Hippocampal volume mediates the association of arterial stiffness with cognitive impairment in adult population

Abstract

Objectives:

To investigate the association of arterial stiffness with brain perfusion, brain tissue volume and cognitive impairment in the general adult population.

Materials and methods:

This prospective study included 1488 adult participants (age range: 22.8–83.9 years) from the Kailuan study. All participants underwent brachial–ankle pulse wave velocity (PWV) measurement, brain MRI, and Montreal Cognitive Assessment (MoCA). The association of PWV with cerebral blood flow (CBF), brain tissue volume and MoCA score was investigated. Mediation analysis was used to determine whether CBF and brain tissue volume changes mediated the associations between PWV and MoCA score.

Results:

A 1 standard deviation (SD) increase in PWV was associated with lower total brain CBF [β (95% CI) −0.67 (−1.2 to −0.14)], total gray matter CBF [β (95% CI) −0.7 [−1.27 to −0.13)], frontal lobe CBF [β (95% CI) −0.59 (−1.17 to −0.01)], parietal lobe CBF [β (95% CI) −0.8 (−1.43 to −0.18)], and temporal lobe CBF [β (95% CI) −0.68 (−1.24 to −0.12)]. Negative associations were found for PWV and total brain volume [β (95% CI) −4.8 (−7.61 to −1.99)] and hippocampus volume [β (95% CI) −0.08 (−0.13 to −0.04)]. A 1 SD increase PWV was significantly associated with elevated odds of developing cognitive impairment [odds ratio (95% CI) 1.21 (1.01–1.45)]. Mediation analysis showed that hippocampal volume partially mediated the negative association between PWV and MoCA scores (proportion: 14.173%).

Conclusion:

High arterial stiffness was associated with decreased total and regional CBF, brain tissue volume, and cognitive impairment. Hippocampal volume partially mediated the effects of arterial stiffness on cognitive impairment.

INTRODUCTION

With the aging of the population, the prevalence of dementia is increasing. The number of individuals with dementia worldwide is expected to rise from the current 50 million to 150 million by 2050 [1]. To date, the exact pathogenesis of dementia has not been clarified, but increasing evidence suggests that dementia and cardiovascular disease share numerous of common risk factors [2,3]. At present, known risk factors for cardiovascular disease, such as smoking, hypertension, and glucose and lipid metabolism disorders, are negatively associated with dementia [4,5]. Controlling or eliminating these risk factors can not only reduce the risk of cardiovascular events in the population but also help delay the progression of dementia.

Arterial stiffness is one of the hallmarks of vascular aging [6]; it reduces the buffering effect of the major arteries on the pulsating pressure produced by cardiac ejection, resulting in damage to low-resistance organs such as the heart, brain and kidney [7,8]. Arterial stiffness has been identified as a risk factor for cardiovascular events including stroke [9,10], and more importantly is associated with cognitive decline and dementia [11,12]. Recent studies have found that arterial stiffness is a risk factor for brain structure and perfusion changes, such as reduced brain tissue volume and cerebral blood flow (CBF) and increased white matter lesions [13–15]. However, the above studies included middle-aged and elderly adults from Europe and the Americas, some of whom were symptomatic, and the sample size of participants who completed brain MRI examination was relatively small. Moreover, brain structure is coupled with brain function, and whether brain structure changes mediate the association between arterial stiffness and cognitive function remains unclear. Thus, we used the multimodal brain imaging data embedded in the Kailuan study to systematically analyze the associations among arterial stiffness, brain volume and perfusion changes, and cognitive function in the general Asian adult population, and further explored whether brain volume and perfusion changes mediate the association between arterial stiffness and cognitive impairment by mediation analysis.

MATERIALS AND METHODS

The study was approved by the Medical Ethics Committee of our institution, and all participants signed informed consent. The data for this study were from the Kailuan study, a prospective cohort study conducted in Tangshan, China. From December 2020 to March 2023, a total of 1585 adult participants were initially included in this study and underwent brachial–ankle pulse wave velocity (PWV) examinations, laboratory tests, cognitive assessment and brain MRI examinations. We excluded 22 participants with missing PWV data, 6 participants who experienced a stroke, 2 participants with a history of tumor, and 67 participants with incomplete images. The flow chart is shown in Fig. 1. Finally, 1488 participants were included in this study.

FIGURE 1.

Flow chart of the inclusion for the participants in this study. PWV, pulse wave velocity.

Pulse wave velocity measurement

The PWV measurement protocol has been described in detail previously [16,17]; in short, this measurement was conducted by a professional using an arterial stiffness detection device [Omron Health Medical Co., Ltd, Dalian, China]. After sitting in a room at a comfortable temperature for at least 5 min, the participants were asked to lie down on an examination bed in a supine position and remain quiet during the measurement. Cuffs were wrapped around both ankles and arms. The average value of the right and left PWV was used for the analysis.

MRI data acquisition

Brain MRI was acquired using a 3.0 T MRI scanner (General Electric 750 W, Milwaukee, Wisconsin, USA). The imaging protocol included 3D pseudocontinuous (3D-pc) ASL images, 3D T1-brain volume (BRAVO) sequence, T2-weighted sequence, 3D T2 FLAIR sequence, and diffusion weighted imaging (DWI) sequence.

3D-pc ASL images were acquired with the following parameters: repetition time (TR) = 5313 ms; echo time (TE) = 10.7 ms; flip angle = 111°; slice thickness = 4 mm; postlabel delay (PLD) = 2525 ms; in-plane resolution, 3.37 mm × 3.37 mm; and FOV = 256 mm × 256 mm.

Brain structural images included T2-weighted images [TE = 99.5 ms; TR = 6099 ms; flip angle) = 142°; matrix = 320 × 320; slice thickness = 5 mm; field of view (FOV) = 256 mm × 230.4 mm], 3D T2 FLAIR images (TE = 118.9 ms; TR = 5000 ms; flip angle = 15°; FOV = 256 mm × 230.4 mm; slice thickness = 1 mm; matrix = 256 × 256), DWI (TR = 5110 ms; TE = 77.2 ms; slice thickness = 5 mm; flip angle = 90°; FOV = 240 mm × 240 mm; matrix = 130 × 160), and 3D T1-BRAVO images (TR/TE = 6.7/2.6 ms; flip angle = 15°; gap = 1 mm; slice thickness = 1 mm; FOV = 256 mm × 256 mm; 170 slices).

MRI data processing

3D-pc ASL data were processed using SPM12 as described previously[18]. First, the CBF maps were automatically generated from 3D ASL Functool software (AW 4.6 Workstation, GE Healthcare, Milwaukee, Wisconsin, USA). Then, the CBF maps were coregistered to the Montreal Neurological Institute space. Next, we used a 6 mm full width at half maximum (FWHM) Gaussian kernel to smooth the CBF maps. Finally, average CBF values were measured in the total brain, total gray matter and regions of interest of relevance to cognitive impairment: frontal lobe, parietal lobe, temporal lobe, and hippocampus (Fig. 2) [19–21]. The ROI was selected from the Anatomical Automatic Labeling Template in SPM12, as reported previously [22].

FIGURE 2.

Cerebral blood flow and brain volume of regions of interests (frontal lobe, parietal lobe, temporal lobe, and hippocampus) were extracted based on anatomical automatic labeling template. CBF, cerebral blood flow.

3D T1-brain volume scans were processed using SPM12. The detailed process, including repositioning, segmentation and normalization, has been reported in previous study [20]. Based on the above preprocessing, total intracranial volume, total gray matter volume and total white matter volume were calculated. The total brain tissue volume was defined as the sum of gray matter volume and white matter volume [23]. In addition, average values were measured using Anatomical Automatic Labeling Template in the regions of interest of relevance to cognitive impairment: frontal lobe, parietal lobe, temporal lobe and hippocampus (Fig. 2) [19–21].

Cognitive assessment

Cognitive function was assessed by a professional using the Montreal Cognitive Assessment (MoCA) prior to brain MRI examination. This scale has a maximum score of 30 points and assesses 7 cognitive domains: visuospatial/executive functions, attention, naming, language, recall, abstraction, and orientation. If a participant has 12 years or fewer of education, 1 point will be added to the total MoCA score. The total MoCA score ranges from 0 to 30, with a lower score reflecting poorer cognitive function. A score of less than 26 indicated cognitive impairment [24].

Covariate measurement

Information on age, sex, smoking, drinking, physical exercise, and past medical history was acquired via questionnaires. Weight, height and blood pressure (BP) were measured by an experienced professional. The BMI was calculated based on height and weight. Low-density lipoprotein cholesterol and total cholesterol were measured using an automated analyzer (Hitachi 747, Hitachi, Tokyo, Japan). Hypertension was defined as BP at least 140/90 mmHg or use of antihypertensive medication. Diabetes was defined as fasting blood glucose at least 7.0 mmol/l or use of antidiabetic medicine. Dyslipidemia was defined as LDL-C at least 4.1 mmol/L or HDL-C 1.0 mmol/l or less or triglycerides at least 2.3 mmol/l or total cholesterol at least 6.2 mmol/l or use of lipid-lowering medication.

Statistical analysis

SAS version 9.4 (SAS Institute, Inc, Cary, North Carolina, USA) was used to perform statistical analyses. A P value less than 0.05 was considered significant. A generalized linear model was used to investigate the effect of PWV increase by 1 SD on CBF, brain tissue volume and MoCA score. A logistic regression analysis was performed to study the effect of PWV increase by 1 SD on cognitive impairment. Three models were constructed in all analyses. Model 1 adjusted for age, sex, BMI, alcohol consumption, smoking status and physical exercise; model 2 included covariates in model 1 and further adjusted for total cholesterol, LDL-C, fasting blood glucose, SBP, and total intracranial volume; and model 3 included covariates in model 2 and further adjusted for use of antihypertensive, hypoglycemic, and hypolipidemic medications.

To test the hypothesis that arterial stiffness affects cognitive function through brain perfusion and structure change, a mediation analysis was performed using the SAS CAUSALMED procedure to examine whether a given mediator (brain perfusion and structure) affects the association of the independent variable (PWV) and the outcome variable (MoCA scores). Covariates adjusted for mediating analysis were age, sex, BMI, drinking, smoking, physical exercise, SBP, total cholesterol, LDL-C, fasting blood glucose, and the use of antihypertensive, hypoglycemic and hypolipidemic medications.

RESULTS

Baseline characteristics

Detailed demographic and MRI marker characteristics are summarized in Table 1 and Supplemental Table 1. Of the 1488 participants, 759 (51.0%) were men, and the age ranged from 22.8 to 83.9 years old with a mean of 54.6 ± 11.9 years. The average PWV was 1502.9 ± 320.9 cm/s. A total of 799 (53.7%) participants had MoCA scores of less than 26 points.

TABLE 1.

Baseline characteristics of participants

Characteristics
N	1488
Age (range) (year)	54.6 ± 11.9 (22.8–83.9)
Male [n (%)]	759 (51.0)
Smoking history [n (%)]	507 (34.1)
Drinking history [n (%)]	651 (43.8)
Physical activity [n (%)]	487 (32.7)
BMI (kg/m2)	25.2 ± 3.5
SBP (mmHg)	132.9 ± 19.4
DBP (mmHg)	79.1 ± 12.3
FBG (mmol/l)	5.6 ± 1.5
Total CHOL (mmol/l)	5.2 ± 1.0
LDLC (mmol/l)	3.1 ± 0.8
Antihypertensive treatment (n (%)]	527 (35.4)
Hypoglycemic treatment (n (%)]	141 (9.5)
Lipid-lowering treatment [n (%)]	172 (11.6)
PWV (cm/s)	1502.9 ± 320.9
MoCA	25 (6–30)
Brain perfusion (ml/100 g/min)
Total brain	47.7 ± 7.2
Total gray matter	50.4 ± 7.7
Frontal lobe	49.1 ± 7.7
Parietal lobe	50.7 ± 8.4
Temporal lobe	49.7 ± 7.5
Hippocampus	45.1 ± 7.3
Brain tissue volume (ml)
Intracranial total volume	1433.4 ± 135.3
Total brain	1095.0 ± 105.1
Total gray matter	596.3 ± 53.4
Frontal lobe	166.3 ± 16.8
Parietal lobe	83.2 ± 7.7
Temporal lobe	117.0 ± 11.7
Hippocampus	7.5 ± 0.8

Data are shown as n (%), mean ± SD or median (range). FBG, fasting blood glucose; CHOL, cholesterol; LDLC, low-density lipoprotein cholesterol; PWV, pulse wave velocity; MoCA, Montreal Cognitive Assessment.

Pulse wave velocity and brain perfusion, brain tissue volume

Table 2 shows the association of PWV with total and regional CBF. In model 1, PWV was negatively associated with CBF in the total brain [β (95% CI) −0.64 (−1.09 to −0.19)] and total gray matter [β (95% CI) −0.7 (−1.18 to −0.22)] as well as the frontal lobe [β (95% CI) −0.68 (−1.17 to −0.2)], parietal lobe [β (95% CI) −0.89 (−1.41 to −0.36)], temporal lobe [β (95% CI) −0.64 (−1.1 to −0.17)] and hippocampus [β (95% CI) −0.5 (−0.96 to −0.04)]. In model 2, PWV was negatively associated with CBF in total brain [β (95% CI) −0.67 (−1.2 to −0.14)] and total gray matter [β (95% CI) −0.7 (−1.27 to −0.14)] as well as the frontal lobe [β (95% CI) −0.6 (−1.17 to −0.02)], parietal lobe [β (95% CI) −0.8 (−1.42 to −0.18)] and temporal lobe [β (95% CI) −0.69 (−1.25 to −0.14)]. Further adjustments in model 3 did not essentially change these associations. The mean CBF and CBF change maps for participants in the different PWV quartiles are shown in Supplemental Figure 1.

TABLE 2.

Association of pulse wave velocity with total and regional cerebral blood flow and brain tissue volume

	Model 1		Model 2		Model 3
	PWV per 1 SD higher β	P	PWV per 1 SD higher β	P	PWV per 1 SD higher β	P
CBF (ml/100 g/min)
Total brain	−0.64 (−1.09 to −0.19)	0.005	−0.67 (−1.2 to −0.14)	0.013	−0.67 (−1.2 to −0.14)	0.014
Total gray matter	−0.7 (−1.18 to −0.22)	0.004	−0.7 (−1.27 to −0.14)	0.015	−0.7 (−1.27 to −0.13)	0.017
Frontal lobe	−0.68 (−1.17 to −0.2)	0.006	−0.6 (−1.17 to −0.02)	0.042	−0.59 (−1.17 to −0.01)	0.045
Parietal lobe	−0.89 (−1.41 to −0.36)	0.001	−0.8 (−1.42 to −0.18)	0.012	−0.8 (−1.43 to −0.18)	0.012
Temporal lobe	−0.64 (−1.1 to −0.17)	0.008	−0.69 (−1.25 to −0.14)	0.015	−0.68 (−1.24 to −0.12)	0.017
Hippocampus	−0.5 (−0.96 to −0.04)	0.032	−0.53 (−1.08 to 0.01)	0.055	−0.54 (−1.09 to 0.01)	0.053
Brain tissue volume (ml)
Total brain	−8.93 (−14.14 to −3.71)	0.001	−5.11 (−7.92 to −2.29)	<0.001	−4.8 (−7.61 to −1.99)	0.001
Total gray matter	−3.31 (−5.9 to −0.72)	0.012	−1.18 (−2.95 to 0.58)	0.189	−0.96 (−2.72 to 0.81)	0.289
Frontal lobe	−1.04 (−1.9 to −0.17)	0.019	−0.08 (−0.74 to 0.57)	0.807	−0.02 (−0.67 to 0.64)	0.954
Parietal lobe	−0.42 (−0.82 to −0.02)	0.041	−0.1 (−0.44 to 0.23)	0.545	−0.08 (−0.41 to 0.26)	0.651
Temporal lobe	−0.84 (−1.41 to −0.28)	0.003	−0.27 (−0.73 to 0.18)	0.238	−0.23 (−0.68 to 0.23)	0.331
Hippocampus	−0.09 (−0.13 to −0.05)	<0.001	−0.08 (−0.13 to −0.04)	<0.001	−0.08 (−0.13 to −0.04)	<0.001

Model 1 was adjusted for age, sex, BMI, alcohol consumption, smoking status, and physical exercise; model 2 included covariates in model 1 and further adjusted for total cholesterol, low-density lipoprotein cholesterol, fasting blood glucose, SBP, and total intracranial volume; Model 3 included covariates in model 2 and further adjusted for use of antihypertensive, hypoglycemic, and hypolipidemic medications. CBF, cerebral blood flow; CI, confidence interval; PWV, pulse wave velocity; SD, standard deviation.

In model 1, PWV was negatively associated with volume in the total brain [β (95% CI) −8.93 (−14.14 to −3.71)] and total gray matter [β [95% CI] −3.31 [−5.9 to −0.72)] as well as frontal lobe [β (95% CI) −1.04 (−1.9 to −0.17)], parietal lobe [β (95% CI) −0.42 (−0.82 to −0.02)], temporal lobe [β (95% CI) −0.84 (−1.41 to −0.28)] and hippocampus [β (95% CI) −0.09 (−0.13 to −0.05)]. In model 2, PWV was negatively associated with volume in the total brain [β (95% CI) −5.11 (−7.92 to −2.29)] and hippocampus [β (95% CI) −0.08 [−0.13 to −0.04)]. Further adjustments in model 3 did not essentially change these associations (Table 2).

Pulse wave velocity and cognitive function

The results of the generalized linear model indicated that elevated PWV exerted a negative effect on the decreased MoCA scores (−0.28 change per SD change in PWV; 95% CI −0.52 to −0.04) after excluding the effects of potential confounding variables (Supplemental Table 2). With MoCA score less than 26 as the dependent variable, logistic regression analysis showed that the odds of developing cognitive impairment increased significantly with each 1SD increase in PWV [odds ratio 1.21 (95% CI 1.01–1.45)] (Supplemental Table 3).

Mediation analysis

Mediation analysis showed that after adjusting for potential confounding variables, decreased hippocampal volume partially mediated the negative association between PWV and MoCA scores [indirect effect: −0.040 (−0.071 to −0.009); direct effect: −0.242 (−0.479 to −0.004); proportion of mediation: 14.173%] (Fig. 3). In addition, significantly altered total and regional CBF did not mediate the association between PWV and MoCA scores.

FIGURE 3.

Mediation effect by hippocampus volume in the association between arterial stiffness and cognitive impairment. MoCA, Montreal Cognitive Assessment; PWV, pulse wave velocity.

DISCUSSION

The main finding of this prospective study was that PWV was negatively associated with CBF and brain tissue volume and was also a risk factor for cognitive impairment. Our results suggest that brain atrophy at specific sites may play a key mediating role in the increased risk of cognitive impairment caused by arterial stiffness. This association is not only independent of traditional risk factors but also shows site differences.

In contrast to previous studies that only explored total brain volume change [13,25], the present study systematically evaluated associations between arterial stiffness and brain perfusion and brain tissue volume in the total brain and regions of interest of relevance to cognitive impairment. Vascular aging can lead to blood–brain barrier breakdown, neuronal loss and microvacuolar changes in these areas of interest, rendering them more susceptible to vascular aging [21,26]. We noted a negative association between PWV and total and regional brain perfusion and brain tissue volume. Arterial stiffness may affect brain health by reducing arterial elasticity and increasing pulsating flow [27]. Increased transmission of harmful pressure pulsatility into the microcirculation leads to microcirculation remodeling [8], resulting in increased resistance, decreased CBF, and brain structural damage [28,29]. More importantly, in this study, we found that PWV was negatively associated with CBF in frontal, parietal and temporal lobes and volume in hippocampus. The hippocampus and the frontal, parietal and temporal lobes are the core brain regions responsible for cognitive function [30–32], and structural and functional changes in these brain regions may lead to cognitive impairment. One possible explanation is that arterial stiffness causes chronic hypoperfusion in the above regions, which ultimately leads to cognitive impairment [33].

In the present study, we also found that arterial stiffness was a risk factor for cognitive impairment. Our results showed that after adjusting for confounding variables, each SD increase in PWV was associated with a 21% increased risk of developing cognitive impairment. The ARIC-NCS (Atherosclerosis Risk in Communities-Neurocognitive Study) found that older participants (aged 67–90 years) with higher PWV had lower scores in general cognition (β = −0.09 z score; 95% CI −0.15 to −0.03 z score) and executive function/processing speed (β = −0.04 z score; 95% CI −0.07 to −0.01 z score) [15]. Bangen et al.[34] further demonstrated in older adults (mean age 74.9 years) that PWV was significantly negatively associated with poorer executive functioning (β = −0.09 ± 0.04, P = 0.03) and that PWV and APOE genotypes had a significant interaction on memory performance (interaction P = 0.01). The Framingham Heart Study of young and middle-aged adults (mean age, 46 years) reported the elevated PWV was associated with poorer processing speed and executive function [35]. However, these studies included age-specific participants, and the association between arterial stiffness and cognitive function in the general population remains unclear. The present study confirmed a negative association between PWV and cognitive impairment in a population aged 22.8–83.9 years in the Kailuan community, which further suggests that long-term cumulative exposure to arterial stiffness at different ages may lead to cognitive impairment. Therefore, early control or elimination of arterial stiffness will help the onset of cognitive decline and delay its progression.

The important finding of this study was that hippocampal volume partially mediated the association between PWV and MoCA score, accounting for 14.173% of the total effects. Hippocampal neurons are involved in episodic coding of events, including spatial and temporal representations [36]. Thus, the hippocampus is the core of the cognitive system and provides an objective spatial framework in which the events and items experienced by the organism are located and related to each other [36]. Hippocampal atrophy is one of the most typical anatomical and imaging manifestations of Alzheimer's disease [37]. Palta et al.[15] demonstrated that participants in the highest quartile of PWV had smaller volumes of the Alzheimer's disease signature region (−1.48 cm³; 95% CI −2.27 to −0.68 cm³), including the hippocampus. Hippocampal atrophy is associated with cognitive decline [38,39]. Although no similar previous studies have been conducted, animal and human studies have confirmed that arterial stiffness can lead to hippocampal microcirculation changes, neuroinflammation, and blood–brain barrier leakage [40,41]. These pathophysiological consequences can further contribute to loss of neuron density and function in the hippocampus, leading to cognitive deficits [42,43]. These studies suggest that elevated PWV may damage hippocampal structure and function and lead to the development of cognitive impairment, which further supports our findings.

There are some limitations to this study. First, this was a cross-sectional study, and the mechanisms of the association observed in this study remain unclear. A future longitudinal study of the association of arterial stiffness with brain perfusion, structure and cognition will help resolve this question. Second, in this study, we evaluated arterial stiffness using the brachial–ankle PWV. Brachial–ankle PWV has good agreement with cervical–femoral PWV [44,45] and has been proposed as a surrogate to cervical–femoral PWV [7]. Moreover, brachial–ankle PWV is simple in operation and is more suitable for early arterial stiffness screening in a large population [46]. A growing number of studies have used brachial–ankle PWV to assess the impact of arterial stiffness on the central nervous system [47–49]. Third, there is an association between age and PWV [50]. Therefore, we adjusted for age in each model, thereby reducing the effect of age on PWV. Fourth, there are many risk factors related to the effect of arterial stiffness on cognitive impairment, such as APOE genotype [51], enlarged perivascular space [52] and neuroinflammation [53]. In this study, we found that the mediating effect of hippocampal volume on the association between arterial stiffness and cognitive decline was only 14.173%, further suggesting the role of other risk factors in this association. In subsequent research, we plan to explore the influence of different risk factors on the effect of arterial stiffness on cognitive impairment in a large sample. Fifth, this study only used MoCA to assess cognitive function. We will follow all participants every 2–4 years and test specific cognitive domains associated with hippocampal impairment. Finally, all the participants in this study were Asian, and it remains uncertain whether these findings can be generalized to other races.

In conclusion, we found that in the general adult population, elevated PWV was associated with decreased CBF and brain tissue volume and was also a risk factor for cognitive impairment. Hippocampal atrophy plays a key mediating role in the increased risk of cognitive impairment caused by arterial stiffness. These findings not only indicate that arterial stiffness plays a key role in the cause of subclinical brain disease, which ultimately leads to cognitive impairment but also provide evidence to improve the early prevention of these pathological conditions.

ACKNOWLEDGEMENTS

Funding: This work was supported by the National Natural Science Foundation of China [Grant No.82202109, No.61931013, No.82171886], Beijing Municipal Natural Science Foundation [No.7232335], Beijing Scholars Program [No. [2015] 160].

Conflicts of interest

Conflicts of interest statement: No.