Until the 1950s, studies of seizures after stroke were largely confined to necropsy studies.33 The advent of computed tomography (CT) in the 1970s and MRI a decade later markedly increased the accuracy of clinical diagnosis of stroke mechanisms, particularly with respect to distinguishing between ischemic and hemorrhagic infarcts. Nevertheless, comparison among different studies has been difficult because of different inclusion and exclusion criteria. Some have examined only ischemic stroke. Others have combined ischemic and hemorrhagic stroke, and some of these have excluded subarachnoid hemorrhage or vascular malformations.4 Fortunately, there have been several recent prospective studies of early,34–36 or both early and late,37,38 seizures that have followed large populations (more than 500 patients) for an adequate period of time to draw valid conclusions and have attempted to separately analyze clinically important subgroups. The smaller prospective or case-control studies,33,39–48 although less valid as sources of incidence data, still provide useful information about risk factors and clinical characteristics. Separate series of patients with intracerebral19,49–51 or subarachnoid52–57hemorrhage or specific vascular malformations58–62 have provided important data about these subgroups.

Prospective analyses of combined groups have found lower overall seizure rates after ischemic stroke than hemorrhagic stroke. Bladin and colleagues,37 for example, found that 8.6% of 1,632 patients with ischemic stroke had seizures, versus 10.6% of 265 patients with hemorrhagic stroke. Because of higher mortality among patients with hemorrhagic stroke, survival analysis revealed an almost twofold increased risk of seizures in that group.

Among the patients in this study who had ischemic stroke, 4.8% had seizures within the first 2 weeks (early), including 3.4% within 24 hours (“onset”). For patients with hemorrhagic stroke, 7.9% had early seizures, with 6.0% at onset. Late seizures occurred with 3.8% of ischemic strokes and 2.6% of hemorrhagic strokes, and were more likely to recur than early seizures, a finding verifying nearly all earlier studies. In multivariate analysis, cortical location was the only risk factor for seizures after either type of stroke. Increased disability predicted seizures among the ischemic group. Hemorrhagic transformation conferred greater risk on univariate analysis, but a high likelihood of embolism did not, a finding confirmed by a large cohort in the National Institute of Neurological Disorders and Stroke’s Stroke Registry,63 despite conflicting results in earlier studies45,64 and in one large retrospective study.34

Lacunar infarction was associated with seizures in 2.6% of cases, although further analysis questioned this relationship. However, risk factors for lacunar disease, including hypertension, serum cholesterol, and left ventricular hypertrophy,65–67have been associated with the development of seizures or epilepsy, even in those without overt stroke.

Kilpatrick and coworkers35 found early seizures in 24 (4%) of 604 patients with ischemic stroke (or 4.4%, excluding brain stem and cerebellar strokes, as did Bladin et al.37). All 24 patients with early seizures had cortical infarcts of the anterior circulation, although again embolism was not a risk factor. So and colleagues38 found 4.7% onset seizures and 6.2% early (1 week) seizures among those with infarction; initial late seizures occurred in an additional 5% of patients, and epilepsy in 3.3%. Early seizures were a significant risk factor for late seizures and epilepsy. The cumulative risk for initial late seizures was 7.4% by 5 years and 8.9% by 10 years.

A British prospective study of 675 patients with a first stroke, 545 of whom had infarction, found a 5-year actuarial risk of seizure of 11.5%.40 Such population-based studies have found seizure risk relative to the general population to be elevated by factors of 20 to 40. Early (2 weeks) seizures in a Danish cohort of 1,197 patients with ischemic infarct occurred in 4.2% of patients, 2.8% within 24 hours.36 Stroke severity was the only risk factor in a multivariate analysis.

Seizures have been attributed to transient ischemic attack (TIA) in 1–4% of patients.34,35 In some earlier studies,48,68 the possibility of seizures heralding TIA or stroke was advanced, although this relationship is tenuous.

Smaller studies examining occurrence of late seizures or epilepsy39,47 confirm the importance of large cortical infarcts and suggest a role for apparently preserved cerebral tissue within the infarcted area.39 Extensive white matter disease in combination with cortical infarction is also important.

No specific incidence figures have been published for seizures following border zone infarcts. One would expect that when these infarcts extend to the cortex, there would be risk similar to that of thrombotic lesions. However, the common electroencephalogram (EEG) occurrence of periodic lateralizing epileptiform discharges (PLEDs) after border zone infarcts69argues that the frequency of seizures may be higher and that epilepsia partialis continua or other forms of partial status epilepticus might occur.

There are no data concerning seizures with less common stroke mechanisms, such as vasospasm as a result of migraine or vasculitis—only limited evidence linking seizures to specific cortical locations. An excess of seizures with strokes involving the anterior rather than posterior circulation may relate to a higher likelihood of cortical involvement. Within a given vascular territory, the probability of seizures or epilepsy may also relate to the intrinsic epileptogenicity of specific cortical regions. Limited evidence parallels that for brain tumors70 or unselected patients71 and suggests the highest probability associated with perirolandic cortex, followed by the temporal lobe, then prefrontal, parietal, and occipital regions.

Specific studies of intracerebral hemorrhages confirm that early seizures are common, occurring in 4.6–17.0%.50,51 All studies show lobar hemorrhages to have the highest rates, 15–24% or more.19,35,50,72 Most of these patients had hypertension. Among those with deep hemorrhages, the rate of early seizures ranges from 0% to 11%.19 The caudate and perhaps putamen are most often involved, and the thalamus least involved. As in infarctions, the majority of early seizures occur within the first day or two. About half of these patients had one or more seizures as the first symptom. Late seizures and epilepsy are much less common, even in those with early seizures. Between 2 and 5 years, only 6.5% of survivors had any seizures, in contrast to an observed cumulative incidence of 32%.19 Furthermore, life-table analysis in the same study suggested a cumulative prevalence of seizures to be 50% had all patients survived to 5 years.

Many patients with lobar hemorrhage in early studies may have had cerebral amyloid angiography, although specific data for seizure risk in this syndrome are not available. Nor are figures available specifically for hemorrhage caused by coagulopathies, but location probably plays an important role here also. In a retrospective study to evaluate the incidence of seizures in chronic subdural hematomas (some of which could result from coagulopathies), the rate of occurrence was less than 2%.73 After hemorrhage from dural arteriovenous fistulas, 5% of patients presented with a generalized seizure.18

In general, the incidence of epilepsy as a late sequela of stroke has been estimated at 3% to 10%, with those who have a late-onset seizure at higher risk (a second unprovoked, late-onset seizure by definition constituting epilepsy). Late seizures may occur earlier after hemorrhage than after infarction.45 The timing of the initial late seizure (after 1–2 weeks) does not predict later recurrence.74

Status epilepticus can occur, involving 31% of stroke patients with seizures in one large study. Status epilepticus was the initial seizure type in more than half of these patients.75

With respect to hemorrhagic infarcts due to vascular malformations, AVMs commonly (17– 40%) present with seizures.58,61,62Presentation with a seizure alone is more common in younger patients. The cumulative risk of epilepsy in untreated patients is estimated at 1% per year, relative to a 2–4% risk of hemorrhage. Cavernous angiomas are commonly undiagnosed until a seizure occurs, with 40–70% presenting with seizures, usually in midlife; epilepsy is typical if the lesion is untreated.59,60Venous angiomas rarely bleed and are usually incidental findings, but when hemorrhage occurs, seizures are common.

Subarachnoid hemorrhage presents with a seizure in a sizable minority (6.3–18.0%) of patients.55,56 In one study, thickness of the cisternal clot was the only predictive factor. Another 7% may have had presenting seizures, but loss of consciousness at the time of aneurysm rupture more commonly reflects temporary cessation of cerebral perfusion secondary to an acute rise in intracranial pressure. Acute symptomatic seizures may occur in 24–26%, and epilepsy can later develop in 25%, with a higher risk if there are neurologic sequelae and if acute seizures occurred.57,76 In a small retrospectively studied cohort with unruptured intracranial aneurysms, the overall risk of postoperative seizures in initially seizure-free patients was 15.7%.77

With cerebral venous thrombosis, seizures have been described in 40% of cases,28 but subsequent epilepsy appears rare, despite the high incidence of hemorrhage.

Acute symptomatic seizures are one of the hallmarks of hypertensive encephalopathy and related conditions (e.g., eclampsia, hyperperfusion syndrome), occurring in the majority. Later epilepsy is rare, however, paralleling the resolution of the often posterior white matter and cortical lesions typically seen on MRI.23,24

In the United States, migraine occurs in 18% of women and 6% of men.78 The prevalence of migraine with aura (which includes hemiplegic migraine) is lower, around 4%,79 and only a tiny fraction of patients ever experience the neuroimaging changes or persistence of symptoms for more than 1 week that defines migrainous infarction. No data are available on the frequency of seizures or epilepsy as a complication of this cause of stroke.

Adapted from: Bromfield, EB, and Henderson GV. Seizures and cerebrovascular disease. In: Ettinger AB and Devinsky O, eds. Managing epilepsy and co-existing disorders. Boston: Butterworth-Heinemann; 2002;269–289.
With permission from Elsevier (www.elsevier.com). 

Reviewed By: 
Steven C. Schachter, MD
Thursday, April 1, 2004