Sage Journals: Discover world-class research

Abstract

Changes in cerebral blood flow are an essential feature of Alzheimer’s disease and have been linked to apolipoprotein E-genotype and cerebral amyloid-deposition. These factors could be interdependent or influence cerebral blood flow via different mechanisms. We examined apolipoprotein E-genotype, amyloid beta-deposition, and cerebral blood flow in amnestic mild cognitive impairment using pseudo-continuous arterial spin labeling MRI in 27 cognitively normal elderly and 16 amnestic mild cognitive impairment participants. Subjects underwent Pittsburgh Compound B (PiB) positron emission tomography and apolipoprotein E-genotyping. Global cerebral blood flow was lower in apolipoprotein E ɛ4-allele carriers (apolipoprotein E4+) than in apolipoprotein E4− across all subjects (including cognitively normal participants) and within the group of cognitively normal elderly. Global cerebral blood flow was lower in subjects with mild cognitive impairment compared with cognitively normal. Subjects with elevated cerebral amyloid-deposition (PiB+) showed a trend for lower global cerebral blood flow. Apolipoprotein E-status exerted the strongest effect on global cerebral blood flow. Regional analysis indicated that local cerebral blood flow reductions were more widespread for the contrasts apolipoprotein E4+ versus apolipoprotein E4− compared with the contrasts PiB+ versus PiB− or mild cognitive impairment versus cognitively normal. These findings suggest that apolipoprotein E-genotype exerts its impact on cerebral blood flow at least partly independently from amyloid beta-deposition, suggesting that apolipoprotein E also contributes to cerebral blood flow changes outside the context of Alzheimer’s disease.

Keywords

Mild cognitive impairment amnestic arterial spin labeling beta-amyloid apolipoprotein E

Introduction

Alterations in regional cerebral blood flow (CBF) in subjects with Alzheimer’s disease (AD) or mild cognitive impairment (MCI) have been consistently described.^1,2 These changes largely overlap with abnormalities in cerebral metabolism measured by 18^F-fluorodeoxyglucose positron emission tomography (FDG-PET).³ While CBF has been measured traditionally in the clinical setting using radioactive tracers, arterial spin labeling (ASL) MRI offers a non-invasive approach to CBF measurement⁴ that can be obtained in conjunction with structural MRI scanning. ASL-MRI utilizes magnetically labeled blood protons as an endogenous tracer for quantification of brain perfusion and does not require injections or exposure to ionizing radiation. In several recent reports, ASL-MRI has been applied to prodromal and clinical AD patient cohorts, displaying sensitivity to early disease stages such as amnestic MCI.^5,6 Additionally, CBF measured with ASL-MRI showed predictive value in the detection of cognitive decline and conversion from MCI to dementia.⁷ However, conflicting results were found in a recent large multi-site study, in which ASL-MRI was performed across a spectrum including cognitively normal (CN) adults to cognitive impaired subjects (early MCI and late MCI) and AD.⁸ Significant hypoperfusion relative to CN was only observed in late MCI and AD, questioning the sensitivity of ASL-MRI for detecting early cognitive impairment.

However, as MCI may be an etiologically heterogeneous diagnostic entity, pathological factors that might contribute to CBF changes in preclinical or prodromal AD need to be considered. Non-fibrillar, soluble assembly states of amyloid beta (Aβ)-peptides constituting the neurofibrillary tangles⁹ are believed to disrupt synaptic function, even before significant neurodegeneration occurs. By the time cognitive impairment becomes clinically significant, a significant number of synapses have already been lost.¹⁰ Currently, biomarkers are systematized as either pathological markers, e.g. amyloid-PET, CSF-Aβ, CSF-Tau/pTau versus topographical or progression markers, such as FDG-PET or structural MRI¹¹ or amyloid-markers versus markers of neuronal injury.¹² Alterations of brain structure and CBF were previously demonstrated in CN¹³ and in MCI⁶ with significant amyloid load. However, in animal models investigating the effect of Aβ on CBF, changes in CBF have only been found in aging Aβ knock-in mice,¹⁴ while direct injections of Aβ in rat hippocampus inhibited excitatory synaptic function but had no effect on CBF.¹⁵ In addition, it has been shown in (amyloid precursor protein) transgenic mouse models of AD that in the presence of cerebral beta-amyloid angiopathy (CAA) a decline of regional CBF could be observed, but not in a mouse model of beta-amyloidosis and negligible CAA, suggesting that CAA may be essential for CBF reductions.¹⁶

The APOE ɛ4 allele conveys the strongest genetic risk for late-onset AD¹⁷ and is a robust predictor of elevated amyloid-deposition in cognitively healthy subjects.¹⁸ However, there are conflicting results on the influence of the ɛ4 allele on brain metabolism.¹⁹ APOE genotype alters CBF²⁰ and the changes reported depended on cognitive/disease status.²¹ Using ASL in CN with a family history of AD, APOE4+ was associated with elevated baseline perfusion in the medial temporal lobe.¹⁹ However, locally reduced CBF (e.g. in the frontal association cortex) has also been reported for APOE4+ CN and AD,^20,21 while APOE4+ MCI demonstrated elevated regional CBF.²⁰ ApoE4+ CN elderlies were found to have higher regional CBF at baseline but a greater longitudinal decline in regional CBF over an observation period of ten years.²²

Although some studies have looked either at APOE4− or Aβ-induced CBF changes,²³ current literature reporting CBF differences between CN and MCI or AD does not take into account both of these early disease-sensitive markers.

In this study, we examined CN and aMCI subjects using ASL-MRI for CBF, PiB-PET for amyloid load, and APOE genotyping. Based on the previous studies examining the role of amyloid-deposition or APOE ɛ4 allele on CBF, we hypothesize that CBF is specifically altered in PiB+ and APOE4+ participants but not in the aMCI group independently of PiB and APOE status. To determine the factor (i.e. group (controls/aMCI), amyloid (PiB−/PiB+) and genetic risk (APOE4−/APOE4+)) with the strongest influence on global CBF, we applied univariate general linear model analyses.

Methods

Participants

Study participants were recruited from two cohort studies and the resulting study sample overlaps with a previously published sample.²⁴ All subjects provided written informed consent. Cognitive health was defined by a Mini-Mental State Examination (MMSE) score ≥27 and a clinical and neuropsychological assessment not indicative of MCI or other cognitive disorders. Amnestic MCI was diagnosed according to standard criteria.²⁵ Participants had to be of age ≥55 years. Exclusion criteria were somatic or psychiatric conditions or use of any medication that would significantly affect cognition. Patients were excluded for this project if they had contraindications for MR or PET imaging or evidence of focal lesions in critical memory structures. ASL analyses were performed by investigators blinded to the APOE and PiB status. The study was approved by the ethics committee of canton Zurich, Switzerland (E_22_2009 and E_64_2009) and was done in accordance to the World Medical Association Declaration of Helsinki (2013).

Clinical and neuropsychological assessment

All participants underwent a comprehensive clinical assessment including clinical history, physical and neurological examination. Further, individuals underwent neuropsychological examination using the CERAD-Plus test battery,²⁶ including MMSE, letter fluency, verbal learning, recall and recognition, figure copy and recall, Boston naming test, trail making tests A and B, category, and letter fluency. Also the German version of the Verbal Learning Memory Test (VLMT, subtests: delayed and recognition), the Visual Paired Associates test from the Wechsler Memory Scale Revised, and the Stroop Test were analyzed. Raw scores were z-transformed based on test-specific normative data adjusted for age and gender.

Structural imaging

MR scans were acquired on a 3T Philips Ingenia scanner (Philips Healthcare, Best, The Netherlands), equipped with a 16-channel head coil. A T1-weighted 3D-MPRAGE anatomical scan was recorded prior to the ASL scan (TR/TE: 8.2/3.8 ms, voxel resolution: 1 × 1 × 1 mm, field of view: 240 mm and 170 slices). Grey matter (GM) and white matter (WM) differences were compared between CN and aMCI using voxel-based morphometry (data will be reported elsewhere).

ASL acquisition

ASL data were acquired using the Philips 2D pseudo-continuous ASL (pCASL) sequence.²⁷ The acquisition parameters were: time of repetition (TR)/time of echo (TE) = 4200/16 ms, flip angle: 90°, FOV = 240 mm, voxel size: 3 × 3 mm, 20 slices, thickness: 6 mm (no gap), imaging matrix = 80 × 80, labeling duration: 1.65 s, post-labeling delay: 1.53 s, SENSE factor: 2.5, and duration 6:26 min. Background suppression was used with two pulses: 1.68 s and 2.76 s.

Equilibrium brain tissue magnetization (M0) images were recorded in a separate run for each subject using the same parameters as described for the pCASL sequence, apart from the TR (10,000 ms) and TE (15 ms).

ASL analysis

ASL images were preprocessed using the pipeline implemented in ASLtbx.²⁸ The first step was motion correction and denoising. Subjects were excluded from subsequent analysis if any of the three translation parameters exceeded half of the voxel size (i.e. 3 mm) or if rotation values exceeded 1°. Denoising included spatial smoothing with an isotropic Gaussian at full-width-at-half-maximum (FWHM) of 6 mm³ to reduce interindividual anatomical differences and further increase the signal-to-noise ratio. The next step was pair-wise subtraction and CBF quantification using the one-compartment model.²⁹ All CBF images were normalized to the Montreal neurological image (MNI) template space to allow for statistical group comparison (see below).

CBF quantification

CBF was calculated on a voxel-wise basis according to the formula

CBF (ml / 100 g / min) = \frac{60 \times 100 \times λ \times (M_{control} - M_{label})}{2 α T_{1 blood} \times M_{0} (e^{\frac{- w}{T_{1 blood}}} - e^{\frac{- w - τ}{T_{1 blood}}})}

M_control − M_label reflects the subtraction of label and control images, and λ = blood brain partition coefficient for water = 0.9,³⁰ T_1blood = 1664 ms,³¹ τ = labeling pulse train length = 1.68 s, α = labeling efficiency = 0.85,³² as background suppression was used, and w (post-tagging delay) = 1.53 s.

The labeling efficiency and the T1 of blood were taken from literature values, derived from previous experimental studies.^31,32 The equilibrium magnetization of blood was calculated from the equilibrium magnetization of CSF, measured in 4 ROIs, multiplied by a correction factor for T2* decay and the relevant blood H₂O partition coefficient taken from the literature.³⁰

For the calculation of the perfusion difference images (M_control– M_label), simple subtraction was used because it has been demonstrated to efficiently minimize spurious BOLD contaminations within the CBF signal in the case of resting-state recordings.³³ Furthermore, it has been demonstrated that simple subtraction in resting-state CBF data works with the same performance as special filtering approaches.³³ CBF maps were shown in mL/100 g/min.

Global CBF calculation

First, a brain mask template was created (using the FSL software package, http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/) to exclude all non-brain voxels. Then, a mean whole-brain CBF map (uncorrected for partial volume effects) was calculated by the mean of all voxels within the mask for each subject.

Voxel-wise CBF analysis

Additionally, partial volume effect (PVE) correction was applied to the normalized CBF maps using the SPM8 and MATLAB compatible (open-source) toolbox Grocer (version 2.11, http://www.nitrc.org/projects/fmri_grocer/) to correct for GM atrophy, as recommended for ASL analysis in neurodegenerative disorders. The Grocer toolbox offers several tools to analyze ASL-based CBF data (and also fMRI data). For our study, we used a batch to extract global PVE (here grey matter) corrected CBF values for each individual. In order to match the spatial resolution of the (smoothed) CBF images, the same isotropic Gaussian filter (6 × 6 × 6 mm) was applied to the GM mask as for the CBF images. Local CBF (and amyloid load) analysis was performed in 90 cortical and subcortical (excluding all sub-regions of the cerebellum) regions of interest (ROI) from the Anatomical Automatic Labeling (AAL) atlas.³⁴

Time of flight angiography

To envisage whether baseline cerebral perfusion within- or across groups was compromised by any vascular pathology in the large and medium sized vessels supplying the brain (e.g. shows lateralization effects in any group), we performed a 3D time of flight angiography (parameters: TR/TE: 25/3.5 ms, slices: 140 (gap 10 mm), scan resolution: 0.35 × 0.61 × 0.55 mm, FOV: 200, SENSE factor: 3, duration: 4.39 min) focusing on the internal carotid arteries and the circle of Willis.

PiB-PET

The method for PET-scanning has been described in detail elsewhere.³⁵ The cut-off for amyloid-positivity (i.e., PiB+) was ≥1.265 and was derived from 93 CN, which underwent PiB-PET in the context of a previously published cohort study.²⁴ Image processing was performed using PMOD V3.5.

Genotyping

APOE genotyping was performed as previously described.³⁶ Participants were classified (APOE4− status) as either carriers (APOE4+) or non-carriers (APOE4−) of the ɛ4 allele.

Statistical analysis

Chi-square (χ²) test was applied for comparisons of categorical data (e.g. gender, PiB status, APOE status). Mann-Whitney U test for continuous variables. For the statistical analysis on global CBF effects, a two-tailed t-test was applied for the three contrasts of interest: group (CN versus aMCI), amyloid status (PiB− versus PiB+), and genetic risk status (APOE4− versus APOE4+). Next, a univariate general linear model (GLM) was applied to assess voxel-wise (not masked by any brain tissue content) CBF effects for these three contrasts. Participants’ specific GM volume was used as covariate of no interest to eliminate any residual PVE at the intra-voxel level so that the observed group differences are not due to a reduced GM fraction in MCI subjects (relative to controls). Next, voxel-wise PVE-corrected ASL differences were assessed using unpaired t-test statistics for four contrasts: (1) CN versus aMCI (2) APOE4− versus APOE4+ (3) PiB− versus PiB+ and (4) CN APOE4− versus CN APOE+. All results are reported at a voxel-threshold of p < 0.01 (t ≥ 2.44, with an additional cluster threshold of k > 51 voxel to achieve a cluster-corrected threshold of p < 0.05). The total intracranial volume (TIV) and age were included as covariates of no interest. Local CBF group comparisons were only tested for the contrast (1) to (3), and were reported at p < 0.05 (uncorrected). We refrained from testing for local associations of CBF and local Aβ, as visual inspection of scatterplots did not reveal a monotone relationship (data not shown), as Aβ was not normal distributed. A univariate GLM was applied to test for the influence of factors ‘APOE4,’ ‘PiB,’ and ‘diagnostic group’ on global CBF. Age was included as covariate. We subsequently eliminated some of the interactions to optimize the GLM and report the effect sizes indicated by partial eta². Further we provide parameter estimates for the three main contrasts (i.e., contrasts (1) to (3)) of the PVE-corrected analyses. All statistical analyses were performed with SPSS (IBM, Version 22).

Results

Sample characteristics

In total, ASL data were recorded from 48 subjects. Five subjects were excluded from further analysis; two because of strong head motion (translation > 3 mm) during ASL-MRI, two due to failed CBF-MNI coregistration (one of whom also showed intraluminal spins indicative of transit time artefacts), and one participant due to structural abnormalities (ischemic lesions, periventricular hyperintensities, and dilated right ventricle). Based on the MR angiography, none of the subjects showed occlusion or significant stenosis of any major vessels, assessed by an experienced neuroradiologist. The final sample consisted thus of 27 CN and 16 patients with aMCI (one left-hander). Diagnostic groups did not differ significantly in age, sex, and education (Table 1), neither did these parameters differ comparing APOE4− vs. APOE4+ or PiB+ vs. PiB− individuals. However, VLMT scores were significantly lower in aMCI than in CN.

Table 1.

Sample characteristics.

	Group
	CN (N = 27)		aMCI (N = 16)
	Mean	SD	Mean	SD	p
Age (years)	71.8	4.4	75.5	8.0	0.06
Gender (male n)	17		12		0.42^a
Education (years)	14.4	2.6	14.8	4.2	0.71
MMSE score (/30)	29.6	0.7	28.5	1.2	<0.001
Neuropsychological scores
VLMT	0.38	0.70	−1.71	1.21	<0.001
Digit span forward	0.15	1.04	−0.66	0.86	0.01
Digit span backward	0.39	1.03	−0.36	0.99	0.02
Trail making	−0.45	1.12	−0.33	0.98	0.72
Letter fluency	0.48	1.45	−0.56	0.96	0.01
Category fluency	0.54	1.34	−0.28	0.66	0.02
Nr. of (recognized) figures	0.49	1.04	−0.37	0.79	<0.01
Stroop	0.56	0.47	0.02	1.12	0.03
Boston naming	0.37	0.75	−0.37	1.21	0.01
CERAD word learning	0.76	1.06	−0.33	1	<0.01
CERAD recall	0.53	1	−0.56	0.68	<0.001
APOE ɛ4 carrier, APOE4+ (n)	9		6		0.78^a
PiB(+) status (n)	7		7		0.22^a
Global cortical PiB (median with IQR)	1.15	0.24	1.26	0.76	0.03
Global cortical PiB (in subgroups)	Median				IQR
All individuals with APOE4+ (n = 15)	1.55				0.96
All individuals with APOE4− (n = 28)	1.14				0.05
All individuals with PiB+ (n = 14)	1.79				0.63
All individuals with PiB− (n = 29)	1.14				0.04

Chi-square.

Note: For PiB values, median with IQR is given, as PiB shows a bimodal distribution.

Group comparisons were either performed by Mann-Whitney U-test or by Chi-square statistics depending on scale of measure. Significant p-values are indicated in bold letters (p < 0.05).

IQR: interquartile range; VLMT: German version of the Rey Auditory Verbal Learning Test.

Diabetes was present in three individuals. The drug record (nearest to the date of the study) is provided in Table 2. In brief, it shows that only few participants (including the aMCI) were using medication (except for statins, antihypertensives, and acetylsalicylic acid). There were no significant differences (unpaired two-tailed t-test) in the number of category-specific drugs between CN and aMCI (all p ≥ 0.1).

Table 2.

Drug regimen for controls (CN) and aMCI.

	Anti-	Benzo-	Anti-	Anti-	Antihyper-		Anti-	Antico-	Acetylsalicylic	Clopido-
nr.	psychotics	diazepines	depressants	Parkinsonians	tensives	Statins	diabetics	agulants	acid	grel	NSAID
aMCI (average and fraction)	0.00	0.06 (1/16)	0.06 (1/16)	0.00	0.31 (5/16)	0.31 (5/16)	0.00	0.06 (1/16)	0.38 (6/16)	0.00	0.06 (1/16)
CN (average)	0.00	0.07 (2/27)	0.07 (2/27)	0.00	0.44 (12/27)	0.30 (8/27)	0.00	0.00	0.26 (7/27)	0.00	0.00
CN vs. aMCI (p-value)	n.a.	0.889	0.889	n.a.	0.404	0.914	n.a.	0.197	0.436	n.a.	0.197

NSAID: nonsteroidal anti-inflammatory drugs.

Note: The level of medication did not differ between groups.

CBF analysis

Mean PVE-corrected CBF was inversely correlated to age (across all subjects: p = 0.002, r = −0.46). Anatomically detailed mean PVE-corrected CBF perfusion maps are shown in Figure 1 for the contrast APOE4− and APOE4+ (pooled across CN and MCI), indicating generally lower CBF in the APOE4+ group. Global CBF (corrected for grey matter) differed between groups, as indicated in Figure 2 (CN > aMCI, p = 0.042, APOE4− > AOPE4+, p = 0.007). The contrast PiB− > PiB+ revealed a trend for lower CBF in the PiB+ (p = 0.11). For illustration and quality assurance, global (i.e. uncorrected for PVE) mean whole-brain CBF maps are shown for all aMCI participants in supplementary Figure 1. Global mean whole-brain CBF was higher for CN than for aMCI (p = 0.079), for APOE4− relative to APOE4+ (p = 0.039), and for PiB− relative to PiB+ (p = 0.059). For the GLM on mean whole-brain CBF, we used GM as covariate of no interest, which was not different between CN and aMCI (mean CN: 554.8 ml, mean aMCI: 551.0, p = 0.98) but differed between APOE4− and APOE4+ (mean APOE4−: 565.9 ml, mean APOE4+: 529.2, p = 0.018) and PiB− and PiB+ (mean PiB−: 569.3 ml, mean PiB+: 531.7, p = 0.009). The GLM analysis revealed that the results look similar for all contrasts (supplementary Figure 2), suggesting that the type of atrophy correction (PVE correction versus GM as covariate in the GLM) does not alter the statistical results substantially. For example, the APOE4 contrast still demonstrated the strongest between group differences (relative to the other contrasts), with a spatially similar distribution for the PVE-corrected (Figure 3) and uncorrected CBF maps (supplementary Figure 2B). Voxel-wise PVE-corrected analysis revealed CBF reductions in aMCI in the left putamen and middle occipital gyrus (Brodmann area, BA 19), as well as in frontal and temporal regions (Table 3a, Figure 3). In contrast, CBF reductions in PiB+ subjects were seen in the ACC (BA 32), middle cingulate cortex, visual cortex, and cerebellum (Table 3b and Figure 3). The most spatially widespread CBF reductions were found for the contrast APOE4− versus APOE4+ (Figure 3, Table 3c). APOE4+ subjects exhibited lower CBF in several frontal regions but also in parietal regions as well as in the hippocampus, putamen, and thalamus. Additionally a within-group (CN only) comparison revealed lower CBF in APOE4+ than in APOE4− participants (p = 0.029). A within-group analysis (only CN) revealed APOE4+ was related to hypoperfusion in the inferior frontal gyrus (IFG) as well as in the ACC and middle frontal gyrus, cingulate cortex, thalamus, caudate nucleus, insula, putamen, and parietal regions (Table 3e, Figure 4).

Table 3.

Between group differences in cerebral blood flow.

Region	Hemisphere	Brodmann area	MNI coordinates
(a) Voxel-wise PVE-corrected CBF differences for CN > aMCI (not separated by PiB and APOE4 status).
Anterior cingulate cortex	Left	32	−5 53 −4
Medial frontal gyrus	Left	10	−8 50 15
Inferior frontal gyrus	Left	47	−43 27 −10
Insula	Left	13	−39 10 −6
	Left		−35 −4 12
Superior temporal gyrus	Left	41	−52 −30 13
	Right	n.a.	38 15 −27
	Left	n.a.	−33 10 −18
	Right	22	40 −58 13
Middle temporal gyrus	Left	21	−52 −24 −10
Inferior temporal gyrus	Left	37	−51 −39 −23
Precuneus	Left	7	−21 −59 42
Angular gyrus/inferior parietal lobe	Left	39/40	−35 −59 48
Middle occipital gyrus	Left	19	−30 −82 17
	Right		39 −72 10
Putamen	Left	n.a.	−25 2 7
	Right	n.a.	28 7 3
(b) Voxel-wise PVE-corrected differences for PiB− > PiB+ (not separated by APOE4 and clinical status).
Cuneus	Left		−22 −97 −4
Lingual gyrus	Left	18	−32 −93 0
Anterior cingulate cortex	Left	32	−11 41 0
	Right	32	1 41 14
Middle cingulate cortex	Right		1 33 30
Cerebellum (posterior lobe)	Right	n.a.	13 66 −31
	Right	n.a.	22 −65 −42
(c) Voxel-wise PVE-corrected CBF differences for APOE4− > APOE4+ (not separated by PiB and clinical status).
Thalamus (pulvinar)	Left	n.a.	−14 −27 5
Thalamus (medial dorsal nucleus)	Left	n.a.	−5 −15 6
Hippocampus	Right	n.a.	19 −27 −8
Superior temporal gyrus	Right	21	66 −22 −4
	Right	42	67 −32 18
Putamen	Left	n.a.	−26 12 −4
	Right	n.a.	25 14 −3
Middle frontal gyrus	Left		−40 40 −13
	Right		20 −40 −12
	Right		41 29 36
	Right	10	30 43 26
Medial frontal gyrus	Right	11	5 55 −10
Anterior cingulate cortex	Left	32	−6 49 3
Superior prefrontal gyrus	Right		11 60 −6
Inferior frontal gyrus	Right		32 28 −14
	Right	47	41 14 −6
Insula	Right		41 12 −5
Paracentral lobe	Left	5	−6 −40 53
Precuneus	Left	7	−1 −51 62
	Left		−10 −41 62
Middle temporal gyrus	Left		−38 −77 19
	Right	21	66 −21 −6
Cerebellum (culmen)	Left		−31 −55 −31
Cerebellum (tonsil)	Right		46 −46 −47
	Left		−35 −60 −46
Cerebellum (inferior semi-lunar lobule)	Right		18 −63 −50
Inferior parietal lobe	Right	40	65 −30 22
(d) Voxel-wise PVE-corrected CBF differences for APOE4+ > APOE4− (not separated by PiB and clinical status).
Medial frontal (subgenual anterior cingulate cortex)	Left	11/25	−2 28 −14
(e) Voxel-wise PVE-corrected CBF differences for CN APOE− > CN APOE+ (not separated by PiB status).
Anterior cingulate cortex	right	10/32	9 47 15
Medial frontal gyrus	Left	10	−8 59 8
	Right		10 42 43
	Right	9	10 25 35
Inferior frontal gyrus	Right	47	38 29 −8
Middle cingulate cortex	Left		−4 31 32
Supplementary motor area	Left		−4 24 46
Thalamus	Left		−3 −10 6
Thalamus	Left		−15 −32 4
Hypothalamus	Left		−4 −4 −9
Caudate nucleus	Left		−10 15 7
Insula	Right		37 8 6
Putamen	Right		21 15 6
	Left		−24 6 −5
Paracentral lobe	Right		3 −32 57
Precuneus	Right		2 −40 51
	Right		3 −52 55
	Left	7	−10 −68 54
Inferior parietal lobe	Right	40	61 −42 30
	Left		−56 −53 36
(Para)hippocampus	Left		−23 −36 −4
(Para)hippocampus	Right		23 −35 −1
Lingual gyrus	Left		−21 −54 −8
Middle occipital gyrus	Right		31 −94 −3
Posterior cingulate cortex	Left	31	−8 −47 36
Angular gyrus	Right		44 −64 35
	Left		−51 −64 36
Cerebellum (culmen)	Right		26 −53 −18
Cerebellum (tonsil)	Right		39 −54 −46
	Left		−42 −53 −44
Middle temporal gyrus	Right	21	56 9 −27
Superior temporal gyrus	Right		66 −19 1
Fusiform gyrus	Left		−28 −52 −14
	Right		28−60 −14

Note: Brodmann areas are given if applicable. n.a.: not applicable; PVE: partial volume corrected (using grey matter).

Figure 1.

PVE-corrected CBF maps for the groups APOE4− and APOE4+. Lighter areas represent areas with higher CBF. CBF maps are shown on axial slices: range z = −32 to z = 64 (gaps: 12 mm) and are given in mL/100 g/min. PVE: partial volume corrected (using grey matter).

Figure 2.

Statistical grey matter based CBF group comparisons. All comparisons (CN versus aMCI, APOE4− versus APOE4+, and CN APOE4− versus APOE4+) showed statistical differences (p < 0.05) or a trend (PiB− versus PiB+). The comparison PiB− versus PIB+ yields a trend, with higher mean CBF for PiB−.

Figure 3.

Voxel-wise PVE-corrected CBF group differences. The different colors indicate three statistical comparisons: APOE4− versus APOE4+ (red), PiB− versus PiB+ (green), CN versus aMCI (blue). All results are shown at p < 0.01 (cluster-enhancement correction, k > 51 voxel; and corrected for age, total intracranial volume). APOE4+: carriers of the ɛ4 allele, PiB+: global Aβ > 1.265 (group cut-off value). PVE: partial volume corrected (using grey matter).

Figure 4.

Voxel-wise PVE-corrected CBF differences for the contrast CN APOE4− versus CN APOE4+ (yellow) and APOE4− versus APOE4+ (red). APOE4+: carriers of the ɛ4 allele; CN: controls. Orange/dark red areas indicate overlap of CBF across both contrasts. PVE: partial volume corrected (using grey matter).

To test for the strongest influence of the factors of main interest (APOE4, PiB, and group) on PVE-corrected CBF, we applied a univariate GLM (corrected for age). The overall significant GLM (F_(6,36) = 4.07, p = 0.003, partial eta²= 0.404) contained the following independent factors and interactions: group, PiB, APOE4, age, PiB x age interaction, and PiB × group interaction. We found the highest effect sizes (partial eta²) for the factor APOE4 (F_(1,36) = 5.01, p = 0.031, partial eta²= 0.122), followed by a trend for PiB (F_(1,36) = 3.3, p = 0.078, partial eta²= 0.084). In contrast, factor group was not significant. The analysis of the parameter estimates further revealed the highest partial eta² values for the contrasts: APOE4− versus APOE4+ (partial eta²: 0.122) relative to the PiB contrast (PiB− versus PiB+: 0.092) and group contrast (CN versus aMCI: 0.092). A full list of PVE-corrected CBF group differences is reported in Table 3.

ROI-based CBF analysis

The results of this ROI-based analysis are reported uncorrected and would not have survived Bonferroni correction for 90 comparisons. Henceforth, we also did not correlate local CBF with (multiple) neuropsychological test scores. A detailed summary of the local CBF group differences is provided in Table 4.

Table 4.

Between group differences in cerebral blood flow in regions of interest.

Region	Hemisphere	p-value
(a) Local ROI-based CBF differences for CN versus aMCI (independent of PiB and APOE status).
Inferior frontal gyrus, triangular	Left	0.04
Inferior frontal gyrus, orbital	Left	0.03
superior temporal pole	Left	0.04
(b) Local ROI-based CBF differences for PiB− versus PiB+ (across aMCI and CN).
Inferior frontal gyrus, orbital	Left	0.04
Inferior frontal gyrus, medial orbital	Left	0.1^a
Anterior cingulate cortex	Left	0.04
Anferior occipital lobe	Left	0.09^a
Superior temporal pole	Right	0.03
(c) Local ROI-based CBF differences for APOE- versus APOE+ (across aMCI and CN)
Precentral gyrus	Right	0.04
Middle frontal gyrus	Right	0.02
Middle frontal gyrus, orbital	Right	0.02
Inferior frontal gyrus, triangular	Right	0.04
Inferior frontal gyrus, orbital	Left	0.02
	Right	0.01
Superior frontal gyrus, medial	Left
	Right	0.02
Superior frontal gyrus, medial orbital	Left	0.03
	Right
Anterior cingulate cortex	Right	0.03
Inferior frontal gyrus, opercular	Right	0.05
Middle occipital lobe	Left	0.04
Superior parietal lobe	Right	0.03
Putamen	Right	0.02
Cerebellum crus	Left	0.02
	Right	0.04
Cerebellum	Left	0.01
	Right	0.03

Trend.

Discussion

In this study, we evaluate differences in CBF in CN and aMCI using ASL-MRI. Furthermore, we have studied the influence of elevated amyloid-deposition and APOE-genotype on CBF. Our study demonstrates that ASL-MRI is capable of assessing perfusion changes associated with amnestic MCI, APOE-status and cerebral amyloid-deposition. The major finding of this study was that APOE4+ show lower global CBF than APOE4−. Lower CBF was found in frontal, parietal and temporal areas. Furthermore, PiB+ and aMCI also showed reduced overall CBF (relative to PiB− and CN, respectively) but effects were weaker and did not reach statistical significance for the contrast PiB− versus PiB+. Local CBF reductions were more widespread for the contrast APOE4−carriers versus non-carriers.

ASL in discrimination of MCI and CN

Global CBF was lower in MCI than in CN which was not caused by occlusion of larger cerebral vessels. Several regions survived PVE-, TIV-, age, and cluster-size correction for all group differences. PCASL has been recommended as the most sensitive ASL technique for the assessment of perfusion differences driven by metabolic changes due to neurodegeneration.⁸ Recently, one 3D PCASL study found weaker CBF differences between CN and MCI in fronto-parieto-occipital regions than the contrast CN > AD.³⁷ In our study, CN > aMCI CBF differences were seen in some frontal regions, the (middle) occipital and temporal cortex and additionally in the left putamen. The relatively small CBF differences between CN versus aMCI (i.e. relative to the spatially more widespread CBF differences for the contrast APOE4− versus APOE4−) are in line with a recent study which did not find CBF differences between MCI and CN if other structural markers (here Aβ) were not taken into account.⁶ The subjects analyzed in this study are part of a larger sample, where lower CBF, estimated by early tissue uptake of the PiB tracer (ePiB), was found in the left posterior cingulate cortex, bilateral hippocampus, and left IFG in aMCI. Though the methodology among the studies is not entirely comparable (regions selected hypothesis based, different read out, atlas, absence of partial volume correction in the former study) we also identified a group difference in the left IFG in this smaller sample, which is in line with the importance of this region for a variety of cognitive abilities.^38–40 Furthermore, we were able to identify clusters of lowered CBF in the left angular gyrus/inferior parietal lobe and in the left precuneus in the voxel-based approach in regions where FDG hypometabolism has also been described.^41,42 We did not identify changes in the hippocampus and in the posterior cingulate as in the previous study and as has been consistently found in the FDG-PET-literature.

Influence of APOE and PiB on CBF

APOE exerted the strongest effect on global CBF assessed by ASL. The most widespread CBF differences were seen for the contrasts: APOE4− versus APOE4+ (across all participants) and for APOE4− versus APOE4+ (within controls), and these findings were not driven by vascular abnormalities (as assessed by TOF angiography). Similar to our study, one recent pulsed ASL-MRI study found reduced regional CBF in APOE4+ participants in several brain regions.²⁰ However, the effects were found in CN and AD but not in MCI, where CBF was increased in discrete regions. We first decided to pool CN and aMCI groups to increase the sensitivity to APOE effects due to the overall small number of ɛ4 allele carriers within aMCI. However, even in the analysis within the CN group (19 APOE4− versus 9 APOE4+) we found widespread fronto-parietal CBF reductions in APOE4+, also including important subcortical regions, such as the thalamus, caudate nucleus, and putamen, which have been shown to be structurally or functionally abnormal in aMCI (or AD).^43–45

In the ROI-based approach, we did not identify differences in typical AD regions, with the exception of the right precuneus, consistent with our previous report on ePiB.²⁴ Additionally, the reductions in CBF were not seen in typical AD regions for the contrast PiB+ versus PiB−. A very recent study examined the link between amyloid load and CBF in AD-dementia and MCI⁶ in 182 subjects in total, including 105 MCI subjects. Similar to our findings, CBF was lower in subjects with high Aβ (independent of diagnostic group). In the two MCI groups (i.e. early and late MCI), however, ASL reductions in PiB+ subjects were seen in the inferior parietal and temporal cortex but not in the precuneus, entorhinal cortex or in the frontal cortices (where changes were observed only in AD dementia). As the authors acknowledge, the APOE genotype and amyloid-pathology may have independent statistical effects on CBF. This notion is also supported by a recent FDG-PET study.⁴⁶ Our findings on the CBF difference between APOE4 carriers and non-carriers are consistent with findings of FDG-PET results showing hypometabolism as widespread throughout the brain.⁴⁷ In addition, the regions we examine partially co-localize with regions that exerted a longitudinal decline in CBF assessed by H₂0-PET. However, this study described higher baseline CBF in ApoE4 carriers at baseline, which we only observed in the anterior cingulate cortex. A recent meta-analysis⁴⁸ identified the right middle frontal gyrus, the region which is most consistently identified to be hypometabolic in FDG-PET studies. This region also showed reduced CBF in our analysis. Interestingly, one study only identified effects of APOE4 and age on cerebral glucose metabolism but not of amyloid-status measured by florbetapir-PET.⁴⁹ Together with these findings, our results strongly suggest that APOE4 contributes to changes in global CBF and that these changes are widespread and not bound to amyloid-pathology or limited to typical regions showing early hypometabolism in AD. Several reasons may account for this. First APOE4 may contribute to neurovascular dysfunction independently of amyloid-deposition.⁵⁰ In mice expression of APOE4 but not E3 or E2 led to blood brain barrier dysfunction and CBF reductions.⁵¹ Other factors by which APOE4 may cause changes in CBF are its interaction with mitochondrial function, and its influence on neuroinflammation, which have been reviewed in detail elsewhere.⁵²

Limitations

As our study is cross sectional we cannot elaborate on the predictive diagnostic value of ASL, e.g. whether reduced CBF in addition to PiB+ and APOE4+ status is predictive for more rapid cognitive decline. The data for the regional CBF analyses are not corrected for multiple comparisons and must thus be considered explorative. However, the regions identified do include clusters where we also found lowered global CBF. Additionally, further evidence for the value of perfusion measurements with ASL-MRI as a biomarker for neuronal dysfunction and injury (e.g. structural MRI, CSF total tau/phosphorylated tau) in MCI and AD must be collected. Technically, the ASL signal is directly related to local blood flow but quantification in physiological units critically depends on the transit delay time. Although large changes in the transit time due to the presence of significant stenosis were excluded in this study by the MR angiography, observed group differences might also reflect inter-individual differences in transit time.

In contrast to multi inversion-time ASL recording schemes, the standard pCASL implementation does not allow for estimation of the arterial transit time. However, the advantage of PCASL is that the SNR is maximized. Due to the known decrease of perfusion in cognitively impaired elderly patients, the perfusion SNR is expected to be lower than in other (e.g. younger healthy) populations. For this reason, pCASL was selected over other ASL implementations for the present study. However, the population examined is also arguably likely to demonstrate prolonged transit times. As it is known that the delivery time is slower in the elderly,⁵³ it is possible that complete delivery of the labeled bolus was not fully achieved in our study, as the labeling delay was 1.53 s (albeit identical for both CN and subjects with aMCI). However, the use of longer than optimal post-labeling delay times would result in strong T1 decay of the label and reduced signal to noise ratio.

In cases with prolonged transit times and without a separate estimate of the arterial transit time, perfusion may be either overestimated (due to the longer T1 of blood relative to tissue, resulting in intraluminal spins at the time of image readout with artefactually increased perfusion) or underestimated. In our data, the masked, PVE-corrected GM CBF maps mostly did not include vascular regions prone to transit time artefacts from delayed perfusion, so transit time effects are more likely to result in an underestimate rather than an overestimate of the CBF, since transit time effects are not accounted for in the single time point pCASL measurement. One subject did show focused (parietal) hyperintensities indicative of a potential transit time artifact (MCI38, see supplemental Figure 1). However, this artifact seems unlikely to affect the statistical result, as this subject was APOE4+ and group-results rather indicate higher CBF in parietal regions for the APOE4− group (Figure 3).

Finally, it would have been informative to perform a hierarchical (regression) analysis, in order to examine the most causal factor (e.g. APOE4 > PiB > group) on CBF. However, we want to emphasize that only a path analysis (using structural equation modeling or partial least square path modeling) can provide such information about the causal interactions. This was not applicable for our study. Firstly, because of our small sample size (at least 30 observations would be required in each sample group, considering just the three predictors APOE4, PiB, and group) and second because such an analysis would require a priori knowledge about both the main variables (e.g. APOE4) and relevant latent variables (e.g. lifestyle, other risk factors for AD such as diabetes, depression, etc.), which might also influence CBF.

Conclusion

From a clinical diagnostic perspective, we provide additional evidence that ASL-MRI is capable of identifying subjects at risk of Alzheimer’s dementia. From the neurobiological perspective, we conclude that CBF is strongly influenced by inheritance of an APOE epsilon4 allele and that this influence is at least partly independent of the contribution of APOE to amyloid-deposition.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was made possible by SNF Grant Nos. 33CM30_124111, 33CM30_140335 and 320030_125378.

Acknowledgments

We thank all subjects for their participation in this study. We furthermore thank Isabella Blum, Esmeralda Gruber and Faith Sieber (study nurses), Diana Bundschuh and Wiebke Buck (technical assistance), Sabine Spoerri, Stefan Kluge and Stefan Doppler (data management), and all study physicians and neuropsychologists for help with data acquisition. We also thank Sarah Grimm for her help on the statistical analyses.

Declaration of conflicting interests

Christoph Hock is co-founder and board member of Neurimmune Holding AG, Schlieren. The other authors have no conflicts of interest.

Authors’ contributions

LM, AG, CH, and SK designed the study. LM, FR, RM collected the data. LM, GM, ROG, GW, AG, AK, and SEL analyzed the data. LM, AG, FR, AB, SK, ROG, and GW wrote the paper. LM and GW contributed equally to the article. SK and AFG shared senior authorship.

Supplementary material

Supplementary material for this paper can be found at

References

Johnson

Jones

Holman

. Preclinical prediction of Alzheimer's disease using SPECT. Neurology 1998; 50(6): 1563–1571.

Ishii

Sasaki

Yamaji

. Demonstration of decreased posterior cingulate perfusion in mild Alzheimer's disease by means of H215O positron emission tomography. Eur J Nucl Med 1997; 24(6): 670–673.

Landau

Harvey

Madison

. Associations between cognitive, functional, and FDG-PET measures of decline in AD and MCI. Neurobiol Aging 2011; 32(7): 1207–1218.

Detre

Leigh

Williams

. Perfusion imaging. Magn Reson Med 1992; 23(1): 37–45.

Alexopoulos

Sorg

Forschler

. Perfusion abnormalities in mild cognitive impairment and mild dementia in Alzheimer's disease measured by pulsed arterial spin labeling MRI. Eur Arch Psychiatry Clin Neurosci 2012; 262(1): 69–77.

Mattsson

Tosun

Insel

. Association of brain amyloid-beta with cerebral perfusion and structure in Alzheimer's disease and mild cognitive impairment. Brain 2014; 137(Pt 5): 1550–1561.

Chao

Buckley

Kornak

. ASL perfusion MRI predicts cognitive decline and conversion from MCI to dementia. Alzheimer Dis Assoc Disord 2010; 24(1): 19–27.

Wang

Das

Xie

. Arterial spin labeled MRI in prodromal Alzheimer's disease: A multi-site study. NeuroImage Clin 2013; 2: 630–636.

Walsh

Klyubin

Fadeeva

. Naturally secreted oligomers of amyloid beta protein potently inhibit hippocampal long-term potentiation in vivo. Nature 2002; 416(6880): 535–539.

10.

Scheff

Price

Schmitt

. Synaptic loss in the inferior temporal gyrus in mild cognitive impairment and Alzheimer's disease. JAD 2011; 24(3): 547–557.

11.

Dubois

Feldman

Jacova

. Advancing research diagnostic criteria for Alzheimer's disease: the IWG-2 criteria. Lancet Neurol 2014; 13(6): 614–629.

12.

Sperling

Aisen

Beckett

. Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimer's Dementia: J Alzheimer's Assoc 2011; 7(3): 280–292.

13.

Sojkova

Beason-Held

Zhou

. Longitudinal cerebral blood flow and amyloid deposition: an emerging pattern? J Nucl Med 2008; 49(9): 1465–1471.

14.

Abrahamson

Foley

Dekosky

. Cerebral blood flow changes after brain injury in human amyloid-beta knock-in mice. J Cereb Blood Flow Metab 2013; 33(6): 826–833.

15.

Smith

Walsh

. Soluble amyloid-beta peptides potently disrupt hippocampal synaptic plasticity in the absence of cerebrovascular dysfunction in vivo. Brain 2008; 131(Pt 9): 2414–2424.

16.

Maier

Wehrl

Schmid

. Longitudinal PET-MRI reveals beta-amyloid deposition and rCBF dynamics and connects vascular amyloidosis to quantitative loss of perfusion. Nature Med 2014; 20(12): 1485–1492.

17.

Farrer

Cupples

Haines

. Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. JAMA 1997; 278(16): 1349–1356.

18.

Mielke

Wiste

Weigand

. Indicators of amyloid burden in a population-based study of cognitively normal elderly. Neurology 2012; 79(15): 1570–1577.

19.

Fleisher

Podraza

Bangen

. Cerebral perfusion and oxygenation differences in Alzheimer's disease risk. Neurobiol Aging 2009; 30(11): 1737–1748.

20.

Kim

Rhee

. Regional cerebral perfusion in patients with Alzheimer's disease and mild cognitive impairment: effect of APOE epsilon4 allele. Neuroradiol 2013; 55(1): 25–34.

21.

Hogh

Knudsen

Kjaer

. Single photon emission computed tomography and apolipoprotein E in Alzheimer's disease: impact of the epsilon4 allele on regional cerebral blood flow. J Geriatric Psych Neurol 2001; 14(1): 42–51.

22.

Thambisetty

Beason-Held

. APOE epsilon4 genotype and longitudinal changes in cerebral blood flow in normal aging. Arch Neurol 2010; 67(1): 93–98.

23.

Lowe

Weigand

Senjem

. Association of hypometabolism and amyloid levels in aging, normal subjects. Neurology 2014; 82(22): 1959–1967.

24.

Gietl

Warnock

Riese

. Regional cerebral blood flow estimated by early PiB uptake is reduced in mild cognitive impairment and associated with age in an amyloid-dependent manner. Neurobiol Aging 2015; 36(4): 1619–1628.

25.

Winblad

Palmer

Kivipelto

. Mild cognitive impairment–beyond controversies, towards a consensus: report of the international working group on mild cognitive impairment. J Intern Med 2004; 256(3): 240–246.

26.

Morris

Heyman

Mohs

. The consortium to establish a registry for alzheimer's disease (CERAD). Part I. clinical and neuropsychological assessment of Alzheimer's disease. Neurology 1989; 39(9): 1159–1165.

27.

Wong

Buxton

Frank

. Implementation of quantitative perfusion imaging techniques for functional brain mapping using pulsed arterial spin labeling. NMR Biomed 1997; 10(4-5): 237–249.

28.

Wang

Aguirre

Rao

. Empirical optimization of ASL data analysis using an ASL data processing toolbox: ASLtbx. Magn Reson Imag 2008; 26(2): 261–269.

29.

Buxton

Frank

Wong

. A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 1998; 40(3): 383–396.

30.

Herscovitch

Raichle

. What is the correct value for the brain–blood partition coefficient for water? J Cereb Blood Flow Metab 1985; 5(1): 65–69.

31.

Clingman

Golay

. Determining the longitudinal relaxation time (T1) of blood at 3.0 Tesla. Magn Reson Med 2004; 52(3): 679–682.

32.

Dai

Garcia

de Bazelaire

. Continuous flow-driven inversion for arterial spin labeling using pulsed radio frequency and gradient fields. Magn Reson Med 2008; 60(6): 1488–1497.

33.

Liu

Wong

. A signal processing model for arterial spin labeling functional MRI. Neuroimage 2005; 24(1): 207–215.

34.

Tzourio-Mazoyer

Landeau

Papathanassiou

. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. NeuroImage 2002; 15(1): 273–289.

35.

Riese

Gietl

Zolch

. Posterior cingulate gamma-aminobutyric acid and glutamate/glutamine are reduced in amnestic mild cognitive impairment and are unrelated to amyloid deposition and apolipoprotein E genotype. Neurobiol Aging 2015; 36(1): 53–59.

36.

Hixson

Vernier

. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res 1990; 31(3): 545–548.

37.

Binnewijzend

Kuijer

Benedictus

. Cerebral blood flow measured with 3D pseudocontinuous arterial spin-labeling MR imaging in Alzheimer disease and mild cognitive impairment: a marker for disease severity. Radiology 2013; 267(1): 221–230.

38.

Costafreda

Lee

. A systematic review and quantitative appraisal of fMRI studies of verbal fluency: role of the left inferior frontal gyrus. Hum Brain Mapp 2006; 27(10): 799–810.

39.

Liakakis

Nickel

Seitz

. Diversity of the inferior frontal gyrus–a meta-analysis of neuroimaging studies. Behav Brain Res 2011; 225(1): 341–347.

40.

Katzev

Tuscher

Hennig

. Revisiting the functional specialization of left inferior frontal gyrus in phonological and semantic fluency: the crucial role of task demands and individual ability. J Neuroscience 2013; 33(18): 7837–7845.

41.

Mosconi

Tsui

Herholz

. Multicenter standardized 18F-FDG PET diagnosis of mild cognitive impairment, Alzheimer's disease, and other dementias. J Nucl Med 2008; 49(3): 390–398.

42.

Chen

Wolk

Reddin

. Voxel-level comparison of arterial spin-labeled perfusion MRI and FDG-PET in Alzheimer disease. Neurology 2011; 77(22): 1977–1985.

43.

Qiu

Fennema-Notestine

Dale

. Alzheimer's disease neuroimaging I. regional shape abnormalities in mild cognitive impairment and Alzheimer's disease. Neuroimage 2009; 45(3): 656–661.

44.

Tang

Holland

Dale

. Baseline shape diffeomorphometry patterns of subcortical and ventricular structures in predicting conversion of mild cognitive impairment to Alzheimer's disease. JAD 2015; 44(2): 599–611.

45.

Tang

Holland

Dale

. Alzheimer's disease neuroimaging I. Shape abnormalities of subcortical and ventricular structures in mild cognitive impairment and Alzheimer's disease: detecting, quantifying, and predicting. Hum Brain Mapp 2014; 35(8): 3701–3725.

46.

Knopman

Jack

Jr. Wiste

. 18)F-fluorodeoxyglucose positron emission tomography, aging, and apolipoprotein E genotype in cognitively normal persons. Neurobiol Aging 2014; 35(9): 2096–106.

47.

Reiman

Chen

Alexander

. Correlations between apolipoprotein E epsilon4 gene dose and brain-imaging measurements of regional hypometabolism. Proc Natl Acad Sci USA 2005; 102(23): 8299–8302.

48.

Liu

Wang

. APOE genotype and neuroimaging markers of Alzheimer's disease: systematic review and meta-analysis. J Neurol Neurosurg Psych 2015; 86(2): 127–134.

49.

Jagust

Landau

. Alzheimer's Disease Neuroimaging I. Apolipoprotein E, not fibrillar beta-amyloid, reduces cerebral glucose metabolism in normal aging. J Neurosci 2012; 32(50): 18227–18233.

50.

Zlokovic

. Cerebrovascular effects of apolipoprotein E: implications for Alzheimer disease. JAMA Neurol 2013; 70(4): 440–444.

51.

Bell

Winkler

Singh

. Apolipoprotein E controls cerebrovascular integrity via cyclophilin A. Nature 2012; 485(7399): 512–516.

52.

Liu

Kanekiyo

. Apolipoprotein E and Alzheimer disease: risk, mechanisms and therapy. Nat Rev Neurol 2013; 9(2): 106–118.

53.

Alsop

Detre

Golay

. 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. Magnetic Reson Med. Epub ahead of print 8 April. 2014. DOI: 10.1002/mrm.25197.

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

4.30 MB

Arterial spin labeling imaging reveals widespread and Aβ-independent reductions in cerebral blood flow in elderly apolipoprotein epsilon-4 carriers

Abstract

Keywords

Introduction

Methods

Participants

Clinical and neuropsychological assessment

Structural imaging

ASL acquisition

ASL analysis

CBF quantification

Global CBF calculation

Voxel-wise CBF analysis

Time of flight angiography

PiB-PET

Genotyping

Statistical analysis

Results

Sample characteristics

CBF analysis

ROI-based CBF analysis

Discussion

ASL in discrimination of MCI and CN

Influence of APOE and PiB on CBF

Limitations

Conclusion

Footnotes

Funding

Acknowledgments

Declaration of conflicting interests

Authors’ contributions

Supplementary material

References

Supplementary Material