

Barkhof

, et al. Cerebral amyloid burden is associated with white matter hyperintensity location in specific posterior white matter regions. Neurobiol Aging 2019; 84: 225–234.

31.

McAleese

Walker

Graham

, et al. Parietal white matter lesions in Alzheimer’s disease are associated with cortical neurodegenerative pathology, but not with small vessel disease. Acta Neuropathol 2017; 134: 459–473.

32.

McAleese

Firbank

Dey

, et al. Cortical tau load is associated with white matter hyperintensities. Acta Neuropathol Commun 2015; 3: 60.

33.

Graff-Radford

Arenaza-Urquijo

Knopman

, et al. White matter hyperintensities: relationship to amyloid and tau burden. Brain 2019; 142: 2483–2491.

34.

Pietroboni

Scarioni

Carandini

, et al. CSF β-amyloid and white matter damage: a new perspective on Alzheimer’s disease. J Neurol Neurosurg Psychiatry 2018; 89: 352–357.

35.

Brun

Englund

. A white matter disorder in dementia of the Alzheimer type: a pathoanatomical study. Ann Neurol 1986; 19: 253–262.

36.

Dadar

Manera

Ducharme

, et al. White matter hyperintensities are associated with grey matter atrophy and cognitive decline in Alzheimer’s disease and frontotemporal dementia. Neurobiol Aging 2022; 111: 54–63.

37.

Griffanti

Jenkinson

Suri

, et al. Classification and characterization of periventricular and deep white matter hyperintensities on MRI: a study in older adults MRI. Neuroimage 2018; 170: 174–181.

38.

Wang

Y-L

Chen

Cai

W-J

, et al. Associations of white matter hyperintensities with cognitive decline: a longitudinal study. J Alzheimers Dis 2020; 73: 759–768.

39.

Jack Jr

Andrews

Beach

, et al. Revised criteria for diagnosis and staging of Alzheimer’s disease: alzheimer’s Association Workgroup. Alzheimers Dement 2024; 20: 5143–5169.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB