Cerebrovascular and Neurodegenerative Pathologies in Long-Term Stable Mild Cognitive Impairment

Abstract

Background:

Mild cognitive impairment (MCI) is considered by some to be a prodromal phase of a progressive disease (i.e., neurodegeneration) resulting in dementia; however, a substantial portion of individuals (ranging from 5–30%) remain cognitively stable over the long term (sMCI). The etiology of sMCI is unclear but may be linked to cerebrovascular disease (CVD), as evidence from longitudinal studies suggest a significant proportion of individuals with vasculopathy remain stable over time.

Objective:

To quantify the presence of neurodegenerative and vascular pathologies in individuals with long-term (>5-year) sMCI, in a preliminary test of the hypothesis that CVD may be a contributor to non-degenerative cognitive impairment. We expect frequent vasculopathy at autopsy in sMCI relative to neurodegenerative disease, and relative to individuals who convert to dementia.

Methods:

In this retrospective study, using data from the National Alzheimer’s Coordinating Center, individuals with sMCI (n = 28) were compared to those with MCI who declined over a 5 to 9-year period (dMCI; n = 139) on measures of neurodegenerative pathology (i.e., Aβ plaques, neurofibrillary tangles, TDP-43, and cerebral amyloid angiopathy) and CVD (infarcts, lacunes, microinfarcts, hemorrhages, and microbleeds).

Results:

Alzheimer’s disease pathology (Aβ plaques, neurofibrillary tangles, and cerebral amyloid angiopathy) was significantly higher in the dMCI group than the sMCI group. Microinfarcts were the only vasculopathy associated with group membership; these were more frequent in sMCI.

Conclusion:

The most frequent neuropathology in this sample of long-term sMCI was microinfarcts, tentatively suggesting that silent small vessel disease may characterize non-worsening cognitive impairment.

Keywords

Autopsy cerebrovascular cognition dementia longitudinal neuropathology

INTRODUCTION

Mild cognitive impairment (MCI) is considered by some to be a phase of cognitive decline between normal aging and dementia [1]. As a construct, MCI is not well defined. Its diagnosis rests almost entirely on clinical presentation. Multiple different criteria currently exist, though all share major components (i.e., objective and subjective evidence of cognitive impairment, with preserved functional independence). This heavy reliance on clinical presentation, combined with a lack of consensus as to the extent or severity of ‘cognitive impairment’ necessary to warrant a diagnosis, contributes to substantial heterogeneity in the overall group of individuals diagnosed with MCI. For instance, 5–10% of MCI cases convert annually to dementia, usually Alzheimer’s disease (AD) [2], but a considerable subset do not show any worsening in cognition or function. The exact percentage of these stable MCI cases (sMCI) varies from study to study, but ranges from 5–30% when considering follow-up data up to 10 years [3 –6]. The reason for this stability remains unclear; however, it suggests that the etiology of the cognitive impairment in these individuals is non-progressive (i.e., not due to typical neurodegenerative pathologies, such as amyloid-β [Aβ] plaques, neurofibrillary tangles, Lewy bodies, transactive response DNA-binding protein 43 (TDP-43), or cerebral amyloid angiopathy [CAA]) and may be linked to a potentially reversible or treatable cause.

A number of prior studies have sought to identify baseline factors that predict subsequent longitudinal trajectories of decline versus stability in individuals with MCI. Clinical factors that have been consistently linked to conversion include baseline functional abilities [7 –9], multi-domain amnestic baseline impairments [7 , 10–16], diabetes [17 –20], and neuropsychiatric symptoms [17, 21]. Certain biomarkers have been identified in this regard; unquestionably, the most robustly associated with conversion is positive APOE ɛ4 status [7 , 22]. Hippocampal atrophy has also proven useful as a predictor of conversion from MCI to dementia [23 –28], though it is often not significantly superior to clinical measures when both types of information are considered together [23, 26]. By default, the absence of these factors was linked to cognitive stability in all aforementioned studies.

Beyond predicting the future course of disease, an effective personalized medicine approach relies on pinpointing the underlying causes of impaired cognition in cases of sMCI to guide intervention studies and ultimately allow for targeted treatment. Two studies to our knowledge have examined neurodegenerative disease biomarkers in sMCI in an attempt to elucidate its underlying causes. Parnetti and colleagues (2014) examined cerebrospinal fluid (CSF) Aβ₄₂, p-tau, and t-tau levels over four years in a group of individuals with MCI who remained cognitively stable throughout the study, as well as individuals with MCI who declined to AD dementia. Ratios of Aβ₄₂:Aβ₄₀, Aβ₄₂:t-tau, and Aβ₄₂:p-tau were significantly decreased in the MCI group that declined to AD dementia compared to the sMCI group, indicating less neurodegenerative pathology in individuals with sMCI than in those who decline to dementia [29]. A second study conducted by Cerami and colleagues (2018) examined CSF Aβ₄₂ levels and conducted amyloid and glucose positron emission tomography (PET) imaging over a mean of 58.3 months in a sample of sMCI. CSF Aβ₄₂ was significantly higher than what is normally reported in prodromal AD dementia, and was consistent with Aβ burden on PET imaging [30]. The results of the two studies suggest a lack of significant neurodegenerative processes in individuals with sMCI, but findings from these studies only provide conclusions about what the cause of sMCI is not (i.e., neurodegeneration) and further clarification is needed to identify factors that contribute to impaired cognition in sMCI. Inferences based on current literature point to cerebrovascular disease (CVD) as a potential candidate.

CVD is highly comorbid with neurodegenerative disorders [31, 32], occurring in up to 80–90% of patients with AD and producing synergistic effects on cognition [33 –35]. Independently of AD, CVD has been identified as a contributing factor to cognitive impairment [36] and can cause dementia (i.e., vascular dementia, VaD [37]), primarily by jeopardizing the integrity of vessels carrying blood to key brain areas, thus compromising function. Vascular brain changes—particularly subcortical—can be degenerative (e.g., progressive thinning of vessel walls; progressive accumulation of lipids within vessels) and are associated with declines in cognition; however, these declines are generally mild (≤0.05 standard deviation change per year [38]). Several longitudinal studies of cognition in individuals with mild vascular cognitive impairment have found that 3–44% decline to dementia [39 –43], implying that most cases remain relatively stable over the longitudinal period. The cognitive rehabilitation literature also indicates that cognitive functioning in post-stroke patients may not worsen over time, and can even improve [44 –46]. Overall, the CVD literature indicates that a subset of individuals with vasculopathies remain cognitively stable over time. As such, we hypothesize that the effects of CVD on cognition in MCI are distinct from those of cascading pathological changes that cause neuronal degeneration (e.g., amyloid [47]).

It must be acknowledged that many previous studies of MCI have identified vascular features as factors that increase risk for conversion from MCI to dementia (e.g., [17, 48]). They are hypothesized by some authors to play a causal role in the development of neurodegenerative AD pathology (e.g., [49, 50]). It may seem counterintuitive, then, to propose that they may characterize stable MCI. The key consideration relates to the directionality of the research question. Previous studies have investigated how vascular burden affects conversion prospectively; in other words, the question of interest in previous research has been: “Of people with MCI who have comorbid vascular disease burden, how many decline versus how many remain stable?” Important information about the underlying cause(s) of sMCI can be gleaned by examining the relationship between disease burden and cognitive outcomes from the opposite perspective: “Of people with MCI who decline versus remain stable, how many have vascular disease?” Classifying participants retrospectively according to their cognitive outcomes (degenerative versus non-degenerative) can provide a framework within which to examine the causes of different types of cognitive impairment (stable versus declining). Addressing the same relationships between variables from alternative perspectives can yield vastly different but equally valuable outcomes; for instance, findings from oncology research have described that the risk of developing lung cancer among heavy smokers is approximately 20% [51], whereas nearly 90% of lung cancer deaths are attributable to smoking [52]. In the same way, it is expected that the current project will provide a novel outlook on the association between vascular burden and cognitive impairment in individuals who are assumed to be in the early stages of a dementia process but whose cognitive impairment does not worsen.

The purpose of the current study is to empirically determine whether cerebrovascular disease is significantly associated with sMCI. We hypothesize that 1) individuals with sMCI will show significantly less neurodegenerative pathology (i.e., neurofibrillary tangles, Aβ plaques, TDP-43, CAA, and/or Lewy bodies) at autopsy than individuals with MCI who decline cognitively over at least a five-year longitudinal period (i.e., dMCI); and 2) the sMCI group will show greater CVD burden compared to the dMCI group.

MATERIALS AND METHODS

The current study is a retrospective analysis on neurodegenerative and CVD data acquired at autopsy from the National Alzheimer’s Coordinating Center data repository (NACC; https://www.alz.washington.edu/index.html). The NACC data repository is a collection of data from 2005 to present and contains data from several thousand subjects collected across 39 Alzheimer’s Disease Centers (ADCs) in the United States of America [53, 54]. Participants in the NACC repository are recruited via clinician referral, self-referral by participants or family members, and recruitment through the community. The ADCs typically collect longitudinal data at approximately yearly intervals by trained clinicians and clinic personnel. The NACC variables of interest for the current study include sociodemographic variables (age, sex, education, ethnicity), MCI subtype [55], autopsy data, APOE status, and vascular health characteristics (current or history of: heart attack/cardiac arrest; angioplasty, endarterectomy, or stent; other cardiovascular disease; transient ischemic attack [TIA]; diabetes; hypertension; hypercholesterolemia; and years smoking). Years smoking were recorded as a continuous variable based on participant’s self-report. Medication at baseline was also obtained. All other vascular health characteristics are scored as absent, recent/active, or remote/inactive in the NACC data repository. For the purpose of the current analysis, recent/active or remote/inactive factors were pooled into a single rating of ‘present’ because both mid- and late-life vascular risk factors may increase dementia risk [35, 56]. Although an ideal approach would have been to consider ‘absent’, ‘recent/active’, or ‘remote/inactive’ as three separate vascular risk factor categories, this reduced the samples within each cell so that statistical comparisons were no longer possible.

Participants

Participant data of individuals aged 50 years or older were requested from NACC (n = 42,022). From this sample, participants were excluded if they had not completed at least five annual visits (this was necessary to determine a robust operational definition of cognitive stability; no upper bound follow-up duration was set; n = 30,447); if they did not have autopsy data available (n = 9,567); if their health history included any other self-reported or clinician-diagnosed neurological condition (e.g., multiple sclerosis, Guillain-Barré syndrome, essential tremor) according to the NACC data repository (n = 332); if they were cognitively normal or diagnosed with dementia at baseline visit (n = 1,277); and if they had < 1.5 years between last visit and autopsy (n = 232). The final sample consisted of 167 participants (Fig. 1).

Fig. 1

Flow chart depicting how many participants were excluded from the original dataset based on inclusion and exclusion criteria.

Cognitive status in the NACC data repository is based on ADC-specific procedures but includes at least objective cognitive impairment and evidence of functional independence determined by a either a single clinician or consensus from a team of clinicians. The Clinical Dementia Rating (CDR® Dementia Staging Instrument) and/or neuropsychological test results are used to evaluate cognition, in which normal cognition is a global CDR score of 0, and/or neuropsychological testing within normal range. MCI criteria are met if there are subjective and objective impairments in cognition, and independence in functional activities is maintained. Participants are considered demented when objective cognitive impairment is present, with a loss of functional ability, and a decline from previous functioning that cannot be explained by delirium or major psychiatric disorder. Individual ADCs each determine this using their own protocols, so exact operationalization of MCI may differ slightly between ADC sites. We used the cognitive status classifications determined by ACD clinicians (rather than algorithmically determining cognitive status using NACC’s available cognitive and functional measures) because we surmised that the evaluators’ clinical judgement may have incorporated additional information not explicitly listed in the data repository (e.g., environmental factors or behavioral observations that may better account for the cognitive symptoms). We estimate that relying on these clinical judgements is akin to real-world neuropsychological assessment and supports the generalizability of our results. Participants were assigned to the ‘stable MCI’ group (sMCI, n = 28) if their cognitive status was ‘MCI’ at their first visit and ‘MCI’ at their last visit (at least 5 years later). They were assigned to the ‘declined MCI’ group (dMCI, n = 139), if their cognitive status was ‘MCI’ at the initial visit and ‘demented’ at the last visit. The NACC data repository further provides data on MCI subtype (i.e., single versus multiple domain and amnestic versus non-amnestic MCI [1]), which is determined based on each ADC’s protocol. Although participants’ first and last visits were considered for sMCI versus dMCI classification purposes, we must acknowledge the variability in cognitive trajectories that is known to be common in MCI (e.g., [57]): 19 of the sMCI cases (68%) reverted to normal cognition at least once prior to their last visit. We remain reasonably confident that these cases do not represent false positive classifications of MCI, as ‘fluctuating’ cases of MCI usually ultimately re-transition to MCI (or further decline to dementia) [58]. None of the participants in the sMCI group ever declined to dementia.

APOE status (presence of 0, 1, or 2 ɛ4 alleles), a known major risk factor for AD [59], was also considered. The APOE ɛ4 variable was dichotomized into absence or presence of at least 1 ɛ4 allele (0/1), an approach taken in previous research (e.g., [60]).

Medications at baseline were obtained from the NACC data repository and categorized into anticonvulsants, anticoagulants, or antihypertensives, as all are used to treat vasculopathy or may have an effect on cognition. Medications that do not fall in these categories were not considered in the current study.

Neurodegenerative pathology at autopsy

Evidence of neurodegenerative disease at autopsy was recorded at each ADC for Braak and Braak neurofibrillary tangles (NFT) [61], Thal phase for Aβ plaques [62], neuritic plaque score using the Consortium to Establish a Registry for AD (CERAD) criteria [63, 64], Lewy body pathology [64], CAA, and TDP-43 pathology. Thal phase and CERAD neuritic plaque scores are both measures of Aβ pathology; however, there is no evidence to suggest one is superior to the other. The Thal rating system may be a better predictor of dementia showing greater sensitivity (94%) compared to CERAD scores (57%) but has worse specificity (87.7%) compared to CERAD specificity (100%) [65]. Current guidelines from the National Institute on Aging-Alzheimer’s Association (NIA-AA) for AD neuropathologic change recommend obtaining an overall score, called an ABC score, by evaluating Thal (A), Braak (B), and CERAD (C) ratings together [64]. The ABC ratings consist of four labels: “None,” “Low,” “Intermediate,” or “High” AD neuropathologic change, of which Intermediate or High ratings are considered a sufficient for a diagnosis of AD [66].

For the purposes of this project, most neuropathologic variables described below were dichotomized into clinically significant/not clinically significant based on evidence from existing literature (e.g., [64]). This was done in order for the results of this study to have the most clinical applicability (i.e., by focusing on levels of disease burden which have been shown to be associated with clinical manifestations). The Braak staging in the NACC database ranges from 0–6, with each stage indicating more diffuse NFTs beginning from the hippocampus and spreading anteriorly, dorsally and posteriorly through the cortex, in a predictable sequence [61]. Braak staging was dichotomized into clinically significant Braak stage (III-VI) and clinically non-significant Braak stage (0-II) [66, 67]. The Thal phase of Aβ plaques ranges from 0–5, and refers to the anatomical location of Aβ plaques, with more diffuse Aβ plaques rated as higher staging [62]. Thal ratings were dichotomized into clinically non-significant (Thal phases 0–2) and clinically significant (Thal phases 3–5) according to the ABC scoring of the NIA-AA guidelines [66]. The CERAD staging ranges from 0–3 indicating the density of neocortical neuritic plaques [63, 64]. CERAD ratings were dichotomized into clinically non-significant (CERAD ratings 0–1) and clinically significant (CERAD ratings 2–3) corresponding to the ABC scoring of the NIA-AA guidelines [66]. Lewy body ratings from the NACC data repository range from 0–5 with additional options for “not assessed” or “missing data” [64], modified from the NIA-AA classification [68]. Lewy body pathology ratings were dichotomized into clinically significant (Lewy body rating 3, present in cortical regions) [64] and clinically non-significant (Lewy bodies not present or present in non-cortical regions: ratings 0–2 & 4–5). CAA ratings from the NACC data repository range from 0–3 and were adapted from previous studies [69, 70], indicating absence of CAA to severe CAA burden. To our knowledge, there is no established threshold for CAA that indicates clinical significance, therefore this variable was not dichotomized. TDP–43 ratings in the NACC data repository are dichotomized into present or absent across five variables based on location (i.e., spinal cord, amygdala, hippocampus, entorhinal/inferior temporal cortex, and neocortex). These five variables were combined into a single variable to represent presence or absence of TDP-43 pathology.

Cerebrovascular disease at autopsy

The NACC data repository records evidence of cerebrovascular disease at autopsy as a dichotomous variable (present or absent), with revisions made across NACC versions (versions 1–3) to remain consistent with the most current literature. It includes infarcts (small and large arteries), lacunes, microinfarcts, hemorrhages, and microbleeds according to standard definitions [71]. In the NACC data repository, information on lacunes, infarcts, hemorrhages, and microbleeds were collected differently across neuropathology form versions. Therefore, the NACC created two variables to best capture the available information in all forms, which the authors decided was most appropriate for the current study. The first variable pooled large artery infarcts and lacunes and is recorded under a single variable called “Infarcts and lacunes.” The second variable pooled hemorrhages and microbleeds under a single variable called “Hemorrhages and microbleeds.” Therefore, ratings of “present” for these variables do not distinguish between infarcts and lacunes or hemorrhages and microbleeds. The third variable of interest is microinfarcts, small ischemic lesions that can be seen in the cortex at autopsy [71]. Microinfarcts are recoded as present or absent.

Statistical analyses

Independent samples Student t-tests were used to compare sMCI and dMCI groups on the following demographic information: age at baseline, age at death, number of visits, education, and years of smoking. Chi-square tests of independence were used to examine group differences in APOE4 status, sex, heart attack/cardiac arrest, angioplasty/endarterectomy/stent, other cardiovascular disease, TIA, diabetes, hypertension, and hypercholesterolemia (see Table 1).

Table 1

Sociodemographic and clinical characteristics of the study sample

Characteristic	sMCI (n = 28)	dMCI (n = 139)	Test statistic	p	Effect size
Age at baseline (SD)	80.21 (7.48)	76.20 (8.08)	t = 2.39	0.018	d = 0.51
Age at death (SD)	87.61 (8.18)	82.87 (8.73)	t = 2.65	0.009	d = 0.56
Number of visits (SD)	7.39 (1.62)	6.97 (1.74)	t = 1.18	0.239	d = 0.25
Sex (% male)	71.4%	63.3%	χ ² = 0.67	0.412	φ= 0.06
Years of education (SD)	15.57 (2.44)	15.95 (2.70)	t = –0.69	0.493	d = 0.15
Years smoking (SD)	9.93 (11.94)	9.17 (13.79)	t = 0.266	0.790	d = 0.06
White ethnicity	89.3%	96.4%	χ ² = 2.65	0.266	φ= 0.13
MCI type % sdaMCI/mdaMCI/	64.3/25.0/ 10.7/0.0	39.6/45.3/ 8.6/6.5		0.077	φ = 0.21
sdnaMCI/mdnaMCI^a
APOE ɛ4 0/1	23/5	77/62	χ ² = 6.94	0.008	φ = 0.20
Heart attack/cardiac arrest^a	7.1%	7.9%		1.000	φ = 0.01
Angioplasty/ endarterectomy/stent^a	3.6%	1.4%		0.425	φ = 0.06
Other cardiovascular disease^a	7.1%	8.6%		1.000	φ = 0.02
TIA^a	3.6%	3.6%		1.000	φ = 0.001
Diabetes^a	10.7%	10.1%		1.000	φ = 0.008
Hypertension	75.0%	48.9%	χ ² = 6.37	0.012	φ = 0.20
Hypercholesterolemia	71.4%	48.9%	χ ² = 4.74	0.030	φ = 0.17
Medication % of sample
Anticonvulsant^a	3.6%	4.3%		1.000	φ = 0.01
Anticoagulant^a	14.3%	7.9%		0.283	φ = 0.08
Antihypertensive	57.1%	46.0%	χ ² = 1.15	0.283	φ = 0.08

sMCI, stable mild cognitive impairment; dMCI, declined mild cognitive impairment; d, Cohen’s d; φ, Phi; sdaMCI, single domain amnestic mild cognitive impairment; mdaMCI, multiple domain amnestic mild cognitive impairment; sdnaMCI, single domain non-amnestic mild cognitive impairment; mdnaMCI, multiple domain non-amnestic mild cognitive impairment; TIA, transient ischemic attack. ^aIndicates Fisher’s exact test was conducted, which does not yield a test statistic. Bolded p -values indicate statistical significance according to Benjamini-Hochberg critical p value.

To test the hypothesis that individuals with sMCI will show significantly less neurodegenerative pathology (i.e., Thal phase, Braak stage, CERAD neuritic plaque density, Lewy bodies, TDP-43, and CAA burden) compared to dMCI, chi-square tests of independence were conducted to examine group associations with each neurodegenerative pathology variable. CAA burden was compared between groups using the Mann-Whitney U non-parametric test. To test the hypothesis that the sMCI group will show greater CVD burden compared to the dMCI group, three chi-square tests of independence were conducted to examine the association between group and each CVD variable (i.e., infarcts and lacunes, hemorrhages and microbleeds, and microinfarcts).

For all chi-square analyses, if assumptions were violated, Fisher’s exact test was reported. For all significant chi-square test of independence, follow-up Z-tests were conducted to compare proportions between the sMCI and dMCI groups. To account for inflated Type I error, the Benjamini-Hochberg procedure [72] was used. This procedure is done by first organizing the p values in ascending order. Each p value is then compared to a Benjamini-Hochberg critical value, calculated with (i/m)Q, where i is the ranked p value, m is the total number of tests and Q is the false discovery rate decided by the researchers (here, 5%). An online calculator [73] was used to calculate the Benjamini-Hochberg critical value. All results are presented alongside standard p values, but statistical significance is based on the Benjamini-Hochberg critical value. For our main hypotheses, a total of nine analyses were conducted. In examining participant health characteristics, 19 analyses were conducted.

Results from a power analysis indicated that for power of 0.80 for a Z-test examining proportion, where α< 0.01, with a sample allocation ratio of 4.96, a total sample size of 511 would be required where sMCI n = 86 and dMCI n = 425. A power analysis for t-test with power of 0.80 and medium effect size, a total sample size of 258 would be required where sMCI n = 37 and dMCI n = 221. The results from analyses below must therefore be interpreted conservatively with the knowledge that the study was underpowered to detect many between-group differences. Effect sizes are included for all analyses. Cohen’s d is reported for parametric group comparisons, and indicates the magnitude of the between group effect [74]. Eta squared (η ²) is reported for non-parametric comparisons, and indicates the variance of the dependent variable explained by the independent variable (group membership) [75]. Phi (φ) is reported for chi-square tests, and indicates the strength of the relationship between both variables [76].

RESULTS

Study participants and demographics

The study’s sample characteristics are summarized in Table 1. Notably, participants were followed-up for an average of 7.39 years (range 5–9 years) in the sMCI group, and 6.97 years (range 5–9 years) in the dMCI group, which was not significantly different (t(165) = 1.18, p = 0.239, d = 0.25), with small effect size. Age at baseline in the sMCI group ranged from 64 to 95 years, and 53 to 97 years in the dMCI group. Participants did not differ in age at baseline (t(165) = 2.39, p = 0.018, d = 0.51), with medium effect size, but the sMCI group was significantly older at death compared to the dMCI group (t(165) = 2.65, p = 0.009, d = 0.56), with a medium effect size. The groups did not differ in years of education (t(165) = –0.69, p = 0.493, d = 0.15), or sex (χ ²(1) = 0.67, p = 0.412, φ= 0.06), both showing less than small effect size. After adjusting for familywise error, APOE ɛ4 status was significantly associated with group membership, with dMCI associated with more significant APOE ɛ4 (χ ²(1) = 6.94, p = 0.008, φ= 0.20), with small to medium effect size. Overall, the sample was generally White.

There was no significant association between MCI type and group (p = 0.077, Fisher’s exact test, φ= 0.21), showing small to medium effect size, history of heart attack/cardiac arrest (p = 1.000, Fisher’s exact test, φ= 0.01), showing small effect size, history of angioplasty/endarterectomy/stent (p = 0.425, Fisher’s exact test, ϕ = 0.06), other cardiovascular disease (p = 1.000, Fisher’s exact test, ϕ = 0.02), TIA (p = 1.000, Fisher’s exact test, ϕ = 0.001), diabetes (p = 1.000, Fisher’s exact test, ϕ = 0.008), showing less than small effect size, hypertension (χ ²(1) = 6.37, p = 0.012, φ= 0.20), hypercholesterolemia (χ ²(1) = 4.74, p = 0.030, φ= 0.17), both showing small to medium effect size, or years smoking (t(157) = 0.266, p = 0.790, d = 0.06) with less than small effect size.

There was no association between group and percentage of the sample on anticonvulsant (p = 1.000, Fisher’s exact test, φ= 0.01), anticoagulant (p = 0.283, Fisher’s exact test, φ= 0.08), or antihypertensive (χ ²(1) = 1.15, p = 0.283, φ= 0.08) medications, all showing less than small effect size.

Group differences in main variables of interest

As summarized in Table 2, results indicated a significant association between Thal phase and group (χ ²(1) = 18.18, p < 0.001, φ= 0.33), with medium to large effect size, specifically that the sMCI sample included fewer clinically Thal cases than the dMCI group (Z = –4.26, p < 0.001). Group membership was also significantly associated with Braak stage (χ ²(1) = 21.40, p < 0.001, φ= 0.36), showing medium to large effect size, where the sMCI group had significantly fewer cases of clinically significant Braak pathology (Z = –4.63, p < 0.001). Group membership was significantly associated with CERAD neuritic plaque density (χ ²(1) = 23.80, p < 0.001, φ= 0.38), showing medium to large effect size, with fewer clinically significant CERAD cases in the sMCI group (Z = –4.88, p < 0.001). Lewy body ratings showed no significant relationship with group membership after adjusting for familywise error (p = 0.040, Fisher’s exact test, φ= 0.18), showing small to medium effect size. In the NACC data repository, TDP-43 became a variable of interest in later versions of neuropathology data collection, and therefore resulted in a significant amount of missing data (sMCI n = 15; dMCI = 68). With this reduced sample size, TDP-43 showed no significant relationship with group membership (p = 0.790, Fisher’s exact test, φ= 0.03). CAA burden was significantly higher in the dMCI group compared to the sMCI group (U = 1219.5, p = 0.001, η ² = 0.06), explain 6% of the total variance in CAA burden accounted for by group membership.

Table 2

Suspected etiologies of sMCI and dMCI

Hypothesized features contributing to MCI	sMCI (n = 28)	dMCI (n = 139)	Test statistic	p	Effect size
Neurodegenerative features (autopsy)
Thal Aβ plaques (% clinically significant)	46.43	83.45	Z = –4.26	< 0.001	φ= 0.33
Braak NFTs (% clinically significant)	29.63	75.18	Z = –4.63	< 0.001	φ= 0.36
CERAD neuritic plaques (% clinically significant)	32.14	78.26	Z = –4.88	< 0.001	φ= 0.38
Lewy bodies (% present in cortical regions)	21.42	42.72	Z = –2.07	0.040	φ= 0.18
TDP-43 (% present)^∗	6.67	8.82	Z = –0.27	0.790	φ= 0.03
CAA burden (score 0–3)	0.50	1.14	U = 1220	< 0.001	η ² = 0.06
Vasculopathies (autopsy)
Microinfarcts (% present)	50.00	25.36	Z = 2.61	0.009	φ= 0.20
Infarcts and lacunes (% present)	7.14	16.67	Z = –1.28	0.257	φ= 0.10
Hemorrhages and microbleeds (% present)	3.57	12.21	Z = –1.34	0.311	φ= 0.11

sMCI, stable mild cognitive impairment; dMCI, declined mild cognitive impairment; η ², Eta squared; φ, Phi; NFTs, neurofibrillary tangles; CERAD, Consortium to Establish a Registry for Alzheimer’s Disease; TDP-43, TAR DNA-binding protein 43; CAA, cerebral amyloid angiopathy; U, Mann-Whitney U. ^∗Sample size for TDP-43 is sMCI (n = 15), dMCI (n = 68).

Analysis of the relationship between CVD and group membership indicated no significant association between the ‘infarcts and lacunes’ variable and group (p = 0.257, Fisher’s exact test, φ= 0.10) or the ‘hemorrhages and microbleeds’ variable and group (p = 0.311, Fisher’s exact test, φ= 0.11). There was a significant association between microinfarcts and group membership (χ ²(1) = 6.79, p = 0.009, φ= 0.20), with greater frequency of clinically significant microinfarcts found in the sMCI group (Z = 2.61, p = 0.009).

Post hoc analyses

As stated above, age was significantly different between groups at death. Therefore, a post hoc analysis was conducted to control for the effect of age on all variables of interest. To do this, the interquartile range (IQR) for age at last visit was calculated in SPSS for the sMCI group. The IQR (= 12) was then subtracted from the sMCI median age (= 90), resulting in a lower limit age of 78 to be applied as an inclusion criterion for both groups in the post hoc analysis. This method was employed to preserve as many participants as possible in the sMCI group, with the understanding that more of the dMCI sample would be excluded from this post-hoc analysis. All analyses were rerun between the sMCI (n = 25) and the dMCI (n = 103) groups. The post hoc analyses showed similar results for all neurodegenerative variables (i.e., Thal phase, Braak Stage, CERAD neuritic plaque density, Lewy bodies, CAA, and TDP-43). The CVD variables, ‘infarcts and lacunes’ and ‘hemorrhages and microbleeds’ results were also similar; however, microinfarcts were no longer statistically significant between groups (χ ²(1) = 3.52, p = 0.061, φ= 0.17). TDP-43 sample was further reduced to sMCI (n = 12) and dMCI (n = 49) and remained non-significantly different between groups.

DISCUSSION

The present study sought to elucidate the neuropathological characteristics of a small sample of individuals with MCI who remained cognitively stable (sMCI) for at least five years, and up to nine years until death. In a preliminary test of the hypothesis that CVD may be a cause of sMCI, we quantified neurodegenerative and cerebrovascular brain pathologies at autopsy, and expected to find frequent CVD pathology and infrequent neurodegenerative pathology in this group. Our results indicate lower occurrences of AD pathology in sMCI, relative to individuals with worsening cognitive impairment (i.e., dMCI), which aligns with others’ previous findings of nonsignificant levels of AD biomarkers in sMCI using in vivo measuring methods [29, 30]. The presence of microinfarcts was also greater in the sMCI group, providing tentative support for our hypothesis, though our analytic approach was only descriptive and cannot allow us to draw direct causal interpretations. We discuss these results in detail below.

Neurodegenerative pathology in sMCI and dMCI

The current study’s prevalence estimate of sMCI is 16.7%. Although many individuals within this group will be assumed to decline to dementia, the presence of clinically significant AD pathologies (amyloid and tau) was substantially lower in this group than in individuals who did eventually decline. The presence of neurodegenerative disease was not nil, however—nearly half the sMCI sample had notable amyloid plaque buildup, and roughly one third had significant neurofibrillary tangles. A recent molecular imaging study using in vivo measures of amyloid brain pathology reported that 60% of MCI individuals who were amyloid-PET positive remained cognitively stable for an average of 39 months [77], suggesting that significant brain amyloid plaque deposition is not inextricably linked to decline. Presence of tau may be more detrimental to cognition than amyloid [78]. Blennow et al. [79] found that up to 20% of their tau-positive MCI sample continued to meet criteria for MCI five years into the study, but most declined to dementia in subsequent years. Therefore, it is conceivable that our subset of sMCI individuals with clinically significant tau pathology would have eventually developed dementia had they lived long enough, and did not represent true cases of sMCI.

CAA was also reduced in sMCI relative to dMCI in this sample. CAA impacts the vascular system and has historically been conceptualized primarily as a vasculopathy, but emerging evidence over the last decade suggests that it is associated with neurodegenerative brain atrophy and progressive cognitive decline [80]. Significantly greater CAA burden in dMCI compared to sMCI aligns with other authors’ suggestions that CAA is a cause of progressive cognitive decline and should be considered neurodegenerative in nature [80, 81], although this was not directly tested in this study as our analyses were descriptive in nature. Along with more frequent CAA and AD pathology, dMCI participants were also more likely to have at least one APOE ɛ4 allele. APOE ɛ4 status is known to be robustly associated with neurodegenerative pathology severity [82, 83], and has previously been linked to more severe CAA burden in individuals with comorbid AD pathology [84].

TDP-43, a neurodegenerative pathology that has also been linked to amyloid deposition and tau hyperphosphorylation [85], occurred at similarly low frequencies in both the sMCI and dMCI groups (6.7% and 8.8%, respectively). This result is likely a function of our reduced sample size, as only roughly half of this study’s participants had available TDP-43 data. More comprehensive analyses of the NACC dataset (n = 929) have revealed that TDP-43 pathology is associated with more severe Thal Aβ phase, Braak NFT stage, and CERAD neuritic plaques [86], and we expect we would have seen significantly more frequent TDP-43 in a larger dMCI sample.

Of our dMCI group, approximately 16% had clinically significant Lewy body pathology in the neocortex. Previous research suggests Lewy bodies occur comorbidly with AD pathology in approximately 41% –57% of cases [87 –89]; this discrepancy is due to our decision to dichotomize the variable into presence or absence of clinically significant Lewy body pathology (i.e., sufficient to have impact on cognition) [64], whereas other studies report on whole-brain pathology [87 –89]. The prevalence of whole-brain Lewy body pathology in the dMCI group was within previously reported ranges for cohorts with AD (41% [89]; data not shown).

Although frequencies of AD and vascular pathologies differed between sMCI and dMCI groups, there were no associations with MCI subtypes (i.e., amnestic versus non-amnestic, single versus multiple domain). This was rather surprising, because previous research suggests that neurodegeneration (i.e., dMCI) is most likely to begin as an amnestic clinical presentation [90], and that further decline is positively associated with number of impaired cognitive domains [91, 92]. On the other hand, single-domain non-amnestic presentations have been linked to non-degenerative, usually vascular, underlying etiologies with high likelihood of cognitive stability or reversion to normal (i.e., sMCI) [90]. In the current study, although there was no significant association between group and MCI subtype, the dMCI mostly consisted of multi-domain amnestic MCI (45%), while the sMCI group consisted mostly of single-domain amnestic MCI (64.3%), suggesting a possible trend of greater decline when multiple cognitive domains are impacted.

Cerebrovascular disease burden in sMCI and dMCI

Microinfarcts were the most commonly occurring neuropathology in our small sMCI sample (50%), and significantly more frequent than in dMCI (25%). Microinfarcts are commonly found in the brains of aging individuals [93], and are emerging as important correlates of vascular cognitive impairment based on results from the Medical Research Council Cognitive Function and Ageing Study [94]. In individuals with MCI or dementia, they can occur in isolation, though this is uncommon [93 , 96], or they can be associated with neurodegenerative pathologies such as CAA [97, 98]. In and of themselves, microinfarcts do not appear to be progressive in nature, but when present in sufficient numbers can cause clinically significant cognitive impairment [93], with greater number of microinfarcts being associated with greater cognitive impairment [99 –101]. Here, we propose that microinfarcts may be a silent manifestation of small vessel disease associated with seemingly stable cognitive impairment. Small vessel disease has been strongly linked to hypertension and hypercholesterolemia, both of which were also significantly elevated in our sMCI sample (though this was not true for diabetes, which is also a key risk factor for small vessel disease) [102]. Structural measures of white matter integrity (e.g., white matter hyperintensities [WMH]) would be valuable to corroborate this interpretation, as WMH have been proposed as a marker of more widespread microscopic vascular lesions [93]. Although structural neuroimaging is available in NACC, unfortunately only three aMCI participants had available WMH data in our sample.

Microinfarcts were no longer significantly associated with group membership after age-matching the sMCI and dMCI groups, suggesting that their presence was primarily driven by age and that they were especially common in the oldest participants, as prior work has shown [93]. sMCI participants were generally older at baseline and significantly older at death than those with dMCI, consistent with shorter life expectancy in dementia [103], and therefore had more opportunity to develop microinfarcts. Considered with previous reported associations between earlier age at symptom onset and more rapid cognitive decline [104, 105], it is possible that later onset of MCI might be suggestive of a more stable course potentially caused by silent small vessel disease. An empirical test of this possibility should involve quantification of small vessel disease (e.g., WMH) tracked prospectively in early- and late-onset cases of MCI with and without biomarker evidence of AD pathology. Some studies have been conducted to address the effect of small vessel disease on neurodegeneration (e.g., [106]) but do not usually parse out the effects of AD pathology from those of vasculopathy, which is important considering their comorbidity is common [31, 32] and seems to have a particularly deleterious effect on cognition [33 –35].

The variable ‘infarcts and lacunes’ was not significantly associated with group membership in this study, which was contrary to the hypothesis that cerebrovascular disease would be associated with sMCI. Lacunes and infarcts, although similar in pathophysiology, may differ in terms of severity, and therefore may have different levels of impact on cognition. While the range of impact on cognition overlaps in lacunes and in infarcts [107 –110], the range of severity is significantly larger for infarcts. It may be that the dMCI group have relatively more infarcts compared to the sMCI group, while the sMCI group shows relatively more lacunes, however this can unfortunately not be determined in the current study as both pathologies were pooled into a single variable in the NACC dataset.

The variable ‘hemorrhages and microbleeds’ was not associated with group membership, contrary to the hypothesis that vasculopathy would be more frequent in sMCI. Hemorrhages, in this context, refer to intracerebral hemorrhage (ICH). Microbleeds and ICHs may occur due to CAA [110, 111] but can also occur due to hypertensive vasculopathy (i.e., in the absence of CAA) [112]. These two causes can be differentiated based on location. Specifically, CAA related microbleeds and ICH occur in the cerebral lobes, while microbleeds and ICH due to hypertensive vasculopathy are seen in the brain stem and deep brain structures [98 , 114]. Further, the impact on cognition appears to differ between the two etiologies. In CAA, microbleeds and ICH contribute to a progressive decline in cognition [110, 111], but when due to hypertensive vasculopathy, cognition remains stable [112]. Thus, the current study’s results showing no association between group membership and ‘hemorrhages and microbleeds’ may be explained by the presence of microbleeds and ICHs in both groups due to distinct causes. It is possible that cognitive decline in dMCI and stable cognitive deficits in sMCI are respectively associated with microbleeds caused by CAA and microbleeds caused by hypertensive vasculopathy; NACC’s pooled variables unfortunately do not allow us to answer this question directly.

Limitations

This study provides descriptions of postmortem brain pathologies that may have contributed to MCI stability during life, using data collected across multiple sites in the USA. A significant limitation to this work is that, as a construct, MCI is not well defined; there is no consensus as to which neuropsychological measures should be used to operationalize cognitive performance, or what cut-offs should be used to define impairment. Consequently, individual ADCs in this study collected data according to their own protocols, and definitions of MCI may have varied from one site to the next. The neuropsychological measures used, number of tests administered, and cut-off scores to operationalize objective cognitive impairment were not clearly defined, and likely differed between sites. As such, the definition of MCI used in the current study is broad, likely resulting in a wide range of severity of cognitive impairment in the sample. Further, we included any participant classified as MCI at first and last visits, with no consideration of ‘in-between’ visits where cognition returned to normal, which occurred in roughly two-thirds of the sample. Pooling stable and fluctuating cases in this way also contributed to our sample’s heterogeneity.

In addition, NACC’s current classifications of CVD are crude and provide less than ideal estimates of CVD. Due to changes in the data curation process since NACC’s inception, the CVD data that were available for the purposes of our analyses included pooled variables (‘microbleeds and hemorrhages’ and ‘lacunes and infarcts’). This confounded aspects of our interpretation, as we were unable to parse out the effects of large cortical infarcts or lobal hemorrhages on cognitive impairment. Similarly, recent/active and remote/inactive in vivo vascular risk factors were pooled together to allow a sufficiently large sample for statistical analyses, but this precluded us from being able to investigate separate associations. While midlife vascular risk factors have been robustly associated with later-life cognitive decline, there is currently a lack of consensus as to the role of later-life vascular risk in brain health [56]. Their individual contributions unfortunately could not be addressed here. Lastly, the current study’s sample was predominantly White. A more diverse sample would increase the generalizability of the results and provide further information on stability/conversion in different ethnicities. Extending MCI and dementia research efforts to racially diverse samples has been a specific recommendation of the NIA-AA [115].

CONCLUSION

In conclusion, our findings describe microinfarcts as the primary neuropathology in individuals with ≥5-year stable MCI, and tentatively suggest that small vessel disease may characterize non-progressive cognitive impairment in older adults. Future work should explore vascular pathology at varying degrees of severity in sMCI (e.g., lacunes versus larger infarcts).

Footnotes

ACKNOWLEDGMENTS

Dr. Callahan is supported through a Canada Research Chair. The NACC database is funded by NIA/NIH Grant U01 AG016976. NACC data are contributed by the NIA-funded ADCs: P30 AG019610 (PI Eric Reiman, MD), P30 AG013846 (PI Neil Kowall, MD), P50 AG008702 (PI Scott Small, MD), P50 AG025688 (PI Allan Levey, MD, PhD), P50 AG047266 (PI Todd Golde, MD, PhD), P30 AG010133 (PI Andrew Saykin, PsyD), P50 AG005146 (PI Marilyn Albert, PhD), P50 AG005134 (PI Bradley Hyman, MD, PhD), P50 AG016574 (PI Ronald Petersen, MD, PhD), P50 AG005138 (PI Mary Sano, PhD), P30 AG008051 (PI Thomas Wisniewski, MD), P30 AG013854 (PI M. Marsel Mesulam, MD), P30 AG008017 (PI Jeffrey Kaye, MD), P30 AG010161 (PI David Bennett, MD), P50 AG047366 (PI Victor Henderson, MD, MS), P30 AG010129 (PI Charles DeCarli, MD), P50 AG016573 (PI Frank LaFerla, PhD), P50 AG005131 (PI James Brewer, MD, PhD), P50 AG023501 (PI Bruce Miller, MD), P30 AG035982 (PI Russell Swerdlow, MD), P30 AG028383 (PI Linda Van Eldik, PhD), P30 AG053760 (PI Henry Paulson, MD, PhD), P30 AG010124 (PI John Trojanowski, MD, PhD), P50 AG005133 (PI Oscar Lopez, MD), P50 AG005142 (PI Helena Chui, MD), P30 AG012300 (PI Roger Rosenberg, MD), P30 AG049638 (PI Suzanne Craft, PhD), P50 AG005136 (PI Thomas Grabowski, MD), P50 AG033514 (PI Sanjay Asthana, MD, FRCP), P50 AG005681 (PI John Morris, MD), P50 AG047270 (PI Stephen Strittmatter, MD, PhD).

Authors’ disclosures available online ().

References

Petersen

(2004) Mild cognitive impairment as a diagnostic entity. J Intern Med 256, 183–194.

Mitchell

, Shiri-Feshki

(2009) Rate of progression of mild cognitive impairment to dementia - meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 119, 252–265.

Buchhave

, Minthon

, Zetterberg

, Wallin

ÅK

, Blennow

, Hansson

(2012) Cerebrospinal fluid levels of β-amyloid 1-42, but not of tau, are fully changed already 5 to 10 years before the onset of Alzheimer dementia. Arch Gen Psychiatry 69, 98–106.

Busse

, Hensel

, Guhne

, Angermeyer

, Riedel-Heller

(2006) Mild cognitive impairment Long-term course of four clinical subtypes. Neurology 67, 2176–2185.

Ganguli

, Dodge

, Shen

, DeKosky

(2004) Mild cognitive impairment, amnestic type: An epidemiologic study. Neurology 63, 115–121.

Alves

, Cardoso

, Maroco

, De Mendonça

, Guerreiro

, Silva

(2018) Neuropsychological predictors of long-term (10 years) mild cognitive impairment stability. J Alzheimers Dis 62, 1703–1711.

, Farnum

, Yang

, Verbeeck

, Lobanov

, Raghavan

, Novak

, DiBernardo

, Narayan

, Alzheimer’s Disease Neuroimaging Initiative (2012) Sparse learning and stability selection for predicting MCI to AD conversion using baseline ADNI data. BMC Neurol 12, 46.

Farias

, Mungas

, Reed

, Harvey

, DeCarli

(2009) Progression of mild cognitive impairment to dementia in clinic- vs community-based cohorts. Arch Neurol 66, 1151–1157.

Nowrangi

, Rosenberg

, Leoutsakos

J-MS

(2016) Subtle changes in daily functioning predict conversion from normal to mild cognitive impairment or dementia: An analysis of the NACC database. Int Psychogeriatr 28, 2009–2018.

10.

Gómez-Tortosa

, Mahillo-Fernández

, Guerrero

, Montoya

, Alonso

, Sainz

(2012) Outcome of mild cognitive impairment comparing early memory profiles. Am J Geriatr Psychiatry 20, 827–835.

11.

Perri

, Serra

, Carlesimo

, Caltagirone

(2007) Preclinical dementia: An Italian multicentre study on amnestic mild cognitive impairment. Dement Geriatr Cogn Disord 23, 289–300.

12.

Ellendt

, Voβ

, Kohn

, Wagels

, Goerlich

, Drexler

, Schneider

, Habel

(2017) Predicting stability of mild cognitive impairment (MCI): Findings of a community based sample. Curr Alzheimer Res 14, 608–619.

13.

Clem

, Holliday

, Pandya

, Hynan

, Lacritz

, Woon

(2017) Predictors that a diagnosis of mild cognitive impairment will remain stable 3 years later. Cogn Behav Neurol 30, 8–15.

14.

Summers

, Saunders

(2012) Neuropsychological measures predict decline to Alzheimer’s dementia from mild cognitive impairment. In Neuropsychology, School of Psychology, University of Tasmania, Tasmania, Australia, and Wicking Dementia Research and Education Centre, Hobart, Tasmania, Australia, pp. 498-508.

15.

Callahan

, Ramirez

, Berezuk

, Duchesne

, Black

(2015) Predicting Alzheimer’s disease development: A comparison of cognitive criteria and associated neuroimaging biomarkers. Alzheimers Res Ther 7, 68.

16.

Michaud

, Su

, Siahpush

, Murman

(2017) The risk of incident mild cognitive impairment and progression to dementia considering mild cognitive impairment subtypes. Dement Geriatr Cogn Dis Extra 7, 15–29.

17.

Cooper

, Sommerlad

, Lyketsos

, Livingston

(2015) Modifiable predictors of dementia in mild cognitive impairment: A systematic review and meta-analysis. Am J Psychiatry 172, 323–334.

18.

Pal

, Mukadam

, Petersen

, Cooper

(2018) Mild cognitive impairment and progression to dementia in people with diabetes, prediabetes and metabolic syndrome: A systematic review and meta-analysis. Soc Psychiatry Psychiatr Epidemiol 53, 1149–1160.

19.

, Wu

, Miao

, Xiao

, Zhang

, Huang

(2015) Conversion of mild cognitive impairment to dementia among subjects with diabetes: A population-based study of incidence and risk factors with five years of follow-up. J Alzheimers Dis 43, 1441–1449.

20.

, Caracciolo

, Wang

H-X

, Winblad

, Bäckman

, Qiu

, Fratiglioni

(2010) Accelerated progression from mild cognitive impairment to dementia in people with diabetes. Diabetes 59, 2928–2935.

21.

Sugarman

, Alosco

, Tripodis

, Steinberg

, Stern

(2018) Neuropsychiatric symptoms and the diagnostic stability of mild cognitive impairment. J Alzheimers Dis 62, 1841–1855.

22.

Korolev

, Symonds

, Bozoki

, Initiative

ADN

(2016) Predicting progression from mild cognitive impairment to Alzheimer’s dementia using clinical, MRI, and plasma biomarkers via probabilistic pattern classification. PLoS One 11, e0138866.

23.

Borgio

JGF

, Baldaçara

, Moraes W dos

, Lacerda

ALT

, Montaño

, Jackowski

, Tufik

, Ramos

, Bressan

(2012) Hippocampal volume and CDR-SB can predict conversion to dementia in MCI patients. Arq Neuropsiquiatr 70, 839–842.

24.

Moscoso

, Silva-Rodríguez

, Aldrey

, Cortés

, Fernández-Ferreiro

, Gómez-Lado

, Ruibal

, Aguiar

(2019) Prediction of Alzheimer’s disease dementia with MRI beyond the short-term: Implications for the design of predictive models. Neuroimage Clin 23, 101837.

25.

Eckerström

, Olsson

, Borga

, Ekholm

, Ribbelin

, Rolstad

, Starck

, Edman

, Wallin

, Malmgren

(2008) Small baseline volume of left hippocampus is associated with subsequent conversion of MCI into dementia: The Göteborg MCI study. J Neurol Sci 272, 48–59.

26.

Devanand

, Pradhaban

, Liu

, Khandji

, De Santi

, Segal

, Rusinek

, Pelton

, Honig

, Mayeux

, Stern

, Tabert

, de Leon

(2007) Hippocampal and entorhinal atrophy in mild cognitive impairment. Neurology 68, 828–836.

27.

Apostolova

, Dutton

, Dinov

, Hayashi

, Toga

, Cummings

, Thompson

(2006) Conversion of mild cognitive impairment to Alzheimer disease predicted by hippocampal atrophy maps. Arch Neurol 63, 693–699.

28.

Jack

, Petersen

, Xu

, O’Brien

, Smith

, Ivnik

, Boeve

, Waring

, Tangalos

, Kokmen

(1999) Prediction of AD with MRI-based hippocampal volume in mild cognitive impairment. Neurology 52, 1397–1403.

29.

Parnetti

, Chiasserini

, Eusebi

, Giannandrea

, Bellomo

(2014) Performance of Aβ1-40, Aβ1-42, total tau, and phosphorylated tau as predictors of dementia in a cohort of patients with mild cognitive impairment. J Alzheimers Dis 29, 229–238.

30.

Cerami

, Dodich

, Iannaccone

, Magnani

, Santangelo

, Presotto

, Marcone

, Gianolli

, Cappa

, Perani

(2018) A biomarker study in long-lasting amnestic mild cognitive impairment. Alzheimer Res Ther 10, 42.

31.

Lee

, Seo

, Yang

, Jang

, Lee

, Kim

, Chin

, Lee

, Kim

, Lee

, Kim

, Yoo

, Lee

, Na

, Kim

(2018) Longitudinal cortical thinning and cognitive decline in patients with early- versus late-stage subcortical vascular mild cognitive impairment. Eur J Neurol 25, 326–333.

32.

Santos

, Snyder

, Wu

, Zhang

, Echeverria

, Alber

(2017) Pathophysiologic relationship between Alzheimer’s disease, cerebrovascular disease, and cardiovascular risk: A review and synthesis. Alzheimers Dement (Amst) 7, 69–87.

33.

Kalaria

, Akinyemi

, Ihara

(2012) Does vascular pathology contribute to Alzheimer changes? J Neurol Sci 322, 141–147.

34.

Attems

, Jellinger

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

35.

Gorelick

, Scuteri

, Black

, Decarli

, Greenberg

, Iadecola

, Launer

, Laurent

, Lopez

, Nyenhuis

, Petersen

, Schneider

, Tzourio

, Arnett

, Bennett

, Chui

, Higashida

, Lindquist

, Nilsson

, Roman

, Sellke

, Seshadri

(2011) Vascular contributions to cognitive impairment and dementia: A statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 42, 2672–713.

36.

van der Flier

, van Straaten

ECW

, Barkhof

, Verdelho

, Madureira

, Pantoni

, Inzitari

, Erkinjuntti

, Crisby

, Waldemar

, Schmidt

, Fazekas

, Scheltens

(2005) Small vessel disease and general cognitive function in nondisabled elderly: The LADIS study. Stroke 36, 2116–2120.

37.

Elahi

, Miller

(2017) A clinicopathological approach to the diagnosis of dementia. Nat Rev Neurol 13, 457–476.

38.

Okonkwo

, Cohen

, Gunstad

, Tremont

, Alosco

, Poppas

(2010) Longitudinal trajectories of cognitive decline among older adults with cardiovascular disease. Cerebrovasc Dis 30, 362–373.

39.

Harrison

, Tang

EYH

, Keage

HAD

, Taylor

J-P

, Allan

, Robinson

, Jagger

, Rockwood

, Stephan

BCM

(2016) A systematic review of the definitions of vascular cognitive impairment, no dementia in cohort studies. Dement Geriatr Cogn Disord 42, 69–79.

40.

Ingles

, Wentzel

, Fisk

, Rockwood

(2002) Neuropsychological predictors of incident dementia in patients with vascular cognitive impairment, without dementia. Stroke 33, 1999–2002.

41.

Hsiung

, Donald

, Grand

, Black

, Bouchard

, Gauthier

, Loy-English

, Hogan

, Kertesz

, Rockwood

, Feldman

(2006) Outcomes of cognitively impaired not demented at 2 years in the Canadian Cohort Study of Cognitive Impairment and Related Dementias. Dement Geriatr Cogn Disord 22, 413–420.

42.

Sachdev

, Chen

, Brodaty

, Thompson

, Altendorf

, Wen

(2009) The determinants and longitudinal course of post-stroke mild cognitive impairment. J Int Neuropsychol Soc 15, 915–923.

43.

Meguro

, Akanuma

, Meguro

, Kasai

, Ishii

, Yamaguchi

(2012) Prognosis of vascular mild cognitive impairment includes vascular dementia onset and death by cardiovascular disease: Reanalysis from the Osaki-Tajiri project. J Stroke Cerebrovasc Dis 21, 607–611.

44.

Chung

, Pollock

, Campbell

, Durward

, Hagen

(2013) Cognitive rehabilitation for executive dysfunction in adults with stroke or other adult non-progressive acquired brain damage. Cochrane Database Syst Rev, CD008391.

45.

Mijajlović

, Pavlović

, Brainin

, Heiss

, Quinn

, Ihle-Hansen

, Hermann

, Assayag Ben

, Richard

, Thiel

, Kliper

, Shin Il

, Kim

, Choi

, Jung

, Lee

, Sinanović

, Levine

, Schlesinger

, Mead

, Milošević

, Leys

, Hagberg

, Ursin

, Teuschl

, Prokopenko

, Mozheyko

, Bezdenezhnykh

, Matz

, Aleksić

, Muresanu

, Korczyn

, Bornstein

(2017) Post-stroke dementia - a comprehensive review. BMC Med 15, 1–12.

46.

Nair

, Lincoln

(2007) Cognitive rehabilitation for memory deficits following stroke. Cochrane Database Syst Rev, CD002293

47.

Jack

, Knopman

, Jagust

, Petersen

, Weiner

, Aisen

, Shaw

, Vemuri

, Wiste

, Weigand

, Lesnick

, Pankratz

, Donohue

, Trojanowski

(2013) Tracking pathophysiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers. Lancet Neurol 12, 207–216.

48.

Clerici

, Caracciolo

, Cova

, Fusari Imperatori

, Maggiore

, Galimberti

, Scarpini

, Mariani

, Fratiglioni

(2012) Does vascular burden contribute to the progression of mild cognitive impairment to dementia? Dement Geriatr Cogn Disord 34, 235–243.

49.

de la Torre

(2010) Vascular risk factor detection and control may prevent Alzheimer’s disease. Ageing Res Rev 9, 218–225.

50.

de la Torre

(2004) Is Alzheimer’s disease a neurodegenerative or a vascular disorder? Data, dogma, and dialectics. Lancet Neurol 3, 184–190.

51.

Brennan

, Crispo

, Zaridze

, Szeszenia-Dabrowska

, Rudnai

, Lissowska

, Fabiánová

, Mates

, Bencko

, Foretova

, Janout

, Fletcher

, Boffetta

(2006) High cumulative risk of lung cancer death among smokers and nonsmokers in Central and Eastern Europe. Am J Epidemiol 164, 1233–1241.

52.

US Department of Health and Human Services (2014) The health consequences of smoking - 50 years of progress: A report of the Surgeon General. Center for Disease Control and Prevention.

53.

Besser

, Kukull

, Knopman

, Chui

, Galasko

, Weintraub

, Jicha

, Carlsson

, Burns

, Quinn

, Sweet

, Rascovsky

, Teylan

, Beekly

, Thomas

, Bollenbeck

, Monsell

, Mock

, Zhou

, Thomas

, Robichaud

, Dean

, Hubbard

, Jacka

, Schwabe-fry

, Wu

, Phelps

, Morris

, Neuropsychology Work Group, Directors, and Clinical Core leaders of the National Institute on Aging-funded US Alzheimer’s Disease (2018) Version 3 of the National Alzheimer’s Coordinating Center’s Uniform Data Set. Alzheimer Dis Assoc Disord 32, 351–358.

54.

Weintraub

, Besser

, Dodge

, Teylan

, Ferris

, Goldstein

, Giordani

, Kramer

, Loewenstein

, Marson

, Mungas

, Salmon

, Welsh-bohmer

, Zhou

, Shirk

, Atri

, Kukull

, Phelps

, Morris

(2018) Version 3 of the Alzheimer Disease Centers’ Neuropsychological Test Battery in the Uniform Data Set (UDS). Alzheimer Dis Assoc Disord 32, 10–17.

55.

Petersen

(2004) Mild cognitive impairment. Continuum (N Y) 57, 869–874.

56.

Iadecola

, Yaffe

, Biller

, Bratzke

, Faraci

, Gorelick

, Gulati

, Kamel

, Knopman

, Launer

, Saczynski

, Seshadri

, Zeki Al Hazzouri

, American Heart Association Council on Hypertension; Council on Clinical Cardiology; Council on Cardiovascular Disease in the Young; Council on Cardiovascular and Stroke Nursing; Council on Quality of Care and Outcomes Research; and Stroke Council (2016) Impact of hypertension on cognitive function: A scientific statement from the American Heart Association. Hypertension 68, e67–e94.

57.

Overton

, Pihlsgård

, Elmståhl

(2019) Diagnostic stability of mild cognitive impairment, and predictors of reversion to normal cognitive functioning. Dement Geriatr Cogn Disord 48, 317–329.

58.

Koepsell

, Monsell

(2012) Reversion from mild cognitive impairment to normal or near-normal cognition: Risk factors and prognosis. Neurology 79, 1591–1598.

59.

Liu

, Kanekiyo

, Xu

, Bu

(2013) Apolipoprotein E and Alzheimer disease: Risk, mechanisms and therapy. Nat Rev Neurol 9, 106–118.

60.

Tsapanou

, Scarmeas

, Gu

, Manly

, Schupf

, Stern

, Barral

(2015) Examining the association between Apolipoprotein E (APOE) and self-reported sleep disturbances in non-demented older adults. Neurosci Lett 606, 72–76.

61.

Braak

, Alafuzo

, Arzberger

, Kretzschmar

, Tredici

K Del

(2006) Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry. Acta Neuropathol 112, 389–404.

62.

Thal

, Rüb

, Orantes

, Braak

(2002) Phases of A[beta]-deposition in the human brain and its relevance for the development of AD. Neurology 58, 1791–1800.

63.

Mirra

, Heyman

, McKeel

, Sumi

, Crain

, Brownlee

, Vogel

, Hughes

, van Belle

, Berg

(1991) The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part II. Standardization of the neuropathologic assessment of Alzheimer’s disease. Neurology 41, 479–86.

64.

Montine

, Phelps

, Beach

, Bigio

, Cairns

, Dickson

, Duyckaerts

, Frosch

, Masliah

, Mirra

, Nelson

, Schneider

, Thal

, Trojanowski

, Vinters

H V

, Hyman

(2012) National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: A practical approach. Acta Neuropathol 123, 1–11.

65.

Boluda

, Toledo

, Irwin

, Raible

, Byrne

, Lee

VMY

, Trojanowski

(2014) A comparison of Aβ amyloid pathology staging systems and correlation with clinical diagnosis. Acta Neuropathol 128, 543–550.

66.

Hyman

, Phelps

, Beach

, Bigio

, Cairns

, Carrillo

, Dickson

, Duyckaerts

, Frosch

, Masliah

, Mirra

, Nelson

, Schneider

, Thal

, Thies

, Trojanowski

, Vinters

H V

, Montine

(2012) National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease. Alzheimers Dement 8, 1–13.

67.

Dallaire-Théroux

, Beheshti

, Potvin

, Dieumegarde

, Saikali

, Duchesne

(2019) Braak neurofibrillary tangle staging prediction from in vivo MRI metrics. Alzheimers Dement (Amst) 11, 599–609.

68.

McKeith

, Dickson

, Lowe

, Emre

, O’Brien

, Feldman

, Cummings

, Duda

, Lippa

, Perry

, Aarsland

, Arai

, Ballard

, Boeve

, Burn

, Costa

, Del Ser

, Dubois

, Galasko

, Gauthier

, Goetz

, Gomez-Tortosa

, Halliday

, Hansen

, Hardy

, Iwatsubo

, Kalaria

, Kaufer

, Kenny

, Korczyn

, Kosaka

, Lee

VMY

, Lees

, Litvan

, Londos

, Lopez

, Minoshima

, Mizuno

, Molina

, Mukaetova-Ladinska

, Pasquier

, Perry

, Schulz

, Trojanowski

, Yamada

, Consortium on DLB (2005) Diagnosis and management of dementia with Lewy bodies: Third report of the DLB consortium. Neurology 65, 1863–1872.

69.

Vonsattel

, Myers

, Hedley-Whyte

, Ropper a

, Bird

, Richardson

(1991) Cerebral amyloid angiopathy without and with cerebral hemorrhages: A comparative histological study. Ann Neurol 30, 637–649.

70.

Olichney

, Hansen

, Hofstetter

, Lee

, Katzman

, Thal

(2000) Association between severe cerebral amyloid angiopathy and cerebrovascular lesions in Alzheimer disease is not a spurious one attributable to apolipoprotein E4. Arch Neurol 57, 869–874.

71.

Wardlaw

, Smith

, Biessels

, Cordonnier

, Fazekas

, Frayne

, Lindley

, O’Brien

, Barkhof

, Benavente

, Black

, Brayne

, Breteler

, Chabriat

, DeCarli

, de Leeuw

, Doubal

, Duering

, Fox

, Greenberg

, Hachinski

, Kilimann

, Mok

, Oostenbrugge

, Pantoni

, Speck

, Stephan

, Teipel

, Viswanathan

, Werring

, Chen

, Smith

, van

, Norrving

, Gorelick

, Dichgans

(2013) Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 12, 822–838.

72.

Benjamini

, Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Ser B 57, 289–300.

73.

McDonald

(2014) Multiple comparisons. In Handbook of Biological Statistics (3rd ed.). Sparky House Publishing, Baltimore, Baltimore, Maryland, pp. 254–260.

74.

Cohen

(1988) Statistical power analysis for the behavioral sciences. Lawrence Erlbaum Associates, Hillsdale, New Jersey, pp. 273–406.

75.

Cohen

(1973) Eta-squared and partial eta-squared in fixed factor Anova designs. Educ Psychol Meas 33, 107–112.

76.

Cohen

, Cohen

, West

, Aiken

(2013) Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences, Lawrence Erlbaum Associates, Routledge.

77.

Iaccarino

, Sala

, Perani

, Initiative

the ADN

(2019) Predicting long-term clinical stability in amyloid-positive subjects by FDG-PET. Ann Clin Transl Neurol 6, 1113–1120.

78.

Betthauser

, Koscik

, Jonaitis

, Allison

, Cody

, Erickson

, Rowley

, Stone

, Mueller

, Clark

, Carlsson

, Chin

, Asthana

, Christian

, Johnson

(2020) Amyloid and tau imaging biomarkers explain cognitive decline from late middle-age. Brain 143, 320–335.

79.

Blennow

, Shaw

, Stomrud

, Mattsson

, Toledo

, Buck

, Wahl

, Eichenlaub

, Lifke

, Simon

, Trojanowski

, Hansson

(2019) Predicting clinical decline and conversion to Alzheimer’s disease or dementia using novel Elecsys Aβ(1-42), pTau and tTau CSF immunoassays. Sci Rep 9, 19024.

80.

Smith

(2018) Cerebral amyloid angiopathy as a cause of neurodegeneration. J Neurochem 144, 651–658.

81.

Wermer

MJH

, Greenberg

(2018) The growing clinical spectrum of cerebral amyloid angiopathy. Curr Opin Neurol 31, 28–35.

82.

Polvikoski

, Sulkava

, Haltia

, Kainulainen

, Vuorio

, Verkkoniemi

, Niinisto

, Halonen

, Kontula

(1995) Apolipoprotein E, dementia, and cortical deposition of β-amyloid protein. N Engl J Med 333, 1242–1247.

83.

Farfel

, Yu

, De Jager

, Schneider

, Bennett

(2016) Association of APOE with tau-tangle pathology with and without β-amyloid. Neurobiol Aging 37, 19–25.

84.

, Boyle

, Nag

, Leurgans

, Buchman

, Wilson

, Arvanitakis

, Farfel

, De Jager

, Bennett

, Schneider

(2015) APOE and cerebral amyloid angiopathy in community-dwelling older persons. Neurobiol Aging 36, 2946–2953.

85.

Huang

, Zhou

, Tu

, Ba

, Huang

, Luo

(2020) TDP-43: From Alzheimer’s disease to limbic-predominant age-related TDP-43 encephalopathy. Front Mol Neurosci 13, 26.

86.

Katsumata

, Fardo

, Kukull

, Nelson

(2018) Dichotomous scoring of TDP-43 proteinopathy from specific brain regions in 27 academic research centers: Associations with Alzheimer’s disease and cerebrovascular disease pathologies. Acta Neuropathol Commun 6, 142.

87.

Hamilton

(2000) Lewy bodies in Alzheimer’s Disease: A neuropathological review of 145 cases using α-synuclein immunohistochemistry. Brain Pathol 10, 378–384.

88.

Uchikado

, Lin

, Delucia

, Dickson

(2006) Alzheimer disease with amygdala Lewy bodies: A distinct form of α-synucleinopathy. J Neuropathol Exp Neurol 65, 685–697.

89.

Robinson

, Lee

, Xie

, Rennert

, Suh

, Bredenberg

, Caswell

, Van Deerlin

, Yan

, Yousef

, Hurtig

, Siderowf

, Grossman

, McMillan

, Miller

, Duda

, Irwin

, Wolk

, Elman

, McCluskey

, Chen-Plotkin

, Weintraub

, Arnold

, Brettschneider

, Lee

VMY

, Trojanowski

(2018) Neurodegenerative disease concomitant proteinopathies are prevalent, age-related and APOE4-associated. Brain 141, 2181–2193.

90.

Petersen

(2016) Mild cognitive impairment. Continuum (Minneap Minn) 22, 404–418.

91.

Han

, Kim

, Lee

, Park

, Lee

, Huh

, Park

, Jhoo

, Lee

, Kim

(2012) Predictive validity and diagnostic stability of mild cognitive impairment subtypes. Alzheimers Dement 8, 553–559.

92.

Hughes

, Snitz

, Ganguli

(2011) Should mild cognitive impairment be subtyped? Curr Opin Psychiatry 24, 237–242.

93.

Smith

, Schneider

, Wardlaw

, Greenberg

(2012) Cerebral microinfarcts: The invisible lesions. Lancet Neurol 11, 272–282.

94.

Richardson

, Stephan

, Ince

, Brayne

, Matthews

, Esiri

(2012) The neuropathology of vascular disease in the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Curr Alzheimer Res 9, 687–696.

95.

Schneider

, Arvanitakis

, Leurgans

, Bennett

(2009) The neuropathology of probable Alzheimer’s disease and mild cognitive impairment. Ann Neurol 66, 200–208.

96.

White

, Petrovitch

, Hardman

, Nelson

, Davis

, Ross

, Masaki

, Launer

, Markesbery

(2009) Cerebrovascular pathology and dementia in autopsied Honolulu-Asia Aging Study participants. Ann N Y Acad Sci 977, 9–23.

97.

Niwa

, Ii

, Shindo

, Matsuo

, Ishikawa

, Taniguchi

, Takase

, Maeda

, Sakuma

, Akatsu

, Hashizume

, Tomimoto

(2017) Comparative analysis of cortical microinfarcts and microbleeds using 3.0-Tesla postmortem magnetic resonance images and histopathology. J Alzheimers Dis 59, 951–959.

98.

, Maeda

, Ishikawa

, Ito

, Matsuo

, Umino

, Shindo

, Kida

, Satoh

, Niwa

, Taniguchi

, Tomimoto

(2019) Cortical microinfarcts in patients with multiple lobar microbleeds on 3 T MRI. J Neurol 266, 1887–1896.

99.

Arvanitakis

, Leurgans

, Barnes

, Bennett

, Schneider

(2011) Microinfarct pathology, dementia, and cognitive systems. Stroke 42, 722–727.

100.

Kapasi

, Leurgans

, James

, Boyle

, Arvanitakis

, Nag

, Bennett

, Buchman

, Schneider

(2018) Watershed microinfarct pathology and cognition in older persons. Neurobiol Aging 70, 10–17.

101.

Sonnen

, Larson

, Crane

, Haneuse

, Li

, Schellenberg

, Craft

, Leverenz

, Montine

(2007) Pathological correlates of dementia in a longitudinal, population-based sample of aging. Ann Neurol 62, 406–413.

102.

Shi

, Wardlaw

(2016) Update on cerebral small vessel disease: A dynamic whole-brain disease. Stroke Vasc Neurol 1, 83–92.

103.

Todd

, Barr

, Roberts

, Passmore

(2013) Survival in dementia and predictors of mortality: A review. Int J Geriatr Psychiatry 28, 1109–1124.

104.

Jacobs

, Sano

, Marder

, Bell

, Bylsma

, Lafleche

, Albert

, Brandt

, Stern

(1994) Age at onset of Alzheimer’s disease: Relation to pattern of cognitive dysfunction and rate of decline. Neurology 44, 1215–1220.

105.

Van Der Vlies

, Koedam

ELGE

, Pijnenburg

YAL

, Twisk

JWR

, Scheltens

, Van Der Flier

(2009) Most rapid cognitive decline in APOE 4 negative Alzheimer’s disease with early onset. Psychol Med 39, 1907–1911.

106.

Tosto

, Zimmerman

, Hamilton

, Carmichael

, Brickman

(2015) The effect of white matter hyperintensities on neurodegeneration in mild cognitive impairment. Alzheimers Dement 11, 1510–1519.

107.

Pantoni

(2010) Cerebral small vessel disease: From pathogenesis and clinical characteristics to therapeutic challenges. Lancet Neurol 9, 689–701.

108.

Benisty

, Gouw

, Porcher

, Madureira

, Hernandez

, Poggesi

, van der Flier

, van Straaten

, Verdelho

, Ferro

, Pantoni

, Inzitari

, Barkhof

, Fazekas

, Chabriat

(2009) Location of lacunar infarcts correlates with cognition in a sample of non-disabled subjects with age-related white-matter changes: The LADIS study. J Neurol Neurosurg Psychiatry 80, 478–483.

109.

Cannistraro

, Badi

, Eidelman

, Dickson

, Middlebrooks

, Meschia

(2019) CNS small vessel disease: A clinical review. Neurology 92, 1146–1156.

110.

Sachdev

, Kalaria

, O’Brien

, Skoog

, Alladi

, Black

, Blacker

, Blazer

, Chen

, Chui

, Ganguli

, Jellinger

, Jeste

D V

, Pasquier

, Paulsen

, Prins

, Rockwood

, Roman

, Scheltens

(2014) Diagnostic criteria for vascular cognitive disorders: A VASCOG statement. Alzheimer Dis Assoc Disord 28, 206–218.

111.

Caceres

, Goldstein

(2012) Intracranial hemorrhage. Emerg Med Clin North Am 30, 771–794.

112.

Xiong

, Reijmer

, Charidimou

, Cordonnier

, Viswanathan

(2016) Intracerebral hemorrhage and cognitive impairment. Biochim Biophys Acta 1862, 939–944.

113.

Goos

, Kester

, Barkhof

, Klein

, Blankenstein

, Scheltens

, van der Flier

(2009) Patients with Alzheimer disease with multiple microbleeds: Relation with cerebrospinal fluid biomarkers and cognition. Stroke 40, 3455–3460.

114.

Cordonnier

, Al-Shahi

, Wardlaw

(2007) Spontaneous brain microbleeds: Systematic review, subgroup analyses and standards for study design and reporting. Brain 130, 1988–2003.

115.

Jack

CRJ

, Bennett

, Blennow

, Carrillo

, Dunn

, Haeberlein

, Holtzman

, Jagust

, Jessen

, Karlawish

, Liu

, Molinuevo

, Montine

, Phelps

, Rankin

, Rowe

, Scheltens

, Siemers

, Snyder

, Sperling

Contributors (2018) NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement 14, 535–562.