Regional cerebral blood flow and arterial blood volume and their reactivity to hypercapnia in hypertensive and normotensive rats

Abstract

Chronic hypertension induces cerebrovascular remodeling, changing the inner diameter and elasticity of arterial vessels. To examine cerebrovascular morphologic changes and vasodilatory impairment in early-stage hypertension, we measured baseline (normocapnic) cerebral arterial blood volume (CBV_a) and cerebral blood flow (CBF) as well as hypercapnia-induced dynamic vascular responses in animal models. All experiments were performed with young (3 to 4 month old) spontaneously hypertensive rats (SHR) and control Wistar–Kyoto rats (WKY) under ∼1% isoflurane anesthesia at 9.4 Tesla. Baseline regional CBF values were similar in both animal groups, whereas SHR had significantly lower CBV_a values, especially in the hippocampus area. As CBF is maintained by adjusting arterial diameters within the autoregulatory blood pressure range, CBV_a is likely more sensitive than CBF for detecting hypertensive-mediated alterations. Unexpectedly, hypercapnia-induced CBF and blood-oxygenation-level-dependent (BOLD) response were significantly higher in SHR as compared with WKY, and the CBF reactivity was highly correlated with the BOLD reactivity in both groups. The higher reactivity in early-stage hypertensive animals indicates no significant vascular remodeling occurred. At later stages of hypertension, the reduced vascular reactivity is expected. Thus, CBF and CBV_a mapping may provide novel insights into regional cerebrovascular impairment in hypertension and its progression as hypertension advances.

Introduction

The adaptive response of arterial vessels to chronic hypertension includes a narrowing of the lumen and an increase in wall thickness.^{1, 2} Consequently, cerebral arterial blood volume (CBV_a) and arterial vessel elasticity decrease, which may in turn alter arterial vascular reactivity to various stimuli. These cerebrovascular alterations could eventually lead to a regional reduction in cerebral blood flow (CBF) and tissue oxygenation,^{3, 4, 5} thereby increasing the risk of stroke, vascular dementia, and cognitive decline.^{6, 7, 8} However, early-stage alterations in the cerebrovasculature are easily modified. As the cerebral regulation of blood pressure differs from the systemic regulation, treatment aimed at reducing peripheral blood pressure may not provide the necessary therapy to prevent impairments of the brain vasculature. Therefore, the ability to detect the progression of hypertension and cerebrovascular remodeling by in vivo imaging is vital.

Hypertension-induced alterations in CBF have been assessed by magnetic resonance imaging (MRI), single-photon emission computed tomography and positron emission tomography, which reported a slight reduction in baseline (normocapnic) CBF in some,^{8, 9, 10} but not in all hypertensive subjects.^{1, 11, 12, 13, 14} In addition, studies showed that, as compared with normotensive subjects, hypertensive subjects have a blunted CBF response to neural tasks¹¹ and CO₂ stimulus.¹⁵ In general, the brain maintains adequate CBF within a certain autoregulatory range independent of changes in systemic arterial pressure by adjusting vessel diameters of arteries and arterioles. In early-stage hypertension, CBF may maintain constant despite a decrease in CBV_a, before the compensation eventually fails. Thus, CBV_a may be a more sensitive index than CBF for identifying and assessing cerebrovascular risk factors in early-stage hypertension, and its reduction may be an early predictor of altered CBF.

In this study, we measured quantitative CBF and CBV_a, as well as cerebrovascular reactivity in a well-established hypertensive animal model (spontaneously hypertensive rats (SHR)) and a control model (Wistar–Kyoto rats (WKY)) to detect regional hypertensive cerebrovascular changes. Quantitative CBF was measured with arterial spin labeling (ASL) MRI, and quantitative CBV_a was obtained using magnetization transfer (MT)-varied MRI.^{16, 17, 18} Varying the MT effect allowed us to separate MT-sensitive tissue from the MT-insensitive arterial blood signal.^{16, 17, 18} Dynamic cerebrovascular reactivity as a response to hypercapnic manipulation was measured by blood-oxygenation-level-dependent (BOLD) functional magnetic resonance imaging in addition to CBF and CBV_a MRI. As BOLD functional magnetic resonance imaging is a vascular measure complementary to CBF and CBV_a, combining all three measures may provide important indices of vascular reserve and vessel viability.

Materials and Methods

Animal Preparation and Hypercapnia Stimulation

All animal protocols were approved by the University of Pittsburgh Institutional Animal Care and Use Committee, in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. A total of twenty four male rats were studied (Charles River Laboratories, Wilmington, MA, USA); twelve animals each for the SHR and WKY groups. All animals were used for baseline measurements, while ten SHR and eight WKY rats were used for hypercapnic challenge.

The animals were initially anesthetized with 5% isoflurane and intubated. Subsequently, the isoflurane level was reduced to 2% for surgery; the femoral artery and vein were catheterized to monitor arterial blood pressure and to administrate supplemental fluid, respectively. The isoflurane level was then reduced to ∼1% with air supplemented with O₂ to attain a total O₂ level of ∼30% throughout experiment, which assists maintaining a normal PaO₂ level under anesthesia. The head of the animal was carefully secured in a home-built restrainer before placement in the magnet. During experiments, mean arterial blood pressure (MABP), end-tidal CO₂ (EtCO₂), O₂, and isoflurane level were continuously monitored and recorded with Acqknowledge software (MP150, BIOPAC Systems, Goleta, CA, USA). Blood gases were measured (Stat profile pHOx, Nova Biomedical, Waltham, MA, USA) and ventilation rate and volume were adjusted accordingly. Body temperature was monitored by a rectal probe and maintained at 37 to 37.5°C with a feedback hot water circulator.

Thirty-second CO₂ stimulation was performed by switching ventilation gas mixtures of air with 30% oxygen without and with ∼4% CO₂ using an electronically controlled pneumatic valve. An inter-hypercapnia interval was >15 minutes. It should be noted that as longer stimulation and higher CO₂ level require a longer inter-stimulus delay for returning to the baseline physiologic condition, the relatively short stimulus duration and mild hypercapnia were used for repeating experiments rapidly.

Magnetic Resonance Acquisitions

All studies were performed at a magnetic field strength of 9.4T with 310 mm diameter of bore size, interfaced to a Unity INOVA console (Varian, Palo Alto, CA, USA). The actively shielded gradient coil had an inner diameter of 120 mm with a gradient strength of 4 G/mm and a rise time of 130 microseconds (Magnex, Abingdon, UK). Two actively detunable radiofrequency (RF) coils were used: a butterfly-shaped surface coil positioned in the neck region for continuous ASL, and a surface coil of 23 mm diameter positioned on top of the head for tissue saturation via MT effects and for image acquisition. A three-dimensional anatomic image was acquired using a magnetization prepared rapid acquisition gradient echo sequence with TR=10 milliseconds, TE=5 milliseconds, TI=1.4 seconds, and a voxel resolution of 117 × 117 × 117 μm³. From the three-dimensional image, five 2-mm thick coronal slices were selected for cerebrovascular imaging. To compare with cerebrovascular images, T₂-weighted 2-D images were then acquired in the same slices with TR=2 seconds, 8 echoes with 11-millisecond spacing, and an in-plane resolution of 58 × 58 μm².

All physiologic images were acquired using the single-shot spin-echo echo planar imaging (EPI) technique with an in-plane resolution of 469 × 469 μm², TE=18 milliseconds, and TR=3 seconds. A 2.75-second magnetization preparation period was used for applying ASL and MT-inducing RF pulses as described previously.¹⁹ Each experimental set included four ASL runs to obtain CBF and CBV_a with improved temporal resolution; two runs each with two different MT saturation levels. In each ASL run, arterial spin labeled (lab) and control unlabeled (unlab) images were acquired repeatedly in an interleaved manner. The acquisition order of labeled (lab) and unlabeled (unlab) images was changed in alternate runs to increase the temporal resolution of the ASL data. The targeted MT saturation level was obtained by adjusting the power level of MT-inducing radiofrequency pulses with +8500 Hz off-resonance frequency; MT ratios (MTR) were (1−S_sat/S₀)=0 and 0.5, where S_sat and S₀ are the equilibrium signal in the presence and absence of MT saturation. Each CO₂ challenge run consisted of 150-second pre-stimulation (25 lab and unlab images), 30-second hypercapnic stimulation (5 pair images), and 450-second post stimulation (75 pair images). Four to six experimental sets (16 to 24 runs) were obtained in each animal. For animal studies without CO₂ challenge, a similar number of images were also acquired.

Data Processing

Data analysis was performed with STIMULATE (CMRR, http://www.cmrr.umn.edu/stimulate, Minneapolis, MN, USA), ImageJ (NIH, http://rsb.info.nih.gov/ij, Bethesda, MD, USA), and in-house Matlab routines (Mathworks, Natick, MA, USA). For each study, all runs with identical experimental conditions were averaged to generate group data. As two different orders of ASL and control image acquisition with 6-second temporal resolution were used, the practical temporal resolution was 3 seconds.¹⁹

Cerebral blood flow and CBV_a maps were obtained, and regional baseline values and evoked responses were determined in a region-by-region basis. To select regions of interest (ROIs), 2-D anatomic images were generated from the three-dimensional anatomic image to match five-slice cerebrovascular maps. Then, five ROIs corresponding to the sensory cortex, motor cortex, caudate putamen, thalamus, and hippocampus were carefully and conservatively delineated on each of the five-slice anatomic images in both hemispheres, based on a rat atlas.²⁰ Pixels with anatomic variations within the slice, such as those coinciding with the hippocampal and ventricle boundaries, were easily visualized with the high-resolution three-dimensional data, and not included. All pixels within each ROI were averaged for each subject's ROI analysis. Individual subject results were averaged and group data are reported as mean±s.d. A two-sample equal variance t-test was performed for statistical significance for the difference between both animal groups.

Baseline cerebral blood flow and cerebral arterial blood volume

For baseline CBF and CBV_a measurements, all repeated ASL data were averaged during the pre-CO₂ challenge period, or an all period in case of no CO₂ challenge, resulting in one labeled and one unlabeled control image for MTR=0 (no MT effect), and MTR=0.5. Details of data processing were described previously.^{16, 17} Briefly, the ASL signals (ΔS_ASL,MT) were obtained from the difference between control and labeled images for each MTR level. The CBV_a signal was separated from ΔS_ASL,MT using the different MT effects in tissue versus arterial blood. This is possible because tissue signals are selectively reduced by MT, while the arterial blood pool experiences only a minimal MT effect owing to the fast and continuous inflow of fresh blood spins from outside the region of the MT-inducing radiofrequency coil's coverage. Normalized ΔS_ASL,MT signals (ΔS_ASL,MT/S₀) were linearly fitted against normalized control signals (S_MT/S₀), where S₀ is the signal intensity from the control image with MTR=0. Cerebral blood flow (in units of mL/g per second) without any arterial blood volume contribution was determined from the slope of the linear fitting as

where λ is the tissue–blood partition coefficient of 0.9 mL/g;²¹ T₁ is the T₁ value of tissue without MT and in the absence of CBF contributions; α_c is the labeling efficiency of arterial spins at the capillary exchange site within the imaging slice, calculated as α₀·exp(−τ_c/T_1a), where α₀ is the labeling efficiency at carotid artery=0.525 and τ_c is the blood transit time from carotid artery to exchange site=0.6 seconds;²² and ζ is the correction factor to the steady state, which is (1-exp(−T_lab/T_1app)), where T_lab is the time span for the spin-preparation period=2.75 seconds and T_1app (apparent T₁)=1.9 seconds.²³

Quantitative CBV_a (in units of mL/g) was obtained from the intercept and slope of the linear fitting as

where α_a is the labeling efficiency of spins entering the imaging slice, calculated as α₀ × exp(−τ_a/T_1a), where τ_a is the blood transit time from carotid artery to imaging slice=0.3 seconds²² and T_1a is T₁ value of arterial blood=2.3 seconds at 9.4 T.¹⁷

Dynamic changes in blood-oxygenation-level-dependent, cerebral blood flow, and cerebral arterial blood volume to hypercapnia

Hypercapnia-induced relative BOLD and CBF changes were obtained from control and ASL images, respectively, in the run without MT effect (MTR=0). To obtain dynamic CBV_a changes induced by CO₂ stimulation, values should be determined for each time point. However, only 4 to 6 runs were averaged, so this data did not have sufficient sensitivity to reliably quantify CBV_a using the MT-varied ASL method. Instead, hypercapnia-induced CBV_a changes were calculated with the MT-varied BOLD method.^{16, 18, 24} The CO₂ stimulus-induced signal change normalized by S₀ (ΔS_BOLD,MT/S₀) was linearly fitted against the normalized baseline signal, S_MT/S₀. The absolute change of CBV_a (in units of mL/g) was obtained from the intercept, as ΔCBV_a=λ × intercept.

Hypercapnia-induced hemodynamic responses did not reach the steady state (i.e. plateau) because of the 30-second stimulation; mean reactivity values for each animal were determined by averaging the hypercapnic period, which was defined from corresponding hemodynamic response as a time period between 6 seconds before the peak and 24 seconds after the peak, which is agreeing with ∼90% and ∼80% of the peak of BOLD response, respectively, for every measurements.

Results

During the experiments, all animals were maintained within normal physiologic ranges. Throughout the study, physiologic conditions were similar between WKY and SHR; pH=7.48±0.05 and 7.50±0.03, PaCO₂=34.9±3.54 mm Hg and 31.3±3.47 mm Hg, PaO₂=116±13 mm Hg and 133±24 mm Hg, EtCO₂ before hypercapnic challenge=3.23±0.31% and 3.53±0.33%, SO₂=97.8±1.6% and 98.7±0.6%, Hct=36.8±2.39% and 37.8±2.33%, body weight=325±39 g and 321±45 g, and age=16.7±3.2 and 14.7±2.6 weeks, respectively. However, the two groups had a significantly different MABP=98±11 mm Hg in WKY versus 148±18 mm Hg in SHR (P<0.0001). These MABP levels did not change during CO₂ stimulation.

We evaluated the CBF and CBV_a in the hypertensive and normotensive groups. Figure 1 shows T₂-weighted anatomic images and quantitative maps of CBF and CBV_a from one WKY and one SHR under the baseline condition (without CO₂ stimulation). In T₂-weighted images, ventricle areas indicating hyperintensity were enlarged in SHR, which is consistent with previous findings.^{25, 26, 27} Interestingly, CBF maps were similar for both groups, but SHR had much smaller CBV_a values as compared with WKY. To determine similarities and differences between WKY and SHR, we conservatively chose five ROIs (see Figure 1A), and calculated their CBF and CBV_a values (Figure 2). As expected from the CBF maps (Figure 1B), CBF values between the two groups were not significantly different (Figure 2A), despite a small difference in baseline PaCO₂. However, CBV_a values from SHR were lower than those from WKY in all five regions. The largest reduction of CBV_a responding to hypertension was observed in the hippocampus, indicating that region-dependent cerebrovascular changes indeed occur.

Figure 1.

Anatomic T₂-weighted images (A) and quantified cerebral blood flow (CBF) (B) and cerebral arterial blood volume (CBV_a) (C) maps from Wistar–Kyoto rats (WKY) and spontaneously hypertensive rats (SHR). Data from one representative animal for each group are shown. The volume of hyperintense ventricles increased in SHR. Overall, both animal models had similar CBF values, whereas SHR demonstrated smaller CBV_a values than WKY. Five regions of interest were chosen in both hemispheres, but overlaid on one hemisphere in the anatomic images; sensory cortex—orange; motor cortex—blue; caudate putaman—red; thalamus—purple; hippocampus—green.

Figure 2.

Baseline cerebral blood flow (CBF) (A) and cerebral arterial blood volume (CBV_a) (B) of five regions of interest (ROIs) under the normocapnic condition. All 12 animal data for each group were averaged. Spontaneously hypertensive rats (SHR) had slightly lower CBF values as compared with Wistar–Kyoto rats (WKY), whereas CBV_a values were signtificantly lower in most areas, particularly in the hippocampus (*P<0.05, **P<0.01). Error bars: s.e.m. (n=12); CPU, caudate putamen; Th, thalamus; Hip, hippocampus.

Dynamic EtCO₂ responses induced by 30-second hypercapnia are plotted in Figure 3. The baseline EtCO₂ level was slightly lower for WKY than for SHR. Owing to a long air delivery system, EtCO₂ changed around 30 seconds after switching to a gas mixture with 4% CO₂, but dynamic EtCO₂ response was reproducible and consistent across all animals studied. The peak EtCO₂ change was ∼1.8% to 2.1%, equivalent to ∼14 to 16 mm Hg in our anesthetized condition, which is higher than self-breathing awake human studies. The evoked EtCO₂ change was slightly higher for WKY than for SHR.

Figure 3.

End-tidal CO₂ (EtCO₂) responses to CO₂ stimulation. All animal data were averaged. Inhalation of ∼4% CO₂ in a gas mixture of air with 30% oxygen began at time 0 and continued for 30 seconds; however, there was a delay owing to a long gas delivery system to the animal inside the magnet. In all animals, dynamic end-tidal CO₂ responses were quite similar. Error bars: s.e.m. SHR, spontaneously hypertensive rats; WKY, Wistar–Kyoto rats.

Hypercapnic challenge activated most pixels (75% to 95% of pixels for P<0.05) in all ROIs (data not shown). However, to minimize pixel selection bias for the comparison between SHR and WKY, all pixel values within each ROI were averaged for each animal. Group-averaged time courses of BOLD, rCBF, and ΔCBV_a responses to hypercapnia are plotted in Figure 4. All regions had similar temporal patterns, so data of the sensory cortex and hippocampus ROIs are shown as representatives. Generally, the dynamics of BOLD and CBF responses were similar between SHR and WKY. Although a temporal resolution of 3 seconds did not allow us accurate analyses of dynamic properties, the time-to-peak of ΔCBV_a appeared shorter for the SHR animals.

Figure 4.

Group-averaged blood-oxygenation-level-dependent (BOLD) (left), relative cerebral blood flow (rCBF) (middle) and cerebral arterial blood volume (ΔCBV_a) (right) responses of the sensory (upper) and the hippocampus area (lower) to CO₂ stimulation. The dynamics of BOLD and ΔCBF were similar, but their peak response was much higher for spontaneously hypertensive rats (SHR). The ΔCBV_a responses in SHR appeared to have a slightly shorter time-to-peak and faster return with similar amplitude of signal changes. Error bars: s.e.m.. WKY, Wistar–Kyoto rats.

The amplitude changes of BOLD, rCBF, and ΔCBV_a were obtained from hypercapnic data, and ΔCBF and relative CBV_a were calculated with baseline CBF and CBV_a. Group-averaged hypercapnia-induced BOLD, rCBF, relative CBV_a, ΔCBF, and ΔCBV_a are shown in Figure 5. Unlike higher evoked EtCO₂ change for WKY versus SHR (see Figure 3) and similar baseline CBF values, SHR demonstrated a much larger BOLD and rCBF reactivity to hypercapnia than WKY in all five regions (P<0.01). However, ΔCBV_a to hypercapnia did not show any difference between the two models (Figure 5E), but relative CBV_a change was slightly higher for SHR than for WKY even though it was not statistically significant.

Figure 5.

Blood-oxygenation-level-dependent (BOLD) (A), relative cerebral blood flow (rCBF) change (B), relative cerebral arterial blood volume (rCBV_a) change (C), ΔCBF (D) and ΔCBV_a (E) during CO₂ stimulation. Hypercapnia-induced BOLD and rCBF responses in spontaneously hypertensive rats (SHR) were significantly higher than those in Wistar–Kyoto (WKY) rats in all regions examined (*P<0.05). ΔCBV_a responses between both groups were similar, whereas relative CBV_a responses appeared higher for hypertensive animals. Error bars: s.e.m.; CPU, caudate putamen; Th, thalamus; Hip, hippocampus.

Blood-oxygenation-level-dependent and rCBF reactivity was highly correlated in both groups (Figure 6). The ratio of BOLD (%) to rCBF (%) (=BOLD/rCBF) were similar for WKY versus SHR; 0.039±0.008 versus 0.036±0.004 for sensory, 0.030±0.009 versus 0.032±0.009 for motor, and 0.029±0.003 versus 0.027±0.005 for the hippocampus, and slightly different for CPU, 0.023±0.002 versus 0.030±0.005 and for the thalamus, 0.025±0.003 versus 0.035±0.012. This indicates that the BOLD reactivity is a good index of CBF reactivity.

Figure 6.

Relationship between the hypercapnia-induced blood-oxygenation-level-dependent (BOLD) response and relative cerebral blood flow (rCBF) changes. This graph includes data from all five regions of interest (ROIs) (different symbols) in all of the animals. Blood-oxygenation-level-dependent responses were linearly correlated with rCBF changes in both spontaneously hypertensive rats (SHR) (red symbols) and Wistar–Kyoto rats (WKY) (blue symbols). This indicates that BOLD is a good index of CBF reactivity, as no oxygen consumption is expected.

Discussion

We investigated the regional cerebrovascular similarities and differences between SHR and normotensive controls using MRI techniques. These studies revealed several major findings: (1) both animal models had similar CBF values, but SHR had a smaller regional CBV_a value than WKY; and (2) in response to CO₂, SHR displayed greater CBF and BOLD reactivity than WKY, but both groups had a similar ΔCBV_a.

Before we discuss the biologic implications of our studies, it is important to evaluate our methodologies critically. There are several concerns. (1) One concern is whether baseline CBF versus CBV_a can be compared fairly owing to potentially different sensitivities. Both CBF and CBV_a were obtained from exactly the same data sets using equations (1) and (2), respectively. Cerebral blood flow was calculated from the slope, whereas CBV_a was determined primarily from the intercept of linear fitting, as the slope is much less than 2α_a in equation (2). As only two points were used for fitting, fitting errors could not be determined. The ratio of pixel-wise SNR between (slope/T₁) and intercept was 1.62, thus the error of CBV_a would be higher than that of CBF. As CBF values were similar for SHR and WKY, higher errors in CBV_a would induce less significant changes between the two animal models. However, the inter-animal variations of CBF and CBV_a were similar (Figure 2), possibly due to a small error in ROI values. Consequently, significant baseline CBV_a differences between two models cannot be explained by measurement error. (2) In our hypercapnic studies, different data sets were used; rCBF from ASL data (i.e., unlabeled—labeled) without MT, BOLD from unlabeled data without MT, and ΔCBV_a from unlabeled data with and without MT. While both rCBF and BOLD responses were obtained directly by comparing values at baseline versus stimulation, ΔCBV_a was from comparing intercepts of linear fitting. Further, less signal averaging was performed for dynamic studies compared with baseline measurements. Consequently, ΔCBV_a is more prone to errors than BOLD and CBF responses. (3) As we used a relative short hypercapnic challenge, hemodynamic responses did not reach to a steady state. Thus, the averaged amplitudes measured during dynamic–hemodynamic changes were less than steady-state values, which were used in a major portion of the literature. However, relative reactivity values between two different animal models and between different vascular parameters were still valid.

Our results show that hypertensive animals have a reduction in CBV_a as expected given the previous study that employed a microscope methodology.²⁸ Increased MABP (hypertension) induces the constriction of arterial vessels to maintain the same CBF level. Therefore, CBV_a can act as a key indicator for the detection of regional cerebrovascular changes in hypertension. Interestingly, we found that hypertension significantly altered the CBV_a of the hippocampus, an area involved in learning and memory. This cerebrovascular alteration may be related to the hypoxia and neural loss in the hippocampus of SHR, leading to vascular dementia.²⁹

We also observed a similar baseline CBF in SHR and WKY animals, consistent with some human studies^{1, 11, 12, 13, 14} but not others in which hypertensive subjects demonstrated a slight reduction in CBF.^{8, 9, 10} At the same time, in other studies using the SHR and WKY models, hypertensive animals had a much higher baseline CBF value than control animals under 2% isoflurane (104±23 versus 71±19 mL/100 g per minute³⁰ and 150±20 versus120±20 mL/100 g per minute.³¹) However, the fact that we used ∼1% isoflurane while the other studies used 2% may account for this difference in results.³⁰ The 2% isoflurane level, typical for surgical procedures, could impair cerebrovascular regulation during a long experiment. Indeed, Leoni et al³⁰ and colleagues observed a decrease in MABP during hypercapnia in animals anesthetized with 2% isoflurane, unlike our study. As CBF is maintained within a certain autoregulatory range independent of changes in systemic arterial pressure through the adjustment of artery and arteriole diameters,^{3, 32} CBF may not be a sensitive marker of early hypertension.

As the evoked EtCO₂ change was slightly higher for WKY than for SHR, the reactivity of WKY would be higher than that of SHR, if the magnitude of EtCO₂ change is directly related to vascular reactivity. Contrary to the expectation from EtCO₂ change, we found that hypercapnia induced greater CBF and BOLD responses in hypertensive animals, while both hypertensive and control rats showed a similar increase in ΔCBV_a (Figure 5). The CBF change is expected to be associated with the CBV_a change. Even though hypercapnia-induced ΔCBV_a is similar for the both animal models, baseline CBV_a is lower for SHR than WKY. Consequently, the relative CBV_a change is higher for SHR than WKY, leading to much higher CBF and BOLD responses for hypertensive rats. Higher vascular reactivity under the vasoconstricted hypertensive condition is similar to the increased reactivity under the vasoconstricted hypocapnic condition in normal subjects,³³ indicating that the SHR model used in our studies did not impair vascular viability.

Our observations differ from human ¹⁵O water positron emission tomography studies in which hypertensive subjects demonstrated a blunted CBF response to neural tasks as compared with normotensive age-matched 60-year-old subjects.¹¹ This reduction in CBF response may be the result of a decrease in neural activity and/or neurovascular reactivity, possibly confounded by age-related vasodilatory deterioration. Based on blood flow velocity measurements with transcranial Doppler sonography, Meada et al.¹⁵ noted that hypertensive subjects have decreased CO₂ reactivity in the middle cerebral artery. In contrast, the reactivity measured with a ¹³³Xe carotid angiogram is similar between hypertensive and normotensive subjects, either during CO₂ inhalation or during hyperventilation.³⁴ These large vessel measurements would be expected to differ from our measurements in small cortical arteries and arterioles given that these small vessels undergo hypertrophy in chronic hypertension.³⁵ The disparity between human and animal studies may cause differences. Anesthesia is needed to minimize head motion and stress levels during animal experiments, but anesthetics affect the vasculature and possibly the blood pressure. However, the mild isoflurane anesthesia (∼1%) used in this study produced no significant blood pressure changes in either SHR or WKY strains over the duration of the experiment. One possible scenario for discrepancy is that arterial vessels have not yet undergone significant remodeling in our hypertensive animals; thus, vascular reactivity is not compromised. Taking our current and previously reported data together, we hypothesize that the hyperreactivity at an early hypertensive stage without significant arterial remodeling is followed by the hyporeactivity at a later hypertensive stage due to arterial vessel wall thickening and remodeling. Vascular reactivity may be used to determine whether vascular remodeling is significant. Further longitudinal studies are necessary for examining our hypothesis.

As histopathological findings and morphologic damage induced by cerebrovascular alterations in the hypertensive animal brain have been shown to be very similar to those observed in the human brain,^{29, 36, 37, 38, 39} similar cerebrovascular morphologic and functional changes are expected in hypertensive patients. The noninvasive MRI mapping technique would be helpful to understand cerebrovascular physiology of hypertension progression and to guide intervention optimized to combat both systemic and cerebrovascular hypertensive impairments.

The authors declare no conflict of interest.