Thomas P. Bleck

Chapter Outline

History

Although seizures have been recognized at least since Hippocratic times, their relatively high rate of occurrence in critically ill patients has only recently been recognized. Seizures as a side effect of critical care treatments (e.g., as a complication of lidocaine infusion for ventricular arrhythmias) are also a recent phenomenon.

The first recorded description of SE is by Gavasetti in 1586.3,4 Sir Thomas Willis described the complications of untreated SE in 1667:

Attempts at treating SE in the nineteenth century included bromide,⁶ morphine,⁷ and ice applications. Barbiturates were introduced in 1912, followed by the identification and use of phenytoin in 1937; these were the first rational treatments for SE.⁸ Paraldehyde gained brief prominence in the next decades.⁹ The most recent major improvement is the use of benzodiazepines, pioneered by the French in the 1960s.¹⁰

Epidemiology

Few data are available concerning the epidemiology of seizures in ICU patients. A 10-year retrospective study of all ICU patients at the Mayo Clinic reported approximately 7 patients with seizures per 1000 ICU admissions.¹¹ In a 2-year prospective study of a medical ICU, we acquired approximately 35 patients with seizures per 1000 admissions.¹ These analyses are not strictly comparable, because the patient populations and methods of detection differed. The incidence of seizures is probably higher in pediatric ICUs than in medical ICUs.^12–14

Certain ICU patients appear to be at increased risk for seizures, but the degree of that increased risk has not been quantified. Patients with renal failure or with an altered blood-brain barrier who receive imipenem-cilastatin are an obvious example, but other patients receiving this antibiotic (or γ-aminobutyric acid [GABA] antagonists such as penicillin) occasionally seize. Cefepime has emerged as a cause of nonconvulsive seizures and SE, especially in patients with renal insufficiency.¹⁵ Transplant patients, especially those receiving cyclosporine, appear to have an increased risk for convulsions. Patients who rapidly become hypo-osmolar from any cause are also at risk. Nonketotic hyperglycemia patients have a high likelihood of partial seizures; this is a rare instance of a metabolic disorder producing focal neurologic syndromes.¹⁶ Less commonly, diabetic ketoacidosis may also produce partial seizures.¹⁷

The epidemiology of SE is somewhat better understood. Estimates of the incidence of generalized convulsive SE in the United States range from 50,000 cases/year¹⁸ to 250,000 cases/year.¹⁹ Some portion of this discrepancy may be due to differences in definitions. The larger estimate comes from the only population-based data available and may be more accurate. Similarly large variations occur in mortality rate estimates, from 1% to 2% in the former study to 22% in the latter. This disagreement stems, at least in part, from a conceptual discordance: the smaller number attempts to determine mortality rate that the authors directly attribute to SE, while the larger figure reflects the overall mortality rate for SE patients, in whom death was frequently a consequence of the cause of the underlying disease rather than SE itself. In the latter study, for example, anoxia was the cause of SE in adults with the highest mortality rate. In many of the reports surveyed in the earlier review, these patients were not included.

A number of important risk factors have emerged from the Richmond study. When SE lasted longer than 1 hour, the mortality rate was 32%; when it lasted less than 1 hour, the mortality rate was only 2.7%. SE caused by anoxia was associated with a mortality rate of about 70% in adults, but the corresponding rate in children was less than 10%. After the age of 12 months, the mortality rate of SE rose with increasing age. In their study, the commonest cause of SE in adults was stroke, followed thereafter by withdrawal from anticonvulsant drug therapy; cryptogenic (or idiopathic) SE; and SE related to ethanol withdrawal, anoxia, and metabolic disorders. Systemic infection was the most commonly diagnosed cause of SE in children; this was followed by congenital abnormalities, anoxia, metabolic disorders, anticonvulsant drug withdrawal, CNS infections, and trauma. Although brain tumors seldom caused SE in children, such patients experienced a nearly 50% mortality rate.

Towne and colleagues demonstrated that 8% of an unselected series of comatose medical ICU patients had unsuspected nonconvulsive status epilepticus (NCSE).²⁰ In septic ICU patients, Oddo and colleagues showed that about 30% have periodic epileptiform activity or NCSE when recorded for 24 hours or longer.²¹

Hospital-based series of SE patients are usually subject to considerable selection bias regarding cause. The data in Table 65.1, based upon 20 years of experience in San Francisco, are of great interest because almost all patients with SE in the city of San Francisco who began to seize outside the hospital are included.^22–24

Between 6% and 12% of epilepsy patients present with SE,²⁵ and about 20% of seizure patients will experience an episode of SE within 5 years of their first seizure.¹¹

Nosology and Semiology

Numerous systems have evolved for the classification of seizures; the most frequently used today is that of the International League Against Epilepsy²⁶ (Box 65.1). This schema allows classification based primarily on clinical criteria, without inferences about cause. Although a more recent proposal for terminology has been published, it is not yet widely accepted.²⁷ It is important because of its predictive value for cause, prognosis, and treatment decisions in ICU patients. Simple partial seizures arise focally in the cerebral cortex, without taking over either the limbic system or subcortical nuclei. The patient remains aware of the environment during the ictus, and except for the seizure itself appears unchanged. Bilateral limbic system dysfunction results in a complex partial seizure; the patient’s awareness and ability to interact with the environment are diminished (but not always completely abolished). Automatisms are movements that the patient seems to make without being aware of them; typical automatisms include swallowing, masticatory movements, and fumbling with nearby items. Secondary generalization implies invasion of either the other hemisphere (with loss of consciousness) or, more commonly, subcortical structures, with the development of a generalized convulsion.

Primary generalized seizures seem to arise from the entire cerebral cortex and the diencephalon at the same time; there are no visible focal phenomena. Consciousness is lost from the start of the seizure. True absence seizures are usually confined to childhood; they consist of the abrupt onset of a blank stare usually lasting 5 to 15 seconds, without lateralizing phenomena, from which the patient abruptly returns to normal. Atypical absence is usually seen in children who have the Lennox-Gastaut syndrome. Myoclonic seizures begin with brief, bilaterally synchronous jerks without an initial change in consciousness, followed by a generalized convulsion. They occur in several of the genetic epilepsies, but in the ICU are more commonly the consequences of anoxia or metabolic disturbances.²⁸ Clonic seizures involve repetitive movements; they may be generalized (synchronous movements of all extremities and both sides of the face) or partial (e.g., one side of the face and the arm of the same side). Tonic seizures are episodes of tonic extension of the arms, legs, and trunk; they must be distinguished from decerebrate rigidity and from tetanic spasms.²⁹ Tonic-clonic seizures begin with tonic extension, followed by a brief phase of rapid vibration of the extremities, evolving into bilaterally synchronous clonus, and concluding with a postictal phase in which incontinence is common and brief apnea is occasionally noted. They may be primarily generalized or, more commonly, occur as the manifestation of spread of a partial seizure. Only those seizures that are known to involve progression through the tonic and clonic stages should be called tonic-clonic.

When seizures occur in ICU patients, clinical judgment is required to apply this system. Patients whose consciousness is already altered by drugs, hypotension, sepsis, or intracranial disease may be difficult to diagnose regarding the “simple” or “complex” nature of their partial seizures.

SE is classified by a somewhat similar system, with alterations to match the observable clinical phenomena (Box 65.2).³⁰ Again, the ability to use clinical observation without inferences about cause is important. Generalized convulsive SE (GCSE) is the type most commonly encountered in ICUs, and poses the greatest risk to the patient. GCSE may be either primarily generalized, as in the intoxicated patient, or may represent secondary generalization, as in the patient with a brain abscess who develops GCSE. Tonic SE is usually seen in children or adolescents with a history of severe CNS dysfunction. Nonconvulsive SE (NCSE) in the ICU is most commonly the consequence of partially treated GCSE. Some authors use this as a general term for any SE involving altered consciousness without convulsive movements. Although conceptually useful, this blurs the distinctions among absence SE, partially treated GCSE, and complex partial SE (CPSE), which have different causes and treatments. Epilepsia partialis continua (EPC) is a special form of partial SE in which a small area of the body makes repetitive movements, sometimes for months or years following a CNS insult.

Pathogenesis

Clinical SE has a large variety of causes. The relative frequencies of causes depend upon the definition of SE employed (e.g., if repetitive, stereotyped myoclonic activity after a cardiorespiratory arrest is considered SE, then the frequency of such arrests as a cause of SE will rise).

The reported “causes” of SE can be separated, if imperfectly, into predispositions and precipitants. Predispositions are relatively fixed conditions that increase the likelihood of SE, such as a brain tumor, in the presence of a precipitant. Precipitants, in contrast, are transient conditions that can produce SE in most, if not all, people but will tend to affect those with predispositions at lesser degrees of severity (e.g., barbiturate withdrawal).

For experimental purposes, the nosologic division of SE into partial (focal) or generalized based on the type of seizure produced works well, as does the recognition of convulsive and nonconvulsive seizure types. One must recognize that these are models of acute SE; they are not chronic conditions that occasionally produce SE, as are many of the afflictions of patients. Nevertheless, they have substantial explanatory power for understanding the neuronal and systemic processes of SE, for studying its consequences, and for predicting responses to therapy.

Pathophysiology

The causes and effects of SE at the cellular, brain, and systemic levels are interrelated, but their individual analysis is useful for understanding them and their therapeutic implications. One must first understand the consequences of a single seizure and then contrast this information with the effects of prolonged or frequent seizures merging into SE. Longer durations of SE produce more profound alterations with an increasing likelihood of permanence, and of becoming refractory to treatment. Figure 65.1 illustrates the variety of processes involved in a single seizure and in the transition to SE.³¹

The ionic events of a seizure follow the opening of ion channels coupled to excitatory amino acid (EAA) receptors. Although the endogenous ligands of these channels are glutamate and aspartate, the channels are named for synthetic compounds that potently activate them. From the standpoint of the intensivist concerned with SE, three channels are particularly important because their activation may raise intracellular free calcium to toxic concentrations. The first channels primarily conduct sodium ions (the AMPA [amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid] channels). The second are the N-methyl-D-aspartate (NMDA) channels, which admit sodium and calcium when the cell has been depolarized (which relieves the resting blockade of the ionophore by magnesium). The third, the metabotropic or ACPD (aminocyclopentane-trans-1,3-dicarboxylic acid) channels, mobilize calcium from intracellular stores via coupling to G-protein-linked second messengers.

These EAA systems are normally crucial for learning and memory. Many drugs that block these systems, such as ketamine, are too toxic to use as chronic anticonvulsants. However, the deleterious consequences of SE, and the brief period during which they would be needed, suggest that similar agents may prove to have a role in the management of SE. Counterregulatory ionic events are also triggered by the epileptiform discharge; the most important is activation of inhibitory interneurons, which feed back to the bursting cells via GABA_A synapses.

The cellular consequences of the excessive EAA channel activation include (1) accumulation of toxic concentrations of free intracellular calcium; (2) activation of autolytic enzyme systems; (3) production of oxygen-derived free radicals; (4) generation of nitric oxide, which both enhances subsequent excitation and serves as a toxin; (5) phosphorylation of several enzyme and receptor systems, making seizures likely; and (6) increasing intracellular osmolality, thereby producing neuronal swelling. If adenosine triphosphate (ATP) production should fail (because the substrate becomes inadequate or is diverted into EAA-related events), membrane ion exchange systems stop functioning, and the neuron swells further. These events are responsible for the neuronal damage associated with SE.

Many other important biophysical and biochemical alterations occur during and after SE. The intense neuronal activity activates immediate-early genes and produces heat shock proteins, providing strong indications of the deleterious effects of SE and insight into the mechanisms by which neurons protect themselves.³² Wasterlain’s group summarized the many mechanisms through which SE damages the nervous system.³³

Absence SE is an exception among these conditions. It appears to consist of rhythmically increased inhibition and does not produce clinical sequelae or neuropathologic abnormalities.

The mechanisms that terminate seizure activity are uncertain. Given the relative rarity of SE in a population in which at least 1 in 50 patients has had a seizure, one must infer that these mechanisms are generally effective. The leading candidates for seizure terminating systems are inhibitory mechanisms, primarily GABAergic neuronal aggregates. This hypothesis receives strong support from the clinical observation that human SE frequently follows withdrawal from GABA agonists (e.g., benzodiazepines).

The electrical phenomena of SE at the whole brain level, as seen in the scalp electroencephalogram (EEG), reflect the seizure type that initiates SE (Fig. 65.2). Thus, absence SE begins with a generalized 3-Hz wave-and-spike EEG pattern. During the course of SE, there will usually be some slowing of this rhythm, but the wave-and-spike characteristic persists. In contrast, GCSE goes through the sequence of EEG changes outlined in Table 65.2. The initial high-frequency discharge becomes progressively less well formed over minutes; this pattern implies that neuronal activity is less synchronous. Whether this indicates that inhibitory systems are attempting to terminate SE, a progressive decay in the ability of synaptic mechanisms to maintain synchrony, or global deterioration in neuronal function remains to be determined.

The repetitive firing that characterizes SE alters the extracellular microenvironment. The most important change is probably the elevation of the extracellular potassium concentration. Although extruding potassium is an effective strategy to maintain normal electronegativity, the excessive amounts of potassium ejected during SE overcome the ability of astrocytes to buffer it. Patients with cerebral edema, glial scarring, or alien tissue lesions have extracellular space abnormalities that impair the potassium buffering ability of glial cells. Raising extracellular potassium is a potent epileptogenic stimulus.

The tremendously increased cellular activity of SE elevates tissue demand for oxygen and glucose. To meet this demand, cerebral blood flow initially increases threefold or greater. However, after about 20 minutes, energy supplies become exhausted. This accentuates the demand for local catabolism in order to support ion pumps (in a vain attempt to restore the internal milieu during the flood of sodium and calcium). Many researchers believe that this is the major cause of epileptic brain damage in GCSE. Other forms of SE may not be subject to such severe hypercatabolism, but still pose a risk.

When partial seizures generalize, subcortical structures begin to play an active role in the clinical phenomena observed. Spread of the electrical activity into the substantia nigra and other subcortical regions appears to be necessary before a tonic-clonic convulsion occurs.

The brain contains intrinsic systems that terminate seizure activity; both local GABAergic interneurons and inhibitory thalamic neurons are involved. Whether these systems have evolved, at least in part, for protection against seizures, or whether this effect is an epiphenomenon of some other physiologic function, is unresolved.

SE can produce cerebral edema, which follows ictal damage to the blood-brain barrier.