Aberrant vascular architecture in the hippocampus correlates with tau burden in mild cognitive impairment and Alzheimer’s disease

Abstract

Cerebrovascular dysfunction is a significant contributor to Alzheimer’s disease (AD) progression. AD mouse models show altered capillary morphology, density, and diminished blood flow in areas of tau and beta-amyloid accumulation. The purpose of this study was to examine alterations in vascular structure and their contributions to perfusion deficits in the hippocampus in AD and mild cognitive impairment (MCI). Seven individuals with AD and MCI (1 AD/6 MCI), nine cognitively intact older healthy adults, and seven younger healthy adults underwent pseudo-continuous arterial spin labeling (PCASL) and gradient-echo/spin-echo (GESE) dynamic susceptibility contrast (DSC) MRI. Cerebral blood flow (CBF), cerebral blood volume, relative vessel size index (rVSI), and mean vessel density were calculated from model fitting. Lower CBF from PCASL and SE DSC MRI was observed in the hippocampus of AD/MCI group. rVSI in the hippocampus of the AD/MCI group was larger than that of the two healthy groups (FDR-P = 0.02). No difference in vessel density was detected between the groups. We also explored relationship of tau burden from ¹⁸F-flortaucipir positron emission tomography and vascular measures from MRI. Tau burden was associated with larger vessel size and lower CBF in the hippocampus. We postulate that larger vessel size may be associated with vascular alterations in AD/MCI.

Keywords

Alzheimer’s disease hippocampus magnetic resonance imaging vascular alteration vessel size imaging

Introduction

Alzheimer’s disease (AD) is the most common form of dementia, characterized by memory loss, cognitive decline, and functional impairment.¹ Neuropathologically, AD is characterized by the presence of beta-amyloid plaques, tau neurofibrillary tangles, excessive inflammatory response, and cell death in the brain followed by neuronal loss.^2
–4 Cerebrovascular dysfunction also emerges in the early stages of AD before the onset of neurodegeneration and brain atrophy.^5
–7 The cerebrovascular system plays an important role in mediating brain activity, function, and energy metabolism and has been implicated in diverse ways in AD pathogenesis. In cerebral amyloid angiopathy (CAA), which is associated with AD, the accumulation of beta-amyloid along the vessel wall triggers vascular degeneration and subsequent reduction of blood flow.⁸ Our previous research using in vivo two-photon microscopy in a mutant tau-overexpressing AD mouse model demonstrated that progressive tau pathology is correlated with the development of tortuous capillaries in the corresponding areas and accompanies diminished blood flow.⁹ Postmortem brain tissues from individuals with AD have revealed thinning cerebral microvasculature due to collapsed degenerating endothelium and disintegrating basement membrane.¹⁰ The postmortem findings suggest that the reductions in capillary density may accelerate in AD more than what has been observed in normal healthy aging.^11
–16

At present, the ability to investigate mesoscopic changes in vascular structure in individuals with AD has been limited due to the lack of robust imaging biomarkers. The measurements of vessel size and density and their contribution to perfusion abnormalities such as cerebral blood flow (CBF) and cerebral blood volume (CBV) are instrumental in advancing an understanding of cerebrovascular changes in AD. Such properties can be measured from multi-modal perfusion magnetic resonance imaging (MRI) techniques, such as dynamic susceptibility contrast (DSC) MRI. DSC MRI measures the transverse relaxation rates during the first pass of an injected intravascular contrast agent, enabling the visualization of cerebrovascular properties such as CBF, CBV, mean transit time (MTT), and capillary transit time heterogeneity (CTH).^17
–19 Arterial spin labeling MRI is another imaging technique that assesses cerebral perfusion by using radiofrequency tagged blood signal as an endogenous contrast agent.²⁰ Lower CBF has been reported in individuals with AD and mild cognitive impairment (MCI) compared to healthy controls by using DSC and pseudo-continuous arterial spin labeling (PCASL) MRI.^21,22 Poor cognitive performance and regional cortical thinning were associated with lower CBF and CBV and with higher MTT and CTH from DSC MRI in AD patients compared to healthy controls, indicating capillary dysfunction possibly due to abnormal vascular structure.^18,19 These findings motivated the current study to investigate alterations in vessel size and vessel density and their contributions to perfusion deficits, as well as their association with AD pathology as measured by high-sensitivity positron emission tomography (PET).

Vessel size imaging is a technique that utilizes the distinctive sensitivity of the transverse relaxation rates of gradient-echo (GE) and spin-echo (SE) DSC MRI toward the underlying tissue and vessel distribution to estimate mean vessel size within a voxel.^23,24 To date, vessel size imaging has mainly been used to explore alterations in vascular morphology in tumors and stroke, wherein significant vascular network alterations emerge such as vascular remodeling, angiogenesis, and arterial occlusion.^24
–30 A few studies have shown vascular alterations in AD transgenic mouse models with the injection of iron-oxide nanoparticle contrast agents.^31,32 In a study of 10 patients with subcortical vascular dementia, steady-state vessel size imaging, which combines quantitative R₂ and R₂^* maps measured before and after the injection of a gadolinium-based contrast agent, showed larger vessel size in the white matter.³³ Another recent study using steady-state vessel size imaging showed lower vessel size in the prefrontal regions of 11 AD patients compared to cognitively normal controls.³⁴ However, no difference in vessel size was observed in key brain regions affected early in AD such as the hippocampus, entorhinal cortex, and amygdala.³⁴ Furthermore, studies using vessel size imaging techniques have been limited in their ability to reveal vascular alterations in AD patients, and the findings may vary considerably depending on the disease stage. Iron-oxide nanoparticle contrast agents are not yet approved for clinical use, and steady-state technique for vessel size imaging with gadolinium-based contrast injection has limitations due to the short washout time and leakage of the small-sized gadolinium-based contrast agent through the brain-blood barrier, which can bias the R₂ and R₂^* measurement.^35,36 Instead, the vessel size imaging in humans employed the dual-echo DSC MRI to measure $Δ$ R₂ and $Δ$ R₂^* during the first pass of the gadolinium-based contrast agent^25
–27, where the vascular concentration is the highest and hence, may deliver stronger contrast than the steady-state technique.

Here, we applied contrast-enhanced dual-echo GESE DSC MRI and non-contrast PCASL MRI techniques to study vascular architecture in AD. We assessed vessel size and density in conjunction with CBF and CBV in the hippocampus and other key regions affected early in AD in individuals with AD and MCI compared to healthy adults. We hypothesized that previously reported lower CBF and CBV in the hippocampus would be accompanied by larger vessel size and lower mean vascular density in AD and MCI. Additionally, we conducted an exploratory analysis to examine the regional relationship of vessel size, CBF and tau burden.

Material and methods

This research was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Massachusetts General Brigham. Written informed consent was obtained from each participant prior to the experiments.

Participants and inclusion criteria

Participants with AD or MCI were recruited from the Massachusetts General Hospital Alzheimer’s Disease Research Center. They received a diagnosis of AD or MCI based on a clinical assessment conducted by board-certified neurologists who followed the National Institute on Aging-Alzheimer’s Association criteria for cognitive impairment.³⁷ Inclusion criteria consisted of the following: a global Clinical Dementia Rating (CDR) score of 0.5 for questionable dementia (MCI) and 1 for mild dementia (AD); a CDR-Sum of Boxes (CDR-SoB) score ranging from 0.5 to 4.0 for questionable cognitive impairment and from 4.5 to 9.0 for mild dementia; as well as completion of standard neuropsychological test batteries and interviews.^38
–40 Patients with any other neurologic or psychiatric illness aside from AD or MCI were excluded. Older and younger healthy adult volunteers were also recruited. None of the healthy participants had a history of head trauma, dementia, stroke, or any other neurological or psychiatric disorders, and they also had to be free of substantial systemic illnesses that could potentially confound the study, such as cardiovascular diseases, respiratory diseases, endocrine disorders, and cancer. Montreal Cognitive Assessment (MoCA)⁴¹ was performed for the neuropsychological evaluations of older healthy adults and AD/MCI groups.

MRI and PET acquisition

MRI was acquired on a 3 T MRI scanner (Magnetom Prisma, Siemens, Germany) with a 32-channel head coil. T₁-weighted anatomical images were acquired using multi-echo magnetization-prepared rapid acquisition with gradient echo (MEMPRAGE) sequence in the sagittal plane with the following parameters: repetition time (TR) = 2530 ms, echo time (TE) = 1.69 ms, 3.55 ms, 5.41 ms, and 7.27 ms, inversion time (TI) = 1100 ms, field of view (FOV) = 256 mm × 256 mm, 1-mm isotropic resolution, and 192 slices. T₂-weighted fluid-attenuated inversion recovery (FLAIR) images were acquired with the following parameters: TR/TE = 5000/387 ms, FOV = 256 mm × 256 mm, 1-mm isotropic resolution, and 192 slices.

GESE DSC MRI data were acquired with an injection of gadolinium contrast. A bolus of 0.1 mmol/kg of Dotarem (gadoterate meglumine, Guerbet, France) was administered at an injection rate of 5.0 ml/s followed by a 20-ml saline flush. A novel dual-echo sequence was used for simultaneous GE and SE collections with the following parameters: TR = 1500 ms, TE = 30 ms and 90 ms for GE and SE, respectively, Generalized Auto-calibrating Partially Parallel Acquisition factor = 3, Simultaneous Multi Slice factor = 3, FOV = 200 mm × 200 mm, 2.2 mm × 2.2 mm in-plane resolution, 3.5 mm slice thickness, 33 slices, 244 times series, and 1.5 s temporal resolution for dynamic acquisition.⁴² PCASL MR images were acquired using an echo planar imaging sequence with the following parameters: TR/TE = 3580/19 ms, Multi-band acceleration factor = 6, FOV = 215 mm × 215 mm, 2.5 mm × 2.5 mm in-plane resolution, 2.5 mm slice thickness, 60 slices, labeling duration of 1500 ms, and post-labeling delay of 200 ms, 700 ms, 1200 ms, 1700 ms, and 2200 ms (with 12, 12, 12, 20, and 30 measurements, respectively). Diffusion-weighted imaging (DWI) was performed using a diffusion-weighted echo planar imaging sequence with the following parameters: TR/TE = 3133/70.8 ms, FOV = 220 mm ×216 mm, 2 mm isotropic resolution, 68 slices, b-value = 0, 1000, and 3000 s/mm² with 5, 32, and 64 gradient directions, respectively. To correct susceptibility-induced distortions in the PCASL and DWI images, two images with reversed phase-encoding polarity and echo spacing/echo times matched to the PCASL and DWI acquisitions were also acquired without labeling or diffusion weighting.

All AD/MCI participants and a subgroup of older healthy adults underwent ¹⁸F-flortaucipir and ¹¹C-Pittsburgh Compound B (¹¹C-PiB) PET imaging within a maximum of four years prior to the MRI acquisitions. The PET data for AD/MCI group were acquired using an ECAT HR+ scanner (Siemens/CTI, Knoxville, TN, USA), and the PET data for older healthy adults were acquired using 3 T mMR MR/PET (Biograph mMR scanner, Siemens, Germany) in 3D mode at Massachusetts General Hospital. The PET data for the AD/MCI group presented here were previously published.⁴³ For older healthy adults, ¹⁸F-flortaucipir PET was acquired from 75 to 115 minutes after a 10.0 $\pm$ 1.2 mCi bolus injection. Ordered subset expectation maximization algorithm with 4 iterations and 21 subsets was used to reconstruct the ¹⁸F-flortaucipir PET data in 8 $\times$ 5 mins frames.⁴⁴ ¹¹C-PiB PET was acquired with a bolus injection of 14.1 $\pm$ 1.8 mCi followed by a 60-minute dynamic acquisition. Filtered back-projection was used to reconstruct the dynamic ¹¹C-PiB PET data into 39 frames (8 $\times$ 15 s, 4 $\times$ 60 s, and 27 $\times$ 120 s).⁴⁵ Normalization and corrections for attenuation, scatter, random coincidences, and deadtime were applied.⁴⁶

Image processing

Image segmentation was performed on the root-mean-square of the MEMPRAGE images across echo times using the ‘recon-all’ script of FreeSurfer (7.1.1; https://surfer.nmr.mgh.harvard.edu) into cortical and subcortical gray matter, white matter, and cerebrospinal fluid. The subfields of the hippocampus and entorhinal cortex were separately segmented using the ‘segmentHA_T1’ script of FreeSurfer (Supplementary Figure S1). FreeSurfer ‘mri_compute_volume_fractions’ script was utilized to measure gray matter and white matter volume fractions. The volume fractions were applied to the segmented brain masks to obtain partial volume-weighted segmented brain regions and minimize partial volume effects.

Following motion correction, GESE DSC MRI data were analyzed using PGUI Perfusion Analysis Software (Center of Functionally Integrative Neuroscience, Aarhus University Hospital Norrebrogade, Denmark; https://cfin.au.dk/software/pgui/). The arterial input function (AIF) was determined by manually selecting three or four intravascular voxels from the middle cerebral artery branches based on the voxel-wise concentration curves converted from dynamic $Δ$ R₂^* (GE) and $Δ$ R₂ (SE) data of DSC MRI. CBV and CBF maps were derived by deconvolving the AIF from the dynamic curve after leakage correction.⁴⁷ For inter-subject comparison, relative CBV (rCBV) and relative CBF (rCBF) maps were derived for each subject by normalizing to the subject’s mean CBV and mean CBF, respectively, of cerebellar gray matter, a brain region that is relatively less affected in AD.⁴⁸

The PCASL MR images were analyzed by running the ‘oxford_asl’ function of the Bayesian Inference for Arterial Spin Labeling MRI (BASIL) toolbox in FSL (FMRIB Software Library, University of Oxford, UK) after motion correction using AFNI's ‘3dvolreg’ function and susceptibility distortion correction using FSL’s ‘topup’ function. The pair-wise subtraction of control and label images was performed and parametric maps of rCBF and arterial CBV were derived by using the Bayesian inference method for the inversion of a single well-mixed tissue compartment model with no dispersion of the bolus of the labeled blood water.⁴⁹ In the analysis, the T₁ values of the blood and tissue were assumed as 1.65 s and 1.3 s, respectively, along with the labeling efficiency = 0.85 and the blood-brain partition coefficient of water = 0.9 ml/g.⁵⁰ No further normalization was applied to the rCBF and arterial CBV measured from PCASL MRI because the measurements obtained from the ‘oxford_asl’ function are already relative measures that can be used for comparing between different subjects.

From the DWI MRI, mean diffusivity (MD) was calculated using the ‘DTIFit’ function in FSL based on a diffusion tensor model after the correction of susceptibility distortion and eddy current.⁵¹

For the vessel size imaging, the concentration curves from GESE DSC MRI were fitted voxel-by-voxel using the gamma-variate function (Figures 1(a) and (b)) after the leakage correction using in-house codes in MATLAB (MathWorks, Natick, MA).⁵² The $Δ$ R₂^3/2 vs. $Δ$ R₂^* hysteresis plot was generated from the vessel architectural imaging (VAI) technique, where each data point of the hysteresis plot was from the fitted data of the gamma-variate functions (Figure 1(c)).⁵³ The slope of the hysteresis plot was derived from linear regression fitting the data to $y = a x$ equation⁵⁴, where $y$ is $Δ$ R₂^*, $x$ is $Δ$ R₂^3/2, and $a$ is the slope. The relative vessel size index (rVSI) was calculated by combining MD, GE rCBV (from the area under the $Δ$ R₂^* concentration curve), and the slope of the hysteresis plot according to equation (1) (Figure 1(d)).^24,53,55 The mean vessel density was calculated by using the slope of the linear regression between $(Δ$ R₂^*)^2/3 and $Δ$ R₂ in the hysteresis plot following equation (2).⁵⁶ Other outcomes derived from the VAI hysteresis plot analysis included the vortex area, vortex direction, and peak shift of the $Δ$ R₂^* concentration curve versus the $Δ$ R₂ concentration curve (Supplementary Figure S2).⁵³ All perfusion parametric maps were co-registered to T₁-weighted anatomical images using the ‘bbregister’ script of FreeSurfer (Supplementary Figure S3).

rVSI \propto \sqrt{G E rCBV \times MD} \times slope of VAI hyteresis plot

(1)

mean vessel density ({mm}^{- 2}) \approx \frac{1}{{1.678}^{3}} \times \frac{1}{MD} \times {(\frac{Δ R_{2}}{{(Δ R_{2}^{*})}^{\frac{2}{3}}})}^{3}

(2)

Figure 1.

Derivation of rVSI map. (a): Change of relaxation rate ( $Δ$ R₂^*) sensitized by GE DSC MRI with gamma-variate fitting (red line). (b): Change of relaxation rate ( $Δ$ R₂) sensitized by SE DSC MRI with gamma-variate fitting (blue line). (c): Hysteresis plot using the VAI technique with fitted values from (a) and (b). (d): The formula to calculate rVSI map. The unit of MD is $μ$ m²/ms. The unit of other maps is an arbitrary unit (a.u.).

PET data of older healthy adults were motion corrected by the ‘FLIRT’ function in FSL for the frame-to-frame realignment. For ¹⁸F-flortaucipir PET and ¹¹C-PiB PET, the first frame (75-80 minutes) and the average of frames between 12-16 min were used as the reference images, respectively. The reference images for ¹⁸F-flortaucipir PET and ¹¹C-PiB PET were co-registered to T₁-weighted images using the ‘flirt’ function of FSL, and the resulting transformation matrices were applied to all PET frames. The FreeSurfer segmentations in the T₁-weigthted space were then applied to the PET images for regional sampling. For ¹⁸F-flortaucipir PET, the data acquired from 80 to 100 minutes following the tracer injection were averaged to ensure consistency with the processing procedure used for the AD/MCI group, and a standardized uptake value ratio (SUVR) map was generated by normalizing the 3D PET volume to the mean activity of the cerebellar gray matter.⁴⁴ Average regional values were then extracted in all parametric perfusion maps and tau SUVR maps in partial volume-weighted FreeSurfer regions known to be affected early in AD progression: the hippocampus, entorhinal cortex, amygdala, and inferior temporal cortex. Additionally, the volume of each of these regions was quantified. For ¹¹C-PiB PET, the regional time-activity curves were generated for the frontal, lateral temporal, and retrosplenial cortices (FLR) composite and cerebellar gray matter region. To assess beta-amyloid burden, distribution volume ratios (DVR) were generated in the FLR region with reference Logan graphical method (t* = 40 min) using cerebellar gray matter as a reference.^44,57
–59 Beta-amyloid positivity was defined as DVR > 1.3 in the FLR region.^43,60

Statistical analysis

Demographic and clinical variables of the three groups of participants undergoing MRI (AD/MCI, older and younger healthy adults) were compared using the chi-squared ( $χ^{2}$ ) test for the categorical variable (sex) and the analysis of variance (ANOVA) and the Mann-Whitney U test for the numerical variable (age and MoCA scores). The averaged parametric values in the segmented ROIs were compared among the three groups using the Mann-Whitney U test, adjusting for multiple comparisons using the false discovery rate (FDR). Tau SUVR comparison between the AD/MCI group and a subgroup of older healthy adults who had tau PET data was assessed by the Mann-Whitney U test without adjusting the P value. The associations between tau SUVR and both rVSI and rCBF from PCASL MRI were measured for each group and all subjects using Pearson’s correlation coefficients. P values of less than 0.05 were considered statistically significant.

Results

Participant characteristics

The demographic and clinical characteristics of the study participants are summarized in Table 1. A total of 23 participants were recruited, composed of 7 participants in the AD/MCI group, 9 participants in the older healthy adult group, and 7 participants in the younger healthy adult group. The AD/MCI group consisted of six individuals with MCI and one individual with AD, ranging in age from 66- to 86-year-old (76.8 $\pm$ 6.7 years, 3 females). The older healthy adult group was comprised of 55- to 71-year-old adults (65.1 $\pm$ 5.4 years, 5 females). The younger healthy adult group was comprised of 22- to 34-year-old adults (29.6 $\pm$ 3.9 years, 3 females). There was no significant difference in sex among the three groups (all FDR-P values > 0.05). Age was significantly different among the three groups (FDR-P value = 0.002 between the AD/MCI group and older healthy adults; FDR-P value <0.001 between the AD/MCI group and younger healthy adults; FDR-P value <0.001 between the older healthy adults and younger healthy adults). The MoCA scores were also significantly different between the AD/MCI group and older healthy adults (unadjusted P value < 0.001). In the AD/MCI group, the global CDR score and CDR-SoB score indicated they are patients with relatively mild dementia.⁴⁰

Table 1.

Clinical and demographic characteristics of AD/MCI, older healthy and younger healthy adults.

	AD/MCI (N = 7)	Older healthy adult (N = 9)	Younger healthy adult (N = 7)	P value	FDR-P value
	AD/MCI (N = 7)	Older healthy adult (N = 9)	Younger healthy adult (N = 7)	P value	AD/MCI vs. Older	AD/MCI vs. Younger	Older vs. Younger
Gender (M/F)	4/3	4/5	4/3	0.95^a	0.92	1	0.92
Age (years)	76.8 $\pm$ 6.7	65.1 $\pm$ 5.4	29.6 $\pm$ 3.9	<0.001**^b	0.002**	<0.001**	<0.001**
MoCA	20.6 $\pm$ 4.5	28.8 $\pm$ 1.3	.	<0.001**^c
Global CDR	0.6 $\pm$ 0.2	.	.	.
CDR-SoB	2.6 $\pm$ 1.6	.	.	.
Amyloid positivity	71.4 % (5/7)	0 % (0/6)^d	.	.

Data are shown as mean $\pm$ standard deviation.

= $χ^{2}$ test,

= ANOVA,

= Mann-Whitney U test.

= Six older healthy adults underwent [¹¹C] PiB PET scan.

FDR-P = P-values from a post-hoc analysis with $χ^{2}$ test^a and Mann-Whitney U test^b adjusting for multiple comparisons using the FDR.

: P < 0.005.

AD/MCI group consisted of six individuals with MCI and one individual with AD.

AD: Alzheimer’s disease; CDR: clinical dementia rating; CDR-SoB: clinical dementia rating-sum of boxes; FDR: false discovery rate; MCI: mild cognitive impairment; MoCA: Montreal Cognitive Assessment.

PET data analysis was conducted in all AD/MCI participants and the subgroup of older healthy adults (six participants) who underwent PET scans. The years between MRI and ¹⁸F-flortaucipir PET were 2.7 $\pm$ 1.3 years (range: 0.5–4.2 years) for the AD/MCI group and 0.7 $\pm$ 0.2 years (range: 0.5–1.1 years) for older healthy adults. For ¹¹C-PiB PET, the years from MRI were 2.7 $\pm$ 1.2 years (range: 0.5–4.2 years) for the AD/MCI group and 0.6 $\pm$ 0.2 years (range: 0.5–1.0 years) for older healthy adults. Five out of seven individuals in the AD/MCI group showed beta-amyloid positivity in the FLR region from [¹¹C] PiB PET with an average FLR DVR value of 2.06 $\pm$ 0.69. All older healthy adults who underwent [¹¹C] PiB PET scan showed negative beta-amyloid results with an average FLR DVR value of 1.02 $\pm$ 0.03.

The volume was lower in all selected brain regions of the AD/MCI group than the healthy adults, but no differences were discernible between the two groups of healthy adults (Supplementary Table S1).

Perfusion and diffusion MRI results

Representative T₂-FLAIR, perfusion and diffusion MR images are shown for an older healthy adult and participant with MCI in Supplementary Figure S4. Figure 2 shows box plots comparing CBV and CBF between groups from PCASL, SE DSC, and GE DSC MRI. The arterial CBV from PCASL was significantly lower in the hippocampus of the AD/MCI group compared to the healthy adults (Figure 2(a), Supplementary Table S2). The rCBF from PCASL was significantly lower in all selected brain regions of the AD/MCI group compared to the healthy adults (Figure 2(d), Supplementary Table S2). The rCBV from SE DSC MRI was lower in the inferior temporal cortex of AD/MCI group and older healthy adults compared to younger healthy adults (Figure 2(b), Supplementary Table S2). The rCBF from SE DSC MRI also showed lower blood flow in the hippocampus of the AD/MCI group compared to the older healthy adults (Figure 2(e), Supplementary Table S2). CBV and CBF from GE DSC MRI did not show significant differences amongst the three groups for any brain region (Figures 2(c) and (f), Supplementary Table S2).

Figure 2.

CBV and CBF distribution from PCASL MRI, GE DSC MRI, and SE DSC MRI. (a): arterial CBV from PCASL MRI. (b): rCBV from SE DSC MRI. (c): rCBV from GE DSC MRI. (d): rCBF from PCASL MRI. (e): rCBF from SE DSC MRI. (f): rCBF from GE DSC MRI. * = P value < 0.05 and ** = P value < 0.005 in Mann-Whitney U-test adjusting for multiple comparisons using the FDR.

MD values in all selected brain regions of the AD/MCI group were significantly higher than the MD values in the two healthy groups (Supplementary Figure S5, Supplementary Table S3). Additionally, MD values in the inferior temporal cortex were higher in the older healthy adults than in the younger healthy adults (Supplementary Figure S5, Supplementary Table S3).

Vessel size, vessel density, and VAI analysis results

Representative vessel size and vessel density maps are shown alongside the corresponding T₂-FLAIR images in Figure 3. rVSI in the hippocampus of the AD/MCI group was significantly larger than that in the two healthy groups (Figure 4(a), Supplementary Table S4). However, mean vessel density was not significantly different among groups for any brain region (Figure 4(b), Supplementary Table S4). This was also the case for the vortex area and vortex direction of the VAI hysteresis plot and peak shift of the $Δ$ R₂^* concentration curve versus the $Δ$ R₂ concentration curve (Supplementary Figure S6, Supplementary Table S5).

Figure 3.

Multiparametric maps of the brain with vessel size, vessel density, and tau SUVR. The top row (a) is from a 70-year-old female older healthy adult. The bottom row (b) is from a 67-year-old female individual with MCI. From left to right: T₂-FLAIR, rVSI (a.u.), mean vessel density (mm⁻²), and tau SUVR (a.u.).

Figure 4.

Vessel size and vessel density distributions. (a): rVSI distribution. (b): mean vessel density distribution. * = P value < 0.05 and ** = P value < 0.005 in Mann-Whitney U-test adjusting for multiple comparisons using the FDR.

The association of tau deposition with rVSI and CBF

Representative tau SUVR maps are presented in Figure 3 for an older healthy adult and AD participant. Tau burden was elevated in all evaluated regions in the AD/MCI group relative to older healthy adults, as shown in Supplementary Figure S7 and Supplementary Table S6. While tau SUVR was not associated with rVSI in the entorhinal cortex, amygdala, and inferior temporal cortex (Figures 5(a) and (b) and (d), Supplementary Table S7) among all subjects, it exhibited a positive correlation with rVSI in the hippocampus (Figure 5(c), Supplementary Table S7). However, tau SUVR was negatively correlated with rVSI in the inferior temporal cortex of older healthy adults (Supplementary Table S7). Tau SUVR was negatively associated with rCBF from PCASL in all selected brain regions (Figures 5(e) to (h), Supplementary Table S7). Tau SUVR was significantly associated with PCASL rCBF in the entorhinal cortex of older healthy adults (Supplementary Table S7). Additionally, rCBF from PCASL was associated with rVSI in the hippocampus among all subjects (Supplementary Figure S8C, Supplementary Table S8).

Figure 5.

Correlation of tau burden with rVSI and PCASL rCBF. (a-d): The correlation between tau SUVR and rVSI in the amygdala, entorhinal cortex, hippocampus, and inferior temporal cortex. (e-h): The correlation between tau SUVR and rCBF from PCASL MRI in the amygdala, entorhinal cortex, hippocampus, and inferior temporal cortex. The association was tested using Pearson’s correlation coefficient with unadjusted P values.

Figure 6 illustrates the hypothesized alteration vessel architecture, particularly in the hippocampus, associated with AD and MCI based on the presented results.

Figure 6.

(a) The schematic diagram of vascular alteration with disease progression and (b) the vessel size distribution in the normal stage and diseased stage.

Discussion

In this study, we used a combined analysis of contrast (GESE DSC) and non-contrast (PCASL) perfusion MRI parameters in selected brain regions preferentially affected in AD to investigate cerebrovascular structural and functional alterations in individuals with AD and MCI. Consistent with our hypothesis, we observed larger vascular size corresponding to areas of lower CBF in individuals with AD and MCI compared to cognitively intact healthy adults, suggesting the presence of vascular architecture alteration in these areas, particularly in the hippocampus. The observation that CBF was negatively correlated with tau burden is in agreement with previous work showing that vascular dysfunction reflected by lower CBF is associated with elevated tau and that these relationships are stronger in individuals with the elevated amyloid burden.⁶¹ Furthermore, our study identified a positive correlation between vessel size and tau burden in the hippocampus. Our results support the hypothesis that vascular architecture alteration occurs with a lower blood flow in the hippocampus of patients with even relatively mild dementia.^62,63

The hippocampus is central to memory and cognition in humans⁶⁴ and is known to be involved early in AD progression.^65,66 We observed a significant increase in the average size of blood vessels in the hippocampus of individuals with AD and MCI compared to healthy adults. Our in vivo imaging results are consistent with previous postmortem studies showing microvascular architectural alteration in the AD brain.^10,11,67 as well as the previous preclinical studies. For example, others have demonstrated that vessel size is larger in the hippocampus in the rTg4510 transgenic mouse model of AD expressing human tau protein containing the P301L mutation.⁶⁸ The larger vessel size with lower vessel density in the hippocampal area was shown in the APP23 AD mouse model.³² In another study, 5xFAD AD mice exhibited larger vessel size in the somatosensory cortex, but not in the hippocampus.³¹ From histological analysis, APP/PS1 AD mice temporarily showed a larger vessel size in the frontoparietal cortex and thalamus at an early age but subsequently lower with disease progression.⁶⁹

Vessel size imaging allows the estimation of mean vessel size within a voxel. Consequently, the larger vessel size can represent two different scenarios: 1) the dilation and enlargement of vessels and 2) the loss of smaller vessels, or even a combination of both occurrences. In our study, lower CBF measured from SE DSC MRI was observed in the hippocampus of the AD/MCI group. This was not observed with GE DSC MRI. These differences may originate from the distinct characteristics of SE and GE signals: SE signals are predominantly sensitive to microscopic vessels such as capillaries, whereas susceptibility-sensitive GE signals are dominated by both microscopic and macroscopic vessels.^23,70 The selective loss of microscopic vessels may produce the observed shift of the mean vessel size in the AD/MCI patients studied here, particularly in the hippocampus (Figure 6).

We found differences between the CBV and CBF from PCASL MRI and DSC MRI in group comparisons and in their correlations with tau burden. The CBF from PCASL MRI differentiates between the AD/MCI and healthy groups better than the CBF from GE DSC MRI or SE DSC MRI in all four brain regions. PCASL MRI CBF also correlated significantly with tau SUVR but such correlations were not significant in the DSC MRI CBF (not shown). Here, we consider the differences between the two techniques and assumptions made in the data analysis that may lead to the discrepancies. DSC MRI utilizes a bolus injection of the gadolinium-based contrast agent to achieve a high signal-to-noise ratio, which depends on the delivery and concentration of the contrast agent at the tissue level. Assessment of perfusion relies on the deconvolution of AIF from the derived concentration curve assuming a linear relationship between the relaxation rates and contrast agent concentration. Such an assumption may not hold in regions of relatively large arterioles or high arterial density. Furthermore, the best AIF was not always found in the same brain regions among subjects and the chosen AIF may not be in close proximity or feeding directly into the brain regions of interest. All these may contribute to large variations among subjects and thus, compromise the sensitivity of resulting CBF and CBV to the perfusion deficits. However, the simultaneous GESE DSC MRI⁴² used in this work can resolve both macroscopic and microscopic vessels^23,70 in the brain, allowing us to derive important quantitative features that reflect vascular size and density with VAI modeling⁵³ beyond standard cerebral perfusion measures. A common practice in DSC MRI is to normalize the resulting parametric maps to a brain region that is least affected by the disease. In AD research, the cerebellum is a common reference region. It is important to note that in the progression of neurodegenerative diseases, global changes in DSC MRI parameters can occur throughout the brain, and the choice of reference region may contribute to variance in the group comparisons with DSC MRI. On the other hand, PCASL MRI employs electromagnetically labeled water molecules in arterial blood as an endogenous contrast agent. The labeled blood flows into the capillaries and exchanges with the tissue, contributing to the overall perfusion signal. The labeled blood water eventually drains into the venous system while the labeling disappears through spin relaxation mechanisms and thus, PCASL MRI primarily captures arterial/capillary blood flow. It also offers the advantage of not requiring signal normalization for intersubject comparison.

We also hypothesized that the selected brain regions would show lower mean vessel density in the AD/MCI group based on the human brain autopsy findings.^10,15,16 However, while the mean vessel density of the AD/MCI group was lower than the other two groups in the hippocampus, this finding was not statistically significant. Previous studies on 2-month-old 5xFAD and 8-month-old APP/PS1 amyloid depositing AD mice revealed an increase in vessel size, but there was no significant change in vessel density.^31,69 In the immunohistochemical analyses of the APP/PS1 AD model, 18-month-old mice showed no difference in capillary density in the hippocampus in one study,⁷¹ but another study reported lower vessel density in the hippocampus and cortex compared to wild-type littermates at 12 months, but not at 6 months.⁷² The lower vessel density in the hippocampus of APP23 mice was observed in 9- and 20-month-old mice on vessel size imaging using MRI, but not in 3-, 6-, and 14-month-old mice.³² The Tg2576 mouse exhibited a lower vessel density in the hippocampus at 17 months, but not at 3 and 9 months on capillary immunostaining.⁷³ Most preclinical studies were performed in transgenic AD mouse models overexpressing transgenes that promote severe amyloid plaque formation in the brain but exhibit varied degrees of CAA, and the lower vascular density was observed at the later stage of at least 9 months. However, our previous work demonstrated that in vivo two-photon microscopy in Tg4510 mice developing tau tangles revealed tortuous capillaries with lower blood flow, lower blood vessel diameters, and higher blood vessel density in the cortex after 15-month-old, which appears divergent from our current results.⁹ It is important to consider that the disease duration, severity, brain region, and underlying pathology in AD may influence and modulate cerebrovascular structural alterations.

We observed a significant association between larger vessel size and elevated tau burden in the hippocampus. Additionally, lower CBF from PCASL was associated with higher tau burden in all selected brain regions. These results are consistent with a recent study showing spatial colocalization between tau PET levels and signs of vascular impairment in the temporal-parietal and frontal areas of two independent cohorts.⁶¹ It remains unclear whether structural and functional changes precede or follow the development of tau pathology.^9,74
–76 Our findings suggest the significance of investigating the relationship between vascular structure, functional abnormalities and tau pathology in the context of AD pathogenesis. Systematic investigations of how tau and even beta-amyloid deposition contribute to vessel size alterations and their interactions in AD should be conducted in future work.

The observed reduction in CBF can introduce considerable variance in mean vessel size and density measures in the AD/MCI group. Conversely, the vessel architectural alterations may lead to changes in CBF. Functional (CBF) and structural (vessel size and density) alterations in the vasculature likely interplay with each other in a complex manner that is not accounted for in the current model, underscoring the importance of further modeling and experimental studies to account for the relationship between vascular structure and reduced CBF.

To increase the robustness of measuring the vessel size, the slope from linear regression using the $y = a x$ equation rather than the slope of the long axis of the vortex in $Δ$ R₂^3/2 and $Δ$ R₂^* hysteresis plot was utilized for the rVSI equation. All data points of the first passage of the contrast agent in dual-echo GESE DSC MRI could then be included in the measurement (Supplementary Figure S9).^53,54 In cases (2) and (3) shown in Supplementary Figure S9, the slope of the long axis of the vortex is infinite, whereas the slope from the linear regression is zero in case (2) and has a specific value in case (3). The identical approach was also applied to calculate the mean vessel density.

A recent study reported that the vessel size was lower in various brain regions of AD patients in voxel-based analysis from the steady-state vessel size imaging with gadolinium contrast injection.³⁴ However, there was no difference in ROI-based analysis. The hippocampus showed no significant difference in both analyses. Furthermore, they were unable to detect the difference in vessel density between the two groups. The steady-state vessel size imaging technique, by combining quantitative R₂ and R₂^* maps obtained before and after an injection of superparamagnetic iron-oxide nanoparticles (SPION), has been considered a standard method in preclinical research because SPION has a long washout time and changes R₂ and R₂^* values within the blood vessel under the steady-state condition for a couple of hours.^36,77 It should be cautiously considered that the fast washout and leakage of small-sized gadolinium contrast agent³⁴ in steady-state vessel size imaging. Another previous study compared the measured vessel sizes from dual-echo DSC MRI using gadolinium-based contrast agent and steady-state vessel size imaging using USPIO (ultrasmall SPION) in rats and demonstrated that there was strong correlation and no significant difference between the two methods in tumor regions, and applying a preload of gadolinium improved VSI estimates in the bolus approach.³⁶

Limitations

In this study, a limited number of participants in each group were studied without longitudinal examinations. The disease severity of the AD/MCI group was also limited. Our aim was to evaluate individuals with relatively mild dementia in order to evaluate early vascular changes occurring in MCI and AD. Even in these early stages, we were able to detect significant differences in vessel size between the AD/MCI group and healthy volunteers. To understand how the changes in vasculature evolve with AD progression better, it will also be important to investigate the variations in vascular size, CBV, CBF, and vascular density, along with their correlations with cognitive test (such as MoCA), in patients with more severe cognitive impairment or late-stage AD, over a longer course of disease progression. We also considered the potential bias introduced by including only one subject with AD in the study. We found that the statistical results remained consistent across all comparisons or associations after excluding the individual with AD. Regarding the analysis of perfusion parameters and tau PET, CBV and CBF from GESE DSC MRI and PET uptake values were semi-quantitatively normalized by the mean uptake in the cerebellar gray matter for inter-subject comparison. The results may depend on the reference region chosen, and we chose the cerebellum based on prior work.⁴⁴ The optimal reference region for PET uptake values remains an open question, with the cerebellum being the region most commonly used as a reference for perfusion MRI and PET in AD studies.^48,78,79 While alternative reference regions could be chosen, the reference region for the normalization would be expected to vary less with disease progression than affected areas like the hippocampus. The long intervals between MRI and PET scans in the AD/MCI group could also lead to a distortion in the correlation between tau burden and MRI results, as the accumulation of additional tau deposits can occur during this time. Finally, age was different between the older healthy adult group and the AD/MCI group and could be contributing to the observed changes in addition to AD-related pathology. We examined the differences in vessel size, perfusion parameters between younger and older healthy adults in this study. Although aging is known to be linked to vessel loss and increased capillary tortuosity,^{11,12,15,16,67} the younger and older healthy adult groups only show significant differences in CBV from SE DSC MRI and MD in the inferior temporal cortex. Our findings merit further investigation and corroboration across a wider range of ages and larger cohorts of patients and healthy subjects, particularly with the inclusion of longitudinal examinations.

Conclusions

In conclusion, this study presents in vivo observations of larger vessel size with corresponding lower CBF in the hippocampus of individuals with AD and MCI. Furthermore, tau deposition is significantly associated with alterations in vascular structure (larger vessel size) and dysfunction (lower CBF). Our findings suggest that vessel size may be a promising imaging marker of aberrant vascular architecture in AD and MCI, allowing discrimination between functional and structural changes taking place in the brain.

Supplemental Material

sj-pdf-1-jcb-10.1177_0271678X231216144 - Supplemental material for Aberrant vascular architecture in the hippocampus correlates with tau burden in mild cognitive impairment and Alzheimer’s disease

Supplemental material, sj-pdf-1-jcb-10.1177_0271678X231216144 for Aberrant vascular architecture in the hippocampus correlates with tau burden in mild cognitive impairment and Alzheimer’s disease by Hansol Lee, Jessie Fanglu Fu, Kyla Gaudet, Annie G Bryant, Julie C Price, Rachel E Bennett, Keith A Johnson, Bradley T Hyman, Trey Hedden, David H Salat, Yi-Fen Yen and Susie Y Huang in Journal of Cerebral Blood Flow & Metabolism

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by NIH R21AG067562, P41EB030006, P30AG062421, R00AG061259, R01NR010827, R01AG053509, and P30AG066514. This work was also supported by the Korea Health Technology R&D Project through the Korea Healthy Industry Development Institute (KHIDI), funded by the Ministry of Health and Welfare, Republic of Korea [grant number: HI19C1095].

Acknowledgements

We thank Dr. Leif Østergaard, Dr. Kim Mouridsen, and Dr. Mikkel Bo Hansen of the Center of Functionally Integrative Neuroscience, Aarhus University Hospital Norrebrogade, Denmark, for their generous support of the PGUI Perfusion Analysis Software. We also thank Dr. Kyrre E. Emblem of the Department of Physics and Computational Radiology, Oslo University Hospital, Norway, for his guidance on the VAI technique.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

Hansol Lee: Involved in study conception and design; data acquisition; analysis and interpretation of data; manuscript drafting and revision. Jessie Fanglu Fu: analysis and interpretation of data; manuscript revision. Kyla Gaudet: Involved in data acquisition; manuscript revision. Annie G. Bryant: Involved in analysis and interpretation of data; manuscript revision. Julie C. Price: Involved in analysis and interpretation of data; manuscript revision. Rachel E. Bennett: Involved in analysis and interpretation of data; manuscript revision. Keith A. Johnson: Involved in subject recruitment; analysis and interpretation of data; manuscript revision. Bradley T. Hyman: Involved in analysis and interpretation of data; manuscript revision. Trey Hedden: Involved in subject recruitment; analysis and interpretation of data; manuscript revision. David H. Salat: Involved in subject recruitment; analysis and interpretation of data; manuscript revision. Yi-Fen Yen: Involved in study conception and design; data acquisition; analysis and interpretation of data; manuscript drafting and revision. Susie Y. Huang: Involved in study conception and design; subject recruitment; data acquisition; analysis and interpretation of data; manuscript drafting and revision.

ORCID iDs

Jessie Fanglu Fu

Annie G Bryant

Supplementary material

Supplemental material for this article is available online.

References

Cummings

Cole

Alzheimer disease. Jama 2002; 287: 2335–2338.

Hyman

Phelps

Beach

, et al. National Institute on Aging–Alzheimer's Association guidelines for the neuropathologic assessment of Alzheimer's disease. Alzheimers Dement 2012; 8: 1–13.

Montine

Phelps

Beach

, et al. National Institute on Aging–Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: a practical approach. Acta Neuropathol 2012; 123: 1–11.

Braak

Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 1991; 82: 239–259.

Iturria-Medina

Sotero

Toussaint

, et al. Early role of vascular dysregulation on late-onset Alzheimer’s disease based on multifactorial data-driven analysis. Nat Commun 2016; 7: 11934.

Sweeney

Kisler

Montagne

, et al. The role of brain vasculature in neurodegenerative disorders. Nat Neurosci 2018; 21: 1318–1331.

Govindpani

McNamara

Smith

, et al. Vascular dysfunction in Alzheimer's disease: a prelude to the pathological process or a consequence of it? J Clin Med 2019; 8: 20190510. DOI: 10.3390/jcm8050651.

Thal

Griffin

de Vos

, et al. Cerebral amyloid angiopathy and its relationship to Alzheimer's disease. Acta Neuropathol 2008; 115: 599–609.

Bennett

Robbins

, et al. Tau induces blood vessel abnormalities and angiogenesis-related gene expression in P301L transgenic mice and human Alzheimer's disease. Proc Natl Acad Sci U S A 2018; 115: E1289–E1298.

10.

Kalaria

Hedera

Differential degeneration of the cerebral microvasculature in Alzheimer's disease. Neuroreport 1995; 6: 477–480.

11.

Farkas

Luiten

PG.

Cerebral microvascular pathology in aging and Alzheimer's disease. Prog Neurobiol 2001; 64: 575–611.

12.

Sonntag

Lynch

Cooney

, et al. Decreases in cerebral microvasculature with age are associated with the decline in growth hormone and insulin-like growth factor 1. Endocrinology 1997; 138: 3515–3520.

13.

Amenta

Ferrante

Mancini

, et al. Effect of long term treatment with the dihydropyridine-type calcium channel blocker darodipine (PY 108-068) on the cerebral capillary network in aged rats. Mech Ageing Dev 1995; 78: 27–37.

14.

Amenta

Cavallotti

Del Valle

, et al. Age-related changes in brain microanatomy: sensitivity to treatment with the dihydropyridine calcium channel blocker darodipine (PY 108-068). Brain Res Bull 1995; 36: 453–460.

15.

Buee

Hof

Delacourte

Brain microvascular changes in Alzheimer's disease and other dementias. Ann N Y Acad Sci 1997; 826: 7–24.

16.

Brown

Thore

CR.

Cerebral microvascular pathology in ageing and neurodegeneration. Neuropathol Appl Neurobiol 2011; 37: 56–74.

17.

Shiroishi

Castellazzi

Boxerman

, et al. Principles of T2*‐weighted dynamic susceptibility contrast MRI technique in brain tumor imaging. J Magn Reson Imaging 2015; 41: 296–313.

18.

Nielsen

Egefjord

Angleys

, et al. Capillary dysfunction is associated with symptom severity and neurodegeneration in Alzheimer's disease. Alzheimers Dement 2017; 13: 1143–1153.

19.

Eskildsen

Gyldensted

Nagenthiraja

, et al. Increased cortical capillary transit time heterogeneity in Alzheimer's disease: a DSC-MRI perfusion study. Neurobiol Aging 2017; 50: 107–118.

20.

Petcharunpaisan

Ramalho

Castillo

Arterial spin labeling in neuroimaging. World J Radiol 2010; 2: 384–398.

21.

Chen

Song

Beyea

, et al. Advances in perfusion magnetic resonance imaging in Alzheimer's disease. Alzheimers Dement 2011; 7: 185–196.

22.

Zhang

Wang

Lyu

, et al. Cerebral blood flow in mild cognitive impairment and Alzheimer’s disease: a systematic review and meta-analysis. Ageing Res Rev 2021; 71: 101450.

23.

Yablonskiy

Haacke

EM.

Theory of NMR signal behavior in magnetically inhomogeneous tissues: the static dephasing regime. Magn Reson Med 1994; 32: 749–763.

24.

Kiselev

Strecker

Ziyeh

, et al. Vessel size imaging in humans. Magn Reson Med 2005; 53: 553–563.

25.

Chakhoyan

Yao

Leu

, et al. Validation of vessel size imaging (VSI) in high-grade human gliomas using magnetic resonance imaging, image-guided biopsies, and quantitative immunohistochemistry. Sci Rep 2019; 9: 2846–2848.

26.

Guo

Kang

Tong

, et al. Microvascular characteristics of lower-grade diffuse gliomas: investigating vessel size imaging for differentiating grades and subtypes. Eur Radiol 2019; 29: 1893–1902.

27.

Schmidt

Villringer

, et al. Vessel size imaging reveals pathological changes of microvessel density and size in acute ischemia. J Cereb Blood Flow Metab 2011; 31: 1687–1695.

28.

Stokes

Skinner

Quarles

CC.

Assessment of a combined spin-and gradient-echo (SAGE) DSC-MRI method for preclinical neuroimaging. Magn Reson Imaging 2014; 32: 1181–1190.

29.

Stokes

Quarles

CC.

A simplified spin and gradient echo approach for brain tumor perfusion imaging. Magn Reson Med 2016; 75: 356–362.

30.

Stokes

Ragunathan

Robison

, et al. Development of a spiral spin‐and gradient‐echo (spiral‐SAGE) approach for improved multi‐parametric dynamic contrast neuroimaging. Magn Reson Med 2021; 86: 3082–3095.

31.

Chang

Kim

Lee

, et al. Mapping of microvascular architecture in the brain of an Alzheimer's disease mouse model using MRI. NMR Biomed 2021; 34: e4481.

32.

Meng

Wen

, et al. Dynamic changes in vascular size and density in transgenic mice with Alzheimer’s disease. Aging (Albany NY) 2020; 12: 17224–17234.

33.

Choi

H-I

Ryu

C-W

Kim

, et al. Changes in microvascular morphology in subcortical vascular dementia: a study of vessel size magnetic resonance imaging. Front Neurol 2020; 11: 545450.

34.

Guo

X-Y

Kwon

Rhee

, et al. Microvascular morphology alteration using relaxation rate change with gadolinium-based magnetic resonance imaging contrast agent in patients with Alzheimer’s disease. Quant Imaging Med Surg 2023; 13: 1–16.

35.

Bjornerud

Sorensen

Mouridsen

, et al. T1-and T* 2-dominant extravasation correction in DSC-MRI: part I – theoretical considerations and implications for assessment of tumor hemodynamic properties. J Cereb Blood Flow Metab 2011; 31: 2041–2053.

36.

Pannetier

Lemasson

Christen

, et al. Vessel size index measurements in a rat model of glioma: comparison of the dynamic (Gd) and steady‐state (iron‐oxide) susceptibility contrast MRI approaches. NMR Biomed 2012; 25: 218–226.

37.

McKhann

Knopman

Chertkow

, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 2011; 7: 263–269.

38.

Hughes

Berg

Danziger

, et al. A new clinical scale for the staging of dementia. Br J Psychiatry 1982; 140: 566–572.

39.

Morris

Weintraub

Chui

, et al. The uniform data set (UDS): clinical and cognitive variables and descriptive data from Alzheimer disease centers. Alzheimer Dis Assoc Disord 2006; 20: 210–216.

40.

O'Bryant

Waring

Cullum

, et al. Staging dementia using clinical dementia rating scale sum of boxes scores: a Texas Alzheimer's research consortium study. Arch Neurol 2008; 65: 1091–1095.

41.

Nasreddine

Phillips

Bédirian

, et al. The Montreal cognitive assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005; 53: 695–699.

42.

Manhard

Bilgic

Liao

, et al. Accelerated whole‐brain perfusion imaging using a simultaneous multislice spin‐echo and gradient‐echo sequence with joint virtual coil reconstruction. Magn Reson Med 2019; 82: 973–983.

43.

Bryant

Manhard

Salat

, et al. Heterogeneity of tau deposition and microvascular involvement in MCI and AD. Curr Alzheimer Res 2021; 18: 711–720.

44.

Johnson

Schultz

Betensky

, et al. Tau positron emission tomographic imaging in aging and early Alzheimer disease. Ann Neurol 2016; 79: 110–119.

45.

Johnson

Gregas

Becker

, et al. Imaging of amyloid burden and distribution in cerebral amyloid angiopathy. Ann Neurol 2007; 62: 229–234.

46.

Izquierdo-Garcia

Hansen

Förster

, et al. An SPM8-based approach for attenuation correction combining segmentation and nonrigid template formation: application to simultaneous PET/MR brain imaging. J Nucl Med 2014; 55: 1825–1830.

47.

Hansen

Tietze

Kalpathy‐Cramer

, et al. Reliable estimation of microvascular flow patterns in patients with disrupted blood–brain barrier using dynamic susceptibility contrast MRI. J Magn Reson Imaging 2017; 46: 537–549.

48.

Larner

AJ.

The cerebellum in Alzheimer's disease. Dement Geriatr Cogn Disord 1997; 8: 203–209.

49.

Buxton

Frank

Wong

, et al. A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 1998; 40: 383–396.

50.

Alsop

Detre

Golay

, et al. Recommended implementation of arterial spin‐labeled perfusion MRI for clinical applications: a consensus of the ISMRM perfusion study group and the European Consortium for ASL in dementia. Magn Reson Med 2015; 73: 102–116.

51.

Andersson

Sotiropoulos

SN.

An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage 2016; 125: 1063–1078.

52.

Boxerman

Schmainda

Weisskoff

Relative cerebral blood volume maps corrected for contrast agent extravasation significantly correlate with glioma tumor grade, whereas uncorrected maps do not. Am J Neuroradiol 2006; 27: 859–867.

53.

Emblem

Mouridsen

Bjornerud

, et al. Vessel architectural imaging identifies cancer patient responders to anti-angiogenic therapy. Nat Med 2013; 19: 1178–1183.

54.

Hsu

Yang

Lim

, et al. Vessel size imaging using dual contrast agent injections. J Magn Reson Imaging 2009; 30: 1078–1084.

55.

Kim

Park

Emblem

, et al. Vessel type determined by vessel architectural imaging improves differentiation between early tumor progression and pseudoprogression in glioblastoma. Am J Neuroradiol 2021; 42: 663–670.

56.

Troprès

Pannetier

Grand

, et al. Imaging the microvessel caliber and density: principles and applications of microvascular MRI. Magn Reson Med 2015; 73: 325–341.

57.

Logan

Fowler

Volkow

, et al. Graphical analysis of reversible radioligand binding from time – activity measurements applied to [N-11C-methyl]-(−)-cocaine PET studies in human subjects. J Cereb Blood Flow Metab 1990; 10: 740–747.

58.

Lopresti

Klunk

Mathis

, et al. Simplified quantification of Pittsburgh compound B amyloid imaging PET studies: a comparative analysis. J Nucl Med 2005; 46: 1959–1972.

59.

Mintun

Larossa

Sheline

, et al. 11C] PIB in a nondemented population: potential antecedent marker of Alzheimer disease. Neurology 2006; 67: 446–452.

60.

Gomperts

Rentz

Moran

, et al. Imaging amyloid deposition in Lewy body diseases. Neurology 2008; 71: 903–910.

61.

Albrecht

Isenberg

Stradford

, et al. Associations between vascular function and tau PET are associated with global cognition and amyloid. J Neurosci 2020; 40: 8573–8586.

62.

Baloyannis

IS.

The vascular factor in Alzheimer's disease: a study in Golgi technique and electron microscopy. J Neurol Sci 2012; 322: 117–121.

63.

Bryant

Jayakumar

, et al. Endothelial cells are heterogeneous in different brain regions and are dramatically altered in Alzheimer's disease. bioRxiv 2023; 2023: 2002. 2016.528825.

64.

Bird

Burgess

The hippocampus and memory: insights from spatial processing. Nat Rev Neurosci 2008; 9: 182–194.

65.

Frisoni

Fox

Jack

Jr , et al. The clinical use of structural MRI in Alzheimer disease. Nat Rev Neurol 2010; 6: 67–77.

66.

DeTure

Dickson

DW.

The neuropathological diagnosis of Alzheimer’s disease. Mol Neurodegener 2019; 14: 32–18.

67.

Steinman

Sun

H-S

Feng

Z-P.

Microvascular alterations in Alzheimer's disease. Front Cell Neurosci 2020; 14: 618986.

68.

Baek

Bennett

Hyman

, et al. MRI assessment of vascular disruption in tau pathology animal model (rTg4510 mouse) of Alzheimer disease. 2020 International Society of Magnetic Resonance in Medicine Virtual Conference & Exhibition 2020, 1941.

69.

Guo

Zhang

, et al. Age‑and brain region‑associated alterations of cerebral blood flow in early Alzheimer's disease assessed in AβPPSWE/PS1ΔE9 transgenic mice using arterial spin labeling. Mol Med Rep 2019; 19: 3045–3052.

70.

Pathak

Ward

Schmainda

KM.

A novel technique for modeling susceptibility-based contrast mechanisms for arbitrary microvascular geometries: the finite perturber method. Neuroimage 2008; 40: 1130–1143.

71.

Hooijmans

Graven

Dederen

, et al. Amyloid beta deposition is related to decreased glucose transporter-1 levels and hippocampal atrophy in brains of aged APP/PS1 mice. Brain Res 2007; 1181: 93–103.

72.

Cao

Zhu

, et al. ADAMTS13 maintains cerebrovascular integrity to ameliorate Alzheimer-like pathology. PLoS Biol 2019; 17: e3000313.

73.

Paris

Patel

DelleDonne

, et al. Impaired angiogenesis in a transgenic mouse model of cerebral amyloidosis. Neurosci Lett 2004; 366: 80–85.

74.

Qiu

Tan

, et al. Chronic cerebral hypoperfusion enhances tau hyperphosphorylation and reduces autophagy in Alzheimer’s disease mice. Sci Rep 2016; 6: 23964–23969.

75.

Shimada

Shindo

Matsuyama

, et al. Chronic cerebral hypoperfusion upregulates leptin receptor expression in astrocytes and tau phosphorylation in tau transgenic mice. Neurosci Lett 2019; 704: 133–140.

76.

Bryant

Carlyle

, et al. Cerebrovascular senescence is associated with tau pathology in Alzheimer's disease. Front Neurol 2020; 11: 575953.

77.

Farrar

Kamoun

Ley

, et al. In vivo validation of MRI vessel caliber index measurement methods with intravital optical microscopy in a U87 mouse brain tumor model. Neuro Oncol 2010; 12: 341–350.

78.

Karbe

Kertesz

Davis

, et al. Quantification of functional deficit in Alzheimer's disease using a computer-assisted mapping program for 99m Tc-HMPAO SPECT. Neuroradiology 1994; 36: 1–6.

79.

Lacalle-Aurioles

Aleman-Gomez

Guzman-De-Villoria

, et al. Is the cerebellum the optimal reference region for intensity normalization of perfusion MR studies in early Alzheimer’s disease? PLoS One 2013; 8: e81548.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

2.33 MB