Hypertension and Cerebral Vasoreactivity: A Continuous Arterial Spin Labeling MRI study

Abstract

Hypertension is associated with microvascular and macrovascular brain injury but its direct influence on the cerebral circulation is not fully clear. Our objective was to investigate the association of hypertension with global and regional cerebral vasoreactvity to CO₂ using continuous arterial spin labeling magnetic resonance imaging, independent of stroke and white matter hyperintensities. Participants (n=62; mean age 66.7±1.0 years, 55% women, 84% white, 65% hypertension, 47% stroke) underwent arterial spin labeling perfusion MRI during normal breathing, 5% CO₂ rebreathing, and hyperventilation as well as 24-hour ambulatory blood pressure monitoring. Vasoreactivity was the slope of the regression between cerebral perfusion and end-tidal CO₂. White matter hyperintensity volumes were quantified. Nighttime dipping was calculated as the percent decline in nighttime/daytime blood pressure. After accounting for stroke and white matter hyperintensity volume, hypertensive participants had lower global vasoreactivity (1.11±0.13 vs. 0.43±0.1 ml/100gm/min/mmHg, p= 0.0012). Regionally, this was significant in the frontal, temporal and parietal lobes. Higher mean systolic blood pressure was associated with lower vasoreactivity (decreased by 0.11 units/10 mm Hg increase in systolic blood pressure, p=0.04), but nighttime dipping was not (p=0.2). The magnitude of decrease in vasoreactivity in hypertension without stroke was comparable to the magnitude of decrease in vasoreactivity in stroke without hypertension. Hypertension has a direct negative effect on the cerebrovascular circulation independent of white matter hyperintensities and stroke that is comparable to that seen with stroke. Since lower vasoreactivity is associated with poor outcomes, studies of the impact of antihypertensive on vasoreactivity are important.

Introduction

Decline in cerebrovascular reactivity is associated with cognitive decline,^1–4 slower gait speed and possibly falls. ⁵ Prior studies suggest that hypertension may have an influence on cerebrovascular reactivity.⁶ Most of these studies have assessed cerebrovascular reactivity in hypertension by measuring changes in cerebral blood flow at the middle cerebral artery in response to changes to end-tidal carbon dioxide (CO₂) using Transcranial Doppler (TCD).⁷ Although these studies have shown that hypertensives may have lower cerebral vasoreactivity to CO₂, gaps in our knowledge remain unanswered. Hypertension is a major risk factor for stroke and white matter hyperintensities (WMH), both associated with impaired cerebral vasoreactivity.^8–9 TCD cannot assess if the lower vasoreactivity noted in hypertension is related or independent of stroke or WMH. TCD, moreover, does not provide simultaneous assessment of vasoreactivity in multiple brain regions. Identifying specific regions that may have lower vasoreactivity in the brain offers an insight to the potential processes by which the brain is affected by hypertension.

Continuous arterial spin labeling (CASL) perfusion magnetic resonance imaging (MRI) can detect cerebral perfusion changes, provide detailed cerebral blood flow mapping in several brain regions, and simultaneously assesses structural brain changes.^10–12 CASL-MRI hence addresses the gaps in our knowledge regarding hypertension and cerebral vasoreactivity. Our group has validated the use of CASL-MRI to assess cerebral vasoreactivity in diabetics and stroke survivors.^{8, 13} In this study, we expanded the application of CASL-MRI to assess the relation between hypertension and cerebral vasoreactivity.

Hypertension is also associated with changes in circadian rhythms, most notably lack of nighttime dipping.¹⁴ This in turn has been associated with stroke and greater brain atrophy.^15–16 It is not known if the potentially impaired cerebrovascular reactivity seen in hypertension is related to elevated blood pressure, lower nighttime dipping or both.

Our objective was to determine the association between hypertension and cerebral vasoreactivity to CO₂, after accounting for WMH and prior stroke using CASL-MRI. Our second objective was to investigate the relation between nighttime dipping and vasoreactivity.

Methods

Subjects

Potential subjects were invited to the study using local advertisement in the greater Boston area. All evaluations were conducted at the Beth Israel Deaconess Medical Center. Inclusion criteria were: Fifty years or older, able to perform study procedures including ambulatory blood pressure monitoring (ABPM) and brain MRI. Individuals with stroke were allowed in this study as long as they were more than 6 months after acute event that affected < 1/3 of middle cerebral artery territory and had a modified Rankin Scale score < 4. The average time post stroke was 6.1 years. Individuals who were receiving antihypertensive medications were also allowed but these medications were gradually tapered over 3 days and discontinued prior to the study for at least 2 days. Anticoagulation and antihyperlipidemic medications were allowed.

Exclusion criteria were: diabetes mellitus (history or hemoglobin A1C levels >7.0 mg/dl), dementia or Alzheimer’s disease, coronary heart disease or congestive heart failure (identified by history, medications or clinical examination), hospitalization, chronic renal and liver disease, transplantation, active cancer treatment or prior exposure to chemotherapy or radiation, or symptomatic arrhythmias. The study protocol and the investigators adhered to the principles of the Declaration of Helsinki and Title 45, U.S. Code of Federal Regulations, Part 46, Protection of Human Subjects, Revised November 13, 2001, effective December 13, 2001. All subjects provided written informed consent and the protocol was approved by the Beth Israel Deaconess Medical Center Institutional Review Board.

We screened potential subjects with detailed medical history and physical and neurological examinations, electrocardiogram, and routine laboratory tests. A trained research nurse performed manual blood pressure measurements according to the American Heart Association guidelines¹⁷. If the participant was receiving antihypertensive medications, then he or she was given instructions to taper off the medication according to a standard protocol. Participants were then admitted for 2 days to the General Clinical Research Center. Two manual blood pressure and heart rate measurements were performed three times per day for 2 days. Cognitive assessment was performed using the mini mental status examination (MMSE).¹⁸

Ambulatory Blood Pressure Monitoring

ABPM was recorded from 8:00 am the first admission day to 8:00 am the next day using a portable automatic monitor Dynapulse (Pulse Metric Inc., Vista CA). This monitor has been previously validated against intra-arterial blood pressure measurement with a correlation of 98% (p<0.001).¹⁹ Systolic (SBP) and diastolic (DBP) blood pressure and heart rate (HR) were measured at 20-minute intervals during the day and at 30-minute intervals during the night. They were asked to follow their daily routine resembling the usual daily activities at home documented in a personal diary completed prior to study initiation. All activities performed during the daytime and the sleep and wake times were confirmed using the diary and direct observation.

Dipping and hypertension definition

Participants were considered hypertensive if they had history of hypertension or were receiving antihypertensives. Those who were not hypertensive based on their history or medication profile but had elevated blood pressure measurements (>=135/85 mm Hg or greater on ABPM^20–21 and average casual blood pressure>=140/90 mm Hg) were also considered hypertensive (only 2 participants). Dipping was calculated according to the standard formula: (daytime blood pressure-nighttime blood pressure)/daytime blood pressure × 100. Those who had >10% nighttime decline in SBP or DBP were considered dippers.^{20, 22}

Magnetic resonance imaging protocol and data analysis

Brain imaging protocol is described elsewhere^{8, 23} and was performed 3 hours after completing the ABPM monitoring at the 3-Tesla MRI scanner. Briefly, high-resolution anatomical image 3D magnetization prepared rapid gradient echo (MP-RAGE) were acquired to quantify volume of white matter and gray matter (cm³). WMH volume was measured on FLAIR images. In our analyses, we used the WMH volume (cm³) divided by intracranial volume to account for head size differences between participants. Cerebral perfusion was measured using Continuous Arterial Spin Labeling (CASL). CASL technology uses proton labeling of the water molecules in the blood flowing in the blood vessels to allow for accurate and reproducible measures of perfusion.¹¹ Labeled and unlabeled images were collected over 2 minute periods during normal breathing, CO₂ rebreathing with 95% air and 5% CO₂ (hypercapnia), and hyperventilation (hypocapnia). Two scans were obtained during each phase and data was then averaged. End-tidal CO₂ was continuously monitored and averaged over 15-second intervals. Cerebral vasoreactivity was calculated as the slope of the regression between cerebral perfusion and end-tidal CO₂, in units of ml/100gm/min/mmHg, during normal breathing, CO₂ rebreathing, and hyperventilation. We calculated a global cerebral vasoreactivity slope and regional vasoreactivity slopes in the frontal, parietal, temporal, occipital and cerebellar regions. In the stroke participants, vasoreactivity measures were calculated for both the stroke and non-stroke hemispheres. (please see http://hyper.ahajournals.org for more details)

Statistical Analyses

We used t-test or chi-square to compare characteristics of the hypertensive and normotensive groups. We used general linear models to investigate the association between hypertension and cerebral vasoreactivity (global and regional). Models were adjusted for age, sex, race/ethnicity, body mass index, antihypertensives, and white matter hyperintensities volume/intracranial cavity. We included WMH because of it effect on vasoreactivity and its close association with hypertension.⁹ We calculated the least square mean vasoreactivity adjusted for covariates when comparing the hypertensive and normotensive groups. Effect size for hypertension was measured using partial eta square (η²).²⁴ Higher values indicate stronger effect size.

We used regression analysis to investigate the association between blood pressure and dipping magnitude (both as a continuous measures) with cerebral vasoreactivity. Models were adjusted for the same covariates. We reported the regression coefficient per 10 mm Hg in the blood pressure analyses and per 10% in the dipping analyses, adjusted for these covariates. Effect size was measured by partial correlations.

Finally, we performed stratified analysis by stroke. We used generalized linear models adjusted for the same covariates, except for stroke. We performed two sets of analyses in the stroke group (stroke side and opposite side). To assess if there were regional variation in the impact of hypertension on cerebral vasoreactivity we used mixed models for correlated data for these analyses since there was a high correlation between the brain regions within an individual.²⁵ We calculated difference in the adjusted least square mean vasoreactivity between hypertensives and normotensives by region. We then tested if the differences by region were significant.

Results

Sample

Of the 68 participants, 6 did not have complete ABPM data. Therefore, data on 62 individuals were used for this analysis (mean age 66.7±1.0 years, 55% women, 84% White). Of those, 40 (65%) were hypertensive (38 were on antihypertensives or were diagnosed with hypertension; and additional 2 had elevated blood pressure based on ABPM and casual blood pressure measurements). As shown in Table 1, there were no significant differences between hypertensive and normotensive participants in demographic, clinical, laboratory, and brain gray and white matter volume measures. ABPM indices and lipid measures were different between the 2 groups. Global WMH volume was higher in the hypertensive group but this did not reach statistical significance.

Table 1.

Demographic, vascular, and brain MRI characteristics of the normotensive and hypertensive participants.

Characteristic	Normotensives	Hypertensives	p
N	22	40
Age, years	67±2	67±1	0.96
Women, %	56%	54%	0.81
White, %	83%	84%	0.98
Body Mass Index, kg/m2	25.8±0.9	26.1±0.7	0.77
Alcohol drinks per week	1.7±0.8	7.9±2.9	0.18
Current-smokers, %	6%	11%	0.71
Stroke, %	32%	57%	0.052
Mini-Mental State Examination	28.1±0.5	26.9±1	0.097
Blood Pressure
Systolic blood pressure, mm Hg	118±1	133±2	<0.0001
Diastolic blood pressure, mm Hg	68±1	68±1	0.9
24 ABPM
Systolic blood pressure, mm Hg	119±2	136±2	<.0001
Diastolic blood pressure, mm Hg	63±1	67±1	0.049
Daytime systolic blood pressure, mm Hg	121±2	137±2	<.0001
Daytime diastolic blood pressure, mm Hg	65±1	68±1	0.11
Nighttime systolic blood pressure, mm Hg	115±2	133±2	<.0001
Nighttime diastolic blood pressure, mm Hg	59±2	64±1	0.033
Heart rate, beats per minute	67±2	67±1	0.89
Dippers, %	54%	27%	0.03
Laboratory
Hemoglobin, g/dl	13.8±0.3	13.9±0.2	0.69
Hematocrit, mg/dl	41.1±0.9	40.9±0.5	0.87
Cholesterol, mg/dl	207.4±9.9	184±5	0.02
Low density lipoprotein, mg/dl	117.5±9.1	94.5±4.1	0.009
Glucose, mg/dl	75.4±3.4	83.4±2.5	0.077
Medications
Hypertension treatment	NA	85%	NA
Diuretics		34%	NA
Angiotensin converting enzyme inhibitors		30%	NA
Beta Blockers		25%	NA
Calcium Channel Blockers		18%	NA
Statins	31%	50%	0.16
Structural Brain Measures
Gray Matter Volume/Intracranial Volume	0.42±0.008	0.41±0.004	0.41
White Matter Volume/Intracranial Volume	0.28±0.008	0.28±0.004	0.62
White Matter Hyperintensities, cm3	8.8±2.8	16.0±3.5	0.12

Numbers are presented as Mean±Standard Error for continuous measures or percentages for discrete variables.

Hypertension and Vasoreactivity

The mean vasoreactivity in the overall sample was 0.78±0.05 ml/100gm/min/mmHg. As shown in Table 2 after adjusting for covariates, stroke and WMH/Intracranial cavity volume, hypertensive participants had significantly lower global vasoreactivity compared to normotensives (p=0.0012). The effect size of hypertension was considerably high (partial η²=0.32). The association of hypertension with lower vasoreactivity was significant in all brain regions except the cerebellum. Figure 1 demonstrates the differences in perfusion at baseline and during hypercapnia and hypocapnia phases between a hypertensive and a normotensive participant.

Table 2.

Global and regional vasoreactivity to CO₂ in normotensive and hypertensive participants

Brain Region	Normotensives	Hypertensives	p-value	partial eta sq
Whole Brain	1.11±0.13	0.43±0.1	0.0012	0.32
Cerebellum	0.69±0.13	0.34±0.1	0.0711	0.12
Frontal	1.2±0.15	0.45±0.12	0.002	0.3
Occipital	1.2±0.13	0.52±0.1	0.0013	0.32
Parietal	1.38±0.17	0.47±0.13	0.0007	0.35
Temporal	0.95±0.12	0.37±0.09	0.0015	0.31

Numbers are least square means adjusted for age, sex, race/ethnicity, body mass index, stroke, antihypertensives and white matter hyperintensities/intracranial cavity

Figure 1.

Anatomical and perfusion images for non stroke normotensive subject (A1 - A5) and non stroke hypertensive subject (B1 - B5).

Footnote: A1 and B1: MPRAGE images used for brain volume analysis. A2 and B2: FLAIR images for WMH analysis. A3 and B3: baseline CBF. A4 and B4: CBF during CO₂ rebreathing. A5 and B5: CBF during hyperventilation. The perfusion scale is 0 to 150 ml/100g/min on all CBF maps. Non-stroke normotensive subject has similar baseline CBF, higher CBF during hypercapnia and lower CBF during hyperventilation, as compared to non-stroke hypertensive subject.

Blood pressure, nighttime dipping, and vasoreactivity

Higher SBP, but not DBP, was associated with lower vasoreactivity. For each 10 mm Hg increments in SBP, global vasoreactivity decreased by 0.1 units (partial R²=0.16) after adjusting for demographics, body mass index, stroke, antihypertensives and WMH/intracranial volume. Regionally, this was only significant in the parietal and occipital lobes. Dipping status was not associated with vasoreactivity: non-dippers: 0.77±0.08 and dippers: 0.77±0.10 ml/100gm/min/mmHg, p=0.76). The magnitudes of dipping in SBP and DBP (as continuous measure) were not related to vasoreactivity. These results are shown in the online supplement (please see Table S1 at http://hyper.ahajournals.org).

Estimating the magnitude of decreased vasoreactivity from hypertension without stroke and from stroke without hypertension

Normotensive participants without stroke had the highest global vasoreactivity. On the other hand, hypertensive participants without stroke had similar vasoreactivity as stroke participants. (Normotension stroke free1.38±0.16 ml/100gm/min/mmHg; hypertension stroke free:0.59±0.010 ml/100gm/min/mmHg (p=0.001 after adjusting for covariates). Normotension stroke: 0.47±0.40 ml/100gm/min/mmHg; hypertension stroke 0.58±0.35 ml/100gm/min/mmHg (p=0.25 in the stroke hemisphere; p=0.57 the non-stroke hemisphere). The regional vasoreactivity results followed the same pattern and are shown in Figure 2. Hypertensive participants without stroke had comparable vasoreactivity to those with stroke in all regions after adjusting for all related covariates. When we compared the hypertension-related differences in vasoreactivity in the various brain regions, the decline was greatest in the parietal and frontal regions in non-stroke hypertensive participants as shown in Table 3.

Figure 2.

Regional cerebral vasoreactivity in stroke and non-stroke normotensive and hypertensive participants. A: non-stroke side for the stroke participants and corresponding hemisphere in those without stroke. B: Stroke side for the stroke participants and the corresponding hemisphere in those without stroke.

Footnote: NTN: normotension, HTN: hypertension. Sample size: NTN= 22 (7 stroke and 15 non-stroke); HTN=40 (23 stroke and 17 non-stroke); Values are least square means adjusted for age, sex, body mass index, WMH/intracranial volume, and hypertensive therapy. In the non-stroke group, p-value for the impact of hypertension on vasoreactivity was 0.035 and in the stroke group 0.97 (stroke side) and 0.90 (opposite side).

Table-3.

Adjusted differences (± standard error) in vasoreactivity between hypertensives and normotensive participants by brain region and history of stroke

Group	Region	Difference*	P-Value
Non-stroke group	cerebellum	0.22±0.16	0.18
	frontal	0.35±0.16	0.03
	occipital	0.29±0.16	0.08
	parietal	0.45±0.16	0.007
	temporal	0.3±0.16	0.07
Stroke Group
Stroke Side	cerebellum	0.09±0.16	0.57
	frontal	0.02±0.16	0.87
	occipital	−0.09±0.16	0.55
	parietal	−0.11±0.16	0.49
	temporal	0.05±0.16	0.76
Non-stroke side	cerebellum	0.01±0.16	0.95
	frontal	−0.11±0.16	0.51
	occipital	0.02±0.16	0.91
	parietal	−0.05±0.16	0.76
	temporal	0.06±0.16	0.71

^*Differences= Least square mean vasoreactivity in the normotensive group-hypertensive group. The differences between lobes was significant in all three groups (p<0.001 non stroke group, p=0.003 stroke side, p=0.005 opposite side). Model adjusted for age, sex, BMI, race/ethnicity, WMH/intracranial volume, and hypertensive therapy

Discussion

Our findings extend our prior knowledge, which is mostly based on TCD and Positron Emission Tomography scan technology, by demonstrating that hypertension is directly associated with lower cerebral vasoreactivity to CO₂ independent of stroke and WMH. The magnitude of decrease in cerebral vasoreactivity from hypertension in the absence of stroke is comparable to that of stroke. Moreover, there is an inverse association between lower cerebral vasoreactivity and higher SBP but not with nighttime dipping. Regionally, hypertension is associated with lower vasoreactivity in all cortical brain regions but is more prominent in the frontal and parietal areas.

To our knowledge this is the first study using CASL-MRI to assess vasoreactivity in hypertensives. We have previously reported that higher SBP is associated with poor global and regional perfusion using CASL-MRI.²³ Dai et al demonstrated that in 19 hypertensive participants in the Cardiovascular Health Study, regional blood flow was lower than normotensives using CASL-MRI.²⁶ Hypertension is also associated with lower vasoreactivity in addition to decreased perfusion. This may further impair ability to maintain cerebral blood flow during increasing demands and lead to further brain injury. Since lower vasoreactivity maybe associated with cognitive and physical declines,^{3, 5} this observation may provide a partial explanation for the relation between hypertension and both cognitive impairment and physical disability noted in other studies.^27–28

The mechanisms of low vasoreactivity in hypertensive individuals are not fully understood. Our study suggests that there are at least 2 potential mechanisms that impair vasoreactivity in hypertension: one indirect mechanism via its relation with stroke and WMH which eventually lower cerebral vasoreactivity²⁹ and another independent of both. Hypertension is associated with significant decrease in nitric oxide production in the cerebral circulation, ³⁰ and the ability of the cerebral circulation to respond appropriately to CO₂ is related to endothelial nitric oxide production.^31–34 Non-nitric oxide function of the endothelium are also impaired in hypertension.³⁵ Adenosine-5'-triphosphate-dependent K+ channel activation may also mediate CO₂-induced nitric oxide activity in the pial arterioles.³⁶ Hypertension is also associated with structural changes in the blood vessels such as increased tortuosity and reduced arteriolar branching, which may also affect cerebral vasoreactivity.³⁷

Of particular concern is that vasoreactivity in stroke free hypertensives is comparable to that of the stroke participants. This is of great significance since from the cerebrovascular function perspective the impact of hypertension is as severe as that of stroke. This suggests that hypertensive individuals without stroke have impaired cerebrovascular function comparable to those who already have stroke. Since impaired cerebrovascular reactivity to CO₂ is related to poor vascular and functional outcomes in various populations^{5, 38}, future studies that investigate the impact of antihypertensive medications on cerebral vasoreactivity are important.

Non-dipping is associated with worse vascular macrovascular and microvascular brain injury in hypertension.^39–40 In this study we did not identify an association between dipping and cerebral vasoreactivity. The lack of association in this study may be related to our relatively small sample size. It is also possible that nighttime dipping may not be intimately related to brain perfusion and cerebrovascular reactivity. We have previously reported that nocturnal dipping in blood pressure was not associated with global cerebral perfusion.²³

Regionally, hypertension is related to lower vasoreactivity in all cortical regions. The magnitude of decrease in vasoreactivity from hypertension was greatest in the fornto-parietal regions. These 2 regions are involved in mood and in cognitive and physical performance. ^41–42 We have recently reported that hypertensive individuals have impairments in mobility, cognition and mood even in the absence of clinical symptoms or disease.⁴³ This may shed more light on the possible mechanisms by which hypertension may be related to these impairments.

Although CASL-MRI is still not widely used clinically, this study provides evidence that it can be used to assess vasoreactivity. CASL-MRI requires no contrast. CASL-MRI also addresses prior limitations of TCD and has a relatively low signal-to-noise ratio and hence is ideal to assess vasoreactivity in the brain.

One limitation of our study is the cross-sectional design which precludes investigating the temporal relation between hypertension and vasoreactivity. As in most hypertension studies, antihypertensive exposure is an important confounder and limits our ability to study the natural impact of elevated blood pressure. In this study, we addressed this issue by performing our measures off antihypertensives, albeit only for few days for ethical reasons. We also adjusted all our models for antihypertensive therapy.

Conclusions

Hypertension is associated with low cerebrovascular reactivity to CO₂ independent of existing stroke or WMH. This decrease is similar in magnitude to that of stroke. Although all cortical regions are affected, the greatest impact was in the fronto-parietal region. Low vasoreactivity to CO₂ in hypertension may further increase the risk of developing brain-related end-organ damage, beyond its role in microvascular ischemic injury and macrovascular brain disease.

Perspectives

Hypertension is associated with significant impairment in cerebral vasoreactivity. This impairment is comparable to that from stroke. Since impaired cerebral vasoreactivity may be associated with poor functional and cognitive outcomes, prospective studies are needed to investigate the role of hypertension treatment in reversing these effects.