Pathophysiology

TL;DR — Vascular cognitive impairment emerges from repeated focal injury and diffuse failure of the neurovascular unit, including endothelial dysfunction, impaired autoregulation, blood–brain barrier (BBB) leakage, mural-cell injury, inflammation, and white-matter disconnection (Santisteban 2023, PMID 36129176; Uemura 2020, PMID 32317958). Chronic hypoperfusion is one mechanism, not a complete theory: measured low flow can be cause, consequence, or compensation. Visible MRI lesions underestimate microinfarction and tissue-level dysfunction, while Alzheimer pathology can both coexist with and be amplified by vascular failure (Duering 2023, PMID 37236211; Schneider 2007, PMID 17568013). Translational progress depends on linking a mechanism to a defined human phenotype and an outcome that can change.

Neurovascular unit

Component	Normal role	Candidate failure
Endothelium	barrier, signaling, tone, antithrombotic surface	leakage, inflammation, impaired dilation
Pericytes	capillary stability and flow regulation	capillary dysfunction (Uemura 2020, PMID 32317958)
Smooth-muscle cells	resistance-vessel tone	impaired autoregulation, wall degeneration
Astrocytes	metabolic/vascular coupling	dysregulated neurovascular signaling
Oligodendrocytes	myelin maintenance	white-matter vulnerability
Microglia	surveillance/repair	chronic inflammatory injury
Extracellular matrix	vessel/tissue structure	fibrosis and impaired transport

Two of these components have mechanistic evidence specific enough to name a step. Pericyte-deficient mice show that pericyte degeneration diminishes both global and individual-capillary CBF responses to neuronal stimulation, producing neurovascular uncoupling, reduced oxygen supply and metabolic stress, with impaired neuronal excitability and neurodegenerative change following over time — establishing the ordering vascular-first, neuronal-second at the capillary level (Kisler 2017, PMID 28135240). Human tissue then shows the same association is regional rather than global: in precuneus and underlying white matter from 49 Alzheimer and 37 control brains, the pericyte marker PDGFRβ fell and fibrinogen (barrier leakage) rose in the precuneus, correlating with reduced ante-mortem oxygenation and with fibrillar plaque load — while in the underlying white matter fibrinogen leakage and reduced oxygenation occurred with PDGFRβ unchanged, which the authors attribute to reduced transcortical perfusion rather than local mural-cell loss (Miners 2018, PMID 28151041). Pericyte injury and barrier leakage are therefore not interchangeable, and white-matter leakage in particular need not imply pericyte loss.

The neurovascular unit concept prevents reduction of disease to lumen narrowing alone. Hypertension can remodel arterioles, blunt vasoreactivity, increase oxidative stress, and disturb BBB function across decades (Santisteban 2023, PMID 36129176). Iadecola's JACC panel treats large- and small-vessel alterations, including subcortical white-matter microcirculation, as key modifiers of cognitive expression of other pathologies including Alzheimer disease (Iadecola 2019, PMID 31248555). Wardlaw et al. list endothelial dysfunction, BBB leakage, impaired vasodilation, stiffening, dysfunctional flow and interstitial drainage, white-matter rarefaction, ischaemia, inflammation, myelin damage, and secondary neurodegeneration as the human mechanistic inventory (Wardlaw 2019, PMID 31097385).

Focal and diffuse injury

Route	Lesion/output	Cognitive consequence
Occlusion	infarct or microinfarct	tissue loss and network interruption
Vessel rupture	ICH or microbleed	tissue destruction and iron-related injury
Chronic wall disease	WMH/lacune progression	distributed disconnection
BBB leakage	edema/protein extravasation	inflammatory and myelin injury
Impaired reactivity	reduced functional hyperemia	inefficient network activation
Drainage failure	perivascular solute accumulation	possible interaction with amyloid

Neuropathology shows that vascular dementia encompasses infarcts, small-vessel changes, haemorrhages, and mixed lesions; no single lesion is necessary across all cases (Jellinger 2002, PMID 12417375; Jellinger 2008, PMID 18525132).

Which vessel makes which lesion

The default account of sporadic small-vessel disease is arteriolar: hypertension thickens and hyalinizes arteriolar walls, and the white matter suffers. Brain arteriolosclerosis is genuinely near-universal — present in over 80% of autopsied individuals beyond 80 years — and in clinicopathological studies adjusting for comorbid disease it is independently associated with impaired global cognition, episodic memory, working memory and perceptual speed, and with autonomic and parkinsonian features; the review that establishes this also states that there is no universally applied diagnostic methodology for it and that it probably comprises subcategories with different pathogenesis (Blevins 2021, PMID 33098484).

Two autopsy studies then complicate the arteriolar account by separating lesions by vessel type.

Vessel change	Associated parenchymal damage	Not associated with
Arteriolosclerosis	lacunes (P=0.004); frontal parenchymal haemosiderin (P=0.024)	leukoaraiosis severity (Cao 2023, PMID 36581493)
Venular collagenosis	white-matter pallor, periventricular (β=−0.430, P=0.028) and deep (β=−0.437, P=0.025)	lacune count; haemosiderin (Cao 2023, PMID 36581493)

Quantitative morphometry in 25 brains selected to span the full WMH range reaches the same split from the imaging side: venous collagen thickness predicted periventricular WMH volume (P=0.017) while arterial collagen did not (P=0.128), and perivascular-space width predicted nothing in either vessel type (arteries P=0.937, veins P=0.133) despite correlating with collagen thickness in both (Lahna 2022, PMID 36054513). If WMH is principally a venular lesion and lacunes principally an arteriolar one, then treating "small-vessel disease burden" as one exposure conflates two diseases with different upstream physiology — and antihypertensive treatment, which acts on the arteriolar side, would be expected to move lacunes more reliably than WMH.

Is hypoperfusion cause or consequence?

The chronic-hypoperfusion model predicts that low flow precedes and produces white-matter injury. The best longitudinal human test available reverses that arrow. In 575 patients with manifest arterial disease (mean age 57) scanned at baseline and after ~3.9 years, baseline parenchymal cerebral blood flow did not predict progression of WMH or lacunes, whereas baseline periventricular and deep WMH volumes predicted subsequent decline in perfusion — per 1% intracranial-volume increase in lesion volume, pCBF fell by 0.70 mL/min/100 mL (95% CI −1.40 to −0.00) for periventricular and 1.01 (−1.64 to −0.38) for deep WMH, the deep-WMH association surviving adjustment for cardiovascular risk factors (0.92, −1.56 to −0.28). Lacunes predicted no perfusion change (van der Veen 2015, PMID 25804924).

That is one cohort, of middle-aged patients with established arterial disease, using 1.5T ASL-free perfusion estimation, and it cannot exclude hypoperfusion operating at a spatial scale the measurement misses — the perilesional and local findings above suggest exactly that. But it means "chronic hypoperfusion causes white-matter disease" is a hypothesis with contrary longitudinal evidence, not an established mechanism, and the oxygen-extraction data showing compensation before decompensation fit the same picture: measured flow is a poor proxy for tissue oxygen adequacy.

Pulsatility as an upstream exposure

If the penetrating arterioles are damaged by pressure waves rather than by mean pressure, the treatable target is aortic rather than cerebral. In 188 OXVASC patients restudied a median 5.8 years after TIA or non-disabling stroke under intensive medical management, pulse-wave velocity rose 2.4% (P<0.0001) and aortic pulse pressure 3.5% (P<0.0001) despite treatment, with acceleration above age 55 (aoPP P<0.0001; middle-cerebral-artery pulsatility index P=0.009); PWV and aoPP progression were predicted by aortic systolic and diastolic pressure, while MCA-PI predominantly reflected concurrent aortic pulse pressure (P<0.001) (Webb 2022, PMID 34852644). Two implications: current best medical treatment does not arrest the stiffening process, and because cerebral pulsatility is largely a downstream readout of aortic pulse pressure, an intervention aimed at cerebral pulsatility must act on the aorta.

Neurovascular coupling as a common final pathway

Coupling — the flow increase that follows neural activity — is reduced across essentially every form of small-vessel disease. Of 29 studies (19 case-control, 10 cohort), 26 reported reduced coupling with increasing SVD severity; 25 of 28 studies of WMH, 8 of 9 of microbleeds, 3 of 5 of lacunar stroke, 6 of 7 in cerebral amyloid angiopathy and 4 of 4 in CADASIL found reductions, while no study found an association with enlarged perivascular spaces. In limited meta-analysis, occipital %BOLD response to visual stimulation fell with more severe WMH (7 studies, SMD −1.51, P<0.01) and with increasing microbleeds (7 studies, SMD −1.31, P<0.01) (Yang 2023, PMID 37697725). The consistency across sporadic, amyloid and monogenic disease argues for endothelial dysfunction as a shared physiological end-point; the perivascular-space exception argues that PVS enlargement is not a vascular-function marker at all.

Hypertension and autoregulation

Chronic hypertension shifts vascular structure and regulation, increasing susceptibility to small-vessel injury. Lowering pressure reduces stroke and modestly reduces cognitive outcomes at population level, but very low pressure can be harmful in people with impaired autoregulation, stenosis, orthostatic symptoms, or frailty (Hughes 2020, PMID 32427305; Santisteban 2023, PMID 36129176).

This creates a mechanistic U-shape that observational studies cannot resolve alone: low pressure may cause hypoperfusion, or it may signal cardiac disease, frailty, autonomic failure, or prodromal dementia.

The physiology under that U-shape is better specified than the clinical literature usually acknowledges, and it is not one mechanism. Cerebral blood flow is regulated by four partly independent systems — autoregulation (response to perfusion-pressure change), vascular reactivity to vasoactive stimuli including CO₂, neurovascular coupling, and endothelium-dependent responses — which interact and can fail separately, with distinct clinical consequences in ageing, hypertension, stroke, mild cognitive impairment, anaesthesia and dementia, and in everyday challenges such as posture change and exercise (Claassen 2021, PMID 33769101). A patient with intact autoregulation but impaired CO₂ reactivity, and one with the reverse, would both be described as having "impaired cerebral autoregulation" in most clinical writing, and would be expected to respond differently to blood-pressure lowering. No vascular-dementia trial has stratified on which of the four is impaired.

Ageing degrades these systems together with the rest of the neurovasculature. Arterial, venous and lymphatic vessels all change structurally and functionally with age, impairing oxygen and glucose delivery to active regions, disrupting endothelial transport, compromising proteostasis by reducing clearance of toxic proteins, weakening immune surveillance and privilege, and depriving the brain of growth factors needed for repair (Santisteban 2025, PMID 39788087). The venous and lymphatic components of that list are the least studied and align with the venular-collagenosis findings above.

Blood–brain barrier

BBB leakage is measurable by dynamic contrast MRI and fluid markers, but protocols and thresholds vary. High-resolution DCE-MRI in the living human brain showed age-dependent BBB breakdown in the hippocampus and specifically in its CA1 and dentate-gyrus subdivisions, worsening with mild cognitive impairment and correlating with CSF evidence of injury to BBB-associated pericytes (Montagne 2015, PMID 25611508). That measurable leakage is hippocampal rather than confined to the deep white matter is awkward for a purely small-vessel account of vascular cognitive impairment and is one reason the barrier is treated in this knowledge base as a shared mechanism with Alzheimer's disease rather than a vascular-specific one.

A 2025 review catalogs BBB/neurovascular-unit biomarkers across cognitive impairment, illustrating breadth rather than validating a VaD test (French 2025, PMID 40145342).

Candidate measure	What it samples	Limitation
DCE-MRI permeability	regional contrast leakage	acquisition/model dependence
CSF/serum albumin ratio	global barrier permeability	lumbar/systemic confounding
soluble adhesion molecules	endothelial activation	nonspecific systemic signal
pericyte-associated proteins	mural-cell injury	limited clinical validation
diffusion/free-water	tissue microstructure	indirect

White-matter vulnerability

Long penetrating vessels and watershed-like white-matter territories may be vulnerable to pulsatility, impaired autoregulation, and oligodendrocyte injury. Axonal damage manifests as slowed network communication before gross cavitation. Diffusion measures may therefore detect disease beyond visible WMH, but scanner harmonization is essential (Duering 2023, PMID 37236211).

A WMH is not one lesion type

Multimodal MRI has dissolved "white matter hyperintensity" into mechanistically distinct entities along four axes — WMH versus normal-appearing white matter, periventricular versus deep, lesion core versus perilesional penumbra, and longitudinal evolution. Periventricular WMH associate with blood–brain barrier dysfunction, interstitial fluid accumulation, and venous remodelling; deep WMH associate more with impaired glymphatic/perivascular clearance and enlarged perivascular spaces, with demyelination and macromolecular compromise varying by context (Chung 2026, PMID 41656632). The perilesional penumbra shows distance-dependent gradients of microstructural rarefaction, extracellular fluid expansion, perfusion deficit, and reduced vascular reactivity that extend beyond the FLAIR-defined border and predict subsequent lesion growth; abnormalities in diffusion, perfusion, and vascular reserve within normal-appearing white matter precede new WMH. Two consequences follow: total lesion burden is the wrong stratifier if the mechanisms differ by location, and the therapeutic target (BBB integrity, glymphatic clearance, cerebrovascular reactivity) should be chosen to match the WMH phenotype rather than applied to "SVD" generically.

The causal direction — vascular dysfunction first, tissue damage second — has now been tested prospectively in the cleanest available human model. In 22 CADASIL patients, voxelwise BOLD cerebrovascular reactivity to hypercapnia was measured at 7 T at baseline, and white-matter change was tracked on 3 T MRI over two years. Lower local baseline reactivity and higher dispersion predicted where mean diffusivity worsened and where normal-appearing white matter converted to WMH — yet whole-brain vascular-function measures predicted nothing at the global level (Pham 2026, PMID 40832975). Vascular dysfunction therefore appears to be locally causal and globally uninformative, which explains why trials using whole-brain reactivity or perfusion as an endpoint may be measuring the right mechanism at the wrong spatial scale.

Remyelination failure as a tractable step

Rodent chronic-hypoperfusion models locate a specific, druggable bottleneck. After unilateral common carotid artery occlusion, IL-1β rises early in the corpus callosum; IL-1 receptor antagonism or IL-1R1 knockout rescued myelin basic protein loss and improved remyelination, and the mechanism was narrow — oligodendrocyte precursor cell recruitment, not proliferation or differentiation, was the only compromised step, with IL-1β inhibiting OPC migration through IL-1R1 in transwell assay (Zhou 2017, PMID 27663285). Independently, optogenetic or chemogenetic activation of medial prefrontal glutamatergic neurons in a bilateral carotid artery stenosis model drove OPC differentiation in the corpus callosum via axonal Wnt2 secretion, improving myelin repair and working memory (Deng 2023, PMID 36529961). Both results argue that white-matter injury in chronic hypoperfusion is a repair-failure phenotype rather than pure tissue destruction, which is the strongest mechanistic reason to expect that a late intervention could still help. Neither has been tested in humans, and rodent chronic-hypoperfusion models lack the arteriolosclerosis, ageing, and copathology of the human disease — this is step 1 of the translational hierarchy below, not step 3.

Glymphatic and perivascular clearance

Perivascular spaces are visible markers linked to vessel and fluid dynamics. The hypothesis that impaired clearance connects vascular dysfunction to amyloid accumulation is biologically attractive, but human causal evidence and intervention targets remain incomplete (Solé-Guardia 2025, PMID 40289686). Enlarged spaces should not be described as direct proof of “glymphatic failure.”

Inflammation and glial response

Stroke triggers acute innate immune responses; chronic SVD may sustain lower-grade endothelial and glial activation. Circulating cytokines are heavily confounded by age, obesity, infection, and systemic vascular disease. Mechanistic claims therefore require spatial or longitudinal linkage to brain injury, not cross-sectional blood association alone.

APOE4 as a vascular, not only amyloid, risk factor

The strongest genetic risk factor for Alzheimer disease acts partly through the vessel wall, which makes it a shared node between this condition and Alzheimer's disease. In APOE transgenic mice, APOE4 expression and Apoe deficiency — but not APOE2 or APOE3 — caused BBB breakdown by activating a proinflammatory cyclophilin A → NF-κB → MMP-9 pathway in pericytes, leading to neuronal uptake of blood-derived neurotoxic proteins and reduced microvascular and cerebral blood flow; the vascular defects preceded neuronal dysfunction, and astrocyte-secreted APOE3 suppressed the pathway through a lipoprotein receptor (Bell 2012, PMID 22622580).

The human counterpart is direct. APOE4 carriers (ε3/ε4 or ε4/ε4) showed BBB breakdown in the hippocampus and medial temporal lobe that was present in cognitively unimpaired carriers, worse in impaired carriers, and unrelated to CSF or PET amyloid-β and tau. High baseline CSF soluble PDGFRβ predicted future cognitive decline in APOE4 carriers but not in non-carriers, after controlling for amyloid and tau status, and correlated with cyclophilin A–MMP-9 pathway activity in CSF (Montagne 2020, PMID 32376954). A pericyte-injury marker that is prognostic only in one genotype is the clearest existing case for genotype-stratified vascular trials, and it places cyclophilin A among the few named molecular targets in this condition.

Neuroinflammation: two processes, not one

Simultaneous PET-MR measurement separates microglial activation from barrier failure rather than merging them. Using ¹¹C-PK11195 for translocator protein and DCE-MRI for permeability in 20 patients with sporadic SVD, patients with CADASIL and 20 controls, sporadic SVD showed increased hotspot volumes of both microglial binding (P=0.003) and barrier permeability (P=0.007) in normal-appearing white matter plus increased mean permeability (P<0.001) — but the hotspots were not spatially related. In CADASIL there was no detectable increase in permeability and only a non-significant trend in microglial binding (P=0.073). Of 93 blood markers of cardiovascular disease, inflammation and endothelial activation, the first principal component was associated with permeability in sporadic SVD (β=0.976, P=0.003 for mean NAWM permeability) but not with microglial binding, and with nothing at all in CADASIL (Walsh 2021, PMID 34000009). Three consequences: inflammation and leakage are dissociable processes, systemic inflammatory markers index the barrier rather than the microglia, and the monogenic model does not reproduce the sporadic disease's barrier phenotype — a caution for using CADASIL as the trial population for a barrier-targeted drug.

Interaction with Alzheimer pathology

Vascular injury can lower the amount of Alzheimer pathology required for clinical dementia by reducing reserve, while impaired vascular clearance might influence amyloid handling. Autopsy evidence strongly supports coexistence, but additive and interactive causal models remain difficult to separate (Schneider 2007, PMID 17568013; Jellinger 2007, PMID 17324442).

Model	Prediction	Test
Additive burden	independent pathology effects sum	joint biomarker longitudinal model
Synergy	combined effect exceeds sum	interaction with adequate power
Mediation	vascular dysfunction increases amyloid/tau	serial mechanistic biomarkers
Common cause	shared aging/exposure drives both	causal inference/genetics
Diagnostic ascertainment	one disease makes the other more visible	pathology-based sampling

Translational hierarchy

1. Establish mechanism in tissue/model.
2. Show a reliable human marker.
3. Link marker longitudinally to cognitive/function change.
4. Modify marker with an intervention.
5. Demonstrate clinical benefit and acceptable harms.

Many candidate mechanisms remain at steps 1–2. Neuroimaging reviews describe perfusion, metabolism, connectivity, and BBB measures but emphasize heterogeneity and lack of validated diagnostic use (Frantellizzi 2020, PMID 31929166).

Open questions

• Which of the four CBF-regulatory systems — autoregulation, CO₂ reactivity, neurovascular coupling, endothelium-dependent response — fails first in vascular cognitive impairment, and should trials stratify on it? (Claassen 2021, PMID 33769101)
• If barrier leakage in white matter occurs without pericyte loss while precuneal leakage occurs with it, are cortical and white-matter barrier failure different diseases? (Miners 2018, PMID 28151041; Kisler 2017, PMID 28135240)
• Why does the earliest measurable human BBB breakdown appear in the hippocampus rather than in deep white matter, if small-vessel disease is the vascular-dementia substrate? (Montagne 2015, PMID 25611508)

• Do venous and lymphatic neurovascular ageing contribute independently to vascular cognitive impairment, or only through arterial change? (Santisteban 2025, PMID 39788087; Cao 2023, PMID 36581493)

• Is BBB leakage an upstream driver, downstream marker, or both? (French 2025, PMID 40145342)

• Which neurovascular dysfunction measure changes early enough for prevention trials? (Santisteban 2023, PMID 36129176)
• Does restoring pericyte function improve human cognition? (Uemura 2020, PMID 32317958)
• Can amyloid-clearance and vascular hypotheses be tested in the same longitudinal cohort? (Solé-Guardia 2025, PMID 40289686)
• Should periventricular and deep WMH be treated as separate diseases with separate targets? (Chung 2026, PMID 41656632)
• If cerebrovascular reactivity predicts white-matter damage locally but not globally, what is the correct spatial unit for an SVD trial endpoint? (Pham 2026, PMID 40832975)
• Is impaired OPC recruitment the rate-limiting step in human white-matter injury, as it is in rodent chronic hypoperfusion? (Zhou 2017, PMID 27663285; Deng 2023, PMID 36529961)
• If WMH is principally venular and lacunes principally arteriolar, should "total SVD burden" scores be abandoned as a stratifier? (Cao 2023, PMID 36581493; Lahna 2022, PMID 36054513)
• Given that baseline perfusion did not predict lesion progression while lesions predicted perfusion decline, what is left of the chronic-hypoperfusion model? (van der Veen 2015, PMID 25804924)
• Can any treatment slow aortic stiffening, given that pulse-wave velocity and aortic pulse pressure progressed under best medical management? (Webb 2022, PMID 34852644)
• Why is neurovascular coupling reduced with every SVD marker except enlarged perivascular spaces? (Yang 2023, PMID 37697725)
• Would a cyclophilin A or MMP-9 inhibitor benefit APOE4 carriers specifically, and should vascular trials stratify by genotype? (Bell 2012, PMID 22622580; Montagne 2020, PMID 32376954)
• Why do microglial-activation and barrier-leakage hotspots occupy different tissue, and which one should a trial target? (Walsh 2021, PMID 34000009)
• Is CADASIL an appropriate model for barrier-targeted therapy given that it shows no measurable barrier leakage? (Walsh 2021, PMID 34000009)

Related pages

• Cerebral small-vessel disease — lesion phenotype.
• Biomarkers — measurable candidates.
• Mixed pathology — interaction models.
• Prevention — modifiable exposures.

References

1. Santisteban MM, Iadecola C. Hypertension, neurovascular dysfunction, and cognitive impairment. Hypertension. 2023. PMID 36129176
2. Uemura MT, et al. Brain microvascular pericytes in vascular cognitive impairment and dementia. Front Aging Neurosci. 2020. PMID 32317958
3. French SR, et al. Biomarkers of BBB and neurovascular unit integrity in cognitive impairment. Alzheimers Dement. 2025. PMID 40145342
4. Duering M, et al. STRIVE-2. Lancet Neurol. 2023. PMID 37236211
5. Jellinger KA. Pathology of ischemic-vascular dementia. J Neurol Sci. 2002. PMID 12417375
6. Jellinger KA. Pathology of vascular dementia: critical update. J Alzheimers Dis. 2008. PMID 18525132
7. Schneider JA, et al. Mixed brain pathologies. Neurology. 2007. PMID 17568013
8. Jellinger KA, Attems J. Neuropathological evaluation of mixed dementia. J Neurol Sci. 2007. PMID 17324442
9. Frantellizzi V, et al. Neuroimaging in VCI and dementia: systematic review. J Alzheimers Dis. 2020. PMID 31929166
10. Solé-Guardia G, et al. Imaging brain fluid dynamics and waste clearance in SVD. Alzheimers Dement. 2025. PMID 40289686
11. Hughes D, et al. Blood pressure lowering and cognitive outcomes. JAMA. 2020. PMID 32427305
12. Iadecola C, et al. Vascular cognitive impairment and dementia. J Am Coll Cardiol. 2019;73:3326-3344. PMID 31248555
13. Wardlaw JM, et al. Small vessel disease: mechanisms and clinical implications. Lancet Neurol. 2019;18:684-696. PMID 31097385
14. Chung J, et al. Multimodal magnetic resonance imaging signatures of white matter hyperintensities: mechanistic insights into pathobiological heterogeneity. J Stroke. 2026;28:1-28. PMID 41656632
15. Pham SD, et al. Small vessel dysfunction at 7T MRI locally predicts white matter damage progression in CADASIL. J Cereb Blood Flow Metab. 2026;46:521-533. PMID 40832975
16. Zhou Y, et al. Interleukin-1β impedes oligodendrocyte progenitor cell recruitment and white matter repair following chronic cerebral hypoperfusion. Brain Behav Immun. 2017;60:93-105. PMID 27663285
17. Deng S, et al. Optogenetic stimulation of mPFC alleviates white matter injury-related cognitive decline after chronic ischemia through adaptive myelination. Adv Sci (Weinh). 2023;10:e2202976. PMID 36529961
18. Blevins BL, et al. Brain arteriolosclerosis. Acta Neuropathol. 2021;141:1-24. PMID 33098484
19. Cao Y, et al. Arteriolosclerosis differs from venular collagenosis in relation to cerebrovascular parenchymal damages: an autopsy-based study. Stroke Vasc Neurol. 2023;8:267-275. PMID 36581493
20. Lahna D, et al. Venous collagenosis as pathogenesis of white matter hyperintensity. Ann Neurol. 2022;92:992-1000. PMID 36054513
21. van der Veen PH, et al. Longitudinal relationship between cerebral small-vessel disease and cerebral blood flow: the SMART-MR study. Stroke. 2015;46:1233-8. PMID 25804924
22. Webb AJS, et al. Aortic stiffness, pulse pressure, and cerebral pulsatility progress despite best medical management: the OXVASC cohort. Stroke. 2022;53:1310-1317. PMID 34852644
23. Yang S, Webb AJS. Associations between neurovascular coupling and cerebral small vessel disease: a systematic review and meta-analysis. Eur Stroke J. 2023;8:895-903. PMID 37697725
24. Bell RD, et al. Apolipoprotein E controls cerebrovascular integrity via cyclophilin A. Nature. 2012;485:512-6. PMID 22622580
25. Montagne A, et al. APOE4 leads to blood-brain barrier dysfunction predicting cognitive decline. Nature. 2020;581:71-76. PMID 32376954
26. Walsh J, et al. Microglial activation and blood-brain barrier permeability in cerebral small vessel disease. Brain. 2021;144:1361-1371. PMID 34000009
27. Kisler K, et al. Pericyte degeneration leads to neurovascular uncoupling and limits oxygen supply to brain. Nat Neurosci. 2017;20:406-416. PMID 28135240
28. Miners JS, et al. Differing associations between Aβ accumulation, hypoperfusion, blood-brain barrier dysfunction and loss of PDGFRB pericyte marker in the precuneus and parietal white matter in Alzheimer's disease. J Cereb Blood Flow Metab. 2018;38:103-115. PMID 28151041
29. Montagne A, et al. Blood-brain barrier breakdown in the aging human hippocampus. Neuron. 2015;85:296-302. PMID 25611508
30. Claassen JAHR, et al. Regulation of cerebral blood flow in humans: physiology and clinical implications of autoregulation. Physiol Rev. 2021;101:1487-1559. PMID 33769101
31. Santisteban MM, et al. The pathobiology of neurovascular aging. Neuron. 2025;113:49-70. PMID 39788087