Introduction
Intracerebral hemorrhage (ICH) occurs in ~15 % of stroke patients, leading to ~50 % mortality and significant disability in survivors [
1,
2]. So far, there are no specific neuroprotective therapies for ICH, although survivors benefit from rehabilitation. Thus, it is important to fully understand those factors that affect outcome after ICH in order to improve medical management and further limit death and disability. For instance, seizures are a common occurrence after ICH or even a presenting sign of an ICH. About 4–20 % of ICH patients will suffer from clinical seizures (e.g., convulsions), whereas 30 % of ICH victims will have subclinical seizures observable on an electroencephalogram (EEG; [
2‐
4]). Current data suggests that the risk of seizures occurring within the first month is 8 % [
3], and the risk of a seizure occurring after the first month and within the first year is 3 % [
4]. Still, this could be an underestimate caused by the lack of continuous EEG monitoring in patients.
Intuitively, seizures are expected to worsen outcome after an ICH. Seizures can exacerbate excitotoxicity and oxidative stress [
5,
6], augment metabolic rate [
7], and cause re-bleeding or increased bleeding due to elevated blood pressure and blood flow during seizures [
3,
8,
7]. Intracranial pressure (ICP) also rises due to seizures [
9], which can cause complications after ICH (e.g., herniation) and increase mortality [
2]. Seizures may also cause aberrant brain plasticity (e.g., larger cortical maps) and impair recovery [
10]. Lastly, even though the number of ICH patients that develop epilepsy is relatively low (2–5 %), the incidence of one seizure increases the chance of developing epilepsy [
2,
4,
3].
All of this illustrates why seizures could be harmful. Clinical studies on this topic, however, have not consistently found that seizures are detrimental [
11,
2‐
4], although some support this notion [
12‐
14]. This variability among clinical studies could be attributed to several factors, such as inclusion criteria, methods for measuring EEG, lack of continuous EEG monitoring, use of anti-epileptic drugs (AEDs), among others. Current guidelines suggest that ICH patients with a depressed mental state should have continuous EEG monitoring, as most seizure activity occurring after ICH is subclinical [
2‐
4,
15]. Any seizure activity ought to be treated intravenously with an AED, and if seizure activity persists, AED treatment may continue orally. Prophylactic administration of AEDs, however, has been discouraged after evidence from studies indicating a worsening of outcome caused by administration of phenytoin before any signs of seizure activity after not only ICH [
15,
16] but also traumatic brain injury [
17].
The incidence and consequences of seizures after an ICH have not been well studied in animal models. Thus far, swine studies have shown that excitability increases in certain areas of the brain after ICH [
18]. Others have shown that in rodents, the most widely used ICH model, intracerebral infusions of blood components such as thrombin [
19] and iron [
20] cause seizures. To our knowledge, there have been no formal evaluations of seizure activity in the common rodent models of ICH, which involve injecting autologous blood [
21] or collagenase [
22] into the brain. Unlike the whole blood injection, bacterial collagenase, an enzyme that breaks down the basal lamina, causes bleeding over hours mimicking what frequently occurs in ICH patients [
22,
23]. Often, investigators target the striatum as it is a common site of ICH in humans and because it can contain a large hematoma that results in persistent, easily quantified, behavioral impairments [
24]. In this study, we induced a moderate-sized striatal ICH in rats by injecting collagenase or whole blood, and we monitored rats with an implanted EEG telemetry probe for a week after the stroke. By using telemetry, we were able to continuously record EEG in freely moving untethered animals, which is the least stressful method for these animals. The objective of this study was to determine the incidence and characteristics of seizures that occur in these animal models of ICH.
Discussion
We expected seizures to occur in the blood infusion model, but this was not observed. Our results did confirm our hypothesis that seizures commonly occur after striatal ICH in the collagenase model. Sixty-six percent of rats in the collagenase group suffered seizures during the first 36 h following their stroke. Seizures commonly occur after brain injury and stroke, both in patients [
30,
2,
31] and in other animal models of brain injury [
32‐
34]. Thus, it is not surprising that collagenase rats would also display abnormal electrical activity as we demonstrated in this study, including both full-blown ictal and abnormal interictal activity, which mostly occurred bilaterally. Increased cross-hemispheric coherence coinciding with increased power suggests that the activity in both hemispheres during seizure activity was coupled and that more severe seizures recruited both hemispheres. Although there were some exceptions to this, coherence remained significantly increased, especially at the higher frequencies. We also demonstrated that even normal-looking electrical activity has an increased RMS for the first 3 days after collagenase-induced ICH, which could indicate that there were abnormalities in non-epileptic EEG activity during this limited time frame. Although robust epileptiform activity was a consistent phenomenon in our collagenase group, we did not find lesion volume to be a predictor for any of the seizure characteristics. Likewise, the lack of seizures in the blood model, which had a comparable lesion, argues against lesion or hematoma volume as key predictors of seizure activity.
The incidence of electrographic seizures in our collagenase group (66 %) is more than double that documented in ICH patients [
2]. The difference in incidence might be attributed to the greater range in patient characteristics (e.g., ICH locations, severity) in clinical studies, along with other factors such as species differences. Interestingly, other pre-clinical studies of brain insults, namely, hypoxic-ischemic injury [
33], focal ischemia [
34], and traumatic brain injury [
32], all report a much higher percentage of animals developing seizures and epilepsy than what is reported in the clinic. However, continuous monitoring of EEG for many weeks or months is rare in clinical studies, making it difficult to compare animal and clinical data.
There are key differences between the collagenase and whole blood models that may explain the discrepancy in seizure incidence between these models. For instance, the whole blood model of ICH provides a somewhat different profile of injury (see Fig.
7a) with less secondary injury, inflammation, blood brain barrier damage, and smaller intracranial pressure spikes [
23,
35,
36], but these may vary by species [
37]. As with inflammation [
38], it is possible that the timing, extent, and localization of thrombin production vary between models and this might account for differences in seizure activity. Note that intracerebral infusions of thrombin induce seizure activity [
19]. Iron infusions also cause epileptogenic activity [
20], as we presently confirmed. However, given the timing of iron release, which in our collagenase model occurs between 24 and 72 h [
39], it is unlikely that iron causes seizures as they began between 10 and 22 h, and stopped by 36 h. As well, the hematoma volume is expected to be larger in rats infused with 100 μL of blood than those given collagenase [
23]. Thus, if iron were the primary cause of seizures, there should have been more seizures in the whole blood model.
Early seizures are predictors of future epilepsy in stroke patients [
13], although a study by Bladin and colleagues [
4] showed that all of the patients that had late onset seizures, which were those at 2 or more weeks after the stroke, developed epilepsy. We did not find recurrent seizures after the first 36 h of the stroke, even though we screened EEG for up to a month after collagenase infusion. While this suggests that this model does not lead to epilepsy, a much larger sample size is needed especially given the small percentage expected to develop that condition. There is also the possibility that seizures may develop later than a month after ICH in rats, as occurs in other animal models such as traumatic brain injury [
32].
There are some limitations to this study. First, we did not video record any of the seizure events, so the type of behavioral manifestations with these electrographic seizures remains unknown. Although, we occasionally noticed behavioral signs of focal seizures, such as clonic paw movements. Second, the relatively limited number of animals in this study cannot exclude the possibility that occasional seizure activity occurs in the whole blood model. Third, with larger sample sizes, a modest relationship between lesion volume and seizure characteristics may have been detected. Indeed, others have reported a relationship between epileptiform activity and infarct size after focal ischemia [
40]. In a clinical study, however, small lesion size was a better predictor of seizure incidence [
3]. Fourth, while we recommend use of telemetry probes, tethered systems have the advantage of allowing monitoring from more locations, which would be advantageous in future studies (e.g., to identify seizure focus). The use of telemetry probes also had some additional disadvantages (e.g., greater cost) including technical problems we encountered with the use of lead extenders and of course the inevitable loss of battery power. Lastly, while EEG eventually returned to normal, it is likely that seizure thresholds were altered as found in traumatic brain injured rats given a pro-convulsant challenge [
32].
Further studies should be carried out to advance our knowledge of seizures occurring after ICH. Seizure incidence should be studied after changing the location of the lesion, as lobar/cortical location has been associated with more seizure activity in patients [
3,
4]. Even though the striatal model of ICH is a common one, other structures have also been targeted, such as cortex [
18] and hippocampus [
41], and we are presently evaluating these models. It is possible that whole blood injections in different locations may elicit seizure activity. Also, patients with an ICH also have increased ICP after the insult [
15], which is also common after a collagenase-induced ICH in rats [
36]. This sustained rise in ICP could be associated either to the mass effect arising from the hematoma and edema or to seizure activity, which could especially be related to ICP spiking [
9]. This could be elucidated by simultaneous EEG and ICP monitoring. Furthermore, future research should focus on the relationship between seizures, cell death, and recovery. Some clinical studies have related seizures with worsened outcome and mortality [
12‐
14], although others failed to do so [
11,
2‐
4]. In animal models, we can experimentally increase seizure activity with convulsant drugs or diminish it with AEDs and test its impact on several markers of cell death (e.g., neurodegeneration) and functional outcome.
In conclusion, seizures occur in the majority of rats subjected to a collagenase-induced striatal ICH but did not occur after infusion of whole blood—models widely used to study the pathophysiology of ICH and to assess neuroprotectants and rehabilitation therapies [
35]. As ICH patients also suffer from seizures early after the stroke, the rat collagenase model has good face validity to model seizures occurring after ICH, although further work is needed to determine whether the underlying cause is the same as in patients. This is a key factor for translational purposes, as others have raised concerns regarding differences between animal and human ICH pathophysiology [
42,
43,
35,
44‐
46]. Researchers should also consider that seizures could potentially impact their studies. For instance, seizure activity may exacerbate the damage caused by the stroke, altering the effectiveness of neuroprotective therapies. Also, treatments may indirectly reduce cell death by ameliorating seizure activity. We recommend the use of the whole blood model when seizures may be a confounding factor. As this is the first study to find that seizures occur after collagenase-induced ICH in rats, we encourage further research to understand the relationship between seizures, cell death, and recovery after ICH. This way, we will be able to enhance therapies currently provided to ICH patients.