Low frequency blood-oxygen-level-dependent oscillations,APOE4,and plasma pTau 217

Abstract

Background

Low frequency oscillations in blood-oxygen-level-dependent signal (BOLD-LFOs) are generally considered nuisance signal in connectivity analysis and discarded. However, recent evidence suggests BOLD-LFOs shed light on cerebrovascular dysfunction and preclinical Alzheimer's disease, but the mechanisms remain unclear. No investigations have assessed the relationship between BOLD-LFOs and plasma pTau₂₁₇, or how it differs in apolipoprotein ε4 (APOE4) carriers who are vulnerable to cerebrovascular dysfunction and genetically predisposed to AD.

Objective

To study the relationship between BOLD-LFOs and plasma p-Tau₂₁₇ in APOE4 carriers compared to non-carriers.

Methods

Independently living older adults (N = 118) were recruited and underwent resting-state fMRI and venipuncture. BOLD-LFOs were quantified as signal power within the 0.01–0.10 Hz frequency range. Plasma pTau₂₁₇ was assessed and linear regression quantified the interactive effect of APOE4 carrier status and BOLD-LFOs on plasma pTau₂₁₇. 2×2 ANCOVA was used to compare BOLD-LFOs across APOE4 carrier and amyloid positivity statuses based on previously reported pTau₂₁₇ cutoffs.

Results

The interactive effect of APOE4 carrier status and BOLD-LFO power was significantly associated with plasma pTau₂₁₇ (β = -0.78, p = 0.001). This relationship was driven by an inverse relationship between BOLD-LFOs and plasma pTau₂₁₇ in APOE4 carriers (β = −0.57, p = 0.0007). Amyloid-β (+) APOE4 carriers displayed lower BOLD-LFOs than amyloid-β (-) APOE4 carriers (p = 0.008) and amyloid-β (+) non-carriers (p = 0.03). Models were adjusted for age, sex, vascular risk factors, and total intracranial volume.

Conclusions

Findings suggests BOLD-LFOs are implicated in preclinical AD in an APOE4 dependent manner, adding support for the continued study of BOLD-LFOs in the context of cerebrovascular contributions to AD genetic risk.

Keywords

Alzheimer's disease BOLD oscillations neuroimaging biomarker

Introduction

Spontaneous blood-oxygen-level-dependent signal low-frequency oscillations (BOLD-LFOs) in the 0.01–0.1 Hz frequency range were once considered nuisance signal in functional magnetic resonance imaging (fMRI) connectivity analysis, caused by cardiac pulsations and respirations.¹ However, more recent work has demonstrated that other neurophysiological processes of potential relevance to brain health contribute to BOLD-LFOs, including spontaneous cerebrovascular reactivity,^2,3 astrocyte-mediated vasomotion,^4–6 and neuron-astrocyte crosstalk.⁷ A growing body of evidence also indicates that there is disease-dependent variation in BOLD-LFOs,^8,9 suggesting changes in BOLD-LFOs may have relevance to the pathophysiology of some brain diseases and could hold unique diagnostic and clinical value relative to standard fMRI.⁹

For example, widespread dampening of the amplitude of BOLD-LFOs has been observed in Alzheimer's disease (AD) dementia patients,⁸ where it correlates specifically with neuronal hypometabolism on ¹⁸F fluorodeoxyglucose-PET (FDG-PET). Another recent study found that BOLD-LFOs are altered early in the preclinical stages of AD, even in the absence of macrostructural atrophy.⁹ This study also found that decreased amplitude of BOLD-LFOs correlates with the degree of diffuse cerebral amyloidosis and with accumulation of tau pathophysiological change in the entorhinal cortex.⁹ Together these findings suggest that BOLD-LFO power attenuation occurs in the early stages of AD pathophysiology and is related to key amyloid, tau, and neurodegeneration markers of the disease. However, no studies to date have investigated how BOLD-LFOs relate to AD pathophysiology in the context of the AD risk gene, apolipoprotein ε4 (APOE4). Finally, no studies have examined whether the amplitude of BOLD-LFOs are related to blood-based biomarkers of AD, including plasma pTau₂₁₇.

The present study aims to address this knowledge gap by investigating the relationship between BOLD-LFOs and plasma pTau₂₁₇, a reliable diagnostic and prognostic marker of AD progression,¹⁰ in older APOE4 carriers compared to non-carriers to test the hypothesis that the relationship between BOLD-LFOs and pTau₂₁₇ differs by APOE4 status. The present investigation also examined changes in BOLD-LFOs associated with early AD pathophysiological change by comparing BOLD-LFO power in APOE4 carriers who have not yet developed pTau₂₁₇ abnormality to APOE4 carriers who have. Elevated p-Tau₂₁₇ closely correlates with the presence of amyloid-β plaques and is associated with early amyloid changes.^10,11 Therefore, the pTau₂₁₇ abnormality cutoff used in the present study was based on a previous investigation identifying a single cutoff with high sensitivity and specificity to cerebral amyloidosis confirmed by cerebrospinal fluid (CSF) amyloid-β measures, enabling the comparison of BOLD-LFOs between groups categorized as either amyloid-β positive or negative.

Methods

Participants

Participants were recruited from Los Angeles County and Orange County communities through outreach events, mailing lists, word-of-mouth, online portals, a research volunteer registry, and through the local Alzheimer's Disease Research Center (ADRC). All procedures were conducted as part of the Vascular Senescence and Cognition (VaSC) Study at the University of Southern California (USC) and the University of California Irvine (UCI). Older adults aged 60 to 89 years who were living independently were included (Table 1). Study exclusions were a prior diagnosis of dementia, history of clinical stroke, family history of dominantly inherited neurodegenerative disorders, current neurological or major psychiatric disorders that may impact cognitive function, history of moderate-to-severe traumatic brain injury, active substance abuse, current use of medications impairing the central nervous system, current organ failure or other uncontrolled systemic illness, and contraindications for brain MRI. Eligibility for the study was verified by a structured clinical health interview and review of current medications with the participant and, when available, a knowledgeable informant study partner. This study was approved by the USC (HS-14-00784) and UCI (HS-2019-5324) Institutional Review Boards, and all participants gave informed consent in accordance with the Declaration of Helsinki. The anonymous data that support the findings of this study are available upon reasonable request from the corresponding author, DAN, through appropriate data sharing protocols.

Table 1.

Participant characteristics and demographics grouped by APOE4 carrier status.

Variable name	APOE4 non-carriers (n = 69)	APOE4 carriers (n = 49)	p
Age (years)	70.5±7.4 (60–89)	69.2±6.7 (60–89)	0.30^a
Sex (n, female %)	49 (71.0)	31 (63.3)	0.42^b
BOLD-LFOs (0.01–0.1 Hz power)	98.82±81.1	103.7±93.1	0.77^a
Plasma pTau₂₁₇ (pg/ml)	0.30±0.17	0.42±.27	0.01^a
Amyloid-β status^c (n, amyloid+ %)	12 (17.4)	17 (34.7)	0.04^b
≥2 Vascular Risk Factors (n, %)	36 (52)	24 (49)	0.85 ^b
APOE2/3 heterozygotes (n, %)	2 (3)	-	-
APOE3/3 homozygotes (n, %)	67 (97)	-	-
APOE2/4 heterozygotes (n, %)	-	2 (4)	-
APOE3/4 heterozygotes (n, %)	-	45 (92)	-
APOE4/4 homozygotes (n, %)	-	2 (4)	-

APOE4: apolipoprotein ε4 allele; BOLD-LFO: gray matter blood oxygen level dependent low frequency oscillations (0.01–0.1 Hz). ^aindependent t-test, ^bchi-squared test, ^camyloid-β status determined based on previously reported pTau₂₁₇ cutoff of 0.44 pg/ml which displays high combination sensitivity/specificity for detecting cerebral amyloidosis based on cerebrospinal fluid Aβ_42/40.¹⁴

Neuroimaging

All participants underwent brain MRI scans conducted on a 3T Siemens Prisma scanner with 20-channel head coil. High-resolution 3D T1-weighted anatomical (Scan parameters: TR = 2300 ms; TE = 2.98 ms; TI = 900 ms; flip angle = 9 deg; FOV = 256 mm; resolution = 1.0 × 1.0 × 1.2 mm3; Scan time = 9 min) images were acquired, using 3-dimensional magnetization-prepared rapid gradient-echo (MPRAGE) sequences. Resting state fMRI scans comprised 140 contiguous echo-planar imaging (EPI) functional volumes (TR = 3000 ms, TE = 30 ms, FA = 80°, 3.3 × 3.3 × 3.3 mm voxels, matrix = 64 × 64, FoV = 212 mm, 48 slices).

Resting-state functional MRI data were preprocessed using a custom Python pipeline that integrates FMRIB Software Library (FSL v6.0), NiBabel, and SciPy. Preprocessing was conducted on each participant's BOLD image series and corresponding T1-weighted anatomical scan.

Each participant's T1-weighted anatomical image was skull-stripped using FSL's BET and segmented into gray matter and white matter probability maps using the FAST tool. These maps were linearly registered to the participant's functional space using FLIRT,¹² and binary tissue masks were generated by thresholding the probabilistic segmentations at 0.5. The fully preprocessed functional images were then registered to MNI152 standard space (2 mm resolution) using affine registration with 12 degrees of freedom.

The first 10 volumes of each BOLD time series were discarded. Slice timing correction was applied to adjust for inter-slice acquisition delays, as well as motion correction using FSL's MCFLIRT tool to realign all volumes to reference image. After realignment, linear trends were removed from each voxel's time series using a Python-based detrending procedure implemented with SciPy. The detrended data were spatially smoothed with a Gaussian kernel (4 mm full width at half maximum, approximated by sigma = 1.7 voxels) and temporally band-pass filtered to extract frequencies between 0.01–0.10 Hz using FSL's fslmaths command.

Mean BOLD time series were extracted from the filtered functional images using the previously created gray matter mask. For each extracted time series, power spectral density (PSD) was calculated via fast Fourier transform, and total power within the 0.01–0.10 Hz frequency band was computed.

Routine nuisance regression of global signal, CSF signal, and motion parameters was not performed in order to avoid loss of hemodynamic information, as demonstrated elsewhere.^12,13

Plasma AD biomarkers

pTau₂₁₇ concentrations were quantified in human plasma samples using the ultra-sensitive Simoa (Single Molecule Array) assay platform developed by Quanterix (Lexington, MA, USA). The pTau₂₁₇ assay utilized a bead-based sandwich 3 step immunoassay format run on the Quanterix HD-X Analyzer, which allows for digital quantification of low-abundance biomarkers in biological fluids. Plasma samples were collected in K2EDTA tubes, centrifuged at 2000 x g for 10 min at 4°C, and stored at −80°C until analysis. Prior to assay, samples were thawed on ice and centrifuged again to remove any debris. The pTau₂₁₇ assay was performed according to the manufacturer's instructions (Quanterix pTau₂₁₇ Advantage Kit, Catalog # 104588), using 100 μL of plasma per well. Capture antibodies specific for pTau₂₁₇ were immobilized on paramagnetic beads, while detector antibodies were labeled with a proprietary enzyme tag. Following incubation and washing steps, individual immunocomplexes were transferred to a microwell array, allowing for single-molecule detection. A chemiluminescent substrate was added, and signals were captured digitally by the HD-X Analyzer.

A plasma pTau₂₁₇ cut off of 0.44 pg/ml has previously displayed high combination specificity and sensitivity (>85%) for detecting cerebral amyloidosis based on CSF Aβ_42/40.¹⁴This cutoff was used in the present study to categorize participants as either amyloid-β positive or negative.

APOE genotyping

Fasted blood samples were obtained by venipuncture and used to determine participant APOE genotype. Genomic DNA was extracted using the PureLink Genomic DNA Mini Kit (Thermo). APOE genotyping was performed as previously described.^15–17 APOE4 carriers were defined as participants with at least one copy of the APOE ε4 allele. All analyses were performed at the same lab at the University of Arizona (KR).

Vascular risk factors

Vascular risk factor (VRF) burden was determined through clinical interviews with the participant and informant (when available), and review of current medications and medical history. The assessed VRFs included history of cardiovascular disease (e.g., heart failure, angina, stent placement, coronary artery bypass graft, intermittent claudication), hypertension, hyperlipidemia, type 2 diabetes, atrial fibrillation, left ventricular hypertrophy, current smoking status, and transient ischemic attack. Hypertension was operationally defined as self-reported diagnosis of hypertension or taking antihypertensive medications. Type 1 diabetics were excluded, and type II diabetes was operationally defined as self-reported diagnosis at clinical interview. Total VRFs were summed for each participant and elevated VRF burden was defined as ≥2 VRFs (versus 0–1) as described previously.^18,19

Statistical analyses

121 participants underwent brain fMRI, anatomical T1w MRI, and venipuncture. All data were screened for outliers (±3 SD) and three gray matter BOLD-LFO outliers were identified and removed with values of +4.84 SD, +4.22 SD, and +3.79 SD. After outlier screen, the total analyzed sample was N = 118. All assessed variables were compared between APOE4 carriers and non-carriers using independent t-tests.

Data analyses were performed in R. Hayes PROCESS Model 1 (simple moderation) was used to assess the potential moderating effect of APOE4 carrier status on the relationship between BOLD-LFOs and plasma pTau₂₁₇ (primary analysis), where x = BOLD-LFO, Y = pTau₂₁₇, and w = APOE4 carrier status. Linear regression assumptions regarding linearity, multicollinearity (VIF<5), and homoscedasticity (Breusch-Pagan test) were met. Lastly, within group analyses were performed to assess the relationship between BOLD-LFOs and pTau₂₁₇ within the APOE4 carrier and non-carrier groups. A second multivariate model was ran adjusted for age, sex, vascular risk factor burden, and intracranial volume. Additional sensitivity analysis of the interactive effect was tested with APOE2 carriers (n = 4) excluded.

2X2 ANCOVA was used to compare BOLD-LFOs by amyloid-β positivity and APOE4 carrier status and pairwise comparisons were performed (secondary analysis). A second 2X2 ANCOVA was ran adjusted for age, sex, vascular risk factor burden, and intracranial volume.

For illustration purposes, power spectral density plots were generated to compare BOLD-LFOs power between amyloid-β (+) and amyloid-β (-) APOE4 carriers. A power spectral density plot is a frequency-domain graph that shows how the power of a signal is distributed across a frequency range, in this case, 0 to 0.15 Hz which represents the frequency band captured by a 3000 ms TR during fMRI.

Minimum detectable effect sizes given a power of 0.80, alpha of 0.05, two covariates, and a sample size of N = 118 was calculated using the pwr package in R as ΔR² = 0.05 for the BOLD-LFO*APOE4 interaction effect. False discovery rate correction was used to account for multiple comparisons in the primary analysis including APOE4*BOLD-LFOs interaction effect and subgroup analyses (within APOE4 carrier and non-carrier group effects).²⁰

Results

118 participants were included for analysis, participant characteristics and demographics for this sample are shown in Table 1 grouped by APOE4 carrier status.

The interactive effect of APOE4 carrier status and gray matter BOLD-LFOs was significantly associated with plasma pTau₂₁₇ in the univariate analysis (β = −0.59, 95% CI (−1.06, −0.12), p = 0.01) and also after adjustment for age, sex, vascular risk factor burden, and total intracranial volume (β = −0.78, 95% CI (−1.24, −0.32), p = 0.001 Table 2). This was driven by an inverse relationship between BOLD-LFOs and pTau₂₁₇ in APOE4 carriers (within-group) as shown in Figure 1.

Figure 1.

The relationship between 0.01–0.1 Hz gray matter BOLD low frequency oscillations (BOLD-LFO) and pTau₂₁₇ in APOE4 carriers (n = 49) compared to APOE3 homozygotes (n = 69). The effect of BOLD-LFOs on pTau₂₁₇ conditional upon APOE4 carrier status (interaction effect) is reported as standardized regression coefficient (β) and p-value in the total sample (N = 118) with and without covariate adjustment (age, sex, vascular risk factor (vrf) burden, and total intracranial volume (TIV)). Within group statistics reported in figure legend as β and p-value within APOE4 carrier status groups with and without covariate adjustment.

Table 2.

Regression coefficients for model predicting plasma pTau₂₁₇.

Variable name	β	95% CI	p
BOLD-LFOs (0.01–0.1 Hz power)	0.21	(−0.11, 0.53)	0.20
Sex (female)	0.04	(−0.53, 0.44)	0.86
Age	0.35	(0.18, 0.53)	0.0001
APOE4	0.51	(0.16, 0.86)	0.005
≥2 Vascular Risk Factors	−0.22	(−0.58, 0.13)	0.22
Total Intracranial Volume	0.21	(−0.003, 0.43)	0.05
BOLD-LFO*APOE4	−0.78	(−1.24, −0.32)	0.001

APOE4: apolipoprotein ε4 allele; β: standardized beta coefficient; BOLD-LFO: gray matter blood oxygen level dependent low frequency oscillations (0.01–0.1 Hz).

To visualize this effect across pTau₂₁₇ cutoffs and APOE4 carrier status groups, a 2X2 ANCOVA with pairwise comparisons was performed. The amyloid-β positivity status*APOE4 carrier status interaction was statistically significant (β = −0.91, 95% CI (−1.53, −0.29), p = 0.004), and remained so after adjustment for age, sex, vascular risk factor burden, and total intracranial volume, (β = −0.95, 95% CI (−1.64, −0.25), p = 0.008). In pairwise comparisons amyloid-β positive APOE4 carriers displayed lower BOLD-LFO power than amyloid-β positive APOE3 homozygotes (Δz = 0.61, 95% CI (0.08, 1.14), p = 0.03), and amyloid-β negative APOE4 carriers (Δz = 0.58, 95% CI (0.15, 1.00), p = 0.008) (Figure 2).

Figure 2.

Gray matter blood oxygen level dependent low frequency 0.01–0.10 Hz oscillations (BOLD-LFO) comparisons of APOE4 status (APOE3 homozygotes versus APOE4 carriers) by amyloid-β positivity status. 2X2 ANCOVA interaction effect p-value in the total sample (N = 118) is reported with and without covariate adjustment (age, sex, vascular risk factor (vrf) burden, and total intracranial volume (TIV)). Pairwise comparisons were conducted using unadjusted t-tests of group means. Amyloid-β status determined based on previously established pTau₂₁₇ cutoff of 0.44 pg/ml which displays high combination sensitivity/specificity for detecting cerebral amyloidosis based on cerebrospinal fluid Aβ_42/40.¹⁴.

A power spectral density plot is shown in Figure 3 to visualize the BOLD-LFO differences between amyloid-β positive and negative APOE4 carriers across the full frequency spectrum collected during fMRI. This graph is for illustrative purposes to visualize which frequency bands drive the previous finding that amyloid-β positivity status is related to BOLD-LFO power (<0.1 Hz) in APOE4 carriers. This finding appears to be driven by decreased power in the <0.05Hz frequency range as shown in Figure 3.

Figure 3.

Gray matter blood oxygen level dependent low frequency oscillations (BOLD-LFO) power spectral density plot. Comparisons across amyloid-β positivity status (amyloid-β positive = pTau₂₁₇>0.44 pg/ml) in APOE4 carriers. Amyloid-β positive n = 17, amyloid-β negative n = 32. Amyloid-β status determined based on previously reported pTau217 cutoff of 0.44 pg/ml which displays high combination sensitivity/specificity for detecting cerebral amyloidosis based on cerebrospinal fluid Aβ_42/40.¹⁴.

Sensitivity analyses and multiple comparisons

To account for potential residual confounding the interactive effect of APOE4 carrier status and gray matter BOLD-LFOs on plasma pTau₂₁₇ was tested after excluding participants with a copy of the APOE2 allele (n = 4) which did not attenuate the interactive effect of APOE4 carrier status and gray matter BOLD-LFOs on plasma pTau217 (β = −0.82, 95% CI (−1.32, −0.31) p = 0.002). The BOLD-LFO*APOE4 interaction term effects and subsequent within group analysis results survived correction for multiple comparisons.

Discussion

The present study finds that BOLD-LFO power is inversely associated with plasma pTau₂₁₇ levels in APOE4 carriers but not in non-carriers, suggesting a role for the APOE4 gene in modifying the relationship between BOLD-LFOs and AD pathophysiological change. Findings also suggest this alteration in BOLD-LFOs may be observed in preclinical AD, as evidenced by increased BOLD-LFO power in amyloid-β negative APOE4 carriers compared to amyloid-β positive APOE4 carriers. Together these results are consistent with prior work suggesting changes in BOLD-LFOs are related to AD pathophysiological changes measured by PET and CSF biomarkers9 and extend prior findings by demonstrating the relationship between decreased BOLD-LFOs and plasma pTau₂₁₇ in APOE4 carriers specifically.

APOE4 is associated with vascular, neuroinflammatory, and metabolic changes in neurons^21,22 and neuroglia.^23–26 One or more of these neurophysiological changes could be implicated in the relationship between BOLD-LFOs and pTau₂₁₇ in APOE4 carriers. For example, APOE4 conveys susceptibility to cerebrovascular dysfunction,^27–29 which could contribute to the observed decreased in BOLD-LFOs through decreased pulsatile or intrinsic vasomotion.^2,3 Astrocytes are also increasingly thought to be important contributors to BOLD-LFO signal^7,30 and APOE4 is associated with numerous changes to astrocyte function caused by an accumulation of lipid droplets and a buildup of unsaturated fatty acids within astrocytes.²³ This accumulation of lipid droplets is sufficient to induce astrocyte reactivity, triggering the secretion of inflammatory chemokines and cytokines.^31,32

An examination of BOLD-LFO differences in amyloid-β positive and amyloid-β negative APOE4 carriers in the present study suggests that the largest differences in power may exist in the lower frequency ranges, which are associated with astrocyte-mediated vasomotion,^4,5,7,30,33 particularly vasodilation.⁶ Thus, the observed relationship between reduced BOLD-LFOs and AD pathophysiology in APOE4 carriers may be related to changes in vascular function, astrocyte reactivity, or astrocyte-vascular interactions.³⁴ However, other systemic physiological processes like cardiac pulsations and respiratory dynamics contribute to BOLD-LFO signal, and further studies are needed to determine the specific mechanistic contributions to BOLD-LFOs that may be implicated in APOE4 associated AD pathophysiology.

This study highlights the relationship between cerebral hemodynamics and preclinical AD pathophysiology. Future studies should seek to determine if BOLD-LFOs represent a causal contribution to amyloid-β or tau pathology, or if the relationship is purely associative. If BOLD-LFOs represent a distinct mechanistic contribution to AD pathophysiology this would have significant implications for the study of cerebrovascular function in the context of AD. If instead, they are merely associated with early AD pathophysiological changes in APOE4 carriers then they may represent a novel imaging biomarker of AD pathophysiology. Strengths of the present investigation include the study of BOLD-LFOs in preclinical AD by comparing APOE4 carriers to non-carriers with and without pTau₂₁₇ abnormality. Limitations include the associative, cross-sectional nature of the study and the inference of cerebral amyloidosis status using a plasma pTau₂₁₇ cutoff rather than measuring it directly using Aβ PET, plasma, or CSF assay. The present study findings suggest that BOLD-LFO changes may be implicated in preclinical AD, particularly in APOE4 carriers.

Footnotes

Acknowledgements

The authors have no acknowledgments to report.

ORCID iDs

Trevor Lohman

Arunima Kapoor

Basant Lashin

Daniel A. Nation

Ethical considerations

This study was approved by the USC (HS-14-00784) and UCI (HS-2019-5324) Institutional Review Boards.

Consent to participate

All participants gave informed consent in accordance with the Declaration of Helsinki.

Consent for publication

All study participants consented to the publication of findings associated with the data generated by this research study.

Author contribution(s)

Trevor Lohman: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Visualization; Writing – original draft.

Arunima Kapoor: Data curation; Writing – review & editing.

Allison C. Engstrom: Data curation; Writing – review & editing.

Jillian Joyce: Data curation; Writing – review & editing.

Melanie Quiring: Data curation; Writing – review & editing.

John Paul M. Alitin: Data curation; Writing – review & editing.

Aimee Gaubert: Data curation; Writing – review & editing.

Amy Nguyen: Data curation; Writing – review & editing.

Elizabeth Head: Data curation; Methodology; Writing – review & editing.

Kathleen E. Rodgers: Data curation; Methodology; Writing – review & editing.

David Bradford: Data curation; Writing – review & editing.

Basant Lashin: Data curation; Writing – review & editing.

S. Duke Han: Conceptualization; Writing – review & editing.

Mara Mather: Conceptualization; Writing – review & editing.

Daniel A. Nation: Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Project administration; Resources; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Southern California Clinical and Translational Science Institute (TL: KL2TR001854), the National Institutes of Health grants (grant numbers DAN: R01AG064228, R01AG082073, P01AG052350, P30AG066530), (grant number SDH: K24AG081325), (grant number EH: P30AG066519), and the American Heart Association (grant number AK: 23PRE1014192). Foundation for the National Institutes of Health, (grant number R01AG060049).

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: TL, EH, and DN are Editorial Board Members of this journal but were not involved in the peer-review process of this article nor had access to any information regarding its peer-review.

Data availability statement

The anonymous data that support the findings of this study are available upon reasonable request from the corresponding author, DAN, through appropriate data sharing protocols.

References

Tong

Hocke

Frederick

. Low frequency systemic hemodynamic “noise” in resting state BOLD fMRI: characteristics, causes, implications, mitigation strategies, and applications. Front Neurosci 2019; 13: 787.

Drew

. Vascular and neural basis of the BOLD signal. Curr Opin Neurobiol 2019; 58: 61–69.

Liu

Lin

Hazel

, 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.

Haidey

Peringod

Institoris

, et al. Astrocytes regulate ultra-slow arteriole oscillations via stretch-mediated TRPV4-COX-1 feedback. Cell Rep 2021; 36: 109405.

Wang

Sejnowski

, et al. Brain-state dependent astrocytic Ca(2+) signals are coupled to both positive and negative BOLD-fMRI signals. Proc Natl Acad Sci U S A 2018; 115: E1647–E1656.

Chen

Volkow

, et al. Synchronized astrocytic Ca2+ responses in neurovascular coupling during somatosensory stimulation and for the resting state. Cell Rep 2018; 23: 3878–3890.

DiNuzzo

Gili

Maraviglia

, et al. Modeling the contribution of neuron-astrocyte cross talk to slow blood oxygenation level-dependent signal oscillations. J Neurophysiol 2011; 106: 3010–3018.

Kazemifar

Manning

Rajakumar

, et al. Spontaneous low frequency BOLD signal variations from resting-state fMRI are decreased in Alzheimer disease. PLoS One 2017; 12: e0178529.

Hrybouski

Das

Xie

, et al. BOLD Amplitude correlates of preclinical Alzheimer’s disease. Neurobiol Aging 2025; 150: 157–171.

10.

Lai

Bishnoi

. P-tau217 as a reliable blood-based marker of Alzheimer’s disease. Biomedicines 2024; 12: 1836.

11.

Ashton

Brum

Di Molfetta

, et al. Diagnostic accuracy of a plasma phosphorylated tau 217 immunoassay for Alzheimer disease pathology. JAMA Neurol 2024; 81: 255–263.

12.

Zhang

Wang

, et al. Reduced coupling between the global blood-oxygen-level-dependent signal and cerebrospinal fluid inflow is associated with the severity of small vessel disease. Neuroimage Clin 2022; 36: 103229.

13.

Fultz

Bonmassar

Setsompop

, et al. Coupled electrophysiological, hemodynamic, and cerebrospinal fluid oscillations in human sleep. Science 2019; 366: 628–631.

14.

Lehmann

Schraen-Maschke

Vidal

J-S

, et al. Clinical value of plasma ALZpath pTau217 immunoassay for assessing mild cognitive impairment. J Neurol Neurosurg Psychiatry 2024; 95: 1046.

15.

Kapoor

Gaubert

Marshall

, et al. Increased levels of circulating angiogenic cells and signaling proteins in older adults with cerebral small vessel disease. Front Aging Neurosci 2021; 13: 711784.

16.

Lohman

Sible

Kapoor

, et al. Blood pressure variability, central autonomic network dysfunction, and cerebral small-vessel disease in APOE4 carriers. J Am Heart Assoc 2024; 13: e034116.

17.

Lohman

Sible

Engstrom

, et al. Beat-to-beat blood pressure variability, hippocampal atrophy, and memory impairment in older adults. Geroscience 2025; 47: 993–1003.

18.

Ferrari-Souza

Brum

Hauschild

, et al. Vascular risk burden is a key player in the early progression of Alzheimer’s disease. Neurobiol Aging 2024; 136: 88–98.

19.

Nation

Sweeney

Montagne

, et al. Blood–brain barrier breakdown is an early biomarker of human cognitive dysfunction. Nat Med 2019; 25: 270–276.

20.

Benjamini

Hochberg

. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B Methodol 1995; 57: 289–300.

21.

Koutsodendris

Blumenfeld

Agrawal

, et al. Neuronal APOE4 removal protects against tau-mediated gliosis, neurodegeneration and myelin deficits. Nat Aging 2023; 3: 275–296.

22.

Fernández-Calle

Konings

Frontiñán-Rubio

, et al. APOE In the bullseye of neurodegenerative diseases: impact of the APOE genotype in Alzheimer’s disease pathology and brain diseases. Mol Neurodegener 2022; 17: 62.

23.

Sienski

Narayan

Bonner

, et al. APOE4 Disrupts intracellular lipid homeostasis in human iPSC-derived glia. Sci Transl Med 2021; 13: eaaz4564.

24.

Arnaud

Benech

Greetham

, et al. APOE4 Drives inflammation in human astrocytes via TAGLN3 repression and NF-κB activation. Cell Rep 2022; 40: 111200.

25.

Chen

Jin

Chen

, et al. The multifaceted roles of apolipoprotein E4 in Alzheimer’s disease pathology and potential therapeutic strategies. Cell Death Discov 2025; 11: 312.

26.

Lin

Seo

Gao

, et al. APOE4 Causes widespread molecular and cellular alterations associated with Alzheimer’s disease phenotypes in human iPSC-derived brain cell types. Neuron 2018; 98: 1141–1154. e1147.

27.

Montagne

Nation

Sagare

, et al. APOE4 Leads to blood-brain barrier dysfunction predicting cognitive decline. Nature 2020; 581: 71–76.

28.

Kaufman

Morris

Vidoni

, et al. Apolipoprotein E4 moderates the association between vascular risk factors and brain pathology. Alzheimer Dis Assoc Disord 2021; 35: 223–229.

29.

Onos

Lin

Pandey

, et al. Assessment of neurovascular uncoupling: APOE status is a key driver of early metabolic and vascular dysfunction. Alzheimers Dement 2024; 20: 4951–4969.

30.

Takata

Sugiura

Yoshida

, et al. Optogenetic astrocyte activation evokes BOLD fMRI response with oxygen consumption without neuronal activity modulation. Glia 2018; 66: 2013–2023.

31.

Kwon

Y-H

Kim

C-S

, et al. Hypothalamic lipid-laden astrocytes induce microglia migration and activation. FEBS Lett 2017; 591: 1742–1751.

32.

Tcw

Qian

Pipalia

, et al. Cholesterol and matrisome pathways dysregulated in astrocytes and microglia. Cell 2022; 185: 2213–2233. e2225.

33.

Tesler

Linne

M-L

Destexhe

. Modeling the relationship between neuronal activity and the BOLD signal: contributions from astrocyte calcium dynamics. Sci Rep 2023; 13: 6451.

34.

Serrano-Pozo

, et al. Astrocyte transcriptomic changes along the spatiotemporal progression of Alzheimer’s disease. Nat Neurosci 2024; 27: 2384–2400.