Carotid Arterial Stiffness and Risk of Incident Cerebral Microbleeds in Older People: the AGES-Reykjavik Study

Abstract

Objective

Age and high blood pressure are major risk factors for cerebral microbleeds (CMBs). However, the underlying mechanisms remain unclear and arterial stiffness may be important. We investigated whether carotid arterial stiffness is associated with incidence and location of CMBs.

Approach and Results

In the prospective, population-based AGES-Reykjavik study, 2,512 participants aged 66–93 years underwent a baseline brain MRI examination and carotid ultrasound in 2002–2006, and returned for a repeat brain MRI in 2007–2011. Common carotid arterial stiffness was assessed using a standardized protocol and expressed as carotid arterial strain (CAS), distensibility coefficient (DC) and Young’s elastic modulus (YEM). Modified poisson regression was applied to relate carotid arterial stiffness parameters to CMBs incidence. During a mean follow-up of 5.2 years, 463 people (18.4%) developed new CMBs, of whom 292 had CMBs restricted to lobar regions and 171 had CMBs in a deep or infratentorial region. After adjusting for age, sex and follow-up interval, arterial stiffness measures were associated with incident CMBs (Risk ratio [RR] per SD decrease in CAS, 1.11 [95%CI, 1.01–1.21]; per SD decrease in natural log-transformed DC, 1.14[1.05–1.24]; per SD increase in natural log-transformed YEM, 1.13[1.04–1.23]). These measures were also significantly associated with incident deep CMBs (1.18[1.02–1.37]; 1.24[1.08–1.42]; 1.23[1.07–1.42]) but not with lobar CMBs. When further adjusted for blood pressure and other baseline vascular risk factors, carotid plaque, prevalent CMBs, subcortical infarcts and white matter hyperintensities, the associations persisted.

Conclusions

Our findings support the hypothesis that localized increases in carotid arterial stiffness may contribute to the development of CMBs, especially in a deep location atttributable to hypertension.

Introduction

Cerebral microbleeds (CMBs), visualized as hypointense lesions on T2*-weighted gradient echo magnetic resonance imaging (MRI), frequently occur in older people^1,2 and are associated with an increased risk of (recurrent) stroke, cognitive impairment and dementia.^3–6 Histopathological studies show CMBs represent hemosiderin deposits from microvascular leakage and commonly relate to the two different small vessel pathologies of cerebral amyloid angiopathy and hypertensive arteriopathy. CMBs resulting from cerebral amyloid angiopathy are predominantly located in lobar regions, whereas those from hypertension are in deep hemispheric or infratentorial locations.⁷ Although advancing age and high blood pressure are established risk factors for CMBs,^1,2 the mechanisms through which these processes lead to CMBs are not fully understood.

One hypothesis concerning the underlying pathways leading to CMBs is arterial stiffening in the media in the vessel wall. The elasticity and compliance of the arterial wall decreases with advancing age and in the presence of hypertension. In particular, arterial stiffening impairs the cushioning function of large to medium-sized arteries, which increases pulsatility of blood flow that transmits excessive power distally. This power may damage small vessel walls and lead to tearing of endothelial and smooth muscle cells, particularly in high-flow organs like the brain.^8–11 Early endothelial damage to the blood-brain barrier may initiate a pathological cascade via CMBs, causing insufficient perfusion and subsequent parenchymal damage.¹²

Carotid arteries are the main conduits that supply blood to the brain and undergo a more pronounced age-related increase in arterial stiffness than peripheral muscular arteries.¹³ Local measurement of carotid arterial stiffness has been found to be associated with incident ischemic stroke.¹⁴ Recently, a cross-sectional report from the 3C-Dijon study showed that local carotid stiffness was associated with larger volume of white matter hyperintensities, which is another manifestation of cerebral small vessel disease that is primarily ischemic in origin.¹⁵

As yet, no study has explored the intriguing link between carotid ultrasound-based local arterial stiffness and bleeding-prone CMBs. We thus investigated whether local carotid arterial stiffness at baseline is associated with incident CMBs in a well-characterized large population-based cohort of older men and women. We hypothesized that carotid arterial stiffness, would be associated with the development of new CMBs. Given the spatial distributions of the underlying arteriopathies in which hypertensive arteriopathy typically affects the small perforating end-arteries of the deep structures,¹⁶ we further hypothesized that the associations would be more robust for deep CMBs attributed to hypertensive arteriopathy.

Materials and Methods

Materials and Methods are available in the online-only Supplement.

Results

The mean age of the study population at baseline was 74.6 years and 58.5% were women. In total, 18.4% (n=463) of the participants had incident CMBs, of whom 63.1% (n=292) had incident strictly lobar CMBs and 36.9% (n=171) had incident deep CMBs (Table 1). Of those who had incident CMBs located in a deep brain region, 63 people also had one or more incident lobar CMBs. Among participants who had no prevalent CMBs at baseline (n=2,082), 329 (15.8%) had completely new onset CM Bs (Supplementary Figure I). Of those with baseline and incident CMBs (n=134), 64 people had baseline and incident CMBs occurring in a strictly lobar location and 31 had both in a deep location.

Table 1.

Baseline characteristics (2002–2006) of the study population (n=2,512) according to cerebral microbleeds (CMBs) incidence category

	No incident CMBs (N=2,049)	All incident CMBs (N=463)	Strictly lobar incident CMBs (N=292)	Deep incident CMBs (N=171)
Age, mean (SD), years	74.5 (4.7)	75.3(4.9)*	75.1 (5.0)*	75.8(4.8)*
Men, %	39.9	48.7*	48.6*	48.8*
Cardiovascular risk factors
Body Mass Index, mean (SD), kg/m2	27.4(4.2)	26.8(4.0)*	26.5*	27.2
Current smokers, %	9.9	14.0*	15.1*	12.2
Current heavy alcohol drinkers, %	0.7	1.7*	1.4	2.4*
Systolic blood pressure, mean (SD), mmHg	140.9 (19.7)	142.7(19.9)	142.0(19.0)	143.9(21.3)
Diastolic blood pressure, mean (SD), mmHg	74.0(9.1)	75.2(10.0)*	74.8(9.2)	75.8(11.2)*
Pulse pressure, mean (SD), mmHg	66.9(17.4)	67.5(17.7)	67.2(17.7)	68.1(17.7)
Mean arterial pressure, mean (SD), mmHg	96.3(10.9)	97.7(11.4)*	97.2(10.4)	98.6(12.9)*
Hypertension, %	77.1	78.7	77.7	80.2
Type 2 diabetes,%	8.8	11.7	11.0	12.8
Serum total cholesterol, mean (SD), mmol/L	5.67(1.13)	5.61(1.14)	5.58(1.11)	5.66(1.18)
LDL cholesterol, mean (SD), mmol/L	3.53(1.03)	3.48(1.00)	3.46(0.99)	3.51(1.00)
Medication use
Blood pressure lowering drugs, %	60.1	60.0	57.9	63.7
Anticoagulants use, %	5.1	7.3	5.8	9.8*
Aspirin use, %	22.4	23.7	24.6	22.2
Statin use, %	24.1	22.2	23.0	20.9
Cardiovascular disease
Coronary artery disease, %	18.1	19.6	20.2	18.6
Stroke, %	4.8	5.4	5.5	5.2
Apolipoprotein E ε4 allele carrier	25.6	29.0	30.2	27.0
Carotid artery measures
Any carotid plaque, %	89.5	90.4	92.0	87.6
Mean carotid intima-media thickness, mean (SD), m/103	0.96(0.14)	0.98(0.14)*	0.97(0.14)	0.99(0.14)*
Carotid arterial strain, mean (SD), %	10.14(2.64)	9.76(2.81)*	9.82(2.97)	9.67(2.52)*
Distensibility coefficient, median (quartile range), kPa−1×10−3	22.4(17.5–28.0)	20.8(15.9–25.9)*	21.1(16.3–26.8)*	20.0(15.5–24.3)*
Young elastic modulus, median (quartile range), kPa	287.7(222.7–374.7)	315.2(234.6–414.4)	310.5(229.5–410.3)	321.8(252.9–422.4)

^*p<0.05 (compared with the no-incident CMBs group)

All carotid stiffness parameters showed significant intercorrelations, with effect sizes of correlations being moderate to high (Supplementary Table II). People with decreased distensibility coefficient (DC) (in the lowest quartile) were slightly older and more likely to be current heavy alcohol drinkers and use blood pressure lowering drugs and had higher prevalence of hypertension, diabetes, stroke and carotid plaque and higher levels of blood pressure (Table 2). In particular, the strong relationship between blood pressure and DC can be seen in the large differences in the distribution of blood pressure between quartiles of DC. The quartiles of DC are shown only for illustration of the association between DC and other baseline characteristics; the whole continuum of the values was assessed in the main analyses.

Table 2.

Baseline characteristics (2002–2006) of the study population by categories of distensibility coefficient

	Quartiles of distensibility coefficient (kPa−1×10−3)
	1st Quartile (4.1–16.6) n=616	2nd Quartile(16.6–21.1) n=616	3rd Quartile (21.1–26.3) n=616	4th Quartile (26.3–77.8) n=616	p-value
Age, mean (SD), years	76.4 (5.1)	74.8(4.6)	74.1 (4.6)	73.3(4.6)	<0.001
Men, %	43.3	43.2	41.2	37.2	0.10
Cardiovascular risk factors
Body Mass Index, mean (SD), kg/m2	26.9(4.1)	27.2(3.9)	27.3(4.2)	27.3(4.1)	0.35
Current smokers, %	9.6	6.2	10.4	15.6	<0.001
Current heavy alcohol drinkers, %	2.0	0.5	0.5	0.8	0.05
Systolic blood pressure, mean (SD), mmHg	157.3 (21.0)	144.7(15.2)	136.7(14.1)	126.0(12.5)	<0.001
Diastolic blood pressure, mean (SD), mmHg	75.9(10.5)	74.9(9.4)	74.1(8.6)	72.2(8.2)	<0.001
Mean arterial pressure, mean (SD), mmHg	103.0 (12.2)	98.2 (9.7)	95.0(9.0)	90.1(8.3)	<0.001
Pulse pressure, mean (SD), mmHg	81.4(18.1)	69.8(13.7)	62.6(12.5)	53.8(11.2)	<0.001
Hypertension, %	92.5	83.3	71.9	60.1	<0.001
Type 2 diabetes,%	13.8	9.9	6.8	5.7	<0.001
Serum total Cholesterol, mean (SD), mmol/L	5.65(1.20)	5.60(1.09)	5.69(1.10)	5.71(1.13)	0.39
LDL cholesterol, mean (SD), mmol/L	3.53(1.07)	3.48(1.00)	3.56(1.00)	3.59(1.01)	0.21
Medication use
Blood pressure lowering drugs, %	65.4	60.6	57.5	54.6	0.001
Anticoagulants use, %	7.1	6.3	5.2	3.7	0.08
Aspirin use, %	25.1	19.9	21.5	22.0	0.20
Statin use, %	24.5	24.5	21.3	22.4	0.44
Cardiovascular disease
Coronary artery disease, %	18.0	19.2	17.7	16.6	0.70
Stroke, %	7.1	4.2	5.5	2.6	0.002
Any carotid plaque, %	93.2	89.5	88.4	87.5	0.006
Mean carotid intima-media thickness, mean (SD), m/103	1.01(0.14)	0.97(0.14)	0.95(0.13)	0.94(0.13)	<0.001
Apolipoprotein E ε4 allele carrier	25.0	29.0	23.7	25.0	0.17

^*P value represents difference in characteristic by distensibility coefficient quartiles, using analysis of variance or chi square test.

After adjustment for age, sex, brain MRI scan interval and head coil, all arterial stiffness measures were individually associated with any incident CMBs (risk ratio [RR] per SD decrease in carotid arterial strain [CAS], 1.11 [95%CI, 1.01–1.21]; RR per SD decrease in natural log-transformed DC, 1.14[1.05–1.24]; RR per SD increase in natural log-transformed Young’s elastic modulus [YEM], 1.13[1.04–1.23]) (Table 3). When stratified according to CMBs location, the associations were strongest for incident deep CMBs (RR per SD decrease in CAS, 1.18[1.02–1.37]; RR per SD decrease in natural log-transformed DC, 1.24[1.08–1.42]; RR per SD increase in natural log-transformed YEM, 1.23[1.07–1.42]), whereas there were no significant associations with strictly lobar CMBs. The associations with any or deep CMBs remained essentially unchanged after additional adjustment for blood pressure and other major vascular risk factors, presence of carotid plaque and prevalent CMBs. When further adjusted for subcortical infarcts and white matter hyperintensities, these associations persisted. There were no significant interactions of carotid arterial stiffness measures with any of the covariates.

Table 3.

Associations of carotid arterial stiffness parameters with incident cerebral microbleeds (CMBs) according to location

	Risk Ratios for incident CMBs (95% CI)
	Any incident CMBs (n=463)	Strictly lobar incident CMBs (n=292)	Deep incident CMBs (n=171)
Model 1: age, sex, brain MRI scan interval & coil type
Carotid arterial strain, per SD (2.7%) decrease	1.11(1.01–1.21)	1.09(0.96–1.23)	1.18(1.02–1.37)
Distensibility coefficient, per SD (0.36) decrease in natural log-transformed	1.14(1.05–1.24)	1.11(0.99–1.24)	1.24(1.08–1.42)
Young’s elastic modulus, per SD (0.41) increase in natural log-transformed	1.13(1.04–1.23)	1.10(0.98–1.23)	1.23(1.07–1.42)
Model 2: model 1 plus vascular risk factors, prevalent carotid plaque & CMBsa
Carotid arterial strain, per SD (2.7%) decrease	1.11(1.02–1.21)	1.10(0.98–1.24)	1.17(1.01–1.36)
Distensibility coefficient, per SD (0.36) decrease in natural log-transformed	1.18(1.06–1.30)	1.15(1.01–1.32)	1.28(1.09–1.49)
Young’s elastic modulus, per SD (0.41) increase in natural log-transformed	1.15(1.04–1.26)	1.12(0.99–1.27)	1.24(1.06–1.44)
Model 3: model 2 plus subcortical infarcts & white matter hyperintensitiesb
Carotid arterial strain, per SD (2.7%) decrease	1.10(1.01–1.20)	1.08(0.96–1.21)	1.17(1.01–1.36)
Distensibility coefficient, per SD (0.36) decrease in natural log-transformed	1.16(1.05–1.28)	1.13(0.99–1.29)	1.25(1.07–1.46)
Young’s elastic modulus, per SD (0.41) increase in natural log-transformed	1.13(1.03–1.24)	1.11(0.98–1.25)	1.22(1.05–1.42)

^aModel 2 was adjusted for age, sex, brain MRI scan interval, head coil, systolic blood pressure, use of blood pressure lowering drugs, body mass index, current smoking, total cholesterol, statin use, use of anticoagulants/aspirin, type 2 diabetes, the presence of carotid plaque & prevalent CMB at baseline;

^bModel 3 was adjusted for age, sex, brain MRI scan interval, head coil, systolic blood pressure, use of blood pressure lowering drugs, body mass index, current smoking, total cholesterol, statin use, use of anticoagulants/aspirin, type 2 diabetes, the presence of carotid plaque, prevalent CMBs at baseline, subcortical infarct and relative measure of white matter hyperintensities (% intracranial volume); Numbers in bold indicate significant p-values at 0.05 level.

Sensitivity analyses

When we analyzed people with ‘strictly’ deep CMBs (n=108) excluding those with CMBs with a mixed location profile, associations were moderately strengthened as compared to those for mixed deep and lobar CMBs (RR per SD decrease in CAS, 1.29[1.08–1.54]; RR per SD decrease in natural log-transformed DC, 1.33[1.10–1.62]; RR per SD increase in natural log-transformed YEM, 1.29[1.07–1.56]) in the fully adjusted models (model 3). The results were qualitatively similar when adjusting for diastolic blood pressure or pulse pressure or mean arterial pressure instead of systolic blood pressure in the models (data not shown). Additional adjustment for prevalent stroke or alcohol intake or eGFR/CKD or Apolipoprotein E ε4 genotype in model 3 did not change the associations substantively (data not shown). In location-specific analyses, exclusion of those with discordant locations between baseline and incident CMBs did not essentially change the findings for incident strictly lobar (i.e. people with baseline deep and incident lobar CMBs were excluded [n=14]) or deep (i.e. people with baseline lobar and incident deep CMBs were excluded [n=25]) CMBs (data not shown). In stratified analyses, most associations persisted in participants without baseline CMBs (n=2,082) and were found to be in the same direction, though not significant, in the smaller sample of people with baseline CMBs (n=430) (Supplementary Tables V&VI). We also repeated the analyses for subgroups stratified by APOE ε4 carriership (data not shown) and formal testing of interactions by APOE ε4 carriership did not reach statistical significance (all p>0.05). Analyses with imputed datasets yielded results similar to those reported in the primary analyses.

Discussion

In this prospective population-based cohort study of men and women older than 65 years who had no dementia at baseline, we found local measures of carotid arterial stiffness were associated with an increased risk of incident CMBs. The associations were restricted to CMBs in the deep or infratentorial brain regions. These associations with CMBs were consistent across the different stiffness indices studied and also persisted after additional adjustment for cardiovascular risk factors including blood pressure, the presence of carotid plaque and ischemic cerebral small vessel disease, suggesting that they were not simply due to confounding by carotid atherosclerosis or other vascular mechanisms.

To the best of our knowledge, the present study is the first to examine the association between ultrasound-based local carotid stiffness indices and incidence of CMBs. We are aware of several previous studies that examined the association of brachial-ankle or carotid-femoral pulse wave velocity with the presence of CMBs.^10,17,18 Those results based on indirect measures of regional arterial stiffness over a certain arterial length, are consistent with our results. More recently, local carotid stiffness measures have been reported to be associated with larger volume of white matter hyperintensities.¹⁵ Our results thus lend further support to an association of carotid arterial stiffness with cerebral small vessel disease, and demonstrate the association with incident CMBs, which are pathologically rather specific to leakage and hemorrhage from damaged arteriolar walls and may reflect a crucial stage in the pathogenesis of cerebral small vessel disease.¹²

Although the underlying mechamisns of the observed associations between carotid arterial stiffness and CMBs have not been fully elucidated, there are several explanations. The brain is a high flow organ and particularly sensitive to excessive pressure and flow pulsatility.⁹ Arterial stiffening may prevent the cushioning of flow pulsations created by left ventricular ejection. In this scenario, high local blood flow in the brain is associated with low microvascular impedance, which facilitates transmission of excessive pulsatile energy into the cerebral microcirculation.⁹ Increased pulsatile stress may lead to damage in the endothelium and smooth muscle cells disrupting cerebral small vessels.¹¹ A stiffened artery may also increase pulse-wave velocity, which potentially causes wave reflections to appear earlier in the cardiac cycle and may further augment central systolic blood pressure.^19,20 However, this hypothesis has recently been disputed.²¹ As carotid artery is a predominantly elastic artery, like the aorta, stiffness of the carotid may represent a surrogate for aortic stiffness. In the subsample (n=506) of our study population who had available carotid-femoral pulse wave velocity measurements (a global estimate of arterial pulse wave velocity through the entire aorta that reflects aortic stiffness), we found moderate correlations between carotid stiffness indices and carotid-femoral pulse wave velocity (correlation coefficients for YEM=0.32, DC=−0.32; both p<0.001). However, it should be noted that the measure of carotid-femoral pulse wave velocity reflects the properties of a mixed elastic and muscular part of the arterial tree²² and the ascending aorta, which is a prime location of aortic stiffening, is not directly accounted for.²³ Alternatively, carotid stiffness is a proxy for middle-size cerebral artery stiffness and increased aortic pulsatility is transmitted through stiff large vessels (middle cerebral arteries) to the cerebral microvasculature.²⁴

Our associations were restricted to CMBs located in the deep regions, suggesting the role of hypertensive vasculopathy. CMBs occur in deep regions that are supplied by directly penetrating branches of the deep cerebral circulation. As described earlier by Mitchell,⁹ these vessels differ significantly from the long, circuitous pial network that supplies the cortical lobes and may have an additional level of protection from excessive central pressure and flow pulsatility.⁹

Major strengths of the present study include the large population-based sample of old individuals, the use of standard MRI protocol at both time points, the use of well-validated image processing tool to assess different parameters of carotid arterial stiffness, and accounting for all necessary covariates including ischemic cerebral small vessel disease in the statistical models. A possible limitation of the study is that the selection bias may have influenced the results. People who were included in the analysis were younger and healthier at baseline than those who were excluded (Supplementary Table I). In particular, people with worse vascular risk profile or more severe cerebral small vessel disease (those more likely to develop new CMBs) died or were lost to follow-up before they could be recruited into the follow-up examination. This may have led us to underestimate the true incidence of CMBs and as such the findings in relation to carotid measures may be affected. For instance, the excluded people had increased carotid arterial stiffness and bias would occur if the association between carotid arterial stiffness and CMBs in the excluded people differed from what we found in the included sample. The study also has the limitations of observational research and there remains scope in these estimates for residual bias due to unmeasured or imprecisely measured confounding factors (e.g. repeated BP measurements over time instead of BP assessment on a single occasion). Further on the exposure side, given that the effect of a single baseline (carotid stiffness) measurement on future events is subject to regression dilution, we may have underestimated the true associations with CMBs incidence. Individuals with hypertension or other atherosclerotic risk factors were more likely to have stiffer arteries and would likely be put on blood pressure lowering therapies through the course of the study. The use of blood pressure-lowering therapies may reduce arterial stiffness²⁵ and thus bias any associations between arterial stiffness and incident CMBs toward the null. DC and YEM were calculated from peripheral brachial rather than central carotid pulse pressure, which would provide more accurate stiffness values. However, because central and peripheral pulse pressures converge with aging and the brachial pulse pressure has been shown as a good reflection of carotid pulse pressure in older individuals,^26,27 this is unlikely to have affected our findings greatly. If prevalent and incident CMBs were located in different locations (deep vs. lobar), then it may be more difficult to identify location-specific risk factors. In our sensitivity analyses, we excluded people whose baseline and follow-up CMBs differed in location; results were essentially unaltered for incident strictly lobar or deep CMBs, suggesting this location difference is unlikely to have a significant impact on our findings. Our study did not set out to examine the contribution of hemodynamic alterations associated with carotid arterial stiffness to CMBs development. Further studies are needed to evaluate whether hemodynamic manifestations of carotid arterial stiffness such as carotid pulsatile flow load²⁸ are related to CMBs.

The effect size for carotid arterial stiffness in these older individuals was small. Such a small effect size for risk factors associated with incident CMBs is not uncommon, including the effects of other potentially important factors (e.g., systolic blood pressure).^29,30 It is possible that even very modest effects of a risk factor may lead to increased size or number of CMBs and possibly symptomatic intracerebral haemorrhage and cognitive impairment with further aging. However, while our findings support the hypothesis that reduction of carotid arterial stiffness might reduce the rate of CMBs, the clinical significance of our findings remains to be established.

In conclusion, independent of carotid plaque and ischemic cerebral small vessel disease, localized increases in carotid arterial stiffness were associated with an increased risk of CMBs. Our findings support the hypothesis that carotid arterial stiffness may contribute to the pathophysiology of CMBs, especially in a deep location. If the above findings are substantiated, CMBs may be amenable to lifestyle or pharmacological interventions targeted specifically at reducing carotid stiffness of elastic arteries in addition to antihypertensive therapies in an aging population.

Figure 1.

Study population.

Significance.

Advancing age and high blood pressure are established risk factors for cerebral microbleeds; however, the mechanisms through which these processes affect cerebral microbleeds are not fully understood and arterial stiffening in the media has been hypothesized to play a major role. This is the first large scale, population-based, prospective study to demonstrate that increased carotid arterial stiffness is independently associated with an increased risk of developing new cerebral microbleeds over approximately 5 years of follow-up. Our findings support the hypothesis that carotid arterial stiffness may contribute to the pathophysiology of cerebral microbleeds, especially those occurring in a deep location attributable to hypertensive vasculopathy, and underscore the importance of lifestyle or pharmacological interventions targeted specifically at improving carotid stiffness of elastic arteries to reduce future cerebral microbleeds risk in an aging population.

Abbreviations

APOE

Apolipoprotein E

CAS

carotid arterial strain

CMBs

cerebral microbleeds

DBP

diastolic blood pressure

DC

distensibility coefficient

IMT

intima-media thickness

MRI

magnetic resonance imaging

PP

pulse pressure

RR

risk ratio

SBP

systolic blood pressure

YEM

Young’s elastic modulus