Skip to content

Mixed pathology and Alzheimer overlap

TL;DR — Vascular and Alzheimer pathology frequently coexist, especially in older community populations, so “pure” categories are selected ends of a continuum rather than the modal autopsy finding (Schneider 2007, PMID 17568013). In one community autopsy series, 38.0% of dementia cases had Alzheimer pathology plus infarcts, versus 30.0% with Alzheimer pathology alone and 12% with vascular dementia alone (Schneider 2007, PMID 17568013). Vascular injury may add to, interact with, or lower reserve against amyloid/tau pathology; routine clinical data rarely identify proportional contribution with precision (Jellinger 2007, PMID 17324442). This page defines the shared border; Alzheimer-specific biomarkers and therapies remain in Alzheimer's disease.

Why the binary fails

Binary assumption Evidence-based replacement
One dementia, one cause Multiple lesions commonly contribute (Schneider 2007, PMID 17568013)
Amnesia means Alzheimer strategic vascular injury and mixed disease can be amnestic
Executive slowing means vascular phenotype is not pathology-specific (Desmond 2004, PMID 15537510)
WMH prove vascular dementia WMH are common and nonspecific (Duering 2023, PMID 37236211)
Positive amyloid biomarker excludes vascular contribution biomarkers establish pathology, not monopoly
“Mixed dementia” is one entity combinations and relative burdens differ

Autopsy anchor

Schneider et al. studied the first 141 autopsies in the Rush Memory and Aging Project. Among 50 dementia cases, Alzheimer pathology plus infarcts (38.0%, n=19) exceeded pure Alzheimer pathology (30.0%, n=15) and vascular dementia alone (12%, n=6); 12% had AD plus PD/LBD. After age adjustment, multiple diagnoses raised dementia odds almost three-fold versus one pathologic diagnosis (OR 2.8, 95% CI 1.2–6.7) (Schneider 2007, PMID 17568013).

In the Nun Study, among 61 participants meeting neuropathologic AD criteria, lacunar infarcts in basal ganglia, thalamus, or deep white matter were associated with dementia OR 20.7 (95% CI 1.5–288); fewer AD lesions appeared sufficient for dementia when those infarcts were present, whereas infarcts were only weakly associated with dementia among 41 participants without AD pathology (Snowdon 1997, PMID 9052711). Esiri et al. compared 24 vascular-dementia, 19 vascular-but-undemented, and 18 control brains selected for absent/minimal Alzheimer pathology: severe cribriform change (P=0.0006) and microinfarction (P=0.031) correlated with dementia, whereas single macroscopic infarcts were more common in the undemented vascular group (P=0.0028) (Esiri 1997, PMID 9416809). Kalaria's later synthesis likewise lists small/lacunar infarcts, microinfarcts, perivascular-space dilation, myelin loss, arteriolosclerosis, and leptomeningeal CAA as the best neuropathologic predictors of vascular cognitive impairment (Kalaria 2018, PMID 29273521).

Autopsy category Share of dementia cases
Alzheimer pathology plus infarcts 38.0%
Alzheimer pathology alone 30.0%
Vascular dementia alone 12%

These proportions are cohort-specific. They should not be transported as universal prevalence, but they establish that mixed disease is not an edge case.

How much is "mixed"? The person-level answer

Schneider's 141-autopsy series has been superseded in scale by the same group's later work, and the numbers get more extreme, not less, as sample size grows.

Study N autopsied Key quantities
Boyle 2018 (PMID 29244218) 1,079 with 2+ cognitive evaluations (max 22) 94% had ≥1 neuropathology, 78% ≥2, 58% ≥3, 35% ≥4. Alzheimer disease was the most frequent (65%) but occurred in isolation in only 9%. More than 230 distinct neuropathologic combinations were observed, each in <6% of the cohort. AD accounted for ~50% of cognitive loss on average, but the person-specific proportion ranged from 22% to 100%
Lamar 2022 (PMID 34601898) 1,474 decedents (mean ~88 y, 65% female) 80% had cerebrovascular pathology; of these, 37% had a single CVD type and 63% had mixed CVD profiles across 32 possible combinations of three vessel diseases (atherosclerosis, arteriolosclerosis, CAA) and two tissue injuries (macroinfarcts, microinfarcts). Only the mixed CVD group declined faster than participants with no CVD; single CVD profiles did not
Yu 2026 (PMID 41543852) 1,633 (ROS + MAP, mean age at death 90.4) >80% mixed pathology; 280 unique copathology combinations. Hierarchical clustering yielded five profiles: infarcts + vessel disease (15.9%), LATE-NC + hippocampal sclerosis (12.3%), Lewy bodies (21.7%), ADNC + CAA (9.7%), and low pathology (40.4%). Fastest decline in the LATE-NC/hippocampal-sclerosis and ADNC/CAA profiles

Three findings from this series change how "vascular dementia" should be read.

1. The person-specific attributable fraction is the right quantity, and it is enormously variable. Boyle's central result is not that AD explains half of cognitive loss on average — it is that the same pathology explains 22% of decline in one person and 100% in another (Boyle 2018, PMID 29244218). A group-level statement that vascular pathology "accounts for X% of dementia" is therefore a statement about a distribution, and prevention or treatment aimed at a single pathology will have wildly heterogeneous individual benefit even if it works. This is the concrete form of OQ-11.

2. Vascular pathology is itself mixed, and only the mixtures matter. Lamar's decomposition is the sharpest available refutation of "cerebrovascular disease" as a unit: among 1,474 brains, single CVD types — one vessel disease or one kind of tissue injury on its own — showed no faster cognitive decline than no CVD at all, while mixed CVD profiles did, with combinations involving both atherosclerosis and arteriolosclerosis performing worst and different profiles mapping onto different cognitive domains (Lamar 2022, PMID 34601898). Any study that codes cerebrovascular pathology as present/absent is averaging a group with no measurable effect together with a group that declines rapidly.

3. When pathologies are clustered empirically, the vascular cluster is not the most damaging one. Yu's five-profile solution assigned 15.9% of participants to an infarct-and-vessel-disease profile, but the fastest cognitive decline occurred in the LATE-NC/hippocampal-sclerosis and ADNC/CAA profiles; degenerative pathologies clustered into distinct profiles while vascular conditions distributed more diffusely across them (Yu 2026, PMID 41543852). Read together with Lamar, this suggests vascular pathology behaves less like a discrete disease entity and more like a modifier that raises the impact of whatever else is present — which is exactly the reserve-threshold model described below, now with a population-scale test behind it (Kapasi 2017, PMID 28488154).

The uncomfortable denominator: most of the variance is unexplained

The mixed-pathology literature is usually read as showing that dementia is caused by several pathologies rather than one. The same cohorts show something stronger and less often quoted: even all of them together explain a minority of cognitive decline. In 856 decedents with a mean of 7.5 annual evaluations, global Alzheimer pathology, amyloid, tangles, macroscopic infarcts and neocortical Lewy bodies individually explained 22%, 6%, 34%, 2% and 8% of the variance in cognitive decline; considered simultaneously they accounted for 41%, and for less than a third of the variation in onset of terminal decline and in preterminal and terminal decline rates (Boyle 2013, PMID 23798485).

Expanding to eleven pathological indices — adding LATE-NC, hippocampal sclerosis, microinfarcts, CAA, atherosclerosis and arteriolosclerosis — in 1,164 decedents with up to 24 annual evaluations barely moved the total: 43% of the variance in decline, of which Alzheimer-related indices accounted for 30–36%, non-Alzheimer neurodegenerative indices 4–10%, and cerebrovascular indices 3–8%; the eleven together explained 28% of the variation in onset of terminal decline, 32% of preterminal and 19% of terminal decline rates (Boyle 2021, PMID 33742668).

Two consequences for this condition specifically. First, the entire cerebrovascular block explains 3–8% of the variance in late-life cognitive decline in these community cohorts — less than tangles alone. This is a cohort- and model-specific descriptive contribution, not a causal ceiling on the benefit possible from vascular intervention. Second, adding six pathologies to the model raised explained variance by two percentage points, showing that broad residual variance remains; the analysis cannot identify whether that residual reflects unmeasured pathology, resilience, reserve, systemic illness, measurement error, or other factors.

The third pathology on this border: LATE-NC

Vascular disease and Alzheimer disease are not the only pair. Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC) is a stereotypical TDP-43 proteinopathy of older adults, with or without hippocampal sclerosis, producing an amnestic syndrome that mimicked Alzheimer-type dementia in retrospective autopsy studies; in community-based cohorts roughly 25% of brains carry enough LATE-NC to be associated with discernible cognitive impairment. It is distinguished from frontotemporal TDP-43 disease by age and by restricted anatomical distribution, is staged by TDP-43 immunohistochemistry in amygdala, hippocampus and middle frontal gyrus, and has five identified risk genes — GRN, TMEM106B, ABCC9, KCNMB2, APOE. There is still no ante-mortem biomarker for it (Nelson 2019, PMID 31039256).

That matters here for two reasons. First, an amnestic older patient with vascular lesions on MRI may have LATE-NC rather than either Alzheimer or vascular disease driving the memory failure, and no test currently distinguishes them in life. Second, LATE-NC and small-vessel pathology co-occur far more than by chance: among 749 community-dwelling decedents, 409 (54.6%) had LATE-NC, and 354 of those (86.5%) had at least one microvascular pathology — moderate-severe arteriolosclerosis in basal ganglia 32.3%, anterior watershed 47.6%, posterior watershed 35.2%; moderate-severe CAA 41.5%; microinfarcts 36.6%. On regression adjusting for demographics, Alzheimer pathology and other age-related pathologies, only posterior watershed arteriolosclerosis (OR 1.12, 95% CI 1.01–1.25) and capillary CAA (OR 1.71, 1.13–2.58) — not CAA severity, and not arteriolosclerosis elsewhere — were associated with more advanced LATE-NC stage, with results unchanged after controlling for APOE ε4, vascular risk factors or vascular disease (Agrawal 2021, PMID 33624322). The regional specificity is the interesting part: a global "arteriolosclerosis present" variable would have missed it.

Vascular risk and amyloid interact rather than add

The autopsy data above describe endpoints; two longitudinal biomarker studies describe the process, and both find synergy, not addition.

In 223 clinically normal older adults from the Harvard Aging Brain Study (mean age 73.7, mean follow-up 3.7 years), faster decline on the Preclinical Alzheimer Cognitive Composite was associated with both Framingham cardiovascular risk score (β = −0.064, 95% CI −0.094 to −0.033, P<0.001) and PiB amyloid burden (β = −0.058, −0.079 to −0.037, P<0.001), and their interaction with time was significant (β = −0.040, −0.062 to −0.018, P<0.001). Vascular risk remained associated with decline (β = −0.055, −0.086 to −0.024) after adjustment for amyloid burden, hippocampal volume, FDG-PET and white matter hyperintensities — that is, the vascular risk score carried information the vascular imaging marker did not (Rabin 2018, PMID 29799986).

The mechanism appears to run through tau. In 175 cognitively unimpaired adults with longitudinal flortaucipir PET over 3.6 years and cognition over 7.0 years, baseline vascular risk and amyloid interacted on inferior-temporal tau accumulation (P=0.004) — including in individuals whose amyloid burden was below the conventional positivity threshold — with systolic blood pressure and body mass index each independently interacting with amyloid (both P<0.0001), and tau accumulation mediating 33% of the vascular-risk × amyloid effect on cognitive decline (Yau 2022, PMID 35880989). If this is right, vascular risk factors are not a parallel insult but an accelerant of Alzheimer's own molecular cascade, and the therapeutic implication — treat hypertension and obesity to reduce amyloid-related tau — is testable.

Whose brains have been counted

The mixed-pathology literature above is overwhelmingly derived from predominantly White North American volunteer cohorts, and the two studies that break that pattern disagree.

Study Sample Finding
Graff-Radford 2016 (PMID 27094726) NACC: 110 African American and 2,500 White decedents, all demented before death Alzheimer, Lewy body and cerebrovascular pathology all more common in African Americans; TDP-43 and FTLD-tau less common; APOE accounted for most of the AD difference
Suemoto 2024 (PMID 39052291) Biobank for Aging Studies, São Paulo: 1,815 population-based autopsies, 617 (34%) Black, collected 2004–2023 Small-vessel disease (OR 1.74, 95% CI 1.29–2.35) and siderocalcinosis (OR 1.70, 1.23–2.34) more frequent in Black participants; neuritic plaques more frequent in White participants (OR 0.61, 0.44–0.83). Neuropathologic AD diagnosis 39% vs 33% (White vs Black); vascular dementia 32% vs 24% (Black vs White). Race was not associated with cognitive ability and did not modify the pathology–cognition relationship

They agree that cerebrovascular pathology is more common in Black decedents and disagree about Alzheimer pathology, which is what one would expect if the NACC comparison — a referral-based, dementia-only, unbalanced sample — carries ascertainment bias that a population-based autopsy series does not. Suemoto's null for race as an effect modifier is the more useful result for this condition: it implies that reported differences in dementia burden between groups are differences in lesion frequency, not in how much damage a given lesion does — which locates the cause upstream, in exposure to vascular risk, rather than in the brain's response to it.

A third non-Western series adds a different axis. Among 610 community-dwelling autopsies from the National Human Brain Bank in China, LATE-NC (n=341) was more frequent than Alzheimer neuropathologic change (n=331), with α-synucleinopathies in 124, primary age-related tauopathy in 231, argyrophilic grain disease in 107, age-related tau astrogliopathy in 144, CAA in 183 and hippocampal sclerosis in 46; frontotemporal lobar degeneration and ALS were rare. APOE ε4 allele frequency was 13.63%, well below typical European figures, and ε4 carriage was linked to advanced ADNC stage and to more comorbidities. Advanced ADNC and higher LATE-NC stage were each associated with the other, and high-level ADNC, neocortical Lewy body disease and LATE-NC stage 3 independently predicted severe cognitive status (Wu 2025, PMID 39901730). If LATE-NC is at least as common as Alzheimer change in a Chinese community brain bank, then the vascular fraction of dementia in China (see epidemiology) is being estimated in a population whose non-vascular copathology mix differs from the cohorts that calibrated the attribution rules.

The oldest-old picture is different again and comes from the largest single series. Among 1,700 consecutive autopsies of demented elderly in Vienna (mean age 84.3 ± 5.4), "pure" vascular dementia was 12.3% of the cohort and declined from 15.0% to 8.7% between age 60 and 90+, while Alzheimer disease alone (45.6%), mixed dementia (5.5%) and Alzheimer disease with minor cerebrovascular lesions (22.3%) increased with age. The vascular cases were systemically ill: 85% had diabetes, 75% prior stroke, 95% morphological signs of hypertension, 65% myocardial infarction, 97% hypertensive-arteriosclerotic microangiopathy (with CAA in 23%) and 90% severe large-artery atherosclerosis. Vascular dementia had subcortical infarcts in over 60% of cases against 43% in mixed dementia (Jellinger 2010, PMID 21504129). Pure vascular dementia therefore becomes rarer with age as a proportion, while the mixed categories expand — the opposite of what the rising absolute stroke burden in the oldest old would suggest, and a reason to distrust age-stratified vascular fractions built from clinical rather than pathological series.

Sex

Mixed pathology is patterned by sex, and the vascular combination is the one that differs. In more than 1,500 community-dwelling older adults, women were significantly more likely than men to have Alzheimer disease plus cerebrovascular pathology, adjusted for age at death, education, race and APOE ε4, while men were more likely to have pure Lewy body disease; the excess of Alzheimer-plus-TDP-43/hippocampal-sclerosis in women did not survive adjustment (Barnes 2019, PMID 31128096). This sits directly against the EURODEM finding that clinical vascular-dementia incidence is sex-neutral, and the two can only be reconciled if the additional cerebrovascular pathology in women is disproportionately labelled Alzheimer disease in life.

Mechanistic models

Model Description Expected data pattern
Additive each pathology independently reduces cognition separate main effects
Synergistic combined effect exceeds sum positive interaction
Reserve-threshold infarcts reduce reserve so less Alzheimer burden becomes symptomatic lower AD burden at same impairment
Vascular-to-amyloid impaired clearance promotes amyloid accumulation vascular change precedes amyloid change
Shared causes aging/metabolic exposures influence both correlated pathways without direct mediation
Ascertainment one lesion leads to imaging/workup finding another detection changes apparent association

Neuropathological reviews find heterogeneous combinations and warn that thresholds for “vascular dementia” themselves influence classification (Jellinger 2002, PMID 12417375; Jellinger 2008, PMID 18525132).

Clinical attribution matrix

Vascular evidence Alzheimer evidence Most defensible formulation
strong absent/negative probable predominant vascular contribution, with test limitations
strong positive mixed vascular and Alzheimer disease
weak/incidental positive Alzheimer disease with vascular comorbidity
weak unavailable etiologically uncertain cognitive disorder
CAA pattern amyloid positive vascular and parenchymal amyloid require separate interpretation

“Strong” vascular evidence means more than any WMH: strategic/recurrent infarction, severe coherent SVD, or a clear temporal-anatomical relationship. Alzheimer biomarker positivity likewise must be interpreted against age, assay, and clinical stage.

Cognitive phenotype cannot partition pathology

Executive and processing-speed impairment occurs commonly in VCI, but Desmond found no single specific neuropsychological deficit that reliably defines vascular etiology (Desmond 2004, PMID 15537510). Mixed cases can show amnestic, dysexecutive, language, visuospatial, or global patterns depending on lesion distribution and neurodegenerative stage.

Observation Possible vascular explanation Possible Alzheimer explanation
poor delayed recall thalamic/hippocampal infarct medial temporal neurodegeneration
slowed set shifting white-matter disconnection later multidomain disease
gradual decline progressive SVD Alzheimer progression
step after stroke direct infarct effect stroke unmasks low reserve
hippocampal atrophy vascular injury or age Alzheimer-pattern neurodegeneration

Imaging and fluid biomarkers

MRI identifies infarcts, WMH, lacunes, microbleeds, siderosis, and atrophy; it does not visualize all microinfarcts or prove causal contribution (Duering 2023, PMID 37236211). Amyloid PET and CSF/plasma amyloid measures support amyloid pathology, while tau biomarkers refine neurofibrillary involvement; these are covered in the companion condition to avoid duplication.

CAA sits directly on the border because amyloid accumulates in vessels and is linked to lobar haemorrhagic and cognitive phenotypes. Boston v2.0 provides a probabilistic in-vivo CAA diagnosis, not a measure of parenchymal Alzheimer contribution (Charidimou 2022, PMID 35841910).

Trial implications

Trial choice Benefit Cost
Exclude amyloid-positive participants cleaner vascular signal lower generalizability
Include and stratify represents clinical population larger sample needed
Require severe MRI SVD enriches vascular burden selects late/heterogeneous injury
Use factorial/combination therapy tests joint biology complexity and safety
Analyze continuous burdens avoids arbitrary labels measurement harmonization

Older vascular-dementia drug trials often used clinical criteria without contemporary Alzheimer biomarkers, so small cognitive effects cannot be assigned confidently to pure vascular biology (Battle 2021, PMID 33704781).

Repository boundary

This condition owns vascular lesions, vascular attribution, SVD/CAA, post-stroke cognition, and vascular prevention. Alzheimer's disease owns amyloid/tau staging and anti-amyloid treatment. Both conditions own the junction through reciprocal overlap pages; neither should claim a clean separation.

Open questions

  • How should cohort-specific variance attributed to cerebrovascular pathology be interpreted without treating it as a causal treatment-effect ceiling? (Boyle 2021, PMID 33742668; Boyle 2013, PMID 23798485)
  • Why is LATE-NC at least as frequent as Alzheimer neuropathologic change in a Chinese community brain bank, and does that change how the vascular fraction should be estimated there? (Wu 2025, PMID 39901730)
  • Why does the proportion of "pure" vascular dementia fall from 15.0% to 8.7% between age 60 and 90+ while stroke burden rises? (Jellinger 2010, PMID 21504129)
  • If women carry more Alzheimer-plus-cerebrovascular pathology at autopsy but vascular-dementia incidence is sex-neutral clinically, is the difference in labelling rather than in biology? (Barnes 2019, PMID 31128096)
  • Can continuous vascular and Alzheimer biomarker burdens generate individual attributable fractions? (Schneider 2007, PMID 17568013)
  • Are pathology effects additive or synergistic across cognitive domains? (Jellinger 2007, PMID 17324442)
  • Should vascular trials stratify, exclude, or target amyloid-positive participants? (Battle 2021, PMID 33704781)
  • Can treating vascular injury delay symptoms even when Alzheimer biomarkers are positive? (Gorelick 2011, PMID 21778438)
  • If single cerebrovascular pathologies do not accelerate decline but mixtures do, is "cerebrovascular disease present/absent" a usable variable at all? (Lamar 2022, PMID 34601898)
  • Can an in-vivo biomarker panel reproduce Boyle's person-specific attributable fractions, which range from 22% to 100% for the same pathology? (Boyle 2018, PMID 29244218)
  • Does the empirical clustering of copathology into five profiles yield better trial strata than criteria-based labels? (Yu 2026, PMID 41543852)
  • With >230 observed pathology combinations each present in <6% of a cohort, is any single-target therapy realistic for late-life dementia? (Boyle 2018, PMID 29244218)
  • How many patients labelled vascular or Alzheimer dementia actually have LATE-NC, given that no ante-mortem biomarker exists? (Nelson 2019, PMID 31039256)
  • Why is LATE-NC stage associated with posterior watershed arteriolosclerosis and capillary CAA specifically, and not with arteriolosclerosis generally? (Agrawal 2021, PMID 33624322)
  • If vascular risk predicts decline after adjustment for WMH, what vascular injury is the risk score capturing that the imaging is not? (Rabin 2018, PMID 29799986)
  • Would treating hypertension and obesity reduce tau accumulation, as the 33% mediation estimate predicts? (Yau 2022, PMID 35880989)
  • Are racial differences in dementia burden differences in lesion frequency rather than in the cognitive cost of a lesion? (Suemoto 2024, PMID 39052291; Graff-Radford 2016, PMID 27094726)

References

  1. Schneider JA, et al. Mixed brain pathologies account for most dementia cases. Neurology. 2007. PMID 17568013
  2. Jellinger KA, Attems J. Neuropathological evaluation of mixed dementia. J Neurol Sci. 2007. PMID 17324442
  3. Jellinger KA. Pathology of ischemic-vascular dementia. J Neurol Sci. 2002. PMID 12417375
  4. Jellinger KA. Pathology of vascular dementia: critical update. J Alzheimers Dis. 2008. PMID 18525132
  5. Desmond DW. Neuropsychology of vascular cognitive impairment. J Neurol Sci. 2004. PMID 15537510
  6. Duering M, et al. STRIVE-2. Lancet Neurol. 2023. PMID 37236211
  7. Charidimou A, et al. Boston criteria v2.0 for CAA. Lancet Neurol. 2022. PMID 35841910
  8. Battle CE, et al. Cholinesterase inhibitors for vascular cognitive impairment. Cochrane Database Syst Rev. 2021. PMID 33704781
  9. Gorelick PB, et al. Vascular contributions to cognitive impairment and dementia. Stroke. 2011. PMID 21778438
  10. Snowdon DA, et al. Brain infarction and the clinical expression of Alzheimer disease: the Nun Study. JAMA. 1997;277:813-7. PMID 9052711
  11. Esiri MM, et al. Neuropathological assessment of the lesions of significance in vascular dementia. J Neurol Neurosurg Psychiatry. 1997;63:749-53. PMID 9416809
  12. Kalaria RN. The pathology and pathophysiology of vascular dementia. Neuropharmacology. 2018;134:226-239. PMID 29273521
  13. Boyle PA, et al. Person-specific contribution of neuropathologies to cognitive loss in old age. Ann Neurol. 2018;83:74-83. PMID 29244218
  14. Lamar M, et al. Complex profiles of cerebrovascular disease pathologies in the aging brain and their relationship with cognitive decline. Stroke. 2022;53:218-227. PMID 34601898
  15. Yu L, et al. Neuropathologic profiles and associated cognitive trajectories in community-living older adults. JAMA Netw Open. 2026;9:e2554354. PMID 41543852
  16. Kapasi A, et al. Impact of multiple pathologies on the threshold for clinically overt dementia. Acta Neuropathol. 2017;134:171-186. PMID 28488154
  17. Nelson PT, et al. Limbic-predominant age-related TDP-43 encephalopathy (LATE): consensus working group report. Brain. 2019;142:1503-1527. PMID 31039256
  18. Agrawal S, et al. Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change and microvascular pathologies in community-dwelling older persons. Brain Pathol. 2021;31:e12939. PMID 33624322
  19. Rabin JS, et al. Interactive associations of vascular risk and β-amyloid burden with cognitive decline in clinically normal elderly individuals: findings from the Harvard Aging Brain Study. JAMA Neurol. 2018;75:1124-1131. PMID 29799986
  20. Yau WW, et al. Tau mediates synergistic influence of vascular risk and Aβ on cognitive decline. Ann Neurol. 2022;92:745-755. PMID 35880989
  21. Graff-Radford NR, et al. Neuropathologic differences by race from the National Alzheimer's Coordinating Center. Alzheimers Dement. 2016;12:669-77. PMID 27094726
  22. Suemoto CK, et al. Neuropathological lesions and cognitive abilities in Black and White older adults in Brazil. JAMA Netw Open. 2024;7:e2423377. PMID 39052291
  23. Boyle PA, et al. Much of late life cognitive decline is not due to common neurodegenerative pathologies. Ann Neurol. 2013;74:478-89. PMID 23798485
  24. Boyle PA, et al. To what degree is late life cognitive decline driven by age-related neuropathologies? Brain. 2021;144:2166-2175. PMID 33742668
  25. Wu W, et al. Neuropathological comorbidity, genetics and cognition in a Chinese community-based autopsy cohort. Brain. 2025;148:2481-2492. PMID 39901730
  26. Jellinger KA, Attems J. Prevalence and pathology of vascular dementia in the oldest-old. J Alzheimers Dis. 2010;21:1283-93. PMID 21504129
  27. Barnes LL, et al. Sex differences in mixed neuropathologies in community-dwelling older adults. Brain Res. 2019;1719:11-16. PMID 31128096