The figure highlights the heterogeneity of brain WMH and pathologies accumulating in aging brains. The bar chart in the lower left corner shows the frequency of each of the WMH and pathologies in the analytic sample and also illustrates number of participants without any of the pathologies or WMH. Connected dots on the x-axis indicate the 60 most frequent combinations of WMH and the pathologies. The bar plot shows the frequencies of the combinations of WMH and the pathologies in the participants with high (orange) vs low (blue) square root–transformed parkinsonism scores (stratified based on the median square root of parkinsonism score, which was 3.9), ordered by their frequency. The height of each bar corresponds to the number of participants with each individual or combination of WMH and the pathologies. The figure illustrates that WMH is more common than any of the examined pathologies, and most participants have combinations of the pathologies with or without WMH compared with having WMH or each pathology alone. AD indicates Alzheimer disease; PD, Parkinson disease.
eFigure 1. Flowchart of the Religious Orders Study (ROS) participants included in the current study.
eFigure 2. Flowchart of the Rush Memory and Aging Project (MAP) participants included in the current study.
eFigure 3. Flowchart of the Minority and Aging Research Study (MARS) participants included in the current study.
eFigure 4. Association of white matter hyperintensity with progressive parkinsonism in older adults without a clinical diagnosis of PD.
eTable 1. MRI protocols used at 4 scanners to acquire ex vivo images.
eTable 2. A linear mixed-effects model examining the association of white matter hyperintensity (WMH) burden with progressive parkinsonism prior to death.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Oveisgharan S, Yu L, Poole VN, et al. Association of White Matter Hyperintensities With Pathology and Progression of Parkinsonism in Aging. JAMA Neurol. 2021;78(12):1494–1502. doi:10.1001/jamaneurol.2021.3996
Is a higher burden of cerebral white matter hyperintensity (WMH), a common manifestation of cerebrovascular disease pathologies, associated with the rate of progressive parkinsonism in older adults without a clinical diagnosis of Parkinson disease?
In this cohort study including 516 decedents, higher WMH was associated with faster progressive parkinsonism.
Both WMH and cerebrovascular disease pathologies may contribute to progressive parkinsonism in older adults; further studies are needed to determine if aggressive treatment of vascular risk factors and diseases can reduce the occurrence or severity of late-life parkinsonism.
Progressive parkinsonism is common in older adults without a diagnosis of Parkinson disease and is associated with adverse health outcomes, but its pathologic basis is controversial.
To examine if the burden of cerebral white matter hyperintensity (WMH), a common manifestation of cerebrovascular disease pathologies, is associated with the rate of progressive parkinsonism.
Design, Setting, and Participants
This community-based cohort study included participants recruited in 3 ongoing cohorts that began enrollment in 1994, 1997, and 2004. Prior to death, participants were observed for a mean of 7.5 years, with annual clinical assessments. From 4427 participants enrolled in the 3 cohorts, 2134 died. Postmortem autopsy was performed in 1725 decedents, and 598 also had ex vivo brain magnetic resonance imaging. Participants were excluded if they were missing any of the 9 postmortem pathology indices (n = 22) or repeated parkinsonism assessment (n = 41) or had received a clinical diagnosis of Parkinson disease at any point before or during the study (n = 19). Data were analyzed from April 2020 to August 2021.
WMH burden was assessed using a modified Fazekas rating scale.
Main Outcomes and Measures
Parkinsonism was assessed annually using 26 items of a modified motor portion of the Unified Parkinson’s Disease Rating Scale. A summary score was developed from the item scores, with higher scores indicating more severe parkinsonism.
Of 516 included decedents, 364 (70.5%) were female, and the mean (SD) age at death was 90.2 (6.4) years. Higher WMH was associated with faster progressive parkinsonism (estimate, 0.024; SE, 0.008; P = .002). The attenuation of this association was greater when controlling for indices of cerebrovascular disease pathologies than when controlling for neurodegenerative pathologies (cerebrovascular disease: estimate, 0.019; SE, 0.008; P = .02; neurodegenerative: estimate, 0.022; SE, 0.008; P = .003), but both remained significant.
Conclusions and Relevance
In this cohort study, higher levels of both WMH and indices of cerebrovascular disease pathologies in aging brains were associated with more rapid progressive parkinsonism. Further studies are needed to determine if in vivo brain imaging of older adults for evidence of WMH and aggressive medical treatment of vascular risk factors and diseases can reduce the occurrence or severity of late-life parkinsonism.
Parkinsonism, a constellation of 4 parkinsonian signs including parkinsonian gait, bradykinesia, tremor, and rigidity, is common in old age, affecting approximately half of community-dwelling older adults without a clinical diagnosis of Parkinson disease (PD).1 Parkinsonism is not benign, as it is associated with a number of adverse health outcomes, including disability, falls, dementia, and death.2-4 Understanding the biology underlying parkinsonism is crucial for developing therapies to decrease its public health consequences in aging populations.
Although parkinsonism is common in older adults, its pathologic basis remains controversial. Epidemiological studies have estimated PD prevalence to be 2% to 5% in adults older than 80 years,5 which is much lower than the estimated prevalence of up to 50% for parkinsonism.1 A genetic variant associated with PD was not associated with parkinsonism in elderly individuals.6 Brain magnetic resonance imaging (MRI) and autopsy studies suggest that one or more markers of cerebrovascular disease pathologies are commonly associated with the severity of parkinsonism in older adults without PD.6-10 Thus, converging evidence suggests that only a minority of cases of parkinsonism in the general population may represent preclinical or subclinical PD.
Few longitudinal studies have examined the association of brain imaging11 or postmortem cerebrovascular disease pathologies with progressive parkinsonism in older adults without PD.12,13 In recent studies, we found that cerebrovascular disease brain pathologies were associated with a faster progression of parkinsonism.12,13 However, these latter studies did not assess the contribution of brain white matter hyperintensities (WMHs) to the progression of parkinsonism.11
Here, we extend our prior work by examining if postmortem WMH is associated with progressive parkinsonism when controlling for neurodegenerative pathologies, including PD and cerebrovascular disease pathologies. To ensure that the indices of WMH were obtained at the same time as postmortem indices of brain pathologies, we used WMH ratings of ex vivo brain imaging.
The study participants were enrolled in one of 3 ongoing longitudinal clinical-pathological studies of aging conducted at the Rush Alzheimer’s Disease Center of Rush University Medical Center in Chicago, Illinois: the Religious Orders Study (ROS), the Rush Memory and Aging Project (MAP), and the Minority Aging Research Study (MARS). ROS began enrollment in 1994 of older nuns, priests, brothers, and sisters across the US. MAP began enrollment in 1997 of older adults from homes, retirement facilities, and subsidized housing across the Chicago metropolitan area. Inclusion criteria for both ROS and MAP included being 65 years or older without a known dementia diagnosis and consenting for annual clinical evaluation and for brain donation at the time of death. MARS began enrollment in 2004 and recruits only Black individuals living across the Chicago metropolitan area who are 65 years or older and without a known dementia diagnosis at study entry. Consenting for annual assessment is required at study entry of MARS, while consenting for brain donation is optional to allow study staff to establish respect and gain the trust of the participants. Harmonized clinical assessments and autopsy protocols were used in all 3 studies. The same study personnel collected clinical and autopsy data for all 3 studies, facilitating joint analyses. Additional details are provided in prior publications.14,15 This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
The primary objective of this study was to examine the association between WMH and progressive parkinsonism, controlling for postmortem brain pathologies. Therefore, we only included decedents with completed brain autopsy, 2 or more valid clinical assessments of parkinsonism, and no clinical diagnosis of PD. The flowchart of the analytic sample (N = 516) is shown in eFigures 1 to 3 in the Supplement.
Each cohort was separately approved by an institutional review board of Rush University Medical Center. Written informed consent was obtained from each participant at study entry, as was a signed anatomical gift act for brain donation.
Trained nurse practitioners annually assessed 4 parkinsonian signs using 26 items of the motor portion of a modified Unified Parkinson’s Disease Rating Scale.16 Item scores for each of the 4 signs are used to compute the total severity score of each sign. The 4 scores were averaged into a global parkinsonian score, with a higher score indicating more severe parkinsonism.
In prior publications, we have shown that the nurse clinicians’ assessment of parkinsonism is reliable, showing high interrater reliability, and stable during a short-term test-retest evaluation.17 The nurse assessments were highly correlated with concurrent assessments of the same individuals by a movement disorders physician.17 A higher global parkinsonism score is associated with a number of adverse health outcomes18 and higher burden of brain pathologies,13 supporting the use of this instrument.
We have also developed and validated categorical classification of parkinsonism using scores of the Unified Parkinson’s Disease Rating Scale items.19 A parkinsonian sign was present if 2 or more of its items showed at least mild impairment in the assessments. Parkinsonism was present if at least 2 of 4 parkinsonian signs were present.
Participants were asked if they had ever received a clinical diagnosis of PD from a physician for which they had been prescribed levodopa or a dopamine agonist. Similar to other epidemiologic studies, the diagnosis of PD was based on self-reported history, review of current medications, and available medical records.19
The mean (SD) interval between death and autopsy procedure was 8.9 (6.5) hours. Details of the procedure are provided elsewhere.20,21 After brain removal, one hemisphere is frozen and the second hemisphere is fixed in a 4% formaldehyde solution. Following formalin fixation, postmortem brain MRI was obtained prior to the collection of indices of brain pathologies.20
Data were acquired using four 3-T MRI scanners (GE Signa [GE Healthcare], MAGNETOM Trio [Siemens Medical Solutions], Philips Achieva [Philips], and MAGNETOM Verio [Siemens Medical Solutions]; eTable 1 in the Supplement). Two MRI sequences were acquired for each brain: 2-dimensional fluid-attenuated inversion recovery and 2-dimensional multiecho spin echo. T2-weighted images acquired at 49.5 to 55 milliseconds echo time were used to maintain consistency across scanners. The images were intensity-normalized for further processing.
One of the novel aspects of this study is that MRI imaging was conducted in brain hemispheres ex vivo at the time of autopsy while immersed in formaldehyde solution. This offers a unique opportunity to measure both the burden of WMH and other brain pathologies at the same time. Since imaging of the brain hemispheres was performed ex vivo while immersed in formaldehyde solution, none of the publicly available software for segmenting and quantifying WMH could be applied. Therefore, this study used visual rating of WMH burden. Details of the WMH rating are provided elsewhere.20 Briefly, a trained rater examined fluid-attenuated inversion recovery and T2-weighted images to separately rate WMH burden in periventricular and deep white matter regions according to the original 4-level Fazekas scale. The whole-brain WMH score was the maximum of the periventricular and deep white matter regions ratings. Considering uncertainty in separating none from the mild level of WMH owing to incomplete cancellation of formalin solution signals, we combined these 2 levels. Therefore, a 3-level scale was used for the whole-brain WMH rating: none or mild, moderate, and severe.
To assess reliability and validity of the WMH ratings, the rater repeated the ratings in 50 participants, and an expert rated WMH burden in 32 participants. The analyses showed that the WMH rating had excellent reliability and good validity.20 In a second validation study, ex vivo WMH rating was compared with in vivo WMH rating in 79 participants.20 Level of WMH was the same in 59 participants (75%) and was higher in ex vivo imaging in the rest (20 [25%]). Notably, participants with more severe ex vivo WMH rating had a longer time interval between in vivo and ex vivo image acquisitions, which could explain more severe ex vivo WMH level owing to accumulating WMH.20 These prior studies support validity of examining ex vivo WMH ratings.
We assessed and quantified indices of 5 cerebrovascular disease and 4 neurodegenerative pathologies, including PD. Details of the pathological assessment are provided in the eMethods in the Supplement and in prior publications.12,13,21-23
Sex and race were reported by participants at the baseline evaluation. Self-reported medical history of vascular risk factors and diseases was obtained, which were summarized by summary scores.24 At each annual visit, all the participants’ medication containers were inspected, and the names and dosages were recorded and subsequently coded using the Medi-Span Drug Database System.25 Antipsychotic medication use was present if the drug codes began with numbers from 590000 to 594999.
Spearman correlation coefficient and χ2 tests were used to examine bivariate associations between WMH level and covariates, as appropriate. The global parkinsonian score was positively skewed; the scores were square root transformed.12,13 Linear mixed-effects models were used to examine the association between WMH level and annual rate of change in parkinsonism. The core model included a term for time (indicating rate of change in parkinsonism) as well as 8 additional terms that provide the associations of each of the covariates, including age at death, sex, education, and WMH with the level of and rate of change in progressive parkinsonism. In subsequent models, we added terms for potential confounders and their interactions with time to examine whether the associations between WMH and parkinsonism were attenuated by these additional covariates. Then, we repeated the core model after addition of terms for 4 neurodegenerative and 5 cerebrovascular disease pathology indices and their interactions with time to determine if these pathologies attenuated the association between WMH and parkinsonism. To examine association of WMH with the presence of parkinsonian signs, we used logistic regressions controlled for demographic characteristics. Analyses were conducted using SAS version 9.4 (SAS Institute), and 2-tailed P values less than .05 were used for rejection of null hypotheses.
Of 516 included decedents, 364 (70.5%) were female, and the mean (SD) age at death was 90.2 (6.4) years; their clinical characteristics at their last visit (a mean 1 year before death) and postmortem indices are summarized in Table 1. Almost half of participants (244 [47.3%]) had severe WMH burden, more than one-third (181 [35.1%]) had moderate WMH burden, and 91 (17.6%) had no to mild levels of WMH. Older adults, women, and those with a history of stroke were more likely to have more severe WMH.
Moderate to severe WMH was more common than any other examined pathology, including AD, cerebrovascular disease pathologies, PD pathology, or hippocampal sclerosis (Figure; Table 1), and WMH tended to coexist with other pathologies (Figure). In bivariate associations, all 5 cerebrovascular disease pathologies were associated with a higher burden of WMH, but AD pathology was the only neurodegenerative pathology associated with WMH (Table 1).
Participants were examined annually for a mean (SD) of 7.5 (4.1) years. The mean (SD; range) square root–transformed parkinsonism score was 2.8 (1.2; 0-7.6) at study baseline and increased in severity to 3.8 (1.3; 0-7.4) proximate to death.
In a linear mixed-effects model controlling for demographic characteristics, the severity of parkinsonism increased by about 0.13 units per year (estimate, 0.128; SE, 0.007; P < .001). More severe WMH was associated with more rapid progressive parkinsonism (estimate, 0.024; SE, 0.008; P = .002) and more severe parkinsonism proximate to death (estimate, 0.325; SE, 0.076; P < .001) (eFigure 4 in the Supplement). To contextualize the association of WMH with progressive parkinsonism, we used model-derived estimates (eTable 2 in the Supplement) to compare the rate of progressive parkinsonism in 2 representative 90-year-old women with average education but different burdens of WMH. Progressive parkinsonism was 51% faster in the woman with severe vs mild WMH.
The global parkinsonian score was an average of 4 scores corresponding to 4 parkinsonian signs. The scores of parkinsonian gait and bradykinesia met assumptions of linear mixed-effects model, enabling us to examine the association of WMH with the progression of both signs. A higher level of WMH was associated with more rapid progression of bradykinesia (estimate, 0.036; SE, 0.012; P = .004) and parkinsonian gait (estimate, 0.036; SE, 0.012; P = .003). Rigidity and tremor did not meet linear mixed-effects model assumptions owing to the frequent score of 0 during follow-up assessments. Therefore, we dichotomized these signs as either present or absent prior to death and examined their associations with postmortem WMH. A higher level of WMH was associated with a higher odds of rigidity (odds ratio, 1.75; 95% CI, 1.30-2.36; P < .001) but not tremor (odds ratio, 0.97; 95% CI, 0.75-1.26, P = .82).
We examined the association of WMH with parkinsonism in the presence of potential confounders. The association of WMH burden with progressive parkinsonism remained significant when controlling for vascular risk factors and diseases (Table 2, models 1 and 2). Next, we examined whether the association between WMH and parkinsonism persisted after adjustment for several study-related covariates. The examined covariates were cohort (ROS vs other cohorts), MRI scanner (3 scanners compared with scanner at the Rush University Medical Center [MAGNETOM Verio] as the reference), and the time interval between death and image acquisition (median [IQR] of 33.9 [31.0-43.5] days). In separate mixed-effects models, we added to the core model terms for each of the 3 covariates and their interactions with time. The association between WMH and rate of progressive parkinsonism or level of parkinsonism prior to death did not change after adjustment for these study covariates (Table 2, models 3 to 5).
In further sensitivity analyses, we examined if our findings were driven by those individuals receiving antipsychotic medications, which can increase the severity of parkinsonism in older adults. Exclusion of 74 participants with antipsychotic use did not change the association of WMH with progressive parkinsonism (Table 2, models 6).
Our previous studies showed that a higher burden of PD and cerebrovascular disease pathologies were associated with a more rapid progression of parkinsonism.13 Our current analyses also show that WMH is associated with postmortem brain pathologies (Table 1).20 Therefore, we examined whether WMH was independently associated with progressive parkinsonism when controlling for other brain pathologies in the same model. Adding terms for cerebrovascular disease pathologies attenuated the association between WMH and progressive parkinsonism by 21% (without cerebrovascular disease pathologies: estimate, 0.024; SE, 0.008; P = .002; controlling for cerebrovascular disease pathologies: estimate, 0.019; SE, 0.008; P = .02), but WMH remained independently associated with progressive parkinsonism (Table 3, model 1). In contrast, adding terms for neurodegenerative pathologies, including PD pathology, attenuated the association of WMH with progressive parkinsonism by only 8% (controlling for neurodegenerative pathologies: estimate, 0.022; SE, 0.008; P = .003) (Table 3, model 2). These findings were unchanged when we included terms for all the brain pathologies in a single model (Table 3, model 3). These findings indicate that WMH has an association with progressive parkinsonism apart from markers of both neurodegenerative and other cerebrovascular disease pathologies measured in the current study.
The study findings show that more severe WMH in older adults is associated with more rapid progressive parkinsonism. The association between WMH and progressive parkinsonism remained significant despite partial attenuation in a model including postmortem indices of cerebrovascular disease brain pathologies. These longitudinal findings support a notion that both WMH and cerebrovascular disease pathologies may be underestimated independent contributors to progressive parkinsonism, which is common in older adults. Prospective studies are needed to determine if in vivo brain imaging for signs of WMH and more aggressive medical treatment of vascular risk factors and diseases in older adults may reduce the common occurrence of progressive parkinsonism in aging adults.
Parkinsonism in older adults, while being common26 and a risk factor for adverse health outcomes,27 is a heterogeneous group of disorders associated with varied medical conditions, neurodegenerative diseases, and medications. Yet in most older adults, without overt diseases or medications known to cause parkinsonism, the etiology of this progressive condition remains unknown. While varied neurodegenerative diseases, including PD, may manifest parkinsonism, evidence to date suggests that parkinsonism caused by preclinical PD is rare in the general population.13 Ongoing controversy persists among movement disorders experts about the diagnosis and prevalence of vascular parkinsonism.28-30 Some have suggested that vascular parkinsonism is rare and only occurs when an ischemic or hemorrhagic event affects substantia nigra or nigrostriatal pathway.28 They suggest that the more common occurrence of progressive parkinsonism without an acute event is caused by other diseases, including PD and normal pressure hydrocephalus, or the symptoms to be pseudoparkinsonism symptoms, as seen in depression-associated apathy.28 In their view, WMH is either a confounder for other conditions or an incidental finding that is common in old age, but WMH is not a cause of parkinsonism in older adults. Moreover, such experts have coined the term pseudovascular parkinsonism to indicate inadequate evidence for supporting WMH as a sign of cerebrovascular disease.28 Thus, ongoing controversy continues about the extent to which there is a causal link between cerebrovascular disease and progressive parkinsonism and its clinical nosology.29
Few longitudinal studies have examined indices of WMH together with postmortem indices of cerebrovascular disease pathologies. An important strength of the current study design is that WMH was obtained from ex vivo brain imaging. When antemortem brain imaging is used, there is always a variable temporal delay between the measured brain WMH and the collection of postmortem indices after death. Our study design enabled us to examine the association of WMH indices with parkinsonism controlling for postmortem pathologies collected at the same time. Nonetheless, our study findings can be generalized to in vivo WMH burden, as a high level of agreement exists between ex vivo and in vivo WMH rating.20
In this study, we examined global measures of WMH and brain pathologies in association with progressive parkinsonism. Yet the clinical consequences of brain pathologies like WMH on parkinsonism will vary in different brain regions. Previous studies have suggested that association of disruption of frontal and internal capsule white matter integrity is most strongly related to the severity of vascular parkinsonism.11,31 These data would suggest that accumulation of WMH and brain pathologies in frontal thalamocortical pathways that are crucial for neural control of movement may have stronger associations with the rate of progressive parkinsonism compared with the global measures examined in our study. Further studies that use quantitative methods to measure WMH burden more precisely are needed to identify the specific brain regions that underlie progressive parkinsonism in older adults.
The current study extends our prior postmortem studies providing evidence that more severe WMH and a higher level of indices of cerebrovascular disease pathologies make separate contributions to a higher level of parkinsonism proximate to death as well as with faster progressive parkinsonism.4,5 Contrary to some prior reports that WMHs are pseudovascular or incidental findings with respect to parkinsonism, our findings lend further support for the suggestion that diverse cerebrovascular disease pathologies and markers, including WMH, may in fact be the most common underlying pathologies contributing to the common occurrence of progressive parkinsonism in old age.12,13
The current findings may have important consequences for both aging research and the clinical care of older adults with late-life motor impairment. The current findings derive from an observational study; prospective clinical studies will be needed to determine if in vivo brain imaging for evidence of WMH together with more aggressive medical treatments, ie, antihypertensive or statins, for vascular risk factors and diseases or lifestyle modification22,32,33 in older adults with progressive parkinsonism might reduce the magnitude of incident parkinsonism.34,35 Also, to what extent can in vivo brain MRI imaging be used to improve the homogeneity of adults recruited for clinical trials of late-life parkinsonism? Our findings may also have clinical consequences for adults with a clinical diagnosis of PD. Cerebrovascular disease pathologies commonly accumulate in aging brains. Even in older adults with PD, it is crucial to consider the cooccurrence of cerebrovascular disease pathologies that might account for progressive parkinsonism, especially in adults who respond poorly to dopaminergic treatments.6
This study has several strengths. The data came from 3 cohorts with high follow-up rates, making attrition bias less likely to affect our findings. Harmonized study protocols lend support for joint analyses. However, the study has several limitations. The study findings are from a selected cohort of adults. Although we included participants from a cohort that exclusively recruits older Black adults, 95% of the analytic sample were White adults. Replication of findings in future studies with an educationally and racially and ethnically more diverse sample is required. The current study used a visual rating scale to assess ex vivo WMH burden at the same time that postmortem brain pathologies were measured. Further studies using quantitative methods to measure the burden of ex vivo WMH more precisely will be needed to replicate the current findings. While assessments of parkinsonism were not performed by a movement disorders specialist, postmortem assessments highlighted that only a small minority of participants showed postmortem evidence of PD pathology.
In this cohort study, a higher level of both WMH and indices of cerebrovascular disease pathologies in aging brains were associated with more rapid progressive parkinsonism. Further studies are needed to determine if in vivo brain imaging of older adults for evidence of WMH and aggressive medical treatment of vascular risk factors and diseases can reduce the occurrence or severity of late-life parkinsonism.
Accepted for Publication: September 13, 2021.
Published Online: November 1, 2021. doi:10.1001/jamaneurol.2021.3996
Correction: This article was corrected on March 21, 2022, to fix an error in the bar chart in the Figure.
Corresponding Author: Shahram Oveisgharan, MD, Rush Alzheimer’s Disease Center, Rush University Medical Center, 1750 W Harrison, Ste 1000, Chicago, IL 60612 (email@example.com).
Author Contributions: Drs Bennett and Buchman had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Oveisgharan, Bennett, Buchman.
Acquisition, analysis, or interpretation of data: Oveisgharan, Yu, Poole, Evia, Barnes, Schneider, Arfanakis, Buchman.
Drafting of the manuscript: Oveisgharan, Arfanakis, Buchman.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Yu.
Obtained funding: Barnes, Arfanakis, Bennett, Buchman.
Administrative, technical, or material support: Evia, Barnes, Schneider, Arfanakis, Bennett.
Study supervision: Bennett, Buchman.
Conflict of Interest Disclosures: Drs Oveisgharan, Schneider, Arfanakis, and Buchman have received grants from the National Institutes of Health. No other disclosures were reported.
Funding/Support: This work was supported by grants R01AG043379, R01AG047976, R01AG056352, R01AG017917, RF1AG022018, R01NS078009, P30AG10161, UH3NS100599, R01AG064233, and R01AG15819 from the National Institutes of Health.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank the participants of the Religious Orders Study, the Rush Memory and Aging Project, and the Minority Aging Research Study. We acknowledge the roles of Traci Colvin, MPH, Tracey Nowakowski, MA, Charlene J. Gamboa, PhD, Shayla Calloway, BA, and Karen Skish, MS, for study coordination, Dominika Burba, MS, for conducting statistical analyses, and other staff of the Rush Alzheimer’s Disease Center, Chicago, Illinois. Statistical analyses were conducted by Dominika Burba, MS, under supervision of Lei Yu, PhD, at Rush Alzheimer’s Disease Center. All contributors were compensated for their work.