Chapter 3: Status Epilepticus

The clinical presentation is most consistent with secondary generalized convulsive status epilepticus (GCSE). For practical purposes, most practitioners would categorize any seizure lasting at least 5 minutes or two or more discrete seizures without recovery of consciousness in between seizures as status epilepticus (SE).^1-6 This cutoff is based on animal and human data suggesting irreversible neuronal injury⁷ and pharmacoresistance^2,8,9 may occur after prolonged seizures and the observation that most clinical and electrographic seizures that do not progress to SE last less than 5 minutes.^1,10-12 The clinical presentation of GCSE is typically relatively characteristic; however, postanoxic myoclonus (see detailed discussion below), posturing during herniation, and psychogenic seizures may be considered in the differential diagnosis.

GCSE is a relatively common neurologic emergency with an overall estimated incidence of 41 to 61 cases per 100,000 patients per year.¹³ Rapid diagnosis is crucial since untreated SE rapidly becomes refractory to therapy and carries significant morbidity and mortality rates. GCSE is defined as generalized convulsions that are associated with rhythmic jerking of the extremities.¹⁴ Classic clinical features include generalized tonic-clonic movements or rhythmic jerking of the extremities and mental status impairment (coma, lethargy, confusion); additional neurologic findings may include aphasia, amnesia, staring, automatisms, blinking, facial twitching, agitation, nystagmus, eye deviation, and perseveration. After convulsions have ceased, focal findings such as focal motor impairment, also known as Todd paralysis, may persist.

A number of different classification schemes have been proposed to categorize SE into subtypes based on clinical and electrographic characteristics. For practical purposes, we will focus on convulsive and nonconvulsive SE (NCSE) since these are the most important to recognize for the emergency and intensive critical care (ICU) physician. Clinical features and differentiating characteristics of NCSE will be discussed in detail below.

The most important principle in treatment of GCSE is initiating treatment as early as possible.^15,16 In one study, if therapy was begun within 30 minutes of onset, 80% responded to the first-line antiepileptic drug (AED), whereas only 40% responded if the therapy started beyond the 2-hour window.¹⁵ Following this principle, studies were undertaken to study if initiation of SE treatment in the prehospital setting could be safely done and would lead to a better outcome.⁶ Patients treated with lorazepam by EMS prior to reaching the hospital had better acute seizure control than those who received diazepam or placebo. Importantly, these practitioners found that respiratory compromise was more commonly seen in patients who received placebo than those who were given benzodiazepines, suggesting that administering AEDs as soon as possible is not only more efficacious but also safer than waiting. A number of studies investigated alternate routes of administrating benzodiazepines including diazepam per rectum and intranasal, buccal, or intramuscular midazolam. All of these routes of administration were shown to be effective alternatives to IV administration, and they should be considered in situations where the lack of IV access would delay initiation of benzodiazepine therapy.

Treatment of these patients should be guided by an institutional protocol to allow multiple team members to simultaneously initiate a series of steps including an assessment of airway, breathing, and circulation (ABCs), administration of AEDs, determination and addressing of underlying causes for seizures, and obtaining IV access. Initiation of treatment should begin within 3 to 5 minutes of seizure onset, depending on the circumstance. Most patients with GCSE will not be able to safely protect their airway because of ongoing seizures or the seizure treatment. Therefore, in most cases of GCSE, intubation should be initiated early. Good hemodynamic monitoring is mandatory since SE as well as an antiepileptic used to treat SE have been associated with arrhythmias and hypotension. Diagnose hypoglycemia rapidly since it will only respond to glucose administration and will result in permanent damage if not corrected rapidly (Table 3-1).

In accordance with evidence from prospective randomized controlled trials, lorazepam (see Table 3-2 for dosing and pharmacokinetic information) should be given as the initial AED, with a success rate between 59% and 65%.¹⁴ If IV access cannot rapidly be established, diazepam 20 mg per rectum (may use diastat or IV solution of diazepam), or midazolam 10 mg administered intranasally, buccally, or intramuscularly^6,14,17 may be chosen as alternatives.

All patients with GCSE should be given a second AED since the initial benzodiazepine is not a long-term therapy to prevent recurrence of SE. Most physicians would not give first- and second-line AEDs in a sequential order. Although second-line therapy has not been prospectively evaluated, phenytoin or fosphenytoin (see Table 3-2) is recommended by most neurologists.¹⁸ However, recent evidence has emerged from three small prospective, randomized, open-label trials to suggest that IV valproate is an efficacious and safe first- or second-line treatment (66% and 88% response rate, respectively) and perhaps superior to phenytoin (42% and 86%, respectively).¹⁹ Other alternatives with even less data include levetiracetam and phenobarbital.

Management should focus on three aspects: (1) making sure that the patient is medically stable, (2) diagnosing and addressing the underlying cause of GCSE, and (3) determining if electrographic seizures are present. As a side note, avoid administration of paralytics for intubation whenever possible since they may mask ongoing convulsions.

1. Medical stability: Hypotension and arrhythmias may be seen during loading with phenytoin or fosphenytoin. Monitoring may include placing a central line or arterial line if not done already.

2. Diagnostic workup: The underlying differential is wide (Tables 3-3 and 3-4) and the diagnostic workup should be individualized depending on the clinical scenario (Table 3-5). In a general population-based study, the most common causes of SE were low levels of AEDs (in 34% of the cases), remote symptomatic etiologies (history of neurologic insults remote to first unprovoked seizures), and cerebrovascular disease.²⁰ Nevertheless, the causes of SE in critically ill patients may be different from those in the general population. In general ICUs, metabolic abnormalities and drug withdrawal represent 66% of the SE admissions.²¹ In comatose patients admitted to general ICUs, anoxia-hypoxia followed by stroke and infection are the most common cause of NCSE and refractory status epilepticus (RSE).^22,23 Some of the diagnostic tests should be started in the ED and some may be completed once the patient is admitted to the floor or ICU. Most physicians would at least send basic blood and urine tests (including complete blood cell count [CBC], basic metabolic panel [BMP], calcium, magnesium, toxicology screens), get AED levels (ie, phenytoin, valproate, carbamazepine), and have a high suspicion to obtain imaging (head CT or magnetic resonance imaging [MRI]) and lumbar puncture. The presence of fever at presentation should raise the suspicion for central nervous system (CNS) infection, and empiric treatment with bacterial and viral coverage should be started until the results from the lumbar puncture and imaging ancillary testing are available.

3. Persistent electrographic seizures: The patient should be connected to computer-analyzed electroencephalography (cEEG) monitoring as soon as possible since electrographic seizures persist in 20% to 48% of clinically apparently successfully treated GCSE and 14% are in NCSE without any clinical signs of seizure activity.^14,24 These electrographic seizures in the aftermath of convulsive seizures are clearly associated with a worse prognosis,²⁴ while no study so far investigated if treatment of these patterns results in a better outcome. Currently, most practitioners would treat these electrographic seizures the same way they would treat clinical seizures.

Our patient has refractory status epilepticus (RSE), a neurologic emergency that should be managed in the ICU setting and requires cEEG monitoring. While some controversy regarding the exact definition persists, most experts classify SE that persists despite first- and second-line AED regardless of the elapsed time as RSE. Most commonly patients with RSE appear comatose and have subtle or no clinical manifestations of seizures.²⁵ Prior to the widespread use of cEEG, the incidence of RSE was estimated at 2000 to 6000 cases per year. It can occur at any age and both sexes are equally at risk.²⁶ Overall, the exact incidence and prevalence is readily underestimated because of the absence of population-based studies with cEEG monitoring. For instance, in a VA cooperative study, 38% of patients with "overt" SE and 82% with "subtle" SE continued seizing after receiving full doses of two AEDs.²⁴ Moreover, depending on sampling bias and utilization of cEEG monitoring after GCSE, 9% to 48% of patients continued to seize after initial therapy.^24,25

At hospital discharge, 9% to 21%^{6, 27, 28} and at 30 days 19% to 27%^14,29 of adults with convulsive SE are dead. Among survivors, disability and particularly cognitive impairment are frequent.^{6,15, 30, 31} Factors associated with poor outcome include older age, impairment of consciousness, duration of seizures, and the presence of medical complications.^{27, 32, 33} Recurrent SE may be seen in up to one-third of patients according to one population-based study that followed patients for 10 years.³⁴

RSE carries a dismal outcome with mortality rates close to 23% to 61%,^28,35-41 and it appears to be rather independent of the chosen therapeutic strategy.

Exact mechanisms underlying the development of refractoriness are to date incompletely understood. Evidence has accumulated that impairment of gamma-aminobutyric acid (GABA)–mediated inhibition related to internalization of GABA receptors^42,43 and upregulation of excitatory AMPA (alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionate) and NMDA (N-methyl-d-aspartate) receptors⁴³ may play a role in the development of increasing refractoriness to treatment.

There are no randomized controlled trials that have investigated this question, and practitioners do not agree on the best treatment for RSE. After standard treatment with two AEDs, the likelihood of responding to a third conventional medication regardless of the agent is only 2% to 5%, depending on the type of SE.¹⁴ Despite these findings, half of the neurologists would still use a conventional AED as a third-line agent,¹⁸ while most intensivists would choose continuous drips of midazolam, propofol, or pentobarbital (Table 3-6). If valproic acid has not been chosen already as a second-line agent, particularly in patients with a "do not intubate" order, it may be a good alternative. All patients should be connected to cEEG at this point if it is available.

Traditionally, algorithms advocated for loading with phenobarbital with a success rate ranging between 2% and 58% depending on the previous AEDs, the length of SE, and cause of SE. Subsequently, a transition to pentobarbital followed if patients continued to seize.⁴⁴ Once started, pentobarbital infusion lowers cerebral oxygen demand, intracranial pressure (ICP), and lipid peroxidation, thereby stopping seizures almost invariably. However, barbiturates are heavily sedating agents with a long half-life, and many patients have complications including refractory acidosis due to propylene glycol toxicity and subsequent multiorgan failure.

Recently, a number of studies have reported the effectiveness of continuous drips of propofol or midazolam as alternative treatments for RSE. In a systematic review of published cases and small series, pentobarbital was found to be more effective in terminating seizures (acute treatment failure, breakthrough seizures, and posttreatment seizures) than midazolam or propofol but was also associated with more frequent side effects (ie, hypotension requiring pressors).⁴⁵ However, these results are to be interpreted with caution for a number of reasons: publication bias in small case series, most of the pentobarbital cases were reported a decade before those treated with the other two agents, cEEG monitoring was not available for the pentobarbital cases, no comparisons between medications, differences in underlying etiologies of SE, and the fact that some centers now use more than 10 times the doses for midazolam than those used in the reported case series.

Although some preliminary evidence in a small retrospective series have shown that propofol is associated with a higher mortality rate than midazolam,⁴¹ a recent small single-center study suggested a decrease in mortality (22%) and morbidity rates (two-thirds of the patients had no sequelae) in a large series of 27 patients using the combination of propofol and benzodiazepine infusions to achieve burst suppression and subsequent maintenance on IV clonazepam.⁴⁶ Moreover, practitioners have recommended higher loading doses of midazolam infusions which may be more efficacious but are also associated with more side effects⁴⁶ (see Table 3-8). Interestingly, there was absolutely no difference in mortality rate (approximately 50%) when comparing the three agents.⁴¹ It is conceivable that when a similar titration end point (ie, EEG end point) would be set for the three agents that the difference in side-effect and efficacy profiles would disappear.

All of these treatment decisions are controversial since there are little or no data to base them on. The most recent European Federation of Neurological Societies (EFNS) guidelines recommend general anesthetic IV doses of sedative or barbiturates with the goal of burst suppression on EEG to be maintained for a minimum of 24 hours while conventional AEDs level are optimized.⁴⁷ A small study of 49 patients with RSE treated with propofol or pentobarbital drips (with or without additional mid-azolam) concluded that the outcome was independent of the choice of agent and the EEG titration goal. In the literature, cessation of nonconvulsive seizures,^48-53 achieving diffuse beta activity,^49,52,54 burst suppression,^55-57 and complete suppression of EEG^18,56 have been advocated. Most experts would recommend continuing IV AEDs for at least 24 to 72 hours after cessation of electrographic status to prevent recurrence of seizures.

Yes. If the initial diagnostic tests do not identify the underlying cause of SE, a more comprehensive workup for less common causes of status needs to be initiated (Table 3-7).

Valproate produces inhibition of cytochrome P450 2C9 (CYP2C9), which leads to inhibition of the clearance of phenytoin.⁵⁸ In addition, valproate also displaces the drug from its protein-binding sites, thus increasing both the free fraction and the total amount of phenytoin present. The complex interaction between these two AEDs requires close monitoring, and following total levels as well as free phenytoin levels may be reasonable.⁵⁹

There is a long list of medications used for RSE that have been described in small series or randomized trials (Table 3-8).

Alternative Treatment Options in RSE

Levetiracetam IV was initially approved in 2006 by the Food and Drug Administration (FDA) for patients with epilepsy who could not take oral medication. In a recent retrospective study, IV levetiracetam (mean loading dose of 994 mg over 30 minutes and maintenance dose of 2166 mg/day) was successfully used to treat 16 of 18 episodes of focal SE that were refractory to initial benzodiazepine trial.⁶⁰ In a second study of 24 critically ill patients with mainly focal-onset seizures and status, up to 82% responded to mean IV loading dose of 1780 mg +/– 649 mg bid with the only side effect of transient thrombocytopenia (4% of the patients).⁶¹

Lacosamide, a new AED with both enteral and parenteral forms, was approved by the FDA in October 2008 as adjunctive therapy for partial-onset seizures in adults suffering from epilepsy. Successful use of lacosamide 200-mg bolus over 5 minutes has been reported in only one case of left-hemispheric NCSE.⁶²

In the absence of ileus, enteric topiramate has been used successfully in dosages of 300 to 1600 mg/day to abort RSE to prevent the breakthrough and withdrawal seizures while tapering the continuous IV (cIV) infusions. Other adjunctive oral medications that we and others have used, especially while tapering off the drip, include oxcarbazepine, felbamate, pregabalin, and carbamazepine.

Ketamine is a NMDA-antagonist that has demonstrated positive results in abolishing self-perpetuating SE in animal models.⁶³ Moreover, it seems to have a synergistic effect when combined with benzodiazepines.⁶⁴ Although little experience exists in the adult population, the dosing has been extrapolated from the anesthesia literature (loading dose 1 to 2 mg/kg IV over 1 minute; maintenance 0.6 to 1.8 mg/kg per hour cIV). It is thought to possess other potential neuroprotective effects. Unlike most of the other agents used for RSE, it also increases blood pressure. Caution should be warranted in patients with elevated ICP, traumatic brain injury (TBI), ocular injuries, hypertension, chronic congestive heart failure (CHF), myocardial infarction (MI), tachyarrhythmias, and history of alcohol abuse.⁶⁵

Few studies exist on using inhaled anesthetic including isoflurane, halothane, or desflurane. Although it seems to be highly efficacious in terminating RSE, hypotension, logistical issues in most ICU settings, and frequent seizure recurrence upon withdrawal make it a less attractive choice.

Hypothermia: Very little experience exists with this treatment option.^66,67 A very small case series of four patients has suggested that induced hypothermia with a target temperature 31°C to 35°C may have a potential antiepileptic effect in terminating SE. In all four patients, the treating physicians were able to stop the cIV midazolam, and two of them remained seizure-free. A potentially beneficial side effect of hypothermia is its neuroprotective properties. Disadvantages include shivering, electrolyte abnormalities, immunosuppression, and potential coagulopathy.

Immunomodulatory agents such as steroid IV, immunoglobulins, plasmapheresis, or adrenonocorticotropic hormone (ACTH) may be helpful; in selected cases of immunologic syndromes such as Rasmussen encephalitis, anti–voltage-gated potassium channel–related limbic encephalitis, acute disseminated encephalomyelitis, and paraneoplastic disorders have occurred. However, the role of these agents outside of identified immune processes is unclear.

Lidocaine (1.5- to 2.0-mg/kg bolus, maintenance dose 3 to 4 mg/kg per hour) was efficacious in terminating seizures in 75% of patients with refractory status after the initial bolus. However, narrow pharmacologic range and neurotoxic side effects (>5 μg/mL) limit its use.⁶⁸

Some practitioners have also recommended using pyridoxine hydrochloride early on in patients with SE in an IV or enteral form at 100 to 300 mg/day, as it is a cofactor in the synthesis of the inhibitory neurotransmitter GABA, which may play a role in the initial phase of SE.

Surgery: Small series from the pediatric literature support that if a single, identifiable focus exists on EEG and imaging (single-photon emission CT [SPECT] or positron emission tomography [PET]), surgical removal of a focal structural lesion may be an option after intracranial mapping.⁶⁹

There are no prospective studies to base these recommendations on. Most investigators agree that cIV AEDs should be continued for a minimum of 24 hours after control of seizure has been achieved, with some advocating for 48 hours and others 72 hours; few recommend continuing as long as 96 hours.⁷⁰ If prior tapers have failed, it might be necessary to treat longer and taper more slowly. A small retrospective study of 40 patients found no obvious difference in seizure control or survival comparing less versus more than 96 hours of pentobarbital infusions despite no obvious differences in underlying etiologies.

Figure 3-2

The EEG in Figure 3-2 shows generalized periodic epileptiform discharges (GPEDs), which have been more recently labeled as generalized periodic discharges.⁷¹ A range of periodic epileptiform discharges (PEDs) that do not meet formal seizure criteria are seen frequently in the aftermath of convulsive or nonconvulsive SE.^14,24 PEDs are characterized by a spike, sharp wave, or sharply contoured slow wave that recurs at a rather regular interval (ie, every second). Discharges may be unilateral or generalized, and if generalized may be synchronous or independently asynchronous.

Patients may have PEDs with superimposed fast activity or other superimposed patterns or features.⁷¹ While these still do not meet formal seizure criteria, many investigators would classify them on an ictal-interictal continuum, and the new classification guidelines identity these patterns with a "+."⁷¹ These patterns may also be seen in acute brain injury without prior SE. One study investigated electrographic features of GPEDs in patients with and without SE. While there were some distinguishing features in the SE group (longer GPED duration and higher GPED amplitude), no generalizable conclusions regarding etiology, treatment, and prognosis could be made based on EEG.⁷² In our practice, we consider treatment of PEDs if they have a frequency of two per second or more or are on the ictal-interictal continuum as described above. Supplemental information may be useful in some of these patients to support more aggressive antiepileptic therapy: (1) the clinical scenario and presumed underlying diagnoses, (2) benzodiazepine treatment trial (brief trial with small doses and sometimes a 1-day suppression [Table 3-9]), (3) serial neuron-specific enolase (including before and after the 1-day suppression trial), (4) results of invasive monitoring if available (elevated lactate-pyruvate ratio specifically from decreasing pyruvate levels, elevated glycerol and glutamate, and possibly a decrease in glucose), (5) brain SPECT (for focal increased blood flow at the site of EEG discharges), (6) MRI with diffusion-weighted imaging (DWI), especially at the foci of EEG discharges, and (7) spectroscopy (for increased lactate) (Table 3-10).⁷³ Of note, no study investigated if any of these supplemental tests is accurate in identifying patients that later have definitive seizures or if treatment decisions based on these tests impact outcome.

The cEEG demonstrated NCSE (Figure 3-3), which is more common than previously recognized, particularly in the ICU setting. The exact prevalence of nonconvulsive seizures and NCSE is unknown because of a lack of population-based studies. In patients admitted to NeuroICUs, 18% to 34% and 10% have been reported to have nonconvulsive seizures (NCSz) and NCSE, respectively.^74-76 These frequencies are to be interpreted with caution since selection bias may be substantial as long as only a select subpopulation of patients undergoes the monitoring. In medical ICU settings, the incidence of NCSE is estimated to be near 10%¹⁶ and is particularly prevalent in those with sepsis.⁷⁷ One prospective study conducted in the ER setting found NCSz in 37% of 198 patients that underwent urgent EEG for altered mental status.⁷⁸ The vast majority of seizures seen in the acute brain injury setting are electrographic and have no or minimal clinical signs (see Table 3-11). The large spectrum of clinical presentations for NCSz and NCSE can easily lead to misdiagnosis and delayed treatment (Tables 3-11 and 3-12).

Indications for cEEG include (1) recent clinical seizure or SE without return to baseline for more than 10 minutes, (2) coma, (3) epileptiform activity or periodic discharges on initial EEG, (4) acute brain injury with high likelihood of seizures, and (5) suspected nonconvulsive seizures and altered mental status.

An imaging study such as a CT scan would be a good first step.

As outlined above, unexplained alteration in behavior or mental status warrants an EEG as well as a consideration for NCSE. NCSz have been associated with coma, young age, epilepsy in the past medical history or remote risk factors for seizures, convulsive seizures prior to monitoring, periodic discharges (such as periodic lateral epileptiform discharges [PLEDs] or GPEDs), or burst-suppression, oculomotor abnormalities (ie, nystagmus, hippus, or eye deviation), cardiac or respiratory arrest, and sepsis.^76,77,79 NCSE and periodic EEG patterns are independently associated with poor outcome for patients with acute ischemic stroke, aneurysmal subarachnoid hemorrhage (SAH), central nervous system (CNS) infections, TBI, and nontraumatic ICH after controlling for other predictors of poor outcome.^80,81 Although some practitioners argued that NCSE can be a surrogate marker for severity of brain injury, there is a large body of accumulating animal and human data supporting that these EEG patterns may cause additional harm.^14,50-52,81 Interestingly, recently unilateral NCSz were associated with long-term hippocampal atrophy diagnosed on follow-up MRI.⁵³ There remains some controversy about how aggressively to treat NCSE, but currently most practitioners would agree that NCSE in the acute brain injury setting should be treated very similar to GCSE (Table 3-13).

Consider obtaining an EEG if the clinical examination is out of proportion to the imaging findings. As outlined in detail above, seizures are frequent in a number of acute brain injuries. These seizures are often slower frequency and have many patterns that do not clearly fulfill classic seizure criteria when compared to patients with epilepsy. Additionally, a number of poorly understood EEG findings such as periodic epileptiform discharges or stimulus-induced rhythmic, periodic, or ictal discharges (SIRPIDs) can be seen in the critically ill. These will be discussed in some detail in Chapter 13.

The EEG is consistent with SE and the clinical syndrome is called myoclonic SE (MSE). It is most frequently seen after cardiac arrest. In animal models of hypoxic-ischemic encephalopathy following cardiac arrest, isoelectric EEG is often followed by burst-suppression patterns, which may then evolve depending on the severity of hypoxia and the degree of neurologic recovery.⁸² A distinction should be made between MSE and status myoclonicus, with only the former being epileptiform in nature. Both present with myoclonic jerks of the extremities but only MSE is associated with epileptiform electrical discharges on the EEG. Status myoclonicus can be seen in the setting of anoxic brain injury but also in any type of severe encephalopathy such as acute renal failure. The underlying pathophysiologic mechanism is related to subcortical white matter injury, specifically the corticospinal tract. EEG findings of status myoclonicus are slow waves or burst-suppression patterns but no epileptiform discharges.^83,84 Symptomatic treatments for this phenomenon include benzodiazepine, valproic acid, and levetiracetam. MSE is an epileptiform phenomenon and can be seen in the setting of epilepsy syndromes, but in the acute brain injury setting is most frequently related to hypoxia.⁸⁵ The EEG typically shows generalized polyspike and wave complexes on a nearly flat background. Patients with this EEG pattern are treated similarly to other patients with RSE, but the overall prognosis should be kept in mind.

MSE on EEG in humans has traditionally been associated with a very poor prognosis.⁸⁶ However, most investigators agree that in the era of induced hypothermia for the treatment of patients with cardiac arrest,^87,88 this ominous fate is less established. Although a recent study confirmed the unfavorable prognosis associated with certain EEG features such as very low-voltage activity, burst-suppression, or generalized epileptiform patterns, these investigators did demonstrate that in the era of temperature modulation these EEG features as well as some clinical features previously associated with an invariably poor prognosis should be interpreted with caution.⁸⁹

The presentation is most typical for a subtype of focal motor SE sometimes seen in the ICU called epilepsia partialis continua (ECP). This phenomenon can be very persistent, with epileptic focal jerking activity lasting days, weeks, or even decades.⁹⁰ Characteristically, a focal structural brain lesion can be identified including heterotopias, infectious lesions, vascular abnormalities, subdural hematomas, or neoplasms. Although long-term prognosis generally depends on the prognosis of the underlying lesion,⁹¹ ECP may result in long-term morbidity due to weakness, sensory deficits, language dysfunction, or cognitive deficits.⁹² AEDs, including benzodiazepines, may prevent secondary generalization but usually do not stop the seizure activity. Occasionally, it might resolve by itself without treatment, and surgery can be at times curative for intractable cases.⁹³ Of note, nonketotic hyperglycemia may be associated with EPC in patients with a concomitant focal cerebral lesion.⁹⁴ These patients may respond best to conventional AEDs in concert with correction of the metabolic disorder.