Human imaging extends prior vascular findings
APOE-stratified BBB imaging combines with fluid biomarkers and cognitive follow-up to extend earlier human BBB observations beyond unstratified associations.
↳ Introduction, refs 7–8; Results, Figs. 1–3
Vascular contributions to dementia and Alzheimer's disease are increasingly recognized1-6. Recent studies have suggested that breakdown of the blood-brain barrier (BBB) is an early biomarker of human cognitive dysfunction7, including the early clinical stages of Alzheimer's disease5,8-10. The E4 variant of apolipoprotein E (APOE4), the main susceptibility gene for Alzheimer's disease11-14, leads to accelerated breakdown of the BBB and degeneration of brain capillary pericytes15-19, which maintain BBB integrity20-22. It is unclear, however, whether the cerebrovascular effects of APOE4 contribute to cognitive impairment. Here we show that individuals bearing APOE4 (with the ε3/ε4 or ε4/ε4 alleles) are distinguished from those without APOE4 (ε3/ε3) by breakdown of the BBB in the hippocampus and medial temporal lobe. This finding is apparent in cognitively unimpaired APOE4 carriers and more severe in those with cognitive impairment, but is not related to amyloid-β or tau pathology measured in cerebrospinal fluid or by positron emission tomography23. High baseline levels of the BBB pericyte injury biomarker soluble PDGFRβ7,8 in the cerebrospinal fluid predicted future cognitive decline in APOE4 carriers but not in non-carriers, even after controlling for amyloid-β and tau status, and were correlated with increased activity of the BBB-degrading cyclophilin A-matrix metalloproteinase-9 pathway19 in cerebrospinal fluid. Our findings suggest that breakdown of the BBB contributes to APOE4-associated cognitive decline independently of Alzheimer's disease pathology, and might be a therapeutic target in APOE4 carriers.
Our findings suggest that breakdown of the BBB contributes to APOE4-associated cognitive decline independently of Alzheimer's disease pathology, and might be a therapeutic target in APOE4 carriers.
observational associations do not identify a causal contribution independent of AD pathology
High baseline levels of the BBB pericyte injury biomarker soluble PDGFRβ7,8 in the cerebrospinal fluid predicted future cognitive decline in APOE4 carriers but not in non-carriers, even after controlling for amyloid-β and tau status
unexamined selection into longitudinal follow-up and uncorrected testing materially limit the predictive evidence
individuals bearing APOE4 (with the ε3/ε4 or ε4/ε4 alleles) are distinguished from those without APOE4 (ε3/ε3) by breakdown of the BBB in the hippocampus and medial temporal lobe
cross-sectional genotype comparisons support the regional association within a selected imaging subset
The BBB dysfunction (Fig. 1c,d,l,m) preceded brain atrophy (Fig. 1j,k) and was independent of systemic vascular risk factors (Extended Data Fig. 4).
cross-sectional cognitive-status comparisons cannot establish within-person temporal ordering
living APOE4 carriers, but not APOE3 homozygotes, develop an increase in CypA CSF levels with cognitive impairment (Fig. 4h,i), which correlated with elevated CSF sPDGFRβ (Fig. 4j)
cross-sectional CSF comparisons and correlations support the association without establishing its mechanism
Derived from the full evaluation — not a separate score.
Strengths
APOE-stratified BBB imaging combines with fluid biomarkers and cognitive follow-up to extend earlier human BBB observations beyond unstratified associations.
↳ Introduction, refs 7–8; Results, Figs. 1–3
MRI acquisition uses standardized protocols and phantom calibration across sites. The tau PET analysis additionally subtracts choroid plexus overlap to address off-target hippocampal signal.
↳ Methods, Magnetic Resonance Imaging and Analysis; Positron Emission Tomography and Analysis
Mixed models use baseline sPDGFRβ continuously and adjust for amyloid and tau status. The median split is explicitly restricted to visualization rather than statistical testing.
↳ Fig. 3; Methods, Baseline CSF sPDGFRβ as a Continuous Predictor of Cognitive Decline
Limitations
The claim that BBB breakdown contributes to cognitive decline is not established by observational associations. The separate assertion that BBB dysfunction preceded atrophy comes from cross-sectional cognitive-status comparisons rather than longitudinal imaging trajectories.
↳ Main text, paragraph 3; Abstract and closing summary
DCE-MRI covers 245 of 435 participants, and longitudinal cognition covers 146 of 350 CSF participants without examination of the resulting selection mechanisms. Amyloid PET tracers and different cognitive instruments are also pooled without an assessment of their measurement comparability.
↳ Methods, Study Participants, PET Analysis and Cognitive Domain Impairment Evaluation; Fig. 3
Dichotomized CSF adjustment, small PET subsets and uncorrected secondary analyses limit the independence claim. PET totals are reported as 74 amyloid and 96 tau participants, but the overlap breakdown implies 67 and 87, leaving the applicable populations unclear.
↳ Results, Fig. 2; Methods, Study Participants, PET Analysis and Multiple Comparison Correction
The contribution lies in combining APOE-stratified human BBB imaging with fluid biomarkers and subsequent cognitive assessments. Blinded analyses, MRI calibration, choroid plexus correction and continuous-predictor mixed models support the observational findings. The principal constraint is inferential: neither a causal contribution to decline nor temporal precedence over atrophy follows from the reported comparisons. Unexamined subset selection and pooled measurements add uncertainty, while the discussion does not fully trace these boundaries through its interpretation.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
APOE-stratified BBB imaging and longitudinal sPDGFRβ analysis meaningfully extend earlier human BBB findings. The advance is an integrated observational association rather than a resolution of causal independence from AD pathology.
Blinding, MRI harmonization, artefact correction and continuous-predictor modelling support the association analyses, but causal contribution and temporal precedence exceed what their designs establish. Unexamined selection and measurement pooling further limit execution, while unresolved PET counts are separately surfaced through the Amber reliability flag.
The progression from imaging to pathology measures, longitudinal prediction and mechanistic correlates is followable. Methods distinguish continuous statistical modelling from median-split visualization, although nomenclature and the separation of observation from interpretation are uneven.
The paper connects prior human BBB observations with animal and pericyte mechanisms and explains why an unvalidated biomarker cutoff motivated continuous modelling. It does not comparably trace how selected cohorts and cross-sectional imaging constrain the interpretation.
Caveats4 of 4 checks
One reported value does not reconcile, but the key claims do not rest on it.
Ethics approval, informed consent and data availability are declared. Two typographical or labelling slips do not change the reported conclusions.
Ethics approval: declared · Data availability: declared
58 references in manuscript 55 of 58 checkable references found in an index
References not verified automatically are resolved manually within 24 hours.
No retraction or expression of concern found in Crossref (includes Retraction Watch).
Sources: Crossref (incl. Retraction Watch) ✓
The primary claim is supported by multi-site longitudinal human cohort data with real-world cognitive decline follow-up, though evidence is observational and from a single investigator group.
“might be a therapeutic target in APOE4 carriers”
The paper names a possible real-world use, targeting the BBB pathway therapeutically and using sPDGFRβ to predict decline, but names no actor who would act on the findings.
Drug developers and dementia clinicians could use the CypA-MMP9 pathway as a therapeutic target and CSF sPDGFRβ as a prognostic biomarker for APOE4 carriers.
Causal evidence is observational, the optimal sPDGFRβ cutoff is unknown, and no clinical trial has tested CypA inhibitors for APOE4-related cognitive decline.
Evaluated 16 Nov 2026 · GBD 2023 · glm_v1_3_3 · impact_profile_v1_3_3
AI-generated, human-governed. Something look off? Contact us to request a review.
Keep reading
Other papers Nabu has evaluated blind, starting with the same field.
McQuaid, Sewell, Montagne · 2026 · Basic medicine (neurosciences)
glial-vascular interactions within the NGVU serve as both essential physiological mechanisms and critical pathological contributors across the spectrum of NDDs.
TentativeLaing, Fialova, Wardlaw, et al. · 2026 · Clinical medicine
APOE4 mice showed a consistent reduction in CBF associated with APOE4 genotype (SMD = -2.87, 95% CI: -5.14 to -0.604, df = 2.66)
TentativeDomocos, Bragin, Shanbhag, et al. · 2026 · Neurology and psychiatry
Hypoxia significantly reduced transendothelial electrical resistance (TEER), whereas PBM restored barrier function in endothelial monocultures and tri-cultures.
Supported