50: Seizures and Epilepsy

• Classification of Seizures and the Epilepsies Is Important for Pathogenesis and Treatment

  • Seizures Are Temporary Disruptions of Brain Function

  • Epilepsy Is the Chronic Condition of Recurrent Seizures

• The Electroencephalogram Represents the Collective Behavior of Cortical Neurons

• Focal Seizures Originate Within a Small Group of Neurons Known as a Seizure Focus

  • Neurons in a Seizure Focus Have Characteristic Activity

  • The Breakdown of Surround Inhibition Leads to Synchronization

  • The Spread of Focal Seizures Involves Normal Cortical Circuitry

• Primary Generalized Seizures Are Driven by Thalamocortical Circuits

• Locating the Seizure Focus Is Critical to the Surgical Treatment of Epilepsy

• Prolonged Seizures Can Cause Brain Damage

  • Repeated Convulsive Seizures Are a Medical Emergency

  • Excitotoxicity Underlies Seizure-Related Brain Damage

• The Factors Leading to Development of Epilepsy Are an Unfolding Mystery

  • Among the Genetic Causes of Epilepsy Are Ion Channel Mutations

  • Epilepsies Involving Focal Seizures May Be a Maladaptive Response to Injury

• An Overall View

Until quite recently the function and organization of the human cerebral cortex—the structure of the brain concerned with perceptual, motor, and cognitive functions—has eluded both clinicians and neuroscientists. In the past the analysis of brain function relied in large part on observations of the behavioral consequences of brain damage caused by strokes or trauma. These natural experiments provided much of the early evidence that distinct brain regions serve specific functions (see Chapter 1).

Observation of patients with seizures and epilepsy has been equally important in the study of brain function because the behavioral consequences of these disorders vary with the brain regions from which they originate. Seizures are temporary disruptions of brain function resulting from abnormal, excessive neuronal activity; epilepsy is a chronic condition of repeated seizures. For centuries understanding the neurological origins of seizures was confounded by the dramatic, sometimes bizarre behaviors associated with seizures. Epilepsy was widely associated with possession by evil spirits, while seizures were thought to reflect oracular, prescient, or special creative powers.

The Greeks in the time of Hippocrates (circa 400 bc) were aware that head injuries to one side of the brain could cause seizure activity on the opposite side of the body. In those earlier times the diagnosis of epilepsy was probably much broader than the contemporary definition. Other causes of episodic unconsciousness, such as syncope as well as mass hysteria and psychogenic seizures, were almost certainly classified with epilepsy. Moreover, historical writings typically describe generalized convulsive seizures involving both cerebral hemispheres. Thus it is likely that focal seizures involving a limited area of the brain were misdiagnosed or never diagnosed at all. Even today it can be difficult for physicians to distinguish between episodic loss of consciousness and the various types of seizures. Nevertheless, as our ability to treat and even cure epilepsy continues to improve, these diagnostic distinctions take on increasing significance.

The modern neurobiological analysis of epilepsy began with John Hughlings Jackson's work in London in the 1860s. Jackson realized that seizures need not involve loss of consciousness but could be associated with localized symptoms such as the jerking of an arm. His observation was the first formal recognition of what we now call focal (or partial) seizures. Jackson also observed patients whose seizures began with focal neurological symptoms and progressed to convulsions with loss of consciousness (the so-called Jacksonian march).

Another early development that presaged modern therapy was the first surgical treatment for epilepsy in 1886 by Victor Horsley. Horsley resected cerebral cortex adjacent to a depressed skull fracture and cured a patient with focal motor seizures. The modern surgical treatment for epilepsy dates to the work of Wilder Penfield and Herbert Jasper in Montreal in the early 1950s. Medical innovations include the first use of phenobarbital as an anticonvulsant in 1912 by Alfred Hauptmann, the development of electroencephalography by Hans Berger in 1929, and the discovery of the anticonvulsant properties of phenytoin (Dilantin) by Houston Merritt and Tracey Putnam in 1937.

As in any chronic disease, the physiological features of seizures are not the only consideration in the care and management of patients with epilepsy. Psychosocial factors are also extremely important. The diagnosis of epilepsy has consequences that can affect all aspects of everyday life, including educational opportunities, driving, and employment. Although many limitations imposed on epileptics are appropriate—most would agree that patients with epilepsy should not be commercial pilots—a diagnosis of epilepsy can have inappropriate, negative effects on educational opportunities and employment. To improve this situation, physicians have a duty to educate themselves and the public.

Not all seizures are the same. Thus the pathology of seizures must take into account their clinical features. Seizures, and the chronic condition of repetitive seizures (epilepsy), are common. Based on epidemiological studies in the United States, somewhere around 3% of all individuals living to the age of 80 years are diagnosed with epilepsy. The highest incidence occurs in young children and the elderly.

In many respects seizures represent a prototypic neurological disease in that the symptoms include both positive and negative sensory or motor manifestations. Examples of positive signs that can occur during a seizure include the perception of flashing lights or the jerking of an arm. Negative signs such as impairment of consciousness and self-awareness or even transient blindness or paralysis reflect impairment of normal brain function. These examples underscore a general feature of seizures: The signs and symptoms depend on the location and extent of brain regions that are affected. Finally, the manifestations of seizures result in part from the activity in normal tissue with normal cellular and network properties. The latter is particularly important in the spread of a focal or partial seizure beyond its original boundaries. Seizures quite literally hijack the normal functions of the brain.

Seizures Are Temporary Disruptions of Brain Function

Seizures can be classified conceptually into two categories: focal (or partial) and generalized (Table 50–1). Although the details of seizure classification are under continuous discussion, this simple dichotomy has proven extremely useful to clinicians because anticonvulsant medications often target preferentially to one or the other type of seizure. The terms "focal" and "partial" are used interchangeably in this chapter.

Focal seizures originate in a small group of neurons (the seizure focus) and thus the symptoms depend on the location of the focus within the brain. Focal seizures were formerly classified as simple partial when there is no alteration of consciousness, or complex partial when there is an alteration of consciousness. A typical focal seizure might begin with jerking in the right hand and progress to clonic movements (ie, jerks) of the entire right arm. This seizure could also be called a focal motor seizure. If a focal seizure progresses further the patient may lose consciousness, fall to the ground, rigidly extend all extremities (tonic phase), then have jerking in all extremities (clonic phase). In this case, the focal seizure has secondarily generalized.

The onset of a focal seizure is often preceded by symptoms called auras. Common auras include abnormal sensations such as a sense of fear, a rising feeling in the abdomen, or even a specific odor. The aura is caused by electrical activity originating from the seizure focus and thus represents the earliest manifestations of a focal seizure. The time after a seizure but before the patient returns to his or her normal level of neurological function is called the postictal period.

Generalized seizures constitute the second main category. They begin without an aura or focal seizure and involve both hemispheres from the onset. Thus they are sometimes called primary generalized seizures to avoid confusion with seizures that begin from a focus and then generalize secondarily. Primary generalized seizures can be further divided into convulsive or nonconvulsive types depending on whether the seizure is associated with tonic or clonic movements.

The prototypic nonconvulsive generalized seizure is the typical absence seizure in children (formerly called petit mal). These seizures begin abruptly, usually last less than 10 seconds, are associated with cessation of all motor activity, and result in loss of consciousness but not loss of posture. The patients appear as if in a trance, but the episodes are so brief that their occurrence can be missed by a casual observer. Unlike a focal seizure there is no aura before the seizure or confusion after the seizure (the postictal period). Patients may exhibit mild motor manifestations such as eye blinking but do not fall or have tonic-clonic movements. Typical absence seizures have very distinctive electrical characteristics on the electroencephalogram (EEG).

Other types of generalized seizures can involve only abnormal movements (myoclonic, clonic, or tonic) or a sudden loss of motor tone (atonic). The most common convulsive generalized seizure is the tonic-clonic (formerly called grand mal seizure). Such seizures begin abruptly, often with a grunt or cry, as tonic contraction of the diaphragm and thorax forces expiration. During the tonic phase the patient may fall to the ground in a rigid posture with clenched jaw, lose bladder or bowel control, and become blue (cyanotic). The tonic phase typically lasts 30 seconds before evolving into clonic jerking of the extremities lasting 1 to 2 minutes. This active phase is followed by a postictal phase during which the patient is sleepy and may complain of a headache and muscle soreness.

A primary generalized tonic-clonic seizure can be difficult to distinguish on purely clinical grounds from a secondarily generalized tonic-clonic seizure with a brief aura. This distinction is not simply academic; it can be vital to choosing proper treatment as well as pinpointing the underlying cause.

Epilepsy Is the Chronic Condition of Recurrent Seizures

Recurrent seizures constitute the minimal criterion for the diagnosis of epilepsy. The oft-quoted clinical rule emphasizes this point: "A single seizure does not epilepsy make." Various factors that contribute to a clinical pattern of recurrent seizures—the underlying etiology of the seizures, the age of onset, or family history—are ignored in the seizure classification scheme in Table 50–1. The classification of the epilepsies initially evolved primarily based upon clinical observation rather than a precise cellular, molecular, or genetic understanding of the disorder. The factors influencing seizure type and severity were sometimes recognized as patterns of symptoms, referred to as the epilepsy syndromes.

In the 1989, according to the International League Against Epilepsy (ILAE) classification of the epilepsies, the primary variables were whether or not a focal brain abnormality could be identified (localization-related versus generalized epilepsies), and whether or not a cause could be identified (symptomatic versus idiopathic). Most adult-onset epilepsies could be classified as symptomatic localization-related epilepsy, a category that included such causes as trauma, stroke, tumors, and infections. However, many patients have adult-onset epilepsies without a clearly defined cause. Thus, despite the usefulness of this scheme, many types of epilepsy did not fit neatly into its categories. In 2010 the ILAE recommended focusing on the underlying causes of epilepsies based on three broad variables—genetic, structural/metabolic, and unknown—as well as a list of electroclinical syndromes. One expects and hopes that these classifications will be elaborated as knowledge of the underlying causes increases.

Because neurons are excitable cells, it should not be surprising that seizures result either directly or indirectly from a change in the excitability of single neurons or groups of neurons. This view dominated early experimental studies of seizures. Electrical recordings of brain activity can be made with intracellular or extracellular electrodes. Extracellular electrodes sense action potentials in nearby neurons and can detect the synchronized activity of ensembles of cells called field potentials.

At the slow time resolution of extracellular recording (hundreds of milliseconds to seconds) field potentials can appear as single transient changes called spikes. These spikes reflect action potentials in many neurons and should not be confused with spikes in single neurons, which are individual action potentials that last only 1 or 2 milliseconds. The EEG thus represents a set of field potentials as recorded by multiple electrodes on the surface of the scalp (Figure 50–1).

Because the electrical activity originates in neurons in the underlying brain tissue, the waveform recorded by the surface electrode depends on the orientation and distance of the electrical source with respect to the electrode. The EEG signal is inevitably distorted by the filtering and attenuation caused by intervening layers of tissue and bone that act in the same way as resistors and capacitors in an electric circuit. Thus the amplitude of EEG signals (measured in microvolts) is much smaller than the voltage changes in a single neuron (measured in millivolts). High-frequency activity in single cells, such as action potentials, is filtered out by the EEG signal, which primarily reflects slower voltage changes across the cell membrane, such as synaptic potentials.

Although the EEG signal is a measure of the extracellular current caused by the summated electrical activity of many neurons, not all cells contribute equally to the EEG. The surface EEG reflects predominantly the activity of cortical neurons in close proximity to the EEG electrode. Thus deep structures such as the hippocampus, thalamus, or brain stem do not contribute directly to the surface EEG. The contributions of individual nerve cells to the EEG are discussed in Box 50–1.

The surface EEG shows patterns of activity— characterized by the frequency and amplitude of the electrical activity—that correlate with various stages of sleep and wakefulness (see Chapter 51), and with some pathophysiological processes such as seizures. The normal human EEG shows activity over the range of 1 to 30 Hz with amplitudes in the range of 20 to 100 μV. The observed frequencies have been divided into several groups: alpha (8–13 Hz), beta (13–30 Hz), delta (0.5–4 Hz), and theta (4–7 Hz).

Alpha waves of moderate amplitude are typical of relaxed wakefulness and are most prominent over parietal and occipital sites. During intense mental activity, beta waves of lower amplitude are more prominent in frontal areas and over other regions. Alerting relaxed subjects by asking them to open their eyes results in so-called desynchronization of the EEG with a reduction in alpha activity and an increase in beta activity (Figure 50–1B). Theta and delta waves are normal during drowsiness and early slow-wave sleep; if present during wakefulness, these waves are a sign of brain dysfunction.

As neuronal ensembles become synchronized, as when a subject relaxes or becomes drowsy, the summated currents become larger and can be seen as abrupt changes from the baseline activity. Such paroxysmal activity can be normal; for example, episodes of high-amplitude activity (1–2 s, 7–15 Hz) occur during sleep (sleep spindles). However, a sharp wave or EEG spike can also provide a clue to the location of a seizure focus in a patient with epilepsy (Figure 50–4).

Despite the variety of clinically defined seizures, important insights into the generation of seizure activity can largely be understood by comparing the electrographic patterns of focal and primary generalized seizures. The defining feature of focal (and secondarily generalized) seizures is that the abnormal electrical activity originates from a seizure focus. The seizure focus is nothing more than a small group of neurons, perhaps 1,000 or so, which have enhanced excitability and the ability to occasionally spread that activity to neighboring regions and thereby cause a seizure.

The enhanced excitability (epileptiform activity) may result from many different factors such as altered cellular properties or altered synaptic connections caused by a local scar, blood clot, or tumor. A discrete focus in the primary motor cortex may cause twitching of a finger or jerking of a limb (sometimes called simple partial seizure), whereas a seizure focus in the limbic system is frequently associated with unusual behaviors or an alteration of consciousness (sometimes called complex partial seizure).

The development of a focal seizure can be arbitrarily divided into four phases: (1) the interictal period between seizures followed by (2) synchronization of activity within the seizure focus (Figure 50–4), (3) seizure spread, and finally (4) secondary generalization. Phases 2 to 4 represent the ictal phase of the seizure. Different factors contribute to each phase.

Much of our knowledge about the electrical events during seizures comes from studies of animal models of focal seizures. A seizure is induced in an animal by focal electrical stimulation or by acute injection of a convulsant agent. This approach and the development of in vitro brain slice preparations (Box 50–2) have provided a good understanding of electrical events within the focus during a seizure as well as during the interictal period

Neurons in a Seizure Focus Have Characteristic Activity

How does electrical activity in a single neuron or group of neurons lead to a seizure? Each neuron within a seizure focus has a stereotypic and synchronized electrical response called the paroxysmal depolarizing shift, an intracellular depolarization that is sudden, large (20–40 mV), and long-lasting (50–200 ms), and triggers a train of action potentials at its peak (Figure 50–7B). The paroxysmal depolarizing shift is followed by an afterhyperpolarization.

The paroxysmal depolarizing shift and afterhyperpolarization are shaped by the intrinsic membrane properties of the neuron (eg, voltage-gated Na⁺, K⁺, and Ca²⁺ channels) and by synaptic inputs from excitatory and inhibitory neurons (primarily glutaminergic and GABAergic, respectively). The depolarizing phase results primarily from activation of AMPA- and NMDA-type glutamate receptor-channels (Figure 50–8A), as well as voltage-gated Na⁺ and Ca²⁺ channels. The NMDA-type receptor-channels are particularly suited to enhancing excitability because depolarization relieves Mg²⁺ blockage of the channel. Once unblocked, excitatory current through the channel increases, thus enhancing the depolarization and allowing extra Ca²⁺ to enter the neuron (see Chapter 10).

The normal response of a cortical pyramidal neuron to excitatory input consists of an excitatory postsynaptic potential (EPSP) followed by an inhibitory postsynaptic potential (IPSP) (because of the basic circuitry shown in Figure 50–8B). Thus the paroxysmal depolarizing shift can be viewed as a massive enhancement of the normal depolarizing and hyperpolarizing synaptic components. The afterhyperpolarization is generated by several types of K⁺ channels as well as a GABA receptor-mediated Cl^– conductance (ionotropic GABA_A receptors) and K⁺ conductance (metabotropic GABA_B receptors) (Figure 50–8A). The Ca²⁺ entry through voltage-dependent Ca²⁺ channels and NMDA-type receptor-channels triggers the opening of calcium-activated channels, particularly K⁺ channels. The afterhyperpolarization limits the duration of the paroxysmal depolarizing shift; its gradual disappearance is an important factor in the onset of a focal seizure, as discussed later.

Thus it is not surprising that many convulsants act by enhancing excitation or blocking inhibition. Conversely, anticonvulsants act by blocking excitation or enhancing inhibition. For example, the benzodiazepines diazepam (Valium) and lorazepam (Ativan) enhance GABA_A -mediated inhibition and are used in the emergency treatment of prolonged repetitive seizures. The commonly used anticonvulsants phenytoin (Dilantin) and carbamazepine (Tegretol) cause reduction in the opening of the voltage-gated Na⁺ channels that underlie the action potential. The ability of these drugs to block the Na⁺ channels is enhanced by repetitive activity associated with seizures.

The Breakdown of Surround Inhibition Leads to Synchronization

As long as the abnormal electrical activity is restricted to a small group of neurons, there are no clinical manifestations. The synchronization of neurons in the focus is dependent not only on the intrinsic properties of each individual cell but also on the connections between neurons. During the interictal period the abnormal activity is confined to the seizure focus by inhibitory effects of the excited region on surrounding tissue. This inhibitory surround is particularly dependent on feed-forward and feedback inhibition by GABA-ergic inhibitory interneurons (Figure 50–9A).

During the development of a focal seizure the inhibitory surround is overcome and the afterhyperpolarization in the neurons of the original focus gradually disappears. As a result, the seizure begins to spread beyond the original focus and a nearly continuous high-frequency train of action potentials is generated (Figure 50–10).

An important factor in the spread of focal seizures appears to be that the intense firing of the pyramidal neurons results in a relative decrease in synaptic transmission from the inhibitory GABAergic interneurons. This decrease may result from a change in the release of GABA (presynaptic mechanisms), a change in the chloride gradient responsible for the GABA_A receptor-mediated ion flux, or a change in GABA receptor activity (postsynaptic mechanism). Other factors that may contribute to the loss of the inhibitory surround include changes in dendritic morphology, changes in the density of receptors or channels, or changes in the amount of extracellular K⁺ ion accumulation. Prolonged firing also transmits action potentials to distant sites in the brain, which in turn may trigger trains of action potentials in neurons that project back to neurons in the seizure focus (backpropagation). Reciprocal connections between the neocortex and thalamus may be particularly important in this regard.

Despite our understanding of such mechanisms, we still do not know what causes a seizure to occur at any particular moment. The inability to predict when a seizure will occur is perhaps the most debilitating aspect of epilepsy. Some patients become adept at adjusting their lifestyle to avoid circumstances that can increase the likelihood of a seizure, such as sleep deprivation or stress. But in many individuals seizures do not follow a predictable pattern. In a few patients sensory stimuli such as flashing lights can trigger seizures, suggesting that repetitive excitation of some circuits causes a change in excitability that is dependent on the frequency of neuronal firing.

Both NMDA-type glutamate receptor activity and GABAergic inhibition undergo changes in sensitivity that depend on the frequency of firing of the presynaptic neuron, providing a possible cellular mechanism for altered network excitability. On a longer time scale, circadian rhythms and hormonal patterns may also influence the likelihood of seizures, as demonstrated by patients who have seizures only while sleeping (nocturnal epilepsy) or during their menstrual period (catamenial epilepsy). New devices for continuous monitoring at or near a seizure focus may allow clinicians to predict the timing of seizure generation. Such approaches offer the possibility of acute therapeutic intervention such as direct cortical stimulation to prevent seizures. The modest success of implanted vagal nerve stimulators in epilepsy that does not respond to other treatments provides one example of such an approach.

The Spread of Focal Seizures Involves Normal Cortical Circuitry

If activity in the seizure focus is sufficiently intense, the electrical activity begins to spread to other brain regions. Spread of seizure activity from a focus generally follows the same axonal pathways as does normal cortical activity. For example, the neurons in the primary motor and sensory cortex are organized functionally into vertical columns that run from the pial surface to the underlying white matter (see Chapter 15). The major input to sensory cortex comes from the thalamus and terminates in layer IV, whereas the output cells are in layer V. Reciprocal thalamocortical pathways connect the thalamus and cortex. Intracortical connections occur via short U fibers between adjacent sulci and via the corpus callosum, the major connection between the cerebral hemispheres. These thalamocortical, subcortical, and interhemispheric pathways can all become involved in seizure spread.

Focal seizure activity can spread from the seizure focus to other areas of the same hemisphere or across the corpus callosum to the contralateral hemisphere (Figure 50–11A). Once both hemispheres become involved, the seizure has become secondarily generalized. At this point the patient generally experiences loss of consciousness. The spread of a focal seizure usually occurs within a few seconds but can also take many minutes.

As the focal seizure begins to spread, the patient may experience some warning symptoms (an aura). If the seizure spreads slowly across the cortex, it may lead to a progression of clinical symptoms—a Jacksonian march in the case of a focal seizure involving the motor cortex. Focal seizures that quickly undergo secondary generalization provide little or no warning. Rapid secondary generalization is more likely if the seizure begins in the neocortex than if it begins in the limbic system (in particular, the hippocampus and amygdala).

An interesting unanswered question is what terminates a seizure. The only definitive conclusion at this point is that termination is not caused by metabolic exhaustion. During the initial 30 seconds or so of a typical secondarily generalized tonic-clonic seizure, neurons in the involved areas undergo prolonged depolarization and continuously fire action potentials (caused by loss of the afterhyperpolarization that normally follows a paroxysmal depolarizing shift). As the seizure evolves, the neurons begin to repolarize and the afterhyperpolarization reappears. The cycles of depolarization and repolarization correspond to the clonic phase of the seizure (Figure 50–10A).

The seizure is often followed by a period of decreased electrical activity, the postictal period, that may be accompanied by confusion, drowsiness, or even focal neurological deficits such as a hemiparesis (Todd paralysis). A neurological exam in the postictal period can lead to insights about the locus of the seizure focus.

Unlike the focal seizure a primary generalized seizure disrupts normal brain activity in both cerebral hemispheres simultaneously (Figure 50–11B). Generalized seizures and their associated epilepsies vary both in their manifestations and etiologies. Although the cellular mechanisms of primary generalized seizures differ in a number of interesting respects from those of focal or secondarily generalized focal seizures, a primary generalized seizure can be difficult to distinguish from a focal seizure that rapidly generalizes.

The most studied type of primary generalized seizure is the typical absence seizure (petit mal), whose characteristic EEG pattern (the 3 Hz spike-wave pattern in Figure 50–12A) was first identified by Hans Berger in 1933. F.A. Gibbs recognized the relationship of this EEG pattern to typical absence seizures (he aptly described the pattern as "dart and dome") and he attributed the mechanism to generalized cortical disturbance. The distinctive clinical features of typical absence seizures are clearly correlated with the EEG activity.

The typical absence seizure begins suddenly, lasts 10 to 30 seconds, and produces loss of awareness and only minor motor manifestations such as blinking or lip smacking. Unlike secondarily generalized seizures, primary generalized seizures are not preceded by an aura or followed by postictal symptoms. The spike-wave EEG pattern can be seen in all cerebral areas simultaneously and is immediately preceded and followed by normal background activity. Brief (1–5 s) runs of 3 Hz EEG activity without apparent clinical symptoms are common in patients with childhood absence seizures.

In contrast to Gibbs's hypothesis of diffuse cortical hyperexcitability, Penfield and Jasper noted that the EEG in typical absence seizures is similar to rhythmic EEG activity in sleep, so-called sleep spindles (see Figure 51–1). They proposed a "centrencephalic" hypothesis in which rapid generalization was attributed to rhythmic activity (pacing) by neuronal aggregates in the upper brain stem or thalamus that project diffusely to the cortex.

Research on animal models of generalized seizures and recent studies on the genetics of generalized epilepsy suggest that elements of both hypotheses are correct. In cats parenteral injections of penicillin, a weak GABA_A antagonist, produce behavioral unresponsiveness associated with an EEG pattern of bilateral synchronous slow waves (generalized penicillin epilepsy). During such a seizure thalamic and cortical cells become synchronized through the same reciprocal thalamocortical connections that contribute to normal sleep spindles during slow wave sleep.

Such seizures could in theory represent a form of diffuse hyperexcitability in the cortex. Recordings from individual cortical neurons show an increase in the rate of firing during a depolarizing burst that in turn produces a powerful GABAergic inhibitory feedback that hyperpolarizes the cell for approximately 200 ms after each burst (Figure 50–12C). This depolarization followed by inhibition differs fundamentally from the paroxysmal depolarizing shift in focal seizures in that GABAergic inhibition is preserved. In the typical absence seizure the summated activity of the bursts produces the spike while the summated inhibition produces the wave of the spike-wave EEG pattern.

What are the properties of cells and networks that facilitate this generalized and synchronous activity? An early clue came from studies of the intrinsic bursting of thalamic relay neurons. Henrik Jahnsen and Rodolfo Llinas found that these neurons robustly express the T-type voltage-gated Ca²⁺ channel that is inactivated at the resting membrane potential but becomes available for activation when the cell is hyperpolarized. A subsequent depolarization then transiently opens the Ca²⁺ channel (thus the name T-type) and the Ca²⁺ influx generates low-threshold Ca²⁺ spikes. Consistent with the hypothesis that T-type channels contribute to absence seizures, certain anticonvulsant agents that block absence seizures, such as ethosuximide (Zarontin) and valproic acid (Depakote), also block T-type channels. T-type channels are encoded by three related genes (Cav3.1 –Cav3.3), with Cav3.1 the predominant type in the thalamus.

The circuitry of the thalamus seems ideally suited to the generation of primary generalized seizures. The pattern of thalamic neuron activity during sleep spindles suggests a reciprocal interaction between thalamic relay neurons and GABAergic interneurons in the thalamic reticular nucleus and perigeniculate nucleus (Figure 50–12B). Studies of thalamic brain slices by David McCormick and his colleagues indicate that the interneurons hyperpolarize the relay neurons, thus removing the inactivation of T-type Ca²⁺ channels. This action leads to an oscillatory response: T-type Ca²⁺ channels drive rebound firing in the relay neurons, which stimulates GABAergic interneurons and leads to another round of rebound firing. The relay neurons also excite cortical neurons, as manifested in the EEG by a spindle wave (see Chapter 51). Both the T-type Ca²⁺ channel and the GABA_B receptor-channel play an important role in the generation of this activity that resembles human absence seizures.

Mutations in voltage-gated Ca²⁺ channels have produced several mouse models of generalized epilepsy, including the so-called totterer mouse. Studies of these mutants by Jeffrey Noebels and his colleagues have revealed that the animals develop primary generalized seizures when they reach adolescence. EEGs in these animals show a paroxysmal spike-wave discharge and seizures that are characterized by an arrest of behavior, similar to typical absence seizures.

The pioneering studies of the surgeon Wilder Penfield in Montreal in the early 1950s led to the recognition that removal of the temporal lobe in certain patients with focal seizures of hippocampal origin could reduce or cure epilepsy. As surgical treatment for difficult-to-control (so-called intractable) focal seizures became more common, it became clear that the surgical outcome is directly related to the adequacy of the resection. Thus precise localization of the seizure focus is essential.

Electrical mapping of seizure foci originally relied on the surface EEG, which we have seen is biased toward particular sets of neurons in the cortex immediately adjacent to the skull. However, intractable seizures often begin in deep structures that show little or no abnormality on the surface EEG at the onset of the seizure. Thus the surface EEG is somewhat limited in identifying the location of the seizure focus.

The development of magnetic resonance imaging (MRI) has markedly improved the noninvasive anatomical mapping of seizure foci. This technique is now routine in the evaluation of seizure foci in the temporal lobe and increasingly promising for identifying seizure foci in other locations. Anatomical mapping of seizure foci by MRI has shown that most patients with intractable temporal lobe seizures have atrophy and cell loss in the mesial portions of the hippocampal formation. There is a dramatic loss of neurons within the hippocampus (mesial temporal sclerosis), changes in dendritic morphology of surviving cells, and collateral sprouting of some axons. The anatomical resolution of MRI provides a noninvasive, quantitative assessment of the size of the hippocampus in epilepsy patients. Loss of volume of the hippocampus on one or another side of the brain generally correlates well with the localization of seizure foci in the hippocampus by electrical criteria.

Patients with mesial temporal lobe epilepsy often have unilateral disease, which leads to shrinkage of the hippocampus on one side. This is apparent in brain imaging as an apparent dilatation of the temporal horn of the lateral ventricle (Box 50–3). However, in many patients abnormalities cannot be detected using anatomical MRI; thus several functional imaging methods have been used as well.

Metabolic mapping takes advantage of the changes in cerebral metabolism and blood flow that occur in the seizure focus during the ictal and interictal periods. The electrical activity associated with a seizure places a large metabolic demand on brain tissue. During a focal seizure there is an approximately threefold increase in glucose and oxygen use. Between seizures the seizure focus often shows decreased metabolism. Despite the increased metabolic demands, the brain is able to maintain normal adenosine triphosphate (ATP) levels during a focal seizure. Conversely, the transient interruption of breathing during a generalized convulsive seizure causes a decrease in oxygen levels in the blood. This results in a drop in ATP concentration and an increase in anaerobic metabolism as indicated by rising lactate levels. This oxygen debt is quickly replenished in the postictal period, and no permanent damage to brain tissue results from a single generalized seizure.

Positron emission tomography (PET) scans of patients with focal seizures originating in the mesial temporal lobe frequently show interictal hypometabolism, with metabolic changes extending to the lateral temporal lobe, ipsilateral thalamus, basal ganglia, and frontal cortex. PET scans using nonhydrolyzable glucose analogs have been particularly helpful in identifying seizure foci in patients with normal MRI scans and in some early childhood epilepsies.

Unfortunately, for unclear reasons PET has been less reliable in localizing seizure foci in extratemporal areas such as the frontal lobe. An additional limitation is the expense of the PET scan and the short half-life of the isotopes (a nearby cyclotron is required). PET scanning can also be used to look for functional changes in neurotransmitter receptor binding and transport related to seizure activity.

A related technique that measures cerebral blood flow, single photon emission computed tomography (SPECT), has been used more frequently than PET. SPECT does not have the resolution of PET but can be performed in the nuclear medicine department of many large hospitals. Injection of radioisotopes and SPECT imaging at the time of a seizure (ictal SPECT) reveals a pattern of hypermetabolism followed by hypometabolism in the seizure focus and surrounding tissue. The use of magnetoencephalography and functional MRI may offer further advantages in the mapping of seizure foci.

With rigorous selection of patients for epilepsy surgery, the cure rate for seizures originating in the temporal lobe can approach 80%. Patients with complicating factors (eg, multiple foci) have lower success rates. However, even among these patients the number or severity of seizures is reduced. Patients who have been cured of seizures still experience cognitive problems such as memory loss and social problems such as adjustments to more independent living and limited employment opportunities. These factors emphasize the need for treatment as early in life as feasible.

Repeated Convulsive Seizures Are a Medical Emergency

As noted earlier, brain tissue can compensate for the metabolic stress of a focal seizure or the transient decrease in oxygen delivery during a single generalized seizure. In a generalized seizure, stimulation of the hypothalamus leads to massive activation of the stress response of the sympathetic nervous system. The increased systemic blood pressure and serum glucose initially compensate for increased metabolic demand, but these homeostatic mechanisms fail during prolonged seizures, particularly seizures that are convulsive. The resulting systemic metabolic derangements, including hypoxia, hypotension, hypoglycemia, and acidemia, lead to a reduction in high-energy phosphates (ATP and phosphocreatine) in the brain and thus can be devastating to brain tissue.

Systemic complications such as cardiac arrhythmias, pulmonary edema, hyperthermia, and muscle breakdown can also occur. Repeated generalized seizures without return to full consciousness between seizures, called status epilepticus, is a true medical emergency. This condition requires aggressive seizure management and general medical support because 30 or more minutes of continuous convulsive seizures leads to brain injury or even death. Status epilepticus can involve nonconvulsive seizures (focal or primary generalized) for which the metabolic consequences are much less severe.

Excitotoxicity Underlies Seizure-Related Brain Damage

Repeated seizures can damage the brain independently of cardiopulmonary or systemic metabolic changes, suggesting that local factors in the brain can result in neuronal death. The immature brain appears particularly vulnerable to such damage, perhaps because of greater electrotonic coupling between neurons in the developing brain, less effective potassium buffering by immature glia, and decreased glucose transport across the blood-brain barrier.

Wilhelm Sommer in 1880 first noted the vulnerability of the hippocampus to such insults, with preferential loss of the pyramidal neurons in the CA1 and CA3 regions. This pattern has been duplicated in experimental animals by electrical stimulation of afferents to the hippocampus or by injection of excitatory amino acid analogs such as kainic acid. Interestingly, kainic acid causes local damage at the site of injection and also at the site of termination of afferents originating at the injection site.

These observations indicate that release of the excitatory transmitter l-glutamate during excessive stimulation such as a seizure can itself cause neuronal damage, a condition that has been termed excitotoxicity. Excitotoxicity following seizures probably is caused more by excessive stimulation of synaptically- activated glutamate receptors than by tonic increases in extracellular glutamate. The histological appearance of acute excitotoxicity includes massive swelling of cell bodies and dendrites, which is consistent with the predominant locations of glutamate receptors and excitatory synapses.

Although the cellular and molecular mechanisms of excitotoxicity are still not fully understood, several features are clear. Overactivation of glutamate receptors leads to an excessive increase in intracellular Ca²⁺ that can activate a self-destructive cellular cascade involving many calcium-dependent enzymes, such as phosphatases, proteases, and lipases. Lipid peroxidation can also cause production of free radicals that damage vital cellular proteins and lead to cell death. The role of mitochondria in Ca²⁺ homeostasis and in control of free radicals may also be important. The pattern of cell death was first thought to reflect necrosis because of the autolysis of critical cellular proteins. However, death genes, characteristic of programmed cell death or apoptosis, may also be involved.

Seizure-related brain damage or excitotoxicity can be specific to certain types of cells in particular brain regions, perhaps because of protective factors, such as calcium-binding proteins in some cells, and sensitizing factors, such as the expression of calcium-permeable glutamate receptors in other cells. For example, excitotoxicity induced in vitro by activation of AMPA-type glutamate receptors preferentially affects interneurons that express AMPA-type receptors that have high Ca²⁺ permeability, providing a possible mechanism for their selective vulnerability.

Several outbreaks of amnestic shellfish poisoning provide a vivid example of the consequences of overactivation of glutamate receptors. Domoic acid, a glutamate analog not present in the brain, is a natural product of certain species of marine algae that flourish during appropriate ocean conditions. Domoic acid can be concentrated by filter feeders such as shellfish. Ingestion of domoic acid contaminated shellfish sporadically causes outbreaks of neurological damage, including severe seizures and memory loss (amnesia). The area most sensitive to damage is the hippocampus, providing further support for the excitotoxicity hypothesis and the critical role of the hippocampus in learning and memory (see Chapter 65).

A single seizure does not warrant a diagnosis of epilepsy. Normal people can have a seizure under extenuating circumstances, such as following drug ingestion or extreme sleep deprivation. Clinicians look for possible causes of seizures in such cases but usually do not begin treatment with anticonvulsants following a single seizure. Unfortunately, our understanding of what factors contribute to susceptibility to epilepsy is still rudimentary.

Some forms of epilepsy are caused in part by a genetic predisposition. For example, infants with febrile seizures often have a family history of similar seizures. The role of genetics in epilepsy is supported by the existence of familial epileptic syndromes in humans as well as seizure-prone animal models with such exotic names as Papio papio (a baboon with photosensitive seizures), audiogenic mice (in which loud sounds induce seizures), and reeler and totterer mice (names alluding to the clinical manifestations of cerebellar mutations in these animals). Even with a genetic predisposition or a structural lesion, the evolution of the epileptic phenotype often involves maladaptive changes in brain structure and function.

Among the Genetic Causes of Epilepsy Are Ion Channel Mutations

Recent studies have provided a wealth of new information concerning the molecular genetics of epilepsy. At present more than 70 genes have been linked to an epileptic phenotype; approximately half of these were discovered in humans and the others in animals, mostly mice. The affected proteins include ion channel sub-units, proteins involved in synaptic transmission such as transporters, vesicle proteins, synaptic receptors, and molecules involved in Ca²⁺ signaling. For example, seizures in the totterer mutant mouse are due to a spontaneous mutation in the gene that encodes the Ca_V 2.1-subunit or α_1A -subunit of the P/Q-type voltage-gated Ca²⁺ channel. That mutations in these classes of proteins can cause epilepsy is perhaps not unexpected given the dependence of seizures on synaptic transmission and neuronal excitability. Some of the other genes linked to epilepsy in mice have been more surprising, such as the genes for Centromere BP-B, a DNA-binding protein, and the sodium/hydrogen exchanger, which is affected in the slow-wave epilepsy mouse.

A wide variety of human gene mutations cause neurological disorders of which epilepsy is only one manifestation. For example, Rett syndrome, a disease associated with mental retardation, autism, and seizures, is caused by mutations in the gene that codes for MECP2 (methyl-CpG-binding protein-2), a regulator of gene transcription. Although the exact links are unknown, it is clear that mutations in many different genes may result in epilepsy.

Most genetic epilepsy syndromes in humans have complex rather than simple (Mendelian) inheritance patterns, suggesting the involvement of many rather than single genes. Nevertheless, a number of monogenic epilepsies have been identified in studies of families with epilepsy. Ortrud Steinlein and colleagues reported in 1995 that a mutation in the α₄ -subunit of the nicotinic acetylcholine receptor-channel is responsible for autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE), the first example of an autosomal gene defect in human epilepsy. Since then other voltage- and ligand-gated channel proteins have been identified as critical genes for epilepsy. Mutations in ion channel genes (channelopathies) constitute a major cause of known monogenic epilepsies (Figure 50–16).

In voltage-gated channels, mutations largely affect the main pore-forming subunit(s), but there are also examples of epilepsy-causing mutations in regulatory subunits. When examined in vitro, the mutant channel proteins are most commonly associated with either reductions in the expression of the channel on the surface of the plasma membrane (caused by reduced targeting to the membrane or premature degradation) or altered kinetics of the channels. The problem of how changes in ion channel gating might affect the excitability of neurons and their synchronization during seizure generation is straight-forward. However, mutations in ion channel genes may also affect neuronal development and thus exert their epileptogenic effects through a secondary action on cell migration, network formation, or patterns of gene expression.

In the early days of research on epilepsy genes it was widely expected that the genes would underlie generalized epilepsies, based on the idea that a gene mutation (eg, in an ion channel) would be expected to affect most neurons. However, the first autosomal dominant epilepsy gene discovered by Steinlein and colleagues causes focal seizures. In retrospect this should not be so surprising because channel subunits are rarely expressed uniformly in the brain, and some brain regions are more likely to generate seizures than other regions. For example, totterer mice with mutations in the pore-forming Ca_V 2.1-subunit of P/Q-type Ca²⁺ channels show spike-wave type seizures. The seizures occur principally in immature mice, presumably because P/Q-type Ca²⁺ channels are the predominant isoform early in development, whereas N-type Ca²⁺ channels predominate later.

Moreover, one mutant gene can give rise to different epilepsy phenotypes, whereas different mutant genes can cause the same epilepsy phenotype. As an example of the latter, the ADNFLE syndrome, first discovered as a mutation in the α₄ -subunit of the nicotinic acetylcholine receptor, can also be caused by a mutation in the α₂ -subunit. But not all family members who carry this autosomal dominant mutation have epilepsy, indicating that even in monogenic epilepsy other, perhaps nongenetic, factors can influence the phenotype.

The GEFS+ syndrome (generalized epilepsy with febrile seizures plus) is a good example of this heterogeneity. It can involve different seizure types in different family members and has been seen in families with mutations in the genes for three different Na⁺ channel subunits and two GABA_A receptors. Family studies of primary generalized epilepsy suggest that seizure types may be heritable within families. These issues indicate that monogenic epilepsies are likely modified by other genes, environmental influences, and even experience-dependent changes in synapses.

Altered cortical development may be a common cause of epilepsy. The higher resolution of MRI scans has revealed an unexpectedly large number of cortical malformations and localized areas of abnormal cortical folding in patients with epilepsy. Thus mutations that disturb the normal formation of the cortex or network wiring are candidate genes for epilepsy. This idea is supported by the mapping of two X-linked cortical malformations with epileptic phenotypes: familial periventricular heterotopia and familial subcortical band heterotopia. The genes responsible for these two disorders, filamin A and doublecortin, are presumably important in neuronal migration. Small focal cortical dysplasias can function as seizure foci that give rise to focal and secondarily generalized seizures, whereas more extensive cortical malformations can cause a variety of seizure types and usually are associated with other neurological problems.

Unfortunately, most cases of epilepsy cannot be explained by even the recent surge in the identification of epilepsy genes. The identification of large numbers of patients through online registries, such as the Epilepsy Phenome/Genome Project, may provide the population sample needed to evaluate susceptibility genes that underlie complex inheritance patterns.

Epilepsies Involving Focal Seizures May Be a Maladaptive Response to Injury

Epilepsies involving repeated focal seizures often develop following a discrete cortical injury such as a penetrating head wound. This injury serves as the nidus for a seizure focus, leading at some later point to seizures. This has led to the idea that the early insult triggers a set of progressive physiological or anatomical changes that lead to chronic seizures. That is, the characteristic silent interval (usually months or years) between the insult and the onset of recurrent seizures may be a period of maladaptive changes that might be amenable to therapeutic manipulation. Although an attractive hypothesis, a unified picture of this process has yet to emerge. The most promising evidence has come from studies of tissue removed from patients undergoing temporal lobectomy and rodent models of limbic seizures.

In one experimental model hyperexcitability is induced by repeated stimulation of limbic structures, such as the amygdala or hippocampus. The initial stimulus is followed by an electrical response (the afterdischarge) that becomes more extensive and prolonged with repeated stimuli until a generalized seizure occurs. This process, called kindling, can be induced by electrical or chemical stimuli. Many investigators believe that kindling may contribute to the development of epilepsy in humans.

Kindling is thought to involve synaptic changes that resemble those important in learning and memory (see Chapter 65). These include short-term changes in excitability and persistent morphological changes, including axonal sprouting. Rearrangements of synaptic connections have been observed in the dentate gyrus of patients with long-standing focal seizures of hippocampal origin as well as following kindling in experimental animals (Figure 50–17). In addition to axonal sprouting, changes include alterations in dendritic structure, control of transmitter release, and expression of ion channels and pumps.

The long-term changes that lead to epilepsy also are likely to involve specific patterns of gene expression. For example, the proto-oncogene c-fos and other immediate early genes as well as growth factors can be activated by seizures. Because many immediate early genes encode transcription factors that control other genes, the gene products that result from epileptiform activity could initiate a cascade of changes that contribute to the development of epilepsy by altering such mechanisms as cell fate, axon targeting, dendritic outgrowth, and synapse formation.

Seizures are one of the most dramatic examples of the collective electrical behavior of the mammalian brain. The distinctive clinical patterns of focal seizures and primary generalized seizures can be attributed to the different patterns of activity of cortical neurons. Studies of focal seizures in animals reveal a series of events—from the activity of neurons in the seizure focus to synchronization and subsequent spread of epileptiform activity throughout the cortex. The gradual loss of GABAergic surround inhibition is critical to the early steps in this progression. In contrast, generalized seizures are thought to arise from activity in thalamocortical circuits, perhaps combined with a general abnormality in the membrane excitability of cortical neurons.

The EEG has long provided a window on the electrical activity of the cortex, both in normal phases of arousal and during abnormal activities such as seizures. The EEG can be used to identify certain electrical activity patterns associated with seizures, but it provides limited insight into the pathophysiology of seizures. Several much more powerful and noninvasive approaches are now available to locate a seizure focus. These advances have led to the widespread and successful use of epilepsy surgery for selected patients, particularly those with focal seizures that originate in the hippocampus.

The increasing power of genetic, molecular, and modern cell-physiological techniques applied to the study of seizures and epilepsy gives new hope that this research will provide new therapeutic options for patients afflicted with epilepsy as well as new insights into the function of the mammalian brain. Further neurobiological studies of the progression from an acute seizure to the development of epilepsy should provide alternative strategies for treatment beyond the standard options of anticonvulsants or epilepsy surgery.