Impact of Statins on Hematoma, Edema, Seizures, Vascular Events, and Functional Recovery After Intracerebral Hemorrhage

Introduction

Statin treatment is effective in the prevention of cardiovascular events.¹ Based on the results of the SPARCL trial (Stroke Prevention by Aggressive Reduction in Cholesterol Levels), statin treatment has been discussed to increase the incidence of intracerebral hemorrhage (ICH).² However, after adjustment for lipid levels, statin treatment seemed associated with a lower incidence of ICH presumably by effects on cerebral small-vessel disease.^2–4 In the event of an ICH, statin premedication is often continued in clinical practice, although there is uncertainty if statins should be discontinued in the acute phase of ICH and it is unclear, if and when statins should be initiated in statin-naive patients after ICH.^5,6 Observational studies suggest initiation of statins after ICH to be associated with favorable functional outcome and fewer remote symptomatic seizures.^7,8 However, these results may harbor residual bias by indication, why the impact of early statin initiation on secondary injury after the initial ICH needs to be established.^9,10 The European Stroke Organization, therefore, defined the issue of when statins should be (re-)started after ICH as major priority in ICH research.¹¹ The present study explored the hypothesis that statin treatment may influence hematoma characteristics (1), extent of perihemorrhagic edema (PHE; 2), occurrence of remote symptomatic seizures (3), cardiovascular events and functional recovery (4) following ICH.

Methods

The authors declare that all supporting data are available within the article and in the Data Supplement. Detailed information on the prospective, single-center UKER-ICH registry (Universitätsklinikum Erlangen Cohort of Patients With Spontaneous Intracerebral Hemorrhage) have been published previously.^12,13 Patients admitted to the University Hospital Erlangen with spontaneous ICH were included between January 1, 2006, and December 31, 2015. The UKER-ICH study was approved by the local institutional review board, and consent was obtained by patients or legal representatives.

Definitions and Data Acquisition

Data on demographics, comorbidities, and premorbid modified Rankin Scale (mRS) were obtained as previously described.^14,15 Laboratory and clinical parameters were retrieved from medical charts and institutional databases. If lipid profile was not measured during hospital stay and prehospital cholesterol levels available, these laboratory parameters were used.

ICH and PHE

We excluded ICH due to secondary cause—such as aneurysm, arteriovenous malformation, tumor, or trauma.¹⁵ All available imaging scans during hospital stay were evaluated to assess hematoma characteristics and PHE volume. Deep ICH was defined as hematoma located within the basal ganglia, internal or external capsula, or thalamus. Lobar ICH was defined as hematoma originating at the cortex and cortical-subcortical junction. PHE volumes were measured by a semiautomatic volumetric algorithm using region of interest with a specified range of Hounsfield units.¹⁶ Peak PHE was defined as the maximum PHE volume measured in any of the available imaging scans (in patients with ≥3 scans during hospital stay).^17,18 ICH volumes were measured by the ABC/2 method (A is defined as the largest axial diameter of the bleeding, B the diameter 90° to A, and C the craniocaudal diameter of the hemorrhage) and ABC/3 method, respectively.¹⁹ Hematoma enlargement was defined as relative ICH volume increase of >33% from initial to follow-up imaging.¹⁵

Statin Treatment

We defined statin treatment as daily statin medication independent of drug or dosage. There are different statin treatment groups in this study. For analysis of hematoma characteristics statin treatment was defined as statin treatment at hospital admission. For analysis of PHE statin treatment was sub-classified according to continuation, discontinuation, and initiation of statin therapy during hospital stay. For analysis of remote seizures, cardiovascular adverse events and functional recovery statin treatment was defined as statin treatment at hospital discharge. Exposure to statins at hospital admission was classified according to duration of statin intake before hospital admission (<1, 1–2, or >2 years), statin type (atorvastatin, simvastatin, or hydrophilic statins), and statin dosage. Average atorvastatin equivalent daily dose was calculated and average atorvastatin equivalent daily dose categories defined as <20 and ≥20 mg/d.³

Epileptic Seizures

We defined remote symptomatic seizures as the occurrence of a first seizure event >7 days after the initial ICH.²⁰ Seizure events were defined as (1) electrographic evidence of ictal activity on EEG with or without corresponding clinical symptoms or signs or (2) witnessed clinical symptoms or signs diagnosed as epileptic seizure by the attending neurologist or (3) unwitnessed clinical events evaluated as epileptic seizure by an epilepsy specialist.²⁰ Data on seizure events were assessed during hospital stay, and thereafter by standardized mailed questionnaires or semi-structured telephone-interviews at 3 and 12 months, or retrieved from institutional databases in case of hospital readmission.¹⁵

Functional Recovery and Cardiovascular Events

Functional recovery was defined as improvement of functional status (at least one point on the mRS) between hospital discharge and outcome assessment at 3 months and between hospital discharge and outcome assessment at 12 months after ICH, respectively. We defined cardiovascular adverse events as composite of nonfatal stroke (ischemic or hemorrhagic), nonfatal myocardial infarction, and death of any cause.

Outcome Measures

We assessed (1) the impact of statin premedication on hematoma characteristics on admission, (2) the extent of PHE according to continuation, discontinuation, and initiation of statins, (3) the influence of statin treatment on the occurrence of remote symptomatic seizures, as well as (4) the influence of statin treatment on the occurrence of cardiovascular adverse events and functional recovery after ICH.

Statistical Analysis

SPSS version 21.0 (www.spss.com), R version 2.12.0 (www.r-project.org), and SAS version 9.4 (www.sas.com) were used for statistical analyses. We performed 2-sided statistical tests with a significance level at α=0.05. Categorial variables are expressed as counts (percentage) and compared by the Pearsons χ² test. Normally distributed data are presented as mean (±SD) and analyzed using the Student t test. Non-normally distributed data are presented as median (interquartile rage) and compared using the Mann-Whitney U test.

Propensity score matching (PSM) was applied using parallel, balanced, ratio (1:3) nearest-neighbor approach at a caliper of 0.1. The propensity score was calculated using parameters showing differences for intergroup comparisons (standardized mean differences >0.10). For analysis of hematoma characteristics parameters were included showing differences for intergroup comparisons regarding baseline characteristics (ie, age, arterial hypertension, diabetes, prior ischemic stroke/transient ischemic attack [TIA], coronary artery disease, premorbid mRS), and lipid level (ie, cholesterol). Subgroup analysis for the association between statin premedication and lobar hematoma location was conducted in the PSM cohort. For analysis of remote symptomatic seizures parameters were included showing differences for intergroup comparisons regarding baseline characteristics (ie, arterial hypertension, diabetes, prior ischemic stroke/TIA, coronary artery disease), and hematoma characteristics (ie, ICH volume). For analysis of outcome events (cardiovascular adverse events and functional recovery), parameters were included showing differences for intergroup comparisons regarding baseline characteristics (ie, arterial hypertension, diabetes, coronary artery disease), characteristics at hospital discharge (ie, mRS, antiplatelet therapy), and hematoma characteristics (ie, ICH volume). After adjustment using PSM, all relevant covariates were well balanced with an absolute standardized difference of 0.1 (Figures I through III in the Data Supplement). Kaplan–Meier method was used to evaluate the timing of event occurrences. Binary logistic regression analyses were applied for analysis of functional recovery. Occurrence of remote symptomatic seizures and cardiovascular adverse events according to statin exposure were compared by Cox proportional hazards models. Cumulative incidence function covariate analyses were performed to account for the competing risk of death as it may preclude the occurrence of remote symptomatic seizures. Sensitivity analyses were conducted using inverse probability treatment weighting method adjusted for the same set of variables used for the corresponding outcome analyses. Additional sensitivity analyses were conducted for established parameters associated with outcome events using cox proportional hazards models for remote symptomatic seizures and cardiovascular adverse events, and multivariable binary regression analyses for hematoma characteristics and functional recovery (results).

Peak PHE volumes were log-transformed to approximate the normal distribution.²¹ Associations between different statin regimes and peak PHE were analyzed by multivariable linear regression adjusted for age and ICH volume. To account for varying time points of control imaging, we categorized time frames of follow-up imaging (days 1, 2–3, 4–6, 7–9, 10–12, 13–15, 16–18, 19–22).^16,18,22

Results

Study Population and Clinical Characteristics

The UKER-ICH study included 1322 patients with primary ICH (Figure 1). After exclusion of 47 patients because of missing data on statin treatment, 1275 patients with ICH remained for analysis. Of these, 277 (21.7%) patients were on statin treatment on hospital admission (Table 1). As compared with nonstatin-users upon admission, patients with statins were significantly older (76 [69–80] years versus 72 [61–80] years; P<0.001) and demonstrated more pronounced comorbidities—for example, arterial hypertension (259/277 [93.5%] versus 821/998 [82.3%]; P<0.001), prior ischemic stroke/TIA (104/277 [37.5%] versus 151/998 [15.1%]; P<0.001), coronary artery disease (126/277 [45.5%] versus 180/998 [18.0%]; P<0.001)—and lower total cholesterol levels (174 [148–207] mg/dL versus 191 [163–219]; P<0.001). Overall, missing data was below 2.0% except for symptom onset to admission and cholesterol levels (results in the Data Supplement).

• Download figure
• Download PowerPoint

Hematoma Characteristics According to Statin Premedication

To properly compare hematoma characteristics according to statin premedication, we excluded patients with oral anticoagulation (OAC), that is, vitamin K antagonist and non–vitamin K antagonist oral anticoagulant—associated ICH (228/1275 [17.9%]) and performed PSM for parameters showing differences for intergroup comparisons (ie, age, arterial hypertension, diabetes, prior ischemic stroke/TIA, coronary artery disease, premorbid mRS, and cholesterol) to rule out possible influence on imaging characteristics (Figure I in the Data Supplement). After adjustments, there were significant differences regarding ICH location between patients with and without statins on hospital admission (lobar ICH, 71/125 [56.8%] versus 130/285 [45.6%]; P=0.037; Table 1). There were no significant differences in ICH volume, hematoma enlargement, and intraventricular hemorrhage.

Subgroup analysis revealed a significant association between statins and lobar hematoma location (odds ratio [OR], 1.57 [1.03–2.40]; P=0.038; Figure 2), including subgroups according to antiplatelet medication, age, statin type, statin treatment duration, and statin dosage (average atorvastatin equivalent daily dose <20 mg/d: OR for the association with lobar hematoma location: 1.40 [0.84–2.34]; average atorvastatin equivalent daily dose ≥20 mg/d: OR, 1.85 [1.01–3.40]). Sensitivity analyses yielded similar results (results in the Data Supplement).

• Download figure
• Download PowerPoint

PHE in Different Statin Regimes

Table 2 illustrates the impact of different statin regimes on peak PHE extent. There were no significant associations between peak PHE and statin therapy on admission (n=73 versus 293, β=−0.01, SE=0.05; P=0.853) or any statin therapy during hospital stay (n=123 versus 243, β=0.04, SE=0.07; P=0.070). Initiation versus no initiation of statins was associated with increased peak PHE (n=50 versus 243, β=0.12, SE=0.06; P=0.008), whereas continuation versus discontinuation of statin therapy was not significantly associated with increased PHE (n=39 versus 34, β=0.11, SE=0.08; P=0.216).

Seizures, Cardiovascular Events, and Functional Recovery

To properly compare occurrence of remote symptomatic seizures according to statin premedication, we excluded patients with OAC associated ICH (228/1275 [17.9%]) and performed PSM for parameters showing differences for intergroup comparisons (arterial hypertension, diabetes, prior ischemic stroke/TIA, coronary artery disease, and ICH volume; Figure II in the Data Supplement). Remote symptomatic seizures occurred in 69 of 769 patients within 12 months after ICH—28 patients with statin therapy and 41 patients without statin therapy (crude incidence: statins: 11.5% versus no statins: 7.8%; subdistribution hazard ratio [HR], 1.15 [0.80–1.66], P=0.512; cause-specific HR, 1.36 [0.84–2.19], P=0.214; Figure 3A). Remote symptomatic seizures occurred in 107 patients within 5 years after ICH—43 patients with statin therapy and 64 patients without statin therapy (crude incidence: statins: 17.6% versus no statins: 12.2%; subdistribution HR, 1.31 [0.94–1.69], P=0.156; HR, 1.26 [0.85–1.85], P=0.251; Figure IVA in the Data Supplement).

• Download figure
• Download PowerPoint

To avoid bias by ischemic and hemorrhagic events in patients with atrial fibrillation, we excluded patients with atrial fibrillation from analyses of outcome events (cardiovascular adverse events and functional recovery) and performed PSM for parameters showing differences for intergroup comparisons (arterial hypertension, diabetes, coronary artery disease, mRS, antiplatelet therapy, and ICH volume; Figure III in the Data Supplement). There were differences in the incidence of cardiovascular adverse events within 12 months after the ICH event according to statin exposure at hospital discharge (crude incidence: statins: 11.6% versus no statins: 18.5%; HR, 0.60 [0.36–1.02], P=0.058; Figure 3B) and 5 years (20.9% versus 33.3%; HR, 0.56 [0.38–0.83], P=0.004; Figure IVB in the Data Supplement).

There were no significant differences in the proportion of patients with functional recovery among those with and without statin treatment after 3 months (statins: 62/150 [41.3%] versus no statins: 103/251 [41.0%]; OR, 1.01 [0.67–1.53], P=0.953). There were significant differences in the proportion of patients with functional recovery after 12 months (79/137 [57.7%] versus 103/229 [45.0%]; OR, 1.67 [1.09–2.56], P=0.019). ICH location had no significant interaction with the effect of statin therapy on the incidence of remote symptomatic seizures, cardiovascular adverse events of functional recovery (results in the Data Supplement). Sensitivity analyses yielded similar results (results in the Data Supplement).

Discussion

In this study, we demonstrate that continuation of statin premedication does not aggravate edema formation and, therefore, appears to be safe in patients with ICH. However, initiation of statin therapy during the first days after ICH in patients without statin premedication is associated with increased PHE extent. Thus, if indicated in statin-naive patients, statins should be initiated after the acute phase of ICH (eg, at hospital discharge) as statins seem to prevent cardiovascular events and improve functional recovery during follow-up.

More than 30% of patients with ICH require statin treatment due to comorbidities such as coronary artery disease or previous ischemic stroke. However, in the event of an ICH, it was unclear if statin premedication should be continued or discontinued and if statins should be initiated in patients without such premedication. Recent observational data derived from the Virtual International Stroke Trials Archive suggest that continuation of prior statin use during the first 48 hours is associated with better functional outcomes.⁶ Preclinical studies have reported potential neuroprotective and recovery enhancement effects of statins for ICH by targeting secondary brain injury pathways.²³ However, selection bias of these retrospective observational studies restricted the significance and interpretation of these results, especially regarding the proposed improvement in neurological function. The effect of statin therapy on specific end points such as PHE yielded conflicting results.^10,24 PHE has only been investigated during the first 24 to 48 hours after the ICH event, which is why the effect of statin continuation, discontinuation, or initiation on edema extent have not been evaluated so far and statins may potentially even impair functional outcome by increased edema formation.^10,23 Therefore, the European Stroke Organisation defined the issue of when statins should be (re-) started after ICH as major priority in ICH research.¹¹

By evaluating the maximal extent of PHE formation, that is, peak PHE, we demonstrate that initiation of statin therapy in patients with ICH without such premedication seems to increase peak PHE and may affect functional outcome by aggravating secondary brain injury. Therefore, based on our findings, statins should not be initiated during the acute phase of ICH. Potential explanations include the complex mechanism and interactions of statins resulting in both beneficial and detrimental effects in patients with ICH. Experimental studies demonstrated that statins accelerate red blood cell phagocytosis by activation of microglia and macrophage recruitment in the hematoma boundary zone, which may result in accelerated hematoma absorption, but also increased peak PHE.^23,25,26 This detrimental effect may be more pronounced in statin-naive patients and may be avoided by not initiating statins during the acute phase of ICH.^23,27,28 On the contrary, preclinical studies demonstrated that statins decrease apoptosis and increase neurogenesis, synaptogenesis, and growth factor expression, which may result in improved brain plasticity and neurological outcome.^23,29,30 This beneficial recovery enhancing effect may be utilized by initiating statins after the phase of peak edema formation. However, continuation (versus discontinuation) of statin premedication did not significantly affect peak PHE in our study. Statins are effective for >24 hours after the last intake,³¹ why statin premedication—even if discontinued—has an effect on a very relevant phase of PHE formation, and the beneficial effect of discontinuation (versus continuation) on PHE formation appears less relevant. Given the pharmacological rebound effects of acute discontinuation of statin treatment resulting in oxidative stress, impaired vascular function, and acute vascular events,^5,32 continuation of statin premedication should be preferred over discontinuation to prevent cardiovascular events.

Regarding the associations of statin use and epileptic seizures in patients with ICH, a recent cohort study suggested potential protective effects of statins against remote symptomatic seizures after ICH. However, Lin et al⁸ included subarachnoid hemorrhage and other unspecified intracranial hemorrhage subtypes and statistical models were not adjusted for hematoma location or hematoma volume.^33,34 In contrast, the incidence of remote symptomatic seizures were higher in patients with ICH with statin treatment in our study, although these differences were statistically not significant. Preclinical studies suggest that statins may have both anticonvulsant and proconvulsant effects depending on the experimental model tested. Suppression of inflammatory pathways and glutamate-mediated neurotoxicity could increase the seizure threshold and provide anticonvulsant effects.³⁵ In contrast, inhibition of G protein and modulation of synapse formation could cause proconvulsant effects, notably in patients with compromised blood–brain barrier.^35,36 Future studies addressing the impact of statins in patients with ICH should further investigate the occurrence of seizures and the potential proconvulsant effects of statin therapy.

Regarding the associations of statin use and cardiovascular events in patients with ICH, a recent study reported that continued statin treatment reduced cardiovascular events in patients post-ICH.²³ We here confirm that statin therapy prevents secondary cardiovascular events, which may exert clinical significance beyond the first 3 months after the index event. The reduction of cardiovascular events appears more relevant than the previously reported recovery enhancing effects of statins and in parts may drive the reported improvement in neurological function at 12 months. In line, the recently published iDEF trial (Deferoxamine Mesylate in Patients With Intracerebral Haemorrhage) demonstrated that a relevant proportion of patients showed functional improvement beyond 3 months after ICH and that benefits in functional recovery may become more evident at later time points. Our results suggest that statins should be initiated after the acute phase of ICH in patients to reduce relevant long-term cardiovascular risk factors, eventually in a care-bundle approach.³⁷

Regarding the aspect of ICH incidence and ICH location: Statins have been discussed to increase the incidence of ICH in the subgroup of patients with previous ischemic stroke or ICH according to the SPARCL trial results in 2006 and similarly albeit not significant in HPS (Heart Protection Study).^2,5,38 However, subsequent randomized controlled trials did not show an increased risk of ICH in general. We here demonstrate that statin therapy may be associated with higher rates of lobar hematoma location, both in primary spontaneous ICH and also in OAC associated ICH. This finding would be in line with the previously reported higher incidence of ICH and might be explained by an effect of statins promoting cerebral amyloid angiopathy and implying effects on cortical-subcortical vessels in addition to anticoagulation.^39,40 Yet, the results may also reflect an epiphenomenon of statin treatment and concomitant blood pressure medication compared with patients without antihypertensive and lipid lowering treatment resulting in lower rates of deep hematoma location in patients with statin treatment. Thus, the potential effects of statins on cerebral small-vessels remain a scientific hypothesis to be tested in future studies that also address absolute incidence of ICH subtypes.

Limitations of our study include single-center study design and no prespecified statin regimes. Further, timing and number of imaging scans were based on clinical practice, not on defined criteria and subgroups of different statin types and regimes were small. Lack of randomization and missing data on cholesterol levels may have resulted in bias by indication and influence of confounder variables in this observational study design. There were violations of the proportional hazard assumption, although the cox proportional hazard model is widely accepted as being robust to minor violations. ICH volumes were calculated by ABC method and some data on clinical outcome were missing, both leaving room for uncertainties.

Conclusions

Among statin users, lobar ICH occurs more frequently as compared with nonstatin users. While continuation of prior statin therapy appears to be safe regarding PHE formation, the initiation of statins during the first days after ICH may increase edema extent and should, thus, be avoided during the acute ICH phase. However, statins should be initiated thereafter (eg, at hospital discharge) to prevent cardiovascular events and potentially improve functional recovery.

Acknowledgments

The present work was performed in (partial) fulfillment of the requirements for obtaining the degree Dr. med. (J.I. Saam).

Supplemental Materials

Expanded Results

Figures I–IV

Footnotes