Hemorrhagic Transformation After Large Cerebral Infarction in Rats Pretreated With Dabigatran or Warfarin
Background and Purpose—It is uncertain whether hemorrhagic transformation (HT) after large cerebral infarction is less frequent in dabigatran users than warfarin users. We compared the occurrence of HT after large cerebral infarction among rats pretreated with dabigatran, warfarin, or placebo.
Methods—This was a triple-blind, randomized, and placebo-controlled experiment. After treatment with warfarin (0.2 mg/kg), dabigatran (20 mg/kg), or saline for 7 days, Wistar rats were subjected to transient middle cerebral artery occlusion. As the primary outcome, HT was determined by gradient-recalled echo imaging. For the secondary outcome, intracranial hemorrhage was assessed via gradient-recalled echo imaging in surviving rats and via autopsy for dead rats.
Results—Of 62 rats, there were 33 deaths (53.2%, 17 technical reasons). Of the intention-to-treat population, 33 rats underwent brain imaging. HT was less frequent in the dabigatran group than the warfarin group (placebo 2/14 [14%], dabigatran 0/10 [0%], and warfarin 9/9 [100%]; dabigatran versus warfarin; P<0.001). In all 62 rats, compared with the placebo (2/14 [14.3%]), the incidence of intracranial hemorrhage was significantly higher in the warfarin group (19/29 [65.5%]; P=0.003), but not in the dabigatran group (6/19 [31.6%]; P=0.420). Mortality was significantly higher in the warfarin group than the dabigatran group (79.3% versus 47.4%; P=0.022), but not related to the hemorrhage frequency.
Conclusions—The risk of HT after a large cerebral infarction was significantly increased in rats pretreated with warfarin than those with dabigatran. However, the results here may not have an exact clinical translation.
- cerebral hemorrhage
- cerebral infarction
- magnetic resonance imaging
Oral anticoagulation therapy with a vitamin K antagonist, such as warfarin, reduces the risk of ischemic stroke in patients with nonvalvular atrial fibrillation. However, warfarin has many drawbacks, including its narrow therapeutic window, requiring frequent laboratory monitoring, and the risk of intracerebral hemorrhage (ICH).1 Direct oral anticoagulants, such as direct thrombin inhibitors (ie, dabigatran) or factor Xa inhibitors (ie, rivaroxaban, apixaban, and edoxaban), are widely used because they show similar or superior efficacy over warfarin in patients with nonvalvular atrial fibrillation in randomized trials.1 They also substantially reduce the risk of ICH.
Cerebral ischemia results in a loss of microvascular integrity, as well as cell death.2–4 The loss of microvascular integrity often causes hemorrhagic transformation (HT) and ICH in the ischemic area.5 The risk of ICH is particularly increased in patients with a large cerebral infarction undergoing anticoagulation treatment.6 Therefore, the use of urgent anticoagulation treatment is not recommended within the guidelines for treatment of patients with acute ischemic stroke.7,8 However, the safety of direct oral anticoagulants in the acute stage of large cerebral infarction remains uncertain. Given that the risk of ICH is lower in direct oral anticoagulant users than warfarin users among patients with nonvalvular atrial fibrillation, the risk of ICH in acute cerebral infarction may also be lower in direct oral anticoagulant users than warfarin users. However, it remains uncertain whether the risk of ICH is lower in direct oral anticoagulant users when they develop large cerebral infarction that is accompanied by a loss of vascular integrity. Therefore, we investigated whether the risk of intracerebral HT or ICH in a rat model of large cerebral infarction differed between rats pretreated with dabigatran and those with warfarin or placebo.
Materials and Methods
This was a triple-blind, randomized, and placebo-controlled study. All animal procedures were approved by the Institutional Animal Care and Use Committee of Yonsei University College of Medicine and were performed in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care. We performed all experimental procedures according to Animal Research: Reporting for In vivo Experiments guidelines9 and Stroke Therapy Academic Industry Roundtable guidelines.10,11 Sixty-two male Wistar rats, ≈9 weeks old and weighing 270 to 300 g, were used. Animals were housed in a temperature-controlled animal facility under a 12/12 hour reverse light/dark cycle, in a plastic cage with soft bedding and were allowed free access to food and water. The animals were then randomly allocated to the dabigatran, warfarin, or placebo group. Intention-to-treat population included all rats included for this study. Per-protocol population (PP) included those with a prothrombin time (PT) ratio of 2.0 to 4.0 among warfarin-treated rats. The PT ratio was defined as the ratio of PT of the treated rat to the mean PT of placebo rats after the convention of relying on PT ratio to monitor the effect of oral anticoagulant.
The primary outcome was any HT on gradient-recalled echo (GRE) magnetic resonance imaging (MRI). The secondary outcome was mortality, infarction volume on T2-weighted MRI, neurological deficits measured by Longa and Garcia scales, and any intracranial hemorrhage on GRE MRI or autopsy (in cases of death before MRI).
Randomization, Blinding, and Drug Treatment
A randomization list with random numbers in permuted blocks with a block size of 4 was computer generated by a researcher who was not involved in the animal experiments. After the preparation of study drugs and assignment of a unique study number, different investigators, who were blinded to the study group, were involved in one of the following experiments: (1) administration of study drugs, (2) surgical procedures and clinical assessments, and (3) outcome assessments for HT/ICH and infarct size. Data monitoring investigators were also blinded to the group identity. The animals randomly received dabigatran etexilate (20 mg/kg per day, Pradaxa; Boehringer Ingelheim, Ingelheim, Germany), warfarin (0.2 mg/kg per day, Warfarin sodium; Daehwa Pharmaceutical Co, Gangwon-do, Korea), or placebo (saline). The doses of dabigatran and warfarin were based on previous studies.12–14 Each drug was given by oral gavage using a feeding needle twice a day for 7 days (warfarin once and saline once in the warfarin group, so that warfarin was administered once a day), and the last administration was 1 hour before the induction of middle cerebral artery occlusion (MCAO).
Sample Size and Conduction of Study
This study was performed as a 2-stage experiment. Sample sizes for the χ2 test were calculated according to our preliminary data. We considered 3 groups for detecting differences of HT. We calculated that a sample size of 13 rats per group was sufficient to detect differences with an effect size of 0.51 (placebo, 10%; dabigatran, 35%; warfarin, 70%) at a significance level of 0.05 (2-sided) with 80% power. We determined that a sample size of 14 rats per group was required to account for a 5% dropout rate. Independent data monitoring investigators, who were blinded to the group identity, monitored data and laboratory abnormalities in accordance with the guidelines during the first experiment. After monitoring of the first experiment, the data monitoring investigators recommended an additional experiment. Sample sizes for the additional experiment were calculated by the data monitoring investigators because of the uneven dropout rates between groups observed in the first experiment. In the second experiment, 5 and 15 rats were added to the dabigatran and warfarin group, respectively, to allow for a 1:3 ratio in dropout rates. In the second experiment, the process for randomization, blinding, drug treatment, and outcome assessments was same as in the first experiment.
Induction of Transient Focal Cerebral Ischemia
One hour after the last administration of each drug, the rats were subjected to transient MCAO, as described previously.4 Reperfusion was performed 2 hours after MCAO by pulling on the suture until resistance was felt. Details are reported in the online-only Data Supplement.
Assessment of Bleeding From the Surgical Wound and Autopsied Brain
The presence of excessive bleeding from the surgery incision site during or after the surgical procedure was recorded by the investigator who performed the operations. Any visible bleeding elsewhere was also recorded. This investigator performed autopsies for rats that died before MRI, examined the brain for any visible ICH or subarachnoid hemorrhage, and recorded it. During all these assessments, the investigator was blinded to the group identity.
Neurological evaluation was performed 21 hours after reperfusion. All surviving animals were graded on 2 known scales, modified from the scales by Longa et al15 and Garcia et al.16 Dead animals were graded as having the worst score. Details are reported in the online-only Data Supplement.
Magnetic Resonance Imaging
After 2 hours of MCAO and 21 hours of reperfusion, we acquired T2-weighted and GRE magnetic resonance images. Details of MRI and magnetic resonance sequences are reported in the online-only Data Supplement.
Assessment of Infarction Volume and HT on MRI
For the assessment of imaging outcomes, we predefined exclusion criteria, which included death before obtaining magnetic resonance images and lack of neurological deficits (Longa scale=0) 2 hours after reperfusion.
The presence of any HT was assessed with GRE imaging, based on the European Cooperative Acute Stroke Study definition.17 Briefly, HT was classified into hemorrhagic infarction (small petechial or confluent hemorrhage without mass effect) and parenchymal hemorrhage (parenchymal hemorrhage with mass effect; Figure 1). Two stroke neurologists assessed the HT, and any discrepancy was solved by consensus. The infarction volume was measured on T2-weighted imaging in a semiautomatic manner using image analysis software (Xelis; Infinitt, Seoul, Korea) by 1 of 2 stroke neurologists. All stroke neurologists who assessed imaging outcomes were blinded to the group information.
Measurement of Blood Pressure
Blood pressure was measured in 14 animals from each group during the first experiment. Details are reported in the online-only Data Supplement.
Determination of PT and Plasma Concentrations of Dabigatran
To determine the PT and plasma concentrations of dabigatran, blood (900 μL) was drawn from the left common carotid artery before ligation during the surgery for MCAO. The blood was drawn into a syringe containing 100 μL of 3.8% sodium citrate solution, resulting in a final citrate concentration of 0.38%. Plasma was obtained by centrifuging the anticoagulated blood at 3000 rpm for 10 minutes at 4°C immediately after blood collection and was stored frozen at −70°C until use.
The plasma concentration for dabigatran was determined at the Severance hospital clinical trials center. Briefly, the plasma concentration of dabigatran was determined by a liquid chromatography/tandem mass spectrometry system. Dabigatran was purchased from Selleckchem (S2196; Houston, TX). For liquid chromatography, the Agilent 1200 high-performance liquid chromatography system (Agilent Technologies, Santa Clara, CA) equipped with an autosampler, pump, and degasser was used. The analytic column comprised a Kinetex XB-C18 (100×4.6 mm, 2.6 μm; Phenomenex, Torrance, CA), and the mobile phase consisted of solvent A (distilled water with 0.1% formic acid; Fluka, Steinheim, Germany) and solvent B (acetonitrile; Thermo Fisher Scientific, Waltham, MA). For tandem mass spectrometry, API 3200 (ASICX; Concord, Ontario, Canada) equipped with an electrospray ionization interface was used. Details are described in the online-only Data Supplement.
To examine the anticoagulant activity of dabigatran and warfarin, clotting time (CT), which is the most relevant parameter to examine anticoagulation effects, was assessed using rotational thromboelastometry (ROTEM). Details are reported in the online-only Data Supplement.
Statistical analyses were performed using SPSS (version 23.0; SPSS Inc, Chicago, IL). The normality of distributions was verified using the Kolmogorov–Smirnov test. Differences between the groups were compared with a 1-way ANOVA test, followed by post hoc analysis via the Tukey method (infarction volumes, neurological deficit scores, blood pressures, PT, and CT). The Student t test was performed to compare plasma concentrations of dabigatran between the groups. The χ2 test or Fisher exact test was used to compare categorical variables (mortality and incidence of HT) between the groups. P<0.05 was considered statistically significant.
A total of 62 rats were used for this study. In the first experiment, 42 rats were randomized and allocated to 3 groups (n=14). Among them, 1 in the placebo group, 8 in the dabigatran group, and 13 in the warfarin group died. Blood pressures were not significantly different between the groups (Table I in the online-only Data Supplement). In the second experiment, 20 rats were randomized into 2 groups according to a 1:3 ratio (5 rats to the dabigatran group and 15 rats to the warfarin group). Among them, 1 in the dabigatran group and 10 in the warfarin group died. In total, 14 rats for the placebo group, 19 rats for the dabigatran group, and 29 rats for the warfarin group were used for the intention-to-treat population (Figure 2). The PP included 42 rats, with 20 rats in the warfarin group excluded because of either being outside the target PT ratio range of 2.0 to 4.0 (n = 18) or the failure to obtain PT ratio (n=2; Figure 3A).
Of the intention-to-treat population, 33 rats underwent MRI and were analyzed for the primary outcome (Table 1). HT developed in 2 of 14 rats in the placebo group (14.3%). The incidence of HT was significantly increased in the warfarin group (9/9 rats, 100%; P<0.001), but not in the dabigatran group (0/10 rats, 0%; P=0.486). All HT was observed in the ischemic areas.
The incidence of any intracranial hemorrhage measured via GRE MRI or autopsy was compared in all 62 rats of the intention-to-treat population. The presence of intracranial hemorrhage was assessed via autopsy if rats died before MRI. The incidence of intracranial hemorrhage was 14.3% (2/14 rats) in the placebo group, 31.6% (6/19 rats) in the dabigatran group, and 65.5% (19/29 rats) in the warfarin group. When comparing with the placebo group, the incidence of intracranial hemorrhage was significantly higher in the warfarin group (P=0.002), but not in the dabigatran group (P=0.416). The incidence of intracranial hemorrhage in the dabigatran group was significantly lower than that in the warfarin group (P=0.021). Detailed causes of death are reported Table II in the online-only Data Supplement.
We determined the occurrence of surgical bleeding. Excessive bleeding from the surgical wound developed more frequently in the warfarin group (17/29 rats, 58.6%) than the placebo group (1/14 rats, 7.1%; P=0.001) or the dabigatran group (2/19 rats, 10.5%; P=0.001). The surgical bleeding was not increased in the dabigatran group compared with the placebo group (P>0.999).
Of 62 rats, there were 33 deaths (53.2%). The overall mortalities were higher in the warfarin group (23/29 rats, 79.3%; P<0.001) and the dabigatran group (9/19 rats, 47.4%; P=0.021) than the placebo group (1/14 rats, 7.1%). The mortality was significantly higher in the warfarin group than the dabigatran group (P=0.022). Among 33 deaths, 17 might be related to technical reasons (stress during blood pressure monitoring in 6, anesthesia in 5, and because of subarachnoid hemorrhage induced by arterial dissection in 6). Detailed causes of death are reported in Table II in the online-only Data Supplement.
Infarction Volume and Neurological Deficits
The mean infarction volume was not different between the groups (placebo, 432.5±78.4 mm3; dabigatran, 432.4±60.3 mm3; warfarin, 446.6±69.3 mm3; P=0.882; Figure 4A). However, the neurological deficits were significantly different between the groups (Longa scale, P=0.006; Garcia scale, P=0.003). The warfarin group (Longa scale, 3.6±0.8, P<0.001; Garcia scale, 4.8±3.6, P<0.001) or the dabigatran group (Longa scale, 2.8±1.3, P=0.006; Garcia scale, 8.5±6.0, P=0.001) had more severe neurological deficits than the placebo group (Longa scale, 1.7±0.7; Garcia scale, 14.1±2.5; Figure 4B and 4C). The warfarin group had more severe neurological deficits than the dabigatran group (Longa scale, P=0.012; Garcia scale, P=0.013).
Of the PP, 26 rats underwent MRI and were analyzed for the primary outcome. HT developed in 2 of 14 rats in the placebo group (14.3%; Table 1). The incidence of HT was significantly increased in the warfarin group (2/2 rats, 100%; P=0.050), but not in the dabigatran group (0/10 rats, 0%; P=0.493).
The incidence of any intracranial hemorrhage measured via GRE MRI or autopsy was compared in all 42 rats of the PP. The incidence of intracranial hemorrhage was 14.3% (2/14 rats) in the placebo group, 31.6% (6/19 rats) in the dabigatran group, and 77.8% (7/9 rats) in the warfarin group. When compared with the placebo group, the incidence of intracranial hemorrhage was significantly higher in the warfarin group (P=0.007), but not in the dabigatran group (P=0.416). The incidence of intracranial hemorrhage in the warfarin group was significantly higher than that in the dabigatran group (P=0.042).
Anticoagulant Activity and Plasma Concentrations of Dabigatran
The warfarin group had higher levels of PT than the placebo group or the dabigatran group (16.7±1.2 in the placebo group, 19.9±1.9 in the dabigatran group, and 42.0±19.7 in the warfarin group; P<0.001). On ROTEM, CT was different between the groups. The CT was significantly prolonged in the dabigatran group in all assays, and the CT in the warfarin group was prolonged with borderline significance (P<0.1) in intrinsic and functional fibrinogen thromboelastometry, when compared with the placebo group. The CT in the dabigatran group was significantly prolonged compared with the warfarin group in extrinsic and functional fibrinogen thromboelastometry (Table 2). Full data sets are available in Table III in the online-only Data Supplement.
The mean plasma concentration of dabigatran in dabigatran-treated rats was 209.8±108.2 ng/mL (Figure 3B). The plasma concentration of dabigatran was not different between the animals with HT and those without (with HT, 194.07±198.85; without HT, 224.03±116.20; P=0.572).
We investigated the safety of pretreatment with dabigatran or warfarin in the acute stage of large cerebral infarction. This study showed that MCAO caused large infarctions and, similar to observations in humans, that HT occasionally developed after large cerebral infarction because 14.3% of the placebo group showed HT, measured by GRE MRI. This study demonstrated that although infarction volumes were similar between the groups, the occurrence of ICH or HT was significantly different. HT on GRE images was found in all of the surviving warfarin-treated rats 23 hours after MCAO. This was observed not only in the intention-to-treat population but also in the PP that showed a PT ratio within the therapeutic range (2.0–4.0). However, no dabigatran-treated rats showed HT on GRE images.
For the secondary outcome, we included rats that died before MRI because ICH might lead to death before MRI. The autopsies revealed that 6 of 9 of the dabigatran group showed intracranial hemorrhage. This suggests that some animals in the dabigatran group may have died of intracranial hemorrhage before MRI analysis; thus, they could not be included in the primary outcome analysis. Nevertheless, the overall incidence of HT or intracranial hemorrhage was not significantly different between the dabigatran group and the placebo.
A previous study showed that the amount of hemorrhage, based on photometric analysis, was not different at 24 and 72 hours after the induction of cerebral infarction between dabigatran-pretreated mice and placebo. In the study, the frequency of ICH was unknown because homogenized whole brains were used for the photometric analysis.18 In another study, after tissue-type plasminogen activator treatment in a rat model of transient MCAO, hematoma volume was measured using histological analysis in 2 mm tissue sections. This study showed significantly reduced hemorrhage volume in rats pretreated with dabigatran than those with warfarin.13 Our findings add evidence on the safety of dabigatran in acute large cerebral infarction by showing a similar frequency of HT or intracranial hemorrhage in dabigatran-pretreated and placebo rats. In addition, we observed a lower frequency of HT or intracranial hemorrhage in dabigatran-pretreated rats than warfarin-pretreated rats.
Although the exact mechanism for the lower occurrence of ICH in dabigatran pretreatment compared with warfarin remains uncertain, the protective effect of dabigatran may be mediated by the inhibition of thrombin-induced protease-activated receptor (PAR)-1 activation. PAR-1 is expressed in neurons, astrocytes, microglial, and endothelial cells in the brain tissue.19 Activation of PAR-1 by thrombin induces neuronal cell death, microglial activation, toxin release, and endothelial hyperpermeability.20 The inhibition of PAR-1 by genetic deletion of PAR-1 receptor or treatment with a direct thrombin inhibitor argatroban alleviated the neuronal damages and decreased vascular disruption during acute focal ischemia.21,22 In fact, in the rat MCAO reperfusion model, there was a marked decrease in the detachment between astrocyte end feet and pericytes in the peri-ischemic lesion of dabigatran-pretreated rats compared with warfarin-pretreated rats.13 In a recent study on primary murine brain endothelial cells, alone or in coculture with astrocytes, dabigatran completely reversed the increased permeability induced by thrombin or oxygen-glucose deprivation.23 These studies imply that dabigatran, as an inhibitor of thrombin, could also have protective effects against neurovascular damage and HT during focal ischemia. Further studies with a larger sample size are required to demonstrate the effect of dabigatran on preventing neurovascular damage and reducing the occurrence of HT in cerebral ischemia.
There may be a concern that the different potency of anticoagulant activity between dabigatran and warfarin was responsible for the decreased frequency of intracranial hemorrhage observed in the dabigatran group. Consensus on the optimal dose of dabigatran and warfarin in rodents has yet to be reached. The dose of 0.2 mg/kg per day was based on previous studies that administered warfarin by oral gavage as in our experiment.12,13 In the venous thrombosis model, a therapeutically relevant dose in the rat was suggested as 0.2 mg/kg, which showed ≈65% inhibition of thrombus formation and international normalized ratio ≈3.0.12 We used 20 mg/kg per day of dabigatran based on previous studies in rats. In a previous study, 10 mg/kg of oral dabigatran reduced the thrombus volume by 70% compared with vehicle control.14 In other studies, 20 mg/kg of oral dabigatran was used.13,24
To determine anticoagulant effects of dabigatran and warfarin, we measured CT using ROTEM. CT is known as the most sensitive parameter affected by anticoagulation among various parameters in ROTEM. Especially, the CT in extrinsic thromboelastometry is clearly associated with warfarin-induced international normalized ratio elevation25 and plasma concentration of dabigatran.26 In this study, 7-day treatment with 20 mg/kg dabigatran resulted in a prolonged CT. In addition, the CT was longer in the dabigatran group than the warfarin group, which suggests that anticoagulating effects of dabigatran were at least equipotent to or more potent than those of warfarin in this study. We further measured plasma concentrations of dabigatran, which showed similar mean peak plasma concentrations (184 ng/mL; 95% confidence interval, 64–443 ng/mL) to those in trial patients administered with 150 mg bid.27 These findings indicate that the anticoagulant activity after pretreatment with dabigatran or warfarin was maintained after a large cerebral infarction, and the low incidence of HT in the dabigatran group was not related with its dose or anticoagulant activity in this experiment.
In this study, mortality was very high and neurological deficits were more severe in the warfarin group. More severe neurological deficits and higher mortality in the warfarin group might be ascribed to the increased risk of ICH. Coexisting ICH in the area of large cerebral infarction could aggravate neurological deficits and increase the risk of brain swelling and death.
This study has several merits. This study modeled the clinical situation as closely as possible to answer unresolved clinical questions. First, there are concerns of ICH in patients who develop large cerebral infarction, although they are undergoing oral anticoagulation treatment. In addition, it remains unknown whether the use of dabigatran is safe during the acute stage of large cerebral infarction. This study provides some evidence to answer these questions. Second, we determined the frequency of ICH via MRI. Accordingly, we estimated the risk of ICH as it is performed in clinical practice.
However, this study also had some limitations. First, mortality was high as a result of a very severe stroke being induced in the experimental procedures. Therefore, approximately half of the rats were not able to be evaluated 23 hours after MCAO and, thus, were not included in the MRI analysis of the primary outcome. Although the presence of intracranial hemorrhage was assessed via autopsy as the secondary outcome, this should be considered when interpreting the primary outcome data. Second, although the mortality was high, the exact reason could not be determined in some animals because necropsy was not performed in dead animals. Some of them might have bleedings in the peripheral organ. Rats may be more sensitive or susceptible to toxic systemic effects of warfarin than human, particularly when it is administered repeatedly.28
Therefore, this factor should be considered for interpreting our results on the cause of death. Third, the excess of HT in the warfarin group might be related to overanticoagulation. However, there was no significant correlation between PT ratio and HT in the warfarin group, and the anticoagulant effects of dabigatran were at least equipotent to or more potent than that of warfarin on ROTEM. Fourth, the warfarin group showed a wide range of PT ratio despite that we administered warfarin using a feeding needle. As a result, there is a possibility of type II error in the PP as only 2 rats were included. Although we set the target PT ratio of 2.0 to 4.0 for defining the PP group, it is uncertain what international normalized ratio target is optimal in rodents for representing international normalized ratio target in human. Finally, direct translation of our findings to humans should be cautious because coagulation and hemostasis systems and ROTEM results may be different between human and rodents, and because animals in this study do not have atrial fibrillation while dabigatran is used in patients with atrial fibrillation.
Nevertheless, the findings in this study suggest that the risk of ICH after large cerebral infarction in dabigatran-treated rodents is significantly lower than warfarin-treated rodents and suggests no significantly increased risk when comparing the use of dabigatran with no anticoagulant. Our findings also provide indirect evidence that the use of dabigatran may not significantly increase the risk of ICH during the acute stage of large cerebral infarction. However, because of several differences between animals and humans as we mentioned above, the results here may not have an exact clinical translation.
We thank Dong-Su Jang, MFA, (Medical Illustrator) for his help with the illustrations.
Sources of Funding
This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare, Republic of Korea (HI08C2149) and Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2014R1A1A1005913).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.017751/-/DC1.
- Received April 18, 2017.
- Revision received August 1, 2017.
- Accepted August 2, 2017.
- © 2017 American Heart Association, Inc.
- Hart RG,
- Lock-Wood KI,
- Hakim AM,
- Koller R,
- Davenport JG,
- Coull BM,
- et al
- Jauch EC,
- Saver JL,
- Adams HP Jr,
- Bruno A,
- Connors JJ,
- Demaerschalk BM,
- et al
- Diener HC,
- Kleinschnitz C
- Macleod MR,
- Fisher M,
- O’Collins V,
- Sena ES,
- Dirnagl U,
- Bath PM,
- et al
- Fisher M,
- Feuerstein G,
- Howells DW,
- Hurn PD,
- Kent TA,
- Savitz SI,
- et al
- Kono S,
- Deguchi K,
- Omote Y,
- Yunoki T,
- Yamashita T,
- Kurata T,
- et al
- Longa EZ,
- Weinstein PR,
- Carlson S,
- Cummins R
- Garcia JH,
- Wagner S,
- Liu KF,
- Hu XJ
- Thanvi BR,
- Treadwell S,
- Robinson T
- Chen B,
- Cheng Q,
- Yang K,
- Lyden PD
- Junge CE,
- Sugawara T,
- Mannaioni G,
- Alagarsamy S,
- Conn PJ,
- Brat DJ,
- et al
- Tanoue S,
- Inamasu J,
- Yamada M,
- Toyama H,
- Hirose Y
- Schmidt DE,
- Holmström M,
- Majeed A,
- Näslin D,
- Wallén H,
- Ågren A
- Božič-Mijovski M,
- Vene N,
- Snoj N,
- Gradišek P,
- Mavri A
- van Ryn J,
- Stangier J,
- Haertter S,
- Liesenfeld KH,
- Wienen W,
- Feuring M,
- et al