Early Seizures in Cerebral Vein and Dural Sinus Thrombosis
Risk Factors and Role of Antiepileptics
Backgound and Purpose— The risk of seizure early after the diagnosis of cerebral vein and dural sinus thrombosis (CVT) is not known, and the use of prophylactic antiepileptic (AED) medication in the acute phase of CVT is controversial.
Methods— In a multicenter, prospective, observational study, we analyzed the risk factors for seizures experienced before the diagnosis of CVT was confirmed (presenting seizures) or within the following 2 weeks (early seizures). The risk of occurrence of early seizures was compared in 4 risk strata and related to whether patients received AEDs or not. Criteria for the strata were “presenting seizures” and “supratentorial lesions.”
Results— Two hundred forty-five of 624 (39.3%) patients with CVT experienced presenting seizures, and 43 (6.9%) patients had early seizure. In logistic-regression analysis, supratentorial lesion (odds ratio [OR]=4.05, 95% CI=2.74 to 5.95), cortical vein thrombosis (OR=2.31, 95% CI=1.44 to 3.73), sagittal sinus thrombosis (OR=2.18, 95% CI=1.50 to 3.18), and puerperal CVT (OR=2.06, 95% CI=1.19 to 3.55) were associated with presenting seizures, whereas supratentorial lesion (OR=3.09, 95% CI=1.56 to 9.62) and presenting seizures (OR=1.74, 95% CI=0.90 to 3.37) predicted early seizures. The risk of early seizures in patients with supratentorial lesions and presenting seizures was significantly lower when AED prophylaxis was used (1 with seizures in 148 patients with AEDs vs 25 in 47 patients without AEDs; OR=0.006, 95% CI=0.001 to 0.05).
Conclusions— CVT patients with supratentorial lesions had a higher risk for both presenting and early seizures, whereas patients with presenting seizures had a higher risk of recurrent seizures within 2 weeks. Our results support the prescription of AEDs in acute CVT patients with supratentorial lesions who present with seizures.
Before the diagnosis of cerebral vein and dural sinus thrombosis (CVT) is confirmed, approximately one third of CVT patients experience focal or generalized epileptic seizures.1 Some have new seizures after hospital admission.2–11 Although such seizures do not have a significant independent influence on long-term prognosis, they may be a cause of early death.5,11 Risk factors for early seizures, such as motor9,11 or sensory5 deficits, cortical vein thrombosis,11 any brain lesion,5 or an hemorrhagic brain lesion11 on admission computed tomography/magnetic resonance imaging (CT/MR), were identified in previous series. However, such series had a modest sample size, leading to uncertainty of the hazard of seizures associated with those risk factors. The risk of a seizure early after the diagnosis of CVT is also unknown. Owing to the lack of controlled information, the role of prophylactic antiepileptic (AED) medication in the acute phase of CVT is controversial.12–14
In this study, we analyzed in a large, multinational, CVT cohort the risk and predictors of seizures at presentation and the risk of seizures within 2 weeks of diagnosis in CVT patients, whether or not they were prescribed AEDs.
Patients and Methods
The organization of the International Study on Cerebral Vein and Dural Sinus Thrombosis (ISCVT) has been described in detail in a previous publication.1 In brief, ISCVT is a prospective, multinational, observational study that included 624 consecutive adult patients with symptomatic CVT who were diagnosed at 89 participating centers in 21 countries from May 1998 to May 2001. Diagnosis of CVT was confirmed by MR/MR venography, conventional or CT venography, surgery, or autopsy. No uniform investigation or treatment protocols were imposed on the participants. All treatments and dates of start/stop were recorded. A list of risk factors for CVT was attached to the inclusion form to assist investigators with the etiologic work-up (available at www.iscvt.com), and thrombophilia screening was recommended. Follow-up visits were performed at 6 and 12 months and yearly thereafter, predominantly by direct interview and observation by the local investigators.
The diagnosis of seizures and their types were established by local investigators. Our operational definitions of early seizures followed the recommendations for epidemiologic studies in epilepsy.15 For other subtypes of stroke, these guidelines define early symptomatic seizures as those occurring within 7 days of clinical onset. For CVT, they do not provide a time window for early seizures because CVT can have an acute, subacute, or chronic onset. We decided to follow the methodology of a study on the same topic in a different CVT cohort5 and included the seizures that occurred from clinical onset up to 2 weeks after confirmation of CVT diagnosis (day 0). We subdivided these seizures into (1) presenting seizures, those that occurred at or before confirmation of CVT diagnosis, and (2) early seizures, those experienced after confirmation of the diagnosis of CVT, because they feature different clinical practice scenarios.
To describe the effect of AEDs on the risk of occurrence of early seizures, we used a stratified case-control design that compared patients who were (cases)/who were not (controls) prescribed AEDs. We calculated the risk of early seizures in 4 risk strata. Criteria for the strata were the presence or absence of “presenting seizures” and the presence or absence of the major risk factor (ie, with the highest odds ratio [OR]) identified in the analysis of risk factors for early seizures. Patients were considered to be on AED prophylaxis when they were prescribed any AEDs before the occurrence of an early seizure, irrespective of type, route, or dosage.
We compared demographic, clinical, and imaging variables and risk factors for CVT in patients with and without presenting seizures and early seizures. For bivariate analysis, we used the χ2 test (with Yates’ correction when necessary) or Fisher’s exact test for dichotomous data, the Mann-Whitney U test for ordinal data, and Student’s t test for continuous data. A significant probability value was set at 0.01 to decrease the risk of an α error due to multiple comparisons. Dichotomous variables with a positive significant association in bivariate analysis were entered in a logistic-regression analysis (backward model).16 We calculated the ORs and the 95% CIs for the retained variables, the percentage of subjects correctly classified, and the area under the receiving-operator characteristic (ROC) curve for the selected models. Data were analyzed with SPSS 12.0 for Windows.
Frequency of Presenting and Early Seizures
Two hundred forty-five patients (39.3%) experienced presenting seizures before the diagnosis of CVT. Seizures were focal in 58 (9.3%) patients, generalized (from onset or secondarily) in 123 (19.7%), or of both types in 64 (10.3%). Three (0.5%) patients presented with convulsive status epilepticus. Two patients had previous epilepsy with partial complex seizures but experienced a different type of seizure, ie, motor seizures, secondarily generalized in 1 of them.
Forty-three patients (6.9%) had early seizures within 2 weeks after the diagnosis of CVT. Twenty-six (60.0%) of the 43 early seizures were recurrent, as they were experienced by patients who already had a presenting seizure. None of the 2 patients with prevalent epilepsy had early seizures. No seizures occurred within 2 weeks after diagnosis for patients discharged before this time.
Variables Associated With Presenting Seizures
Table 1 shows the variables that were significantly associated with presenting seizures in bivariate analysis. Presenting seizures were more frequent in patients with motor deficits, superior sagittal sinus and cortical vein thrombosis, any cerebral lesion on admission CT/MR, and pregnancy/puerperal CVT. In logistic-regression analysis, supratentorial lesion (OR=4.05, 95% CI=2.74 to 5.95), cortical vein thrombosis (OR=2.31, 95% CI=1.44 to 3.73), sagittal sinus thrombosis (OR=2.18, 95% CI=1.50 to 3.18), and puerperal CVT (OR=2.06, 95% CI=1.19 to 3.55) were the retained predictors of presenting seizures. This model correctly classified 70.5% of the subjects, and the area under the ROC curve was 0.75.
Variables Associated With Early Seizures
Table 2 shows the variables that were significantly associated in bivariate analysis with early seizures within 2 weeks of the diagnosis of CVT. Early seizures were more frequent in patients with presenting seizures, motor deficits, superior sagittal sinus and cortical vein thrombosis, and any cerebral lesion on admission CT/MR.
In logistic-regression analysis, supratentorial parenchymal lesion on admission CT/MR (OR=3.09, 95% CI=1.56 to 9.62) and presenting seizure (OR=1.74, 95% CI=0.90 to 3.37) predicted new early seizures. This model correctly classified 93% of the subjects, and the area under the ROC curve was 0.68. Supratentorial parenchymal lesion on admission CT/MR (OR=5.15, 95% CI=0.67 to 40) and motor deficit (OR=2.9, 95% CI=1.1 to 7.58) predicted early recurrent seizures in patients with presenting seizures (percentage correctly classified=89.3; area under the ROC curve=0.68). Supratentorial parenchymal lesion on admission CT/MR (OR=3.18, 95% CI=0.52 to 5.99), aphasia (OR=4.16, 95% CI=1.29 to 12.8), and sagittal sinus thrombosis (OR=4.08, 95% CI=0.99 to 10.2) predicted early seizures in subjects without presenting seizures (percentage correctly classified=95.5; area under the ROC curve=0.72).
Risk of Early Seizures and AED Prophylaxis
Table 3 shows the comparison of clinical and imaging features of patients who were and who were not prescribed AEDs before the occurrence of an early seizure. Prophylactic AEDs were prescribed for a median of 15.5 days, and except for 23 patients, they were prescribed until discharge. In logistic regression, AEDs were more often prescribed for patients with presenting seizures (OR=31.3, 95% CI=19 to 53), who were comatose (OR=3.8, 95% CI=1.9 to 7.8), who had parenchymal lesions on admission CT/MR (OR=3.1, 95% CI=1.8 to 5.5), who were aphasic (OR=2.5, 95% CI= 1.3 to 4.7), and who had sagittal sinus (OR=1.9, 95% CI=1.2 to 3.3) or cortical vein thrombosis (OR=1.8, 95% CI=0.9 to 3.4). This logistic-regression model predicted 87% of AED use. No country or center effect was found.
As described in the Methods section, we performed stratification, controlling for “presenting seizure” and “supratentorial parenchymal lesion on admission CT/MR,” which (as described earlier) was the major risk factor for early seizures. Therefore, patients were grouped in 4 risk strata according to whether or not they had “presenting seizure” or “supratentorial parenchymal lesion on admission CT/MR.” Comparison of baseline characteristics of patients who were/were not prescribed AEDs revealed the following statistically significant differences within each stratum: stratum “no supratentorial lesion, no presenting seizures,” lower Glasgow Coma Scale score (13.0 vs 14.7, P<0.05); stratum “no supratentorial lesion, presenting seizures,” lower Glasgow Coma Scale score (14.3 vs 14.9, P<0.05); stratum “supratentorial lesion, presenting seizures,” more patients with mental status disorder (47/54 vs 102/142, P<0.05), stratum “supratentorial lesion, no presenting seizures,” lower Glasgow Coma Scale score (13.0 vs 14.1, P<0.05), more women (32/134 vs 3/35, P<0.05), and more aphasic patients (17/58 vs 18/111, P<0.05). These differences indicate that patients who received AEDs had a somewhat more severe clinical status at presentation. However, in none of the strata could we detect differences in risk factors for early seizures between patients who were/were not prescribed AEDs.
The risk of seizures in each stratum and the effect of AEDs are shown in Table 4. The risk of early seizures in patients without supratentorial lesions or presenting seizures was very low, regardless of AED prophylaxis. Patients with supratentorial lesion, in particular those with presenting seizures, had a moderate risk of early seizures. The risk of early seizures in patients with supratentorial lesions and presenting seizures was significantly lower when AED prophylaxis was used. Patients who received prophylactic AEDs where more likely to be dead or dependent at discharge (26.0%, 60 of 231 with AEDs vs 14.8%, 58 of 393 with no AEDs; P=0.001), but no differences (15.6%, 36/231 vs 12.5%, 49/393; P=0.27) were found at final follow-up.
This work confirms that nearly 40% of patients with CVT have seizures at presentation and shows that ≈7% of CVT patients experience seizures within 2 weeks of diagnosis. The majority of these early seizures were recurrent, but 40% of them occurred de novo. Patients with CVT had a heterogeneous risk of presenting seizures and of early seizures: a higher risk for both types of seizures was found in patients with supratentorial parenchymal lesions, whereas patients with presenting seizures also had a higher risk of early seizures.
The prevention of early seizures is a relevant clinical problem. Such seizures can cause neurologic and systemic deterioration, status epilepticus,11 and death,3,11,13 although in the ISCVT cohort, seizures were not an independent predictor of death and/or dependency.1 A number of previous studies described the correlates of early seizures in CVT and the risk of later seizures2,3,5,7,9–11 (Table 5). Except for Stolz et al,3 who reported recurrent early seizures in 39% of their patients, none of the other studies addressed the topic of early seizures, ie, seizures occurring shortly after admission after confirmation of the diagnosis of CVT. The higher proportion of recurrent early seizures in the single-center study of Stolz et al3 may be due to a different definition of recurrent early seizure or to a selection referral bias toward more severe CVT cases.
Concerning risk factors for presenting seizures, Bousser and Ross Russell12 stated that seizures are more frequent when the lesion is anterior to the central sulcus and in patients who have focal deficits. In the VENOPORT study,5 presenting and early seizures were associated with sensory deficits and parenchymal lesions on admission CT/MR. In a study performed in 2 tertiary-care German centers, motor deficit, intracranial hemorrhage, and cortical vein thrombosis were independent predictors of early seizures.11 The present study is larger than all previous series and identified some clinically plausible variables with robust associations with presenting and early seizures. Parenchymal cerebral lesions, particularly when located supratentorially, may have cortical involvement with an increased risk of seizures. Occlusion of the superior sagittal sinus and cortical veins,17 which drain venous blood from the upper convexity of the brain, including the motor and sensory cortices, could also increase such risk. The risk of seizures was increased in patients with motor deficits, again pointing to the relevance of damage to the motor cortex. The association between presenting seizures and pregnancy/puerperal CVT is less clear. A very high percentage of seizures in pregnancy/puerperal CVT have been mentioned in some series,5,18,19 but not in others.20 Possible factors that raise the likelihood of seizures are the young age of pregnant/puerperal women and the concurrence of preeclampsia. Table 4 summarizes the results of the pooled analysis of grouped data from the VENOPORT,5 ISCVT,1 and the German11 studies. The pooled results confirm that focal signs (aphasia and motor deficits), the location of CVT (superior sagittal sinus and cortical vein), and parenchymal lesions, more so if supratentorial, are associated with a higher risk of presenting seizures.
The risk factors for early seizures had not been investigated previously. We found that supratentorial parenchymal lesions and presenting seizure were the most important predictors of early seizures. Conversely, the risk of early seizures in patients presenting with isolated intracranial hypertension was low (<3%). The increased risk of an early seizure in patients who had already experienced one at presentation indicates the ongoing presence of risk factors for both presenting and early seizures (eg, location of the CVT, supratentorial lesion) and the variable individual susceptibility to seizures.
The major strength of this study is the large sample size and the diversity of participating hospitals in different countries and continents, which diminish potential inclusion bias and increase the external validity of the results. There are, however, a number of potential limitations, including the lack of a central reading of images and of a uniform diagnostic and therapeutic protocol. Other limitations pertain to the study of seizures, which was a secondary objective of ISCVT. Therefore, some information of potential relevance was not requested in the inclusion form for the sake of brevity and study feasibility. In fact, we did not register familial history of seizures and we did not record electroencephalographic data.21 Probably, this additional information would not have modified our results. In ISCVT, no uniform treatment protocols were imposed on the participants. There was considerable variation among participating centers in the use of AEDs, in particular, among patients without presenting seizures.22 This variation reflects the lack of scientific evidence to support clinical decisions on this therapeutic question. Such uncertainty left a window for research, as we could compare the risk of early seizures in patients who received AEDs and in those who did not. We postulated that besides seizure at presentation, the main factor influencing AED use would be tradition and local expert opinion, and therefore, that no major imbalance in baseline characteristics would be found between patients who were treated or not with AEDs. To minimize a potential imbalance, we stratified the patients according to the 2 major risk factors for early seizures. The risk of early seizures in patients without supratentorial lesions or presenting seizures was very low, regardless of AED prophylaxis. On the other hand, AED prophylaxis significantly decreased the risk of early seizures in patients with supratentorial lesions and presenting seizures. Our study was not randomized. Because of its case-control design, this study tended to overestimate AED treatment effects. Despite stratification by major risk factors for early seizures, there was some imbalance of baseline variables between patients who were treated with AEDs and those who were not. The differences were minor, and no differences in variables known to be risk factors for early seizures were detected. Another limitation was the absence of blinding of patients and examiners to treatment. This could lead to detection bias of early seizures. However, this bias was minimized by the fact that all early seizures occurred in hospital. Our analysis should be considered pragmatic, mimicking “mainstream practice,” as we did not consider AED type, dosage, or compliance. In summary, our results support the prescription of AEDs in acute CVT patients with supratentorial lesions who present with seizures.
We wish to thank the investigators who participated in the ISCVT. Their names and centers were listed in a previous publication.1 We thank F. Masuhr and M. Busch for providing us with additional information of their sample.11
Source of Funding
This study was supported by PRAXIS grant C/SAU/10248/1998 from the “Fundação para a Ciência e Tecnologia.”
- Received March 7, 2007.
- Revision received July 11, 2007.
- Accepted August 17, 2007.
Ferro JM, Canhão P, Stam J, Bousser M-G, Barinagarrementeria F, for the ISCVT investigators. Prognosis of cerebral vein and dural sinus thrombosis: results of the International Study on Cerebral Vein and Dural Sinus Thrombosis (ISCVT). Stroke. 2004; 35: 664–670.
Dentali F, Gianni M, Crowther M, Ageno W. Natural history of cerebral vein thrombosis: a systematic review. Blood. 2006; 108: 1129–1134.
Cakmak S, Derex L, Berruyer M, Nighoghossian N, Philippeau F, Adeleine P, Hermier M, Froment JC, Trouillas P. Cerebral venous thrombosis: clinical outcome and systematic screening of prothrombotic factors. Neurology. 2003; 60: 1175–1178.
de Bruijn SFTM, de Haan RJ, Stam J, for the Cerebral Venous Sinus Thrombosis Study Group. Clinical features and prognostic factors of cerebral venous sinus thrombosis in a prospective series of 59 patients. J Neurol Neurosurg Psychiatry. 2001; 70: 105–108.
Preter M, Tzourio C, Ameri A, Bousser MG. Long-term prognosis in cerebral venous thrombosis: a follow-up of 77 patients. Stroke. 1996; 27: 243–246.
Bousser MG, Ross Russell RR. Cerebral venous thrombosis. In: Warlow CP, Van Gijn J, eds. Major Problems in Neurology, vol. 33. London: Saunders; 1997.
Einhäupl KM, Masuhr F. Cerebral venous and sinus thrombosis: an update. Eur J Neurol. 1994; 1: 109–126.
Katz MH. Multivariable Analysis: A Practical Guide for Clinicians, 2nd ed. Cambridge, England: Cambridge University Press; 2006: 74, 135–136.
Jacobs K, Moulin T, Bogousslavsky J, Woimant F, Dehaene I, Tatu L, Besson G, Assouline E, Casselman J. The stroke syndrome of cortical vein thrombosis. Neurology. 1996; 47: 376–382.
Srinivasan K. Cerebral venous and arterial thrombosis in pregnancy and puerperium: a study of 135 patients. Angiology. 1983; 34: 731–746.
Meheraein S, Ortwein H, Busch M, Weih M, Einhäupl K, Masuhr F. Risk of recurrence of cerebral venous ans sinus thrombosis during subsequent pregnancy and puerperium. J Neurol Neurosurg Psychiatry. 2003; 74: 814–816.
Cantu C, Barinagarrementeria F. Cerebral venous thrombosis associated with pregnancy and puerperium: review of 67 cases. Stroke. 1993; 24: 1880–1884.