Biomarkers

TL;DR — No blood or CSF biomarker is validated to diagnose vascular dementia or quantify its causal share in mixed disease. Neurofilament light (NfL) reflects axonal injury but is elevated across neurodegenerative, inflammatory, and vascular disorders (Zhao 2019, PMID 31026486). Endothelial, inflammatory, coagulation, BBB, and extracellular-matrix markers are biologically relevant but systemically confounded (French 2025, PMID 40145342). The most useful near-term strategy is a multimodal panel that separates vascular lesion burden, active injury, and Alzheimer copathology, validated prospectively against cognition and function.

Why no diagnostic blood test yet

Liao et al. screened 2,911 records and meta-analysed circulating markers in SVD-related cognitive impairment; homocysteine and hs-CRP differed between SVD with versus without cognitive impairment, whereas Lp-PLA2 and NfL did not, illustrating that even replicated associations fail intended-use tests (Liao 2024, PMID 39839309). Endothelial-marker effects on cognition are small (β −0.04 to −0.02) and cross-sectional only (Beran 2024, PMID 39428719). Case-control CAA plasma panels with AUC 0.90 still compare Boston v2.0 cases with healthy controls, not memory-clinic differentials (Muir 2025, PMID 40156276).

Intended-use framework

Use	Required evidence
Diagnosis	discrimination from mimics in target clinic population
Etiologic attribution	incremental value beyond MRI and AD biomarkers
Prognosis	calibrated longitudinal prediction
Trial enrichment	identifies mechanism-responsive subgroup
Pharmacodynamic	changes with target engagement
Surrogate endpoint	intervention effect on marker predicts clinical benefit

A marker can succeed for one use and fail for another. Association with WMH does not make it diagnostic; treatment responsiveness does not automatically make it a surrogate.

Candidate classes

Class	Examples	Principal limitation
Axonal injury	NfL	nonspecific and age/renal effects
Glial injury	GFAP, S100B	shared across diseases
Endothelial activation	ICAM-1, VCAM-1, selectins	systemic vascular inflammation
BBB integrity	albumin quotient, DCE leakage, pericyte markers	protocol and systemic confounding
Coagulation/fibrinolysis	D-dimer, fibrin markers	acute illness and thrombosis
Inflammation	cytokines, CRP	low brain specificity
Matrix remodeling	MMPs and inhibitors	temporal variability
Amyloid/tau	plasma/CSF/PET measures	identifies copathology, not vascular burden

Neurofilament light

NfL is released with axonal injury and is measurable in CSF and blood. Network meta-analysis of 42 case-control studies across neurodegenerative dementias supports elevation but also demonstrates lack of etiologic specificity (Zhao 2019, PMID 31026486). Liao et al. meta-analysed circulating candidates in SVD with cognitive impairment and retained homocysteine, hs-CRP, Lp-PLA2, and NfL as possible markers; Hcy and hs-CRP were higher in SVD-CI than SVD without CI, whereas Lp-PLA2 and NfL differences were not statistically significant between those two SVD groups (Liao 2024, PMID 39839309). In VCI, NfL may index total tissue injury or prognosis; it cannot distinguish infarction, SVD, Alzheimer disease, and other axonopathies alone.

Within small-vessel disease specifically, however, NfL behaves more usefully than the null in cognitive-impairment comparison suggests. In 496 non-demented ADNI participants (387, 78%, with longitudinal measurement), plasma NfL was higher in the moderate-severe SVD-burden group than in the no-burden group (45.2 ± 16.0 vs 34.3 ± 15.1 pg/mL; OR 1.71, 95% CI 1.24–2.35), and was associated with microbleeds (OR 1.29, 1.01–1.64), lacunar infarcts (OR 1.43, 1.06–1.93) and moderate-severe WMH (OR 1.67, 1.24–2.25). Longitudinally, a higher rate of change in NfL predicted progression of total SVD burden (OR 1.38, 1.08–1.76), WMH (OR 1.41, 1.10–1.79) and lacunes (OR 1.99, 1.42–2.77) (Qu 2021, PMID 33517704). NfL level therefore marked baseline burden while change tracked subsequent progression; the study did not formally establish that change outperformed level.

Among 5,270 UK Biobank participants aged 40–60 at baseline with plasma biomarkers and MRI a mean 9 years later, higher baseline GFAP predicted all three SVD markers — WMH volume (β 0.06, 95% CI 0.01–0.10, P=0.014), fractional anisotropy (β 0.08, 0.03–0.13, P=0.001) and mean diffusivity (β 0.14, 0.09–0.18, P<0.001) — and, in 1,317 with repeat MRI, predicted 3-year progression of FA (β 0.012, 0.001–0.023) and MD (β 0.020, 0.003–0.038). NfL was not significantly associated with any SVD marker, and longitudinal changes in either biomarker did not track concurrent MRI progression (Prapiadou 2026, PMID 41461058). Read with Qu, these findings support GFAP as a candidate risk marker and NfL as a candidate burden/progression marker. Because baseline MRI was unavailable, the temporal design does not show that astroglial activation preceded existing white-matter injury.

Beran et al. pooled 9,414 participants from eight Dutch cohorts: a composite of sICAM-1, sVCAM-1, and soluble E-selectin associated cross-sectionally with worse executive function, processing speed, delayed memory, and attention (pooled β range −0.04 to −0.02) but not with longitudinal cognitive change (Beran 2024, PMID 39428719). Muir et al. found lower plasma Aβ42/40 and higher p-tau-181 and NfL in 45 Boston v2.0 CAA cases versus 47 controls; the three-marker AUC was 0.90 (0.80–0.95), which is case-control discrimination, not clinic diagnostic validity (Muir 2025, PMID 40156276).

Endothelial and BBB markers

BBB/neurovascular-unit reviews catalog imaging and fluid candidates, but analytical platforms, timing, and populations differ (French 2025, PMID 40145342). Hypertension, diabetes, kidney disease, infection, and medication can change circulating endothelial markers independently of brain injury.

"BBB breakdown" is not one measurement

Three studies show that the phrase covers at least two physically different quantities that behave differently.

Using CSF soluble PDGFRβ as a pericyte-injury marker plus DCE-MRI, individuals with early cognitive dysfunction showed hippocampal capillary damage and BBB breakdown irrespective of amyloid and tau biomarker status — i.e. barrier failure is an early event that is not downstream of Alzheimer pathology (Nation 2019, PMID 30643288). A subsequent study separated the barrier by molecular scale: in 33 MCI patients and 22 controls, BBB permeability to water (measured with water-extraction phase-contrast arterial spin tagging) was increased in MCI and tracked CSF Aβ and p-tau, whereas permeability to albumin (CSF/serum albumin ratio) was not increased and tracked vascular risk factors — hypercholesterolaemia in particular. Only small-molecule permeability predicted cognitive performance (Lin 2021, PMID 34041783). A negative albumin quotient therefore does not exclude barrier dysfunction, and the two measures should not be treated as interchangeable in a panel.

At scale the albumin quotient nonetheless carries a disease signal, and its pattern of associations is informative about what it may measure. Across 1,861 patients — including 255 vascular dementia, 362 mixed AD/VaD, 666 late-onset and 130 early-onset Alzheimer disease, 50 dementia with Lewy bodies, 56 frontotemporal dementia — and two control groups (n=292 and n=20), the CSF/serum albumin ratio was elevated in late-onset Alzheimer disease, vascular dementia, mixed dementia, DLB and unspecified dementia. It correlated with CSF NfL in the late-onset AD, mixed, vascular and "other" groups but with no Alzheimer biomarker in any group (Skillbäck 2017, PMID 28779628). The authors proposed that barrier leakage may mark concomitant cerebrovascular pathology rather than being intrinsic to Alzheimer disease. This makes an elevated ratio a candidate clue to vascular copathology, not a validated positive indicator.

Barrier failure and hypoperfusion also travel together and steepen toward lesions: in 27 patients with SVD scanned with both dynamic susceptibility contrast and DCE, cerebral blood flow fell by 2.2 mL/min/100 g and leakage volume rose by 0.7% per millimetre closer to a WMH, with CBF and leakage inversely correlated in both normal-appearing white matter and WMH (−0.53 < r < −0.40, P<0.05), the correlation strengthening near the lesion (Wong 2019, PMID 30867275). Whether perfusion failure causes leakage, leakage causes perfusion failure, or both are outputs of one neurovascular-unit failure is not resolved by these data.

Because the effect being measured in SVD is subtle, quantification is where most of the between-study disagreement originates. The HARNESS Initiative consensus statement reviews DCE-MRI acquisition, model choice and analysis for subtle leakage and issues recommendations specifically for investigators new to the measurement (Thrippleton 2019, PMID 31031101); results obtained outside those recommendations are not straightforwardly comparable.

The inflammatory (Binswanger) phenotype

The most developed attempt to define a mechanism-coherent VCI subgroup by biomarker rather than by syndrome comes from the Binswanger-disease programme. In 62 VCID patients with clinical, multimodal MRI and CSF measures at entry and long-term diagnostic follow-up, used as a training set with 43 further patients as a test set, N-acetylaspartate on ¹H-MRS was the single best classifying variable, followed by CSF matrix metalloproteinase-2 and BBB permeability on MRI; logistic regression and random forest outperformed both a clinical subcortical-ischaemic-vascular-disease score and exploratory factor analysis, and 2-year follow-up confirmed the statistically derived classifications (Barry Erhardt 2019, PMID 29513153). The mechanistic argument behind it is that MMPs both open the barrier and degrade myelin, producing the two defining pathological features, with cytokines and angiogenic growth factors as accompanying CSF signals (Rosenberg 2022, PMID 36148658). The underlying measurements are specific enough to be checked. In 60 patients with suspected VCI (25 classified as subcortical ischaemic vascular disease, the rest mixed AD/VCI, multiple strokes or uncertain leukoaraiosis), CSF MMP-2 index was reduced across the whole patient group versus spinal-anaesthesia controls (P<0.001) while MMP-3 activity was increased (P<0.01); in SIVD specifically the MMP-2 index correlated negatively with the albumin index — an association absent for MMP-9 — and combining MMP-2 index with MMP-3 activity separated SIVD from controls with high specificity (P<0.0005) (Candelario-Jalil 2011, PMID 21454822). The direction is counterintuitive and matters: the enzyme index that tracks barrier opening is lower, not higher, in the CSF, consistent with consumption or sequestration at the vessel wall rather than with simple overproduction. Any trial using an MMP measure as a pharmacodynamic readout needs to specify which MMP and in which direction benefit would be expected.

This is the clearest existing template for biomarker-defined trial enrichment in this condition — and it remains single-programme work, not externally replicated at scale.

Confounder	Biomarker distortion
Age	baseline NfL and vascular markers rise
Kidney function	changes protein clearance
Recent stroke	acute injury dominates chronic signal
Infection	raises inflammatory markers
Diabetes	systemic endothelial activation
Exercise/trauma	can alter injury markers
Assay platform	shifts absolute thresholds

MarkVCID: the only systematic biomarker-qualification programme in this field

Where the fluid-biomarker literature is a scatter of single-cohort associations, MarkVCID is a NINDS/NIA consortium of seven project sites explicitly built to take candidate VCID biomarkers through staged qualification. It selected 11 candidate biomarker "kits" — each kit being the measure plus the clinical and cognitive data needed to validate it — and defined five intended uses in advance: risk identification, diagnosis of small-vessel pathology, prognostic stratification, target engagement, and progression monitoring (Wilcock 2021, PMID 33480172). Its staged model is worth stating explicitly because most VCID biomarker papers skip a stage: instrumental validation (does the measure reproduce across sites and scans?) precedes biological validation (does it relate to cognition?), which precedes clinical validation (does it work for a stated use in a trial?).

Kit	Instrumental validation	Biological validation	Stage reached
PSMD (peak-width of skeletonized mean diffusivity)	robust across sites (Maillard 2022, PMID 35382232)	higher PSMD → lower general cognition, β −0.8 (95% CI −1.2 to −0.4), in 396 MarkVCID-1 participants, replicated in CHARGE (n=6,172), Rush (n=287), UC Davis (n=567); explains cognition beyond WMH (Luckey 2024, PMID 39569745)	awaiting clinical validation
MRI free water	validated across MarkVCID kits (Maillard 2022, PMID 35382232)	higher mean free water → lower executive composite in three sub-cohorts and five legacy cohorts, and predicted accelerated executive decline over ~1.29 years (P=0.0026) (Maillard 2022, PMID 36523847)	biological validation with a longitudinal signal
Cerebrovascular reactivity MRI (5% CO₂)	harmonized identical procedure at three sites	prespecified hypothesis confirmed at all three sites in 263 subjects: CVR positively associated with MoCA and with an IRT executive score (Liu 2024, PMID 38951718)	biological validation, prespecified
WMH growth/regression protocol	test-retest ICC 0.969 (growth) and 0.937 (regression); cross-site inter-rater ICC 0.995 and 0.990	word-list long-delay free recall associated with WMH growth (P<0.028) but not with regression (Bahrani 2023, PMID 37840499)	instrumental validation excellent; construct validity partial

Three points deserve emphasis. First, PSMD and free water both outperform WMH volume, the marker on which nearly all clinical practice and most trial designs rest — PSMD explains cognitive status above and beyond WMH (Luckey 2024, PMID 39569745), and free water predicts decline rather than merely correlating with a cross-section (Maillard 2022, PMID 36523847). Second, the WMH growth/regression kit's near-perfect reliability alongside only partial construct validity is the clearest demonstration in this knowledge base that reproducibility and meaningfulness are independent properties. Third, that WMH regression has no cognitive correlate while growth does implies WMH change is not a single reversible quantity — a direct problem for any trial that would use WMH volume change as its primary endpoint (see OQ-7).

Alzheimer plasma markers in vascular disease: which one reads which pathology

The arrival of plasma p-tau217, GFAP, and NfL has partially solved the attribution problem — not by measuring vascular disease directly, but by letting vascular and Alzheimer contributions be read off different analytes.

Analyte	Tracks	Evidence
NfL	small-vessel-disease burden	strongest associations with WMH volume, mean skeletonized mean diffusivity, and fiber density in 76 memory-clinic patients, validated in 41 ADNI participants with amyloid/tau PET (Dewenter 2026, PMID 42252508); positively associated with WMH and white-matter free water in 251 MESA participants (Lockhart 2026, PMID 41630612)
p-tau217	Alzheimer-related neurodegeneration	strongest associations with AD-signature cortical thickness and fiber-bundle cross-section (Dewenter 2026, PMID 42252508); associated with amyloid PET, and also with white-matter free water (Lockhart 2026, PMID 41630612)
GFAP	both	linked to AD-related neurodegeneration and to SVD burden in memory-clinic patients (Dewenter 2026, PMID 42252508); positively associated with WMH in MESA (Lockhart 2026, PMID 41630612)
Aβ42/40	amyloid, and microbleeds	lower ratio associated with amyloid PET deposition and with the presence of microbleeds (Lockhart 2026, PMID 41630612)

The cognitive dissociation follows the biomarker dissociation: SVD markers were associated with processing speed, AD markers with memory (Dewenter 2026, PMID 42252508). The practical warning from both studies is the same and is not yet reflected in clinical use of these assays — coexisting small-vessel disease shifts plasma Alzheimer biomarker values, so an elevated p-tau or GFAP in a patient with heavy WMH burden should not be read as a pure amyloid signal. That NfL is the best available vascular plasma marker and is simultaneously the least etiologically specific analyte in neurology (Zhao 2019, PMID 31026486) remains the central irony of this field.

Imaging biomarkers

MRI remains the most established vascular biomarker family: lesion counts, WMH volume, diffusion measures, perfusion, BBB permeability, and network metrics (Duering 2023, PMID 37236211). STRIVE-2 provides terminology but does not certify any measure as a treatment surrogate.

CSF markers that separate vascular from Alzheimer dementia

Most CSF work in this field measures Alzheimer analytes and infers the vascular contribution by subtraction. Extracellular-matrix proteoglycans invert that logic. Quantifying 20 proteolytic peptides from brevican (9) and neurocan (11) by parallel-reaction-monitoring mass spectrometry in two cohorts (75 and 31 individuals), most peptide concentrations were decreased in vascular dementia relative to Alzheimer disease (AUC 0.83–0.93, P≤0.05) and relative to controls (AUC 0.79–0.87); two brevican peptides replicated in the second cohort (AUC 0.86–0.91). Critically, no peptide differed between Alzheimer disease and controls — so the signal is vascular-specific rather than a general dementia marker (Minta 2021, PMID 33337373). Cohort sizes are small (10 and 14 VaD cases) and no memory-clinic differential validation exists; the direction of the finding — matrix degradation products lower in the disease characterized by matrix-degrading enzyme activity — is itself worth resolving.

The retina as an accessible microvascular window

The retinal microvasculature is developmentally related to the cerebral microvasculature and is the only capillary bed that can be imaged non-invasively at capillary resolution. In 123 minor-stroke patients (mean age 68.1 ± 9.9; 115 with usable retinal data) imaged with OCT angiography, lower retinal vessel density remained associated after adjustment for age, eye disease, diabetes, blood pressure and image quality with higher mean diffusivity (β −0.16, 95% CI −0.32 to −0.01) and lower cerebrovascular reactivity (β 0.17, 0.03–0.31 left; 0.19, 0.02–0.36 right) in normal-appearing white matter; sparser retinal branching tracked higher MD (β −0.24, −0.08 to −0.40), lower fractional anisotropy (β 0.17, 0.01–0.33) and lower CVR (β 0.20, 0.02–0.38) (Wiseman 2023, PMID 36300327). The associations are with sub-visible white-matter damage and with vascular function, not with lesion counts — which is the strongest available argument that OCT-A is measuring microvascular biology rather than shared risk-factor burden. Effect sizes are small, eye disease is a competing cause in exactly the age group concerned, and no study has yet shown that a retinal measure adds prediction beyond MRI.

Multi-omics and machine learning

Proteomic, metabolomic, genomic, and retinal signatures may capture distributed vascular biology. A systematic review/meta-analysis of potential biomarkers for SVD with cognitive impairment illustrates many candidates but does not establish a clinical panel (Liao 2024, PMID 39839309). Endothelial-dysfunction meta-analysis similarly shows associations whose causal and diagnostic roles require prospective validation (Beran 2024, PMID 39428719).

Untargeted proteomics has now been run at a scale that produces usable prediction. Across 52,645 dementia-free UK Biobank participants with 1,417 incident cases over 14.1 years, four of 1,463 plasma proteins ranked highest for all-cause dementia, Alzheimer disease and vascular dementia: GFAP, NEFL, GDF15 and LTBP2. Combining GFAP with demographics gave AUC 0.891 for all-cause dementia and 0.872 for Alzheimer disease; GDF15 with demographics gave AUC 0.912 for vascular dementia — the highest of the three. Participants in the higher GFAP stratum were 2.32 times more likely to develop dementia, and GFAP and NEFL began diverging at least 10 years before diagnosis (Guo 2024, PMID 38347190). Two cautions travel with these numbers: AUCs computed with age and demographics included are dominated by those covariates, and the incident-VaD label in a biobank rests on administrative coding, not on the criteria examined in nosology.

Plasma markers also track imaging progression, not only cross-sectional burden. Across 1,240 MRIs from 473 patients with Alzheimer/MCI, Parkinson disease, frontotemporal disorders or cerebrovascular disease followed annually for 2 years, baseline GFAP, NfL, p-tau181 and p-tau217 (and, less strongly, lower Aβ42/40) predicted both baseline atrophy and WMH burden (|B| = 0.02–1.69) and their progression (|B| = 0.001–0.028), with models adjusted for age, sex, APOE, kidney function, vascular risk and disease duration (Sanchez 2025, PMID 40063856).

The renal confound: when it matters and when it does not

Because plasma NfL is renally cleared and rises with age alongside eGFR decline, the standing advice has been to adjust for kidney function. Two studies specify the condition more precisely. In 393 adults plus an independent 969-man twin sample, plasma NfL's associations with CSF NfL and with brain volumes did not meaningfully change after adjusting for creatinine; plasma NfL and creatinine were each heritable (h² = 0.54 and 0.60) and their correlation (r = 0.38) was partly shared-genetic (genetic correlation 0.46). The authors' conclusion is a rule rather than a caveat: adjustment is unnecessary for association analyses but necessary when a plasma value is compared with a cut-point for classification (Tang 2022, PMID 36527130). Consistently, in 242 amyloid-PET-characterized TRIAD participants, eGFR associated with most plasma biomarkers but the associations largely vanished after age and sex adjustment except for Aβ40, Aβ42, NfL and GFAP, and adding eGFR did not improve prediction of amyloid positivity for any biomarker (Arslan 2025, PMID 39972340). The unresolved edge is chronic kidney disease beyond mild impairment, which both studies under-sample and which is common in exactly the vascular-risk population this condition concerns.

Retina as an accessible vascular window

A systematic review of 48 studies (9,147 cases, 12,276 controls) across sporadic SVD, CAA, hypertensive arteriopathy, CADASIL, Fabry disease and MELAS found the most frequently replicated associations to be abnormal retinal vessel diameter (11 reports, n=11,391), increased vessel tortuosity (11 reports, n=617), decreased vessel fractal dimension (5 reports, n=1,597) and thinner retinal nerve fibre layer (5 reports, n=4,509). It also found no longitudinal retinal studies in SVD and no systematic evaluation of diagnostic performance (Biffi 2022, PMID 35421194) — a literature of associations without a single test-accuracy study, which is the same failure mode the endothelial-marker literature shows.

Validation checklist

• Prespecified intended use.
• Representative target population and clinically relevant controls.
• Blinded reference standard.
• Standardized preanalytics and assay.
• Age, renal, vascular, and acute-stroke adjustment.
• External geographic and ancestry validation.
• Incremental value beyond MRI and clinical variables.
• Calibration, not only AUC.
• Longitudinal clinical outcome.
• Decision-curve or treatment-selection evidence.

Mixed-pathology panel

Biological question	Candidate measure
Is there vascular structural injury?	MRI STRIVE markers
Is injury active/progressing?	serial MRI, NfL, recent infarct
Is BBB dysfunction present?	DCE/albumin quotient candidates
Is Alzheimer pathology present?	amyloid/tau biomarkers
Is CAA likely?	Boston v2.0 MRI plus emerging plasma markers (Charidimou 2022, PMID 35841910)
Does the marker matter clinically?	serial cognition and function

Emerging plasma studies distinguish CAA groups from controls, but case-control separation is not yet routine diagnostic validity (Muir 2025, PMID 40156276).

Open questions

• If plasma GFAP predicts small-vessel-disease markers nine years later while NfL does not, should the two analytes be assigned to different intended uses rather than compared head to head? (Prapiadou 2026, PMID 41461058; Qu 2021, PMID 33517704)
• Should an elevated CSF/serum albumin ratio in an amyloid-positive patient be reported as positive evidence of vascular copathology? (Skillbäck 2017, PMID 28779628)
• Why is the CSF MMP-2 index lower in subcortical ischaemic vascular disease when MMP activity is the proposed mechanism of barrier opening? (Candelario-Jalil 2011, PMID 21454822)
• Does OCT-angiography add anything to MRI for vascular cognitive risk, or only substitute for it where MRI is unavailable? (Wiseman 2023, PMID 36300327)
• Which biomarker panel adds calibrated prediction beyond MRI? (Liao 2024, PMID 39839309)
• Can BBB biomarkers identify a treatment-responsive phenotype? (French 2025, PMID 40145342)
• How should acute stroke timing be modeled in NfL trajectories? (Zhao 2019, PMID 31026486)
• Can biomarkers yield individual vascular versus Alzheimer attributable fractions? (Schneider 2007, PMID 17568013)
• Should PSMD or free water replace WMH volume as the default SVD trial endpoint, given that both explain cognition beyond it? (Luckey 2024, PMID 39569745; Maillard 2022, PMID 36523847)
• Why does WMH growth correlate with memory decline while WMH regression correlates with nothing? (Bahrani 2023, PMID 37840499)
• How much do plasma Alzheimer biomarker thresholds need to be adjusted for coexisting small-vessel disease? (Dewenter 2026, PMID 42252508; Lockhart 2026, PMID 41630612)
• Is GFAP's dual association with vascular and Alzheimer pathology a shared astrocytic mechanism or an assay-level confound? (Dewenter 2026, PMID 42252508)
• Are water permeability and albumin permeability measuring two distinct barrier failures, and should a panel contain both? (Lin 2021, PMID 34041783)
• Does barrier leakage cause hypoperfusion, does hypoperfusion cause leakage, or are both outputs of a single neurovascular-unit failure? (Wong 2019, PMID 30867275)
• Can the NAA/MMP-2/BBB-permeability classifier for the Binswanger phenotype be replicated outside its originating programme, and would it enrich a trial? (Barry Erhardt 2019, PMID 29513153)
• Why are brevican and neurocan degradation peptides lower in vascular dementia, when the mechanism invoked is increased matrix-degrading enzyme activity? (Minta 2021, PMID 33337373; Rosenberg 2022, PMID 36148658)
• Does GDF15 add anything to demographics for vascular-dementia prediction once age is properly accounted for? (Guo 2024, PMID 38347190)
• Does the "adjust only for cut-points" rule for renal function hold in CKD stages 4–5, which both validating studies under-sampled? (Tang 2022, PMID 36527130; Arslan 2025, PMID 39972340)
• Why has no longitudinal or test-accuracy study of retinal markers in small-vessel disease been done, given 48 cross-sectional reports? (Biffi 2022, PMID 35421194)

Related pages

• Pathophysiology — biological targets.
• Neuroimaging — imaging markers.
• Mixed pathology — attribution problem.
• Clinical trials — biomarker endpoints.

References

1. Zhao Y, et al. Neurofilament light chain in neurodegenerative dementia. Neurosci Biobehav Rev. 2019. PMID 31026486
2. French SR, et al. Biomarkers of BBB and neurovascular unit integrity. Alzheimers Dement. 2025. PMID 40145342
3. Duering M, et al. STRIVE-2. Lancet Neurol. 2023. PMID 37236211
4. Liao L, et al. Potential biomarkers for cerebral SVD with cognitive impairment. Front Aging Neurosci. 2024. PMID 39839309
5. Beran M, et al. Biomarkers of endothelial dysfunction and cognition. Alzheimers Dement. 2024. PMID 39428719
6. Muir RT, et al. Plasma biomarkers distinguish Boston Criteria 2.0 cerebral amyloid angiopathy from healthy controls. Alzheimers Dement. 2025. PMID 40156276
7. Schneider JA, et al. Mixed brain pathologies. Neurology. 2007. PMID 17568013
8. Charidimou A, et al. Boston criteria version 2.0 for CAA. Lancet Neurol. 2022;21:714-725. PMID 35841910
9. Wilcock D, et al. MarkVCID cerebral small vessel consortium: I. Enrollment, clinical, fluid protocols. Alzheimers Dement. 2021;17:704-715. PMID 33480172
10. Luckey AM, et al. Biological validation of peak-width of skeletonized mean diffusivity as a VCID biomarker: the MarkVCID consortium. Alzheimers Dement. 2024;20:8814-8824. PMID 39569745
11. Maillard P, et al. MRI free water as a biomarker for cognitive performance: validation in the MarkVCID consortium. Alzheimers Dement (Amst). 2022;14:e12362. PMID 36523847
12. Maillard P, et al. Instrumental validation of free water, peak-width of skeletonized mean diffusivity, and white matter hyperintensities: MarkVCID neuroimaging kits. Alzheimers Dement (Amst). 2022;14:e12261. PMID 35382232
13. Liu P, et al. Cerebrovascular reactivity MRI as a biomarker for cerebral small vessel disease-related cognitive decline: multi-site validation in the MarkVCID consortium. Alzheimers Dement. 2024;20:5281-5289. PMID 38951718
14. Bahrani AA, et al. Multi-site cross-site inter-rater and test-retest reliability and construct validity of the MarkVCID white matter hyperintensity growth and regression protocol. J Alzheimers Dis. 2023;96:683-693. PMID 37840499
15. Dewenter A, et al. Differential associations of plasma biomarkers with Alzheimer's disease and small vessel disease: a multimodal imaging study. Alzheimers Dement. 2026;22:e71530. PMID 42252508
16. Lockhart SN, et al. Plasma and neuroimaging biomarkers of small vessel disease and Alzheimer's disease in a diverse cohort: MESA. Alzheimers Dement. 2026;22:e71131. PMID 41630612
17. Nation DA, et al. Blood-brain barrier breakdown is an early biomarker of human cognitive dysfunction. Nat Med. 2019;25:270-276. PMID 30643288
18. Lin Z, et al. Blood-brain barrier breakdown in relationship to Alzheimer and vascular disease. Ann Neurol. 2021;90:227-238. PMID 34041783
19. Wong SM, et al. Blood-brain barrier impairment and hypoperfusion are linked in cerebral small vessel disease. Neurology. 2019;92:e1669-e1677. PMID 30867275
20. Thrippleton MJ, et al. Quantifying blood-brain barrier leakage in small vessel disease: review and consensus recommendations. Alzheimers Dement. 2019;15:840-858. PMID 31031101
21. Barry Erhardt E, et al. Biomarkers identify the Binswanger type of vascular cognitive impairment. J Cereb Blood Flow Metab. 2019;39:1602-1612. PMID 29513153
22. Rosenberg GA. Willis Lecture: biomarkers for inflammatory white matter injury in Binswanger disease provide pathways to precision medicine. Stroke. 2022;53:3514-3523. PMID 36148658
23. Minta K, et al. Brevican and neurocan peptides as potential cerebrospinal fluid biomarkers for differentiation between vascular dementia and Alzheimer's disease. J Alzheimers Dis. 2021;79:729-741. PMID 33337373
24. Guo Y, et al. Plasma proteomic profiles predict future dementia in healthy adults. Nat Aging. 2024;4:247-260. PMID 38347190
25. Sanchez E, et al. Association of plasma biomarkers with longitudinal atrophy and microvascular burden on MRI across neurodegenerative and cerebrovascular diseases. Neurology. 2025;104:e213438. PMID 40063856
26. Tang R, et al. Association of neurofilament light chain with renal function: mechanisms and clinical implications. Alzheimers Res Ther. 2022;14:189. PMID 36527130
27. Arslan B, et al. The impact of kidney function on Alzheimer's disease blood biomarkers: implications for predicting amyloid-β positivity. Alzheimers Res Ther. 2025;17:48. PMID 39972340
28. Biffi E, et al. Retinal biomarkers of cerebral small vessel disease: a systematic review. PLoS One. 2022;17:e0266974. PMID 35421194
29. Qu Y, et al. Association of plasma neurofilament light with small vessel disease burden in nondemented elderly: a longitudinal study. Stroke. 2021;52:896-904. PMID 33517704
30. Prapiadou S, et al. Association of plasma GFAP and NfL in middle-aged adults with MRI markers of cerebral small vessel disease later in life. Neurology. 2026;106:e214481. PMID 41461058
31. Skillbäck T, et al. CSF/serum albumin ratio in dementias: a cross-sectional study on 1861 patients. Neurobiol Aging. 2017;59:1-9. PMID 28779628
32. Candelario-Jalil E, et al. Matrix metalloproteinases are associated with increased blood-brain barrier opening in vascular cognitive impairment. Stroke. 2011;42:1345-50. PMID 21454822
33. Wiseman SJ, et al. Retinal capillary microvessel morphology changes are associated with vascular damage and dysfunction in cerebral small vessel disease. J Cereb Blood Flow Metab. 2023;43:231-240. PMID 36300327