Level of Systolic Blood Pressure Within the Normal Range and Risk of Recurrent Stroke FREE

Although national guidelines suggest maintaining a normal blood pressure (BP) defined as a systolic BP (SBP) of less than 120 mm Hg and a diastolic BP (DBP) of less than 80 mm Hg in persons with a prior stroke,¹ limited data specifically address the role of BP levels within the normal range for vascular risk reduction after stroke.² With recent evidence suggesting no benefit in achieving more aggressive SBP targets in high-risk patients with diabetes,³ and perhaps even harm,⁴ there is mounting interest in exploring the existence and nature of the J-shaped link of BP with outcome in various patient groups at high risk for vascular events. For stroke, the vascular disease entity most highly correlated with BP,⁵ it is generally perceived that a J-shaped association between BP and outcome may not exist.

Pending the conduct of a large randomized controlled trial examining the effects of intensive vs standard BP control on a recurrent stroke end point, information on the differential relationship of BP maintained within a high vs low-normal range with outcome could be beneficial. In this study, we evaluated the independent association of SBP maintained within a low-normal range vs high-normal range with clinical outcomes among patients who recently experienced an ischemic stroke.

We conducted a post hoc observational analysis of the Prevention Regimen for Effectively Avoiding Second Strokes (PROFESS) trial. Details of the trial protocol and the main results have been published.^6- 8 In brief, from September 2003 through July 2006, PROFESS enrolled 20 330 patients from 695 centers in 35 countries who recently had a non–cardioembolic ischemic stroke.

Patients aged 55 years or older, or those aged 50 to 54 years with at least 2 additional vascular risk factors, who experienced an ischemic stroke within 120 days prior to randomization and whose condition was stable, were invited to participate in the study. Ischemic stroke was defined as a new focal neurological deficit of cardiovascular origin persisting for more than 24 hours, but patients whose symptoms persisted for less than 24 hours could be included if they had evidence of a recent ischemic stroke on computed tomography or magnetic resonance imaging. Patients were excluded if they had a primary hemorrhagic stroke, severe disability after the qualifying stroke, contraindications to 1 of the study antiplatelet agents, or other factors making them unsuitable for randomization.^6- 8

Given the influence of race on stroke outcomes,⁵ patient-reported information on race was collected in the PROFESS trial and categorized as white (European/Caucasian, Arab, or Persian), black (colored African or black African), Asian (South Asian, Chinese, Japanese, Malays, or other Asian), or Other (Caribbean Hispanic, Native Latin, or other).

The trial used a 2 × 2 factorial design to compare 4 regimens: a combination of acetylsalicylic acid (aspirin) and extended-release dipyridamole compared with clopidogrel and telmisartan compared with placebo. All patients also received medications at the discretion of the investigators for control of BP. Patients were evaluated at the time of hospital discharge or in a clinic visit at 1 week and then at 1, 3, and 6 months; and every 6 months thereafter. The minimum expected follow-up was 18 months. Interim telephone calls were scheduled halfway between clinic visits. Blood pressure was measured by using a standard and validated Omron sphygmomanometer (Omron Healthcare Inc) with an appropriately sized cuff applied to the upper nondominant arm at heart level. Blood pressure was measured twice, at least 2 minutes apart, and the measurements were averaged.⁸

The median time from the qualifying stroke to randomization was 15 days, and 39.8% of patients were randomized within 10 days of the qualifying stroke. The mean duration of patient follow-up was 2.5 years (range, 1.5-4.4 years); the median duration of patient follow-up was 2.44 years (range, 1.96-3.00 years); and follow-up ended on February 8, 2008. The primary outcome was first recurrence of stroke of any type. The secondary outcome was a composite of stroke, myocardial infarction (MI), or death from vascular causes. A central committee blinded to treatment code adjudicated the primary outcome and the first of the secondary outcomes. If a patient had multiple MIs, all MIs were adjudicated up until the first MI was confirmed. After an MI was confirmed for a patient, all subsequent MIs for that patient were not adjudicated. The PROFESS trial was approved by the ethics committee or institutional review board at each national or local site, and all participants provided written informed consent.

The PROFESS trial did not find any evidence that the combination of acetylsalicylic acid (aspirin) and extended-release dipyridamole was superior to clopidogrel or that telmisartan was superior to placebo in the prevention of recurrent stroke.^6- 7 As such, for this study, data for all enrolled patients were combined and included in these analyses following the intention-to-treat principle.

Patients in the PROFESS trial were divided into 5 prespecified groups according to their mean SBP level: very low–normal (<120 mm Hg), low-normal (120-<130 mm Hg), high-normal (130-<140 mm Hg), high (140-<150 mm Hg), and very high (≥150 mm Hg). Systolic BP was chosen because of its relatively stronger relationship with vascular risk,^9- 12 the J-curve phenomenon,¹³ and to be consistent with data from other recently published studies on this topic.^3- 4 The SBP groups were chosen to reflect levels mentioned in guideline recommendations,¹ for consistency with other studies,^3- 4 and practical considerations in routine clinical settings. A multitiered approach comparing SBP levels within the normal range and abnormal range has been used previously.⁴ Blood pressure measurements from postbaseline clinic visits before the first occurrence of stroke, MI, vascular death, or censoring were used to determine the mean follow-up BP.

Two patients without BP measurements during the trial were excluded from these analyses. Baseline demographic and clinical covariates to be examined for the 5 SBP level groups were preselected based on prior studies of factors that influence the occurrence of vascular events outcomes after ischemic stroke. The additional secondary outcomes that were evaluated were fatal stroke, fatal and nonfatal MI, death from vascular causes, death from all causes, and composite end point of stroke, MI, or death from vascular causes. We only used SBP measurements and events up to the first occurrence of stroke, MI, or death to ensure that patients were in the same BP control category for all of the individual end points. For patients without postbaseline SBP measurements prior to the event (n = 631), we used their baseline values. The group with mean SBP within the high-normal range was the reference group.

Mean follow-up and reduction in BP from baseline was calculated for each patient using all but their baseline measurements until they died, experienced an outcome (stroke or MI), or were censored. Covariate selection was performed by forcing the following prespecified covariates into an initial step-wise model: BP category, age, sex, history of prior stroke (before qualifying stroke), history of congestive heart failure, history of diabetes, and history of MI. The covariates presented in Table 1 also were included and selected based on a P value level of less than .10 to enter into the model and a P value level of less than .10 to remain in the model. The forced and selected covariates were then included in a final model for analysis.

Tests for interaction for prespecified baseline features including age, sex, history of diabetes, stroke before qualifying event, large vessel stroke subtype, small vessel stroke subtype, and treatment assignment (telmisartan vs placebo) also were performed. Step-wise Cox (forward selection) models also were used to separately analyze the association of baseline SBP, mean follow-up BP, and mean BP reduction on the risk of outcome in relation to various mean SBP levels during the trial.

To investigate the association of lower SBP on outcome in patients with known diabetes at baseline, we also evaluated the relationship of mean SBP during the PROFESS trial with adjusted outcome rates. To explore the relationship of timing of SBP level maintained within the normal range with outcome after ischemic stroke, we divided the periods for events after the qualifying stroke as occurring during the first 30 days, more than 30 days to 90 days, more than 90 days to 180 days, or more than 180 days, and evaluated the relationship between achieved SBP level and vascular risk within each of these periods.

The overall significance level for the study was a P value of less than .05 using a 2-sided test. Based on comparisons with the high-normal group at an α level of .05, there was greater than 80% power to detect a 20% risk reduction in recurrent strokes for the low-normal, high, and very high SBP level groups, and greater than 70% power for the very low–normal SBP level group. All analyses were performed using SAS statistical software version 9.2 (SAS Institute Inc).

Among the 20 330 patients, the mean (SD) age was 66.1 (8.6) years and 36% were female. The baseline characteristics of patients within all designated categories of mean SBP are shown in Table 1. Several of these characteristics were comparable across SBP groups, although frequency of hypertension, diabetes, and use of an antihypertensive drug at baseline increased steadily from the very low–normal SBP level group to the very high SBP level group. Of note, specifically comparing the very low–normal SBP level group with the high-normal SBP level group, there was a much lower frequency of hypertension (53.7% vs 73.7%, respectively) and antihypertensive drug use (54.0% vs 65.8%), as well as much higher frequencies of use for an antiplatelet drug (72.4% vs 62.5%) and a lipid modifier drug (53.5 vs 46.0%), and treatment with telmisartan (68.7% vs 48.8%).

Table 2 displays various BP parameters (eg, mean baseline, mean follow-up, and mean change) by designated category of mean SBP. Mean baseline and follow-up BPs were lowest in the very low–normal SBP level group, while mean BP change was greatest in the very low–normal SBP group. Specifically, mean (SD) baseline SBP was 130.0 (14.3) mm Hg in the very low–normal SBP level group vs 142.7 (14.5) mm Hg in the high-normal SBP group; mean (SD) follow-up SBP was 113.4 (5.7) mm Hg in the very low–normal SBP group vs 134.8 (2.89) mm Hg in the high-normal SBP group; and mean (SD) change in SBP was −16.9 (14.3) mm Hg in the very low–normal SBP group vs −8.1 (14.5) mm Hg in the high-normal SBP group.

Results of the unadjusted associations of mean SBP with outcome appear in Table 3. Occurrence of the primary outcome (stroke) was greatest in the very high SBP level group (14.1%; 95% CI, 13.0%-15.2%), followed by the high SBP group (8.7%; 95% CI, 7.9%-9.5%), the very low–normal SBP group (8.0%; 95% CI, 6.8%-9.2%), the low-normal SBP group (7.2%; 95% CI, 6.4%-8.0%), and then the high-normal SBP group (6.8%; 95% CI, 6.1%-7.4%). Occurrence of the secondary outcome (stroke, MI, or vascular death) followed a similar pattern. However, rates of all-cause mortality and death due to vascular causes were highest in the very low–normal SBP group (9.2% [95% CI, 7.9%-10.5%] and 3.1% [95% CI, 2.3%-3.9%], respectively) and very high SBP group (9.2% [95% CI, 8.3%-10.1%] and 3.2% [95% CI, 2.6%-3.7%], respectively). eFigure 1 and eFigure 2) show percentages of recurrent stroke and MI by mean SBP group and reveal near-similar J-curve patterns to this relationship among the patients with a recent ischemic stroke.

In multivariable analyses, compared with the high-normal SBP level group, risks of the primary outcome were significantly higher in the very low–normal SBP group (adjusted hazard ratio [AHR], 1.29; 95% CI, 1.07-1.56), in the high SBP group (AHR, 1.23; 95% CI, 1.07-1.41), and in the very high SBP group (AHR, 2.08; 95% CI, 1.83-2.37) (Table 4). Compared with the high-normal SBP level group, risks of the secondary outcome were significantly higher in the very low–normal SBP group (AHR, 1.31; 95% CI, 1.13-1.52), in the low-normal SBP group (AHR, 1.16; 95% CI, 1.03-1.31), in the high SBP group (AHR, 1.24; 95% CI, 1.11-1.39), and in the very high SBP group (AHR, 1.94; 95% CI, 1.74-2.16). The AHRs of covariates included in the adjusted risk models appear in eTable 1.

Table 5 shows the results of patterns of interactions that were significant between the SBP groups and prespecified variables. Compared with those who were younger than 65 years, patients aged 75 years or older with very low–normal SBP levels (primary outcome: risk ratio [RR], 1.02 [95% CI, 1.01-1.04]; secondary outcome: RR, 1.05 [95% CI, 1.04-1.06]) experienced more events, a disparity more pronounced than among those patients with very high SBP levels (primary outcome: RR, 1.00 [95% CI, 0.99-1.01]; secondary outcome: RR, 1.007 [95% CI, 0.999-1.016]). Patients with or without baseline diabetes and mean SBP level in the very low–normal SBP group did not significantly differ on the secondary outcome (RR, 1.14; 95% CI, 0.87-1.50). However, patients in all of the other SBP groups, including those in the low-normal SBP level group, baseline diabetes was significantly associated with a higher risk of the secondary outcome (RR, 1.75; 95% CI, 1.47-2.09). There was a significant interaction of add-on telmisartan treatment with the composite secondary outcome (ie, fewer events in the very low–normal SBP level group: RR, 0.77 [95% CI, 0.60-0.98], but more events in the high SBP level group: RR, 1.15 [95% CI, 0.99-1.34]).

Models that adjusted separately for mean baseline, mean follow-up, and mean change from baseline in SBP showed generally similar patterns for outcomes across SBP categories (eTable 2). Mean baseline SBP and mean BP change were not selected by the models for either the primary or secondary outcomes, but mean follow-up SBP was independently associated with the primary outcome (HR, 1.010; 95% CI, 1.006-1.014) and secondary outcome (HR, 1.009; 95% CI, 1.005-1.014). eTable 3 and eTable 4 show the unadjusted and adjusted clinical outcomes by mean SBP category in patients with known diabetes at baseline. There were no significant differences in primary outcome (HR, 1.065; 95% CI, 0.727-1.560) or secondary outcome (HR, 1.142; 95% CI, 0.850-1.534) between the very low–normal SBP group compared with the high-normal SBP groups of patients with diabetes.

Analyses by incidence of vascular events by time and mean SBP category (eTable 5) showed that the J-shaped relationship between BP and outcome was most prominent within the first 180 days following the index ischemic stroke event and was seen for both the primary outcome (very low–normal SBP group: 3.33%, low-normal SBP group: 2.49%, high-normal SBP group: 2.26%, high SBP group: 3.01%, and very high SBP group: 5.92%; after 180 days: very low–normal SBP group: 4.64%, low-normal SBP group: 4.72%, high-normal SBP group: 4.50%, high SBP group: 5.71%, and very high SBP group: 8.17%) and the secondary outcome (very low–normal SBP group: 4.43%, low-normal SBP group: 3.47%, high-normal SBP group: 2.93%, high SBP group: 3.70%, and very high SBP group: 7.35%; after 180 days: very low–normal SBP group: 8.03%, low-normal SBP group: 7.89%, high-normal SBP group: 7.26%, high SBP group: 9.38%, and very high SBP group: 12.40%).

Quiz Ref IDWe observed that among individuals with a recent non–cardioembolic ischemic stroke, risk of further vascular events was significantly higher for those patients with a mean SBP level below 120 mm Hg than for those patients with a level between 130 to 139 mm Hg, even after adjusting for major confounders including stroke subtype. Of note, these findings were independent of BP level at baseline, follow-up, and change in level, and occurred within a trial conducted in an era of widespread background risk factor treatment with antiplatelet therapies, statins, and use of multiple antihypertensive drug classes.

Our results indicate that there may indeed be thresholds of benefit or harm with regard to short-term to longer-term SBP levels after a recent non–cardioembolic ischemic stroke, and imply that clinicians regularly caring for stroke patients in the outpatient setting may need to be vigilant about how low a given patient's BP is within the normal range to promote favorable outcomes. On the one hand, the link between low-normal SBP level and untoward vascular risk after stroke could simply represent a marker of known or unknown poor global health, keying health care practitioners to facilitate optimal management of the patient's known chronic severe conditions or to consider further inquiry into potentially undiagnosed major medical problems. In fact, the J-shaped relationship of SBP with vascular morbidity and mortality has been attributed mainly to prevalent medical conditions and not specifically to BP-lowering treatment.¹⁴Quiz Ref IDOn the other hand, our results included adjustments for several major health conditions (eg, congestive heart failure [inability of heart to support normal BP], body mass index [cachexia]) as well as baseline BP level, the latter of which could be an indirect indicator of overall health status suggesting that very low–normal SBP levels during follow-up may not have been a reflection of an underlying severe medical condition at baseline. Furthermore, harm with very low–normal SBP levels was not just associated with all-cause mortality, but for the individual vascular end points as well including stroke.

Quiz Ref IDFollowing the Action to Control Cardiovascular Risk in Diabetes-Blood Pressure ( ACCORD-BP) trial results showing no vascular protective benefit for targeting SBP levels below 120 mm Hg in patients at high risk for vascular events, but higher incidence of adverse events,³ as well as post hoc trial data suggesting no benefit for targeting SBP levels below 130 mm Hg,⁴ the broadly held notion that vascular risk continues to decline well into the normal range¹⁵ should be viewed with some caution. In ACCORD-BP, the only trial so far to specifically investigate this issue, there were significantly fewer strokes in the more intensively treated group of patients with diabetes, but it is not clear what proportion of these patients had previous strokes (34% of the cohort reportedly had a history of previous cardiovascular events); and the ACCORD-BP trial's secondary end point of stroke was rather scarce (approximately 100 events).³ Our analysis of the 5743 patients with diabetes and a recent ischemic stroke in the PROFESS trial was more consistent with the primary composite outcome (nonfatal MI, nonfatal stroke, or death) of ACCORD-BP and found no significant differences in outcome between patients with very low–normal SBP levels and high-normal SBP levels. Quiz Ref IDIt may be prudent to shift guideline emphasis back to less than 140 mm Hg for SBP and less than 90 mm Hg for DBP.

Our results also challenge perceptions that the J-shaped curve applies largely to the cardiovascular arterial bed and not the cerebrovascular bed.^16- 18 For instance, an observational analysis of the Perindopril Protection Against Recurrent Stroke Study (PROGRESS) investigating the association between achieved follow-up BP levels and recurrent stroke risk showed that the lowest risk of recurrence was among the one-quarter of participants with the lowest follow-up BP levels (median SBP of 112 mm Hg and DBP of 72 mm Hg), and that risks increased progressively with higher follow-up BP levels.¹⁹ However, the PROGRESS trial enrolled patients with a history of ischemic or hemorrhagic stroke, which are mechanistically quite different, as well as patients with transient ischemic attacks. As such, the PROGRESS patient population was more heterogeneous than that of PROFESS, presumably more stable and less vulnerable to the effects of low-normal SBP levels given the enrollment time from their index event. Observational studies among other patients at high risk for vascular events generally without established symptomatic cerebrovascular disease also indicate that risk of stroke is especially favorably influenced by low-normal BP levels,^17,20- 21 but it is conceivable patients free of known stroke may tolerate low-normal BP levels better than those with a recent bout of symptomatic cerebral ischemia with brain parenchymal damage.

Timing may be important. We found the J-shaped association of SBP with recurrent vascular risk after stroke to be most pronounced in the first 90 to 180 days after the qualifying event. PROGRESS enrolled patients within 5 years of their index event,¹⁹ while trials of patients generally without symptomatic cerebrovascular disease enrolled most patients more than 1 year after their index event.¹⁸ The results of our study parallel those seen in an analysis of acute ischemic stroke where a J-shaped curve was observed²² and low normal SBP was associated with a higher risk of early recurrence by 2 weeks and poor functional outcome at 6 months compared with high normal SBP. Similarly, in the acute ischemic stroke study, both heart failure and large cortical strokes were more common in the group with low normal SBP.²² More recently, results of a randomized trial in patients with acute stroke and raised BP levels (SBP ≥140 mm Hg) suggested a trend toward greater risk of poor functional outcome at 180 days after careful BP-lowering treatment was initiated within 30 hours of the index stroke.²³

Quiz Ref IDThere were significant interactions between some variables and mean SBP on outcome in this study, including the association of older age with more events in the low-normal SBP level group. This finding may not be surprising because epidemiological data indicate that BP and the risk of death are inversely related among the very elderly, likely due to higher risks of antihypertensive treatment or reverse causation by disease entities linked with low SBP such as MI, congestive heart failure, dementia, and cancer.^24- 25 Of note, we adjusted for the baseline history of MI and congestive heart failure in our study. Interestingly, there were fewer vascular events among those in both normal SBP level groups who received telmisartan (vs placebo), but more events among those in the high SBP level group who received telmisartan. Several trials have hinted that modulators of the renin angiotensin system may provide BP-independent benefits to patients with stroke,^26- 27 so it is possible that such benefits might be more apparent when BP remains under control, but this finding warrants further investigation.

This study has several limitations. It was a post hoc analysis of a completed randomized trial (ie, patients were not randomized a priori to the respective BP level groups but instead were divided according the consistency of BP levels with the context of the study), and so, despite controlling for established prognosticators, we cannot exclude the possibility that unmeasured confounding may explain some of our findings. Office-based BP determinations may not adequately detect major differences in BP measured outside of a physician's office, but assessing persistence and consistency of BP levels as we did may have mitigated this weakness slightly.²⁸ Although the vast majority of ischemic strokes are noncardioembolic in nature,¹ the PROFESS trial excluded patients with cardioembolic stroke, so these results may not apply to patients whose index stroke is due to a presumed cardioembolic mechanism. Our results are strengthened by the rigorous procedures of the PROFESS trial design, large sample size (>20 000 patients), multiple visits for each patient (at 3, 6, 12, 18, 24, 30, 36, and 42 months), inclusion of patients enrolled from around the world (generalizability), and general consistency across individual end points.

In conclusion, these data are hypothesis generating and the notion that aggressively and consistently lowering BP levels within the normal range in the short term to longer term after an index ischemic stroke is not beneficial remains unproven, and will require the conduct of dedicated clinical trials comparing intensive with usual BP reduction in the stable follow-up period after a stroke. The ongoing Prevention of Decline in Cognition After Stroke Trial (PODCAST) trial is examining the impact of achieving low-normal with high-normal SBP levels on cognition after recent ischemic or hemorrhagic stroke,²⁹ while the Secondary Prevention of Small Subcortical Strokes trial is evaluating a subset of ischemic stroke patients (those eligible for magnetic resonance imaging with small vessel disease strokes) within a broader time window (6 months of ictus) and with higher SBP cutoffs (<150 mm Hg vs <130 mm Hg).³⁰ However, in the meantime, the results of this analysis support aiming for consistent SBP levels of less than 140 mm Hg and less than 90 mm Hg for DBP among recent ischemic stroke patients, particularly within the first 6 months.