Brain White Matter Structure and Amyloid Deposition in Black and White Older Adults: The ARIC-PET Study.

Background White matter abnormalities are a common feature of aging and Alzheimer disease, and tend to be more severe among Black individuals. However, the extent to which white matter abnormalities relate to amyloid deposition, a marker of Alzheimer pathology, remains unclear. This cross-sectional study examined the association of white matter abnormalities with cortical amyloid in a community sample of older adults without dementia and examined the moderating effect of race. Methods and Results Participants from the ARIC-PET (Atherosclerosis Risk in Communities-Positron Emission Tomography) study underwent brain magnetic resonance imaging, which quantified white matter hyperintensity volume and microstructural integrity using diffusion tensor imaging. Participants received florbetapir positron emission tomography imaging to measure brain amyloid. Associations between measures of white matter structure and elevated amyloid status were examined using multivariable logistic regression. Among 322 participants (43% Black), each SD increase in white matter hyperintensity volume was associated with a greater odds of elevated amyloid (odds ratio [OR], 1.37; 95% CI, 1.03-1.83) after adjusting for demographic and cardiovascular risk factors. In race-stratified analyses, a greater white matter hyperintensity volume was more strongly associated with elevated amyloid among Black participants (OR, 2.00; 95% CI, 1.15-3.50), compared with White participants (OR, 1.29; 95% CI, 0.89-1.89). However, the race interaction was not statistically significant (P interaction=0.09). We found no association between white matter microstructure and elevated amyloid. Conclusions The results suggest a modest positive relationship between white matter hyperintensity and elevated amyloid in older adults without dementia. Although the results indicate that this association is nonsignificantly stronger among Black participants, these findings will need to be confirmed or refuted using larger multiracial cohorts.

White matter dysfunction is a pervasive feature of Alzheimer disease (AD) that has been consistently associated with cognitive decline and symptomatic progression. Magnetic resonance imaging (MRI)–defined white matter hyperintensities (WMHs) and white matter microstructural abnormalities often emerge one or more decades before the onset of clinically defined dementia, even in relatively younger individuals with autosomal dominant AD and minimal vascular disease. , Cerebral small vessel disease is believed to be a primary cause of white matter abnormalities, and there is accumulating support for a bidirectional relationship between cerebral small vessel disease and AD pathology, including amyloid‐ß.

A 2017 systematic review examining the relationship between WMHs and cortical amyloid concluded that the association between positron emission tomography (PET)–defined amyloid and WMH volume was not consistently supported. However, several recent studies with larger sample sizes (compared with most previous studies) have found a modest positive association between WMH volume and cortical amyloid levels in individuals without dementia. , While these findings hint at a connection between cerebral small vessel disease and cortical amyloid in the preclinical and/or prodromal phase of AD, it remains unknown whether these results translate to non‐White/European populations with differing vascular risk profiles, lifetime environmental exposures, and AD genetic architecture. Thus, there is a clear need to understand the relationship between white matter abnormalities and amyloid accumulation, particularly among Black older adults given that: (1) the prevalence of cerebrovascular disease, vascular risk factors, and dementia is higher in this group, and (2) white matter abnormalities tend to be more severe in Black, compared with White, older adults. To begin to understand whether white matter structural abnormalities and amyloid co‐occur similarly among these race groups, the current study examined the association of WMH volume and WM microstructural integrity with cortical amyloid in a community sample of Black and White older adults without dementia in the ARIC (Atherosclerosis Risk in Communities) study.

METHODS

The data, analytic methods, and study materials will be made available to other researchers for purposes of reproducing the results or replicating the procedure in accordance with ARIC study policies. Data from the ARIC study can be accessed, with appropriate approvals, through the National Heart, Lung, and Blood Institute's Biospecimen and Data Repository Information Coordinating Center (https://biolincc.nhlbi.nih.gov/home/) or by contacting the ARIC Coordinating Center.

Study Design and Participants

The ARIC study is a community‐based cohort study that enrolled 15 792 participants from 4 US communities upon its initiation (1987–1989). Of the 6538 participants who attended ARIC visit 5 (2011–2013), 1978 were selected to undergo brain MRI (Data S1). We used available data from 346 of participants with MRI scans who underwent florbetapir PET imaging as part of the ARIC‐PET study. Study inclusion/exclusion criteria are outlined in Figure 1A. ARIC study protocols were approved by the institutional review boards at each participating center. All participants gave written informed consent at each study visit.

Figure 1

Study inclusion and exclusion criteria and white matter hyperintensity (WMH) volume by race.

ARIC indicates Atherosclerosis Risk in Communities; MRI, magnetic resonance imaging; and PET, positron emission tomography. Figure created with BioRender.com.

Brain MRI and PET Imaging

Brain MRI scans were conducted using a 3T scanner. Images were analyzed at the ARIC MRI Reading Center (Mayo Clinic). Magnetization‐prepared rapid acquisition gradient echo (MP‐RAGE), axial T2*gradient echo, axial T2 fluid‐attenuated inversion recovery (FLAIR), and axial diffusion tensor imaging (DTI) sequences were obtained from all participants. WMH volumes were derived from FLAIR images using a quantitative computer‐aided segmentation program to measure the volumetric burden of leukoaraiosis, defined as increased signal intensity within white matter. The computer‐aided segmentation program is an update of the in‐house semiautomated method previously described by Raz et al. WMHs were segmented on native 2‐dimensional FLAIR images using an automated seed initialization based on location (spatial priors), intensity relative to the distribution of grey matter intensity values, and the intensity relative to its local neighborhood. To reduce the number of false‐positive segmentations of WMHs from FLAIR images, the MP‐RAGE image was resampled in FLAIR space and the MP‐RAGE segmentation was used to generate a white matter mask. Using this method, a continuous measure of WMH volume was derived. T2*gradient echo scans were used to identify lobar and subcortical cerebral microbleeds (CMBs). DTI measures of fractional anisotropy (FA) and mean diffusivity (MD) were used to assess white matter microstructural integrity. Lower FA and higher MD are accepted as measures of reduced white matter integrity. Our primary analysis examined composite FA and MD measures derived using 4 white matter tracts known to be affected early in AD: the cingulate gyrus cingulum, hippocampal cingulum, superior longitudinal fasciculus, and splenium of the corpus callosum. , , Using these tracts, we derived a general factor for FA and MD from the first unrotated principal component of the standardized FA and MD values (Table S1).

Florbetapir PET neuroimaging was used to identify participants with elevated cortical amyloid. PET scans were conducted within 1 year of brain MRIs and were coregistered with MRI MP‐RAGE sequences at 3 ARIC‐PET sites. We used a 20‐minute (4×5 minute) uptake scan that was obtained starting 50 minutes after intravenous injection of the florbetapir isotope. The Wong laboratory (D.F.W. and Y.Z.) at Johns Hopkins University reviewed images for quality and calculated standardized uptake value ratios using a cerebellar gray matter reference region. The current analyses used a global measure of cortical florbetapir uptake defined as the volume‐dependent weighted averages of 9 regions. Elevated cortical amyloid (standardized uptake value ratio >1.2) was defined a priori based on the ARIC‐PET sample median, consistent with previous ARIC‐PET studies. See Data S1 for further description of the PET protocol.

Covariate and Clinical Assessment

Age at index visit (visit 5) and participant demographic data (race [Black/White], education, sex) reported by participants at ARIC visit 1 were used as covariates. The TaqMan assay (Applied Biosystems) was used to define APOE genotype (0 versus ≥1 APOEε4 alleles). Annual combined family income was assessed at visit 4 (1996–1998) based on self‐report. All other covariates were defined at visit 5. Body mass index was defined based on participant height and measured weight. Hypertension was defined based on measured systolic and diastolic blood pressure >140/90 mm/Hg or use of antihypertensive medication. Diabetes mellitus was defined based on self‐report of physician diagnosis, diabetes mellitus medication use, or glycated hemoglobin level ≥6.5%. History of coronary heart disease was defined by self‐report at visit 1 and adjudicated between visits 1 and 5. Current smoking status was defined based on self‐report. Participants' cognitive status (normal/mild cognitive impairment) was defined based on National Institute on Aging (NIA) and the Alzheimer's Association (AA)/Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM‐5) criteria (Data S1). Participants who met criteria for dementia were not included in the ARIC‐PET study.

Statistical Analysis

We used multivariable logistic regression to examine the association of WMH volume and DTI FA/MD (independent variables) with elevated cortical amyloid (dependent variable). We examined 2 models. The first model included demographic risk factors (age, study center, race, sex, education, and APOE genotype). The second model additionally adjusted for cardiovascular risk factors (CRFs), ie, body mass index, diabetes mellitus, hypertension, coronary heart disease, and current smoking status. Based on recent findings suggesting that CMBs may account for the relationship between WMHs and amyloid, we repeated the primary analyses after excluding participants with any/lobar CMBs. To examine the modifying effect of race on the white matter‐amyloid relationship, we also examined race‐by‐WMH and race‐by‐FA/MD interaction terms and conducted stratified analyses. WMH volume and DTI FA/MD were modeled as continuous variables and divided into quartiles to examine nonlinear associations. WMH quartiles were defined after WMH volume was normalized for intracranial volume. Analyses examining WMH volume as a continuous variable were adjusted for intracranial volume. Additionally, WMH volumes were log‐transformed because of skewness. A 2‐sided P value <0.05 was used to designate statistical significance. Analyses were conducted using Stata version 14 (StataCorp).

RESULTS

A total of 338 participants were included in the final analytic sample (age 76 years [SD, 5 years]; 43% [N=144] were of Black race and 57% [N=191] were women). A total of 51% (N=173) of the sample was amyloid‐positive; 64% (N=92) of Black participants and 42% (N=81) of White participants were amyloid‐positive (Figure S1). Sixteen participants were missing ≥1 CRF covariates and were not included in the CRF‐adjusted model. Full sample characteristics are displayed in Table S2. WMH volume did not significantly differ between Black and White participants (Mann‐Whitney U test, z=1.50; P=0.13), although there were more Black participants at the highest end of the distribution (Figure 1B).

In a model adjusted for demographic characteristics, participants with greater WMH volume had a higher odds of elevated cortical amyloid (OR, 1.36 per SD increase WMH volume; 95% CI, 1.03–1.79; Table). The results were similar after adjustment for CRFs, after excluding participants with ≥1 CMB or lobar CMB (Table S3), and after additionally adjusting for the presence of cerebral infarcts (Table S4). In race‐stratified analyses, each SD increase in WMH volume doubled the odds of elevated cortical amyloid among Black participants (Table). The association between WMH volume and amyloid did not extend to White participants, however. A formal assessment of effect modification by race did not yield statistical significance (P interaction=0.09). Overall, the results were similar when WMH volume was examined by quartile (Table; Figure S2). Use of race‐specific WMH quartiles did not change these results (Figure 2A and 2B; Table S5). Results were similar using a common alternative threshold for elevated amyloid (standardized uptake value ratio >1.11; Table S6).

Table 1.

Association of WMH Volume With Elevated Cortical Amyloid

WMH volume*	Model 0 N=338	P value	Model 1 N=338	P value	Model 2 N=322	P value	Model 3 N=309	P value
	Elevated amyloid OR (95% CI)		Elevated amyloid OR (95% CI)		Elevated amyloid OR (95% CI)		Elevated amyloid OR (95% CI)
Quartile 1, 0.5–7.5, cm3 (reference)	1 (reference)	…	1 (reference)	…	1 (reference)	…	1 (reference)	…
Quartile 2, 6.1–13.0, cm3	1.57 (0.86–2.89)	0.15	1.37 (0.69–2.71)	0.37	1.34 (0.65–2.75)	0.42	1.47 (0.70–3.09)	0.31
Quartile 3, 9.5–21.8, cm3	1.40 (0.76–2.57)	0.28	1.08 (0.54–2.16)	0.82	1.05 (0.50–2.18)	0.90	1.07 (0.51–2.27)	0.86
Quartile 4, 15.8–133.5, cm3	2.84 (1.52–5.31)	0.001	2.16 (1.04–4.47)	0.04	2.19 (1.03–4.69)	0.04	2.45 (1.12–5.39)	0.03
WMH (log), Per 1 SD (continuous)	1.51 (1.19–1.90)	0.001	1.36 (1.03–1.79)	0.03	1.37 (1.03–1.83)	0.03	1.40 (1.04–1.88)	0.03

Model 0 is unadjusted (model includes only intracranial volume). Model 1 is adjusted for intracranial volume, age, center, race, sex, education, and APOE ε4 status. Model 2 is additionally adjusted for late‐life (visit 5) body mass index, diabetes mellitus, hypertension, coronary heart disease, and current smoking status. Sixteen participants included in model 1 were excluded from model 2 for missing ≥1 model 2 covariates. Model 3 is additionally adjusted for total combined annual family income. Thirteen participants included in model 2 were excluded from model 3 for missing household family income data. P values for the white matter hyperintensity (WMH) volume by race interaction term derived from model 2 were 0.13 for the quartiled analysis and 0.09 for the continuous analysis. OR indicates odds ratio.

WMH volumes were quartiled after normalization for intracranial volume. As a result, there is some overlap among quartiles for the provided non‐normalized WMH volumes.

Figure 2

Race‐ and brain region–specific associations between white matter hyperintensity (WMH) volume and elevated cortical amyloid.

All models were adjusted for intracranial volume, age, center, race, sex, education, APOE ε4 status, and late‐life (visit 5) body mass index, diabetes mellitus, hypertension, coronary heart disease, and current smoking status (ie, model 2). A, B, The adjusted probability and standard error of elevated cortical amyloid for each quartile of WMH volume, calculated using logistic regression. Race‐specific WMH quartiles were used for analyses. C, The adjusted odds ratio (OR) of regional elevated cortical amyloid per SD increase in WMH volume, calculated using logistic regression. The adjusted OR for Black participants was imprecise: OR, 25.2 (95% CI, 1.9–339.8); coronary heart disease and smoking status covariates were excluded because they predicted the outcome perfectly.*P<0.05; **P<0.01.

Black (Model 0)

N=144

White (Model 0)

N=194

Black (Model 2)

N=139

White (Model 2)

N=183

Quartile 1, 0.5–7.5, cm3

(reference)

In post hoc sensitivity analyses that examined the effect of adjusting for midlife vascular risk factors (which have been previously linked to late‐life WMH and amyloid levels), the association between WMH and elevated amyloid persisted in the full sample (Table S7). However, the magnitude of the association was attenuated among Black participants, suggesting that midlife vascular risk factors account, in part, for the WMH‐amyloid relationship in this group. Adding combined annual family income (a proxy of socioeconomic status) to the primary model did not meaningfully change the results (Table; Table S8). The relationship between greater WMH volume and elevated amyloid among Black participants was maintained when 7 participants with outlier WMH volumes (>3 SD above the total sample mean) were excluded (odds ratio [OR], 1.94 [95% CI, 1.06–3.55] P=0.03; N=132) and when only cognitively normal Black participants were examined (OR, 2.07 [95% CI, 1.11–3.86] P=0.02; N=102). An examination of region‐specific amyloid found the WMH volume‐amyloid association was strongest in the precuneus region (Figure 2C). Results were similar among race groups.

General and tract‐specific DTI measures of white matter microstructural integrity were unrelated to elevated cortical amyloid in demographically adjusted and CRF‐adjusted models, and in race‐stratified analyses (Tables S9–S12).

DISCUSSION

The current study provides evidence for a positive relationship between brain WMH volume and cortical amyloid in older adults without dementia. This association occurred independent of CMBs, a marker of cerebral amyloid angiopathy, and tended to be stronger among Black participants, although not significantly. In contrast to WMH volume, general and tract‐specific measures of DTI white matter microstructure were unrelated to amyloid status.

Previous studies of the relationship between WMH volume and cortical amyloid have been conducted in European, US White, or Korean populations and have yielded mixed results. Among studies showing positive results, modest effect sizes were typically observed, comparable with that found in the current analysis of White participants. By comparison, our analysis revealed a strong association between WMH volume and cortical amyloid in Black participants, even among the subset of Black participants without cognitive impairment. This finding is of particular relevance to understanding of the role that race, perhaps a proxy for lifelong social experiences and physiological and environmental exposures, may play in the development of AD. Black adults are disproportionately affected by Alzheimer dementia compared with their age‐adjusted White counterparts. Although the causes of this disparity are not yet fully understood, there likely exist a multitude of factors, social, environmental, and possibly biologic.

While causal inferences cannot be derived from this cross‐sectional analysis, the findings do suggest a connection between cerebrovascular processes underlying WMHs and amyloid accumulation that is independent of cardiovascular risk factors (late‐life and midlife), cerebral infarcts, and cerebral amyloid angiopathy. If there is a causal relationship between cerebrovascular dysfunction and amyloid accumulation, as has been suggested elsewhere, the strong association between WMHs and amyloid deposition in Black older adults may be relevant to understanding the pattern of increased dementia risk in this population. Regarding predictors of amyloid status, we found that demographic factors accounted for the bulk of the variance in amyloid status in the full sample, and in race‐stratified analyses. In contrast, prevalent (late‐life) CRFs did not account for much additional variance in amyloid status and do not appear to account for the association between WMH volume and amyloid. However, the presence of midlife vascular risk factors did account in part for the stronger association between WMH volume and elevated amyloid among Black participants, suggesting that distal health factors, rather than race itself, may contribute to group differences in the magnitude of the WMH‐amyloid association. Similar to distal health factors, social determinants of health (eg, socioeconomic status/position, stress/discrimination, access to health care, and neighborhood and environmental exposures), which tend to differ between Black and White individuals living within the United States, may account for the stronger link between WMH and amyloid observed in Black participants. Although the current study attempted to account for socioeconomic status using a crude proxy measure (family income level), future work that more carefully accounts for social health determinants will be needed to identify drivers of disparities in AD.

The lack of association between white matter microstructural integrity and amyloid‐positive status also deserves some consideration. While studies examining the association of PET‐ and cerebrospinal fluid‐defined amyloid levels with DTI measures of FA and MD have yielded mixed results, a more consistent association has emerged between DTI white matter microstructure and tau pathology, particularly for white matter tracts in areas that experience early tau pathology. If the associations between amyloid levels and white matter microstructural integrity are indeed driven by tau pathology, it is possible that the amyloid‐positive participants in the present study are not advanced enough in their disease, ie, do not have enough tau neurofibrillary tangle pathology, to show amyloid‐DTI associations.

The current results should be interpreted within the context of several limitations. First, the study is limited by its cross‐sectional design. Understanding the time course of the amyloid‐WMH relationship with respect to the clinical manifestations of symptoms will be essential in advancing the understanding of AD pathophysiology. Second, because the majority of Black participants who underwent amyloid PET neuroimaging were from a single study site (Jackson, Mississippi), it is possible that the demonstrated race‐specific findings are attributable to a nondescribed phenomenon specific to the geographic region. Thus, additional multiethnic community‐based studies in other geographic regions are required to confirm the generalizability of our findings. Third, the lack of region‐specific WMH information limited the ability to look at how WMH occurring in specific brain regions related to amyloid levels. Finally, selection into the ARIC‐PET substudy, which excluded participants with dementia and participants with MRI and PET contraindications, may have introduced a positive selection bias for those participants who have generally better cardiovascular health compared with the population in totum. This selection may also affect the generalizability of the findings.

Within the context of these limitations, the current study provides support for a relationship between WMHs and elevated cortical amyloid in older adults without dementia. This relationship was particularly strong among Black participants. However, there was no relationship between DTI‐defined white matter microstructural integrity and cortical amyloid. These findings will need to be confirmed or refuted using larger multiracial cohorts.

Sources of Funding

The ARIC study is performed as a collaborative study supported by National Heart, Lung, and Blood Institute (NHLBI) contracts (HHSN268201700001I, HHSN268201700002I, HHSN268201700003I, HHSN268201700005I, HHSN268201700004I). Neurocognitive data are collected by U01 2U01HL096812, 2U01HL096814, 2U01HL096899, 2U01HL096902, 2U01HL096917 from the National Institutes of Health (NIH) (NHLBI, National Institute of Neurological Disorders and Stroke [NINDS], National Institute on Aging [NIA], and National Institute on Deafness and Other Communication Disorders [NIDCD]), and with previous brain MRI examinations funded by R01‐HL70825 from the NHLBI. The ARIC‐PET study is funded by the NIA (R01AG040282). Infrastructure was partly supported by grant number UL1RR025005, a component of the NIH and NIH Roadmap for Medical Research. This study was also supported by contracts K23 AG064122 (Dr Walker) and K24 AG052573 (Dr Gottesman) from the NIA. This research was supported in part by the Intramural Research Program of the NIH/NIA. Avid Radiopharmaceuticals provided the florbetapir isotope for the study but had no role in the study design or interpretation of results.

Disclosures

Keenan A. Walker receives research funding from the NIH/NIA Intramural Research Program. Clifford R. Jack Jr serves on an independent data monitoring board for Roche, has served as a speaker for Eisai, and consulted for Biogen, but he receives no personal compensation from any commercial entity. He receives research support from the NIH and the Alexander Family Alzheimer's Disease Research Professorship of the Mayo Clinic. David S. Knopman served on a Data Safety Monitoring Board for the DIAN (Dominantly Inherited Alzheimer Network) study. He serves on a data safety monitoring board for a tau therapeutic for Biogen, but receives no personal compensation. He is a site investigator in the Biogen aducanumab trials. He is an investigator in a clinical trial sponsored by Lilly Pharmaceuticals and the University of Southern California. He serves as a consultant for Samus Therapeutics, Third Rock, Roche, and Alzeca Biosciences but receives no personal compensation. He receives research support from the NIH. Dean F. Wong receives research funds as a contract through Washington University in St. Louis from LB Pharma, New York. Rebecca F. Gottesman is former Associate Editor for the journal Neurology. This article was prepared while Dr Rebecca Gottesman was employed at the Johns Hopkins University School of Medicine. The opinions expressed in this article are the author's own and do not reflect the view of the NIH, the Department of Health and Human Services, or the US government.

Data S1

Tables S1–S12

Figures S1–S2

References 16, 17, 18, 19, 20

Click here for additional data file.