White Matter Hyperintensities and Apolipoprotein E Affect the Association Between Mean Arterial Pressure and Objective and Subjective Cognitive Functioning in Older Adults

Abstract

Background:

White matter hyperintensities (WMH) show a robust relationship with arterial pressure as well as objective and subjective cognitive functioning. In addition, APOE ɛ4 carriership may influence how arterial pressure affects cognitive functioning.

Objective:

To determine the role of region-specific WMH burden and APOE ɛ4 carriership on the relationship between mean arterial pressure (MAP) and cognitive function as well as subjective cognitive decline (SCD).

Methods:

The sample consisted of 87 cognitively unimpaired middle-aged to older adults aged 50–85. We measured WMH volume for the whole brain, anterior thalamic radiation (ATR), forceps minor, and superior longitudinal fasciculus (SLF). We examined whether WMH burden mediated the relationship between MAP and cognition (i.e., TMT-A score for processing speed; Stroop performance for executive function) as well as SCD (i.e., Frequency of Forgetting (FoF)), and whether APOE ɛ4 carriership moderated that mediation.

Results:

WMH burden within SLF mediated the effect of MAP on Stroop performance. Both whole brain and ATR WMH burden mediated the effect of MAP on FoF score. In the MAP–WMH–Stroop relationship, the mediation effect of SLF WMH and the effect of MAP on SLF WMH were significant only in APOE ɛ4 carriers. In the MAP–WMH–FoF relationship, the effect of MAP on whole brain WMH burden was significant only in ɛ4 carriers.

Conclusion:

WMH burden and APOE genotype explain the link between blood pressure and cognitive function and may enable a more accurate assessment of the effect of high blood pressure on cognitive decline and risk for dementia.

Keywords

Aging Alzheimer’s disease apolipoprotein E4 blood pressure dementia executive function subjective cognitive decline white matter hyperintensities

INTRODUCTION

Vascular risk factors are crucial for healthier aging. They increase the risk of a wide variety of diseases prevalent in older adults, such as cerebrovascular disease, depression, and dementia [1 –3]. Studies have suggested that vascular risk factors are associated with cognitive decline in normal aging or dementia, although the results have been somewhat mixed [4 –6]. At the same time, both vascular risk factors and cognitive decline have been consistently associated with white matter hyperintensities (WMH) in healthy older adults and older adults with dementia [7 –10]. WMH, which are considered manifestations of cerebral small vessel disease (CSVD), are commonly found in the magnetic resonance brain images (MRI) of older adults [2, 11]. Studies have found that the major underlying pathologies of WMH include demyelination, axonal loss around perivascular space, and blood-brain barrier (BBB) breakage [12 –14]. WMH have been associated with vascular risk factors such as hypertension [10, 15], high body mass index [16], diabetes [17], and smoking [18]. Particularly, the associations have been more pronounced for blood pressure-related measures [19]. Elevated systolic and diastolic blood pressure [20, 21], hypertension [15, 18], and mean arterial pressure [22, 23] have been consistently associated with higher WMH burden. It was found that among individuals taking antihypertensive medication, those who had normal blood pressure had less WMH burden than those who were still hypertensive [24]. WMH burden also has been associated with cognitive function, particularly, executive function and processing speed, in healthy older adults [25 –27]. WMH burden also predicts increased risk of pathological age-related cognitive decline, such as vascular dementia and Alzheimer’s disease (AD) [28, 29]. Considering that WMH burden associates with blood pressure as well as pathological and non-pathological age-related cognitive decline and considering that there have been mixed results about the direct relationship between vascular risk factors and cognitive functioning in older adults, blood pressure may indirectly affect cognitive functioning as well as the risk of developing dementia through the mediation of WMH burden. An earlier study reported the mediating roles of WMH burden on the relationship between hypertension and global cognition [30]. However, whether WMH burden mediates the effect of the level of blood pressure on specific domains of cognitive function remains to be investigated. Moreover, the extant literature provides evidence of regional specificity in the association between CSVD and cognitive domains, showing that higher CSVD burden in the anterior thalamic radiation (ATR) and forceps minor (Fmin) associates with slower processing speed, and higher CSVD burden in the ATR, Fmin, and superior longitudinal fasciculus (SLF) associates with lower executive performance [31 –35]. Further investigations extending previous research are needed to examine whether WMH burden shows such regional specificity when mediating the effect of blood pressure on specific domains of cognitive function. It is also not clear whether WMH play a mediating role in the link between blood pressure and dementia risk. Previous findings suggest that vascular risk factors play their utmost role as risk factors for dementia at the early stages of the disease [36, 37]. Also, subjective cognitive decline (SCD), which is often related to a higher level of AD biomarkers [38, 39] and is assumed to be a risk factor for future cognitive decline and AD dementia [40 –42], was found to associate with increased WMH burden in our previous observation based on a subset of the current sample [43]. Thus, it is likely that blood pressure indirectly predicts SCD through the mediation of WMH. However, this needs empirical testing.

Moreover, there may be a potential moderator for the mediation relationships among blood pressure, WMH, and cognition or SCD. Apolipoprotein E (APOE) ɛ4 allele, a primary genetic risk factor for AD, has been associated with a greater WMH burden [44 –46]. It was found that expression of APOE ɛ4 contributes to the breakage of BBB, one of the primary causes of WMH, by activating the proinflammatory cyclophilin A-matrix metalloproteinase-9 pathway that triggers pericyte injury [47, 48]. There is also evidence that the APOE ɛ4 allele associates with neurovascular abnormalities, such as higher tortuosity of small veins [49], and that the existence of the APOE ɛ4 allele exacerbates the effect of vascular risk factors on WMH burden [50, 51]. Thus, it seems plausible that ɛ4 carriership moderates the blood pressure–WMH–cognition mediation relationship by interacting with blood pressure.

The main goal of this study was to examine the mediating role of WMH burden in the association between blood pressure and cognitive functioning as well as the risk for pathological cognitive aging, i.e., dementia, and to investigate the moderating role of APOE ɛ4 carriership in the blood pressure–WMH–cognitive function mediation relationship. We specifically chose MAP from among blood pressure-related measures considering the physiological state MAP represents. Because MAP is determined by cardiac output and peripheral vascular resistance, it is likely to be more directly related to the resistance in small vessels [52, 53], as evidenced by the marked association between MAP and WMH [21, 22]. We chose processing speed and executive function as the cognitive domains of interest because these have been most consistently associated with CSVD and other cerebrovascular diseases in previous studies [25 –27]. In addition to whole brain WMH, we also tested the mediating role of WMH burden in specific white matter tracts, ATR, Fmin, and SLF, considering their observed associations with processing speed and executive function [35, 54]. We hypothesized that greater WMH burden would mediate the association between higher MAP and worse cognitive functioning, and that WMH in specific tracts would be a stronger mediator than global WMH. Specifically, we expected that greater WMH burden in ATR and Fmin would mediate the association between higher MAP and slower processing speed, and the WMH burden in ATR and SLF would be a stronger mediator for executive function based on the aforementioned previous findings of the tract specificity in the association between WMH burden and cognitive domains [31 –35]. We also hypothesized that greater WMH burden would mediate the association between higher MAP and more SCD based on the finding that more SCD is significantly associated with more WMH burden [43] and the lack of clear evidence for a direct relationship between blood pressure and SCD. Lastly, we expected that APOE ɛ4 carriership would modulate the mediation relationships by interacting with MAP considering the previous findings on the effect of ɛ4 carriership on cerebrovascular damage [45 –48].

MATERIALS AND METHODS

Participants

The data included 91 healthy late middle-aged to older adults (34 men, 57 women; 50–85 years, M = 67.81, SD = 8.28). Participants were recruited in Metro Detroit, Michigan, and Leiden, the Netherlands from the community and through memory clinics. Exclusion criteria were diagnosed neurological or psychiatric illness, significant cognitive decline (i.e., mild cognitive impairment (MCI) or dementia), and MRI contraindications such as implanted electrical devices or metallic clips. Participants underwent a brain MRI and neuropsychological assessments. We determined whether participants were cognitively unimpaired based on either clinical assessment or performance on the Mini-Mental State Examination (MMSE≥24) [55] and Wechsler Memory Scale–IV (no more than two indices lower than 2 SD below the normative mean). This cutoff was selected to minimize the risk of misclassifying cognitively unimpaired older adults as impaired based on the following rationales. First, older adults often perform worse on neuropsychological tests when they complete multiple tests [56 –58] as our participants did. Moreover, all participants completed the adult battery of the Wechsler Memory Scale–IV, which provides normative scores up to age 69. Therefore, the normed scores of nearly half of the participants were deflated as they were older than 69. Lastly, our Detroit sample largely consisted of African American participants. It is well established that cognitively unimpaired African American adults are likely to score lower on neuropsychological tests than cognitively unimpaired white adults [59, 60]. We aimed to minimize the risk of misclassifying cognitively unimpaired African American participants as impaired, which may occur when there are no appropriate culture- or ethnicity-adjusted test norms. The institutional review boards approved the study in both sites, and we obtained informed consent from all participants.

Procedures

We screened potential participants via phone to determine whether they qualified for the study according to the inclusion criteria listed in the Participants section. For screening purposes, we obtained basic demographic information including age, gender, ethnicity, handedness, level of education, health information including medical history of diagnosis of neurological, psychiatric, and cardiovascular disorders, and medication use. Furthermore, we screened for MRI eligibility and cognitive status using the MMSE [61] and Wechsler Memory Scale–IV [62]. Qualified participants first visited for an MRI session held at either the Wayne State University MR Research Facility in Detroit, MI or the Leiden Institute for Brain and Cognition at Leiden University Medical Center, the Netherlands. The MRI scan session took about three hours, including the scan, paperwork, and prep-time. Participants were scheduled to return for a three-hour cognitive test session within one week after the MRI session. Blood pressure, mood, memory function, and executive function were measured during the cognitive test session. Participants were allowed to take their regular medications on the day of both sessions. No participants were taking psychoactive medication, which was one of the exclusion criteria for the study. The MRI and cognitive test sessions were carried out during regular working hours. All MRI and cognitive test sessions were conducted according to standardized procedures approved by the institutional review boards at both sites.

Vascular risk factors

We measured systolic and diastolic blood pressure and obtained information regarding the use of anti-hypertension medication. Blood pressure was measured using an electric arm blood pressure cuff on the day of a cognitive testing session at least 5 min after participants settled down. The blood pressure cuff was put around the participants’ left upper arm, at about the level of the participant’s heart. We recorded the systolic and diastolic blood pressure shown on the blood pressure monitor. MAP was calculated using the following formula.

$MAP = 1 / 3 \times (systolic blood pressure + diastolic blood pressure \times 2) .$ (1)

Apolipoprotein E genotyping

All apolipoprotein E genotyping was performed at the Wayne State University Applied Genomics Technology Center. DNA was extracted from saliva samples using the Qiagen EZ1 Advanced Nucleic Acid Purification System in conjunction with the EZ1 DNA Tissue Kit and the EZ1 DNA Tissue Card. We determined APOE genotypes based on two single nucleotide polymorphisms (SNPs), rs429358 and rs7412. Samples were analyzed on Applied Biosystems Quantstudio 12K Flex Real-Time PCR Instrument using the following protocol for thermocycling: 95°C for 5 min and 40 cycles of 95°C for 5 s and 60°C for 30 s. Because there is evidence that the APOE ɛ2 allele plays a protective role for cardiovascular disease as well as AD [63, 64], we limited our analyses involving APOE ɛ4 carriership to individuals who had no ɛ2 allele. APOE ɛ4 homozygotes (ɛ4/ɛ4) and heterozygotes (ɛ3/ɛ4) were classified as APOE ɛ4 carriers, and ɛ3 homozygotes (ɛ3/ɛ3) were classified as APOE ɛ4 non-carriers. We obtained APOE genotype information from 85 of 91 participants.

Neuropsychological assessments

We administered the following neuropsychological tests to measure cognitive functioning: part A of the Trail Making Test (TMT-A) [65] and the Stroop test [66]. Time to complete TMT-A was a measure of processing speed. Thus, a lower TMT-A score indicates higher processing speed. The Stroop Interference score, computed by subtracting the average time needed to complete the Stroop Color test (t_C) and the Stroop Word test (t_W) from the time needed to complete the Color-Word test (t_ChboxW) (Interference = t_ChboxW –(t_C + t_W)/2) [67], was our measure of executive function, with a lower interference score denoting better performance.

Assessment of subjective cognitive decline

We used the frequency of forgetting (FoF) subscale of the Memory Functioning Questionnaire [68] to assess subjective cognitive function. The FoF subscale consists of 7-point Likert scale questions on the frequency of experiencing memory problems in specific situations and subjective judgment of general memory performance. As a higher score in the original scale indicates lower perceived cognitive decline, we inversed the FoF score so that a higher score would reflect worse perceived cognitive performance. The possible mean scores of inversed FoF subscale range from 1 to 7.

MRI acquisition

In the United States, we obtained images on a 3T Siemens Magnetom Verio scanner (Siemens Medical AG, Erlangen, Germany) using a 32-channel head coil at the Wayne State University MR Research Facility in Detroit, MI. 3D fluid attenuated inverse recovery (FLAIR) images had the following parameters: repetition time (TR) = 8440 ms, echo time (TE) = 122 ms, inversion time (TI) = 2500 ms, 72 slices, field of view (FOV) = 256×256×144 mm, voxel size = 1.3×1.3×2.0 mm, scan duration ∼4min. High-resolution 3D magnetization-prepared rapid gradient echo (MP-RAGE) sequence T1-weighted structural images, which were used to register individual’s FLAIR image to a high-resolution structural image, had the following parameters: TR =1680 ms, TE = 3.51 ms, TI = 900 ms flip angle = 9°, 176 slices, FOV = 256×256×236 mm, voxel size = 0.67×0.67×1.34 mm, scan duration ∼6 min. In the Netherlands, MRI scanning occurred on a 3T Philips Achieva TX scanner with a 32-channel head coil at the Leiden Institute for Brain and Cognition, and 3D FLAIR scans and 3D MP-RAGE T1-weighted scans had the following parameters. 3D FLAIR scans: TR 4800 ms, TE = 1650 ms, TI =50 ms, 162 slices, FOV = 250×250×180 mm, voxel size = 1.04×1.04×1.11 mm, scan duration ∼2.5min. 3D T1-weighted scans: TR = 9.7 ms, TE = 4.6ms, flip angle 8°, 140 slices, FOV 224×177×168mm, voxel size 1.17×1.17×1.2 mm, scan duration ∼5 min.

White matter hyperintensity segmentation

All WMH segmentation was performed at Wayne State University. FLAIR images were first corrected for field inhomogeneity using the fsl-anat tool from the FMRIB Software Library (FSL, v5.0.11) toolbox (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki) and brain-extracted using the Brain Extraction Tool (BET) [69] from FSL. WMH were segmented on the bias-corrected FLAIR images using the lesion prediction algorithm [70, Chapter 6.1] as implemented in the LST toolbox version 2.0.15 (http://www.statistical-modelling.de/lst.html) for SPM. The WMH probability maps produced with LST were binarized with a threshold of 0.4 to obtain binary whole brain WMH maps. To remove false positives in non-white matter areas, we applied a white matter mask generated using BIANCA (Brain Intensity AbNormality Classification Algorithm) [71] on the WMH maps obtained using LST. We inspected each individual WMH map and manually removed residual false positives and false negatives. Figure 1A shows a whole brain WMH map of an exemplary subject. To obtain tract-specific WMH maps, we masked the whole brain WMH maps with ATR, Fmin, and SLF tract masks (Fig. 1B) from the Johns Hopkins University White Matter Tractography Atlas [72] binarized at the threshold of 0, which is JHU max probability map (JHU-ICBM-tracts-maxprob-thr0-2 mm.nii.gz) included in FSL, and then measured WMH volume (mm³) within the masked area. We chose to use a map binarized with a low threshold to maximize the number of the participants showing WMH on the tracts of interest, considering that most of the participants, who were cognitively unimpaired, did not have a high WMH load. Whole brain WMH volume was measured in each participant’s native space. We defined whole brain WMH burden as the percentage of the volume of WMH out of the person’s total cranial volume to adjust for the individual differences in intracranial volume (ICV) and calculated it as WMH volume (mm³) ÷ ICV (mm³) × 100. Tract-specific WMH burden was measured in MNI standard space, in which the tracts of the JHU White Matter Tractography Atlas are defined. Similar to the definition of whole brain WMH volume, tract-specific WMH volume was defined as the percentage of the volume of WMH within a tract out of the total volume of the tract and calculated as WMH volume within a tract ÷ the tract volume × 100. For tract-specific WMH volume, the individual differences in ICV were not directly accounted for because WMH volume was calculated after individuals’ brains were registered to the same MNI standard space.

Fig. 1

(A) An example of WMH segmentation performed on one of the participants. (B) From the left, ATR, Fmin, and SLF tract masks from Johns Hopkins University White Matter Tractography Atlas. The red-colored area shows an example of a segmented WMH mask registered to MNI space. ATR, anterior thalamic radiation; Fmin, forceps minor; SLF, superior longitudinal fasciculus; WMH, white matter hyperintensities.

Statistical analysis

We excluded outliers (i.e., cases that deviated by more than 3.29 SD from the sample mean on at least one of the variables, following Tabachnick & Fidell [73] prior to analysis. Three participants were excluded for being outliers in whole brain WMH volume (n = 1) and in TMT-A score (n = 2). Missing values were dealt with using listwise deletion on a model-by-model basis. Thus, participant counts slightly varied across models. We square root transformed all measures of WMH volume to satisfy the normality assumption of regression analysis. We first performed multiple linear regression analyses examining the direct association between MAP and WMH volume as well as the association between WMH volume and cognitive function or FoF. In the first set of regression models, MAP was entered as the independent variable, and WMH volume in the ATR, Fmin, SLF, or whole brain was entered as the dependent variable. In another set of regression models, WMH volume in the ATR, Fmin, SLF, or whole brain was entered as the independent variable, and TMT-A, Stroop Interference score, or FoF was entered as the dependent variable. These initial regression models testing for direct associations between variables revealed significant associations between MAP and WMH volume in all regions except Fmin; between Stroop performance and WMH volume in ATR and SLF; between TMT-A and WMH in ATR and Fmin; and between FoF and WMH volume in the whole brain, ATR, and Fmin. See Supplementary Tables 1 and 2 for the full results. Based on the significant associations found in the above regression analyses, we fit mediation models with MAP as the independent variable, FoF, TMT-A score, or Stroop interference score as the dependent variable, and WMH volume as the mediator (See Fig. 2 for the schematics of the mediation models tested). We next fit moderated mediation models to test whether APOE ɛ4 carriership moderated the effect of MAP on WMH volume (Fig. 3A, B) by adding APOE ɛ4 carriership as the moderator to the mediation models in which the mediation effect was significant. We also additionally fit alternative moderated mediation models testing whether APOE ɛ4 carriership moderated the effect of WMH volume on cognitive function or FoF (Fig. 3C, D). All statistical models included baseline age, sex, and test site as covariates. The significance of effects was determined by Bonferroni corrected 95%bootstrap confidence interval (CI) computed using 5000 bootstrap samples. For the two mediation models with Stroop performance as the dependent variable (Fig. 2A) and two other models with TMT-A score as the dependent variable (Fig. 2B), 97.5%(1–(0.05/2) = 97.5 %) CI was applied. For the three mediation models with FoF as the dependent variable (Fig. 2C), 98.3%(1–(0.05/3) = 98.3%) CI was applied. For the moderated mediation models with Stroop performance as the dependent variable (Fig. 3A, C), 95%CI was applied. For the two moderated mediation models as FoF as the dependent variable (Fig. 3B, D), 97.5%(1–(.05/2) = 97.5%) CI was applied. We performed linear regression with SPSS version 26 (SPSS Inc., Chicago, IL), and mediation and moderated mediation analyses using the PROCESS macro for SPSS [74].

Fig. 2

The mediation models with MAP as the independent variable, WMH volume as the mediator, and (A) the Stroop interference score, (B) the TMT-A score, or (C) the inversed FoF score as the dependent variable. The numbers on each path denote the unstandardized coefficient and confidence interval. The significant direct effects were bold-faced. To correct for multiple comparisons, 97.5%CI was applied to the models in (A) and (B), and 98.3%CI to the models in (C) unless otherwise mentioned. The results of indirect effects tested at the uncorrected level (95%CI) were included in case the significance of the results yielded under the corrected and uncorrected CIs differed. ^*The effects significant at the corrected level. ^†The effects significant at the uncorrected level (95%CI). All the effects significant at the corrected or uncorrected level are bold-faced. ATR, anterior thalamic radiation; Fmin, forceps minor; FoF, frequency of forgetting subscore of memory function questionnaire; MAP, mean arterial pressure; SLF, superior longitudinal fasciculus; TMT-A, Trail Making Test A; WMH, white matter hyperintensities.

Fig. 3

The moderated mediation models fitted to test whether the indirect effect of MAP on Stroop or FoF score via WMH is moderated by APOE ɛ4 carriership. The models in (A) and (B) test whether APOE ɛ4 carriership moderates the effect of MAP on WMH volume. The alternative models in (C) and (D) test whether APOE ɛ4 carriership moderates the effect of WMH burden on Stroop (C) or FoF (D) score. Only the results of moderation and moderated mediation effects are shown. 95%CI was applied to the models in (A) and (C), and 97.5%CI to the models in (B) and (D). ^*The effects significant at the corrected level. All significant effects are bold-faced. APOE, apolipoprotein E; ATR, anterior thalamic radiation; FoF, frequency of forgetting subscore of memory function questionnaire; MAP, mean arterial pressure; SLF, superior longitudinal fasciculus; WMH, white matter hyperintensities.

RESULTS

Sample characteristics

The characteristics of the sample are summarized in Table 1. After excluding three outliers and one participant whose MMSE score was 23, a total of 87 people were included in the analyses. The mean age was 67.57 with the SD of 8.32 (range: 50–85 years). There were more women (n = 54; 62.07%) than men (n = 33; 37.93%), χ²(1, N = 87) = 5.07, p = 0.02. Among those who did not have APOE ɛ2 allele, 39.34%(n = 24) were ɛ4 carriers and 60.66%(n = 37) were non-carriers. As expected, our sample that consisted of cognitively unimpaired individuals had a relatively low WMH burden (M = 0.57, SD = 0.31). Out of the 82 participants who had WMH data, all (100.00%) of them showed WMH on ATR and Fmin, and 70 participants (85.37%) showed WMH on SLF. 47.13%(n = 41) of the participants were taking antihypertensive medication. 45.68%(n = 37) of the participants were hypertensive as determined by either a systolic blood pressure≥140 mmHg or a diastolic blood pressure≥90. Of all participants, 69.14%(n = 56) were either hypertensive or taking antihypertensive medication.

Table 1

The sample characteristics

	N	Mean (SD) or %
Age	87	67.57 (8.32)
Sex (M/F)	87	33/54
Racial identity (Black/White)	87	35/52
Test Site (US/NL)	87	47/40
MMSE	83	28.61 (1.48)
APOE genotype
ɛ2/ɛ2, ɛ2/ɛ3, or ɛ2/ɛ4	20/81	24.69%
ɛ3/ɛ3	37/81	45.68%
ɛ3/ɛ4	23/81	28.40%
ɛ4/ɛ4	1/81	1.23%
Vascular risks
MAP	81	101.12 (12.62)
Systolic blood pressure (mm Hg)	81	138.70 (19.12)
Diastolic blood pressure (mm Hg)	81	82.32 (11.49)
Hypertension	37/81	45.68%
Antihypertensive medication	41/87	47.13%
Cognitive functioning
TMT-A	81	40.64 (16.35)
Stroop Interference	81	64.21 (29.52)
Subjective cognitive decline
MFQ-FoF	84	2.94 (0.93)
WMH volume
Whole brain (%ICV)	82	0.57 (0.31)
ATR (%the tract vol.)	82	1.31 (0.61)
Fmin (%the tract vol.)	82	0.80 (0.39)
SLF (%the tract vol.)	82	0.43 (0.41)

MMSE, Mini-Mental State Exam; APOE, apolipoprotein E; MAP, mean arterial pressure; MFQ-FoF, frequency of forgetting subscore of memory function questionnaire; TMT-A, Trail Making Test A; ICV, intra-cranial volume; ATR, anterior thalamic radiation; Fmin, forceps minor; SLF, superior longitudinal fasciculus. A lower score in TMT-A and Stroop Interference denotes a better performance.

Mediating effect of WMH for MAP and cognitive performance

In the mediation models with Stroop performance as the dependent variable (Fig. 2A), WMH volume in SLF mediated the effect of MAP on Stroop performance, unstandardized coefficient B = 0.2259, 97.5%CI [0.0043, 0.4921]. In all the other mediation models with Stroop performance or TMT-A as the dependent variable, the mediation effects of WMH burden were not significant (See Fig. 2A and 2B for the unstandardized coefficients and confidence intervals). There was no significant direct effect of MAP on cognitive function, i.e., Stroop performance or TMT-A, in any of the mediation models with cognitive function as the dependent variable (See Fig. 2A and 2B for the statistics).

Mediating effect of WMH for MAP and SCD

Next, we tested whether MAP indirectly predicted FoF, an index of SCD, via WMH burden (Fig. 2C). The effect of MAP on FoF was mediated by whole brain WMH volume, B = 0.0080, 95%CI [0.0010, 0.0138], as well as WMH burden in ATR, B = 0.0081, 95%CI [0.0011, 0.0145], at the uncorrected significance level. There was no mediating effect found for Fmin WMH volume. As with the results from the mediation analysis with cognitive function as the dependent variable, there was no significant direct effect of MAP on FoF.

APOE ɛ4 carriership as a moderator for the MAP–WMH–cognition relationship

We tested whether APOE ɛ4 carriership moderated the mediation effects of WMH burden by interacting with MAP to determine WMH volume among the models in which the mediation effect of WMH volume was significant (the models with the mediation effect significant at the uncorrected level were also included). In the model with Stroop performance as the dependent variable and WMH burden in the SLF as the mediator (Fig. 3A), APOE ɛ4 carriership moderated the effect of MAP on SLF WMH, unstandardized coefficient, B = 0.0290, 95%CI [0.0139, 0.0442], indicating that higher MAP significantly predicted more WMH burden in SLF only in APOE ɛ4 carriers, B = 0.0178, 95%CI [0.0088, 0.0268], but not in non-carriers, B = –0.0112, 95%CI [–0.0234, 0.0009]. Moreover, APOE ɛ4 carriership significantly moderated the mediating effect of SLF WMH, meaning that the mediation was only significant in APOE ɛ4 carriers, B = 0.4999, 95%CI [0.1074, 0.9060], but not in non-carriers, B = –0.3150, 95%CI [–0.8346, 0.0133] (Fig. 4). Two more moderated mediation models with FoF score as the dependent variable and WMH as the mediator were tested (Fig. 3B). In the model with whole brain WMH as the mediator, APOE ɛ4 carriership significantly moderated the effect of MAP on WMH, B = 0.0124, 97.5%CI [0.0005, 0.0244], meaning that higher MAP significantly predicted whole brain WMH volume only in APOE ɛ4 carriers, B = 0.0085, 97.5%CI [0.0014, 0.0156] but not in non-carriers, B = –0.0039, 97.5%CI [–0.0135, 0.0056]. However, APOE ɛ4 carriership did not moderate the mediation effect of whole brain WMH, B = 0.0165, 97.5%CI [–0.0072, 0.0495]. In the models with ATR WMH as the mediator, APOE ɛ4 carriership moderated neither the effect of MAP on ATR WMH volume, B = 0.0152, 97.5%CI [–0.0128, 0.0432], nor the mediation effect of ATR WMH volume, B = 0.0105, 97.5%CI [–0.0132, 0.0447]. We additionally fit alternative moderated mediation models testing whether APOE ɛ4 carriership moderated the same mediation relationships by interacting with WMH volume to determine cognitive function or FoF (Fig. 3C, D). APOE ɛ4 carriership moderated neither the link between WMH volume and cognitive function or FoF nor the mediation effect of WMH volume in any of those alternative moderated mediation models.

Fig. 4

Scatterplots showing the relationship between mean arterial pressure (MAP) and WMH volume in superior longitudinal fasciculus (SLF) in APOE ɛ4 carriers and non-carriers. The moderated mediation model in Fig. 3A showed that the WMH volume in SLF was significantly associated with MAP in APOE ɛ4 carriers, B = 0.0178, p = 0.0002, 95%CI [0.0088, 0.0268], but not in APOE ɛ4 non-carriers, B = –0.0112, p = 0.0688, 95%CI [–0.0234, 0.0009]. APOE, apolipoprotein E; MAP, mean arterial pressure; SLF, superior longitudinal fasciculus; WMH, white matter hyperintensities.

DISCUSSION

We hypothesized that WMH burden would mediate the link between MAP and objective cognitive function in cognitively unimpaired late middle-aged to older adults, and our results supported this hypothesis. We showed that WMH burden within SLF but not the whole brain mediated the relationship between MAP and Stroop performance. Notably, MAP was not directly associated with objective cognitive functioning (i.e., Stroop performance). These results suggest that arterial pressure indirectly affects cognitive function through damage in cerebral small vessels and perivascular neurons. Our study extends the aforementioned study by Wang and colleagues [30] by demonstrating that the mediation relationship also applies to more specific cognitive domains, executive function, and that executive functioning is specifically mediated by WMH burden in the SLF, rather than by global WMH burden. Our results are in line with previous studies that reported an association between lower executive functioning and a higher burden of CSVD in SLF and ATR [35, 54]. Moreover, the observed link between WMH burden in the SLF and executive function aligns with our understanding of the SLF as the major tract connecting the frontal and parietal regions of the frontoparietal network, which is implicated in executive functions, such as inhibitory control and goal-directed cognition [75]. Our findings also suggest that the significant mediation effect of whole brain WMH on the relationship between vascular risk factors and executive function reported in previous whole brain WMH studies may have been driven by the WMH burden in the white matter areas where the SLF is located.

As we hypothesized, APOE ɛ4 carriership moderated the relationship between MAP and WMH burden as well as the mediation effect of WMH burden within SLF on the relationship between MAP and Stroop performance. However, APOE ɛ4 carriership did not moderate the relationship between WMH burden and cognitive function. These results are in line with previous studies suggesting a more robust association between vascular risk factors and cognitive function in the presence of an ɛ4 allele [76]. Together with this previous finding, our results may provide one possible explanation on why the associations between vascular risk factors and cognition have been inconsistent. A stronger association between vascular risk and cognitive function may be in part due to the magnitude of mediating effect of WMH burden, which can vary by the APOE ɛ4 status, indicating that maintaining a normal range of MAP and vascular health is particularly crucial to prevent cognitive decline among healthy APOE ɛ4 carriers. The results also suggest that APOE ɛ4 contributes to worse cognitive functioning in healthy older adults by interacting with blood pressure to affect cerebrovascular health, rather than through involvement in the processes in which vascular and axonal damage affects cognitive functioning. Considering the findings that the subcortical venules of APOE ɛ4 carriers show more tortuosity compare to non-ɛ4 carriers [49], and APOE ɛ4 contributes to the breakage of BBB by the activation of the proinflammatory cyclophilin A-matrix metalloproteinase-9 pathway [47,48, 47,48], the APOE ɛ4 allele may facilitate the inflammatory effect from elevated arterial pressure to induce neurovascular dysfunction such as more tortuous subcortical venules and BBB breakdown, which are assumed to be the central mechanisms of CSVD including WMH [12, 77].

We also hypothesized that WMH burden would mediate the link between MAP and subjective cognitive function in our sample that consisted of cognitively unimpaired late middle-aged to older adults, and the results supported the hypothesis. We found that WMH burden in the ATR as well as whole brain mediated the association between MAP and SCD. Similar to the mediation results regarding objective cognitive function, SCD was not directly associated with MAP. This is a novel finding in that we showed that arterial pressure indirectly predicts SCD through WMH burden. This result suggests that higher arterial pressure indirectly increases the risk of dementia through the CSVD burden. However, we remain cautious about this interpretation because the mediation effects were significant only at the uncorrected level of 95%CI.

Our multiple linear regression and the mediation results showed that WMH in ATR as well as the whole brain predicted SCD. It is possible that the association between SCD and ATR reflects early signs of AD-related neurodegeneration. Previous diffusion imaging studies have found that ATR and several tracts in temporoparietal areas showed lower integrity in AD and amnestic MCI patients compared to cognitively unimpaired older adults, suggesting the vulnerability of ATR to AD [78, 79]. Also, ATR and other AD-vulnerable tracts project to the regions overlapping with the default mode network, which is a brain functional network showing the earliest changes in AD [80]. Thus, it may be that the white matter areas around ATR are already undergoing vascular and axonal damage in individuals with higher SCD.

In the moderated mediation models with SCD as the dependent variable, although the mediation effect of WMH burden was not moderated by APOE ɛ4 carriership, the link between MAP and whole brain WMH burden was moderated by ɛ4 carriership. This indicates that WMH burden mediates the effect of MAP on dementia risk regardless of one’s ɛ4 carriership status. Instead of modulating the mediation effect of WMH burden, the ɛ4 allele may indirectly contribute to increasing the risk of dementia by exacerbating the damaging effect of elevated MAP on cerebral small vessels (i.e., the earlier steps of the causal link of MAP–WMH–dementia risk). Considering that vascular dysfunction is often comorbid with AD, and APOE ɛ4 is a strong risk factor for AD, this result may provide one possible mechanism of how ɛ4 carriership influences the risk of AD.

This study has the following strengths. First, we measured tract-specific WMH burden as well as whole brain WMH burden and examined whether they mediate MAP and cognitive function. By doing so, we were able to reveal that subjective cognitive decline and different domains of objective cognitive function are mediated by WMH burden in distinct locations of the white matter. Another strength is the diversity in ethnicity and cultural background of our sample. More than 40%of the participants were African American. Moreover, the data were collected on two different continents, Europe and America. The samples of existing aging studies tend to be biased towards white Americans. Our findings from the diverse sample would fill the gap of knowledge on the mechanism of cognitive decline in older adults with diverse ethnic and cultural backgrounds. Despite the strengths of this study, there are several limitations that should be discussed. First, although the moderation and the moderated mediation models we tested allowed us to reveal how MAP, WMH burden, and APOE interplay to influence cognitive functioning and risk for dementia, these models were based on cross-sectional data. Longitudinal investigations on how the change in WMH burden mediates the link between vascular risk factors and conversion from normal cognition to age-related impaired cognition (i.e., MCI or AD), and how APOE ɛ4 influences the mediation relationship would help further understand the temporal relationships among vascular risk factors, WMH burden, cognitive decline, and APOE genotype. Second, this study relied on the limited sample size. To fully generalize the current study results, replicating the results using a large dataset will be needed. Lastly, we used the measure of each cognitive domain that was based on one test: processing speed was measured with TMT-A and executive function was measured with Stroop test. Further investigations using composite measures of these cognitive domains based on multiple tests are warranted to be able to fully generalize the current results.

This study has the following practical implications. It has been well recognized that high blood pressure results in worse cognitive outcome and a higher risk for dementia in older adults. Here, we revealed that WMH burden and APOE ɛ4 carriership can significantly affect the relationship between blood pressure and cognitive status. Thus, when assessing the risk of blood pressure on age-related cognitive decline or the likelihood of developing dementia, factoring in WMH burden and APOE ɛ4 carriership would allow a more accurate assessment of an individual’s risk. Moreover, considering that the mediation effect of WMH for the executive function was only significant in the ɛ4 carriers, examining other brain measures (e.g., gray matter volume) instead of WMH may lead to a more accurate evaluation of the impact of blood pressure on executive function in ɛ4 non-carriers.

In conclusion, this study provides a possible mechanism of how arterial pressure affects executive functioning and the risk for dementia in healthy middle-aged to older adults. Our results indicate that for healthy middle-aged to older adults with elevated blood pressure, specifically those with higher tract-specific WMH burden appear to have lower executive function and more SCD. Furthermore, the detrimental effect of elevated blood pressure on cognition is more prominent in APOE ɛ4 carriers. In clinical settings, considering WMH burden and APOE ɛ4 carriership may facilitate a more accurate assessment of the potential effect of high blood pressure on cognitive decline and risk for dementia.

Footnotes

ACKNOWLEDGMENTS

This work was supported in part by the Netherlands Organisation for Scientific Research [Veni grant: 016.136.072].

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Taylor

, Aizenstein

, Alexopoulos

(2013) The vascular depression hypothesis: Mechanisms linking vascular disease with depression. Mol Psychiatry 18, 963–974.

Prins

, Scheltens

(2015) White matter hyperintensities, cognitive impairment and dementia: An update. Nat Rev Neurol 11, 157–165.

Chutinet

, Rost

(2014) White matter disease as a biomarker for long-term cerebrovascular disease and dementia topical collection on cerebrovascular disease and stroke. Curr Treat Options Cardiovasc Med 16, 292.

Knopman

, Boland

, Mosley

, Howard

, Liao

, Szklo

, McGovern

, Folsom

(2001) Cardiovascular risk factors and cognitive decline in middle-aged adults. Neurology 56, 42–48.

Zhu

, Viitanen

, Guo

, Winblad

, Fratiglioni

(1998) Blood pressure reduction, cardiovascular diseases, and cognitive decline in the Mini-Mental State Examination in a community population of normal very old people: A three-year follow-up. J Clin Epidemiol 51, 385–391.

Van

Vliet P

(2012) Cholesterol and late-life cognitive decline. J Alzheimers Dis 30(Suppl 2), S147–162.

Van Den Berg

, Geerlings

, Biessels

, Nederkoorn

, Kloppenborg

(2018) White matter hyperintensities and cognition in mild cognitive impairment and Alzheimer’s disease: A domain-specific meta-analysis. J Alzheimers Dis 63, 515–527.

Kloppenborg

, Nederkoorn

, Geerlings

, Van

Den Berg E

(2014) Presence and progression of white matter hyperintensities and cognition: A meta-analysis. Neurology 82, 2127–2138.

Wallin

, Sjögren

, Edman

, Blennow

, Regland

(2000) Symptoms, vascular risk factors and blood-brain barrier function in relation to CT white-matter changes in dementia. Eur Neurol 44, 229–235.

10.

Murray

, Staff

, Shenkin

, Deary

, Starr

, Whalley

(2005) Brain white matter hyperintensities: Relative importance of vascular risk factors in nondemented elderly people. Radiology 237, 251–257.

11.

Wardlaw

, Valdés Hernández

, Muñoz-Maniega

(2015) What are white matter hyperintensities made of? Relevance to vascular cognitive impairment. J Am Heart Assoc 4, 001140.

12.

Fazekas

, Kleinert

, Offenbacher

, Schmidt

, Kleinert

, Payer

, Radner

, Lechner

(1993) Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology 43, 1683–1689.

13.

Tomimoto

, Akiguchi

, Suenaga

, Nishimura

, Wakita

, Nakamura

, Kimura

(1996) Alterations of the blood-brain barrier and glial cells in white-matter lesions in cerebrovascular and Alzheimer’s disease patients. Stroke 27, 2069–2074.

14.

Young

, Halliday

, Kril

(2008) Neuropathologic correlates of white matter hyperintensities. Neurology 71, 804–811.

15.

Muñoz Maniega

, Chappell

, Valdés Hernández

, Armitage

, Makin

, Heye

, Thrippleton

, Sakka

, Shuler

, Dennis

, Wardlaw

(2017) Integrity of normal-appearing white matter: Influence of age, visible lesion burden and hypertension in patients with small-vessel disease. J Cereb Blood Flow Metab 37, 644–656.

16.

Griffanti

, Jenkinson

, Suri

, Zsoldos

, Mahmood

, Filippini

, Sexton

, Topiwala

, Allan

, Kivimäki

, Singh-Manoux

, Ebmeier

, Mackay

, Zamboni

(2018) Classification and characterization of periventricular and deep white matter hyperintensities on MRI: A study in older adults. Neuroimage 170, 174–181.

17.

Jongen

, van der Grond

, Kappelle

, Biessels

, Viergever

, Pluim

; Utrecht Diabetic Encephalopathy Study Group (2007) Automated measurement of brain and white matter lesion volume in type 2 diabetes mellitus. Diabetologia 50, 1509–1516.

18.

Debette

, Seshadri

, Beiser

, Au

, Himali

, Palumbo

, Wolf

, DeCarli

(2011) Midlife vascular risk factor exposure accelerates structural brain aging and cognitive decline. Neurology 77, 461–468.

19.

Alber

, Alladi

, Bae

, Barton

, Beckett

, Bell

, Berman

, Biessels

, Black

, Bos

, Bowman

, Brai

, Brickman

, Callahan

, Corriveau

, Fossati

, Gottesman

, Gustafson

, Hachinski

, Hayden

, Helman

, Hughes

, Isaacs

, Jefferson

, Johnson

, Kapasi

, Kern

, Kwon

, Kukolja

, Lee

, Lockhart

, Murray

, Osborn

, Power

, Price

, Rhodius-Meester

HFM

, Rondeau

, Rosen

, Rosene

, Schneider

, Scholtzova

, Shaaban

, Silva

NCBS

, Snyder

, Swardfager

, Troen

, van Veluw

, Vemuri

, Wallin

, Wellington

, Wilcock

, Xie

, Hainsworth

(2019) White matter hyperintensities in vascular contributions to cognitive impairment and dementia (VCID): Knowledge gaps and opportunities. Alzheimers Dement (N Y) 5, 107–117.

20.

Lampe

, Zhang

, Beyer

, Huhn

, Kharabian Masouleh

, Preusser

, Bazin

, Schroeter

, Villringer

, Witte

(2019) Visceral obesity relates to deep white matter hyperintensities via inflammation. Ann Neurol 85, 194–203.

21.

Guo

, Pantoni

, Simoni

, Bengtsson

, Björkelund

, Lissner

, Gustafson

, Skoog

(2009) Blood pressure components and changes in relation to white matter lesions. Hypertension 54, 57–62.

22.

Marcus

, Gardener

, Rundek

, Elkind

MSV

, Sacco

, Decarli

, Wright

(2011) Baseline and longitudinal increases in diastolic blood pressure are associated with greater white matter hyperintensity volume: The Northern Manhattan Study. Stroke 42, 2639–2641.

23.

Tsao

, Himali

, Beiser

, Larson

, Decarli

, Vasan

, Mitchell

, Seshadri

(2016) Association of arterial stiffness with progression of subclinical brain and cognitive disease. Neurology 86, 619–626.

24.

Dufouil

, de Kersaint-Gilly

, Besancon

, Levy

, Auffray

, Brunnereau

, Alperovitch

, Tzourio

(2001) Longitudinal study of blood pressure and white matter hyperintensities: The EVA MRI Cohort. Neurology 56, 921–926.

25.

Jacobs

HIL

, Visser

, Van Boxtel

MPJ

, Frisoni

, Tsolaki

, Papapostolou

, Nobili

, Wahlund

, Minthon

, Frölich

, Hampel

, Soininen

, van de Pol

, Scheltens

, Tan

FES

, Jolles

, Verhey

FRJ

(2012) The association between white matter hyperintensities and executive decline in mild cognitive impairment is network dependent. Neurobiol Aging 33, 201.e1–8.

26.

Hirsiger

, Koppelmans

, Mérillat

, Erdin

, Narkhede

, Brickman

, Jäncke

(2017) Executive functions in healthy older adults are differentially related to macro- and microstructural white matter characteristics of the cerebral lobes. Front Aging Neurosci 9, 1–14.

27.

Van Den Heuvel

DMJ

, Ten Dam

, De Craen

AJM

, Admiraal-Behloul

, Olofsen

, Bollen

ELEM

, Jolles

, Murray

, Blauw

, Westendorp

RGJ

, Van Buchem

(2006) Increase in periventricular white matter hyperintensities parallels decline in mental processing speed in a non-demented elderly population. J Neurol Neurosurg Psychiatry 77, 149–153.

28.

Debette

, Markus

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

29.

Brickman

(2013) Contemplating Alzheimer’s disease and the contribution of white matter hyperintensities. Curr Neurol Neurosci Rep 13, 415.

30.

Wang

, Fratiglioni

, Kalpouzos

, Lövdén

, Laukka

, Bronge

, Wahlund

, Bäckman

, Qiu

(2017) Mixed brain lesions mediate the association between cardiovascular risk burden and cognitive decline in old age: A population-based study. Alzheimers Dement 13, 247–256.

31.

Biesbroek

, Kuijf

, Van der Graaf

, Vincken

, Postma

, Mali

WPTM

, Biessels

, Geerlings

(2013) Association between subcortical vascular lesion location and cognition: A voxel-based and tract-based lesion-symptom mapping study. The SMART-MR Study. PLoS One 8, e60541.

32.

Biesbroek

, Weaver

, Hilal

, Kuijf

, Ikram

, Xu

, Tan

, Venketasubramanian

, Postma

, Biessels

, Chen

CPLH

(2016) Impact of strategically located white matter hyperintensities on cognition in memory clinic patients with small vessel disease. PLoS One 11, e0166261.

33.

Duering

, Gesierich

, Seiler

, Pirpamer

, Gonik

, Hofer

, Jouvent

, Duchesnay

, Chabriat

, Ropele

, Schmidt

, Dichgans

(2014) Strategic white matter tracts for processing speed deficits in age-related small vessel disease. Neurology 82, 1946–1950.

34.

Duering

, Zieren

, Hervé

, Jouvent

, Reyes

, Peters

, Pachai

, Opherk

, Chabriat

, Dichgans

(2011) Strategic role of frontal white matter tracts in vascular cognitive impairment: A voxel-based lesion-symptom mapping study in CADASIL. Brain 134, 2366–2375.

35.

Jiang

, Paradise

, Liu

, Armstrong

, Zhu

, Kochan

, Brodaty

, Sachdev

, Wen

(2018) The association of regional white matter lesions with cognition in a community-based cohort of older individuals. Neuroimage Clin 19, 14–21.

36.

Attems

, Jellinger

(2014) The overlap between vascular disease and Alzheimer’s disease - lessons from pathology. BMC Med 12, 206.

37.

Kloppenborg

, Van den Berg

, Kappelle

, Biessels

(2008) Diabetes and other vascular risk factors for dementia: Which factor matters most? A systematic review. Eur J Pharmacol 585, 97–108.

38.

Buckley

, Hanseeuw

, Schultz

, Vannini

, Aghjayan

, Properzi

, Jackson

, Mormino

, Rentz

, Sperling

, Johnson

, Amariglio

(2017) Region-specific association of subjective cognitive decline with tauopathy independent of global β-amyloid burden. JAMA Neurol 74, 1455–1463.

39.

Amariglio

, Buckley

, Mormino

, Marshall

, Johnson

, Rentz

, Sperling

(2018) Amyloid-associated increases in longitudinal report of subjective cognitive complaints. Alzheimers Dement (N Y) 4, 444–449.

40.

Dufouil

, Fuhrer

, Alpérovitch

(2005) Subjective cognitive complaints and cognitive decline: Consequence or predictor? The epidemiology of vascular aging study. J Am Geriatr Soc 53, 616–621.

41.

Reisberg

, Shulman

, Torossian

, Leng

, Zhu

(2010) Outcome over seven years of healthy adults with and without subjective cognitive impairment. Alzheimers Dement 6, 11–24.

42.

Jessen

, Wiese

, Bachmann

, Eifflaender-Gorfer

, Haller

, Kölsch

, Luck

, Mösch

, van den Bussche

, Wagner

, others (2010) Prediction of dementia by subjective memory impairment: Effects of severity and temporal association with cognitive impairment. Arch Gen Psychiatry 67, 414–422.

43.

Van Rooden

, Van Den Berg-Huysmans

, Croll

, Labadie

, Hayes

, Viviano

, Van Der Grond

, Rombouts

SARB

, Damoiseaux

(2018) Subjective cognitive decline is associated with greater white matter hyperintensity volume. J Alzheimers Dis 66, 1283–1294.

44.

Verghese

, Castellano

, Holtzman

(2011) Apolipoprotein E in Alzheimer’s disease and other neurological disorders. Lancet Neurol 10, 241–252.

45.

Brickman

, Schupf

, Manly

, Stern

, Luchsinger

, Provenzano

, Narkhede

, Razlighi

, Collins-Praino

, Artero

, Akbaraly

, Ritchie

, Mayeux

, Portet

(2014) APOE ɛ4 and risk for Alzheimer’s disease: Do regionally distributed white matter hyperintensities play a role? Alzheimers Dement 10, 619–629.

46.

Schilling

, DeStefano

, Sachdev

, Choi

, Mather

, DeCarli

, Wen

, Høgh

, Raz

, Au

, Beiser

, Wolf

, Romero

, Zhu

, Lunetta

, Farrer

, Dufouil

, Kuller

, Mazoyer

, Seshadri

, Tzourio

, Debette

(2013) APOE genotype and MRI markers of cerebrovascular disease. Neurology 81, 292–300.

47.

Bell

, Winkler

, Singh

, Sagare

, Deane

, Wu

, Holtzman

, Betsholtz

, Armulik

, Sallstrom

, Berk

, Zlokovic

, Apo

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

48.

Montagne

, Nation

, Sagare

, Barisano

, Sweeney

, Chakhoyan

, Pachicano

, Joe

, Nelson

, D’Orazio

, Buennagel

, Harrington

, Benzinger

TLS

, Fagan

, Ringman

, Schneider

, Morris

, Reiman

, Caselli

, Chui

, Tcw

, Chen

, Pa

, Conti

, Law

, Toga

, Zlokovic

(2020) APOE4 leads to blood-brain barrier dysfunction predicting cognitive decline. Nature 581, 71–76.

49.

Shaaban

, Aizenstein

, Jorgensen

, MacCloud

, Meckes

, Erickson

, Glynn

, Mettenburg

, Guralnik

, Newman

, Ibrahim

, Laurienti

, Vallejo

, Rosano

(2017) In vivo imaging of venous side cerebral small-vessel disease in older adults: An MRI method at 7T. Am J Neuroradiol 38, 1923–1928.

50.

De Leeuw

, Richard

, De Groot

, Van Duijn

, Hofman

, Van Gijn

, Breteler

MMB

(2004) Interaction between hypertension, apoE, and cerebral white matter lesions. Stroke 35, 1057–1060.

51.

Livny

, Ravona-Springer

, Heymann

, Priess

, Kushnir

, Tsarfaty

, Rabinov

, Moran

, Hoffman

, Cooper

, Greenbaum

, Silverman

, Sano

, Johnson

, Bendlin

, Beeri

(2016) Long-term variability in glycemic control is associated with white matter hyperintensities in APOE4 genotype carriers with type 2 diabetes. Diabetes Care 39, 1056–1059.

52.

Safar

, Boudier

(2005) Vascular development, pulse pressure, and the mechanisms of hypertension. In Hypertension, pp. 205-209.

53.

Mayet

, Hughes

(2003) Pathophysiology in hypertension. Hear J 89, 1104–1109.

54.

Biesbroek

, Weaver

, Biessels

(2017) Lesion location and cognitive impact of cerebral small vessel disease. Clin Sci 131, 715–728.

55.

Tombaugh

, Mclntyre

(1992) The Mini-Mental State Examination: A comprehensive review. Prog Geriatr 40, 922–935.

56.

Loewenstein

, Acevedo

, Ownby

, Agron

, Barker

, Isaacson

, Strauman

, Duara

(2006) Using different memory cutoffs to assess mild cognitive impairment. Am J Geriatr Psychiatry 14, 911–919.

57.

Brooks

, Iverson

, Holdnack

, Feldman

(2008) Potential for misclassification of mild cognitive impairment: A study of memory scores on the Wechsler Memory Scale-III in healthy older adults. J Int Neuropsychol Soc 14, 463–478.

58.

Mistridis

, Egli

, Iverson

, Berres

, Willmes

, Welsh-Bohmer

, Monsch

(2015) Considering the base rates of low performance in cognitively healthy older adults improves the accuracy to identify neurocognitive impairment with the Consortium to Establish a Registry for Alzheimer’s Disease-Neuropsychological Assessment Battery (CERAD). Eur Arch Psychiatry Clin Neurosci 265, 407–417.

59.

Werry

, Daniel

, Bergström

(2019) Group differences in normal neuropsychological test performance for older non-Hispanic White and Black/African American adults. Neuropsychology 33, 1089–1100.

60.

Manly

, Jacobs

, Sano

, Bell

, Merchant

, Small

, Stern

(1998) Cognitive test performance among nondemented elderly African Americans and whites. Neurology 50, 1238–1245.

61.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”: A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

62.

Wechsler

(2008) Wechsler memory scale–fourth edition (WMS-IV) technical and interpretive manual, Pearson, San Antonio, TX.

63.

Corder

, Saunders

, Risch

, Strittmatter

, Schmechel

, Gaskell

, Rimmler

, Locke

, Conneally

, Schmader

, Small

, Roses

, Haines

, Pericak-Vance

(1994) Protective effect of apolipoprotein E type 2 allele for late onset Alzheimer disease. Nat Genet 7, 180–184.

64.

Bennet

, Di Angelantonio

, Ye

, Wensley

, Dahlin

, Ahlbom

, Keavney

, Collins

, Wiman

, De

Faire U

, Danesh

(2007) Association of apolipoprotein e genotypes with lipid levels and coronary risk. J Am Med Assoc 298, 1300–1311.

65.

Reitan

(1958) Validity of the Trail Making Test as an indicator of organic brain damage. Percept Mot Skills 8, 271–276.

66.

Stroop

(1935) Studies of interference in serial verbal reactions. J Exp Psychol 18, 643.

67.

Valentijn

SAM

, Van Boxtel

MPJ

, Van Hooren

SAH

, Bosma

, Beckers

HJM

, Ponds

RWHM

, Jolles

(2005) Change in sensory functioning predicts change in cognitive functioning: Results from a 6-year follow-up in the Maastricht Aging Study. J Am Geriatr Soc 53, 374–380.

68.

Gilewski

, Zelinski

, Schaie

(1990) The Memory Functioning Questionnaire for assessment of memory complaints in adulthood and old age. Psychol Aging 5, 482–490.

69.

Jenkinson

, Pechaud

, Smith

(2005) BET2: MR-based estimation of brain, skull and scalp surfaces. In Eleventh annual meeting of the organization for human brain mapping, pp. 167.

70.

Schmidt

(2017) Bayesian inference for structured additive regression models for large-scale problems with applications to medical imaging, Doctoral dissertation. Ludwig-Maximilians-Universität München, Germany.

71.

Griffanti

, Zamboni

, Khan

, Li

, Bonifacio

, Sundaresan

, Schulz

, Kuker

, Battaglini

, Rothwell

, Jenkinson

(2016) BIANCA (Brain Intensity AbNormality Classification Algorithm): A new tool for automated segmentation of white matter hyperintensities. Neuroimage 141, 191–205.

72.

Hua

, Zhang

, Wakana

, Jiang

, Li

, Reich

, Calabresi

, Pekar

, van Zijl

PCM

, Mori

(2008) Tract probability maps in stereotaxic spaces: Analyses of white matter anatomy and tract-specific quantification. Neuroimage 39, 336–347.

73.

Tabachnick

, Fidell

(2013) Using Multivariate Statistics, Pearson, Boston, MA, pp. 73.

74.

Hayes

(2018) Introduction to mediation, moderation, and conditional process analysis: A regression-based approach, Guilford publications, New York, NY.

75.

Campbell

, Grady

, Ng

, Hasher

(2012) Age differences in the frontoparietal cognitive control network: Implications for distractibility. Neuropsychologia 50, 2212–2223.

76.

Bangen

, Beiser

, Delano-Wood

, Nation

, Lamar

, Libon

, Bondi

, Seshadri

, Wolf

, Au

(2013) APOE genotype modifies the relationship between midlife vascular risk factors and later cognitive decline. J Stroke Cerebrovasc Dis 22, 1361–1369.

77.

Pantoni

, Garcia

(1997) Pathogenesis of leukoaraiosis: A review. Stroke 28, 652–659.

78.

Serra

, Cercignani

, Petrosini

, Basile

, Perri

, Fadda

, Spanò

, Marra

, Giubilei

, Carlesimo

, Caltagirone

, Bozzali

(2011) Neuroanatomical correlates of cognitive reserve in Alzheimer disease. Rejuvenation Res 14, 143–151.

79.

Shao

, Li

, Zhang

, Yang

, Tao

, Zhang

, Peng

(2019) White matter integrity disruption in the pre-dementia stages of Alzheimer’s disease: From subjective memory impairment to amnestic mild cognitive impairment. Eur J Neurol 26, 800–807.

80.

Sorg

, Riedl

, Perneczky

, Kurz

, Wohlschlager

(2009) Impact of Alzheimer’s disease on the functional connectivity of spontaneous brain activity. Curr Alzheimer Res 6, 541–553.