Jonathan K Kleen and Gregory L Holmes
Although seizures are the most striking clinical manifestation of the epilepsies, children with epilepsy are at risk not only for seizures but also for a number of comorbid health problems defined as conditions that occur in children with epilepsy at a higher rate than would be expected by chance (Committee on the Public Health Dimensions of the Epilepsies et al. 2012). Common comorbidities that occur in children with epilepsy include cognitive dysfunction, such as memory, attention, or processing difficulties; mental health conditions, including depression, anxiety, and oppositional-defiant disorder; and somatic comorbidities, such as sleep disorders and migraines. Epilepsy comorbidities are common and often severe. Many parents of children with epilepsy consider the comorbidities to outweigh the burden of the seizures themselves.
Cognitive abnormalities and behavioral disorders are among the most common and troublesome comorbidities associated with epilepsy in children. In children with epilepsy, there is an associated high rate of cognitive difficulties that compromise educational progress and achievement throughout their life (Berg et al. 2008). In addition to a higher incidence of low IQ scores (Farwell et al. 1985), approximately half of the children with epilepsy have a discrepancy between IQ and achievement (Fastenau et al. 2008). Lower IQ scores are particularly common in children with poorly controlled (pharmacoresistant) seizures (Berg et al. 2012). Likewise, children with epilepsy are at very high risk of attention-deficit–hyperactivity, anxiety, and conduct disorders (Cavazzuti and Nalin 1990; Dunn et al. 1997; Austin and Dunn 2002; Austin et al. 2004; Pellock 2004).
The major determinant of outcome in children with epilepsy is the etiology of their illness. However, there is increasing evidence that seizures, antiepileptic drugs, and interictal EEG abnormalities can contribute to adverse outcomes. In children, it is often difficult to differentiate the adverse cognitive effects of seizures and their treatment from transient EEG abnormalities because they all tend to converge. In animal studies, one can induce seizures, interictal spikes (IISs), or both in the normal brain without antiepileptic drug treatment. This allows investigation into the biological mechanisms underpinning independent influences of IISs or seizures on cognitive impairment. In this chapter, pertinent data from immature animals will be discussed in relationship to human studies.
Substantial literature shows that recurrent seizures in the developing brain can result in long-term adverse consequences. Rat pups that are subjected to multiple recurrent seizures in the first weeks of life have considerable cognitive and behavioral abnormalities later on. This includes deficits of spatial cognition in the Morris water maze (Holmes et al. 1998; Huang et al. 1999; Liu et al. 1999; Karnam et al. 2009a; 2009b) and delayed non-match-to-sample task (Kleen et al. 2011b), auditory discrimination impairments (Neill et al. 1996), abnormal activity level (Karnam et al. 2009a), increased emotionality (Holmes et al. 1993), and reduced behavioral flexibility (Kleen et al. 2011a). Recurrent early-life seizures also produce physiological changes including a persistent decrease in GABA currents in the hippocampus (Isaeva et al. 2006) and neocortex (Isaeva et al. 2009), enhanced excitation in the neocortex (Isaeva et al. 2010), impairment in spike frequency adaptation (Villeneuve et al. 2000), and marked reductions in after-hyperpolarizing potentials following spike trains (Villeneuve et al. 2000). They are also associated with impaired long-term potentiation (Karnam et al. 2009a), enhanced short-term plasticity (Hernan et al. 2013), alterations in theta oscillation power (Karnam et al. 2009b), and impaired place cell coherence and stability (Karnam et al. 2009b).
Although early-life seizures have detrimental effects on cognitive function, recurrent seizures during the first 2 weeks of life do not produce cell loss (Holmes et al. 1998; Liu et al. 1999; Riviello et al. 2002). However, seizures in immature rats can produce synaptic reorganization as evidenced by CA3 sprouting (Holmes et al. 1998; Huang et al. 1999; Sogawa et al. 2001; Huang et al. 2002) and decreased neurogenesis (McCabe et al. 2001).
In the majority of studies, recurrent seizures have been induced in rats with “normal” brains. Using normal developing rats, investigators have been able to differentiate the effects of seizures from the etiological cause of epilepsy. However, in children, seizures do not occur in the “normal brain.” As such, results from studies with animal models of epilepsy must be taken with a degree of reserve before applying to patients. Few investigators have studied the consequences of seizures in animals with structural abnormalities of the brain. In one of the few studies of this type, Lucas et al. (2011) found that seizures induced in rat pups with malformations of cortical development but without seizures had severe spatial cognitive deficits in the water maze. When the rat pups were subjected to recurrent flurothyl-induced seizures and tested at 25 days of age (immediate postweaning), their cognitive performance was worse. In contrast, in animals tested during adolescence, seizures had no additional adverse effect. The authors also investigated whether the severity of the structural abnormality and seizures affected brain weight, cortical thickness, hippocampal area, and cell dispersion area. Early-life seizures did not have a significant impact on any of these parameters, although the size of the dysplasia did have a significant effect.
Childhood epilepsy carries a significant risk for a variety of problems involving cognition. IQ scores in children with epilepsy have a skewed distribution toward lower values (Farwell et al. 1985; Neyens et al. 1999). The number of children with epilepsy and learning disabilities in school or behavioral problems is greater than that of the normal population (Sillanpaa et al. 1998; Williams et al. 1998; Bailet and Turk 2000; Wakamoto et al. 2000). Predictors of poor cognitive outcome include higher seizure frequencies (Hermann et al. 2008) and longer duration of the epilepsy (Farwell et al. 1985; Seidenberg et al. 1986).
Most children with epilepsy do not have a decline in intelligence or behavior, however. It could also be fairly argued that children with a high seizure frequency and a long duration of epilepsy have a more severe underlying brain abnormality than those with less severe epilepsy. Many children that develop epilepsy appear to have cognitive deficits that precede the onset of the seizures. This might suggest again that the etiology of the seizures, not the seizures themselves, is responsible for the impaired cognition (Fastenau et al. 2009). Nevertheless, some children with epilepsy slow in their cognitive development (Neyens et al. 1999) or even have progressive declines of IQ on serial intelligence tests over time (Bourgeois et al. 1983). Similarly, increasing duration of epilepsy in temporal lobe epilepsy is associated with declining performance across both intellectual and memory measures (Hermann et al. 2002). It is difficult to ascertain whether this is related to the underlying etiology causing progressive dysfunction or more so to the cumulative effects of seizures over time.
Animal data suggest that seizures in early childhood are more detrimental than those occurring at an older age. Risk factors for cognitive impairment in children with epilepsy include an early age at onset of seizures (Huttenlocher and Hapke 1990; Glosser et al. 1997; Bulteau et al. 2000; Bjornaes et al. 2001; Hermann et al. 2002; Cormack et al. 2007), particularly during the neonatal period (Glass et al. 2009). Epileptic syndromes in which psychomotor deterioration occurs have an early age at onset. These include early infantile epileptic encephalopathy with suppression burst (Ohtahara syndrome), early myoclonic encephalopathy, migrating partial epilepsy in infancy, infantile spasms (West syndrome), severe myoclonic epilepsy of infancy (Dravet syndrome), Lennox–Gastaut syndrome, myoclonic-astatic epilepsy, continuous spike–wave discharges during sleep (CSWS), and Landau–Kleffner syndrome (LKS) (Genton and Dravet 1997; Panayiotopoulos 2002; Nabbout and Dulac 2003). It is likely that pediatric seizures affect the activity-dependent processes among developing networks. They may thus be more detrimental than seizures in older individuals, in whom neural circuitry is relatively more fixed and less malleable. For example, infants with infantile spasms and hypsarrhythmia have EEGs that have high coherences, a measure of connectivity, predominately at long interelectrode distances. At short interelectrode distances, coherences are decreased in the theta and beta range, particularly in the frontal region (Burroughs et al. 2014). This suggests impaired local cortical integration in frontal regions (which are important for executive function development), yet abnormal congruence between areas that would normally have functional differentiation. Altered connectivity may thus underlie cognitive impairment among children with infantile spasms, and this may extend to other children with both seizures and developmental delay.
Although the etiology of the seizures clearly plays the major role in cognitive and behavioral development, childhood seizures themselves (i.e. independent of etiology) can lead to cognitive impairment (Glass et al. 2009; Korman et al. 2013; Payne et al. 2014). Among children with focal cortical dysplasia, Korman et al. (2013) found that the age at onset of epilepsy and the extent of the dysplasia each contributed independently to cognitive dysfunction. They suggested that the early onset of epilepsy led to poor cognitive outcomes potentially through disrupting critical periods of development. Furthermore, it was suggested that even a localized lesion could engender developmental deficits if the age at onset of epilepsy was early. In a pediatric intensive care unit, it was found that seizures, even without a behavioral correlate, were strongly associated with neurological decline that could not be accounted for by the etiology (Payne et al. 2014).
Although seizures themselves are significantly related to worse cognitive outcomes as described previously, there is an extensive literature on the topic of whether the IIS activity also contributes independently to cognitive impairment. Despite being quite transient and usually without overt clinical manifestations, these brief bursts of activity are hypothesized to affect cognition both dynamically and cumulatively over time.
Both the acute (transient) and chronic (cumulative) effects of IISs have been studied in rats. Animal models fortunately allow intracranial recordings (increasing sensitivity and specificity of IISs on EEG) and more controlled testing on large numbers of trials in cognitive tasks. The latter is often difficult to attain in patients with epilepsy given attentional, motivational, and situational (e.g. EEG lead discomfort) issues. The majority of studies dealing with the acute effects of IISs have been conducted in adult rats because it has been difficult to induce sustained IISs in immature rats. For example, following status epilepticus induced by kainic acid, pilocarpine, or electrical stimulation, IISs occur within days of the status epilepticus in adult rats, whereas in immature rats, status epilepticus rarely results in IISs. Although the acute effects of IISs in adult rats are remarkably similar to those found in adult patients with epilepsy, caution is required in extrapolating results from adult rats to children.
IISs have been shown to result in a task-specific cognitive impairment in adult rats. Using a within-subject analysis to analyze how IISs might independently affect memory processing in the hippocampus, Kleen et al. (2010) studied rats that developed chronic IIS following intrahippocampal pilocarpine in a hippocampal-dependent operant behavior task (delayed match-to-sample test). Although IISs that occurred during memory encoding or memory maintenance did not affect performance, hippocampal IISs that occurred during memory retrieval strongly impaired performance. Memory retrieval is especially reliant on hippocampal activation (Montgomery and Buzsaki 2007). Thus, IISs were most disruptive when the active engagement of neurons involved in performing this stage of the task was critical (Fig 6.1).
The physiological effects of IIS have been investigated. There is a sustained reduction of action potentials in the hippocampus for up to 2 seconds following a local IIS (Zhou et al. 2007a) and up to 6 seconds following flurries of spikes (Zhou et al. 2007b). The extensive inhibitory wave immediately after IIS can also reduce the power of gamma oscillations and other important oscillatory signals in the hippocampus (Urrestarazu et al. 2006). Because oscillations are closely linked with ongoing learning and memory function (Halasz et al. 2005), this type of transient disruption likely contributes to cognitive deficits. As noted previously, whether IIS in immature rats also causes a transient impairment in cognition is known. However, in view of the physiological effects IISs have in the mature brain, the same phenomenon would likely occur in the immature brain. Furthermore, as with seizures, this effect could have more impact on a very young brain because network development is heavily influenced by local dynamic electrophysiological activity, which would be pathological in this case.
There is a considerable amount of animal data showing that IISs during early life have long-term adverse effects on the developing neural circuits. In studies of the effects of IIS on network development, IISs were elicited by either penicillin (Baumbach and Chow 1981; Crabtree et al. 1981) or bicuculline (Campbell et al. 1984; Ostrach et al. 1984) through focal application in the rabbit striate cortex. IISs were elicited in this manner for 6 to 12 hours daily from P8–9 to P24–30. Note that P indicates the number of days after birth (although note that there are substantial species differences; P10 in a rat, for instance, is approximately analogous to the human developmental stage at the time of birth). Despite frequent IISs, none of the rabbits had behavioral seizures. In single-unit recordings from the lateral geniculate nucleus, superior colliculus, and occipital cortex ipsilateral to the hemisphere with IIS, receptive field types were abnormally distributed. Normal recordings were found from the contralateral hemisphere. Remarkably, these findings were age dependent. Adult rabbits with IISs induced in a similar fashion during adulthood maintained normal disruption of receptive field types. This highlights an additional vulnerability of critical developmental periods to cumulative IIS effects over time.