Remote Intracerebral Hemorrhage After Intravenous Thrombolysis
Results From a Multicenter Study
Background and Purpose—Remote parenchymal hemorrhage (rPH) after intravenous thrombolysis with recombinant tissue-type plasminogen activator may be associated with cerebral amyloid angiopathy, although supportive data are limited. We aimed to investigate risk factors of rPH after intravenous thrombolysis with recombinant tissue-type plasminogen activator.
Methods—This is an observational study of patients with ischemic stroke who were treated with intravenous thrombolysis with recombinant tissue-type plasminogen activator and were included in a multicenter prospective registry. rPH was defined as any extraischemic hemorrhage detected in the follow-up computed tomography. We collected demographic, clinical, laboratory, radiological, and outcome variables. In the subset of patients who underwent a magnetic resonance imaging examination, we evaluated the distribution and burden of cerebral microbleeds, cortical superficial siderosis, leukoaraiosis, and recent silent ischemia in regions anatomically unrelated to the ischemic lesion that caused the initial symptoms. We compared patients with rPH with those without rPH or parenchymal hemorrhage. Independent risk factors for rPH were obtained by multivariable logistic regression analyses.
Results—We evaluated 992 patients (mean age, 74.0±12.6 years; 52.9% were men), and 408 (41%) of them underwent a magnetic resonance imaging. Twenty-six patients (2.6%) had a rPH, 8 (0.8%) had both rPH and PH, 58 (5.8%) had PH, and 900 (90.7%) had no bleeding complication. Lobar cerebral microbleeds (odds ratio, 8.0; 95% confidence interval, 2.3–27.2) and recent silent ischemia (odds ratio, 4.8; 95% confidence interval, 1.6–14.1) increased the risk of rPH.
Conclusions—The occurrence of rPH after intravenous thrombolysis with recombinant tissue-type plasminogen activator in patients with ischemic stroke is associated with lobar cerebral microbleeds and multiple ischemic lesions in different regions.
Intracerebral hemorrhage (ICH) is the most serious complication of intravenous thrombolysis with recombinant tissue-type plasminogen activator (IV-r-tPA) in patients with ischemic stroke.1 The European Cooperative Acute Stroke Study (ECASS) defined remote parenchymal hemorrhages (rPH) as single or multiple hemorrhages that appear in brain regions without visible ischemic damage detected by cranial computed tomography (CT), remote from the area causing the initial stroke symptoms.1,2 This type of bleeding has an incidence of 1.3 to 3.7% and is associated with a poor outcome.1–5
The majority of studies on cerebral bleedings after IV-r-tPA in patients with acute ischemic stroke have focused on risk factors for symptomatic and local parenchymal hemorrhage (PH). As a result, little is known about the frequency, associated risk factors, and prognosis of rPH. To date, only 1 multicenter study reported an association of rPH with old age and previous stroke.3 However, the results from magnetic resonance imaging (MRI) were not provided, and variables were obtained from a general stroke database. Two additional observational studies analyzed possible risk factors but were limited by their retrospective design and a small sample size.4,5
It has been hypothesized that rPH after IV-r-tPA may be related to different mechanisms such as undiagnosed coagulopathies, multiple acute embolic areas of ischemia, or a generalized cerebral vasculopathy, such as cerebral amyloid angiopathy (CAA).6,7 The deposition of amyloid fibrils in cortical and leptomeningeal blood vessels leads to structural and functional arterial changes that seem to be involved in the pathogenesis of vessel rupture.6–8 Small neuropathological studies9 linked CAA to IV-r-tPA–associated ICH, as did a positron emission tomography study10 showing higher neocortical amyloid retention among patients with IV-r-tPA–associated ICH when compared with those without. We hypothesized that MRI surrogate markers of CAA, such as lobar cerebral microbleeds (CMB) and cortical superficial siderosis (CSS),11 among other variables including multiple embolic ischemic lesions in diffusion weighted imaging, are associated with the occurrence of rPH. The aim of this study was to analyze frequency, risk factors, clinical and radiological features, and prognosis of patients with rPH.
This is a multicenter, prospective, observational study of consecutive patients with ischemic stroke treated with IV-r-tPA at 9 different hospitals in Catalonia (Spain) between January 2011 and August 2013. All the patients were prospectively and consecutively included in the Sistema Online d´Informació de l´Ictus Agut (SONIIA). SONIIA is a mandatory and externally audited registry that monitors quality of all reperfusion therapies performed in Catalonia under routine practice conditions. The SONIIA registry satisfies all legal requirements mandated by the local law of protection of personal data. The study was performed according to local ethical guidelines.
We included patients aged ≥18 years with a clinical and radiological diagnosis of acute ischemic stroke and treated with IV-r-tPA within the first 4.5 hours of the onset of symptoms. All patients underwent a baseline noncontrast CT or MRI and a follow-up noncontrast CT within the first 36 hours of stroke onset. Cerebral hemorrhagic complications were classified according to the ECASS criteria.2 We defined rPH as any extraischemic intracranial hemorrhagic lesion observed in the follow-up CT.2 Two different investigators independently confirmed the diagnosis of all rPH. Exclusion criteria included any reperfusion therapy beyond the first 4.5 hours or any modality of endovascular treatment and stroke mimics, and patients without a follow-up CT within the first 36 hours were also excluded mainly because of early and unexpected death or hemodynamic instability. In the subgroup of patients included in the MRI analyses, we excluded those patients who performed the MRI scan beyond the 14 days of the onset of stroke.
Clinical and Radiological Assessment
The following variables were recorded: (1) demographics (age and sex); (2) traditional vascular risk factors (hypertension, diabetes mellitus, atrial fibrillation, previous transient ischemic attack, or cerebral infarct); (3) liver disease (if alanine aminotransferase and aspartate aminotransferase were >55/30 U/L and 40/32 U/L for men and women, respectively, or any previous know liver disease); (4) previous medication (antiplatelet agents, anticoagulants, and statins); (5) time from the onset of symptoms to the administration of IV-r-tPA (≤1.5 hours, 1.5–3 hours, and 3–4.5 hours); (6) severity of the neurological deficit as measured by the National Institutes of Health Stroke Scale (NIHSS) score at admission and 24 to 36 hours later by a certified neurologist; (7) hypertensive episodes (≥185/105 mm Hg), hyperglycemia (≥7.7 mmol/L), and hyperthermia (≥37.5 °C) within the first 24 hours of admission; (8) baseline laboratory data (platelet count and basic coagulation tests); (9) type of hemorrhagic complication (ECASS criteria2), number (single/multiple) and localization (deep/lobar/brain stem) of rPH; (10) pathogenesis (TOAST [Trial of Org 10172 in Acute Stroke Treatment] classification12); (11) symptomatic ICH (increase of ≥4 points on the NIHSS score within the first 36 hours associated with a PH or a rPH complication); and (12) functional outcome and mortality were assessed at 3 months by a certified neurologist. A favorable functional outcome was defined as a modified Rankin scale score ≤2.
In a subgroup of patients, a MRI was performed within the first 14 days of the onset of stroke. The MRI protocol was standardized at each hospital. With most patients, imaging was 1.5-T field strength, and it always included T1-weighted, T2-weighted, fluid-attenuated inversion recovery, T2*-weighted gradient-recalled echo sequences, and diffusion weighted imaging. A neuroradiologist or a stroke neurologist evaluated the scans for recent silent ischemia (RSI), leukoaraiosis, CMB, CSS, and vascular malformations. RSI were defined as one or more recent ischemic lesions visible as a hyperintensity on diffusion weighted imaging and a decreased apparent diffusion coefficient in a different vascular region from the stroke that caused the initial symptoms.13 Leukoaraiosis was assessed by a simplified Fazekas rating scale from 0 to 3 (0, no lesions; 1, focal lesions; 2, early confluent; and 3, confluent).14 CMB were defined as small well-demarcated hemorrhages, hypointense, rounded lesions on MRI on T2*-weighted gradient-recalled echo sequences, or sensitive to magnetic susceptibility.15 The distribution of CMB was classified as lobar, deep, or mixed. The number of CMB was divided into 3 groups: no CMB, single CMB, multiple CMB (≥2). CSS was defined as linear residues of blood products in the superficial layers of the cerebral cortex that showed a characteristic gyriform pattern of low signal on T2* gradient-recalled echo images without corresponding hyperintense signal on T1-weighted or fluid-attenuated inversion recovery images.16
We divided the patients according to their bleeding complications. The rPH group was composed of patients with rPH, either with or without a concomitant local PH. The PH group included patients with local PH who did not have concomitant rPH. The group of patients without any rPH or PH included patients without any bleeding. In the group of patients without any rPH or PH, patients with hemorrhagic infarct were included, because these bleeding complications are considered not clinically relevant.
We compared demographic variables, traditional risk factors, previous medications, vital signs, laboratory findings, radiological features, and stroke pathogenesis between patients from the rPH group and patients without any rPH or PH. We also compared patients from the PH group and patients without any rPH or PH. Analyses were performed twice, first in the entire sample and then in patients in whom a MRI was performed. Finally, we compared prognosis (neurological deterioration, functional outcome, and mortality) of all the patients with rPH, symptomatic rPH, PH, and symptomatic PH and patients without any rPH or PH.
Comparison of collected variables between groups was done using contingency tables with the χ2 or the Fisher exact test for categorical variables, Student t test for quantitative variables with a normal distribution, and Mann–Whitney U test when a nonparametric test was required.
Forward multiple logistic regression analysis was performed to determine the clinical factors that independently predicted rPH. A multiple logistic regression analysis was also used to identify independent predictors of PH. Because of the expected low number of events, variables were selected for entry into the multivariate model based on the results of univariate analyses (P<0.05), previous literature about rPH, and other methodological principles (dichotomized variables wherever possible or eliminated predictor variables with too much missing data). The results of the multivariate analysis were assessed by (1) the Hosmer–Lemeshow test, (2) the accuracy of the predictions of the model, and (3) the area under the receiver-operating characteristic curve. We selected the model with the largest area under the curve or, if there were no statistically significant differences in the areas, the model with the simplest variables. All the statistical tests were performed using IBM SPSS 22.0 computer software. A 2-tailed P value of <0.05 was considered to indicate a significant difference.
A total of 1309 patients with ischemic stroke were treated with any reperfusion therapy within the study period, and 992 fulfilled the inclusion criteria (Figure 1). The mean age was 74.0±12.6 years, and 52.9% of them were men. A total of 26 patients (2.6%) had a rPH, 8 (0.8%) a rPH with concomitant PH, 58 (5.8%) had a PH, and 900 (90.7%) had no rPH or PH (Table 1). The ratio between single and multiple rPH was 1.6, and the most frequent distribution of rPH was lobar in 25 patients (73.5%), followed by deep in 6 patients (17.6%) and brain stem/cerebellum in 3 patients (8.8%).
Risk Factors in the Whole Sample
Demographic variables, traditional vascular risk factors, previous treatments, time from the onset of symptoms to the administration of IV-r-tPA, initial NIHSS score, vital signs, laboratory findings, and pathogenesis were not different between patients with rPH and those without any rPH or PH (Table I in the online-only Data Supplement). We did not find any patient with rPH and concomitant liver disease.
Risk Factors in Patients With MRI
The number of patients who underwent a MRI during the study was 408 (41.1%). Compared with those without MRI, these patients were younger (70.9±12.9 versus 76.2±11.9; P<0.001), the proportion of women was lower (43.4% versus 49.9%; P=0.045), and the 24-hour stroke severity was lower (median NIHSS score, 3 [1–7] versus 6 [1–17]; P<0.001). Fourteen patients (3.4%) had a rPH, 4 patients (0.9%) had a rPH with a concomitant PH, 26 patients had PH (6.3%), and 365 patients (89.4%) were without any rPH or PH.
Compared with patients without any rPH or PH (Table 2), atrial fibrillation was more common in patients with rPH (P=0.016). Lobar distribution of CMB (38.8% versus 5.6%; P=0.01) and multiple CMB (22.2% versus 4.4%; P=0.01) and the presence of CSS (11.1% versus 1.1%; P=0.03) were more frequent in patients with rPH compared with those without any rPH or PH. Moreover, the proportion of RSI was higher in the rPH group (43.7% versus 13.1%, P=0.004). We did not find any vascular malformation in the group of rPH.
The multivariable logistic regression analysis indicated that the presence of lobar CMB and RSI were independently associated with rPH (Table 3).
Data about PH are shown in Tables II and III in the online-only Data Supplement.
The severity of stroke at 24 hours was significantly higher in patients with any rPH (median NIHSS score 9 [Interquartile range: 3–21]) and any PH (median NIHSS score 12 [Interquartile range 4–19]) compared with patients without any rPH or PH (median NIHSS score 4 [Interquartile range 1–11]); P=0.008 and P<0.001, respectively). Symptomatic ICH was observed in patients with either rPH (17/34, 50%) or PH (25/57, 43.1%). Most concomitant rPH and PH were symptomatic (7/8, 87.5%).
At 3 months, the proportion of patients with a favorable outcome was lower in patients with rPH (26.4%), symptomatic rPH (11.8%), PH (29.7%), and symptomatic PH (16%) compared with patients without any rPH or PH (57.8%; all P<0.001). Mortality rates (Rankin scale score=6) were higher in patients with rPH (35.3%; P=0.002), symptomatic rPH (52.9%; P<0.001), PH (38.5%; P<0.001), and symptomatic PH (64%; P<0.001) compared with those without any rPH or PH (13%; Figure 2).
In this multicenter study, we observed that 3.4% patients had a rPH that is comparable to the rates reported in previous clinical trials (1.3–3.7%).1,2 The most remarkable finding in this study is that strictly lobar CMB, which appear to be indicative of CAA-related vasculopathy,11 and multiple acute embolic ischemic lesions are associated with rPH. In agreement with previous reports,3–5 rPH increases the risk of long-term functional dependence and mortality.
The results of the bivariate and multivariable analyses indicated that rPH is difficult to predict because most variables were equivalent in patients with or without this complication. Despite the weak association between rPH and previous stroke, higher age, higher blood pressure, and aspirin treatment, Mazya et al.3 acknowledged that they did not find any strong predictive risk factor. Hence, it is important to note that risk factors for rPH may not be reflected in the general stroke databases and seem different from risk factors for PH.
It is interesting to speculate that the association between CAA MRI markers and rPH may indicate a subjacent cerebral vasculopathy. CMB are common in patients with acute ischemic stroke (12.2–38.5%),15 but strictly lobar CMB are characteristic of CAA, whereas deep CMB most likely reflect hypertension-related arteriopathy.7 CSS is uncommon (≈1% of acute ischemic stroke)16 and reflects repeated episodes of hemorrhage into the subarachnoid space from brittle superficial cortical or leptomeningeal CAA-laden vessels, potentially heralding a high risk of future lobar ICH.17 Sporadic CAA renders small cortical vessels brittle and fragile because of vascular amyloid-β deposition, increasing their vulnerability to bleeding when hemostasis is acutely impaired after IV-r-tPA.7 The mechanism accounting for vessel rupture after amyloid deposition is not clear, but fibrinoid necrosis and microaneurysm seem likely to be involved.6 In addition to structural changes, functional changes, antithrombotic effects, and a local anticoagulation environment have been reported to be secondary to amyloid deposition.6,18 Genetic predisposing factors, such as APOE (apolipoprotein E) genotype (ε2 and ε4 alleles are linked with ICH), may be of interest as an indicator of the presence of CAA and CMB but have not been investigated in acute stroke.7Thus, it is likely that CAA may predispose to cerebral bleeding after thrombolytic therapy or oral anticoagulation.10,19,20
Our results agree with previous studies that patients with pre-existing CMB before IV-r-tPA have more frequently new extraischemic CMB and extraischemic hemorrhages after treatment,21,22 and also with a recent meta-analysis that reported that CMB presence doubled the risk of symptomatic ICH after IV-r-tPA.15
In addition to the relevance of CAA, we observed an association between multiple embolic ischemic lesions and rPH after IV-r-tPA. The increased prevalence of atrial fibrillation in patients with rPH could be one of the possible explanations of multiple acute ischemic lesions. Our study is limited by a lack of a MRI before IV-r-tPA, but we speculate that some of the ischemic lesions were present before IV-r-tPA. Hence, the main mechanism involved with this bleeding may be the reperfusion injury on a recently damaged tissue such as it occurs in local PH.7
Hypertensive episodes during or after IV-r-tPA infusion also may cause ICH,7 but this association is not definite in rPH. Kimura et al21 observed new CMB after IV-r-tPA in patients with higher systolic blood pressure. Additionally, higher arterial blood pressure after thrombolytic therapy in myocardial infarction, which is considered pathophysiologically comparable to rPH, increased the rates of cerebral bleeding.21 Whether or not lowering blood pressure during IV-r-tPA reduces the risk of any ICH may be answered by the on-going clinical trial Enhanced Control of Hypertension and Thrombolysis Stroke Study (ENCHANTED).23
It is clear that patients with rPH have higher rates of early neurological deterioration, 3 months functional dependence, and mortality compared with patients without any rPH or PH. These results are consistent with the results of previous studies.3–5 It is remarkable that symptomatic rPH was associated with an unfavorable outcome and a high mortality. This fact may be attributed collectively to the mass effect of PH considering that most concomitant rPH with PH were symptomatic and of the neuronal damage caused by rPH.
There are some limitations: first, our study was retrospective even though case ascertainment and entry of some basic information into the registry were prospective. Second, we decided to exclude patients without basic clinical or radiological data. Thus, the number of patients in this study was reduced. Third, our results are limited by a possible sample selection bias. According to the inclusion criteria, some patients with rPH may have been excluded because of the lack of a follow-up CT within the first 36 hours. Fourth, we included a small number of patients with a rPH, and our model has only been tested in the data set from which it was derived. It has not been prospectively validated with independent data. To avoid overfitting of the model, we limited the number of candidate variables, and we selected the simplest model. In addition, lobar CMB and RSI are included in the literature as possible explanations of rPH.24 Fifth, MRI data were reviewed retrospectively by different investigators without any inter-rater variability analyses. We are aware that there is a selection bias to MRI performance because MRI was performed more frequently in younger patients and in those without a severe neurological impairment. In addition, the MRI studies were performed to a different purpose rather than our study. Sixth, both CMB and CSS may be mimicked by other structural lesions.17,24 Finally, most MRI were performed after IV-r-tPA. CSS or CMB could be caused by IV-r-tPA,17,21 and new ischemic lesions may appear during the first days of hospitalization. Whether or not RSI or CAA markers were present before IV-r-tPA makes our conclusions less reliable.
We have shown that rPH is a relatively uncommon complication of intravenous thrombolysis that increases the risk of poor neurological outcome and mortality. Our findings support an association between this complication and CAA and ischemic damage in different regions. Future studies, including a baseline MRI protocol before intravenous thrombolysis and a larger number of patients are needed to confirm our findings. Our results cannot be used to indicate or exclude patients for intravenous thrombolysis treatment who otherwise fulfill the accepted criteria for this therapy. However, our results can be useful for selecting patients for further studies to improve the safety of this therapy.
We are grateful to all stroke professionals involved in prehospital and in-hospital stroke care across Catalonia.
Sources of Funding
This study was supported in part by the Spanish Ministry of Health (Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III FEDER, RD12/0042/0020) and Redes Temáticas de Investigación Cooperativa (Redes Temáticas de Investigación Cooperativa (RETICS-INVICTUS RD12/0014/0002).
The authors are solely responsible for the design and conduct of the presented study. They confirm the adherence and ethical standards.
Presented in part at the International Stroke Conference, Nashville, TN, February 11–12, 2015, and at the European Stroke Organisation Conference Conference, Glasgow, Scotland, April 17–19, 2015.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.013952/-/DC1.
- Received April 30, 2016.
- Revision received June 6, 2016.
- Accepted June 7, 2016.
- © 2016 American Heart Association, Inc.
- Fiorelli M,
- Bastianello S,
- von Kummer R,
- del Zoppo GJ,
- Larrue V,
- Lesaffre E,
- et al
- Trouillas P,
- von Kummer R
- Mazya MV,
- Ahmed N,
- Ford GA,
- Hobohm C,
- Mikulik R,
- Nunes AP,
- et al
- Karaszewski B,
- Houlden H,
- Smith EE,
- Markus HS,
- Charidimou A,
- Levi C,
- et al
- Smith EE,
- Greenberg SM
- Adams HP Jr,
- Bendixen BH,
- Kappelle LJ,
- Biller J,
- Love BB,
- Gordon DL,
- et al
- Inzitari D,
- Simoni M,
- Pracucci G,
- Poggesi A,
- Basile AM,
- Chabriat H,
- et al
- Charidimou A,
- Linn J,
- Vernooij MW,
- Opherk C,
- Akoudad S,
- Baron JC,
- et al
- Martí-Fàbregas J,
- Delgado-Mederos R,
- Granell E,
- Morenas Rodríguez E,
- Marín Lahoz J,
- Dinia L,
- et al
- Charidimou A,
- Wilson D,
- Shakeshaft C,
- Ambler G,
- White M,
- Cohen H,
- et al
- Kimura K,
- Aoki J,
- Shibazaki K,
- Saji N,
- Uemura J,
- Sakamoto Y
- Gratz PP,
- El-Koussy M,
- Hsieh K,
- von Arx S,
- Mono ML,
- Heldner MR,
- et al
- Craig SA