Intima-Media Thickness and Regional Cerebral Blood Flow in Older Adults

Abstract

Background and Purpose

The relationship between the thickness of the carotid intima (IMT) and brain function remains unclear in those without clinical manifestations of cerebrovascular disease. Understanding the neural correlates of this vascular measure is important in view of emerging evidence linking poorer cognitive performance with increased IMT in individuals without clinical cerebrovascular disease.

Methods

73 participants in the Baltimore Longitudinal Study of Aging (70.9(SD 7.3) years) were evaluated with carotid artery ultrasound and resting [¹⁵O]H₂O-PET.

Results

After adjusting for age, sex, and grey and white matter volumes in the regions where IMT is related to rCBF, we found that higher IMT was associated with lower rCBF in lingual, inferior occipital, and superior temporal regions. Higher IMT was also associated with higher rCBF in medial frontal gyrus, putamen, and hippocampal-uncal regions (p=0.001). While women had lower IMT (p=0.01) and mean arterial pressure (MAP;p=0.05) than men, they showed more robust associations between IMT and rCBF. The relationship between IMT and rCBF was only minimally affected by additional adjustment for MAP.

Conclusions

IMT is related to patterns of resting rCBF in older adults without clinical manifestations of cerebrovascular disease, suggesting that there are regional differences in CBF that are associated with subclinical vascular disease.

Introduction

Increased carotid intima-media thickness (IMT) is a marker of accelerated arterial aging¹. As many of the factors influencing arterial wall thickness are also implicated in the pathogenesis of atherosclerosis, it is not surprising that increased IMT is not only a risk factor for stroke³ but is also associated with MRI-defined cerebral infarcts and white matter disease, as well as sulcal and ventricular widening^4,5. Relatively little is known, however, about the relationship between IMT and resting regional cerebral blood flow (rCBF) in older adults without clinical manifestations of cerebrovascular disease.

It has been shown that IMT is an independent predictor of reduced cognitive speed and poorer performance on tests of verbal and nonverbal memory, semantic fluency, and executive function even in individuals without clinical manifestations of cerebrovascular disease^2,6,7. These findings suggest that accelerated arterial aging is associated with global alterations in brain function. Because rCBF is a marker of brain function and IMT is a modifier of rCBF changes that occur as individuals age, we hypothesized that rCBF and IMT may be directly related in older individuals even in the absence of cerebrovascular disease symptoms. Given that preventative measures and treatment may decrease or even arrest progression of atherosclerosis at early stages, understanding of accelerated aging and its cerebral correlates is important.

In the present study, we examined the cross-sectional relationship of IMT and rCBF in 73 older adults without overt cerebrovascular disease from the neuroimaging study of the Baltimore Longitudinal Study of Aging (NI-BLSA)⁸. We hypothesized that rCBF patterns would differ in individuals with higher IMT compared with lower IMT even in the absence of clinically diagnosed cerebrovascular disease. Given that there are differences between men and women in both IMT⁹ and rCBF¹⁰, sex differences in the relationships between IMT and rCBF were examined. We also evaluated the effects of mean arterial pressure(MAP) on the relationship between IMT and rCBF as pathophysiological circulatory changes affecting arteriolar tone might be related to the association between IMT and rCBF. Finally, to better characterize the degree of vascular disease in this sample, we quantified the white matter lesion (WML) load and examined how it relates to IMT.

Materials and Methods

Study Participants

73 nondemented participants from the NI-BLSA who underwent resting [¹⁵O]H₂O-PET and carotid ultrasound during the same visit were included in the current analyses. Structural MRI was acquired concurrently with PET in all but three individuals who were unable to tolerate MRI at the time of the PET study. For these individuals, MRI obtained 1.3(SD 0.6) years prior to PET imaging was used.

Participant demographic, cognitive and medical history data are shown in Table 1. NI-BLSA initially enrolled individuals with no history of central nervous system disease [epilepsy, stroke, bipolar illness], severe cardiac disease [myocardial infarction, coronary artery disease requiring angioplasty or bypass surgery], or diagnosis of dementia⁸. In this investigation, only participants without significant carotid artery disease [i.e.those who had not undergone carotid endarterectomy] were included. In addition, participants with dementia or cognitive impairment at the time of imaging were excluded from analyses. Cognitive status was determined by consensus diagnosis according to established procedures^11,12. Institutional IRB approval was obtained for the study, and written informed consent was obtained from each participant.

TABLE 1.

Participant Characteristics

Demographic Information	All N=73	Males N=42	Females N=31	Statistical test
Age	70.9 (7.3)	71.7 (7.3)	70 (7.3)	ns
Race	62 Caucasian, 11 African American	37 Caucasian, 5 African American	25 Caucasian, 6 African American	ns
Education (years)	17.0 (2.4)	17 (2.6)	16.8 (2.2)	ns
Cognitive Status
MMSE	28.9 (1.5)	28.9 (1.7)	29 (1.2)	ns
Clinical History
Cardiovascular Disease	6 (8.2%)	3 (7.1%)	3 (9.7%)	ns
Anti-hypertensive therapy	32 (43.8%)	23 (54.8%)	9 (29 %)	0.03
Statin therapy	6 (8.2%)	4 (9.5%)	2 (2.7 %)	ns
Vascular Data
SBP (mm Hg)	123.0 (14.8)	125.1 (15.3)	120 (13.8)	ns
DBP (mm Hg)	66.3 (8.5)	70.8 (8.6)	66.9 (7.9)	0.04
MAP (mm Hg)	85.2 (9.7)	87.1(9.9)	82.7(9.0)	0.05
White matter lesion volume#	1348 (2308.5)	1673.1(2902.4)	866.3 (699.2)	ns
White matter lesion load#	0.1 (0.17)	0.12 (0.22)	0.07 (0.06)	ns
IMT (mm)	0.6 (0.2)	0.7 (0.2)	0.6 (0.2)	0.01

^#White matter lesion load is white matter lesion volume (WMV) adjusted for intracranial volume (ICV; (WMV/ICV)*100). WM volume data for 6 participants are not available. Continuous data reported as mean (standard deviation). Group differences evaluated by Student’s T-test (continuous variables) and Fisher’s exact test (categorical variables).

SBP: Systolic blood pressure; DBP: Diastolic blood pressure; MAP: Mean arterial blood pressure; IMT: Intima-media thickness

PET Scanning Parameters and Conditions

[¹⁵O]H₂O scans were performed on a GE 4096+ scanner (15 slices, in-plane resolution of 6.5 mm FWHM, 60 second acquisition). During rest, participants were instructed to focus on a screen covered with black cloth. Attenuation correction using 2D mode transmission scan (Ge-68 rotating source) was performed.

Carotid Ultrasonography

High resolution B-mode carotid ultrasound was obtained using a linear array, 5–10-Mhz transducer (Ultramark 9 HDI, Advanced Technology Larboratories, Inc., Seattle, Washington)^13,14. Evaluation was performed in the supine position in a dark, quiet room. The IMT was measured on a frozen frame of the region 1.5 cm proximal to the carotid bifurcation after the left common carotid artery was maximized in the longitudinal plane. The IMT measurement was obtained by averaging the distance between the lumen-intima interface and the media-adventitia interface obtained from five contiguous sites 1mm apart. Blood pressure measurements (Critikon1846SX/P, version 085, Dinamap, Critikon, Tampa, Florida) were obtained in a supine position 15 minutes after the onset of testing.

SPM Analysis of PET Scans

Using Statistical Parametric Mapping (SPM2; Wellcome Department of Imaging Neuroscience, London, England), [¹⁵O]H₂O-PET scans were realigned, spatially normalized and smoothed to FWHM of 12 mm. To control for variability in global flow, rCBF values at each voxel were ratio-adjusted to the mean global flow and then multiplied by 50 to scale the data to the range of experimentally derived mean CBF values of 50 ml/100g/min. Using a multiple regression model, the relationship between resting rCBF and IMT was assessed on a voxel-by-voxel basis, adjusting for age and sex. Separate contrasts were used to determine linear associations between IMT and higher rCBF and lower rCBF, respectively. To examine the effects of antihypertensive medications, this analysis was repeated with antihypertensive medication as an additional covariate. The effect of sex on the association between IMT and rCBF was also examined across all participants in the multiple regression model using a sex x IMT interaction. In view of IMT sex differences in our sample (Table 1) and effect of sex on the IMT-rCBF relationship in the multiple regression model, separate age-adjusted regression analyses of IMT and rCBF also were performed in males and females. These analyses were then repeated to include additional adjustments for MAP and for the (grey+white matter) brain volumes of regions showing IMT-rCBF associations in the initial analyses. Significant effects for all analyses were based on the peak magnitude (p≤0.001) with a spatial extent of ≥100 voxels. We chose a relatively large spatial extent threshold to limit spurious correlations as we hypothesized that IMT and rCBF will be related in relatively large brain regions.

MRI

Spoiled gradient recalled MRI (124 slices, matrix 256×256, pixel size 0.93×0.93mm, slice thickness 1.5 mm), proton density (PD), and T2 weighted images (TR=3000, TE=34/100, FOV=24 cm, matrix=256×192, NEX=0.5, slice thickness 5mm) were obtained on a 1.5 Tesla GE Signa system.

Brain Volumes for Significant Clusters of Interest

After segmentation into gray matter, white matter and cerebrospinal fluid, the images were spatially normalized using a high-dimensional elastic warping method and a volume-preserving transformation¹⁵. Masks of regions showing significant relationships between IMT and rCBF from the primary SPM2 regression analyses were then used to extract the total grey and white tissue volumes for each region.

White Matter Lesion Quantitation

A computer-assisted WML segmentation method, based on local features extracted from T1-weighted, T2-weighted and PD sequences, was used for volumetric assessment of WML volume using a support vector machine classifier¹⁶.

Results

Carotid Ultrasound Findings

Overall, the mean IMT was 0.62(SD 0.18)mm. IMT was higher with increasing age: for every decade, IMT increased by 0.08(SE 0.03)mm, adjusting for sex. Females had significantly lower mean IMT than males (p=0.01) (Table 1, Figure 1).

FIGURE 1. Sex Differences in IMT.

Vascular Data

Of these 73 individuals, only 6 had SBP>140 mmHg and none had DBP>90 mmHg (Table 1). 43.8 % of the participants were on antihypertensive medications at the time of the study. While 29% of women as compared to 55% of men were on antihypertensive medications (χ²=4.8, p=0.03), the mean MAP was lower in females than in males (p=0.05) (Table 1). MAP did not correlate with IMT (r=0.06, p=0.6). Additionally, IMT did not correlate with WML burden (r=0.06, p=0.6) after removal of a single outlier.

Resting rCBF and IMT

In the group as a whole (Table 2, Figure 2), higher IMT was associated with lower rCBF in the left lingual gyrus(BA19), right inferior occipital gyrus(BA19), and right superior temporal gyrus(BA38). In addition, higher IMT was associated with greater rCBF in the right medial frontal gyrus(BA9) extending bilaterally and inferiorly to the left inferior frontal gyrus, and also in the right putamen and left hippocampal-uncal region. The additional adjustment for antihypertensive medications in the overall analysis did not alter the pattern of results.

TABLE 2.

Maxima of regions showing significant correlation between IMT and rCBF

		Primary Correlation Analysis Adjusted for age and sex						Secondary Correlation Analysis Adjusted for age, sex, and MAP
Cluster Maxima*†	Side	Coordinates			T-value	p value	Spatial Extent	Coordinates			T-value	p value	Spatial Extent
		x	y	z			(# voxels)	x	y	z			(# voxels)
↑ IMT ↓ rCBF
Lingual gyrus (BA19)	L	−26	−64	−6	3.7	<0.001	476	−26	−64	−6	3.64	<0.001	460
Inferior occipital gyrus (BA19)	R	42	−86	0	3.73	<0.001	419	42	−86	0	3.57	<0.001	351
Superior temporal gyrus (BA38)	R	46	18	−20	3.43	0.001	185	46	18	−22	3.66	<0.001	239
↑ IMT ↑ rCBF
Medial frontal gyrus (BA9)	L	−8	46	28	4.07	<0.001	2126	−6	46	30	4.03	<0.001	2547
Putamen	R	32	−14	0	4.89	<0.001	328	32	−14	0	4.86	<0.001	308
Uncus	L	−34	−20	−34	3.68	<0.001	264	−34	−20	−34	3.45	<0.001	201

^*Brodmann Areas noted in parenthesis.

^†All analyses are adjusted for the combined grey and white matter volume of a given region in which IMT is related to rCBF.

FIGURE 2. Neural Correlates of Higher IMT.

A. The overall relationship between IMT and rCBF, adjusted for age, sex, and brain volume. Regions with lower(blue) and higher(orange) resting rCBF with greater IMT are overlaid on MRI slices. B. After additional adjustment for MAP, only minimal changes in the spatial extent are seen.

Sex-Differences in Associations between IMT and rCBF

Across all participants, the multiple regression model revealed significant effects of sex on the relationship between IMT and rCBF in a number of large regions such as the right middle/inferior occipital gyri (BA18/19), the left middle/inferior (BA19/37) temporal gyri, bilateral anterior cingulate (BA32) and medial frontal (BA10) regions, the left striatum, as well as the left superior frontal gyrus (BA10). To define this relationship further, we investigated the relationship between IMT and rCBF separately by gender. In males, higher IMT was associated with lower rCBF in right superior temporal gyrus (BA38) and greater rCBF in the right putamen and right inferior frontal gyrus (BA47) (Table 3, Figure 3). In women, higher IMT was associated with lower rCBF in the right middle temporal gyrus (BA19), left middle occipital gyrus (BA19), and right inferior parietal lobule (BA40). Higher IMT was also related to greater rCBF in the medial frontal gyri bilaterally, the right insular region, bilateral putamen, and left superior temporal gyrus. Although women comprise only 43% of the participants, the relationship between rCBF and IMT in women more closely resembles the overall pattern in the group as a whole.

TABLE 3.

Maxima of regions showing significant age-adjusted relationship between IMT and rCBF in men and women

		Primary Correlation Analysis Adjusted for age						Primary Correlation Analysis Adjusted for age and MAP
Cluster Maxima*†	Side	Coordinates			T- value	p- value	Spatial Extent	Coordinates			T- value	p value	Spatial Extent
		x	y	z			(# voxels)	x	y	z			(# voxels)
Males: ↑ IMT ↓ rCBF
Superior temporal gyrus (BA 38)	R	46	18	−20	3.37	0.001	143	48	18	−22	3.7	<0.001	179
Males: ↑ IMT ↑ rCBF
Putamen	R	32	−14	0	4.06	<0.001	140	32	−14	0	4.02	<0.001	141
Inferior frontal gyrus (BA 47)	R	30	26	−16	3.51	0.001	112	30	26	−18	3.87	<0.001	138
Females: ↑ IMT ↓ rCBF
Middle temporal gyrus (BA 19)	R	34	−64	18	7.36	<0.001	4671	34	−64	18	5.77	<0.001	1297
Middle occipital gyrus (BA 19)	L	−38	−62	−2	4.92	<0.001	1509	−38	−62	−2	3.41	0.001	100
Inferior parietal lobule (BA 40)	R	64	−32	40	3.94	<0.001	369	-	-	-	-	-	-
Fusiform gyrus (BA 37)	L	−46	−48	−12	3.6	0.001	1534	−46	−48	−12	3.83	<0.001	112
Fusiform gyrus (BA 21)	L	-	-	-	-	-	-	−42	−8	−26	4.24	<0.001	159
Middle temporal gyrus (BA 21)	R	-	-	-	-	-	-	42	0	−16	3.64	0.001	237
Females: ↑ IMT ↑ rCBF
Medial frontal gyrus (BA 10)	L	−6	54	−6	5.01	<0.001	7325	−24	40	38	5.36	<0.001	6083
Superior temporal gyrus (BA 42)	L	−48	−28	18	4.43	<0.001	280	−48	−30	18	5.51	<0.001	807
Insula	R	36	−8	0	4.31	<0.001	489	30	0	8	3.37	0.001	157
Middle frontal gyrus (BA 8)	R	-	-	-	-	-	-	26	38	46	4.51	<0.001	148
Medial frontal gyrus (BA 6)	L	-	-	-	-	-	-	−6	14	44	3.51	0.001	165

^*Brodmann Areas noted in parenthesis.

^†All analyses are adjusted for the combined grey and white matter volume of a given region in which IMT is related to rCBF.

FIGURE 3. Sex Differences in the Neural Correlates of Higher IMT.

A. Regions showing lower(blue) and higher(orange) resting rCBF with higher IMT in females and males separately overlaid on MRI slices. B. After additional adjustment for MAP, changes in spatial extent of neural correlates of higher IMT are most pronounced in females.

Effect of MAP on the relationship between IMT and rCBF

The additional adjustment for MAP in the overall analysis resulted in only minimal changes in the correlation patterns, primarily in the spatial extent of the significant clusters (Table 2, Figure 2). When MAP was added to the models evaluating males and females separately, the effect on spatial extent was more pronounced in females than in males with decreases in spatial extent occurring primarily in the occipitotemporal regions (Table 3, Figure 3).

Discussion

This study investigated whether IMT, a marker of accelerated vascular aging, is associated with differential spatial patterns of rCBF in older adults without clinical manifestations of cerebrovascular disease. We have shown that (1) IMT is related to resting rCBF patterns, (2) greater IMT is associated with lower rCBF in occipitotemporal regions, and with higher rCBF in the frontotemporal regions, (3) there are sex differences in relationship between IMT and rCBF, (4) in individuals with well-controlled blood pressure, MAP only minimally affects the relationship between IMT and rCBF, and (5) WML burden is not associated with IMT in this sample of individuals.

We found that higher IMT is related to lower rCBF in the lingual, inferior occipital, and superior temporal gyri. These findings suggest that accelerated vascular aging may be related to rCBF decreases in occipitotemporal regions. Based on the involvement of these areas in memory function^17,18, decreased rCBF in occipitotemporal regions may be related to previous reports of a relationship between cognitive function and IMT². Indeed, a recent study of BLSA participants which included the subset of NI-BLSA participants in our report, found that higher IMT was associated with declining verbal and nonverbal memory performance over time⁷ supporting this relationship.

We also found that higher IMT was associated with higher rCBF in the medial frontal gyri, putamen and hippocampal-uncal regions. This anterior medial temporal finding complements that of another recent functional MRI study where IMT was positively associated with amygdalar activation and connectivity¹⁹. Whereas the association of IMT with lower rCBF in the posterior occipitotemporal region suggests that these areas may be negatively impacted during accelerated vascular aging, relative increases in rCBF in frontal and medial temporal regions associated with higher IMT may represent an attempt to preserve function. The findings of both increased and decreased rCBF in relation to IMT support functional compensation theories^20,21 and may play a role in the posterior-anterior shift in age-related activation patterns characterized by decreases in occipital activation and increases in prefrontal cortex activation²². As the spatial distribution of the rCBF correlates is very similar to regional differences in vasodilatory capacity during hypercapnia in older adults²³, this distribution may reflect regional differences in neuronal vasoresponsivity through nitric oxide(NO), and possibly in angiotensin system expression with subclinical vascular disease.

We further investigated sex differences in the relationship between IMT and rCBF. Although women had lower IMT, they exhibited a robust relationship between IMT and rCBF. Furthermore, the spatial distribution of rCBF correlates with IMT in women was very similar to the pattern seen in the entire group, suggesting that women significantly contributed to the overall relationship between IMT and rCBF. In women, higher IMT correlates with lower rCBF in temporal, occipital and parietal regions, and the correlates of higher rCBF with higher IMT are seen in medial frontal, superior temporal, insular regions and in bilateral putamen. These findings are in contrast with regional correlates in males, where higher IMT correlates with lower rCBF only in the superior temporal gyrus and with greater rCBF in the putamen, and right inferior frontal gyrus. The differences in location and spatial extent of the rCBF correlates for men and women may reflect sex differences in NO and angiotensin system expression^24–28 in the brain and at the common carotid artery^29,30. Differential susceptibility to angiotensin induced attenuation in rCBF by neural activity³¹ and differences in sex hormone levels that may affect other endothelial factors such as prostanoids and endothelium-derived hyperpolarizing factor²⁷ may further contribute to the sex differences observed here, as many of these factors are related not only to rCBF but also to IMT.

As increased blood pressure can modify cerebral autoregulation, we also evaluated whether MAP significantly affects the relationship between IMT and rCBF. In this study, additional adjustment for MAP only minimally affected the relationship between rCBF and IMT, primarily influencing the spatial extent of the regional correlations. In females, MAP adjustment had a more pronounced impact on the relationship between IMT and rCBF, with decreased spatial extent primarily in the occipitotemporal region. These findings are of interest as women had, on average, lower MAP and fewer women in this study were on antihypertensive medications. Overall, however, in this group of older adults, we found that MAP adjustment has a limited effect on the relationship between IMT and CBF when blood pressure is well controlled. This attenuated effect suggests that, at least in individuals with well-controlled blood pressure, the relationship between carotid IMT and neuronal function is not greatly affected by levels of distending blood pressure.

Finally, the relationship between IMT and WML burden was examined. In a subset of our sample for whom WML data were available, we did not observe associations between IMT and WML. Although cross-sectional⁴ and longitudinal studies³² have reported strong associations between IMT and WML, recent findings suggest that IMT measured at the internal carotid artery³³ and presence of arterial plaque in addition to increased IMT³⁴ may be stronger predictors of WML load and silent cerebral infarcts than IMT alone. This, in conjunction with lower WML load in our sample than in studies of community dwelling elderly⁸, may account for the lack of relationship between IMT and WML load.

A limitation of our study is that participants are not representative of the general population with respect to their vascular health. Individuals who had histories of myocardial infarction, bypass surgery or angioplasty were not accepted into the NI-BLSA at the time of initial enrollment. In addition, 44% of the study participants were on antihypertensive therapy and nearly 10% on statin treatment, medication classes shown to decrease IMT³⁵. Finally, we used a statistical threshold of p=0.001 with spatial extent of 100 voxels as compromise between Type 1 and Type 2 errors due to the limited power afforded by only a single resting PET scan per individual in these analyses. Although our results require replication in other samples, investigation of associations between carotid IMT and rCBF is important for our understanding of the cumulative effects of subclinical vascular disease in the brain, as these studies provide a basis for investigating the CNS effects of therapies targeting markers of subclinical atherosclerosis.