Hemorrhagic stroke in the Stroke Prevention by Aggressive Reduction in Cholesterol Levels study

The Stroke Prevention with Aggressive Reductions in Cholesterol Levels (SPARCL) trial was a prospective, double-blind, randomized clinical trial which showed that treatment with a HMG-CoA reductase inhibitor (atorvastatin 80 mg per day) resulted in a 16% reduction in the combined risk of fatal and nonfatal stroke in patients with a recent (within 1 to 6 months) stroke or TIA and no known coronary heart disease (11.2% vs 13.1% over 4.9 years; HR 0.84, 95% CI 0.71 to 0.99, p = 0.03).¹ A post hoc analysis found the overall benefit of treatment included an increase in the numbers of patients having hemorrhagic stroke (n = 55 for active treatment vs n = 33 for placebo; unadjusted HR 1.68, 95% CI 1.09 to 2.59) with no difference in the incidence of fatal hemorrhagic stroke between the groups (17 in the atorvastatin and 18 in the placebo group).

A variety of clinical factors are associated with an increased risk of hemorrhagic stroke including advancing age, hypertension, cigarette smoking, use of antithrombotic medications, lower blood glucose, and having a prior hemorrhagic stroke.^2–4 Although meta-analysis of previous statin trials carried out predominately in patients with coronary heart disease found no relationship between statin treatment and hemorrhagic stroke risk,^5,6 epidemiologic studies show a relationship between low cholesterol levels and hemorrhagic stroke.^7–10 In SPARCL, atorvastatin treatment was associated with a mean post-randomization low-density lipoprotein (LDL) cholesterol of 72.9 ± 0.5 mg per deciliter (1.88 ± 0.01 mmol per liter) vs 128.5 ± 0.5 mg per deciliter (3.32 ± 0.01 mmol per liter) with placebo treatment (p < 0.001).¹ This analysis explores the relationships between baseline patient characteristics, atorvastatin treatment, most recent blood pressure, and LDL cholesterol levels and the risk of hemorrhagic stroke in patients enrolled in the SPARCL trial.

METHODS

The methods of the SPARCL study have been described in detail previously.^1,11 The local research ethics committee or institutional review board at each participating study center approved the study protocol (15 of 205 centers excluded otherwise suitable patients with an LDL cholesterol level above 160 mg per deciliter [4.1 mmol per liter], as required by their institutional review boards), and all patients gave written informed consent. The primary hypothesis of the SPARCL trial was that treatment with 80 mg of atorvastatin per day would reduce the combined risk of fatal and nonfatal stroke among patients with a history of stroke or TIA. Eligible patients were men and women over 18 years of age who had had an ischemic or hemorrhagic stroke or a TIA (diagnosed by a neurologist within 30 days after the event) 1 to 6 months before randomization. Stroke was defined by focal clinical signs of CNS dysfunction of vascular origin that lasted for at least 24 hours; TIA was defined by the loss of cerebral or ocular function for less than 24 hours. Patients had to be ambulatory, with a modified Rankin score of no more than 3 (scores can range from 0 to 5, with higher scores indicating more severe disability), and to have an LDL cholesterol level of at least 100 mg per deciliter (2.6 mmol per liter) and no more than 190 mg per deciliter (4.9 mmol per liter). Patients who were taking lipid-altering drugs had to stop these medications 30 days before the screening phase of the study and these drugs were prohibited during the course of the trial. Excluded patients included those with atrial fibrillation, mechanical prosthetic heart valves, or subarachnoid hemorrhage. The investigators categorized stroke subtype as ischemic (large vessel atherothromboembolic, cardioembolic, small vessel, other etiology, or unknown cause), hemorrhagic, other, or unable to be determined based on their clinical judgment (diagnostic criteria were not provided and stroke subtype was not adjudicated). Patients with hemorrhagic stroke (2% of the study population) could be included if they were deemed by the investigator to be at risk for ischemic stroke or coronary heart disease. Subjects were enrolled between September 1998 and March 2001.

The primary outcome was the time from randomization to a first nonfatal or fatal stroke. An independent endpoint committee adjudicated all potential endpoints without knowledge of the patients’ treatment status or cholesterol levels. In a post hoc analysis, strokes occurring after randomization were categorized as being ischemic, hemorrhagic, or unclassified based on standard clinical and radiographic criteria.

The SPARCL steering committee developed the study protocol with the sponsor and takes responsibility for the data and data analyses. Medpace (Cincinnati) managed all data. Medpace, Charles River Laboratories Clinical Services (Brussels), and the sponsor provided site monitoring throughout the study. A data and safety monitoring board with independent statistical support performed interim monitoring analyses for safety and efficacy.

For the current secondary analyses, we first explored relationships between individual baseline factors and time to hemorrhagic stroke in separate Cox regression models with adjustment for treatment. Interactions between the factors and treatment assignment were assessed. A factor for treatment and individual baseline factors significant at the <0.10 level were subsequently entered into a single multivariable model and backward elimination removed factors from the model that did not remain significant (p > 0.10).

The effects of two post-randomization variables on the risk of hemorrhagic stroke were then assessed with adjustment for significant variables from the multivariable baseline risk factor model. The association between blood pressure during follow-up and the risk of hemorrhagic stroke was evaluated using the Joint National Committee–7 (JNC-7) classification categories (normal, systolic blood pressure [SBP] <120 mm Hg and diastolic blood pressure [DBP] <80 mm Hg; pre-hypertension, SBP 120 to 139 mm Hg or DBP 80 to 89 mm Hg; stage 1 hypertension, SBP 140 to 159 mm Hg or DBP 90 to 99 mm Hg; stage 2 hypertension, SBP ≥160 mm Hg or DBP ≥100 mm Hg)¹² as a time-varying covariate in a Cox regression model. The value of the time-varying covariate for each subject was updated each time a subject’s blood pressure was measured prior to a hemorrhagic stroke or prior to the end of follow-up for hemorrhagic stroke. The interaction between the time-varying covariate and treatment group was also evaluated. A multivariable analysis including LDL-cholesterol levels during follow-up in the atorvastatin group was also performed with LDL cholesterol as the time-varying covariate.

RESULTS

The figure gives the Kaplan-Meier curves and unadjusted hazard ratios for the occurrence of all fatal and nonfatal (top panel), fatal (middle panel), and nonfatal (bottom panel) ischemic and hemorrhagic strokes based on the intention to treat subjects with atorvastatin or placebo (the net difference in statin use because of drop-ins and drop-outs between groups was 78%¹), regardless of the type (i.e., ischemic or hemorrhagic) of entry event. The median follow-up was 4.9 years. A 21% reduction in fatal and nonfatal ischemic stroke (unadjusted HR 0.79, 95% CI 0.66 to 0.95) is partially attenuated by an increased risk of hemorrhage (unadjusted HR 1.68, 95% CI 1.09 to 2.59) resulting in the previously reported, treatment-related 16% overall reduction in the risk of fatal and nonfatal stroke (adjusted hazard ratio = 0.84; 95% CI 0.71 to 0.99, p = 0.03; unadjusted p = 0.05).¹ The figure shows that the risk of ischemic stroke was higher than hemorrhagic stroke in both treatment groups early after randomization and remained so throughout follow-up. Of those randomized to atorvastatin, 2.3% had a hemorrhagic stroke as compared to 1.4% of those randomized to placebo. Subgroup analyses showed a treatment-associated reduction in the risk of fatal ischemic stroke with no difference in the rate of fatal hemorrhagic strokes.

• Download figure
• Open in new tab
• Download powerpoint

Figure Kaplan-Meier curves (with unadjusted hazard ratios) for the occurrence of all fatal and nonfatal stroke, fatal stroke, and nonfatal stroke

Kaplan-Meier curves (with unadjusted hazard ratios) for the occurrence of all fatal and nonfatal stroke (top panel), fatal stroke (middle panel), and nonfatal stroke (bottom panel) for patients randomized to atorvastatin 80 mg per day or placebo within 1 to 6 months after a nondisabling stroke or TIA, regardless of the type (i.e., ischemic or hemorrhagic) of entry event.

Table e-1 on the Neurology® Web site at www.neurology.org gives baseline characteristics of subjects who later did or did not have a post-randomization hemorrhagic stroke with the associated unadjusted univariable hazard ratios based on Cox regression models (a table providing comparisons of the baseline characteristics of patients randomized to atorvastatin or placebo has been published¹). In addition to randomization to atorvastatin treatment, the risk of hemorrhagic stroke was higher in those with hemorrhage as the entry event, was increased in men, increased with age, and tended to be higher in those with a history of hypertension. There were no statistical interactions between any of the significant baseline factors and atorvastatin treatment for the risk of hemorrhage. Time since entry event, baseline LDL or total cholesterol, smoking status, or the use of antiplatelet agents or anticoagulants did not affect the risk of hemorrhagic stroke.

Table 1 gives the numbers of patients having an outcome ischemic, hemorrhagic, or any stroke and associated unadjusted hazard ratios based on the type of entry event as designated by the investigators. Those having a hemorrhagic stroke as an entry event had an overall higher risk of an outcome stroke with treatment (HR 2.82, 95% CI 0.89 to 9.01).

Cox multivariable regression including baseline variables significant in the univariable analysis showed that having a hemorrhagic stroke as the entry event (HR 5.65, 95% CI 2.82 to 11.30, p < 0.001), male sex (HR 1.79, 95% CI 1.13 to 2.84, p = 0.01), atorvastatin treatment (HR 1.68, 95% CI 1.09 to 2.59, p = 0.02), age (10 y increments, HR 1.42, 95% CI 1.16 to 1.74, p = 0.001), but not a history of hypertension (HR 1.41, 95% CI 0.88 to 2.25, p = 0.15) were independently associated with the risk of hemorrhagic stroke. The overall model explained 1% of the risk of hemorrhagic stroke suggesting most of the risk is related to unmeasured factors (model R² = 0.009). Clinical factors aside from randomization to treatment with atorvastatin were associated with 86% of the variance explained by the model.

Table 2 gives the Cox multivariable regression analysis evaluating the effects of post-randomization time-varying blood pressure, adjusting for significant variables from the multivariable baseline risk factor model (four factor model excluding a history of hypertension). Blood pressure categorization was based on 59,547 measurements (mean of 12.6 measurements per subject); 9% were in the normal, 37% in the pre-hypertension, 36% in the stage 1 hypertension, and 18% in the stage 2 hypertension range. Increasing blood pressure was independently associated with an increased risk of hemorrhagic stroke (four-category time-varying blood pressure, p = 0.01) with those having a blood pressure in the stage 2 hypertension range at highest risk.

Table 3 gives the Cox multivariable regression analysis evaluating the effects of post-randomization time-varying LDL-cholesterol level (divided in quartiles) in the atorvastatin-randomized patients (based on 27,649 values in 2,365 patients), adjusting for significant variables from the multivariable baseline risk factor model. The median LDL cholesterol was 66 mg per deciliter (1.7 mmol per liter) with few values >100 mg per deciliter (2.6 mmol per liter). There was no relationship between LDL-cholesterol level and the risk of hemorrhagic stroke. Very low LDL-cholesterol levels (less than 40 mg per deciliter [1.0 mmol per liter]) were not associated with increased risk, and we found no threshold below which the risk was increased.

DISCUSSION

In addition to treatment with atorvastatin, this exploratory analysis found that having hemorrhagic stroke as an entry event, male sex, and advancing age at baseline accounted for 86% of the increased risk of hemorrhagic stroke explained by the Cox regression model, with the largest risk being associated with having a hemorrhagic stroke as the qualifying event at baseline. Although each of these factors and treatment assignment were related to the risk of having such an event during follow-up, they explained a relatively small proportion of the overall risk (about 1%). The impact of baseline characteristics on hemorrhagic stroke risk was similar in those randomized to atorvastatin vs placebo (i.e., these factors, including hemorrhagic stroke as the entry event, did not disproportionately increase risk in atorvastatin-treated patients).

The risk of recurrent hemorrhagic stroke is estimated at 2.5% per year.^2,13 It is, therefore, not surprising that having a hemorrhagic stroke as an entry event was associated with having an outcome hemorrhage. In SPARCL, investigators were permitted to randomize patients with a hemorrhagic stroke if they were deemed to be at risk for ischemic stroke or coronary heart disease. A prior study found that the rate of recurrent hemorrhagic stroke and ischemic events were similar.² Although the numbers were small and the CIs were wide, there were excess numbers of treatment-associated outcome ischemic and hemorrhagic strokes among patients with an entry hemorrhage whereas those with entry ischemic strokes, regardless of investigator-designated subtype, benefited (table 1). As shown in table 1, there was an increase in outcome hemorrhagic strokes in subjects with an investigator-designated small-vessel distribution stroke at entry (possibly contributing to some of the unexplained variance in hemorrhage risk). Because of the high risk of false-positive findings, conclusions regarding treatment effects on secondary outcomes in isolated subgroups may lack validity. Those with an entry small-vessel distribution stroke also had a reduction in outcome ischemic strokes resulting in a total benefit similar to the overall study cohort (table 1).

A cohort study of patients with prior stroke or TIA found that increasing age, lower blood glucose, systolic blood pressure, and the use of antihypertensive medications were associated with an increased risk of hemorrhagic stroke.⁴ A history of hypertension at baseline tended to be associated with increased risk of hemorrhagic stroke based on univariable analyses, a trend lost after accounting for other factors. An independent relationship between blood pressure recorded at the last assessment prior to a hemorrhagic stroke and bleeding risk was found with the largest risk being in those with stage 2 hypertension. These data support the need for aggressive management of hypertension to reduce the risk of hemorrhagic stroke.^14,15 We do not have data on blood glucose at the time of hemorrhagic stroke and cannot address the possible role of lower levels of blood glucose on this risk.

Epidemiologic studies have found an association between low cholesterol levels and an increased risk of hemorrhagic stroke,^7–10 a relationship not found in more recent clinical trials of statins given for coronary heart disease, including in those patients with major reductions in LDL cholesterol.^5,6,16 Consistent with these observations, we found no relationships between the baseline levels of either total or LDL cholesterol and the risk of hemorrhagic stroke, no disproportionate increase in the risk of bleeding associated with treatment based on baseline cholesterol levels, and no independent effect of LDL-cholesterol levels at the last measurement prior to a hemorrhagic stroke in those treated with atorvastatin. Moreover, we found no threshold of LDL cholesterol below which the risk of hemorrhagic stroke was increased.

Anticoagulants and some antiplatelet regimens may be associated with increased risk of poststroke brain hemorrhage.^17,18 We, however, found no overall effect of antiplatelet drugs or anticoagulants on the risk of brain hemorrhage in SPARCL. Despite statins having antithrombotic properties,^19–22 we also found no statistical interaction between treatment and the use of any individual or combination of antithrombotic drugs (table e-1). An increase in the risk of hemorrhagic stroke related to a statin’s antithrombotic properties cannot be excluded.

Post hoc analysis of data from patients with prior cerebrovascular disease enrolled in the Heart Protection Study found a non-significant increase in hemorrhagic stroke in those treated with simvastatin 40 mg per day vs placebo (n = 21, 1.3% vs n = 11, 0.7%).²³ The overall number of strokes was small, and the Heart Protection Study was not powered to detect these differences. Treatment-associated hemorrhagic stroke in SPARCL remained after accounting for the other factors included in this exploratory analysis (table 2). Although meta-analysis of over 90,000 subjects in statin trials of patients with coronary heart disease found no increase in the risk of hemorrhagic stroke associated with statin treatment,^5,6 it is possible that patients with a prior stroke or TIA are at greater risk of statin-related bleeding. Nevertheless, it remains uncertain whether there is a difference in the risk of bleeding in statin-treated patients with prior stroke compared to those with coronary heart disease, given the small numbers of patients with hemorrhagic stroke in these trials and the post hoc nature of the analyses.

No baseline (including having a hemorrhagic stroke as an entry event) or post-randomization factors were identified that disproportionately increased bleeding in treated compared with placebo patients, but treatment-associated risk remained after adjustment for other significant factors. Although not significant, the numbers of patients with an entry hemorrhagic stroke having an outcome ischemic or hemorrhagic stroke was greater in treated vs placebo patients. Therefore, unlike those having an entry ischemic stroke, there is no evidence that those having a hemorrhagic stroke at baseline benefited from treatment. It is, however, important to re-emphasize the exploratory nature of these analyses that are useful for hypothesis generation, but cannot be conclusive. Outcome hemorrhagic strokes occurred in less than 2% of the study population, the observation was found in a post hoc analysis, and the exploratory statistical models account for only a small proportion of bleeding. In making therapeutic decisions, the increase in the risk of hemorrhagic stroke found in SPARCL, if not due to chance, must be balanced against the benefit of treatment with atorvastatin 80 mg per day in reducing the overall risk of stroke, as well as other cardiovascular events found in the study’s prespecified, intention-to-treat analysis.