Association of Asymmetrical White Matter Hyperintensities and Apolipoprotein E4 on Cognitive Impairment

Abstract

Background:

Asymmetrical patterns of cerebral damage have been widely observed in a range of neurodegenerative diseases, including Alzheimer’s disease (AD).

Objective:

To elucidate the clinical associations of asymmetrical white matter hyperintensities (WMH) in mild cognitive impairment (MCI) and AD.

Methods:

Regional WMH asymmetry of 340 participants (90 healthy controls, 132 MCI, 118 AD) was calculated as the difference in normalized hemispheric WMH volume (WMH/ICV) adjusted for structural brain asymmetry of respective lobar regions and total WMH. WMH asymmetry was analyzed in relation to disease classification, cognition, and APOE4 status using ANCOVA and multiple regression analysis, controlling for gender, age, ethnicity, and correcting for multiple comparisons using Bonferroni correction. Moderation analysis examined interaction effects of APOE4 on associations between cognition and WMH asymmetry.

Results:

Greater left-dominant occipital WMH asymmetry was observed in AD, compared to healthy controls and MCI, and was associated with poorer global cognition, memory, language, and executive functions among cognitively impaired participants (MCI and AD). Cognitively impaired APOE4 carriers displayed greater left-dominant WMH asymmetry in the whole brain and frontal lobe, compared to non-carriers. Importantly, effects were independent of structural brain asymmetry, global cerebral atrophy, and overall WMH burden. Moderation analysis demonstrated associations between left-dominant WMH asymmetry and cognition in cognitively impaired APOE4 non-carriers, but not APOE4 carriers.

Conclusion:

Leftward asymmetry of WMH may be more pathological in nature, compared to symmetrical WMH. Furthermore, the detrimental effects of WMH asymmetry was more relevant in APOE4-negative cognitive impairment, compared to APOE4-positive which may be driven primarily by AD pathology.

Keywords

Alzheimer’s disease apolipoprotein E4 cerebral small vessel diseases cognitive aging leukoaraiosis magnetic resonance imaging mild cognitive impairment

INTRODUCTION

White matter hyperintensities (WMH) represent brain lesions caused by localized changes in tissue composition. On magnetic resonance imaging (MRI), they appear as patchy areas of increased intensity on T2-weighted and FLAIR sequences, and hypointense areas on T1-weighted imaging.

Widely considered to be indicative of underlying cerebrovascular pathology, WMH are more frequently observed in individuals with vascular risk factors such as hypertension and diabetes [1]. On top of increased risk of cerebrovascular disease, WMH have also been linked to increased risk of cognitive impairment, depression, functional disability, and early mortality [2 –8]. In terms of cognitive outcomes, the presence of WMH is closely associated with poorer cognitive functioning across a wide range of cognitive domains, including episodic memory, processing speed, and executive functions [9, 10]. Clinically, WMH burden has also been found to be related to increased risk of mild cognitive impairment (MCI) and dementia [2 , 12]—importantly, longitudinal increase of WMH has also been found to correspond with higher rates of conversion to dementia [13].

However, while WMH is often associated with disease conditions such as cerebrovascular disease and Alzheimer’s disease (AD), they are also commonly observed among healthy elderly individuals [14]. In fact, it has been demonstrated that up to 22% of healthy elderly may have moderate WMH, while 9% may have severe WMH burden [15]. This is similar to previous investigations done in Asian cohorts by our group, whereby 7% of healthy elderly had severe WMH [6]. The prevalence of WMH observed in healthy aging raises the questions of how the pathological processes differ between WMH observed in healthy aging and WMH of a more pathological nature observed in MCI and AD individuals.

At present, the majority of studies investigate WMH using measures of global volume. However, there has been increasing efforts to take WMH location into account [16, 17], and in recent years, novel methods of studying the shape of WMH have also emerged [18]. These are important advancements in the field of WMH research, considering the mounting evidence that anatomical distribution of WMH provides valuable etiological and prognostic information [18, 19].

Across the brain, patterns of asymmetrical cerebral atrophy [20], hippocampal atrophy [21], medial temporal lobe atrophy [22], and cerebral hypometabolism [23] have been observed in a variety of neurodegenerative diseases, including AD, suggesting an abnormal progression of neurodegeneration driven by pathological processes. However, to the best of our knowledge, no studies have investigated the relevance of asymmetrical WMH lesions in relation to cognition (Fig. 1).

Fig.1

Axial scans displaying (A) symmetrical and (B) asymmetrical distribution of white matter hyperintensities (WMH). Patient B displays right-dominant WMH asymmetry in the parietal lobe and left-dominant WMH asymmetry in the frontal lobe.

In this cross-sectional study, we test the hypothesis that hemispheric asymmetry of WMH volume is associated with cognitive impairment in MCI and AD, as opposed to symmetrical WMH observed in healthy aging. Specifically, we postulate that greater WMH asymmetry would be observed in patients with AD and MCI, compared to healthy controls, and that greater WMH asymmetry would be associated with poorer cognition. In addition, the associations between the apolipoprotein E ɛ4 (APOE4) allele and asymmetry in the brain have not been well researched, although the ɛ4 allele has been shown to have region-specific effects on the brain [24]. Therefore, to elucidate the underlying mechanisms of asymmetrical WMH, we further investigate the relationship between APOE4 and patterns of WMH asymmetry.

METHODS

Participants

This present study recruited 395 participants between July 2013 and December 2017 from a specialist outpatient neurology clinic (National Neuroscience Institute, Singapore). 55 participants did not have complete data, and were excluded from the study. Only the 340 participants with complete data (90 healthy controls (HC), 132 MCI, 118 AD) were included in this study. All participants underwent clinical and neuropsychological assessments, MRI imaging, as well as APOE genotyping. Classifications of MCI and AD were based on clinical diagnosis by cognitive neurologists based on the NIA-AA criteria for MCI [25] and AD [26]. All classifications were corroborated by comprehensive neuropsychological assessment and supported by structural magnetic resonance imaging (MRI) scans where available. Healthy controls were recruited from the community and were required to have a Clinical Dementia Rating score of 0 [27]. This study was approved by the Singhealth institutional ethics review board, and written informed consent was obtained from all participants prior to data collection.

Clinical, demographic, and neuropsychological assessments were performed for all participants. All cognitive tests were administered by trained psychologists through one-on-one interviews in a quiet room free of distraction. Global cognitive function was evaluated using the Mini-Mental State Examination (MMSE) and a local adaptation of the Montreal Cognitive Assessment (MoCA). Episodic memory was assessed using the immediate and delayed 10-word recall tasks from the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog); attention was measured using the Color Trails Test 1 and the Digit Span Forward; executive functions were assessed using the Color Trails Test 2 and the Digit Span Backwards; language was evaluated using a test of semantic fluency (naming animals).

APOE genotyping

Blood samples were collected from participants, and DNA was extracted using QIAamp^® DNA Blood Maxi Kit (Qiagen GmbH, Hilden, Germany). APOE genotyping was performed using TaqMan SNP genotyping assay and ABI 7900HT PCR system (Applied Biosystems, Foster City, CA).

MRI acquisition

All participants underwent 3T MRI imaging on a Siemens Allegra (Siemens, Erlangen, Germany) scanner system. Structural scans using high-resolution T1-weighted MPRAGE (axial acquisition, 176 slices, matrix size = 256×256, voxel size= 1.0×1.0×1.0 mm³, echo time (TE) = 3.2 ms, repetition time (TR) = 7 ms, inversion time (TI) = 850 ms, flip angle = 8°, field of view (FOV) = 256×256 mm²) and T2-weighted FLAIR imaging (170 slices, matrix size = 256×256, voxel size = 1.0×1.0×1.0 mm³, TE = 340 ms, TR = 8000 ms,TI =2400 ms, FOV = 240×240 mm²) were acquired for all patients.

Quantification of white matter hyperintensities

WMH volume (WMHV) was quantified using the Statistical Parametric Mapping 8 (SPM8) suite (http://www.fil.ion.ucl.ac.uk/spm/) using an automated script; details on the procedures involved and its validation have been described previously [28, 29]. Briefly, SPM8 was used to perform segmentation of T1-weighted images into grey matter, white matter, and cerebrospinal fluid, based on prior probability maps. Using the grey and white matter maps, a brain mask was created and used to perform removal of non-brain matter from the FLAIR images. WMHV was obtained using threshold-based segmentation at a threshold of 1.40 times the modal pixel intensity, i.e., lesions with pixel intensity more than 1.40 times the modal intensity were included in WMHV. A region of interest map, created in MNI space, was transformed onto the FLAIR image of each subject to segment WMHV into four lobar regions (frontal, parietal, occipital, temporal) (Fig. 2). Total intracranial volume (ICV) was obtained from the same procedure, and total brain WMHV was the sum of the lobar volumes.

Fig.2

Segmentation of white matter hyperintensities (WMH). Fron, frontal lobe; Temp, temporal lobe; Par, parietal lobe; Occ, occipital lobe.

All WMHV were normalized to the ICV to control for differences in head size. Normalized volumes were calculated as such: (WMHV/ICV)^*100%. Traditional computational formulas of asymmetry (e.g., [(WMHV_left – WMHV_right)/WMH_left]×100%) are not feasible, as certain brain regions may have absent WMH or very minimal WMH. When entered as a denominator, a volume of 0 ml on either hemisphere would result in an undefined value, while a very small WMH volume (e.g., 0.001 ml) would result in an inflation of asymmetry score. To avoid these issues, and account for WMH volume and existing asymmetries of the brain parenchymal, we took a two-step approach to quantify WMH asymmetry. Firstly, differences in WMHV between the hemispheres were computed as an asymmetry index (AI): AI = ((WMHV_left/ICV)^*100%) – (WMHV_right/ICV^*100%)). As this study sought to investigate the effects of asymmetry, as opposed to lateralization, the degree of leftward and rightward asymmetry was considered separately, and transformed such that a higher number represented greater asymmetry. Next, the raw AI was adjusted for hemispheric asymmetry of the respective lobes and total WMH volume using linear regressions with the raw regional AI (e.g., WMH asymmetry of frontal lobe) as the dependent variable and the relevant lobar volume (e.g., asymmetry of frontal lobe) and global WMH volume as predictors: raw AI = a + b₁(structural asymmetry of respective lobe) + b₂(total WMH volume) + e. The resulting residuals of this regression (e) are used to represent the extent of asymmetry in each region, independent of regional brain structure asymmetries and WMH volume.

Statistical analysis

Statistical analysis was performed using SPSS (Version 20.0, SPSS, Inc, Chicago, IL, USA). Continuous variables were assessed for normality using the Shapiro-Wilk Test. HC, MCI, and AD groups were compared on demographic variables, vascular risk factors, cognitive scores, and degree of WMHV asymmetry using independent t-tests (for normally distributed data), Mann-Whitney U tests (for nonparametric data), and chi-square tests of independence (for categorical variables). One-way Analysis of Covariance (ANCOVA) was conducted to examine the associations between WMHV asymmetry and diagnostic classification, as well as global cognition (MMSE and MoCA), controlling for possible confounders such as gender, age, ethnicity, and bilateral WMHV. To detect significant differences between each group, post-hoc pairwise comparisons were conducted, correcting for multiple comparisons using Bonferroni correction. To examine the effects of global brain atrophy, further analysis was also conducted by including total brain volume as an additional covariate. ANCOVA, controlling for the same variables, was also conducted to study the association between WMHV asymmetry and APOE4 status. Moderation analyses were conducted using the PROCESS 3.0 macro [30] on SPSS to further investigate whether the relationship between WMH asymmetry and global cognition was moderated by the presence of the APOE4 allele. To do so, we ran five separate moderation models for whole brain WMH asymmetry and each of the four lobar regions. Each moderation model was made up of WMHV asymmetry (independent variable), APOE4 (moderating variable), and MoCA (dependent variable), and was analyzed using 5000 bootstrap samples. In the first step, two variables were included: WMH asymmetry and APOE4. Next, an interaction term between WMH asymmetry and APOE4 is added to the regression model. To avoid issues of multicollinearity, variables were mean-centered prior to regression analysis. While this can be done manually through the subtraction of means, variables in this study were centered automatically as part of the PROCESS 3.0 analysis. Given the relatively large proportion of participants carrying the APOE2 allele, which has been found to be associated with greater WMH burden and left-lateralized cerebral atrophy [31], we conducted additional analysis to investigate the interaction of APOE2 status and WMH asymmetry on global cognition in APOE4-negative participants with cognitive impairment. Significance was set at a two-tailed probability value of 0.05.

RESULTS

Participant characteristics

340 participants (HC: 90, MCI: 132, AD: 118) were included in the analyses (Table 1). The study sample was majority Chinese (90.3%), and had an equal distribution of males (48.8%) and females (51.2%). The mean age of the sample was 64.50 (SD 8.82) and mean years of education was 10.94 (SD 4.17). Compared to healthy controls, MCI and AD participants were older, had fewer years of education, and were more likely to be positive for the APOE4 gene. The three groups were comparable on cardiovascular risk factors such as diabetes, hypertension, and hyperlipidemia, although MCI and AD participants were more likely to have a history of smoking compared to HC. As expected, AD participants performed significantly worse than MCI on measures of global cognition (i.e., MMSE and MoCA), and both groups performed worse than the HC group.

Table 1

Participant characteristics

	HC (n = 90)	MCI (n = 132)	AD (n = 118)	p
Participant Characteristics
Gender¹
% Male	47.8	51.5	46.6	0.721
Age²
Mean (SD)	61.30 (7.11)	65.26 (8.73)	66.08 (9.53)	0.001^**
Ethnicity¹
% Chinese	92.2	92.4	86.4
% Malay	3.3	3.0	5.1
% Indian	3.3	3.8	4.2
% Others	1.1	0.8	4.2	0.603
Education²
Mean in years (SD)	13.21 (3.20)	11.21 (4.01)	8.92 (4.04)	<0.001^**
APOE 2 status¹
% Positive	14.5	11.5	15.5	0.655
APOE 4 status¹
% Positive	14.5	27.0	35.9	0.004^**
Vascular Risk Factors¹
% Diabetes	17.9	25.6	23.8	0.416
% Hypertension	31.1	39.5	44.1	0.266
% Hyperlipidemia	50.6	61.4	50.5	0.186
% History of smoking	10.7	15.5	22.6	0.003^**
Cognition²
MMSE	28.86 (1.35)	26.68 (2.33)	19.54 (5.81)	<0.001^**
MoCA	27.63 (1.97)	24.84 (3.50)	15.96 (6.17)	<0.001^**
Raw WMHV Asymmetry²
Global	1.25 (2.17)	2.70 (5.22)	5.01 (10.69)	<0.001^**
Frontal	0.91 (1.75)	1.60 (3.02)	3.21 (6.25)	<0.001^**
Parietal	0.45 (1.34)	1.17 (2.62)	2.11 (4.73)	<0.001^**
Occipital	0.21 (0.42)	0.39 (0.60)	0.94 (2.06)	<0.001^**
Temporal	0.17 (0.39)	0.41 (0.90)	0.75 (1.39)	<0.001^**

^*p < 0.05, ^**p < 0.01. ¹Chi-square test; ²Mann-Whitney U test.

Degree of WMHV asymmetry in MCI and AD

The absolute degree of bilateral asymmetry was compared between the three diagnostic groups. In unadjusted analysis using the Mann-Whitney U test, the absolute degree of global and regional asymmetry was significantly higher in AD and MCI than healthy controls (p < 0.001) (Table 1). A one-way ANCOVA was performed to compare the degree of normalized WMHV asymmetry between the three groups, while controlling for gender, age, and ethnicity (Table 2). Significant difference in the degree of occipital WMHV asymmetry was observed [F(2,172) = 4.17, p = 0.014] between the three groups. Post hoc tests with Bonferroni correction showed that there were significant differences in occipital WMHV asymmetry between MCI and AD (p = 0.020). These associations remained statistically significant, even after inclusion of total brain volume as an additional covariate in ANCOVA.

Table 2

Group differences in adjusted white matter hyperintensity (WMH) asymmetry

	HC (n = 90)	MCI (n = 132)	AD (n = 118)	p
L > R Asymmetry
Global	–0.04 (0.48)	0.14 (0.85)	–0.09 (1.27)	0.447
Frontal	0.01 (0.48)	0.01 (0.76)	–0.02 (1.38)	0.910
Parietal	–0.08 (0.21)	0.01 (0.53)	0.05 (1.64)	0.772
Occipital	–0.12 (0.65)	–0.18 (0.64)	0.26 (1.35)	0.014^*
Temporal	0.02 (0.54)	0.04 (0.97)	–0.07 (1.33)	0.796
R > L Asymmetry
Global	–0.06 (0.28)	0.02 (0.57)	0.02 (1.66)	0.870
Frontal	–0.06 (0.49)	–0.03 (0.78)	0.09 (1.44)	0.656
Parietal	–0.08 (0.44)	0.04 (0.73)	0.03 (1.48)	0.789
Occipital	–0.06 (0.29)	0.00 (0.44)	0.06 (1.65)	0.791
Temporal	–0.09 (0.54)	0.03 (0.76)	0.03 (1.42)	0.719

^*p < 0.05. Analyses of WMH asymmetry are based on adjusted residuals, accounting for total WMH volume and structural brain asymmetry, and controlling for gender, age, and ethnicity in ANCOVA.

Associations between WMHV asymmetry and cognition

Using multiple regression analysis, we analyzed the associations between cognition and WMHV asymmetry globally and regionally, controlling for gender, age, and ethnicity. In cognitively impaired participants (MCI + AD), poorer MoCA scores were associated with greater L > R occipital lobe (p = 0.010) and L > R parietal lobe (p = 0.019) WMHV asymmetry, while poorer MMSE scores were associated with greater L > R occipital lobe asymmetry (p = 0.006). The degree of R > L asymmetry in the overall brain and in each individual region was not significantly associated with MMSE or MoCA scores. Further analysis was conducted to study individual cognitive domains and their associations to normalized WMHV asymmetry among individuals with cognitive impairment, using multiple linear regression analysis to control for gender, age, and ethnicity. Significant associations were observed between the degree of occipital lobe L > R WMHV asymmetry and performance on the ADAS-Cog Immediate Word Recall task (p = 0.016), ADAS-Cog Delayed Word Recall task (p = 0.018), the Boston Naming Test (p = 0.011), semantic fluency (p = 0.006), and Color Trails Test 2 (p = 0.035). Greater L > R WMHV asymmetry in the parietal lobe was also associated with poorer semantic fluency (p = 0.047) and Digit Span backwards (p = 0.006). Inclusion of total brain volume as a covariate in ANCOVA showed that the occipital WMH asymmetry effects were independent of global cerebral atrophy, although the association between parietal WMH asymmetry with MoCA scores and the ADAS-Cog Immediate Recall task and between occipital WMH asymmetry and Color Trails Test 2 were rendered non-significant. In healthy participants, greater leftward temporal lobe WMH was associated with poorer performance on the MMSE (p = 0.001), Digit Span backwards (p = 0.044), and ADAS-Cog Immediate Recall task (p = 0.043), while greater leftward parietal lobe WMH was associated with poorer scores on the Boston Naming Test (p = 0.034). Upon controlling for total brain volume however, only the associations between leftward WMH asymmetry in the temporal lobe with MMSE and the ADAS-Cog Immediate Recall task remained significant.

Associations between APOE4 and WMHV asymmetry

Further analysis was conducted to investigate the relationship between APOE4 and WMHV asymmetry in cognitively impaired participants (Table 3) using ANCOVA to control for gender, age, and ethnicity. Cognitively impaired participants holding the APOE4 gene displayed greater L > R asymmetry in total WMHV (p = 0.026) and the frontal lobe (p = 0.001). These associations remained significant after the addition of total brain volume as an additional covariate in ANCOVA.

Table 3

Mean and standard deviation of normalized white matter hyperintensity (WMH) asymmetry according to APOE4 status in cognitively impaired participants, i.e., mild cognitive impairment (n = 132) and Alzheimer’s disease (n = 118)

	APOE4 non-carriers (n = 155)	APOE4 carriers (n = 70)	p
L > R Asymmetry
Global	–0.13 (0.77)	0.39 (1.64)	0.026^*
Frontal	–0.19 (0.66)	0.60 (1.81)	0.001^**
Parietal	–0.07 (0.92)	0.00 (0.88)	0.647
Occipital	0.10 (1.21)	–0.03 (0.85)	0.574
Temporal	–0.10 (0.89)	0.21 (1.35)	0.131
R > L Asymmetry
Global	–0.05 (1.05)	0.15 (1.48)	0.278
Frontal	–0.10 (1.15)	0.24 (1.20)	0.119
Parietal	–0.01 (1.27)	0.05 (0.90)	0.826
Occipital	–0.08 (1.08)	0.27 (1.38)	0.054
Temporal	–0.04 (1.10)	0.19 (1.17)	0.080

^*p < 0.05, ^**p < 0.01. Analyses of WMH asymmetry are based on adjusted residuals, accounting for total WMH volume and structural brain asymmetry, and controlling for gender, age, and ethnicity in ANCOVA.

Moderating effect of APOE4 on WMH asymmetry and cognition

Given the significant associations between APOE4 possession and greater L > R WMHV asymmetry, further investigations using moderation analyses were performed to determine whether the relationship between WMH asymmetry and global cognition was moderated by the presence of the APOE4 allele in cognitively impaired individuals. As seen in Table 4, the interaction term between APOE4 and L > R asymmetry of WMH volume in the whole brain accounted for a significant amount of variance in global cognition, as measured by the MoCA: R² change = 0.05, b = 0.489, p = 0.036. Examining the interaction plots (Fig. 3), findings demonstrate that greater L > R WMH asymmetry in the whole brain was related to poorer cognition in APOE4 non-carriers, but not APOE4 carriers. This finding was similarly observed with regard to L > R WMH asymmetry in the temporal lobe: R² change = 0.05, b = 2.99, p = 0.013. Additional analysis was conducted to examine the interaction of APOE2 status and WMH asymmetry on global cognition in APOE4-negative participants with cognitive impairment, with findings demonstrating no significant interaction effects in relation to MoCA (whole brain: p = 0.206; temporal: p = 0.826) and MMSE (whole brain: p = 0.457; temporal: p = 0.669).

Fig.3

Moderating effect of APOE4 on the association between cognition and left-dominant WMH asymmetry.

Table 4

Moderating effect of APOE4 on association between global cognition (MoCA) and left-dominant white matter hyperintensity asymmetry in each lobe in cognitively impaired participants, i.e., mild cognitive impairment (n = 132) and Alzheimer’s disease (n = 118)

	b	Change in R²	p
Direct associations
APOE4 ⟶ MoCA	–1.85	–	0.047^*
Global ⟶ MoCA	–0.12	–	0.228
Frontal ⟶ MoCA	–0.10	–	0.482
Parietal ⟶ MoCA	–0.59	–	<0.001^**,^†
Occipital ⟶ MoCA	–1.98	–	0.002^**,^†
Temporal ⟶ MoCA	–0.90	–	0.067
Interaction effects
Global ^*APOE4 ⟶ MoCA	0.49	0.05	0.036^*
Frontal ^*APOE4 ⟶ MoCA	0.50	0.02	0.191
Parietal ^*APOE4 ⟶ MoCA	0.64	0.01	0.187
Occipital ^*APOE4 ⟶ MoCA	–1.49	0.01	0.335
Temporal ^*APOE4 ⟶ MoCA	2.99	0.05	0.013^*

^*p < 0.05, ^**p < 0.01. ^†Significant after adjusting for multiple comparisons using Bonferroni correction.

DISCUSSION

This present study demonstrated an association between WMH asymmetry and poorer cognitive performance, particularly when the asymmetry was left-dominant (i.e., greater WMH in left hemisphere than right hemisphere). Greater left-dominant occipital lobe asymmetry was observed in demented participants, compared to non-demented participants, and was associated with poorer global cognition, episodic memory, language, and executive functions. Poorer executive functions were also related to greater left-dominant parietal lobe WMH asymmetry. Among the cognitively impaired (MCI and dementia), presence of the APOE4 allele was associated with higher left-dominant WMH asymmetry in the whole brain and frontal lobe. On further investigation, left-dominant WMH asymmetry was found to be related to poorer cognition in APOE4 non-carriers, but not APOE4 carriers. Importantly, the effects of leftward WMH asymmetry were independent of overall WMH burden, total brain atrophy, and lobar-specific structural asymmetry.

Although this study initially sought to investigate the bidirectional asymmetry of WMH across the two hemispheres, our findings consistently point to a striking left-dominant pattern of asymmetry in relation to cognitive impairment. These findings are aligned with existing evidence that AD affects the brain in an asymmetrical pattern, with greater involvement of the left hemisphere, although the effects of leftward WMH asymmetry observed in this present study appeared independent of structural cerebral asymmetries. In prior studies, AD has been demonstrated to be associated with left-dominant asymmetry of cortical atrophy, ventricular enlargement, and reduction of glucose metabolism in the frontal and parietotemporal cortex [20, 32]. This leftward asymmetrical damage has also been observed in other neurological disorders such as semantic dementia and frontotemporal dementia [33 –35]. One possible explanation for the cognitive dysfunction observed in specifically left-dominant, but not right-dominant, patterns of WMH asymmetry may perhaps be related to existing asymmetries in neurotransmitter pathways. An extensive review by Tucker and Williamson (1984) describes distinct patterns of hemispheric asymmetry with regard to the dopamine and norepinephrine systems, whereby dopamine-dependent processes are more prevalent in the left hemisphere, while norepinephrine-dependent processes are more prevalent in the right hemisphere [36]. Tucker and Williamson (1984) propose that the asymmetry observed in neurotransmitter pathways may be responsible for systemic differences between the two hemispheres—specifically, that the left hemisphere is organized around a dopamine-driven activation system, while the right hemisphere is organized around a norepinephrine-driven system. Given the important role of the dopaminergic system on blood vessels and cerebral cortical flow [37], disruptions to dopaminergic pathways, which are commonly observed among AD patients [38], may alter the cerebrovascular integrity of the brain, resulting in greater vascular lesions such as WMH in the more dopamine-dependent left hemisphere [39]. Further research will be required to ascertain the association between asymmetries of the dopaminergic system and WMH development.

Topographically, greater degrees of left-dominant WMH asymmetry in the occipital lobe were observed in AD participants, compared to those without AD. Among cognitively impaired individuals, left-dominant asymmetry of WMH burden in the occipital lobe (Brodmann areas 17, 18, 19) was also associated with poorer cognitive functioning, specifically in the domains of episodic memory, language, and executive functions. Aside from the expected association between leftward WMH and language (given the left hemisphere’s heavier involvement in language functioning), the mechanisms behind the associations between left-dominant WMH asymmetry in the occipital lobe with episodic memory and executive functions remain unclear. One possible explanation involves disruptions to white matter tracts, including the frontal-occipital fiber tract, which connects the frontal regions (Brodmann areas 10, 45, 46) and occipital regions (Brodmann areas 18, 19) of the brain [40, 41], and when disrupted, has been found to impair episodic memory and executive functions [42].

In considering the cognitive battery used in this study, it could be argued that the association between leftward WMH asymmetry and cognition may be due to the predominantly verbal nature of cognitive tests used (e.g., digit span, ADAS-Cog word recall, verbal fluency), or the use of tests involving linear processing, both of which are left-dominant tasks. As such, it is possible that leftward microstructural damage to the existing white matter may explain the poorer performance on the administered tests. Further research is needed to establish whether the effects observed in this present study extend to non-verbal tests of memory, attention, and executive function.

Notably, APOE4 carriers displayed greater left-dominant WMH asymmetry in the whole brain and the frontal lobe, even after taking into account total WMH volume, structural brain asymmetries, gender, age, and ethnicity. It is plausible that this effect may be attributed to the cerebrovascular involvement of APOE4 in blood-brain barrier dysfunction and reduced cerebral blood flow [43], which has been proposed to be responsible, at least in part, for the left hemisphere’s stronger vulnerability to ischemic damage [44]. Given the observed association between APOE4 and WMH asymmetry, further analysis was conducted to determine whether the relationship between WMH asymmetry and cognition was moderated by APOE4, i.e., whether the presence or absence of the APOE4 allele modifies how strongly WMH asymmetry and cognition are related to each other. This moderation analysis demonstrated that possession of the APOE4 allele did indeed have a moderating role on the relationship between WMH asymmetry and cognition, whereby greater left-dominant WMH burden was associated with poorer cognitive functions in APOE4 non-carriers, but not APOE4 carriers. These findings suggest that among APOE4 non-carriers, asymmetrical accumulation of WMH across the hemispheres may play an important role in cognitive impairment, while in APOE4 carriers, other factors such as AD pathology may play a more central role in cognitive dysfunction, rendering WMH asymmetry less relevant in the development of cognitive impairment. Conversely, recent evidence implicates APOE2 in leftward asymmetry of cerebral atrophy, as well as greater WMH burden [31]. As our sample comprised of a relatively large proportion of APOE2 carriers, we conducted further analysis to examine the role of APOE2 in the relationship between WMH asymmetry and cognition, but found that APOE2 was unable to explain the associations observed in APOE4 non-carriers. Future studies on the role of WMH and other small vessel disease markers in APOE4-negative patients are required to uncover the underlying mechanisms involved.

To our knowledge, this is the first study investigating the associations between asymmetrical WMH and cognitive impairment. Few studies have studied the cognitive correlates of WMH properties such as location and shape, and none have investigated cognition with regard to WMH asymmetry. Strengths of this study include its relatively large sample size, the use of well-structured diagnostic criteria in classifying disease groups, high resolution imaging, the measurements of regional WMH volumes, and the inclusion of APOE4 analysis. Furthermore, this study accounted for bilateral WMH burden and structural asymmetries of regional brain volumes, allowing us to determine that the associations observed were independent of WMH burden and structural brain asymmetries. However, this study is limited by the inability to confirm the diagnosis of AD using biomarkers such as amyloid-β. Combined with the relatively low proportion of APOE4 carriers among AD patients, it is a possibility that a proportion of this group might reflect a non-AD dementia pathology instead. Due to the exploratory nature of our moderation analysis, findings were not adjusted for multiple comparisons. As such, further a priori studies will be required to confirm this finding. Furthermore, due to the cross-sectional design of this study, the mechanisms behind asymmetrical WMH cannot be determined. To further elucidate the neuropathology underlying left-dominant asymmetry, more investigation will be needed to study the topographical and temporal relationships between the patterns of asymmetrical cerebrovascular damage, structural changes, and metabolic activity.

Conclusion

This present study demonstrates the significant associations between WMH asymmetry and cognition, highlighting the importance of considering the hemispheric distribution of WMH in clinical evaluations. These novel findings suggest that asymmetry in WMH volume across the left and right hemispheres might be indicative of a more pathological variant of WMH, compared to symmetrical WMH. Importantly, the interaction effects of APOE4 on the relationship between WMH asymmetry and cognition suggests distinct etiology underlying the cognitive dysfunctions in different APOE genotypes, with WMH asymmetry playing a greater role in those without the APOE4 allele.

Footnotes

ACKNOWLEDGMENTS

The research was supported by the National Neuroscience Institute (Singapore) in the design and conduct of the study. This study was funded by the National Medical Research Council (NMRC/CSA/0063/2014).

Authors’ disclosures available online ().

References

Gebeily

, Fares

, Kordahi

, Khodeir

, Labaki

, Fazekas

(2014) Cerebral white matter hyperintensities (WMH): An analysis of cerebrovascular risk factors in Lebanon. Int J Neurosci 124, 799–805.

Debette

, Markus

(2010) The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: Systematic review and meta-analysis. Br Med J 341, c3666.

O’Brien

, Perry

, Barber

, Gholkar

, Thomas

(2000) The association between white matter lesions on magnetic resonance imaging and noncognitive symptoms. Ann N Y Acad Sci 903, 482–489.

Firbank

, O’Brien

, Pakrasi

, Pantoni

, Simoni

, Erkinjuntti

, Wallin

, Wahlund

L-O

, Straaten

I van

, Inzitari

(2005) White matter hyperintensities and depression—preliminary results from the LADIS study. Int J Geriatr Psychiatry 20, 674–679.

Inzitari

, Simoni

, Pracucci

, Poggesi

, Basile

, Chabriat

, Erkinjuntti

, Fazekas

, Ferro

, Hennerici

, Langhorne

, O’Brien

, Barkhof

, Visser

, Wahlund

L-O

, Waldemar

, Wallin

, Pantoni

(2007) Risk of rapid global functional decline in elderly patients with severe cerebral age-related white matter changes: The LADIS study. Arch Intern Med 167, 81–88.

Kandiah

, Chander

, Ng

, Wen

, Cenina

, Assam

(2015) Association between white matter hyperintensity and medial temporal atrophy at various stages of Alzheimer’s disease. Eur J Neurol 22, 150–155.

Chander

, Lam

BYK

, Lin

, Ng

AYT

, Wong

APL

, Mok

VCT

, Kandiah

(2017) Development and validation of a risk score (CHANGE) for cognitive impairment after ischemic stroke. Sci Rep 7, 12441.

Yatawara

, Ng

, Chander

, Kandiah

(2018) Associations between lesions and domain-specific cognitive decline in poststroke dementia. Neurology 91, e45–e54.

Gunning-Dixon

, Raz

(2000) The cognitive correlates of white matter abnormalities in normal aging: A quantitative review. Neuropsychology 14, 224–232.

10.

Burton

, Kenny

, O’Brien

, Stephens

, Bradbury

, Rowan

, Kalaria

, Firbank

, Wesnes

, Ballard

(2004) White matter hyperintensities are associated with impairment of memory, attention, and global cognitive performance in older stroke patients. Stroke 35, 1270–1275.

11.

DeCarli

, Miller

, Swan

, Reed

, Wolf

, Carmelli

(2001) Cerebrovascular and brain morphologic correlates of mild cognitive impairment in the National Heart, Lung, and Blood Institute Twin Study. Arch Neurol 58, 643–647.

12.

Prasad

, Wiryasaputra

, Ng

, Kandiah

(2011) White matter disease independently predicts progression from mild cognitive impairment to Alzheimer’s disease in a clinic cohort. Dement Geriatr Cogn Disord 31, 431–434.

13.

Jokinen

, Kalska

, Ylikoski

, Madureira

, Verdelho

, van der Flier

, Scheltens

, Barkhof

, Visser

, Fazekas

, Schmidt

, O’Brien

, Waldemar

, Wallin

, Chabriat

, Pantoni

, Inzitari

, Erkinjuntti

(2009) Longitudinal cognitive decline in subcortical ischemic vascular disease - The LADIS Study. Cerebrovasc Dis 27, 384–391.

14.

de Groot

, de Leeuw

F-E

, Oudkerk

, Hofman

, Jolles

, Breteler

MMB

(2000) Cerebral white matter lesions and depressive symptoms in elderly adults. Arch Gen Psychiatry 57, 1071–1076.

15.

Hunt

, Orrison

, Yeo

, Haaland

, Rhyne

, Garry

, Rosenberg

(1989) Clinical significance of MRI white matter lesions in the elderly. Neurology 39, 1470–1474.

16.

van der Lijn

, Verhaaren

BFJ

, Ikram

, Klein

, de Bruijne

, Vrooman

, Vernooij

, Hammers

, Rueckert

, van der Lugt

, Breteler

MMB

, Niessen

(2012) Automated measurement of local white matter lesion volume. Neuroimage 59, 3901–3908.

17.

Wen

, Sachdev

(2004) The topography of white matter hyperintensities on brain MRI in healthy 60- to 64-year-old individuals. Neuroimage 22, 144–154.

18.

De Bresser

, Kuijf

, Zaanen

, Viergever

, Hendrikse

, Biessels

, Algra

, Van Den Berg

, Bouvy

, Brundel

, Heringa

, Kappelle

, Leemans

, Luijten

, Mali

WPTM

, Rutten

GEHM

, Vincken

, Zwanenburg

(2018) White matter hyperintensity shape and location feature analysis on brain MRI; Proof of principle study in patients with diabetes. Sci Rep 8, 1893.

19.

Artero

, Tiemeier

, Prins

, Sabatier

, Breteler

, Ritchie

(2004) Neuroanatomical localisation and clinical correlates of white matter lesions in the elderly. J Neurol Neurosurg Psychiatry 75, 1304–1308.

20.

Derflinger

, Sorg

, Gaser

, Myers

, Arsic

, Kurz

, Zimmer

, Wohlschläger

, Mühlau

(2011) Grey-matter atrophy in Alzheimer’s disease is asymmetric but not lateralized. J Alzheimers Dis 25, 347–357.

21.

Fox

, Warrington

, Freeborough

, Hartikainen

, Kennedy

, Stevens

, Rossor

(1996) Presymptomatic hippocampal atrophy in Alzheimer’s disease. Brain 119, 2001–2007.

22.

Thompson

, Patterson

, Hodges

(2003) Left/right asymmetry of atrophy in semantic dementia: Behavioral-cognitive implications. Neurology 61, 1196–1203.

23.

Grady

, Haxby

, Schlageter

, Berg

, Rapoport

(1986) Stability of metabolic and neuropsychological asymmetries in dementia of the Alzheimer type. Neurology 36, 1390–1392.

24.

Liu

, Paajanen

, Westman

, Wahlund

L-O

, Simmons

, Tunnard

, Sobow

, Proitsi

, Powell

, Mecocci

, Tsolaki

, Vellas

, Muehlboeck

, Evans

, Spenger

, Lovestone

, Soininen

, AddNeuroMed Consortium (2010) Effect of APOE ɛ4 allele on cortical thicknesses and volumes: The AddNeuroMed Study. J Alzheimers Dis 21, 947–966.

25.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

26.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

, Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

27.

Morris

(1993) The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology 43, 2412–2414.

28.

Firbank

, Minett

, O’Brien

(2003) Changes in DWI and MRS associated with white matter hyperintensities in elderly subjects. Neurology 61, 950–954.

29.

Smart

, Firbank

, O’Brien

(2011) Validation of automated white matter hyperintensity segmentation. J Aging Res 2011, 391783.

30.

Hayes

(2013) Introduction to mediation, moderation, and conditional process analysis: A regression-based approach, Guilford Press, New York, NY, US.

31.

Groot

, Sudre

, Barkhof

, Teunissen

, van Berckel

BNM

, Seo

, Ourselin

, Scheltens

, Cardoso

, van der Flier

, Ossenkoppele

(2018) Clinical phenotype, atrophy, and small vessel disease in APOEɛ2 carriers with Alzheimer disease. Neurology 91, e1851LP–e1859.

32.

Fazekas

(1989) Magnetic resonance signal abnormalities in asymptomatic individuals: Their incidence and functional correlates. Eur Neurol 29, 164–168.

33.

Boccardi

, Laakso

, Bresciani

, Galluzzi

, Geroldi

, Beltramello

, Soininen

, Frisoni

(2003) The MRI pattern of frontal and temporal brain atrophy in fronto-temporal dementia. Neurobiol Aging 24, 95–103.

34.

Hooten

, Lyketsos

(1996) Frontotemporal dementia: A clinicopathological review of four postmortem studies. J Neuropsychiatry Clin Neurosci 8, 10–19.

35.

Jeong

, Park

, Cho

, Kim

, Kang

, Kim

, Kang

, Na

(2005) Pattern of glucose hypometabolism in frontotemporal dementia with motor neuron disease. Neurology 64, 734–736.

36.

Tucker

, Williamson

(1984) Asymmetric neural control systems in human self-regulation. Psychol Rev 91, 185–215.

37.

Krimer

, Muly

, Williams

, Goldman-Rakic

(1998) Dopaminergic regulation of cerebral cortical microcirculation. Nat Neurosci 1, 286–289.

38.

Storga

, Vrecko

, Birkmayer

JGD

, Reibnegger

(1996) Monoaminergic neurotransmitters, their precursors and metabolites in brains of Alzheimer patients. Neurosci Lett 203, 29–32.

39.

Iadecola

(1998) Neurogenic control of the cerebral microcirculation: Is dopamine minding the store? Nat Neurosci 1, 263–265.

40.

Schmahmann

, Pandya

(2007) The complex history of the fronto-occipital fasciculus. J Hist Neurosci 16, 362–377.

41.

Hua

, Oishi

, Zhang

, Wakana

, Yoshioka

, Zhang

, Akhter

, Li

, Huang

, Jiang

, van Zijl

, Mori

(2009) Mapping of functional areas in the human cortex based on connectivity through association fibers. Cereb Cortex 19, 1889–1895.

42.

Smith

, Salat

, Jeng

, McCreary

, Fischl

, Schmahmann

, Dickerson

, Viswanathan

, Albert

, Blacker

, Greenberg

(2011) Correlations between MRI white matter lesion location and executive function and episodic memory. Neurology 76, 1492–1499.

43.

Bell

, Winkler

, Singh

, Sagare

, Deane

, Wu

, Holtzman

, Betsholtz

, Armulik

, Sallstrom

, Berk

, Zlokovic B

(2012) Apolipoprotein E controls cerebrovascular integrity via cyclophilin A. Nature 485, 512–516.

44.

Mullaart

, Daniëls

, Hopman

, de Haan

, Stoelinga

, Rotteveel

(1995) Asymmetry of the cerebral blood flow: An ultrasound Doppler study in preterm newborns. Pediatr Neurol 13, 319–322.