Incidence and Predictors of Late Seizures in Intracerebral Hemorrhages
Background and Purpose—To identify incidence and predictors of late seizures (LS, occurring >1 week of stroke) in a cohort of patients with intracerebral hemorrhage (ICH).
Methods—Prospective cohort of consecutive adults with spontaneous ICH. Incidence and predictors were identified with Cox regression. We included multivariate analyses on MRI biomarkers (global cortical atrophy, leukoaraiosis, brain microbleeds).
Results—Our study population consisted of 325 patients: 54% men, median age 70 years (interquartile range, 58–79). During 778 person-years of follow-up, the incidence rate was 4 new cases/100 person-years (95% confidence interval, 3–6). The median delay between ICH and LS was 9 months (interquartile range, 3–23). The only factor independently associated with the occurrence of LS was a cortical involvement of the ICH (hazard ratio, 2.8; 95% confidence interval, 1.3–6.1). Concerning MRI biomarkers, multivariate analyses found lobar brain microbleeds to be associated with LS (hazard ratio, 2.4; 95% confidence interval, 1.1–5.4), especially if ≥3 (hazard ratio, 2.7; 95% confidence interval, 1.1–6.8). LS were associated with a worse functional outcome after 3 years of follow-up (P=0.009).
Conclusions—LS frequently occur >9 months after ICH onset, imposing a long-term follow-up. The association of lobar brain microbleeds with the risk of LS might suggest a link with the underlying vasculopathy (cerebral amyloid angiopathy).
Data on intracerebral hemorrhage (ICH)–related seizures are scarce and often come from mixed cohorts with a reported incidence of 4% to 16%.1,2 We previously reported a 14% incidence of early seizures (ES), which were associated with cortical involvement of the ICH and did not influence in-hospital mortality or outcome at 6 months.3 However, their influence on the risk of developing late seizure (LS) remains uncertain.
Our aim was to identify the incidence and predictive factors of LS in patients with a spontaneous ICH.
Patients and Methods
The Prognosis of InTra-Cerebral Hemorrhage cohort is an ongoing observational study.4 We prospectively recruited all adults admitted to the emergency department of Lille University Hospital for a stroke related to spontaneous ICH (from November 2004 to April 2009).
Seizures were defined according to the International League Against Epilepsy criteria and classified as focal or generalized. We recorded dates of seizure incidence and distinguished as ES ≤7 days and LS >7 days after stroke.5 Use of antiepileptic drugs was recorded.
The original cohort consisted of 562 patients. Patients who had a history of seizures before stroke (n=36), for whom the occurrence of previous seizures was unknown (n=4) and who died within the first 7 days of stroke (n=197), were excluded.
We prospectively collected clinical data and medical history, including vascular risk factors.4
Computed tomographic scans were performed at admission in all patients. Lesions were considered to be cortical when they involved cortical areas, even if the origin of the bleeding was in deep structures.
Concerning MRI biomarkers, brain microbleeds (BMB) were counted throughout the brain.6 BMB located in the cortex, in the gray-white matter junction and in the subcortical white-matter, were considered as lobar. We evaluated the severity of global cortical atrophy7 and leukoaraiosis.8
Patients were invited to be followed up at 6 months, then annually. At each visit, the occurrence of seizures (dates of seizures were ascertained from medical reports and interviews) and treatments were recorded.
We performed survival analysis using life tables, Kaplan–Meier statistics, and Cox models, starting at the date of ICH presentation, and censoring on the date of first LS, of death, or end of follow-up. To identify predictors of LS, we performed multivariate analyses (Cox models), including ES, cortical involvement of the ICH and ICH volume. We selected the variables based on P value ≤0.1 in bivariate analyses.
We also analyzed (Cox models) the possible predictive role of BMB defined as total number, BMB in lobar location, severity of lobar BMB (0= reference, 1–2, and ≥3).We applied 4 models as follows: each MRI biomarker was entered univariately (model 1); adjusted for age and sex (model 2); adjusted for model 2 and cortical involvement (model 3); adjusted for model 3, global cortical atrophy, and leukoaraiosis (model 4).
The influence of LS on functional outcome (mRS≥3 after 3 years of follow-up) was evaluated with Kaplan–Meier statistics.
Detailed methods are provided in the online-only Data Supplement.
The cohort consisted of 325 patients (54% males, median age 70 years, interquartile range 58–79).
During 778 person-years of follow-up (median follow-up, 2.2 years [interquartile range, 0.97–4.31]), 31 patients developed ≥1 LS, resulting in an incidence rate of 4 new cases/100 person-years (95% confidence interval [CI], 3–6). The median delay between ICH and LS was 9 months (interquartile range, 3–23).
Bivariate analyses are presented in Table 1. In multivariate analysis, the only factor independently associated with the occurrence of LS was a cortical involvement of the ICH (hazard ratio [HR], 2.8; 95% CI, 1.3–6.1; P=0.01).
ES were not predictors of LS (HR, 1.9; 95% CI, 0.9–4.3; P=0.1). Among 51 patients with ES, the risk of developing LS did not significantly differ between patients treated with long-term antiepileptic drugs and those without long-term antiepileptic drugs treatments (HR, 1.4; 95% CI, 0.4–5.6; P=0.6).
Two hundred thirty-one patients (median age, 68 years [interquartile range, 56–78]) underwent MRI (Table 2).
Cox models showed that lobar BMB were predictors of LS, after adjustment for age, sex, and cortical involvement of ICH (HR, 2.4; 95% CI, 1.0–5.4; P=0.04), whereas total BMB did not predict LS. The predictive effects of lobar BMB remained significant after adjustment for severity of global cortical atrophy and leukoaraiosis, with a strong impact of number of lobar BMB (≥3 BMB; Figure).
The occurrence of LS was associated with a worse functional outcome (mRS≥3) after 3 years of follow-up (P=0.009).
We found an incidence of LS of 4 new cases/100 person-years (95% CI, 3–6) in a cohort of ICH with a strong influence of cortical involvement. Among MRI biomarkers, the presence of lobar microbleeds (HR, 2.4; 95% CI, 1.1–5.4), as well as their number, was associated with the occurrence of LS during long-term follow-up.
The Prognosis of InTra-Cerebral Hemorrhage cohort presents a good external validity, notwithstanding potential limitations attributable to recruitment biases.4 Despite the enrollment of 562 patients with consecutive ICH, only 325 were alive on day 7, including 231 patients who underwent MRI, highlighting limitations to the sample size attributable to monocentric recruitment.
In our cohort, ES did not predict the risk of developing LS. Data in the literature are scarce and mainly come from cohorts that gathered infarcts and ICH.9 In line with our results, the largest ICH study also found that ES were not predictive of LS.10 Interestingly, patients with LS had a worse functional outcome, suggesting that LS may either have a direct influence on outcome or may simply be symptomatic of the severity of the underlying disease.
Our main finding is the association between lobar microbleeds and the occurrence of LS during long-term follow-up. The number of lobar BMB (≥3) was also important, even if there is no clear threshold like in other disease settings.11 The association between lobar BMB and the risk of LS might suggest a link with the underlying vasculopathy (cerebral amyloid angiopathy).12 Future prospective studies on larger cohorts of patients with ICH are warranted to confirm these findings.
Sources of Funding
This work was supported by Equipe d’Accueil 1046, ADRINORD (Association pour le Développement de la Recherche et de l’Innovation dans le NORD PAS DE CALAIS).
Dr Rossi was funded by the European Federation of Neurological Societies and the Société Française NeuroVasculaire. Dr Herdt was supported by scientific advisory board (UCB). Drs Rossi, Hénon, Leys, Cordonnier are investigators in Photothera (NEST3), Brainsgate (Impact-24). Dr Cordonnier received speaker fees from Boehringer Ingelheim, Bayer. Dr Leys is a member of advisory boards, steering committees, adjudication committees for Sanofi-Aventis, Servier, Boehringer Ingelheim, Allergan, Bayer, BMS, Ebewe. Honoraria were paid to Adrinord. The other authors have no conflict to report.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.000232/-/DC1.
- Received November 20, 2012.
- Accepted March 11, 2013.
- © 2013 American Heart Association, Inc.
- Cordonnier C,
- Potter GM,
- Jackson CA,
- Doubal F,
- Keir S,
- Sudlow CL,
- et al
- Henneman WJ,
- Sluimer JD,
- Cordonnier C,
- Baak MM,
- Scheltens P,
- Barkhof F,
- et al