Aneurysmal subarachnoid hemorrhage (SAH) accounts for about 3% of all strokes but it is associated with a high mortality and morbidity, typically affecting patients in the midst of their productivity. Approximately 10–15% of patients with aneurysmal SAH die within the first minutes or hours after hemorrhage, even before reaching a medical center [4]. Outcomes of aneurysmal SAH are better for patients that are treated at high volume medical facilities that have both neurosurgical and endovascular techniques available and treated more than 100 patients per year [5]. Case-fatality rates have been trending down over the past decade as a result of early securing of the aneurysm, better detection of complications including early treatment of hydrocephalus and delayed cerebral ischemia, and neurocritical care management in general [6]. For instance, Hop JW et al. demonstrated an 8% decline in case-fatality rate of aneurysmal SAH per decade between 1960 and 1992 [7]. Similar observation was made by Vergouwen MD et al. in a recent study which showed ninety-day case fatality in 2009–2012 compared to 1999–2002 period [8]. This also supports the value of neurointensive care in improvement of outcome of these patients.

Seizures are a well-known complication of SAH. The incidence of early and late convulsive seizures after SAH has been, respectively, reported to be 1.1–16% [9, 10] and 5.1–25% of patients [11, 12] in literature depending on the definition used. Seizures may occur at different stages of the disease, including at the time of initial presentation typically prior to hospital admission, within 2 weeks of ictus during the ICU or hospital stay, and in a delayed fashion up to years post-SAH [13]. Less frequently, seizure may be seen due to unruptured cerebral aneurysms. Kamali AW et al. reported five patients with complex partial seizures that were found to have cerebral aneurysms arising from middle cerebral or internal carotid arteries adjacent to the presumed epileptogenic area. The mechanisms implicated were calcification of the aneurysmal wall with compression or ischemia in the adjacent tissue or distal emboli originating from the aneurysm itself [14].

Convulsive seizures after SAH have been associated with worse admission neurological function expressed as a higher aneurysm grade based on the Hunt Hess or World Federation of Neurosurgical Societies Scale [or lower Glasgow Coma Scale (GCS)], larger amount of subarachnoid blood on head CT scan, rebleeding of the aneurysm, and overall poor outcome [15, 16]. Seizures can also occur following the treatment of aneurysms. In the International Subarachnoid Aneurysm Trial (ISAT) trial, it has been showed that the risk of seizure is higher in patients who undergo surgical clipping than those patients who get endovascular coiling [17].

In two studies with aneurysmal SAH, the highest incidence of seizure was in anterior cerebral or communicating artery aneurysms, followed by middle cerebral artery aneurysms [18, 19]. In addition to the direct effect of SAH on seizure incidence as the result of cortical irritation from blood products, surgical or less likely endovascular treatment of an aneurysm may also contribute to early post-procedure seizures or epilepsy. Chang TR et al. conducted a prospective study of 1134 patients with aneurysmal SAH. Overall, 16% of 182 patients developed seizures, which were more frequently seen in younger patients (<40 years old), those with poor clinical grade, that also had intracerebral hemorrhage, MCA location of the aneurysm, and cocaine use [20].

Hyponatremia , which is a common electrolyte abnormality in patients with SAH, as a result of either cerebral salt wasting or SIADH, may lower seizure threshold after SAH. In a retrospective study of 316 patients with SAH , Sherlock M et al. showed that 179 patients (56.6%) developed hyponatremia (defined as serum sodium less than 135 mmol/L), which was due to SIADH in most of the patients (69%). Seizures developed in 14.5% of 62 patients with serum sodium less than 130 mmol/L, but in this study there was no significant difference in the incidence of seizures between patients with severe hyponatremia or plasma sodium less than 125 mmol/L (3/18) and patients with mild hyponatremia or plasma sodium between 125 and 130 mmol/L (6/44). Overall, hyponatremia was associated with longer hospital course, but it did not independently affect mortality [21].

Continuous EEG monitoring may detect non-convulsive status epilepticus (NCSE) as a cause of unexplained alteration of mental status or coma in SAH patients. Based on multiple studies, the rate of non-convulsive seizures and NCSE in patients with aneurysmal SAH has been estimated to be 7–18% and 3–13%, respectively (Kondziella, Claassen, Claassen) [22–24]. In poor grade SAH patients non-convulsive seizures may be particularly difficult to detect clinically. Dennis et al. found that almost every third patient with poor grade SAH and unexplained alteration of consciousness who underwent continuous video EEG monitoring was in NCSE. All of these patients received prophylactic anti-epileptics at the time of admission. Four patients were persistently comatose and four revealed deterioration to stupor or coma; only one demonstrated overt generalized tonic-clonic seizure before entering to coma; two patients had subtle eye blinking and facial twitching. NCSE was successfully treated for five patients (63%), but only one experienced clinical improvement , which was transient. In this small study, the mortality rate was 100% in all patients after a period of prolonged coma [25]. NCSE in SAH patients has been associated with older age (mean age 68 years) and high mortality (82–100%) [22]. The following risk factors for NCSE were identified in another study done by Claassen et al.: poor Hunt and Hess grade (IV or V), older age, ventricular drainage, and cerebral edema on CT [26]. In addition to seizures recorded on the surface recent studies indicate that focal seizures may be even more frequently recorded on invasive brain monitoring comatose SAH patients using depth electrodes [26].

EEG monitoring, initially used primarily to detect electrographic seizures, is now routinely applied for many SAH patients and has revealed a number of additional EEG patterns. Some of these, such as periodic discharges, while associated with poor outcome, are of uncertain significance and what is not clear is that their aggressive treatment affects outcome. In a prospective study of 756 patients with SAH, certain EEG findings on continuous video EEG monitoring during the time of hospitalization were associated with poor outcome. These EEG findings include the absence of sleep architecture, the presence of periodic lateralized epileptiform discharges (PLEDs) , bilateral independent lateralized epileptiform discharges (BiPLEDs) , and NCSE [24]. Interestingly, aside from seizure detection, continuous video EEG monitoring in patients with poor grade SAH can provide a diagnostic tool for detection of cerebral ischemia due to vasospasm as predictable EEG changes are observed with decreasing brain perfusion [23, 27, 28].

In the recent guidelines for the management of patients with SAH published by American Heart Association, it is recommended to use anti-epileptics for seizure prophylaxis for 3–7 days following the bleed (Class IIb; Level of Evidence B) [6]. The main reason for prophylactic use of anti-epileptics after SAH in the ICU setting is to decrease the likelihood of catastrophic events such as aneurysm re-rupture and ICP crisis which are usually associated with high mortality and morbidity. In addition, seizures may cause increasing cerebral metabolism in an already stressed brain after acute injury, which may precipitate additional brain injury. Following acute brain injury such as subarachnoid hemorrhage, increased metabolism and blood flow have been demonstrated with the onset of non-convulsive seizures. Interestingly, brain oxygen may drop, intracranial pressure may rise rapidly, but regional cerebral blood flow may increase only minutes and not seconds after the onset of the seizure [26, 29]. One possible explanation for these observations may be that seizures cause more damage in acutely brain injured patients than in those with epilepsy, damaging intrinsic defense mechanisms such as vasoreactivity.

Most series exploring the benefits of seizure prophylaxis after SAH studied the effects of phenytoin, which was associated with worse functional and cognitive outcomes in non-controlled case series [30, 31]. Phenytoin , metabolized by the hepatic cytochrome P450-3A4 system, increases the metabolism of the calcium channel antagonist nimodipine, therefore decreasing its bioavailability [32]. This is problematic since nimodipine is recommended based on clinical trials (Class I; Level of Evidence A) as a treatment to decreased morbidity from delayed cerebral ischemia [6]. With more favorable safety profiles, newer AEDs, such as levetiracetam, are increasingly used for seizure prophylaxis following SAH. In a recent survey from 25 centers in the USA, levetiracetam was used in 94% as the anti-epileptic agent of choice for seizure prophylaxis in SAH [33].

Early identification and treatment of seizures or status epilepticus (SE) is of paramount importance and should be achieved as soon as possible to avoid refractory or super-refractory status epilepticus. The first line agents in standard protocol for treatment of SE are parenteral lorazepam or midazolam. The second line treatment includes intravenous fosphenytoin or phenytoin, valproate, levetiracetam, or lacosamide based on comorbidities, availability of medications, and preference. In case of failure of first and second line agents, the third line therapy should be initiated promptly in conjunction with video EEG monitoring in an intensive care unit. At this point, patients need to be intubated for airway protection. The third line treatment includes phenobarbital, another second line agent or intravenous anesthetics such as midazolam , propofol, pentobarbital, or ketamine [34].

Spontaneous intracerebral hemorrhage (ICH) or intra-parenchymal hemorrhage constitutes about 15% of all strokes and is associated with high mortality and morbidity. The most common causes of spontaneous ICH include hypertension, amyloid angiopathy, hemorrhagic conversion of ischemic stroke , arteriovenous malformation (AVM) , cavernous malformation, other vascular abnormalities, and hemorrhagic complications of primary or metastatic brain tumors [35]. Aggressive management of secondary complications has been associated with improved outcomes after ICH.

Seizures are common following ICH with an estimated 30-day risk of 8% [36]. Most seizures (90%) occur in first 72 h of admission but delayed seizures and later development of epilepsy are also seen [37]. Neshige S et al. most recently performed a retrospective study of 1920 patients with ICH over a period of 8 years [38]. These researchers observed that the rate of seizures was 6.6% (127/1920) and more specifically 4.3% for early seizures (defined as seizures within one week of ICH onset) and 2.3% for late seizures (those occurring after the first week from ICH onset). Use of continuous video EEG monitoring in neuro-ICUs has played a major role in detecting and appropriately treating subclinical non-convulsive seizures in ICH, which are estimated at 18 to 28% [39–41].

A number of characteristics increase the risk of convulsive seizures in ICH including lobar hemorrhage (particularly non-occipital and juxtacortical locations), hematoma size, low GCS, focal neurologic deficit, presence of hydrocephalus, and midline shift on CT head (Bladin CF, 2000; De Reuck J, 2007; Weisberg LA, 1991). In multivariate analyses, the following were significantly associated with the development of both convulsive and non-convulsive seizures following ICH: cortical location of the ICH (odds ratio 7.37, 95% CI 4.77–11.45); non-hypertensive etiology (1.63, 1.02–2.56); high NIH stroke scale at the time of admission (1.04, 1.02–1.05), and younger age (0.97, 0.96–0.98). Interestingly, hematoma volume was the only independent factor associated with recurrence of seizures (1.02, 1.001–1.023) [38].

The association between ICH-related early seizures and prognosis or functional outcome has been recently studied. Brüning et al. in a retrospective study of 484 patients with spontaneous non-traumatic ICH from a single institution showed that there was no statistically significant association between early seizures and functional outcome measured by modified Rankin Scale (mRS) [42]. De Herdt et al. in a prospective study of 562 patients with spontaneous ICH found that early seizures did not affect functional outcome at 6 months [43]. Electrographic seizures, on the other hand, were associated with poor overall 3 months outcome and hematoma expansion [40] and increasing midline shift [43].

The underlying pathophysiology of seizures after ICH depends on the time that seizures are first recorded. Early onset seizures which occur within 2 weeks of ICH are primarily caused by anatomical destruction due to mass effects of hematoma and also biochemical disruption at the cellular level [3]. Delayed seizures that occur after 2 weeks of ICH onset are thought to be related to gliosis and chemico-cellular repair processes creating an epileptogenic focus [3, 40].

The role of seizure prophylaxis, choice of anti-epileptic agent, and its duration are controversial in the absence of randomized clinical trials. In the past, prophylactic AEDs were recommended for 30 days after lobar ICH [44, 45]; however, this is not generally accepted by many authorities any more. Based on the most recent American Heart Association guidelines regarding the management of ICH, prophylactic AEDs are not recommended (Class III, level of evidence B) [46]. Naidech AM et al. in a prospective study of 98 patients with ICH showed that prophylactic use of phenytoin was associated with poor functional outcome defined as mRS of 3–6 [47]. Gilad R et al. in a small randomized clinical trial of 72 patients with ICH showed that 7 patients of the treatment group (total 36) and 8 patients of the placebo group (total 36) developed seizures within a 1 year follow-up period, a non-significant difference. They concluded that prophylactic treatment with valproic acid for 1 month following ICH did not prevent seizures in 1 year follow-up [48]. Sheth KN et al. performed a prospective cohort study of 744 ICH patients. AEDs, mainly levetiracetam (89%), were prescribed for seizure prophylaxis in 289 (39%) patients. The authors found that use of prophylactic AEDs was associated with poor outcome in an unadjusted model (O.R 1.4; 95% CI 1.04–1.88); however, after adjusting for clinical data including age, hematoma volume, GCS, IVH presence, and lobar hemorrhage in multivariate logistic regression, no significant association was found (1.11, 0.74–1.65) [49].

Continuous video EEG monitoring is recommended in ICH patients with unexplained alteration of level of consciousness or any suspicion for subclinical seizures (Class IIa; level of evidence C) [46]. Brain hemorrhages themselves may cause alterations of EEG rhythms depending on their location, Delta activity may be observed over the affected hemisphere in deep capsular/basal ganglia hemorrhages, at times arising in rhythmic runs of moderate amplitude [50, 51]. Thalamic hemorrhages may be associated with ipsilateral delta activity, a reduction or enhancement of the alpha rhythm depending on the precise location within the thalamus, and absence of sleep spindles [50, 52]. Bleeds located in the mesencephalic area may cause diffuse theta activity [50]. Hemorrhages in the mid-lower brainstem can lead to diffuse attenuation or a lack of reactivity [50, 53].

Arteriovenous malformations (AVM) present frequently before age 40 and they are almost equally distributed between men and women. The most common indication for ICU admission for AVMs is ICH, which is seen in 30–82% of patients. The most common clinical presentations of cerebral AVMs include ICH, seizures, headache, and focal neurologic deficits [54]. The rate of seizure as initial manifestation of AVMs has been estimated to be about 16–53%. Seizures associated with the presence of an AVM can occur as a result of: (1) secondary ICH; (2) unruptured AVMs causing epileptogenesis in the surrounding brain tissue; (3) treatment, including post-surgical resection, post-endovascular coiling or even stereotactic radiosurgery [55–60]. In a prospective study of 229 patients with AVM, Josephson et al. found that the risk of developing epilepsy within 5 years after a first ever seizure was 58% (95% CI 40–76%). The risk of a first ever unprovoked seizure in 5 years was higher in patients with AVM and ICH or AVM with focal neurologic deficit compared to asymptomatic AVMs who were found incidentally (23%, 9–37% vs 8%0–20%) [61].

There are several factors that increase the risk for post-operative seizures in patients with AVMs. In a retrospective study , Hoh et al., a multidisciplinary neurovascular team, treated 424 patients with cerebral AVMs. One hundred forty-one (33%) of those had seizures before treatment. Follow-up data was available in 110 (78%) of these patients for a mean period of 2.9 years after treatment. Based on Engel Seizure Outcome Scale, there were 73 (66%) Class I (free of disabling seizures), 11 (10%) Class II (rare disabling seizures), 1 (0.9%) Class III (worthwhile improvement), and 22 (20%) Class IV (no worthwhile improvement) outcomes. Sixteen (5.7%) patients experienced new-onset seizures after treatment. A limited seizure history (<5 seizures before treatment), association of seizures with intracranial hemorrhage , generalized tonic-clonic seizure type, deep and posterior fossa AVM locations, surgical resection, and complete AVM obliteration were statistically associated with Class I outcomes. In the entire cohort, surgery resulted in seizure elimination in 81%, radiosurgery in 43%, and embolization in 50% of patients treated. When only completely obliterated AVMs were considered, no statistically significant differences between surgery, radiosurgery, and embolization were observed [62].

Other factors increasing the post-operative risk for seizures include age less than 30 years at seizure onset, pre-operative seizure duration greater than 12 months, AVM size greater than 3 cm, location in the medial temporal or peri-Rolandic cortex, and previous hemorrhage or hemosiderin deposition [63–66].

Treatment for AVM-related seizures is controversial. In a recent meta-analysis, Josephson et al. concluded that it is still unclear whether invasive treatment for AVM-related seizures is superior to AEDs or not. Interestingly, there is still no randomized clinical trial to compare invasive treatment versus medical management for AVM-related seizures, probably because the focus of the treatment is to obliterate the AVM and decrease the risk for bleeding instead of treating seizures [67].

Cerebral cavernous malformation (CCM) consists of a tangle of blood vessels which lack elastic and muscular layers without intervening brain tissue. In addition, the endothelial cells are not connected together via tight junctions causing them susceptible to hemorrhage (Al-Shahi Salman R, 2012). CCMs can occur sporadically or be familial with an autosomal dominance inheritance pattern with incomplete penetrance and variable clinical presentations. Their overall incidence has been reported to be around 0.5% in the general population. The most common clinical presentation of CCMs is seizures, which occur in 23–79% of patients [68–70]. ICH is another frequent complication of CCMs, particularly in Hispanic patients, and this may also lead to seizures [71]. Of those patients who present with seizures, 40% develop medically refractory epilepsy for which surgical resection is the treatment of choice [72]. The exact mechanism of epileptogenesis in CCM is still unclear. It has been hypothesized that chronic microhemorrhages from the CCM causes deposition of hemosiderin, a blood degradation product, in the adjacent brain tissue. Hemosiderin can irritate the brain parenchyma by generating free radicals and lipid peroxides causing excitotoxicity on the adjacent neuronal tissues [73, 74]. Based on a recent meta-analysis, total surgical resection of CCMs including complete removal of the hemosiderin ring is associated with better seizure control compared to surgical resection without excision of the hemosiderin ring [75].