Status epilepticus (SE) represents a major neurologic emergency, each year affecting 10 to 50/100,000 persons; in adults, its mortality rate ranges from 7 to 20% (1–7). Prompt initiation of treatment is essential, because SE becomes more refractory to treatment with time (5,7–12). SE that fails to respond to first-line therapy [clonazepam (CZP), lorazepam (LZP), or diazepam (DZP)] followed by second-line therapy [phenytoin (PHT), valproate (VPA), or phenobarbital (PB)] is defined as refractory SE (RSE) (5,11,13). RSE occurs in ∼30% of patients with SE and is associated with mortality rates >20% (5). Management of RSE requires coma induction with potent anesthetics under continuous EEG (cEEG) monitoring, implying mechanical ventilation in an intensive care unit (ICU) setting. The aim of treatment is to terminate SE and to prevent late SE-associated complications, including hyperthermia, hypotension, hypoglycemia, rhabdomyolysis, pulmonary edema, cerebral edema, and failure of cerebral autoregulation (11,14). Barbiturates [pentobarbital (PTB) and thiopental (THP)] are the anesthetics most commonly used to induce coma in this setting, especially in Europe (1,15). Propofol (PRO; 2,6-diisopropylphenol), a widely used intravenous anesthetic, has been shown to control SE in animals (16,17) and in humans (18–24) and to shorten the duration of electroconvulsive therapy (25,26). It is still debated whether PRO should be used as an alternative to barbiturates or as first-choice treatment in RSE (3,13,14,27). In the medical ICU of our institution, PRO has been used for several years to induce burst suppression on cEEG. The goal of this study was to determine the efficacy and tolerance of PRO for the management of patients with RSE.
Summary: Purpose: Refractory status epilepticus (RSE) is a critical medical condition with high mortality. Although propofol (PRO) is considered an alternative treatment to barbiturates for the management of RSE, only limited data are available. The aim of this study was to assess PRO effectiveness in patients with RSE.
Methods: We retrospectively considered all consecutive patients with RSE admitted to the medical intensive care unit (ICU) between 1997 and 2002 treated with PRO for induction of EEG-monitored burst suppression. Subjects with anoxic encephalopathy showing pathological N20 on somatosensory evoked potentials were excluded.
Results: We studied 31 RSE episodes in 27 adults (16 men, 11 women; median age, 41.5 years). All patients received PRO, and six also subsequently thiopental (THP). Clonazepam (CZP) was administered with PRO, and other antiepileptic drugs (AEDs) concomitant with PRO and THP. RSE was successfully treated with PRO in 21 (67%) episodes and with THP after PRO in three (10%). Median PRO injection rate was 4.8 mg/kg/h (range, 2.1–13), median duration of PRO treatment was 3 days (range, 1–9), and median duration of ICU stay was 7 days (range, 2–42). In 24 episodes in which the patient survived, shivering after general anesthesia was seen in 10 episodes, transient dystonia and hyperlipemia in one each, and mild neuropsychological impairment in five. The seven deaths were not directly related to PRO use.
Conclusions: PRO administered with CZP was effective in controlling most of RSE episodes, without major adverse effects. In this setting, PRO may therefore represent a valuable alternative to barbiturates. A randomized trial with these drug classes could definitively assess their respective role in RSE treatment.
We retrospectively considered all consecutive adult patients with RSE admitted to the medical ICU between 1997 and 2002 who received PRO for induction of burst suppression. Patients were identified by manual screening of the ICU records, looking for the diagnosis of “seizure,”“epilepsy,” or “status epilepticus.” We also reviewed the cEEG data to determine whether the patients had RSE. RSE was defined as SE that was clinically or electroencephalographically refractory after the administration of first- and second-line treatment within 60 min after SE onset. Subjects with clinical suspected RSE were included only either if EEG showed findings consistent with this diagnosis, or, if EEG was not available immediately (see later), the seizures pattern in a given patient was consistent with her or his previous seizures. Patients with RSE and anoxic encephalopathy showing uni- or bilaterally absent early-latency cortical somatosensory evoked potentials (N20) were excluded from our series, because these patients have a very bad prognosis (28).
Management of RSE in the ICU
On the basis of clinical and EEG criteria, RSE episodes were classified according to seizure types into generalized convulsive (GCRSE), nonconvulsive (NCRSE) (i.e., complex-partial RSE, subtle status, or electromechanical dissociation), and partial motor (PMRSE). On admission, routine laboratory investigations (complete blood cell count, chemistry, blood gas analysis, and urinalysis), electrocardiogram (ECG), chest radiograph, and cerebral computed tomography (CT) imaging were performed on all patients; in the absence of an identified etiology, brain magnetic resonance imaging (MRI) scanning was performed (before/after continuous EEG recording). Subjects with suspected CNS inflammation or infection underwent CSF analysis. After unsuccessful treatment with CZP 2 mg intravenous (IV), as a bolus, followed by a continuous IV infusion) and phenytoin (PHT, 20 mg/kg IV), patients were admitted to the medical ICU, intubated, and mechanically ventilated. Burst-suppression induction with PRO, monitored by cEEG (14 leads, according to a modified 10-20 system) was concomitantly initiated. PRO was given as a loading dose of 2 mg/kg IV, followed by continuous IV infusion at a rate that produced a burst-suppression pattern on the cEEG (interburst interval, 5–15 s). Because, in our center, cEEG initiation is available only between 8 a.m. and 6 p.m., patients fulfilling the clinical criteria for RSE admitted outside this time span were treated with PRO infusion at a rate necessary to control seizures clinically until cEEG was available; cEEG was then continued uninterruptedly. Clinical assessment and cEEG interpretation were made several times daily by certified neurologists. ICU medical and paramedical personnel were personally instructed by one of the authors to recognize a “burst-suppression” pattern. Laboratory tests were performed regularly: chemistry including creatine kinase and liver parameters was assessed daily, whereas triglyceride and AED levels were performed ≥3 times per week. A neuropsychological assessment was performed on all patients on discharge from hospital by neurologists and neuropsychologists, by using a standardized bedside battery (MMS, Stroop test, Poppelreuter's test, recall of 10 words, test of executive functions, and tests of ideomotor and constructive praxia).
Clinical and epidemiologic data
By reviewing the case records, we analyzed the etiology, precipitating factors, duration of mechanical ventilation, and duration of ICU stay. The types and amounts of administered AEDs and anaesthetics were determined. We assessed general medical and neurologic complications during treatment in all patients, neuropsychological impairment in survivors, and mortality. Seizure were considered controlled when both clinical and EEG seizure activity were permanently abolished after anesthetic drug tapering.
We found 95 episodes that met our inclusion criteria, but excluded 59 because they were not related to RSE (six clinically suspected RSEs not confirmed either by EEG or by history, of whom all survived; 26 nonrefractory SE or “simple” seizures, and 27 nonepileptic diseases) and five with anoxia showing pathological somatosensory excitatory potentials (SSEPs). The remaining 31 RSE episodes occurred in 27 adults (16 men and 11 women; median age, 41.5 years; range, 22–86 years). One man had four episodes (GCRSE) and one woman two (PMRSE) (both survived all episodes), whereas the other 25 subjects had a single RSE episode. Diagnosis of RSE was confirmed by EEG in all NCSE and PMRSE episodes, whereas four episodes in two patients with GCRSE were diagnosed only clinically (all were already known for epilepsy).
Subtypes and etiologies of RSE
We observed 18 episodes of GCRSE in 15 patients. Fourteen episodes occurred in 11 patients with a previous history of seizure, eight with symptomatic epilepsy (cavernoma in the patient with four episodes, accounting for three “clinical diagnoses”; previous stroke in four, accounting for one “clinical diagnosis”; and cerebral abscess, melanoma metastasis, or cerebral contusion, each in one), two with idiopathic generalized, and one with cryptogenic epilepsy. An acute trigger was identified in nine of these 14 episodes (noncompliance in four, electrolyte imbalance in three and cerebral hypoperfusion and alcohol intoxication, each in one). In the four patients without previous epilepsy, RSE was associated with a trigger (cerebral contusion, alcohol withdrawal, cocaine intoxication, or anoxia).
Seven episodes of NCRSE were seen in seven patients. Four of these patients had a history of epilepsy (cryptogenic, idiopathic generalized, partial complex on hippocampal sclerosis, and secondarily generalized symptomatic of oligodendroglioma), and an acute trigger was found in two (noncompliance or electrolyte imbalance, each in one). The remaining three patients had NCRSE after viral meningoencephalitis (two) or hepatic encephalopathy (one).
All six episodes of PMRSE seen in five patients were symptomatic: two were secondary to previous ischemic strokes, and four in three patients with known seizures [mitochondrial encephalopathy (MERRF) in the patient with two episodes and cerebral contusion or secondarily generalized epilepsy after stroke, each in one]. RSE was triggered by electrolyte imbalance in one patient.
Overall, the most frequent RSE etiologies were toxic–metabolic factors and previous stroke (seven episodes; six ischemic and one hemorrhagic on arteriovenous malformation). In our 31 episodes, we identified an acute trigger in 19, a remote cause in eight, and a chronic progressive mechanism in four. No episode was completely idiopathic (i.e., in each case, either a trigger or a remote etiology could be identified; Table 1).
|Toxic–metabolic||9 (2)||2 (1)|
|Stroke||7 (1)||3 (1)|
|Viral encephalitis||2 (2)||1 (1)|
|Total||19 (4)||12 (2)||7 (3)|
All but one patient received PRO therapy after unsuccessful treatment with CZP and PHT; one subject with anoxic encephalopathy and tachycardic atrial fibrillation received CZP followed by IV VPA. All patients received PRO at a loading dose of 2 mg/kg IV, followed by a median continuous perfusion rate to achieve a burst-suppression pattern or before the switch to THP of 4.8 mg/kg/h (range, 2.1–13 mg/kg/h), corresponding to a median total dose of 24 g of PRO (range, 3.6–108 g). The two patients with four episodes with a “clinical diagnosis” received PRO at 5 mg/kg/h after the loading dose. The patient with the highest total dose of PRO (108 g over a 9-day period) experienced the only episode of severe but asymptomatic hypertriglyceridemia in our series (32 mg/L, not associated with rhabdomyolysis). The median duration of PRO treatment was 3 days (range, 1–9 days); the median duration of mechanical ventilation was 5 days (range, 1–37 days), and the median duration of ICU stay was 7 days (range, 2–42 days). Other AEDs were continued during PRO infusion, especially CZP and PHT. Additional AEDs were given if the patient was already taking them on admission or were initiated to optimize treatment [VPA in 14, CBZ in 12, lamotrigine (LTG) in six, vigabatrin (VGB) in three, and levetiracetam (LEV) or tiagabine (TGB), each in two patients]. The management of AEDs in patients with RSE was modified daily on the basis of the clinical and cEEG response and blood AED monitoring. In all 31 episodes, we obtained initially a burst-suppression pattern on the cEEG, with abolition of clinical epileptic activity. After ≥12 h, PRO was tapered progressively over a 12- to 24-h period under cEEG control (depending on the baseline infusion rate, patients were given half of the last PRO dose for 6–12 h, and then a fourth for 6–12 h, and then PRO was stopped). In a group of patients, epileptiform discharges reappeared in the first few hours of tapering; the PRO infusion rate was then adapted toward the lowest effective dose.
Five patients experienced at least two treatment failures during PRO reduction, and in one, hyperlipidemia developed (see earlier). In these six subjects, THP was initiated immediately after cessation of PRO to achieve a burst-suppression pattern, after a median length of PRO treatment of 4 days (range, 3–9 days). At that point, the median cumulative PRO dose was 51 g (range, 15–108 g), and median CZP infusion rate was 4 mg/24 h (range, 2–8 mg). THP was administered by a bolus of 3 mg/kg followed by a mean infusion rate of 6 mg/kg/h (range, 2–8 mg/kg/h). CZP was stopped, but other AEDs (particularly PHT) were administered concomitant with THP.
|18 mg/L||16 mg/L||19 mg/L||17 mg/L|
|Time to RSE |
|Time to |
|Time to |
|Cumulative dose |
to RSE control (g)
|Cumulative dose |
to failure (g)
|Cumulative dose |
to death (g)
|PRO||2 (1–9) 21 episodes||4 (3–9) 6 episodes||4 (3–7) 4 episodes||20 (3–95) 21 episodes||51 (15–108) 6 episodes||25 (19–50) 4 episodes|
|THP||14 (4–20) 3 episodes||—||8 (3–22) 3 episodes||117 (42–144) 3 episodes||—||94 (35–258) 3 episodes|
During the ICU stay, antibiotics were prescribed in 22 (71%), and vasopressor drugs, in 15 episodes (48%).
Seven patients died: four treated with PRO alone (three PMRSE and one NCRSE) and three with PRO followed by THP (two GCRSE and one NCRSE). One of them died after successful treatment of RSE (PRO alone) following gastrointestinal bleeding outside the ICU, whereas the other six died in RSE. Death occurred during two (11%) of 18 episodes of GCRSE, two (28%) of seven episodes of NCRSE, and three (50%) of six episodes of PMRSE. RSE etiologies related to death were stroke in three cases and previous idiopathic generalized epilepsy complicated by hypoxia, hepatic encephalopathy, meningoencephalitis, or alcohol withdrawal combined with previous ischemic stroke, each in one case (Tables 1 and 4).
|GCRSE||16 (1)||2 (2)||11%|
|NCRSE||5 (1)||2 (1)||28%|
|Total||24 (3)||7 (3)||22%|
All four episodes diagnosed only clinically were successfully treated with PRO only and survived: after starting cEEG, three did not need any increase in PRO infusion rate, whereas one patient had incomplete electrographic SE control. She improved after adapting the PRO dose.
Considering the 24 episodes from which the patient recovered (21 patients), no sequelae were seen after 16 (67%). Mild neuropsychological impairment (mainly mental slowing and anterograde memory impairment) was seen in five episodes (GCRSE or PMRSE, two each; one NCRSE), critical care polyneuropathy in two episodes (both NCRSE), and Lance–Adams postanoxic encephalopathy in one episode (GCRSE). RSE was successfully treated with PRO in 21 patients and with THP after PRO in three (one each with GCRSE, NCRSE, or PMRSE).
Transient treatment complications related to PRO were shivering at the end of general anesthesia (10 episodes), reversible hyperlipidemia (one episode; see earlier), and dystonia (one episode).
Our series showed that PRO was effective in the management of most of 31 RSE episodes occurring in 27 adults. Despite relatively high total doses of PRO, the incidence of treatment-related adverse events was quite low. In particular, no patient experienced life-threatening adverse effects or so-called propofol infusion syndrome. The seven deaths were attributable to RSE itself (or its etiology) in six patients and to a fatal gastrointestinal bleeding outside the ICU in the remaining subject.
One possible limitation of this study relates to the diagnosis of RSE, which was performed clinically and electroencephalographically in all but four episodes. The latter had only a clinical diagnosis before PRO treatment. We tried, however, to avoid an overestimation of the occurrence of RSE in patients without EEG-confirmed seizures, as we included only subjects with clinical patterns consistent with previously EEG-confirmed seizures in the same patient. One episode was confirmed electroencephalographically as the EEG became available a few hours after beginning of PRO infusion.
The tapering rate of PRO may appear rapid in our series. However, all treatment failures under PRO appeared at the beginning of the tapering, and patients responding to PRO did not show any seizure recurrence until hospital discharge despite this schedule. These considerations suggest that PRO may be safely tapered within 24 h if patients do not show epileptiform activity at the beginning of the dose reduction.
PRO has been used since the early 1980s (29) for the induction and maintenance of anesthesia in the operation room and sedation in the ICU. Metabolized in the liver, it is highly lipid soluble and has very short distribution and elimination half-lives (2–4 min and 30–60 min, respectively) and little propensity to accumulate (13,21). It possibly modulates GABA-α receptors at a site different from that targeted by benzodiazepines (BZDs) and barbiturates (16,17,30). Although PRO is currently considered as an alternative to THP or midazolam (MDZ) for the management of patients with RSE (9,11,14,31), only limited clinical data are available. In a retrospective review of 28 patients receiving PRO for RSE, it was found to be ineffective in only four (14%) (20). Another RSE series found no difference in terms of duration of ICU stay or the incidence of arterial hypotension between patients receiving PRO and those receiving barbiturates (eight patients each), whereas time to seizure control was significantly shorter in the PRO group (2.6 vs. 123 min) (21). A retrospective comparison of PRO and MDZ in RSE (14 and six patients, respectively) did not find any difference in terms of seizure control (clinical and EEG), infectious complications, hemodynamic compromise, duration of mechanical ventilation, and mortality (23). A recently published systematic review of 28 series including 193 RSE patients [54 taking MDZ, 33 taking PRO, and 106 taking pentobarbital (PTB)] (24) showed that PTB was most effective for controlling seizures, followed by PRO, and then MDZ; however, more patients receiving PTB had cEEG monitoring, introducing a possible bias. Furthermore, this study showed less hypotension with PRO, suggesting a subtherapeutic dosage of PRO. Finally, recently PRO was reported to control complex-partial RSE (22).
In our series, 24 (77%) of 31 RSE episodes were successfully treated, allowing a permanent seizure control: 21 (67%) with PRO alone and three (10%) with PRO and subsequent THP therapy. PRO was administered at relatively high dosage (mean, 4.8 mg/kg/h) for several days (mean, 3 days).
To set PRO infusion at the lowest effective dose, we concomitantly prescribed BZDs (in particular, CZP) (32). This possibly induced a better outcome as compared with the administration of PRO alone. It would be interesting to analyze prospectively the outcome of PRO alone versus PRO with BZDs. However, the PRO dose would probably be higher if this drug would be administered alone (32), thus theoretically exposing patients to treatment complications related to PRO.
Seven patients died, corresponding to an overall mortality of 23%, which is identical to that found in another retrospective RSE study (5). Three of the six patients receiving PRO and THP died, suggesting that they had a more refractory SE. The large variation in previously reported mortality rates, ranging from 17 to 80%, can be explained by discrepancies in the definition, management, and etiology of RSE (5,21,33,34).
The present study showed that all episodes of RSE were symptomatic, with an acute precipitating factor in 61% and a remote etiology in the remaining cases (Table 1). Our data confirmed that most of the deaths (71%, five of seven) were related to toxic–metabolic and stroke RSE etiologies (1,2,35). Mortality seemed higher for NCRSE (28%) than for GCRSE (11%). It is known that the former may follow untreated GCRSE or even arise de novo, and is especially difficult to treat (7,8). Our NCRSE mortality rate is within the range of those previously described (18–57%) (35,36), this variation probably being explained by differences in the inclusion criteria for patients with anoxic encephalopathy. We observed a high rate of PMRSE episodes (six of 31), all symptomatic of strokes, cerebral contusions, or MERFF, with an important mortality rate of 50%.
The optimal EEG pattern for the management of patients with RSE is still debated (26,33). Our series showed that burst-suppression induction with an interburst interval of 5–15 s allowed the successful long-term control of RSE in 78% of our episodes.
The median duration of mechanical ventilation and of ICU stay in our series were shorter than previously reported in series of RSE treated with barbiturates (33,34); in particular, the minimal duration of treatment was 1 day in our study as compared with 4 days in a recently published THP series (33). However, discrepancies in patient selection between series may account for these differences. The relatively short duration of mechanical ventilation may be related to the pharmacokinetic properties of PRO, especially its short elimination half-life (13,20,21,27).
As regards complications related to RSE treatment, we observed antibiotic prescription in 71% of episodes, a figure similar to that seen in recent studies on RSE (33). Whereas infectious complications are, to some extent, inherent to ventilator treatment, they may be enhanced by barbiturates possibly in relation to an inhibitory effect on chemotaxis and airway ciliary clearance described in vitro (37,38).
In contrast to a lower incidence of hypotension reported by others (24), we found hypotension requiring vasopressors in 48% of our episodes. This is similar to the data found in two series of patients with RSE given barbiturates (40%, 65%) (5,33).
The so-called PRO infusion syndrome, which is characterized by hyperlipidemia, rhabdomyolysis, and lactic acidosis, has mainly been described in children (39). Some fatal cases also have been reported in adults, all occurring in brain-injured patients (40,41). To our knowledge, no fatal case of PRO infusion syndrome has been described in patients with RSE. Although some of our patients had high PRO infusion rates for a prolonged time (i.e., ≤9 days), isolated hyperlipidemia was seen in only one subject (cumulative dose, 108 g).
Eleven of our patients had transient movement disorders, 10 showing tremor on emergence from general anesthesia, and one, a transient focal dystonia of the upper limb. Whereas some case reports underlined the efficacy of PRO in the management of RSE (18,19), others described abnormal movements, posturing, and seizure-like activity related to its use (42–50). However, no EEG data were available in most of these cases. Conversely, other studies prospectively investigating motor phenomena related to PRO by using EEG concluded that these were nearly always nonepileptic (51–54). A recent systematic review including 81 patients with PRO-related “seizures” highlighted the underuse of EEG (performed in one third of patients only, and showing epileptic activity in two subjects) and the low involvement of neurologists (consulted in 17% of cases only) (55). It is thus likely that the aforementioned “seizures” might have been confused with abnormal movements, including opisthotonus, increased tone with twitching and rhythmic movements (29,55,56), which are probably caused by subcortical dopaminergic excitatory activity induced by a low dose of PRO. These movements disappear as cortical GABA inhibition occurs at higher doses (51). PRO might also act as a glycine antagonist in subcortical and spinal structures (47,56,57), explaining the occurrence of opisthotonus. The subcortical action of PRO is suggested by its antipruritic and antiemetic properties (52). Similar abnormal movements can be encountered with many other anesthetics (58–61).
In conclusion, in our series, PRO administered concomitantly with CZP and PHT was effective in controlling most of RSE episodes by inducing burst suppression on cEEG. PRO was well tolerated, and PRO infusion syndrome was not seen. This seems to make relative recent concerns about possible increased mortality of PRO-treated patients with SE (62). Furthermore, our data suggest that the use of PRO may result in a shorter duration of treatment. Thus our series, which, to our knowledge, is the largest to date, confirmed that PRO may constitute an alternative therapy to barbiturates for the management of patients with RSE. However, a randomized controlled study is required to compare directly the effectiveness of barbiturates and PRO in the treatment of RSE.