Identifying Cardiovascular Risk Factors that Impact Cerebrovascular Reactivity: an ASL MRI Study

Abstract

BACKGROUND:

To maintain cerebral blood flow (CBF), cerebral blood vessels dilate and contract in response to blood supply through cerebrovascular reactivity (CR).

PURPOSE:

Cardiovascular (CV) disease is associated with increased stroke risk, but which risk factors specifically impact CR is unknown.

STUDY TYPE:

Prospective longitudinal

SUBJECTS:

53 subjects undergoing carotid endarterectomy or stenting

FIELDSTRENGTH /SEQUENCE:

3T, 3D PCASL ASL and T1 3D FSPGR

ASSESSMENT:

We evaluated group differences in CBF changes for multiple cardiovascular risk factors in patients undergoing carotid revascularization surgery.

STATISTICAL TESTS:

PRE (baseline), POST (48-hour postop), and 6MO (6 months postop) whole Brain CBF measurements, as 129 CBF maps from 53 subjects were modeled as within-subject ANOVA. To identify CV risk factors associated with CBF change, the CBF change from PRE to POST, POST to 6MO, and PRE to 6MO were modeled as multiple linear regression with each CV risk factor as an independent variable. Statistical models were performed controlling for age on a voxel-by-voxel basis using SPM8. Significant clusters were reported if family wise error (FWE)-corrected cluster-level was p < 0.05 while the voxel-level significance threshold was set for p < 0.001.

RESULTS:

The entire group showed significant (cluster-level p < 0.001) CBF increase from PRE to POST, decrease from POST to 6MO, and no significant difference (all voxels with p > 0.001) from PRE to 6MO. Of multiple CV risk factors evaluated, only elevated systolic blood pressure (SBP, p=0.001), chronic renal insufficiency (CRI, p=0.026), and history of prior stroke (CVA, p<0.001) predicted lower increases in CBF PRE to POST. Over POST to 6MO, obesity predicted lower (p>0.001) and cholesterol greater CBF decrease (p>0.001).

DATA CONCLUSION:

The CV risk factors of higher SBP, CRI, CVA, BMI and Cholesterol may indicate altered CR, and may warrant different stroke risk mitigation and special consideration for CBF change evaluation.

INTRODUCTION

Cerebrovascular reactivity (CR) is the mechanism by which humans maintain adequate and stable cerebral blood flow (CBF) and is critical to neurological function (1). CBF is maintained between mean arterial pressures of 60 and 150 mm Hg, with pressures outside of this range noted to have detrimental effects on the brain, including edema and ischemic injury(1). Under normal physiological conditions, CBF is controlled closely by cerebral perfusion pressure (CPP) and cerebrovascular resistance (CVR). CVR is determined by brain capillary density, which is regulated by the pressure gradient between bordering arterioles and venules (1). To maintain homeostatic conditions, cerebral arteries will dilate in response to small decreases in CPP. However, when CPP decreases too drastically, autoregulation fails and declines in CPP will result in decreases in CBF (2). While insults to the cardiovascular system can result in derangements of cerebral autoregulation (3), the ways that different cardiovascular risk factors impact cerebral autoregulation is not well known.

One type of cardiovascular disease that impacts CBF is carotid artery stenosis, in which one or both cervical internal carotid arteries are narrowed, and is responsible for up to 20% of strokes in the adult population (4). It has been shown that these patients can benefit from carotid revascularization procedures, such as carotid endarterectomy (CEA) and carotid artery stenting (CAS), which restore blood supply to the brain and thereby decrease stroke risk (5). Multiple studies have demonstrated a typical pattern of increased brain CBF shortly after carotid revascularization surgery, followed by a return to baseline level CBF (6–8).

We propose that these surgeries provide a framework for evaluating the impact of different cardiovascular risk factors on cerebral vasoreactivity. Previous work has shown that reduced CBF is common in patients with cardiovascular risk factors such as hypercholesterolemia (9), elevation of fibrinogen (10), smoking, hypertension, diabetes, abdominal obesity, psychosocial factors, lack of consumption of fruits/vegetables, alcohol consumption, physical inactivity (11), age (12,13), renal insufficiency (14,15), and history of prior stroke (16). However, we hypothesize that not all these cardiovascular risk factors will impact CR equally, with some more likely to predict worse clinical outcomes or subsequent ischemic events. By assessing CBF using the noninvasive method of Arterial Spin Labeling (ASL) MRI prior to, shortly after, and several months after carotid revascularization surgery, we can test whether specific cardiovascular risk factors are associated with group differences in CBF change from baseline to 24 hours and 6 months after surgery, implying differences in CR. Not only would this assessment provide new insights into the cerebrovascular impact of cardiovascular risk factors, but it may also provide guidance for patient stroke risk stratification and management.

MATERIALS and METHODS

Under an IRB approved and HIPAA compliant protocol at the Veterans Affairs Palo Alto Health Care System, veterans scheduled to undergo carotid interventions (CEA or CAS) at the Veterans Affairs Palo Alto Health Care System (VAPAHCS) were prospectively enrolled into the study after giving written informed consent. Indications for surgical procedures included severe asymptomatic stenosis (> 80%) of carotid arteries identified on carotid duplex ultrasound or moderate to severe stenosis (>60%) with focal neurological symptoms. Subjects underwent imaging 1 month prior to surgery (PRE), within 24 hours after surgery (POST), and 6 months after surgery (6MO).

Subjects

Fifty-three veterans were recruited into the study. All subjects signed written informed consent under an IRB approved protocol and received some portion of the imaging protocol along with single side CEA or CAS. All subjects achieved the standard of care 0 % residual carotid stenosis after CEA or less than 30% residual stenosis after CAS. All subjects received ASL imaging either twice or three times (during the PRE, POST and / or 6MO sessions). Of all subjects recruited, a total of 129 ASL images were acquired. 23 subjects had PRE, POST, and 6MO ASL imaging. 42 had PRE and POST surgery ASL, 26 had PRE and 6MO ASL, and 31 had POST and 6MO ASL images. Regarding unique subjects, 23 subjects were images during all 3 sessions, an additional 19 subjects were imaged at PRE and POST, an additional 7 had only POST and 6MO, and an additional 4 subjects had only PRE and 6MO imaging. No subjects were reported to have hyperperfusion syndrome during their postsurgical course.

Recruited patients demonstrated the demographic features noted in Table 1 which were recorded from most recently available medical records at time of study enrollment, within 1 month prior to surgery.

Table 1:

Demographics and Cardiovascular Risk Factors for Enrolled Subjects

		Number of subjects (n = 53)
Group	Subgroup	No. (%)
	Any	53 (100)
	Pre-Post	42 (79)
	Pre-6MO	27 (51)
	Post-6MO	31 (58)
Surgery-Type
	CEA	29 (55)
	CAS	24 (45)
Surgery-Side
	Left	31 (58)
	Right	22 (42)
Age
	Mean	70.1
	Median	68
	Range	56–87
Gender
	M	53 (100)
	F	0 (0)
Ethnicity
	African American	2 (4)
	Asian	1 (2)
	Caucasian	41 (77)
	Hispanic	7 (13)
	Native American	1 (2)
	Pacific Islander	1 (2)
Systolic BP
	Mean	134.6
	Median	132
	Range	108–180
	Subjects >120	41 (77)
Diastolic BP
	Mean	71.4
	Median	72
	Range	49–95
	Subjects >80	10 (19)
Renal Function
	Normal	41 (77)
	Cr > 1.5 but not on HD	12 (23)
	on HD	0 (0)
Risk Factors
	Diabetes	19 (36)
	Tobacco (quit w >10 pack/yr or current)	46 (87)
	Alcohol Hx (mod or heavy use)	26 (49)
	Hypertension	52 (98)
	Hypercholesteremia	47 (89)
	Obesity	19 (36)
	Congestive Heart Failure	1 (2)
	Chronic Obstructive Pulmonary Disease	8 (15)
	Peripheral Vascular Disease (PVD)	10 (19)
	Atrial Fibrillation	2 (4)
	Left Ventricular Hypertrophy	0 (0)
	Anti-platelets	42 (79)
	Anti-coagulants	3 (6)
	Statin Usage	44 (83)
	Symptomatic (0=asymptomatic, 1=TIA/Amaurosis fugax, 2=stroke)	27 (51)
	Prior Stroke	9 (17)
	Amaurosis Fugax	4 (8)
APOE
	Single E4	7 (13)
	Double E4	1 (2)
	Unknown	15 (28)
	Double E3	25 (47)
	Single E3	5 (9)
	Double E2	0 (0)
	Single E2	5 (9)
	Single / Double E1	0 (0)

Imaging Acquisition:

Subjects received neuroimaging under a clinical protocol using a General Electric (GE) Discovery MR750 3.0 T MRI scanner and an 8-channel head coil (G.E., Waukesha, WI, USA). The protocol included a volumetric 3D fast spoiled gradient echo (FSPGR) (Echo Time [TE] = Min Full; TR = 5327 ms; flip angle = 11°; voxel size = 1.2 mm³; Inversion Time [TI] = 400 ms; Field of View [FOV] = 27 cm; number of averages = 1; matrix size = 256 × 256), and whole brain 3D GE product Pseudo-continuous Arterial Spin Labeling (PCASL) imaging (axial acquisition, FOV = 24 cm, 36 slices, with slice thickness = 4–5mm, spiral trajectory with 6–8 arms, number of averages = 3–4, Bandwidth [BW] = 62.5 kHz, Labeling Duration [LD] = 1450ms, Post-labeling Delay [PLD]= 2525 ms). As our patient population was expected to be older and have cardiovascular disease, our expectation was that these subjects would have slower cerebral vascular peak velocities(17).Post-labeling delay of 2525 ms, which is the longest delay in the product PCASL sequence, was adopted to minimize the delayed transit effect and maximize detected blood flow in this older population (see Table 1). PCASL was adopted for CBF measurements because it offers reliable noninvasive CBF measurements as validated with a PET study(18).

Image Processing:

CBF maps were quantified using the one-compartment standard kinetic model (19–21) from the PCASL difference and reference images. The individual CBF maps were then normalized to the standard Montreal Neurological Institute (MNI) template space using SPM 8 (22). Specifically, T1 FSPGR images were segmented to generate the individual gray matter images. PCASL difference images were registered to the segmented gray matter images and the corresponding registration transformation parameters were used to transform the individual CBF map to be registered to the T1 weighted image. The individual gray matter images were normalized to the gray matter template in the standard MNI space and the obtained normalization transformation parameters were used to transform the CBF maps from T1-weighted image to the standard space (Figure 1). All normalized CBF maps were left-right flipped if the side of surgery was on the right side, allowing all patients side of surgery to be considered as the left side of the images. All CBF maps were smoothed using a Gaussian kernel with full width at half maximum (FWHM) of 8 mm.

Figure 1: Arterial Spinal Label (ASL) MRI image post processing pipeline for group analyses of cardiovascular risk factors.

All ASL images were normalized into a standard space by utilizing additional information from patients structural T1 MRI images.

Statistical Analysis:

In order to examine CBF differences among three groups (PRE, POST, and 6MO) across the whole brain volume, the 129 CBF maps from all 53 subjects were modeled as within-subject ANOVA on a voxel-by-voxel basis using SPM8. The within-subject ANOVA model was used because repeated CBF maps were measured on each subject. In the ANOVA model, the independent variables were three conditions, which are binary variables representing whether the CBF map is from PRE, POST, or 6MO condition. As CBF is known to decrease with age,(23–25) this parameter was included as a confounding variable. The CBF maps were compared for different contrasts: PRE vs. POST, PRE vs. 6MO, and POST vs. 6MO. The voxel-level significance threshold was set for p < 0.001 while the family wise error (FWE)-corrected cluster-level threshold was set for p < 0.05 in order to minimize false positive findings because of multiple comparisons. Due to much higher CBF values in gray matter and increased specificity of significant regions, the voxel-level significant regions were masked with gray matter mask (the value of a voxel is 1 if its value in the prior gray matter probability map of SPM8 is greater than 0.3) before correction of cluster-level multiple comparisons.

To investigate the relationship between CBF changes and cardiovascular risk factors, post-hoc regression analysis was performed using SPM8 for each of the cardiovascular risk factors listed in Table 1. Multiple linear regression models were used to examine the relationship between the longitudinal changes in CBF maps and cardiovascular risk factors on a voxel-by-voxel basis with age included as a covariate. The longitudinal changes in CBF maps were calculated between PRE and POST, between POST and 6MO, and between PRE and 6MO, represented by CBF_POST - CBF_PRE, CBF_6MO - CBF_POST, CBF_6MO – CBF_PRE, for those subjects who have the corresponding pairs of ASL scans. Longitudinal changes of CBF maps were calculated: 42 subjects for CBF_POST - CBF_PRE group, 26 subjects for CBF_6MO - CBF_POST group, and 31 subjects for CBF_6MO – CBF_PRE group respectively. The longitudinal changes of CBF maps within each group were assessed for its association with each cardiovascular risk factor separately. The voxel-level significance threshold and the FWE-corrected cluster-level threshold were set as 0.001 and 0.05, respectively. For the regions that were detected from the voxel-level analysis, post-hoc scatter plots were generated to visualize the average association between the longitudinal CBF changes in the observed significant clusters and cardiovascular risk factors. For each subject, the longitudinal perfusion change in each observed significant cluster were calculated as its regional mean over the cluster.

For some subjects, the MRI imaging for the PRE session was not included due to patients receiving preoperative imaging at an outside institution, which did not include ASL imaging (n=7). For other subjects POST (n=4) or 6MO (n=19) session MRI imaging was not obtained due to patient refusal to undergo or not showing up for imaging. For the association analysis with a cardiovascular risk factor, we chose to remove subjects from individual analysis to avoid bias from random filling if the baseline cardiovascular risk factor being evaluated was missing for that subject. The number of subjects included in the association analysis (n), regression coefficient (β), correlation coefficient after controlling for age effects (r), and 95% confidence interval were reported for each cardiovascular risk factor.

The MNI Automated Anatomical Labeling (AAL) atlas (26) was used to generate the anatomical locations for the observed areas with significant CBF differences among PRE, POST, and 6MO timepoints, as well as the areas with significant association between CBF change and cardiovascular risk factors. Anatomical regions containing more than 1% of each cluster size and 1% of each anatomical region are reported.

RESULTS

Perfusion Changes from PRE to POST and 6MO, Controlling for Age.

All 53 subjects demonstrated a CBF increase across the entire cerebral cortex from PRE to POST CAS/CEA ASL images with the surgical side demonstrating less CBF increase relative to the contralateral side (cluster-level p<0.001, cluster size = 136,603) (Figure 2). A decrease of CBF was observed on 6MO images also across the entire cerebral cortex compared to POST (cluster-level p<0.001, cluster size = 104,802) (Figure 2). Again, this perfusion change was more pronounced on the side contralateral to surgery. No significant difference in CBF was observed between PRE and 6MO time points because no voxel had p value below the voxel-level threshold of 0.001 (voxel-level p > 0.001 for all the brain voxels).

Figure 2: ASL MRI derived CBF changes between PRE, POST, and 6MO timepoints.

Patients with right sided carotid surgery were right / left flipped so that the side of surgery is on the left (L) for all subjects during group analysis. (A) Significant (cluster-level p < 0.001) CBF increases from PRE to POST and (C) CBF decreases from POST to 6MO overlaid on a rendered brain. No significant difference in CBF was observed between PRE and 6MO time points (all voxel-level threshold p > 0.001). (B) Axial images of average CBF (i) at baseline, (ii) when subtracting POST from PRE, (iii) when subtracting PRE from 6MO, and (iv) when subtracting POST from 6MO. Color bar indicates range of CBF (row i) or CBF change in ml/100g.min (rows ii through iv). CBF increase from PRE to POST were observed across the entire brain cortex (cluster-level p < 0.001). However, CBF increased on the surgical side was less extensive (see the regions pointed by solid arrows) relative to the contralateral side from PRE to POST. CBF decrease from POST to 6MO were also observed across the entire cerebrum (cluster-level p < 0.001). However, CBF decrease on the non-surgical side was slightly less extensive (see the regions pointed by the dashed arrows) relative to the contralateral side.

ASSOCIATION OF CBF CHANGES WITH CEA or CAS, CONTROLLING FOR AGE

There was no significant difference for PRE to POST, POST to 6MO or PRE to 6MO between CEA or CAS.

ASSOCIATION OF CBF CHANGES WITH BASELINE CARDIOVASCULAR RISK FACTORS, CONTROLLING FOR AGE

PRE to POST:

ASL CBF changes from PRE to POST demonstrated significant negative association with several cardiovascular risk variables (Figure 3).

Figure 3: Cardiovascular (CV) Risk Factors Associated with PRE TO POST CBF Change Differences for (A) Systolic Blood Pressure (SBP), (B) Chronic Renal Insufficiency (CRI), and (C) Prior Stroke (CVA), overlaid on standard structural images.

Color bar indicates range of t scores for association of PRE to POST CBF. The significant regions where greater CBF increases from PRE to POST were associated with (A) lower SBP values, (B) no chronic renal insufficiency, and (C) no prior stroke at baseline. Crossing blue lines indicate corresponding region across planes. The side of surgery is on the left (L).

Systolic blood pressure (SBP, continuous variable, n=42):

These patients demonstrated less CBF increase from PRE to POST ipsilateral and contralateral to the side of surgery relative to patients without this risk factor (Figure 3A). Cluster statistics, including mean beta, standard error, peak-t MNI coordinates, and anatomical regions, of all significant clusters (clusters 1–3) are shown in Table 2.

Table 2.

Clusters with significant association between longitudinal perfusion changes and cardiovascular risk factors

Risk Factor	Cluster	N Voxels	mean beta±std error	Peak-t MNI coordinates	Anatomical Locations	% Cluster	% Region
Systolic Blood Pressure (PRE to POST)	1	1958	−0.77 ± 0.018	−34, −10, 66	Frontal Lobe  Precentral_L  Frontal_Mid_L  Frontal_Sup_L Parietal Lobe  Postcentral_L	54.70 17.93 6.03 20.17	30.37 7.22 3.28 10.05
Systolic Blood Pressure (PRE to POST)	2	4728	−0.89 ± 0.022	−26, −98, −4	Occipital Lobe  Occipital_Mid_L  Occipital_Inf_L  Fusiform_L Temporal Lobe  Temporal_Mid_L Parietal Lobe  Angular_L  Parietal_Inf_L	45.85 10.68 1.76 17.81 9.12 6.62	66.30 53.67 3.59 17.04 36.74 12.79
Systolic Blood Pressure (PRE to POST)	3	1915	−0.8 ± 0.013	40, −58, 18	Occipital Lobe  Occipital_Mid_R  Calcarine_R  Occipital_Inf_R  Occipital_Sup_R  Cuneus_R  Fusiform_R  Lingual_R Temporal Lobe  Temporal_Mid_R Parietal Lobe  Angular_R	31.49 12.17 8.83 5.07 4.49 2.66 1.41 19.53 5.27	28.74 12.52 17.09 6.86 6.04 2.03 1.17 8.48 5.76
Chronic Renal Insufficiency (PRE to POST)	4	2670	−26.05± 0.99	−24, 62, −16	Frontal Lobe  Frontal_Inf_Orb_L  Frontal_Mid_L  Frontal_Inf_Tri_L  Frontal_Mid_Orb_L  Frontal_Sup_Orb_L  Frontal_Inf_Oper_L Temporal Lobe  Temporal_Pole_Mid_L  Temporal_Pole_Sup_L  Temporal_Inf_L Occipital Lobe  Fusiform_L Insular  Insular_L	22.85 13.82 12.36 8.91 2.81 1.24 11.20 5.24 2.17 5.32 1.91	36.09 7.59 13.05 26.80 7.79 3.18 39.60 10.89 1.81 6.15 2.74
Chronic Renal Insufficiency (PRE to POST)	5	1170	−27.09± 1.06	−14, 26, −2	Basal Ganglia  Caudate_L  Caudate_R Frontal Lobe  Olfactory_L  Olfactory_R  Rectus_L Limbic System  Cingulum_Ant_L	18.12 2.22 13.25 5.13 2.05 15.47	22.04 2.62 55.36 20.76 2.82 12.93
Prior Stroke (PRE to POST)	6	5516	−45.27± 0.97	16, 6, −6	Temporal Lobe  Temporal_Sup_R  Temporal_Pole_Sup_R  Temporal_Inf_R  Heschl_R  Temporal_Mid_R Insular  Insular_R Basal Ganglia  Caudate_R  Putamen_R  Pallidum_R  Thalamus_R Frontal Lobe  Rolandic_Oper_R  Olfactory_R  Frontal_Inf_Orb_R  Frontal_Inf_Oper_R  Rectus_R Limbic System  Hippocampus_R  Parahippacampal_R Occipital Lobe  Lingual_R	9.05 4.01 2.23 1.83 1.56 8.45 7.70 7.45 4.24 2.56 4.51 2.52 2.01 1.65 1.38 2.77 2.30 1.14	15.89 16.52 3.46 40.56 1.95 26.33 42.76 38.63 83.57 13.34 18.71 48.10 6.50 6.50 10.20 16.17 11.22 2.74
Obesity (POST to 6MO)	7	2772	−36.04±1.22	62, −38, 42	Parietal Lobe  SupraMarginal_R  Parietal_Inf_R  Angular_R  Postcentral_R  Parietal_Sup_R Temporal Lobe  Temporal_Sup_R   Temporal_Mid_R Frontal Lobe  Rolandic_Oper_R  Precentral_R	33.77 23.48 13.64 13.46 1.26 4.87 1.98 2.45 2.42	47.42 48.40 21.58 9.76 1.58 4.30 1.25 5.11 1.98
Obesity (POST to 6MO)	8	3741	−34.17±1.23	−8, −22, 62	Parietal Lobe  Precuneus_L  Paracentral_Lobule_L  Parietal_Sup_L  Postcentral_L Limbic System  Cingulum_Mid_L Frontal Lobe  Supp_Motor_Area_L  Frontal_Sup_L  Frontal_Mid_L	21.38 16.39 12.40 5.72 13.63 12.86 6.55 2.97	22.68 45.44 22.47 5.50 26.28 22.40 6.81 2.28
Cholesterol (POST to 6MO)	9	1301	0.38±0.009	−24, −12, 66	Frontal Lobe  Precentral_L  Supp_Motor_Area_L  Frontal_Sup_L  Supp_Motor_Area_R Parietal Lobe  Postcentral_L  Paracentral_Lobule_L Limbic System  Cingulum_Mid_L	33.21 20.91 7.15 4.77 10.53 8.46 7.30	12.25 12.67 2.58 2.61 3.52 8.15 4.89
Cholesterol (POST to 6MO)	10	1018	0.37±0.01	−20.−32,14	Limbic System  Hippocampus_L  Parahippacampal_L Basal Ganglia  Thalamus_L Occipital Lobe  Fusiform_L Temporal Lobe  Temporal_Inf_L Cerebellum  Cerebellum_6_L	23.38 3.73 14.44 14.44 9.63 5.30	25.54 3.89 13.36 6.36 3.06 3.19

%Cluster indicates the percentage of each cluster that falls within the defined region, %Region indicates the percentage of each defined region that falls within the cluster. The listed anatomical regions are either “%Cluster” > 1% and “%Region” > 1%. Longitudinal perfusion changes from PRE to POST were significantly associated with systolic blood pressure (clusters 1–3), with chronic renal insufficiency (clusters 4–5), and history of prior stroke (cluster 6). Longitudinal perfusion changes from POST to 6MO were significantly associated with obesity (clusters 7–8) and with cholesterol levels (clusters 9–10).

Chronic renal insufficiency (CRI, 0=normal Cr level, 1=Cr>1.5mg/dl, 2=on dialysis/kidney transplant, n=42, no patients had CRI=2): Patients with worse renal function (as indicated by higher value for this variable) demonstrated less CBF increase from PRE to POST ipsilateral to the side of surgery relative to patients with better renal function (as indicated by a lower value of this variable) (Figure 3B). Cluster statistics of the significant clusters (clusters 4–5) are also listed in Table 2.

History of prior stroke (CVA, n=42):

Contralateral to surgery, patients exhibited less CBF increase from PRE to POST (Figure 3C).

Other CV Risk Factors (Table 1):

CBF changes from PRE to POST did not show significant association with any other examined variables (Table 1). Cluster statistics of the significant cluster (cluster 6) are also listed in Table 2.

POST to 6MO:

ASL CBF changes from POST to 6MO demonstrated significant negative correlation with obesity, and positive correlation with cholesterol level (Figure 4). Cluster statistics of the significant clusters (clusters 7–8) are also listed in Table 2.

Figure 4: CV risk factors associated with POST to 6MO CBF change differences.

Color bar indicates range of t scores for association of POST to 6MO CBF change with (A) obesity and (B) cholesterol. The side of surgery is on the left (L). Crossing blue lines indicate corresponding region across planes. (A) Regions where CBF decrease between POST and 6MO are significantly less in patients with obesity at baseline (obesity is 1 if BMI >30 and 0 otherwise) relative to subjects without obesity. (B) Regions where CBF decrease between POST and 6MO are significantly greater in patients with higher cholesterol level at baseline relative to patients without higher cholesterol at baseline.

Obesity (0=no, 1=BMI>30, n=31): Obesity prior to surgery was significantly associated with a lower decrease in ASL CBF from POST to 6MO in regions both ipsilateral and contralateral to surgery (Figure 4A). Cluster statistics of the significant clusters (clusters 7–8) are also listed in Table 2.

Cholesterol (continuous variable, n=29, baseline cholesterol unavailable for 2 patients with available ASL MRI): Patients with higher cholesterol (Figure 4B) prior to surgery demonstrated a larger decrease in ASL CBF from POST to 6MO in regions ipsilateral to surgery in the regions listed in Table 2.

Other CV Risk Factors (Table 1):

ASL CBF changes from POST to 6MO did not show significant association with any other examined variables (Table 1).

PRE to 6MO:

No baseline variables were noted to correlate with the change in CBF from PRE to 6MO. Similarly, ASL CBF changes from PRE to 6MO did not show significant association (all voxels with p > 0.001) with any examined CV risk factors (Table 1).

Largest Significant Cluster Correlation Analysis:

The scatter plots for the association of the longitudinal CBF changes from PRE to POST (CBF_POST - CBF_PRE) with systolic blood pressure at cluster 2 (n = 42, β = −0.87, r = −0.44, p = 0.0036, 95% confidence interval [−1.45, −0.31]), with chronic renal inefficiency at cluster 4 (n = 42, β = −26.10, r = −0.51 p = 0.0007, 95% confidence interval [−40.49, −11.71]), and with history of prior stroke at cluster 6 (n = 42, β = −45.27, r = −0.55, p = 0.0002, 95% confidence interval [−67.44, −23.10]) are shown in Fig. 5a–5c. The scatter plots for the association of the longitudinal CBF changes from POST to 6MO (CBF_6MO – CBF_POST) with obesity at cluster 8 (n = 31, β = −36.04, r = −0.56, p = 0.0013, 95% confidence interval [−56.68, −15.40]) and with cholesterol level at cluster 9 (n = 29, β = 0.38, r = 0.63, p = 0.0004, 95% confidence interval [0.19, 0.57]) are shown in Fig. 5d–5e. The locations of the clusters (which showed the significant association from the voxel-based analysis) are shown in Table 2. For each cardiovascular risk factor, only the largest cluster was used to derive the scatter plots because all the other clusters showed similar results. For the completeness of reporting, the significance levels for the association of the longitudinal CBF changes from PRE to POST with systolic blood pressure at cluster 1, 2, and 3 were shown as 0.0043, 0.0036, and 0.0044 respectively, with chronic renal inefficiency at the cluster 4 and 5 as 0.0007 and 0.0017 respectively, with history of prior stroke at the cluster 6 as 0.0002. The significance levels for the association of the longitudinal CBF changes from POST to 6MO with obesity at the cluster 7 and 8 were shown as 0.0013 and 0.0016 respectively, with cholesterol level at the cluster 9 and 10 as 0.0004 and 0.0005 respectively.

Figure 5: Association of longitudinal CBF changes with CV risk factors at the corresponding largest significant cluster from the voxel-level analysis.

Significant association of the longitudinal CBF changes from PRE to POST (CBF_POST - CBF_PRE) with (A) systolic blood pressure at cluster 2 (n=42, β= −0.87, r = −0.44, p = 0.0036, 95% confidence interval [−1.45, −0.31]), (B) chronic renal inefficiency at cluster 4 (n = 42, β= −26.10, r = −0.51 p = 0.0007, 95% confidence interval [−40.49, −11.71]), and (C) history of prior stroke at cluster 6 (n = 42, β = −45.27, r = −0.55, p = 0.0002, 95% confidence interval [−67.44, −23.10]), and significant association of the longitudinal CBF changes from POST to 6MO (CBF_6MO - CBF_POST) with (D) obesity at cluster 8 (n = 31, β = −36.04, r = −0.56, p = 0.0013, 95% confidence interval [−56.68, −15.40]), (E) cholesterol level at cluster 9 (n = 29, β = 0.38, r = 0.63, p = 0.0004, 95% confidence interval [0.19, 0.57]). The AAL labels of the clusters are shown in Table 2.

DISCUSSION

We demonstrated group longitudinal differences in CBF change associated with individual cardiovascular risk factors. Specifically, we found that when faced with the increased blood flow after surgical reopening of a chronically narrowed carotid artery, either via CEA or CAS, the blood vessels in patients with baseline CRI, history of prior stroke, higher SBP, obesity (BMI >30), or higher cholesterol all demonstrated significant differences in change in CBF from subjects without those risk factors. While it is conceivable that a difference in residual stenosis after surgery may explain these findings, the fact that all of our subjects were reported to achieve the clinical standard of 0% residual stenosis after CEA and <30% after CAS and that our study found no significant difference in CBF change for any time interval (PRE to POST, POST to 6MO or PRE to 6MO) between CEA and CAS patients suggests that residual stenosis did not influence our results. This further suggests that the differences in CBF changes noted are likely related to differences in physiologic response of cerebral arteries in patients with these specific risk factors.

The lower increase in CBF seen with CRI and higher SBP from PRE to POST, and greater decrease in CBF from POST to 6MO seen with higher cholesterol levels are consistent with endothelial molecular dysfunction known to be associated with these conditions. Nitric oxide (NO) is a vasodilator and has been shown to maintain homeostatic levels of CBF (27). Mice with chronic renal failure have been shown to express high levels of the endothelial nitric oxide synthase inhibitor ADMA, associated with lower NO levels (28,29). Uremic toxins in hemodialysis patients has been associated with endothelial cell apoptosis, interfering with NO production (30). While our study did not include patients on dialysis, it is likely that patients with impaired renal function had more circulating uremic toxins than patients with normal renal function. Superoxide has been shown to decrease NO availability, and is noted to be increased in hypertensive patients (31). Additionally, animal models have shown hypertension to be associated with decreased pial artery density and diameter (32), increasing CVR and decreasing CBF. Similarly, increased superoxide release has been demonstrated in mice with hyperlipidemia, as seen in elevated cholesterol levels (33). High cholesterol has also been shown to increase vessel constriction via smooth muscle contraction by increased cytosolic calcium concentration and activation of the Rho/Rho-kinase pathway (34).

Our study also demonstrated lower increase in CBF after carotid revascularization to be associated with history of prior stroke. Patients with history of lacunar strokes, and non-hypertensive patients after atherothrombotic strokes have been noted to show decreased CBF up to 40 months after stroke (35), with others suggesting reduced CBF up to 6 years post stroke, although there were large discrepancies in flow values in this study (36). A different study found that despite spontaneous reperfusion immediately after stroke, CBF was not maintained in a 150 day follow up (37). It may be possible that the mechanisms contributing to lower CBF after stroke in these other study populations may be contributing to lower increase in CBF after blood flow is increased by carotid revascularization.

Our finding of smaller decrease in CBF from POST to 6MO in obese patients compared to controls has also not been previously demonstrated. The fact that we did not see a significant difference in CBF from PRE to 6MO, or PRE to POST suggests that this finding from POST to 6MO does not mean there was a significant difference in residual CBF at 6MO in obese patients. This result conflicts with other studies utilizing SPECT and ASL imaging techniques, which observed decreased CBF in patients with high BMI (38). As animal models have shown endothelial cells to dilate in response to circulating insulin (39), which is known to be higher in obese patients relative to their non-obese counterparts (40), it may be possible that this difference in ability to decrease CBF may reflect an impaired ability to constrict vessels after carotid revascularization.

Our results of overall group longitudinal CBF changes were consistent with earlier work, demonstrating increases in perfusion shortly after CEA / CAS, with return to baseline CBF months later (6–8,41). We also found that the CBF increase shortly after surgery was greater in the cerebral hemisphere contralateral to the side of surgery, corroborated by some studies (42), while others found no statistically significant differences in CBF restoration in the ipsilateral and contralateral hemispheres (6,43). The source of this disagreement may be related to follow-up time variations in the post-operative CBF evaluation, with two of these studies performing this measurement 1 month after surgery (6,42), and the other at 2 hours after surgery (43). Regarding a possible mechanism for this finding, we hypothesize that the cerebral hemisphere for the side of surgery may have vessels already near maximum dilatation to compensate for the chronic reduced blood flow caused by the carotid stenosis on that side. Consequently, restoring blood flow by CEA or CAS to that hemisphere may not show as great an increase in CBF than the contralateral side.

Limitations of this work include the lack of female subjects and a variable number of subjects in the PRE to POST, POST to 6MO, and PRE to 6MO analysis groups. Additionally, the single label time ASL techniques used in this study may not have captured some CBF in patients with lower or higher cardiac output. Future studies employing multi-delay ASL perfusion imaging may better characterize CBF across these groups, providing additional insights into differences in perfusion between groups. Also, while this study included patients with at least 60% carotid stenosis on the side of surgical intervention, including more granular carotid stenosis measurements on the ipsilateral and contralateral sides may provide more nuanced insights into differences in cerebrovascular reactivity that correlate with baseline risk factors in future studies. Additionally, another limitation is that evaluation of the nonsurgical side carotid stenosis from PRE through 6MO timepoints was not performed. As this study evaluated change in CBF between time points, there is no reason to expect contralateral vessel stenosis progression between PRE and POST time points. However, it is possible the extent of contralateral carotid artery stenosis may have progressed by the 6MO time point.

In conclusion, the CV risk factors of elevated systolic blood pressure, chronic renal insufficiency, history of prior stroke, obesity, and cholesterol level are associated with different responses to blood flow changes than other risk factors and may warrant more aggressive management for stroke risk mitigation. Further study of cerebral blood flow in patients with these risk factors may provide new insights into derangement of CR, especially examining the physiologic mechanisms described above. Additionally, the presence of these specific risk factors may impact evaluations of patients where models of hemodynamic response are implicit in interpretation (e.g. functional MRI studies).

Abbreviations

CR

Cerebrovascular reactivity