Carotid Stiffness is Associated with Brain Amyloid-β Burden in Amnestic Mild Cognitive Impairment

Subjects

Thirty-two aMCI subjects (59% female, aged 55–78 years) were recruited from local newspaper advertisements, senior centers, and the UT Southwestern Alzheimer’s Disease Center participated. These individuals were a subgroup of study participants enrolled in a proof of concept study of effects of exercise training on neurocognitive function in aMCI (ClinicalTrials.gov, NCT01146717) [16]. Data presented herein were obtained at the study baseline, i.e., before the initiation of study interventions in these subjects. aMCI diagnosis was based on Petersen criteria as modified by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (http://adni-info.org) [13]. Alzheimer’s Disease Cooperative Study (ADCS) recommendations (http://www.adcs.org) further guided clinical evaluations using standard diagnostic criteria. Table 1 shows diagnostic cognitive assessments including the Mini-Mental State Examination (MMSE), Clinical Dementia Rating (CDR) scale, and Wechsler Memory Scale Logical Memory (LM) for immediate and delayed recalls. Subjects with major psychiatric disorders, major or unstable medical conditions, uncontrolled hypertension, diabetes mellitus, or chronic inflammatory diseases were excluded, as were subjects with cardiac pacemaker or any metal plates or pins in their body contraindicating MRI (detailed inclusion and exclusion criteria are provided in ClinicalTrials.gov NCT01146717). All subjects and/or their study partners gave informed consent. This study was approved by the Institutional Review Boards of the UT Southwestern Medical Center and Texas Health Presbyterian Hospital of Dallas.

Vascular assessment

All experiments were conducted in a temperature-controlled laboratory of about 22°C. Subjects were asked to refrain from vigorous exercise, caffeine, and alcohol for at least 24 h before experiments. Vascular measurements were gathered following ≥10 min of supine rest. Brachial cuff blood pressure was measured ≥3 times with an ECG-gated electrosphygmomanometer (Suntech, Morrisville, NC, USA). Obtained values were averaged to obtain systolic (SBP) and diastolic blood pressures (DBP). Pulse pressure (PP) was calculated as SBP-DBP. Mean arterial pressure (MAP) was calculated as DBP + 1/3(SBP-DBP). Heart rate was recorded from a 3-lead ECG (GE Solar 8000 M, Milwaukee, WI, USA). Aortic stiffness was determined with the gold standard method of measuring the carotid-femoral pulse wave velocity (cfPWV) [17, 18]. Sequential arterial tonometery of the common carotid and femoral arteries (SphygmoCor 8.0; AtCor Medical, West Ryde, NSW, Australia) captured the pulse pressure waveform gated with an ECG. ≥10 cardiac cycles were recorded to calculate cfPWV outlined in greater detail elsewhere [17, 18].

The right common carotid artery was insonated perpendicularly to the vessel 1–2 cm proximal to the carotid bulb in the C2–C4 region using 2-D vascular ultrasound with a 3 to 12 MHz linear array transducer (CX-50, Philips Healthcare). Mean luminal diameters and cIMT were obtained from ≥5 consecutive cardiac cycles and measured with wall tracking (Carotid Analyzer, Medical Imaging Applications) [19]. cIMT was defined as the distance between the leading edge of the lumen-intima interface and the leading edge of the media-adventitia interface of the far wall [20]. Carotid pressure waveforms were obtained by applanation tonometry and calibrated to brachial blood pressure [21, 22]. Carotid β stiffness index and distensibility were calculated by the following equations where cSBP and cDBP represent carotid systolic and diastolic pressures respectively and D represents carotid diameters [23]. Carotid β stiffness index is representative of carotid artery stiffening whereas carotid distensibility is representative of carotid flexibility, a complementary measurement of arterial stiffness [23]. Carotid β stiffness index = Ln ( cSBP - cDBP ) / [ ( D systole - D diastole ) / D diastole ] Carotid distensibility = ( D systole - D diastole ) / [ ( cSBP - cDBP ) × D diastole ]

Positron emission tomography image acquisition

Participants underwent an IV bolus injection of 10 mCi ¹⁸F-florbetapir and were situated on an imaging table of a Siemens (Munich, Germany) ECAT HR PET scanner for data acquisition using the laser guide for head position. Velcro straps and foam wedges were used to secure the participant’s head. To ensure the brain was completely in the field of view and absent of rotation in either the transverse or sagittal planes, a 2 min scout scan was acquired. At 50 min post injection, 2 frames of 5 min PET emission scan and a 7 min transmission scan were acquired in 3D mode using the following parameters: matrix size = 128×128, resolution = 5 mm×5 mm, slice thickness = 2.42 mm, and field of view = 58.3 cm. Emission images were processed by iterative reconstruction, 4 iterations and 16 subsets with a 3 mm full width at half maximum (FWHM) ramp filter. The transmission image was reconstructed using back-projection and a 6 mm FWHM Gaussian filter for attenuation correction [24, 25].

Image processing

Participant’s PET images were spatially normalized to a florbetapir uptake template (2×2×2 mm³ voxels) in MNI space using SPM8 (Wellcome Department of Cognitive Neurology, London, UK) and in-house MATLAB (MathWorks, Sherborn, MA) scripts, and visually inspected for registration quality. Standardized uptake value ratio (SUVR) was computed using mean cerebellar uptake as a reference. The mean cortical SUVR was calculated as the average of the anterior cingulate cortex (ACC), posterior cingulate cortex (PCC), precuneus, temporal, dorsolateral prefrontal, orbital frontal, parietal, and occipital SUVRs [24]. Participants with a mean brain Aβ≥1.11 were classified dichotomously as amyloid positive [25]. For the purpose of this study, partial volume corrections were not applied because of potential bias may be introduced by the different modeling methods used [26].

To confirm region of interest (ROI) based analyses, whole-brain voxel-based analysis was performed. After the Aβ SUVR maps of each subject were normalized to the florbetapir uptake template, voxel-based correlations were calculated between Aβ SUVR maps and cfPWV, carotid β stiffness index, and carotid distensibility using the “3dfim+” routine in AFNI software. To assess the effect of the multiple comparisons, Monte Carlo simulation was performed according to the image matrix and voxel sizes of the average-size brain of these subjects and the approximate spatial smoothness. A spatial smoothness of 6.71 mm FWHM was estimated based on the image filters applied in PET data pre-processing steps. Assuming that amyloid plaques tend to change in clusters and based on the Monte Carlo simulation using the “AlphaSim” routine in AFNI software, we applied a minimal connected cluster size of 840 mm³ to correct the voxel-based p < 0.005 to reach a corrected p < 0.0475 for significant correlations between arterial stiffness and Aβ SUVR maps.

White matter hyperintensity (WMH) has been used widely to assess cerebral small vessel disease and has been related to aortic arterial stiffness [27]. To quantify WMH burden, full-brain 2D T2 FLAIR images were collected on a subset of participants (n = 24) on a Philips Achieva 3T scanner (Philips Healthcare, Best, the Netherlands) with an 8-channel receiver coil with the following parameters: axial, time of echo (TE) = 125 ms, time of repetition (TR) = 11 s, time of inversion (TI) = 2,800 ms, field of view (FOV) = 23 cm×23 cm, slice thickness = 5 mm, number of slices = 24 with 1 mm gaps, acquisition matrix size = 352×212, and reconstructed matrix size = 512×512. WMH regions were segmented on each 2D image through the lesion prediction algorithm (LPA) implemented in the Lesion Segmentation Toolbox (LST) version 2.0.12 for Statistical Parametric Mapping (SPM12) (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/). A threshold of 0.5 was used on the obtained lesion probability maps to identify WMH regions. WMH volumes were expressed as a percentage of total intracranial volume [28].

APOE genotyping

Peripheral blood mononuclear cells (PBMC) were obtained from the antecubital vein via IV blood draw and centrifugal Ficoll-based separation. PBMCs were cryopreserved in the media containing 50% human serum on the day of collection. One million cells were thawed, and DNA extracted using the DNeasy Blood and Tissue kit (Qiagen; Venlo, Netherlands) for genotype analysis. APOE genotype was identified using TaqMan SNP genotyping assays (Life Technologies; Carlsbad, CA). APOE genotype data were available from 26 participants.

Statistical analysis

Descriptive data are summarized by means, standard deviations, and ranges for continuous variables. Categorical variables are represented as frequency and percent of sample. Independent and dependent variable distribution normality was determined with the Shapiro-Wilk test with all variables displaying normal distributions (p > 0.05) except for carotid β stiffness index (p = 0.008) and WMH (p < 0.001). Upon visual inspection of Q-Q plots, no outliers were observed for carotid β stiffness index. The WMH volume distribution was severely skewed and a logarithmic transformation (logWMH) was performed to acquire normality.

To determine the relationship of arterial stiffness (independent variables) and potential covariates (e.g., age, sex, blood pressure, body mass index) with a priori ROIs of the mean cortical, ACC, PCC, and precuneus SUVR (dependent variable), simple correlations of covariates were performed (model 1). Model 2 aimed to determine the unique effect of arterial stiffness with regional Aβ SUVR independent of age, sex, blood pressure, and APOE4 status. A critical alpha of p < 0.05 was used for all ROI based analyses. All statistical analyses were performed with SPSS 23.0 (Chicago, IL, USA).

Arterial stiffness and brain Aβ

Previous work has explored the relationship of arterial stiffness with Aβ [9–12, 29–31]. A two-year longitudinal investigation of non-demented elderly adults showed that measurement of cfPWV was associated with accelerated Aβ accumulation [11]. In that study, Aβ positive individuals also had significantly elevated brachial-ankle (baPWV) and femoral-ankle pulse wave velocity when compared with Aβ negative peers [11]. In very elderly dementia free adults, baPWV was associated with increased Aβ deposition while cfPWV was not [10]. In the Atherosclerosis Risk in Communities Study, heart-carotid pulse wave velocity was associated with greater Aβ deposition in 320 older adults with mixed cognitive status [12]. However, a relation between global Aβ burden and cfPWV was absent [12]. In a separate study, pulse blood pressure, a measure often used as a surrogate for central arterial stiffening, was positively correlated with Aβ deposition as measured by PiB distribution volume ratio in frontal, temporal, and posterior-cingulate/precuneus regions in healthy middle-age individuals [31]. Recently, an investigation assessed the relationship of aortic stiffness by cardiac magnetic resonance imaging with cerebral blood flow (CBF) and cerebrovascular reactivity to hypercapnic stimulus (CVR) in cognitively normal volunteers and patients with MCI [32]. Participants with elevated aortic stiffness who were also APOE4 carriers demonstrated lower regional CBF, but higher CVR [32]. In a study of cognitively normal older adults, individuals who were amyloid positive (global Aβ SUVR > 1.1) demonstrated an attenuated cerebrovascular response from rest to moderate intensity exercise as assessed by absolute change in middle cerebral artery blood flow velocity [33]. Taken together, these studies, in general, illustrate a potential relationship between large arterial stiffening and brain Aβ accumulation and CBF regulation in older adults. However, the inconsistencies regarding which measures of arterial stiffening are associated with brain Aβ highlight the need to better understand the effects of systemic versus regional arterial stiffness on brain Aβ. Still, none of these studies have investigated the relationships of aortic and carotid artery stiffening with brain Aβ in patients with aMCI.

The present study found that although cfPWV was correlated positively with mean cortical, PCC, and precuneus Aβ burden, these associations were no longer significant after adjusting for age, sex, MAP, PP, and APOE4 status. Conversely, carotid β stiffness index and carotid distensibility were associated with mean cortical and precuneus Aβ even after adjusting these covariates and these observations are consistent with the whole brain voxel-based analysis.

Central arterial stiffness with advanced aging, reflected mainly by aortic stiffening, has been well established [7]. Pathophysiologically, central arterial stiffening can lead to increases in both systemic blood pressure and blood flow pulsatility which may have deleterious effects on the brain and other end-organs due to increased biomechanical stress alteration of blood flow profile such as from laminar to turbulent flow, or both, thus damages to the microcirculation [6]. Of note, the effects of these age-related vascular and hemodynamic changes on the end-organs may depend more on the local than the distant arterial stiffness, which serve as the circulatory gateways that act to dampen systemic blood pressure and blood flow oscillations.

In this regard, carotid artery stiffening reflects directly reduced buffering capacities of the feeding arteries to the cerebral circulation on blood pressure and blood flow pulsatility. Consequently, increases in blood pressure/flow pulsatility may propagate downstream into the cerebral microcirculation leading to damage in endothelial cell structure and function and blood-brain barrier which are essential for brain Aβ clearance [6]. In addition, damage in arteriolar smooth muscle cells may diminish its capability to dilate or constrict in response to changes in arterial pressure (cerebral autoregulation), resulting in intermittent ischemic-hypoxic like perfusion which may also impair brain Aβ clearance [5]. Finally, altered blood pressure/flow pulsatiltiy associated with cerebral arterial stiffening may influence Aβ clearance via peri- and paravascular drainage mechanisms [34–36]. In support of this hypothesis, data from the Northern Manhattan Study showed an association of the middle and anterior cerebral artery diameters, used as surrogates of arterial stiffness, with brain paravascular spaces, which was modified by pulsatile systemic hemodynamics [37]. This finding suggests that increasing blood pressure or flow pulsatility is related to dilated brain arteries and increases in cerebral paravascular spaces which may influence the clearance of brain Aβ via the glymphatic drainage systems [38].

However, it must be acknowledged that brain Aβ accumulation or neuronal degeneration may also lead to cerebral arteriolar dysfunction, which then may lead to increases in cerebrovascular resistance, reductions in CBF, and arterial stiffening [39].

The observation that carotid artery stiffening is correlated positively with precuneus Aβ after adjusting for age, sex, MAP, PP, and APOE4 status is worthy of discussion. The precuneus is part of the default mode network (DMN) which is vulnerable to AD [40]. The increases in Aβ and decreases in perfusion and metabolic activity in the precuneus have also been associated with the progression from normal cognitive function to AD in older adults [41]. Although the role of the DMN in age-related cognitive decline and AD pathogenesis and development remains to be fully elucidated, there is evidence that the posterior cingulate and precuneus give rise to recollection of prior experiences [42]. The findings of the current study suggest a vascular mechanism by which carotid artery stiffening may contribute to precuneus Aβ accumulation in patients with aMCI. However, no associations of cIMT were observed with neither brain Aβ nor WMH. The absence of this association suggests a more important role for arteriosclerotic than atherosclerotic processes in brain Aβ accumulation.

Clinical implications

Multiple investigations have linked large arterial stiffening or remodeling to the risk of AD [43–45]. Recently, a large multi-ethnic cohort study from the Washington Heights-Inwood Columbia Aging Project found that larger carotid artery diameters were associated with increased risks of AD [43]. Data from the Framingham Offspring study as well as Cardiovascular Health Study Cognition Study (CHSCS) indicated that higher aortic stiffness was associated with an increased risk of MCI, all-cause dementia, and AD [44, 45]. However, after controlling for MAP, diabetes, and high-density lipoprotein cholesterol, cfPWV only predicted future incident MCI [44]. The CHSCS also found that cfPWV was significantly associated with increased risk for incident dementia independent of education, race, APOE4 status, diabetes, body mass index, mean arterial pressure, and anti-hypertensive medication [45]. Additionally, a 5-year longitudinal follow-up of MCI patients found modifiable vascular risk factors, such as hypertension, diabetes, and hypercholesterolemia known to contribute to arterial stiffening, were related to increased risk of conversion to dementia [46].

The current investigation extends these previous works by demonstrating that it is the regional carotid rather than central arterial stiffening measured by cfPWV that are associated positively with brain Aβ burden in patients with aMCI after adjusting for age, sex, MAP, PP, and APOE4 status. These findings highlight the potential role of vascular contributions to AD onset and development. A critical question then is whether early treatment of arterial stiffening may prevent or slow AD progression [46]. Recently, a large randomized controlled trial of older adults demonstrated that intensive blood pressure control (SBP < 120 mm Hg) significantly reduced the risk of MCI when compared with standard care (SBP < 140 mm Hg) [47]. Despite the well-established relationship between hypertension and arterial stiffening, it remains to be determined if aggressive blood pressure treatment or other interventions to improve vascular health such as exercise and a healthy diet may attenuate large artery, particularly carotid artery stiffening, thus leading to a reduction in brain Aβ and improvement in brain function.

Strengths and limitations

This study is strengthened by the direct measurement of brain Aβ plaque burden using ¹⁸F-florbetapir PET in a well characterized clinical population of aMCI patients. In addition, well-established methods for assessment of both aortic and carotid artery stiffness were used to examine their relations with brain Aβ. Major limitations are the relatively small sample size and cross-sectional data which limits the understanding of causality between arterial stiffness and brain Aβ. Consequently, we cannot surmise definitively that carotid stiffening precedes brain Aβ accumulation, vice-versa, or if the two processes happen simultaneously. The aMCI patients in the present study were otherwise healthy (i.e., without uncontrolled hypertension, diabetes mellitus, and other major medical conditions) which may limit the generalizability of this study, thus our findings await confirmations from future investigations.

Conclusions

By measuring both aortic and carotid stiffness and examining their associations with brain Aβ burden in patients with aMCI, we found that carotid rather than aortic stiffening was independently associated with brain Aβ after adjustment for age, sex, MAP, PP, and APOE4 status. These findings indicate that carotid artery stiffening, which has a direct impact on and is more pertinent to the cerebral circulation than the aorta, may contribute to AD pathology in aMCI patients.