Cerebral magnetic resonance images of 14 patients with cerebral infarct–related early seizures. Patients were numbered in the order they were admitted to the stroke unit. Asterisks indicate patients with watershed infarcts.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Denier C, Masnou P, Mapoure Y, et al. Watershed Infarctions Are More Prone Than Other Cortical Infarcts to Cause Early-Onset Seizures. Arch Neurol. 2010;67(10):1219–1223. doi:10.1001/archneurol.2010.263
Early-onset seizures(ESs) have been reported in 2% to 6% of strokes. Most previous studies have been retrospective and did not systematically perform cerebral magnetic resonance imaging (MRI).
To determine the prevalence and determinants of ESs in a prospective cohort.
Prospective cohort study.
Stroke unit in an academic hospital.
Six hundred sixty-one consecutive individuals admitted to our stroke unit during an 18-month period for suspected stroke.
Main Outcome Measures
Initial investigations systematically included cerebral MRI. Among patients with MRI-confirmed cerebral infarction, individuals with ES, defined as occurring within 14 days of stroke, were identified.
Three hundred twenty-eight patients had MRI-confirmed cerebral infarcts and 178 had cortical involvement. The ESs, all initially partial seizures, occurred in 14 patients (4.3%) and at stroke onset in 5 patients. The ESs occurred exclusively in patients with cortical involvement (P <.001). With infarcts involving the cerebral cortex, there was a higher risk of ESs in watershed infarctions than in territorial strokes (6 of 26 [23.1%] vs 8 of 152 [5.3%], P = .007). Logistic regression analysis showed an almost 4-fold increased risk of ES in patients with watershed infarctions compared with other cortical infarcts (odds ratio, 4.7; 95% confidence interval, 1.5- 15.4; P = .01). Age, sex, diabetes mellitus, hypertension, smoking, National Institutes of Health Stroke Scale score, and cardioembolic origin were not significant risk factors for ES.
The cortical hemispheric location of ischemic strokes is associated with a higher risk of ES. Among these patients, the watershed mechanism is a strong and independent determinant of stroke-related ES.
Stroke-related seizures occur in 1.8% to 9.7% of infarcts, 2.8% to 26.1% of primary intracerebral hemorrhages, and 2.7% to 34.3% of subarachnoid hemorrhages1-8 (see Silverman et al9 for a review). Stroke-related seizures are classically subdivided into early-onset seizures (ESs), occurring within 2 weeks of stroke onset, and late-onset seizures (LSs), occurring later, knowing that the risk of recurrent seizures is higher in patients with LSs (vascular epilepsy).9,10 Interpretation and comparison of previous studies remain difficult because of the heterogeneity of their designs, stroke classification by computed tomography or magnetic resonance imaging (MRI), and lengths of follow-up.10
Independent risk factors for stroke-related seizures have long been recognized, including cortical location and hemorrhagic subtypes.3,6,11-19 Other risk factors, such as a cardioembolic mechanism or the severity of initial deficit, remain debated (see Camilo and Goldstein10 for a review). More recently, some cortical regions have been proposed to be more prone to provoke seizures.6,20
The aim of this study was to determine the prevalence and determinants of ESs in a prospective cohort of consecutive patients with MRI-confirmed cerebral infarction admitted to a stroke unit. We herein report watershed infarcts as a new risk factor for ESs.
This prospective stroke cohort includes 661 consecutive patients admitted for stroke or transient ischemic attack suspicion to the stroke unit (Stroke Center, Department of Neurology, Centre Hospitalo–Universitaire de Bicêtre) between June 1, 2007, and December 31, 2008. Initial diagnostic studies systematically included routine laboratory workup and cerebral MRI within 48 hours after admission (protocol: axial T1-weighted imaging, fluid-attenuated inversion recovery imaging, diffusion-weighted imaging, gradient-echo sequences, and intracranial time-of-flight and cervical vessel magnetic resonance angiography). All consecutive individuals have been explored by means of cerebral MRI except 8 (because of a pacemaker in 7 and severe claustrophobia in 1 [MRI in 98.8%]. In this cohort, 495 patients presented with various acute cerebrovascular diseases, including 328 MRI-confirmed cerebral arterial infarctions.
Patients with ischemic events underwent etiologic investigations, including systematic cervical and transcranial Doppler, echocardiography, and electrocardiographic monitoring for at least 48 hours. Additional explorations were performed if necessary (complementary biological tests, lumbar puncture, transesophageal echocardiography, and cerebral angiography). Based on the results of diagnostic studies, causes were classified according to the TOAST (Trial of Org 10172 in Acute Stroke Treatment) classification.21 Cerebral infarctions were also classified according to vascular territories determined by cerebral MRI in (1) total middle cerebral artery (MCA) infarction (cortical and subcortical ischemia due to an occlusion of the main stem of the MCA or of the internal carotid artery, (2) partial anterior circulation infarction (ischemia due to occlusion of a branch of the MCA, anterior choroidal artery, or anterior cerebral artery), (3) posterior circulation infarction (ischemia in the posterior cerebral artery [PCA] territory), (4) deep hemispheric infarcts (mainly lacunar strokes due to small vessel disease), (5) brainstem or cerebellar infarction, (6) multifocal infarctions, and (7) hemispheric watershed infarcts. Two types of hemispheric watershed infarcts associated with cortical involvement were considered in this study on diffusion-weighted MRI patterns: the border zone area between the superficial territories of the MCA and the anterior cerebral artery and between the superficial territories of the MCA and the PCA.22,23
All ESs were defined as occurring within 14 days of stroke in a paradigm comparable with posttraumatic epilepsy in which an arbitrary cutoff point of 2 weeks was considered to distinguish ESs and LSs.9,10 The exclusion criteria were a history of seizures before stroke, severe metabolic abnormalities, and a history of substance abuse, particularly alcohol.
Patient characteristics, including age, sex, history of hypertension, diabetes mellitus, current tobacco use, stroke severity based on National Institutes of Health Stroke Scale score, involved vascular territories, categories of etiologies of stroke, and ES occurrence, were routinely entered into the database. Different groups were compared using χ2 or Fisher exact tests for categorical variables and the Kruskal-Wallis test for quantitative variables. Among patients with cortical infarctions, relations between ES occurrence and several risk factors were first evaluated in univariate analysis using Fisher exact or Kruskal-Wallis tests (age, sex, presence or absence of diabetes, hypertension, tobacco use, hypercholesterolemia, cardioembolic origin, stroke disability [National Institutes of Health Stroke Scale score], and presence of watershed vs territorial cerebral infarctions). Then, variables related to ESs in univariate analysis with P < .15 were introduced into a multivariate logistic regression model to determine the independent factors of ES occurrence. The significance level was set at α = .05.
Of 328 consecutive patients with MRI-confirmed cerebral infarctions included in this study (200 males and 128 females; mean age, 63.6 years; age range, 17-99 years), 31 had total MCA infarction, 88 had partial anterior circulation infarction (due to occlusion of a branch of the MCA [n = 70], anterior choroidal artery [n = 11], or anterior cerebral artery [n = 7]), 26 had partial posterior circulation infarction (due to occlusion of a branch of the PCA), 70 had deep hemispheric infarcts (mainly lacunar stroke), 26 had hemispheric watershed infarcts, 62 had brainstem or cerebellar infarctions, and 25 had multifocal infarctions (Table 1 and Table 2). Age, sex, history of hypertension, diabetes mellitus, hypercholesterolemia, and current tobacco use did not significantly differ between these groups (data not shown).
Seizures within 14 days of stroke occurred in 14 of the 328 patients with cerebral infarctions (4.3%; mean age, 63.1 years; age range, 43-77 years; 8 men) (Tables 1 and 2). The ESs were initially partial seizures in all 14 patients, followed by secondary generalization in 9 and complicated by status epilepticus in 3. The ESs occurred at stroke onset in 5 patients (36%), after stroke onset but within the first 24 hours in 3 (21%), and between 24 hours and 14 days in 6 (43%) (Table 1).
Findings from electroencephalography within the first 24 hours after seizures were normal in 4 patients, showed asymmetrical background activities in 5, and had focal slow waves in 5. Periodic lateralized epileptiform discharges were not observed in any patients.
Of 328 patients with cerebral infarctions, cortical involvement was present in 178 (including 108 anterior and 26 posterior circulation infarcts, 26 watershed infarcts, and 18 of the 25 multifocal infarcts). All ESs occurred in this group (14 of 178 vs 0 of 150; Fisher exact test, P < .001). No difference between the groups with and without ESs was observed concerning age, sex, history of hypertension, diabetes, hypercholesterolemia, current tobacco use, stroke severity (initial National Institutes of Health Stroke Scale score), or cardioembolic origin (Table 2). Although a history of symptomatic stroke or transient ischemic attacks was not systematically entered into the database, only 1 of the 14 patients with ES previously had symptomatic cerebrovascular disease (PCA infarction 2 years earlier) (patient 27).
Among infarcts with cortical hemispheric involvement, ESs occurred more frequently in patients with watershed infarcts (6 of 26 [23.1%]) than in patients with territorial infarcts (8 of 152 [5.3%]) (Fisher exact test, P = .007). Logistic regression analysis showed that there was an almost 4-fold increased risk of ESs in patients with watershed infarcts compared with other cortical infarcts (multivariate logistic regression: odds ratio, 4.7; 95% confidence interval, 1.5-15.4; P = .01). Other variables between these 2 groups did not differ (age, sex, diabetes, hypertension, hypercholesterolemia, tobacco use, and stroke severity) (data not shown).
Stroke-related ESs occurred in 4.3% of patients with cerebral infarcts, within the range (1.4%-6.0%) in most recent series.1-7,14,15,24-28 The classic association between ESs and cortical localization is confirmed in the present study.1-5,15,26,29-32 The new finding is the association between ESs and watershed infarcts. This association has been previously reported in 1 article in different conditions24: the authors studied ESs and LSs occurring after cerebral hematomas and infarctions explored by means of computed tomography (including two-thirds of patients with seizures occurring more than 1 month after stroke).24
While LSs are thought to be caused by permanent changes, such as gliosis and meningocerebral scars, stroke-related ESs are thought to result from cellular biochemical dysfunctions leading to electrically irritable tissue (see Camilo and Goldstein10 for a review). Acute ischemia is known to lead to increased extracellular concentrations of glutamate, an excitatory neurotransmitter, and intracellular calcium and sodium, which result in depolarization of the transmembrane potential and other calcium-mediated effects. Neuronal discharges can also occur from surviving neurons exposed to glutamate facilitated by local ionic shifts that lower the seizure threshold. While LSs are independent high-risk factors for the subsequent development of recurrent epilepsy, ESs may, according to recent experimental data, increase infarct size and hamper later recovery. Moreover, antiepileptic drug therapy may also impair recovery after stroke (see Camilo and Goldstein10 for a review).
Although electroencephalographic recordings showing periodic lateralized epileptiform discharges have been previously reported to be associated with watershed infarcts33 and might reflect a key pattern for focal hyperexcitability in the penumbra zone of ischemic stroke,34 periodic lateralized epileptiform discharges were not identified in any of the present patients.
Two distinct etiopathologic hypotheses—hemodynamic compromise and microembolism—have been proposed for hemispheric watershed infarcts.35 Both could promote seizures via focal hypoperfusion or distal artery-to-artery embolism. In the present patients, low flow existed because we observed severe carotid stenosis or occlusion in 19 of 26 patients with watershed infarcts (Figure and Table 1, [data not shown]). Future explorations, including brain diffusion-weighted and perfusion imaging and ultrasound detection of microembolic signals, should help us better understand the physiopathologic features of these infarcts and their association with ESs.35
Although limb shakings have been described in association with high-grade stenosis or occlusion of the internal carotid artery,36 we did not observe any limb-shaking episodes in the present series. Some of these episodes could have been mistaken as seizures. The fact that most of the ES episodes generalized in this population suggests that these, indeed, were seizure disorders.
In conclusion, ESs are rare, occurring in 4.3% of patients with cerebral infarcts; all ESs occurred in cortical infarcts; and ESs are 4 times more frequent in watershed infarcts than in territorial infarcts.
Correspondence: Christian Denier, MD, PhD, Department of Neurology, Centre Hospitalo–Universitaire de Bicêtre, 78 rue du Général Leclerc, 94275 Le Kremlin-Bicêtre Cedex, France (firstname.lastname@example.org).
Accepted for Publication: March 15, 2010.
Author Contributions:Study concept and design: Denier, Masnou, and Adams. Acquisition of data: Denier, Masnou, Souillard-Scemama, Guedj, Théaudin, Fagniez, Join-Lambert, Lozeron, and Ducreux. Analysis and interpretation of data: Denier, Masnou, Mapoure, Souillard-Scemama, Ducot, and Adams. Drafting of the manuscript: Denier. Critical revision of the manuscript for important intellectual content: Masnou, Mapoure, Souillard-Scemama, Guedj, Théaudin, Fagniez, Join-Lambert, Lozeron, Ducot, Ducreux, and Adams. Statistical analysis: Denier and Ducot. Administrative, technical, and material support: Denier, Masnou, Souillard-Scemama, and Ducreux. Study supervision: Denier and Adams.
Financial Disclosure: None reported.
Additional Contributions: M.-G. Bousser, MD, provided helpful discussions and excellent critical reading of the manuscript.
Create a personal account or sign in to: