Effect of Intensive Versus Standard Blood Pressure Control on Stroke Subtypes

Abstract

In the Systolic Blood Pressure Intervention Trial (SPRINT), the number of strokes did not differ significantly by treatment group. However, stroke subtypes have heterogeneous etiologies that could respond differently to intensive blood pressure control. SPRINT participants (N=9361) were randomized to target systolic blood pressures of <120 mm Hg (intensive treatment) compared to <140 mm Hg (standard treatment). We compared incident hemorrhage, cardiac embolism, large- and small-vessel infarctions across treatment arms. Participants randomized to the intensive arm had mean systolic blood pressures of 121.4 mm Hg in the intensive arm (N=4,678) and 136.2 mm Hg in the standard arm (N=4,683) at one year. Sixty-nine strokes occurred in the intensive arm and 78 in the standard arm when SPRINT was stopped. The breakdown of stroke subtypes across treatment arms included hemorrhagic (intensive treatment, n=6, standard treatment, n=7) and ischemic stroke subtypes (large artery atherosclerosis: intensive treatment n=11, standard treatment, n=13; cardiac embolism: intensive treatment n=11, standard treatment n=15; small artery occlusion: intensive treatment n=8, standard treatment n=8; other ischemic stroke: intensive treatment n=3, standard treatment n=1). Fewer strokes occurred among participants without prior cardiovascular disease in the intensive (n = 43) than the standard arm (n=61), but the difference did not reach pre-defined statistical significance level of 0.05 (p=0.09). The interaction between baseline cardiovascular risk factor status and treatment arm on stroke risk did not reach significance (p=0.05). Similar numbers of stroke subtypes occurred in the intensive BP control and standard control arms of SPRINT.

Graphical Abstract

INTRODUCTION

Stroke is the fifth leading cause of death in the United States and the leading cause of adult disability. The estimated annual cost of stroke is expected to increase by $240 billion by 2030 ¹. The most recent AHA/ASA Statement on the primary prevention of stroke states, “The relationship between BP and stroke risk is strong, continuous, graded, consistent, independent, predictive, and etiologically significant”.² Observational studies show a benefit of lower BP down to 115/75 for both men and women aged 40 to 89 years in relation to risk of first fatal or non-fatal stroke.^{3, 4} Multiple randomized controlled trials have also shown a benefit of moderate BP lowering in primary stroke prevention, and some have shown a benefit of more intensive BP lowering to various targets as well, but the issue of what BP target is optimal for stroke prevention remains unsettled.⁵

The Systolic Blood Pressure Intervention Trial (SPRINT) randomized 9,361 participants with an increased cardiovascular risk, but without diabetes or a history of stroke, to a systolic BP target of less than 120 mm Hg (intensive treatment) or a target of less than 140 mm Hg (standard treatment) and was stopped early when the combined primary endpoint was reached. The combined primary outcome favored the intensive BP arm, with significantly fewer fatal and nonfatal major cardiovascular events (including stroke) and death from any cause. However, although fewer strokes occurred in the intensive arm, only 147 stroke events had accrued by study termination.⁶ Since stroke is a broad term referring to events caused by multiple mechanisms, most of them modifiable by treatment of hypertension, it is important to examine the effect of intensive BP lowering on different stroke subtypes to better understand the mechanisms that link hypertension with different types of brain damage. Blood pressure lowering appears to be beneficial for preventing both ischemic and hemorrhagic stroke, but only limited data are available on the effects of BP treatment on different types of stroke, especially ischemic stroke subtypes.^{3, 4} Cerebral small vessel disease due to hypertensive vasculopathy can manifest as small subcortical (i.e., lacunar) infarctions and intraparenchymal hemorrhages and may be the most direct link between high BP and stroke. Hypertension also leads to heart disease, including myocardial infarction, non-valvular atrial fibrillation, and heart failure predisposing to cardiac embolism that results in stroke. Large vessel atherosclerosis leads to local thrombosis or artery to artery embolism. Given the established links between hypertension and small vessel stroke, and the reductions in heart failure and cardiovascular mortality due to intensive BP lowering seen in SPRINT, we hypothesized that intensive BP lowering would also result in fewer strokes attributable to cerebral small vessel disease or those attributable to a cardiogenic mechanism.

METHODS

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Design

Stroke sub-typing was a pre-specified outcome. The design and cardiovascular outcome results for SPRINT have been described previously.^{6, 7} Adults 50 years of age or older with systolic BP (SBP) between 130 and 180 mm Hg at screening were enrolled. Participants were at elevated cardiovascular risk, defined as either having chronic kidney disease with an estimated glomerular filtration rate of 20 to <60 ml/min/1.73m², a 10-year Framingham cardiovascular disease risk ≥15%, or being 75 years of age or older. Exclusions included having diabetes mellitus, a history of prior stroke or dementia, or living in a nursing home. Enrolled participants were randomly assigned to either an intensive treatment strategy with a systolic BP goal of <120 mm Hg or a standard treatment strategy with a systolic BP goal of <140 mm Hg. SPRINT was funded by the National Heart, Lung, and Blood Institute, and co-funded by the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Neurological Disorders and Stroke, and the National Institute on Aging. An independent data and safety monitoring board monitored unblinded trial results and safety events. The study was approved by the institutional review board at each participating study site (NCT01206062).

Stroke Adjudication and Subtyping

Randomization was stratified by clinical site and participants and study personnel were aware of study-group assignments, but outcome adjudicators were blinded and did not adjudicate cases from their home networks. Medical records and electrocardiograms were obtained for documentation of events and whenever clinical site staff became aware of a stroke, the approved protocol was followed to obtain information on the event.⁷ A member of the Stroke Subcommittee (CW) of the Morbidity and Mortality Committee of SPRINT adjudicated stroke subtypes using the Causative Classification System (CCS), a standardized, automated, evidence-based, and web-based subtype classification system adapted for use in SPRINT as specified in the design and rationale publication.^{8, 9} The CCS generated five mutually exclusive categories for each case: large artery atherosclerosis (LAA; supra-aortic), cardiac embolism (CE), small artery occlusion (SAO), other uncommon causes, and undetermined causes.^{8, 10, 11} The undetermined category was further divided into cryptogenic, multiple competing etiologies, and incomplete evaluation.^{8, 10} We also stratified CE into major-CE and minor-CE where major CE denotes cardio-embolic sources with high potential to cause a stroke such as atrial fibrillation, and minor CE indicates sources with a low or uncertain potential to cause a stroke such as mitral annulus calcification (1–4).^{8, 10–12} Finally, we generated an additional cryptogenic group that also included low- or uncertain-risk cardiac sources. The CCS software stratified each CCS category into three confidence levels as evident, probable, and possible depending on the level of causal evidence. Because of the small number of strokes, we determined the effects of the intervention by aggregating the confidence levels.

Statistical Analysis

We compared the number of participants with stroke types and subtypes across treatment arms using Chi-Squared Tests or Fisher’s Exact Tests (for expected cell counts < 5). The three major stroke types were hemorrhagic, ischemic, and unknown. Hemorrhagic strokes were classified as subarachnoid, intraparenchymal, or other. We then used Cox proportional hazard models stratified by clinical site to estimate hazard ratios and 95% confidence intervals comparing stroke rates across pre-specified subgroups using tests of interaction with significance defined as Hommel-adjusted alpha levels smaller than 0.05.¹³

RESULTS

Characteristics of the SPRINT participants are shown in Table 1 and were well balanced across BP treatment arms. Of 9,361 participants randomized to intensive (N=4,678) or standard (N=4,683) BP control, 69 participants in the intensive arm versus 78 participants in the standard arm had strokes during a mean follow-up of 3.33 years (Figure 1). There was no significant difference in the number of strokes overall by treatment arm.⁶ Likewise, baseline cardiovascular risk factor status did not significantly modify the effect of the intervention. However, for those having no prior history of cardiovascular disease at baseline, results favored intensive BP control, but this did not reach significance (p=0.05, Figure 2).

Table 1:

Baseline Clinical Characteristics by Treatment Arm

Characteristic	Intensive Treatment (N=4678)	Standard Treatment (N=4683)
Criterion for increased cardiovascular risk – no. (%)
Age≥75 years	1317 (28.2%)	1319 (28.2%)
Chronic Kidney disease	1329 (28.5%)	1316 (28.3%)
Cardiovascular disease – no. (%)	940 (20.1%)	937 (20.0%)
Clinical	779 (16.7%)	783 (16.7%)
Subclinical	247 (5.28%)	246 (5.3%)
Framingham 10-yr cardiovascular disease risk score >= 15%	3556 (76.0%)	3547 (75.7%)
Female sex – no. (%)	1684 (36.0%)	1648 (35.2%)
Age – yr.
Overall	67.9 ± 9.4	67.9 ± 9.5
Among those ≥ 75 yr of age	79.8 ± 3.9	79.9 ± 4.1
Race or Ethnic Group – no. (%)
Non-Hispanic black	1379 (29.5%)	1423 (30.4%)
Hispanic	503 (10.8%)	481 (10.3%)
Non-Hispanic white	2698 (57.7%)	2701 (57.7%)
Other	98 (2.1%)	78 (1.7%)
Black race*	1454 (31.1%)	1493 (31.9%)
Baseline blood pressure – mm Hg
Systolic	139.7 ± 15.8	139.7 ± 15.4
Diastolic	78.2 ± 11.9	78.0 ± 12.0
Distribution of systolic blood pressure – no. (%)
≤ 132 mmHg	1583 (33.8%)	1553 (33.2%)
>132 mm Hg to <145 mm Hg	1489 (31.8%)	1549 (33.1%)
≥145 mm Hg	1606 (34.3%)	1581 (33.8%)
Serum creatinine - mg/dl	1.07 ± 0.34	1.08 ± 0.34
Estimated GFR – ml/min/1.73 m2
Among all participants	71.8 ± 20.7	71.7 ± 20.5
Among those with estimated GFR ≥60 ml/min/1.73 m2	81.3 ± 15.5	81.1 ± 15.5
Among those with estimated GFR <60 ml/min/1.73 m2	47.8 ± 9.5	47.9 ± 9.5
Ratio of urinary albumin (mg) to creatinine (g)	44.1 ± 178.7	41.1 ± 152.9
Fasting total cholesterol - mg/dl	190.2 ± 41.4	190.0 ± 40.9
Fasting HDL cholesterol - mg/dl	52.9 ± 14.3	52.8 ± 14.6
Fasting total triglycerides - mg/dl	124.8 ± 85.8	127.1 ± 95.0
Fasting plasma glucose - mg/dl	98.8 ± 13.7	98.8 ± 13.4
Statin use – no./total no. (%)	1978/4646 (42.6%)†	2076/4640 (44.7%)†
Aspirin use – no./total no. (%)	2406/4662 (51.6%)†	2350/4666 (50.4%)†
Smoking Status – no. (%)
Never smoked	2051 (43.8%)	2072 (44.2%)
Former smoker	1977 (42.3%)	1996 (42.6%)
Current smoker	639 (13.7%)	601 (12.8%)
Missing data	11 (0.2%)	14 (0.3%)
Atrial fibrillation – no./total no. (%)	390/4661 (8.4%)†	364/4666 (7.8%)†
Framingham 10-yr cvd risk score – %	24.8 ± 12.6	24.8 ± 12.5
Body-mass index	29.9 ± 5.8	29.8 ± 5.7
Anti-Hypertensive Medications Prescribed – no./patient	1.8 ± 1.0	1.8 ± 1.0
Not using antihypertensive agents – no. (%)	432 (9.2%)	450 (9.6%)

^*Black race includes Hispanic black and black as part of a multiracial identification.

^†Denominator smaller than overall treatment arm totals due to missing data.

Figure 1. Cumulative Incidence Plot of Stroke Events by Treatment Arm in SPRINT.

The cumulative incidence of stroke events (Y-axis) in each SPRINT study arm is plotted for the number of participants at risk by month of follow-up (X-axis).

Figure 2. Forest Plot of Stroke Outcome by Subgroups.

The forest plot for the groups of interest. Note that all interactions between covariate and treatment arm are non-significant for stroke as an outcome at the <0.05 significance level.

Hemorrhagic and ischemic strokes and their subtypes across treatment arms are shown in Table 2. There were 32 events that could not be classified as to type, and five with incomplete stroke evaluations that prevented subtype classification. Roughly 70% of strokes were ischemic, with cryptogenic (31%) and cardioembolic (25%) being the most common subtypes. The treatment effects of intensive BP were consistent for the different stroke subtypes with generally fewer strokes in the intensive arm regardless of subtype. Likewise, treatment effects were similar across arms when the undetermined category was broken into subgroups, when the CE category was stratified into major and minor, and when minor CE was combined with the cryptogenic category. The number of ischemic and hemorrhagic stroke subtypes was similar across age (<75 versus ≥75 years), sex, and race/ethnic strata (p=0.05, Figure 2).

Table 2:

Stroke subtypes in the Systolic Blood Pressure Intervention Trial (SPRINT)

Stroke Outcome	Overall (N=9361)	Intensive (N=4678)	Standard (N=4683)	p-value
Absolute Overall — no. (%)	147 (1.6)	69 (1.5)	78 (1.7)	0.51
Hemorrhagic Stroke — no. (%)	13 (0.1)	6 (0.1)	7 (0.2)	1.00
Subarachnoid Hemorrhage	3 (0.0)	2 (0.0)	1 (0.0)	0.65*
Intraparenchymal Hemorrhage	6 (0.1)	3 (0.1)	3 (0.1)	1.00*
Other Hemorrhage	4 (0.0)	1 (0.0)	3 (0.1)	0.62*
Ischemic Stroke — no. (%)	102 (1.1)	48 (1.0)	54 (1.2)	0.62
Large Artery Atherosclerosis	24 (0.3)	11 (0.2)	13 (0.3)	0.84
Cardiac Embolism	26 (0.3)	11 (0.2)	15 (0.3)	0.56
Major Cardiac Embolism	13 (0.1)	5 (0.1)	8 (0.2)	0.58
Minor Cardiac Embolism	13 (0.1)	6 (0.1)	7 (0.2)	1.00
Small Artery Occlusion	16 (0.2)	8 (0.2)	8 (0.2)	1.00
Other Uncommon Causes	4 (0.0)	3 (0.1)	1 (0.0)	0.38*
Undetermined Etiology	32 (0.3)	15 (0.3)	17 (0.4)	0.86
Unknown: Cryptogenic Embolism	19 (0.2)	8 (0.2)	11 (0.2)	0.65
Multiple Competing Etiologies	8 (0.1)	4 (0.1)	4 (0.1)	1.00*
Incomplete Evaluation	5 (0.1)	3 (0.1)	2 (0.0)	0.69*
Unknown Stroke Type — no. (%)	32 (0.3)	15 (0.3)	17 (0.4)	0.86

^*Denotes Fisher’s exact test.

DISCUSSION

In this pre-specified analysis of SPRINT data, intensive BP lowering resulted in similar numbers, types, and subtypes of strokes compared to standard BP control during follow-up. The numbers of hemorrhagic and ischemic stroke subtypes were similar across arms.

As indicated in the primary outcomes report, intensive BP control in SPRINT participants reduced the risk of heart failure and death from cardiovascular causes, with a non-significant decrease in the number of myocardial infarctions as well.⁶ Given these findings, a lower risk of ischemic strokes caused by cardiac embolism in the intensive BP control group than the standard control group might be expected. However, the number of stroke events was small and SPRINT was not powered to detect these differences. Although the anticipated difference in SBP between randomized arms (>10 mm Hg) was achieved, it is possible a difference in strokes may have been observed if greater SBP differences across arms had been attained.

Hypertension is a major risk factor for small vessel arteriopathies that lead to end-organ damage affecting the brain, heart, kidney, and eye.^14–17 Intensive BP lowering could limit such damage in patients with hypertension, thereby protecting against stroke. In the Secondary Prevention of Small Subcortical Strokes (SPS3) trial that enrolled participants with recent lacunar strokes who were thus at greater risk of stroke than those in SPRINT, maintaining systolic BP below 130 mm Hg compared to 130–139 mm Hg resulted in significantly fewer incident intraparenchymal hemorrhages but not new ischemic strokes.¹⁸ Since intraparenchymal hemorrhage often results in severe morbidity and mortality, it is notable that intensive BP lowering did not result in fewer hemorrhages compared to standard control in the current study.

Intensive lowering of BP in patients with longstanding hypertension could also place some organs at risk of ischemia should BP be lowered too aggressively. In the brain, long-standing hypertension has been hypothesized to require greater BPs to maintain adequate cerebral perfusion pressures and avoid ischemia due to a rightward shift in the autoregulatory curve.¹⁹ Data are limited, but a small study showed cerebral hemodynamic responses were stable acutely and up to four months later after initiating treatment of mild and moderate hypertension.²⁰ In SPRINT, intensive BP lowering also did not increase the risk of stroke subtypes usually attributable to cerebral small vessel disease such as intracerebral hemorrhage and lacunar stroke. These findings are consistent with the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial where intensive BP control targeted below 130 mm Hg significantly lowered stroke risk, and with pooled data from SPRINT and ACCORD that showed no increase in stroke risk from intensive BP lowering.^{21, 22} Further, the SPRINT MIND study found intensive BP control lowered the risk of mild cognitive impairment (MCI) and the combined outcome of MCI and probable dementia.²³ Intensive BP control did not prevent probable dementia alone (the primary outcome) compared to standard BP control, but there were only about half as many dementia as MCI events due to early termination.

In the current study we did not find a notable difference between treatment arms in the number of hemorrhagic or ischemic strokes usually attributed to small vessel arteriopathies, namely intraparenchymal hemorrhages and lacunar infarctions. However, the SPRINT-MIND MRI sub-study showed that participants in the intensive BP treatment arm had less progression of white matter hyperintensities (WMH) than those in the standard arm and that this effect did not differ across various sub-groups (those with and without prior CVD; those with and without a history of orthostatic hypertension; those older versus younger than 75; and across baseline BP levels).²⁴ Since there was reason to be concerned about the possible detrimental effects of intensive BP control in all of these subgroups, these findings are reassuring. In addition, since only 22 strokes in the small vessel sub-type categories occurred during follow-up, lack of concordance with the MRI sub-study findings could be attributable to the small number of events. In addition, WMH are areas of extracellular water detected on T2-weighted MRI sequences and, though strongly associated with small vessel damage, are non-specific.

SPRINT was not designed to examine hypertensive medication class effects, and the small number of ischemic strokes does not allow for a meaningful analysis of the potential class effects of different BP agents. Some classes have been posited to provide additional benefits beyond BP lowering. For example, the Heart Outcomes Prevention Evaluation (HOPE) trial showed a benefit to adding ramipril to standard BP treatment for both ischemic and hemorrhagic stroke, and the Systolic Hypertension in the Elderly Program (SHEP) showed a benefit of chlorthalidone versus placebo for various stroke subtypes.^{25, 26}

The CCS was used for ischemic stroke subtype classification in this study and differs from conventional systems in that it is fully rule- and evidence-based as an algorithm, using objective criteria to assign stroke etiologies into easily replicable subtypes.²⁷ The reported kappa values for CCS range between 0.75 and 0.90 depending on the data source, number of cases, and number of raters.^{8, 10–12} In contrast, reports from independent investigators demonstrate only moderate reliability with kappa values ranging between 0.42 and 0.54 for conventional etiologic classification systems.^28–32 Likewise, CCS provides higher discriminative validity compared to other classification systems, because the CCS generates more distinct subtypes with discrete clinical, genetic, and prognostic features.^{27, 33} However, the ability of etiologic classification, regardless of the system used, to unambiguously assign the cause of stroke is limited because of the absence of pathology data. This problem is greater for the category of cardiac embolism as this category includes several abnormalities with discrete embolic potential. The CCS provided the flexibility to examine the effect of intensive BP reduction across a wide range of etiologies generated based on the strength of causal evidence. For instance, we stratified cardiac sources into high-risk and low- or uncertain-risk CE. Likewise, we generated a new cryptogenic category that included low- or uncertain-risk cardiac pathologies. We found that the intervention effect was similar across such categories, suggesting that the etiologic mechanism of stroke is not a strong determinant of benefit from intensive BP reduction, with the caveat noted above about the small number of strokes in each category.

Limitations.

This study was not powered to detect differences in stroke subtypes. The generalizability of these findings is limited to people with higher cardiovascular risk than the general population due to the enrollment criteria that excluded those with diabetes mellitus, prior stroke, and nursing home residents. A single adjudicator did the ischemic stroke subtype classifications using the CCS system and intra-rater reliability was not measured during the adjudication process.

In summary, we found similar numbers of stroke subtypes in the intensive BP control and standard control arms of SPRINT.

Perspectives

Hypertension leads to stroke through heart disease, especially atrial cardiopathy and atrial fibrillation, as well as large vessel atherosclerosis and small vessel arteriolosclerosis. In addition, longstanding hypertension may place the brain at elevated risk of ischemia if blood pressure is treated aggressively. It is important to understand if intensive blood pressure control affects the risk of certain types of stroke. In SPRINT, intensive blood pressure control did not reduce the risk of stroke overall, but the study was stopped early and the number of strokes was small. Different kinds of damage lead to diverse types of stroke, and intensive blood pressure control could affect them in different ways. In this study we compared the effect of intensive blood pressure control to standard blood pressure control on the risk of different types of stroke and found that hemorrhagic and ischemic subtypes were similar across treatment arms and neither blood pressure control strategy showed greater benefits or harms. Future well-powered stroke studies are needed to understand the role of intensive blood pressure control on different types of stroke in those at high risk of cardiovascular disease and longstanding hypertension.

Novelty and Significance.

What is new?

• SPRINT provided a unique opportunity to compare intensive blood pressure control to standard blood pressure control in relation to the risk of hemorrhagic and ischemic stroke and their subtypes.

What is relevant?

• Hypertension is a strong risk factor for stroke through its effects on the heart and systemic arteries.

• Intensive blood pressure lowering could halt the deleterious effects of hypertension and lower the risk of stroke or cause brain ischemia in the setting of longstanding hypertension.

Summary

• The number of hemorrhagic and ischemic stroke subtypes were similar across the intensive and standard blood pressure arms of SPRINT.

• Intensive blood pressure lowering was not associated with an elevated risk of cerebral small vessel strokes compared to standard control.