Blood Pressure Variability and Cerebral Perfusion Decline: A Post Hoc Analysis of the SPRINT MIND Trial

Abstract

Background

Blood pressure variability (BPV) is predictive of cerebrovascular disease and dementia, possibly though cerebral hypoperfusion. Higher BPV is associated with cerebral blood flow (CBF) decline in observational cohorts, but relationships in samples with strictly controlled blood pressure remain understudied. We investigated whether BPV relates to change in CBF in the context of intensive versus standard antihypertensive treatment.

Methods and Results

In this post hoc analysis of the SPRINT MIND (Systolic Blood Pressure Intervention Trial–Memory and Cognition in Decreased Hypertension) trial, 289 participants (mean, 67.6 [7.6 SD] years, 38.8% women) underwent 4 blood pressure measurements over a 9‐month period after treatment randomization (intensive versus standard) and pseudo‐continuous arterial spin labeling magnetic resonance imaging at baseline and ≈4‐year follow‐up. BPV was calculated as tertiles of variability independent of mean. CBF was determined for whole brain, gray matter, white matter, hippocampus, parahippocampal gyrus, and entorhinal cortex. Linear mixed models examined relationships between BPV and change in CBF under intensive versus standard antihypertensive treatment. Higher BPV in the standard treatment group was associated with CBF decline in all regions (ß comparing the first versus third tertiles of BPV in whole brain: −0.09 [95% CI, −0.17 to −0.01]; P=0.03), especially in medial temporal regions. In the intensive treatment group, elevated BPV was related to CBF decline only in the hippocampus (ß, −0.10 [95% CI, −0.18, −0.01]; P=0.03).

Conclusions

Elevated BPV is associated with CBF decline, especially under standard blood pressure–lowering strategies. Relationships were particularly robust in medial temporal regions, consistent with prior work using observational cohorts. Findings highlight the possibility that BPV remains a risk for CBF decline even in individuals with strictly controlled mean blood pressure levels.

Registration

URL: http://clinicaltrials.gov. Identifier: NCT01206062.

Nonstandard Abbreviations and Acronyms

BPV

blood pressure variability

CBF

cerebral blood flow

EC

entorhinal cortex

pCASL

pseudo‐continuous arterial spin labeling

PHG

parahippocampal gyrus

SPRINT

Systolic Blood Pressure Intervention Trial

SPRINT MIND

Systolic Blood Pressure Intervention Trial–Memory and Cognition in Decreased Hypertension

WM

white matter

CLINICAL PERSPECTIVE.

What Is New?

• In this post hoc analysis of the SPRINT MIND (Systolic Blood Pressure Intervention Trial–Memory and Cognition in Decreased Hypertension) clinical trial, higher blood pressure variability, independent of mean blood pressure levels, was associated with cerebral perfusion decline in brain regions vulnerable to Alzheimer disease and cerebrovascular disease, especially under standard blood pressure lowering.

What Are the Clinical Implications?

• Blood pressure variability remains a risk for cerebral perfusion decline even in individuals with strictly controlled mean blood pressure levels, suggesting blood pressure variability may be an understudied vascular factor associated with dementia risk.

Blood pressure (BP) remains one of the most accessible therapeutic targets for prevention of cerebrovascular disease and dementia. ¹ , ² , ³ Results from the SPRINT (Systolic Blood Pressure Intervention Trial) trial highlight this approach and suggest that intensive BP lowering (<120 mm Hg systolic BP target), when compared with standard BP lowering (<140 mm Hg systolic BP target), is associated with reduced risk for cardiovascular events, ⁴ slower progression of white matter (WM) hyperintensities, ⁵ and reduced risk for mild cognitive impairment and a composite of mild cognitive impairment and probable dementia. ³ Blood pressure variability (BPV) may represent a distinct aspect of BP control that is increasingly associated with poor brain health outcomes, independent of and oftentimes above and beyond traditionally studied mean BP levels. ⁶ A growing number of studies suggest that elevated BPV is predictive of stroke, progression of cerebrovascular disease, cognitive decline, and dementia, including Alzheimer disease. ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ It has been hypothesized that BPV may be related to these outcomes through links with cerebral hypoperfusion. ⁶ , ⁸ , ⁹ Consistently, 2 recent studies found that elevated BPV was associated with cerebral hypoperfusion ¹¹ and cerebral blood flow (CBF) decline ¹² in older adults without major neurocognitive impairment, especially in medial temporal regions vulnerable to vascular insult and Alzheimer disease. ¹³ , ¹⁴ , ¹⁵ However, these studies—and the majority of BPV studies—relied on observational study designs following cohorts with varying degrees of BP control in terms of BP medication use/initiation/discontinuation/adherence, dosing, type of agent(s), and level of monitoring. ⁶ , ⁸ , ⁹ , ¹⁶ Less is known about relationships between BPV and CBF in cohorts with strictly and uniformly controlled BP, where BPV effects may be better studied independent of mean BP levels. Furthermore, studying the relationship between BPV and CBF in the context of intensive versus standard BP lowering may inform our understanding of intensive BP lowering effects on CBF, given a recent SPRINT MIND (Systolic Blood Pressure Intervention Trial–Memory and Cognition in Decreased Hypertension) study found that intensive BP lowering was associated with increased CBF. ¹⁷ Thus, we conducted a post hoc analysis of the SPRINT MIND trial to examine the longitudinal relationship between BPV and CBF change under intensive versus standard BP lowering.

Methods

Participants

Data were obtained from the SPRINT MIND trial, a publicly available deidentified data set from the National Heart, Lung, and Blood Institute that has been described in detailed elsewhere. ¹⁸ , ¹⁹ The present investigation was a post hoc analysis of these data. SPRINT was a multicenter randomized controlled study cohort trial in the United States and Puerto Rico conducted between November 2010 and March 2013 investigating whether intensive BP lowering could reduce cardiovascular risk when compared with standard BP treatment. Participants were recruited from the local community and a variety of clinical settings, such as primary care, nephrology, and geriatrics. Participants were aged ≥50 years, had hypertension (systolic BP 130–180 mm Hg at screening), and had at least 1 risk factor for cardiovascular disease (history of cardiovascular disease, chronic kidney disease [estimated glomerular filtration rate <60 mL/min/1.73 m²], 10‐year Framingham cardiovascular disease risk ≥15%, aged ≥75 years). Participants were excluded for history of stroke, diabetes, or heart failure; residing in a nursing home; diagnosis of dementia on the basis of medical record review; or receiving medication primarily used to treat dementia. Participants were randomized 1:1 to either standard treatment (<140 mm Hg systolic BP target) or intensive treatment (<120 mm Hg systolic BP target). SPRINT was approved by an institutional review board at each site. All participants provided their informed consent before treatment randomization.

Measures

BP Assessment

The SPRINT BP protocol has been described in detail previously. ³ , ⁴ , ⁵ , ¹⁷ , ¹⁸ Briefly, BP was measured using an automated BP device (Professional Digital Blood Pressure Monitor, model 907XL [Omron Healthcare, Kyoto, Japan]) at baseline; 1‐, 2‐, and 3‐month follow‐up; and then every 3 months for up to 6 years’ follow‐up. BP values from each visit were recorded as the average of 3 serial seated BP measurements after a 5‐minute period of rest. Participants were prohibited from completing questionnaires, talking, or texting during the rest period and BP collection. At 3‐month follow‐up, BP levels reached a relatively stable plateau in both treatment groups. ²⁰ Therefore, BPV was determined from BP collected at 3‐, 6‐, 9‐, and 12‐month follow‐up to minimize the effect of initial BP fluctuation in the intensive treatment group, consistent with other BPV studies using the SPRINT data set. ²⁰ , ²¹ , ²² We calculated intraindividual BPV from the 4 BP measurements (eg, from 3‐, 6‐, 9‐, and 12‐month follow‐up visits) as variability independent of mean, a newer index of BPV uncorrelated with mean BP across visits. ²³ , ²⁴ Variability independent of mean was calculated as variability independent of mean=SD/mean ^x , where the power x was derived from nonlinear curve fitting of BP SD against mean BP using the nls package in R Project (R Foundation for Statistical Computing, Vienna, Austria), as previously described. ¹¹ , ¹² , ²³ , ²⁵ , ²⁶ We conducted a bivariate correlation to confirm that BPV variability independent of mean was not significantly correlated with mean BP (r=0.03, P=0.18). BPV values were then divided into tertiles and used in all analyses. Mean BP was calculated from BP values collected over the same 9‐month period as BPV (eg, 3‐, 6‐, 9‐, and 12‐month follow‐up).

CBF Assessment

Participants underwent 3‐Tesla T1‐weighted structural magnetic resonance imaging (MRI) and pseudo‐continuous arterial spin labeling (pCASL)‐MRI at study baseline and planned 48‐month follow‐up and were scanned on the same scanner at each time point (n=7 total MRI sites). However, due to early stoppage of the SPRINT trial due to cardiovascular benefit finding in the intensive treatment group, ⁴ the majority (273 [94.5%]) of the follow‐up MRI scans were collected after study closeout when participants transitioned to having their primary care physician manage their hypertension (and the study still provided antihypertensive medications). ¹⁷ The median time interval between baseline MRI and follow‐up MRI was 1451 days (interquartile range, 121 days). T1‐weighted magnetization‐prepared rapid acquisition with gradient echo images were collected using the following parameters ¹⁷ : 1900 ms repetition time; 2.89 ms echo time; 250 mm field of view; 176 slices; 1 mm isotropic native resolution. As previously described, ¹⁷ pCASL‐MRI data were obtained with 1.5/1.5‐second labeling/postlabeling delay at 90 mm below the center of the imaging volume (Siemens scanners) or 30.5 mm below the inferior slice (Phillips scanners). Forty tag‐control image pairs were collected using the following parameters: 4000 ms repetition time; 11 ms echo time; 220 mm field of view; 3.4×3.4×5 mm³ voxel size; 20% distance factor; 20 slices. pCASL‐MRI processing has been described in detail elsewhere. ¹⁷ Briefly, processing steps included motion correction of raw tag‐control time series data, CBF quantification using a single compartment model, and denoising. Only pCASL‐MRI images with whole brain CBF above an automated quality evaluation index threshold of 0.35 ¹⁷ , ²⁷ were included in the present analyses. CBF maps were then coregistered to T1 images to extract CBF from gray matter (GM) and WM regions of interest, consistent with a recent study of CBF in the SPRINT MIND trial. ¹⁷ To reduce the likelihood of type 1 error and take a hypothesis‐driven approach building off previously reported findings linking BPV to regional cerebral hypoperfusion ¹¹ and CBF decline ¹² in observational cohort studies of BPV, we also assessed CBF in the hippocampus (HC), parahippocampal gyrus (PHG), and entorhinal cortex (EC). These regions are also vulnerable to Alzheimer disease. ¹⁴ , ²⁸ , ²⁹ CBF values from each hemisphere were averaged and used in all analyses.

Other Measurements

Baseline clinical evaluation determined the following variables ¹⁸ : education (self‐report; < college/other versus college versus graduate), race and ethnicity (self‐report; Black versus White versus Hispanic versus Other), body mass index (BMI), Framingham Risk Score, number of antihypertensive medications used.

Data Availability Statement

All data are available through the SPRINT group.

Statistical Analysis

Linear mixed models investigated the effect of BPV × time on CBF levels in each of the regions of interest (ie, whole brain, GM, WM, HC, PHG, EC) separately. CBF values in the HC, PHG, and EC were divided by whole brain CBF to reflect region‐specific CBF. Random intercepts for participant and MRI facility were included in the models. Time was calculated as days since treatment randomization. We focused our analyses on systolic BPV given the target of systolic lowering of the SPRINT trial, ¹⁸ consistent with other BPV studies using the SPRINT data set. ²⁰ , ²² , ³⁰ Models were stratified by treatment group (intensive versus standard) and included the following covariates: age, sex, race and ethnicity, education, history of cardiovascular disease, ¹⁸ and mean BP over the same 9‐month period BPV was determined. Sensitivity analyses additionally controlled for (1) number of antihypertensive medications used, (2) body mass index, (3) Framingham Risk Score, and (4) whether the follow‐up pCASL‐MRI scan was collected before versus after study closeout (Tables S1 and S2). Multiple comparison corrections using the false discovery rate method ³¹ were set at P<0.05. All analyses were 2‐tailed, with significance set at P<0.05. All analyses were carried out in R. ³²

Results

A total sample of 289 participants had valid pCASL‐MRI scans at baseline and follow‐up and valid BPV calculated from BP measurements at 3‐, 6‐, 9‐, and 12‐month follow‐up. There were 156 participants in the intensive treatment group (mean, 68.3 [7.5 SD] years, 41.0% women, 51.9% less than college/other education; mean, 23.9 [4.0 SD] Montreal Cognitive Assessment) and 133 participants in the standard treatment group (mean 66.8 [7.7 SD] years, 36.1% women, 54.9% less than college/other education; mean, 24.0 [4.2 SD] Montreal Cognitive Assessment). Mean BP was significantly different between the treatment groups (119.0 [8.0] mm Hg intensive versus 134.5 [7.3] standard; between‐group difference, 15.5 mm Hg; P<0.001) while BPV was not (9.8 [6.1] mm Hg intensive versus 9.4 [5.6] standard; between‐group difference, 0.4 mm Hg; P=0.56). Demographic information is summarized in Table 1.

Table 1.

Baseline Clinical and Demographic Information

	Intensive (n=156)	Standard (n=133)	F or x 2	P value
Age, y	68.3 (7.5)	66.8 (7.7)	2.98	0.09
Sex, female, n (%)	64 (41.0)	48 (36.1)	0.54	0.46
Race and ethnicity, n (%)			1.64	0.65
Black	45 (28.9)	40 (30.1)
Hispanic	8 (5.1)	9 (6.8)
White	101 (64.7)	80 (60.2)
Other	2 (1.3)	4 (3.0)
Education, n (%)			0.39	0.82
Less than college/other	81 (51.9)	73 (54.9)
College	26 (16.7)	19 (14.3)
Graduate school	49 (31.4)	41 (30.8)
MoCA	23.9 (4.0)	24.0 (4.2)	0.91	0.34
Body mass index, kg/m2	29.4 (5.3)	29.3 (6.0)	0.02	0.90
FRS 10‐year risk score	18.6 (10.5)	18.1 (9.4)	0.18	0.68
Medical history, n (%)
Cardiovascular disease	18 (11.2)	16 (12.0)	0.00	0.99
Hypertension	146 (93.6)	124 (93.2)	0.00	0.99
Medication use, n (%)
Antihypertensive agents	145 (93.0)	124 (93.2)	0.00	0.99
No. antihypertensive agents used	1.8 (1.0)	1.9 (1.0)	0.28	0.60
Systolic BP, mm Hg
Baseline	135.7 (18.1)	137.8 (15.5)	1.10	0.30
Mean*	119.0 (8.0)	134.5 (7.3)	288.3	<0.001
VIM*	9.8 (6.1)	9.4 (5.6)	0.35	0.56
Diastolic BP, mm Hg
Baseline	75.6 (10.9)	77.8 (11.2)	2.89	0.09
Mean*	67.0 (7.7)	76.0 (8.7)	86.92	<0.001
VIM*	4.9 (1.0)	4.8 (0.9)	1.10	0.30
CBF, mL/100 g/min
Whole brain	38.6 (9.5)	38.2 (9.0)	0.31	0.58
GM	50.2 (11.6)	49.5 (10.8)	0.43	0.511
WM	20.1 (6.5)	19.7 (6.3)	0.37	0.55
HC	44.1 (9.8)	43.7 (9.2)	0.30	0.58
PHG	41.2 (10.5)	39.7 (10.1)	2.48	0.12
EC	35.2 (10.4)	34.0 (10.6)	1.00	0.32

Means and SDs shown unless otherwise indicated.

BP indicates blood pressure; CBF, cerebral blood flow; EC, entorhinal cortex; FRS, Framingham Risk Score; GM, gray matter; HC, hippocampus; MoCA, Montreal Cognitive Assessment; PHG, parahippocampal gyrus; VIM, variability independent of mean; and WM, white matter.

^*Mean and BPV values determined from BP measurements collected at 3‐, 6‐, 9‐, and 12‐month follow‐up, after BP levels reached a relatively stable plateau in both treatment groups at 3‐month follow‐up.

BP Variability

As shown in Table 2 and in the Figure, higher BPV in the standard treatment group was associated with CBF decline in all regions (P for trend comparing 1st versus 3rd tertiles of BPV≤0.001–0.04). Relationships in medial temporal regions were particularly robust (ß comparing first versus third tertiles of BPV: HC: ß, −0.16 [95% CI, −0.24, −0.08], P<0.001; PHG: ß, −0.15 [95% CI, −0.24, −0.07], P<0.001; EC: ß, −0.12 [95% CI, −0.23, −0.02], P=0.03).

Table 2.

Linear Mixed‐Model Estimates (ß [95% CI]) of the 2‐Way Interaction of BPV × Time on Change in CBF Levels in the Standard Treatment Group

Region of interest	Tertile 1	Tertile 2	Tertile 3	P value for trend
Whole brain	Ref	−0.03 [−0.11 to 0.05]	−0.09 [−0.17 to −0.01]	0.03
GM	Ref	0.02 [−0.06 to 0.10]	−0.09 [−0.17 to −0.01]	0.03
WM	Ref	−0.14 [−0.23 to −0.07]	−0.09 [−0.18 to −0.01]	0.04
HC	Ref	−0.09 [−0.17 to −0.01]	−0.16 [−0.24 to −0.08]	<0.001
PHG	Ref	−0.04 [−0.13 to 0.05]	−0.15 [−0.24 to −0.07]	<0.001
EC	Ref	0.03 [−0.08 to 0.14]	−0.12 [−0.23 to −0.02]	0.03

Models adjusted for age, sex, race and ethnicity, education, history of cardiovascular disease, and mean BP over the same 9‐month period BPV was determined.

BP indicates blood pressure; BPV, blood pressure variability; CBF, cerebral blood flow; EC, entorhinal cortex; GM, gray matter; HC, hippocampus; PHG, parahippocampal gyrus; and WM, white matter.

Figure . Elevated BPV is associated with CBF decline in several regions in the standard treatment group.

Conditional effects of BPV by time on change in CBF (mL/100 g/min) in (A) WB, (B) GM, (C) WM, (D) HC, (E) PHG, and (F) EC in the standard treatment group. Lines represent rate of change for each tertile of BPV. Lines are shaded with 95% CI for each BPV tertile. Models adjusted for age, sex, race and ethnicity, education, history of cardiovascular disease, and mean BP. BPV indicates blood pressure variability; CBF, cerebral blood flow; EC, entorhinal cortex; GM, gray matter; HC, hippocampus; PHG, parahippocampal gyrus; WB, whole brain; and WM, white matter.

As shown in Table 3, elevated BPV in the intensive treatment group was associated with decline in CBF in the HC only (ß comparing first versus 3rd tertiles of BPV: −0.10 [95% CI, −0.18, −0.01], P=0.03). BPV was not significantly associated with CBF change in other regions (P's for trend comparing first versus third tertiles of BPV=0.07–0.49).

Table 3.

Linear Mixed‐Model Estimates (ß [95% CI]) of the 2‐Way Interaction of BPV × Time on Change in CBF Levels in the Intensive Treatment Group

Region of interest	Tertile 1	Tertile 2	Tertile 3	P value for trend
Whole brain	Ref	0.02 [−0.05 to 0.10]	0.05 [−0.02 to 0.13]	0.16
GM	Ref	0.01 [−0.06 to 0.09]	0.07 [−0.01 to 0.15]	0.07
WM	Ref	0.03 [−0.05 to 0.10]	0.03 [−0.05 to 0.11]	0.49
HC	Ref	−0.02 [−0.10 to 0.07]	−0.10 [−0.18 to −0.01]	0.03
PHG	Ref	0.02 [−0.08 to 0.11]	0.04 [−0.05 to 0.14]	0.39
EC	Ref	−0.14 [−0.24 to −0.04]	−0.07 [−0.17 to 0.03]	0.17

Models adjusted for age, sex, race and ethnicity, education, history of cardiovascular disease, and mean BP over the same 9‐month period BPV was determined.

EC indicates entorhinal cortex; GM, gray matter; HC, hippocampus; PHG, parahippocampal gyrus; and WM, white matter.

After adjustment for multiple comparisons, associations between higher BPV and CBF decline in the HC and PHG in the standard treatment group remained significant (P=0.002 and P=0.004, respectively). Findings in other regions in the standard treatment group (P=0.055–0.06) and in the intensive treatment group (P=0.055–0.43) did not survive after false discovery rate correction.

Sensitivity Analysis

Consistent associations were observed in sensitivity analyses controlling for (1) number of antihypertensive medications used, (2) body mass index, (3) Framingham Risk Score, and (4) whether the follow‐up pCASL‐MRI scan was collected before versus after study closeout (Tables S1 and S2).

Discussion

Findings indicate that elevated systolic BPV is associated with CBF decline in several brain regions, even in individuals with rigorously controlled mean systolic BP levels. Under standard treatment conditions, patients with higher BPV exhibited widespread CBF decline across all regions examined, including medial temporal regions implicated in cognitive decline and Alzheimer disease dementia. Under intensive treatment conditions, associations between BPV elevation and CBF decline were less widespread, yet despite strictly controlled mean BP, patients with higher BPV still exhibited decline in hippocampal CBF. The present findings are consistent with prior work using observational cohort samples linking BPV to cerebral hypoperfusion ¹¹ and CBF decline ¹² and add to our understanding of how BPV may be related to change in cerebral perfusion in the context of rigorous antihypertensive strategies.

Although higher BPV in the standard treatment group was robustly associated with decline in global and regional CBF, it is interesting to note that associations with CBF decline were stronger in the HC, PHG, and EC than in the whole brain, GM, or WM. Additionally, BPV was associated with CBF decline in the HC only in the intensive treatment group. Prior work on BPV and CBF has found that older adults with greater BPV show CBF changes specifically in medial temporal regions known to have susceptibility to both vascular insult and Alzheimer disease pathophysiology. ¹¹ , ¹² Together, these findings suggest a possible selective vulnerability of medial temporal regions to excessive BP fluctuations and hypoperfusion, even in adults with strictly controlled mean BP. Other recent studies also indicate strong, longitudinal associations between elevated BPV and Alzheimer disease pathophysiologic changes in the medial temporal lobe, including tau accumulation ³³ and GM volume loss. ³⁴ The medial temporal lobe critically supports memory and is an early site for neuropathological change in Alzheimer disease. ¹⁴ , ¹⁵ The HC is especially vulnerable to hypoxic–ischemic injury, ¹³ , ³⁵ , ³⁶ , ³⁷ , ³⁸ and large fluctuations in BP may leave this region prone to the effects of pressure dipping, including neuronal and capillary damage to hippocampal CA1 area. ³⁸ Therefore, it is interesting that BPV was related to hippocampal CBF decline in both treatment conditions aimed at lowering mean BP, whereas associations in other brain regions were observed only in the standard treatment group. Although causality cannot be inferred from the present observational study, it is possible that intensive BP lowering constrains the effect of high BPV on CBF to particularly vulnerable regions, such as the HC. The present study is a post hoc analysis of a clinical trial, and future interventional studies directly investigating the role of BPV on CBF change are needed. Nevertheless, it is interesting that elevated BPV in both treatment groups was associated with change in hippocampal CBF as measured by pCASL‐MRI, which captures perfusion in the smallest cerebrovascular compartments (eg, capillaries and arterioles) that are especially vulnerable to erratic blood flow. ³⁵

Similar to the present study findings with CBF change largely in the standard treatment group, a recent SPRINT MIND study found that elevated BPV was associated with cognitive decline only in the standard treatment group, ²² particularly in cognitive domains vulnerable to cerebrovascular disease. ³⁹ , ⁴⁰ Together, these findings are relevant to the larger discussion that intensive BP lowering could increase the risk for cerebral hypoperfusion, ³ , ¹⁷ and the additional concern that excessive BP fluctuations (eg, dipping in already low mean BP) could exacerbate this risk. ⁸ , ⁹ However, intensive BP lowering was recently associated with an increase in CBF, ¹⁷ and our findings suggest that BPV may be most relevant to CBF change in the context of standard BP lowering. Additionally, when compared with standard BP lowering, intensive BP lowering has been associated with slower progression of WM hyperintensities, ⁴¹ , ⁴² which are strongly related to lower CBF. ⁴³ Interestingly, we found that higher BPV was associated with CBF decline in WM only in the standard treatment group, although pCASL‐MRI has limited ability to accurately quantify CBF in the WM due to low levels of perfusion and extended transit time in this tissue compartment. A growing number of studies link BPV elevation with severity and progression of cerebrovascular disease, including WM hyperintensities on MRI ⁸ , ⁹ and WM rarefaction at postmortem evaluation, independent of mean BP levels. ⁴⁴ It has been hypothesized that chronic large fluctuations in BP could have a “tsunami effect” ⁴⁵ on cerebral arterial walls, distending cerebrovascular beds beyond repair and promoting microvascular damage and cerebrovascular disease. ⁸ , ⁹ , ¹⁰ In addition to cerebrovascular disease, higher BPV may be related to markers of cerebrovascular dysfunction, including deficits in cerebrovascular reactivity. ⁴⁶ Other conditions characterized by impaired cerebral perfusion, such as carotid artery disease, can result in cerebrovascular reactivity deficits and increased risk for brain infarcts. ⁴⁷ , ⁴⁸ Intensive BP lowering may also reduce BPV to a greater extent than standard BP lowering and decouple the impact of BPV on cerebral arterial health. Additionally, some classes of antihypertensive medications may lower both mean BP and BPV, whereas others may just lower mean BP. ⁴⁹ , ⁵⁰ Interventional studies explicitly designed to test questions regarding BPV and cerebrovascular function/disease are warranted and could have therapeutic implications.

In addition to a large literature linking Alzheimer disease risk gene apolipoprotein e4 to cognitive decline and medial temporal pathologic changes, ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ apolipoprotein e4 carriers have been shown to have lower CBF when compared with noncarriers. ⁵⁶ , ⁵⁷ A growing number of studies suggest apolipoprotein e4 also modifies the relationship between BPV and several brain health outcomes. Specifically, higher BPV is associated with medial temporal GM volume loss, ³⁴ alterations in cerebral spinal fluid Alzheimer disease biomarkers amyloid‐beta and tau, ⁵⁸ and tau accumulation in the medial temporal lobe ³³ in apolipoprotein e4 carriers and not in apolipoprotein e4 noncarriers. Studies that are able to address potential moderating effects of apolipoprotein e4 on BPV and CBF are needed. Importantly, apolipoprotein e4 carriers exhibit blood–brain barrier breakdown in the medial temporal lobe, ⁵⁹ , ⁶⁰ where BPV was most strongly associated with CBF decline, especially in the standard treatment group. Future studies investigating the role of apolipoprotein e4 in BP clinical trials may improve our understanding of vascular factors in precision medicine approaches to dementia care.

The present investigation used a cohort of adults in a randomized control trial with strictly controlled BP. Therefore, we were able to assess potential antihypertensive treatment effects in a more rigorous way than prior work relying on observational cohorts with varied BP control. ⁶ , ⁸ Relatedly, findings provide novel evidence that BPV is related to CBF decline even in individuals with strictly controlled mean BP and add to recent studies using observational cohorts. ¹¹ , ¹² Findings suggest that intensive lowering may limit the effects of BPV on CBF to only highly vulnerable areas, whereas BPV may pose a risk to more pervasive CBF changes under standard treatment. While we did examine differences based on the intensive versus standard treatment group, we were not able to explore potential class effects of monotherapy or combination therapy—an area of active investigation and an important step in translating BPV research to the clinic. However, BPV was calculated from BP measurements using methods that are standard in clinical settings and support the role of BPV as an important aspect of BP control. ⁶¹ , ⁶² CBF was quantified using pCASL‐MRI and was thus able to capture regional and global cerebral perfusion (as opposed to transcranial Doppler, which is limited to capturing flow in the larger intracranial arteries ⁶³ ). Thus, we were able to appreciate treatment effects on the relationship between BPV and CBF in brain regions with known susceptibility to cerebrovascular disease and Alzheimer disease. ¹⁴ , ¹⁵ Additionally, the SPRINT cohort is racially and ethnically diverse, with variability in educational level, ⁴ which helps generalize study findings perhaps more broadly than the majority of BPV work relying on largely non‐Hispanic White and highly educated samples. ⁶ , ⁸ , ⁹ , ¹² There are several study limitations. As part of inclusionary criteria, ¹⁸ SPRINT participants were at higher risk for cardiovascular disease before treatment randomization, which could bias findings related to vascular function (eg, BP and CBF). However, findings are consistent with 2 prior observational studies using samples with limited vascular risk. ¹¹ , ¹² The majority of follow‐up MRI scans were collected after study closeout, similar to another post hoc analysis of SPRINT MIND CBF data. ¹⁷ Although findings were consistent in sensitivity analyses controlling for whether the follow‐up MRI was collected before versus after study closeout, transitioning BP management at this stage could have effects on BP and CBF. Additionally, the specific parameters used for pCASL‐MRI were consistent with a recent post hoc analysis of SPRINT MIND CBF data, ¹⁷ but scan sequences with a longer label duration and postlabel delay may provide more accurate estimates of CBF, especially within the WM. The present study is further limited by the post hoc nature of the analyses. Future randomized control trials directly examining/targeting BPV beyond traditionally studied mean BP levels have the potential to improve our understanding of BPV as an emerging BP risk indicator strongly associated with cerebrovascular disease and dementia. ⁶ , ⁸ , ⁹

Sources of Funding

The study data analysis was supported by National Institutes of Health/National Institute on Aging grants (R01AG064228, R01AG060049, P30AG066519, P01AG052350) and Alzheimer's Association grant AARG‐17‐532 905.

Disclosures

None.

Supporting information

Table S1

Table S2