A “Malignant” Variant of Status Epilepticus

Background  Status epilepticus (SE) frequently does not respond to common first-line anticonvulsants. In a substantial portion of patients, administration of anticonvulsant anesthetics is inevitable. Even this aggressive approach fails to terminate SE in an undefined number of cases. We have coined the term malignant SE for this most severe variant of SE.

Objective  To assess frequency, risk factors, and in-hospital outcome of malignant SE.

Design  Retrospective cohort study.

Setting  Neurologic intensive care unit of a large university hospital.

Patients  Sample of 35 episodes of SE not responding to first-line anticonvulsants in 34 patients.

Main Outcome Measures  Predictive and prognostic features of episodes of malignant SE with persistent epileptic activity after high-dose anesthetics compared with features of the remainder of cases with refractory SE and persistent epileptic activity after failure of first-line anticonvulsants.

Results  Status epilepticus that could not be controlled by first-line anticonvulsants resulted in malignant SE in 20% of cases. Patients with malignant SE were significantly younger than patients with refractory SE (P = .03). Encephalitis was identified as an independent risk factor for malignant SE (P = .008). Outcome in malignant SE was poor, with significantly longer duration of seizure activity (P<.001), longer stay in the neurologic intensive care unit (P<.001) and in the hospital (P = .007), and more patients with functional dependency at discharge from the hospital (P = .04).

Conclusions  Malignant SE is not rare after failure of first-line anticonvulsants. The patient at risk is typically young and suffers from encephalitis. Such patients should be treated aggressively early in the course of SE to prevent malignant SE.

Status epilepticus (SE) may occur with various degrees of severity. In many cases, the condition can be terminated by first-line anticonvulsants, but about 30% to 50% of cases are refractory.^1,2 There are no standardized treatment guidelines for refractory SE (RSE). In current practice, anticonvulsant anesthetics are applied after failure of first-line drugs in almost all patients with generalized convulsive SE and in most patients with complex partial SE.³ However, even anesthetics may fail to terminate SE,^4-12 a condition for which we suggest the term malignant SE (MSE). Malignant SE cannot be terminated by any substance and continues for weeks or months. In some of the reported cases, the condition has been suggested to represent a distinct disease entity, but a cause could not be determined.¹¹

While RSE has been characterized as to frequency, risk factors, and outcome,² similar data on MSE are lacking. In the current study, we compared predictive, therapeutic, and prognostic features of MSE with those of the remainder of cases of SE not responding to first-line anticonvulsants. The identification of patients at risk may help to prevent the development of MSE by the use of aggressive therapeutic interventions early in the course of SE.

Malignant SE was defined as persistent clinical and/or electrophysiologic epileptic activity immediately recurring within 5 days after tapering of the maximal dose of intravenous anesthetic anticonvulsants required to achieve an electroencephalographic (EEG) burst suppression pattern previously. This time gap was chosen to ascertain that the suppressive effect of high-dose anesthetics on the EEG, which may last hours to days,^13,14 had ceased.

Refractory SE was defined as all forms of continuous clinical and/or electrophysiologic epileptic activity not responding to first-line anticonvulsants regardless of the delay from seizure onset, and excluding patients with MSE. First-line anticonvulsant drugs had to be applied in adequate form and dose. A treatment regimen was considered adequate if it included intravenous administration of at least 10 mg of diazepam, 1 mg of clonazepam, 6 mg of lorazepam, or 5 mg of midazolam hydrochloride followed by phenytoin in a minimum dose of 10 mg/kg.² Status epilepticus not treated with phenytoin was considered refractory with continuing epileptic activity after a double dose of benzodiazepines.

In comatose patients with no or only subtle motor phenomena, SE was defined by the presence of repetitive generalized or focal epileptiform discharges (spikes, sharp waves, and spike waves) excluding periodic lateralized epileptiform discharges, generalized periodic epileptiform discharges, and stimulus-induced rhythmic, periodic, or ictal discharges. Termination of SE was defined in patients treated with anesthetics by cessation of seizure activity and absence of a burst suppression pattern or flat line on the EEG. Patients who did not receive anesthetics were regarded as successfully treated if seizure activity ceased clinically. Two or more unprovoked epileptic seizures that had occurred more than 4 weeks before the onset of SE defined preexisting epilepsy.

Encephalitis was defined as encephalopathy (depressed or altered level of consciousness lasting 24 hours or longer, lethargy, or change in personality) and 1 or more of the following symptoms: fever, focal neurologic findings, cerebrospinal fluid (CSF) pleocytosis, and EEG or neuroimaging findings consistent with encephalitis,¹⁵ after exclusion of systemic infections. Encephalitis was infectious, noninfectious, or of unknown etiology.

All episodes of SE not responding to first-line anticonvulsants in patients aged 18 years or older admitted between January 1, 1993, and December 31, 2002, to the Neurological Intensive Care Unit, Charité–Universitätsmedizin Berlin, Germany, were analyzed retrospectively. The study had the approval of the institutional review board. In a first step, the computer-assisted patient files were searched by using the keywords status epilepticus, seizure clustering, and prolonged epileptic seizures. By these criteria, 138 episodes in 129 patients suitable for a diagnosis of SE were identified. We then excluded all episodes of SE that responded to adequate application of first-line anticonvulsants. We also excluded all episodes of nonepileptic origin (psychogenic seizure, prolonged convulsive syncope, transient ischemic attack, etc) that had initially been regarded as SE. The records of 5 patients with 6 episodes could not be retrieved in the archives. Thus, 35 episodes in 34 patients fulfilled our definition of SE refractory to first-line anticonvulsants and were included in the present analysis.

We used a structured data collection grid completed by 2 independent reviewers to analyze the clinical variables. For each episode of SE, patients’ demographic data (age, sex) and history of epilepsy were documented. Furthermore, data concerning cause, symptoms, and treatment course were evaluated. Finally, duration of SE, treatment duration in the neurologic intensive care unit, length of in-hospital stay, in-hospital mortality, and functional outcome at hospital discharge using the Glasgow Outcome Scale¹⁶ were analyzed.

Data were collected with the database program Access 2000 (Microsoft, Redmond, Wash). Statistical calculations were performed with SPSS version 11.0 software (SPSS Inc, Chicago, Ill). Frequency distributions of predictive, therapeutic, and prognostic features were compared between patients with MSE and RSE to identify characteristics of MSE, and were calculated by the χ² test. For analysis of continuous data with normal distribution, the unpaired, 2-tailed t test was used, and for data with nonnormal distribution, the Mann-Whitney test was used. Predictors associated with MSE (P<.20) were entered into a backward stepwise logistic regression analysis to identify independent risk factors. Differences were considered significant if P<.05.

A total of 35 episodes in 34 patients fulfilled the diagnostic criteria for SE not responding to first-line anticonvulsants. The mean ± SD age was 52.1 ± 18.4 years (range, 18-88 years). Seven episodes (20%) were compatible with our definition of MSE. Six patients (86%) with MSE were female, but sex distribution was not significantly different from that of patients with RSE (16 [57%] female; P = .17). The mean ± SD age of patients with MSE (38.7 ± 13 years) was significantly lower than that of patients with RSE (55.4 ± 18.2 years; P = .03).

Only 1 of the 7 patients with MSE had preexisting epilepsy, compared with 8 (29%) of those with RSE. Encephalitis was the only single cause significantly more frequently seen in MSE (5 [71%]) than RSE (3 [11%]; P = .003). All 8 patients with encephalitis (5 with MSE, 3 with RSE) had had fever and symptoms of encephalopathy for at least 24 hours before the onset of seizure activity. In 6 of the patients, CSF pleocytosis (median white blood cell count, 52/μL; range, 16-256/μL) with predominantly mononuclear cells was found. Oligoclonal bands were found in the CSF of 3 patients, and in 2 of the 8 patients a causative agent was proven (herpes simplex virus and cytomegalovirus). In a multivariate analysis, only encephalitis was an independent risk factor for MSE (odds ratio, 31.5; 95% confidence interval, 2.5-396; P = .008). Further individual causes are listed in the Table. The most common clinical presentation was complex partial SE in both groups, found in 5 (71%) of MSE episodes and 12 (43%) of RSE episodes. No symptomatology of SE was seen significantly more often in 1 of the 2 study groups. Focal onset of seizure activity occurred in 5 (71%) of cases of MSE and thus had an incidence not significantly different from that in RSE (22 [79%]).

Titration of anesthetic anticonvulsants to a burst suppression pattern was performed by definition in all cases of MSE. An anesthetic was administered in 10 cases of RSE (36%) (P = .003), and in 4 (40%) of those cases the anesthetic was titrated to a burst suppression pattern (P = .02) (Figure). Duration of anesthetic therapy was significantly longer in MSE (median, 15 days) than RSE (median, 3 days; P = .03). Anesthetics of first choice in MSE were thiopental sodium (n = 5), midazolam (n = 1), and propofol (n = 1). Anesthetics of first choice in RSE were propofol (n = 6), midazolam (n = 2), and thiopental (n = 2). Propofol was used significantly more often in RSE than MSE (P = .04).

Seizure duration was significantly longer in MSE (median, 17 days) than RSE (median, 2 days; P<.001). The median length of stay in the neurologic intensive care unit was significantly longer in MSE (53 days) than RSE (10 days; P<.001). Median in-hospital stay was also significantly longer in MSE (100 days) than RSE (25 days; P = .007). In-hospital mortality was similar in MSE (1 [14%]) and RSE (5 [18%]). One patient with MSE and 2 patients with RSE died during persistent seizures. The other patients died of medical complications. Five of 6 surviving patients with MSE and 4 of 23 surviving patients with RSE were discharged from the hospital with a reduction of 2 points or more on the Glasgow Outcome Scale compared with admission values, indicating marked functional dependency (P = .04).

In the present study, the frequency of MSE amounted to 20% of all patients with SE refractory to first-line anticonvulsants. In a systematic review including almost 200 patients, the efficacy of RSE treatment with anesthetic anticonvulsants was assessed. Epileptic activity recurred between 60 minutes and 6 hours after the initial loading dose in 8% to 20% of cases and after 6 hours of intravenous treatment in 12% to 51% of cases, depending on the drug used.¹⁷ However, these figures indicate treatment failure with anesthetics and do not represent the frequency of malignant or other difficult-to-treat forms of SE. Thus, our study provides the first data, to our knowledge, about the frequency of MSE in a large series of patients with SE not responding to first-line anticonvulsants.

In this study, encephalitis was identified as an independent predictor of MSE. An infectious agent was identified in a quarter of our patients, which is in line with studies on encephalitis reporting specific microbiological agents only in a minority of cases.¹⁵ Therefore, encephalitis frequently has to be diagnosed on the basis of clinical, CSF, EEG, and magnetic resonance imaging findings¹⁵; in the present study, symptoms of encephalopathy and fever before seizure onset, as well as mononuclear CSF pleocytosis, were regarded as sufficient.

Previous studies found encephalitis to be a primary cause of SE refractory to anesthetic anticonvulsants as well,^5,12 but in most reported cases the cause remained unknown.^4,7-9,11 Van Lierde et al¹¹ reported a series of 6 young patients with refractory multifocal febrile SE, most of whom required long-lasting general anesthesia. A cause could not be established in a single case, and the existence of a distinct RSE syndrome was discussed. Interestingly, mild mononuclear CSF pleocytosis has been described in many patients but has been attributed to SE rather than to CNS infection.^4,8,11 Some studies have focused on SE-induced pleocytosis, describing increased white blood cell counts in the CSF up to 71/μL in 10% to 20% of patients with SE lacking concomitant infectious diseases.^18,19 In a clinical landmark study by Aminoff and Simon,¹⁹ a polymorphonuclear CSF pleocytosis was reported in the vast majority of SE cases. However, the typical CSF feature of viral encephalitis is mononuclear pleocytosis, as seen in the current patients and in most previously described patients with malignant or otherwise-termed forms of long-lasting SE.^4,8,11 Thus, at least some of the reported cases of SE refractory to anesthetics with unknown cause may have been caused by encephalitis. The present findings indicate that a malignant course of SE depends on the underlying cause. This argues against the hypothesis that MSE or similar forms of difficult-to-treat SE display a distinct clinical entity.

In the present study, some cases were regarded to be refractory to first-line anticonvulsants after administration of a double dose of benzodiazepines without subsequent phenytoin.²⁰ The rationale of this approach is based on the preliminary finding that the addition of second- and third-line medications rarely controls seizures.²¹ Treatment failure with anesthetics did not depend on the drug used, but barbiturates, midazolam, and propofol share similar mechanisms of action, since all exert γ-aminobutyric acid (GABA)–ergic properties. In continuing SE, a progressive erosion of efficacy of GABAergic drugs is well known in patients²¹ and animal models,²² and the administration of non-GABAergic drugs such as ketamine hydrochloride has been encouraged in RSE.²³ Although experimental studies on ketamine in prolonged SE have been promising,²⁴ clinical evidence is anecdotal.²⁵ Further studies focusing on alternatives to or endorsement of GABAergic drugs are urgently needed in RSE and MSE.

The small number of patients, the retrospective design, and the lack of long-term prognostic data do not allow firm conclusions about the possible impact of MSE on clinical outcome. However, it should be noted that more patients in the MSE group than in the RSE group were severely disabled at hospital discharge. Thus, MSE not only prolongs the in-hospital stay but also is associated with adverse effects on the clinical outcome.

In summary, the present study indicates that the typical patient at risk to develop a malignant course of SE is young and has encephalitis. Early aggressive treatment strategies seem to be reasonable in these patients. To further investigate predictors, clinical features, and outcome of patients with MSE, the establishment of a prospective multicenter database appears to be appropriate.

Correspondence: Martin Holtkamp, MD, Department of Neurology, Charité–Universitätsmedizin Berlin, Schumannstr 20/21, 10117 Berlin, Germany (martin.holtkamp@charite.de).

Accepted for Publication: December 7, 2004.

Author Contributions:Study concept and design: Holtkamp, Othman, Buchheim, and Meierkord. Acquisition of data: Holtkamp, Othman, Buchheim, and Schielke. Analysis and interpretation of data: Holtkamp, Othman, Buchheim, Masuhr, and Schielke. Drafting of the manuscript: Holtkamp, Othman, Buchheim, Masuhr, and Meierkord. Critical revision of the manuscript for important intellectual content: Holtkamp, Othman, Buchheim, Schielke, and Meierkord. Statistical analysis: Holtkamp, Othman, and Buchheim. Administrative, technical, and material support: Holtkamp and Meierkord. Study supervision: Masuhr and Meierkord.

Acknowledgment: We thank Brigitte Wegner, MD, Institute of Medical Biometry, Charité–Universitätsmedizin Berlin, for statistical advice.