The neuropsychological evaluation of children with epilepsy involves the assessment and interpretation of a broad range of variables that have the potential to alter the natural course of brain–behavior development. These variables include seizure type and frequency, lesion lateralization and localization, type of neuropathology, side effects of anticonvulsants, age of epilepsy onset, and elapsed time since onset. Reliable neuropsychological diagnosis and prognosis of a child or adolescent with epilepsy requires the careful consideration of all of these factors and how they interact to affect future cognitive development and organization within the developing brain. Subsequent neuropsychological deficits can range from relatively specific impairments in one neurocognitive system to severe generalized intellectual disability. A child’s ability to compensate for and/or overcome neuropsychological deficits will depend in part on the child’s exposure to risk or protective factors within his or her environment, such as quality of educational support, family attitudes toward epilepsy and learning difficulties, socioeconomic status, access to specialist rehabilitative programs, behavioral management strategies, pharmacologic management, and/or surgical management. A combination of these variables interacting with neuropsychological impairments will influence the child’s overall quality of life and mental health.

Contemporary pediatric neuropsychology for children with epilepsy plays an important role in both enhancing clinical practice and increasing the knowledge base of developmental cognitive neuroscience. From a clinical standpoint, neuropsychological evaluations in children are carried out to establish cognitive status preoperatively, identify eloquent tissue (subserving language, memory, visuospatial perception, and motor function) that may be at risk during surgical intervention, determine the response of cognitive and/or behavioral systems to neuropharmacologic agents, provide prognostic indicators of cognitive and behavioral outcome to help both clinical management and guide educational support and intervention, monitor postoperative outcome and long-term progress through transitional stages of late childhood and adolescence, and identify children who might be at risk of decline or regression. From a scientific viewpoint, investigations of children with epilepsy using a systems neuroscience approach can advance understanding of the extent and limits of the plasticity and reorganizational capacity of the immature brain, determine the costs of reorganization and the adaptive/maladaptive compensatory responses to brain damage, document the developmental trajectory of functional brain organization in the presence of focal or diffuse pathology, establish brain structure/function relationships, and help identify the neural substrates of cognitive function and dysfunction, with the ultimate goal of translating the basic neuroscience findings to inform clinical diagnosis, prognosis, and management.

Over recent years the neuropsychological study of children with epilepsy has gained considerable momentum due to the rise in neurosurgical procedures conducted at younger developmental ages. Pediatric neurosurgery has increased because of the development of enhanced presurgical investigative techniques (e.g., quantitative methods of structural and functional neuroimaging, electroencephalogram [EEG] telemetry, ictal and interictal single photon emission computed tomography [SPECT], and interictal positron emission tomography [PET] examinations) that help to identify seizure type and focus, site and source of seizures, and type of neuropathology more reliably than before.⁵ There have also been improvements in the safety of neurosurgical practice²¹ and increasing evidence that early surgery may counteract pronounced restriction or regression of cognitive and psychosocial development due to persisting epilepsy.^23,70

To reflect the increasing interest in neuropsychological evaluations in relation to pediatric neurosurgery, this chapter will consider many of the developmental issues peculiar to pediatric neuropsychology by focusing on studies of children pre- and post-neurosurgery. Alternative models of the ontogeny of hemispheric specialization are introduced to help understand the potential effects that epilepsy can have on cognitive development early in life. Such models are important for pediatric neuropsychologists to consider when interpreting results of evaluations carried out pre- and post-neurosurgery. The relevance of such models to one of the most common neurosurgical procedures for children with epilepsy, namely hemispherectomy, is discussed based on studies of intellectual function and studies of speech and language. We include data from our series of children who have undergone hemispherectomy to highlight the variables that influence intellectual outcome, in particular the interaction between age at onset of epilepsy and hemispheric side of lesion. The effects on speech and language development in children who have bilateral or unilateral left- and right-sided lesions are outlined, and issues of potential reorganization of speech and language following early left-sided lesions are discussed in relation to recent functional neuroimaging studies that demonstrate the importance of hippocampal involvement in language reorganization. The complexity of understanding brain–behavior development following the onset of epilepsy in childhood will become clear, as will the need for clinical pediatric neuropsychologists to be integrated within multidisciplinary pediatric epilepsy surgery services and work collaboratively alongside developmental cognitive neuroscientists to mutually advance clinical practice and scientific understanding.

There is often an implicit assumption that brain–behavior relationships identified in adults can be applied to children. For
example, it is well known that the adult human brain is relatively asymmetric in function for some cognitive processes: Acquired lesions to Broca’s area and the perisylvian regions of the left hemisphere affect speech and language,^17,77 whereas lesions to the right hemisphere more often affect aspects of visuospatial ability.^64,83 However, the application of such knowledge to the pediatric population is usually made in the absence of supporting empirical data⁹ and without an understanding of normal and abnormal patterns of brain maturation and functional specialization during the course of development.

Historically, the evolution of ideas relating to the ontogeny of hemispheric specialization has been dominated by three models. The equipotentiality model first proposed by Lenneberg³⁵ states that the two cerebral hemispheres have equal potential at the start to subserve different aspects of cognitive function but gradually acquire specialization with increasing age and learning experience up to adolescence, when the pattern becomes crystallized and resembles that of adults. The major implication of the equipotentiality model is that the earlier the lesion, the more effective is the compensatory processes, and, consequently, the better is the cognitive outcome. In fact, it is now recognized that age at injury per se is not the sole determinant of outcome, and other factors (e.g., severity and extent of damage, locus of injury, etiology of the lesion, elapsed time since the onset of the lesion, presence or absence of epilepsy, unilateral or bilateral pathology) each contributes to the eventual outcome. However, this model can account for the impressive recovery of function (especially for speech and language) that ensues after early unilateral lesions and for the emergence of selective deficits (e.g., aphasic symptoms), signifying a relative decline in plasticity, with increasing age.

The early-specialization model posits that early onset of unilateral lesions leads to material-specific deficits that correspond to the side of damage, suggesting that the hemispheric division of labor for the processing of verbal and nonverbal information is present at or shortly after birth (for reviews see Bates et al.³ and Vargha-Khadem et al.⁶⁹). The early-specialization view was invoked by a number of researchers^{10,22,32,80,81,82} to provide a post hoc account for a series of empirical findings on children with unilateral lesions, including hemispherectomy, showing material-specific deficits corresponding to the hemispheric side of damage. A major implication of this model is that brain organization in the young child is modular just as it is in the adult, and that the double dissociation of function that is the hallmark of adult neuropsychology is present from infancy before the emergence of different aspects of cognitive function. Despite its appeal because of continuity with adult models, the early-specialization view does not satisfactorily account for the impressive rescue of articulate and grammatically formulated speech that is frequently seen after extensive left hemisphere pathology of early origin or the absence of chronic dysphasic symptoms after such early insult. A more fundamental problem for the model is the dearth of empirical evidence supporting a clear pattern of hemispheric specialization in cohorts of healthy children at different ages or a double dissociation of function in pediatric patients with early unilateral lesions.^2,3,39,69,72

In an attempt to reconcile the two extreme views, that of equipotentiality, which is compatible with developmental plasticity, versus early specialization, which is consistent with modular organization of brain function across the developmental span, Satz et al.⁵⁵ proposed a compromise position, referred to by Bates et al.³ as “constrained plasticity” and by Vargha-Khadem et al.⁶⁹ as “ontogenetic specialization.” This model assumes that there is a genetically determined anatomic basis to hemispheric specialization, and that during normal brain development, the functional expression of this genetic predisposition can unfold early in life. The model also assumes, however, that during normal development, there is an interaction between environmentally evoked neural activity that becomes progressively complex and nonredundant and neural plasticity that gradually declines with increasing age. Early brain injury can counteract the genetic predisposition toward specialization and change the normal trajectory of the activity/plasticity interaction. Thus, children who sustain brain damage early in life suffer two setbacks, one relating to the direct consequences of the actual lesion and the other relating to an altered developmental trajectory that is likely to be different from normal. The ontogenetic specialization model fits well with theories of brain and cognitive development (e.g., Alexander Luria, Jean Piaget) insofar as it conceives of the interactions between neural activity and plasticity emerging in stages of increasing complexity during childhood and adolescence. Furthermore, it also fits well with research on normal development of intelligence and language functions, in which evidence suggests that (a) whole-brain gray matter volume obtained from magnetic resonance imaging (MRI) is strongly correlated with IQ in older but not younger children,⁷⁶ (b) the two cerebral hemispheres are more nearly equal in linguistic processing capacity in infancy^7,8 than they are later in life,^45,59 and (c) the degrees of left lateralized functional MRI (fMRI) activation on a verb generation task²⁵ and blood oxygenation level–dependent (BOLD) fMRI signal during sensorimotor and language tasks⁵⁶ increase with age. Although the fMRI activation patterns on a variety of language tasks performed by normal children are the same as those in adults,^15,16,79 there are qualitative and quantitative differences suggesting that maturational features in functional neuroanatomy and/or differences in ability are distinct in development.^57,79 Finally, the model is also consistent with increasing recognition that in the aftermath of early unilateral injury, many factors can influence and counteract the normal developmental trajectory toward hemispheric specialization. For example, a small unilateral lesion acquired in the second decade of childhood can leave undisturbed the normal crystallization of hemispheric specialization and yield a selective deficit that is consistent with the early-specialization view. In contrast, an extensive unilateral lesion acquired early in life, before the development of complex cognitive repertoires, can lead to extensive interhemispheric reorganization and relative stunting of cognitive development across all domains, consistent with the equipotentiality model. With advances in quantitative neuroimaging techniques delineating the site and extent of neuropathology and functional imaging paradigms revealing activation patterns associated with specific cognitive tasks, it is now possible to examine structure/function relationships in pediatric patients with epilepsy and translate the findings to aid surgical decision making and subsequent management.

Hemispherectomy—complete or partial removal or disconnection of one cerebral hemisphere—is one of the commonest forms of pediatric neurosurgery for the relief of epilepsy, constituting up to 30% of neurosurgical procedures,⁵ particularly during the first decade of life. When children are assessed for hemispherectomy, one of the central questions concerns the status of cerebral lateralization and the consequences of surgery for intellectual functions and speech and language. Candidates for surgery usually present with a unilateral structural lesion and seizures, the origins of which date to pathologic congenital or perinatal processes or events, such as Sturge-Weber syndrome or hemimegalencephaly, although some children have later acquired disorders such as Rasmussen syndrome that can develop from about 1 year of age through to adolescence.¹³

In 2004, Pulsifer and colleagues reported on the neuropsychological outcome of the largest series (n = 71) of
hemispherectomized patients studied to date.⁵¹ Patients were divided into groups depending on the etiologic category (cortical dysplasia; Rasmussen encephalitis; vascular malformations or stroke). Mean IQ scores were almost two standard deviations below normal (i.e., ∼70) in the Rasmussen and the vascular malformation/stroke groups but considerably lower (i.e., more than five standard deviations below normal) in the group with cortical dysplasia. These results are broadly consistent with those reported in our series of 33 children who underwent hemispherectomy and were divided into developmental, acquired, and progressive etiologic groups.¹¹ Results from these two recent studies suggest that both cognitive and seizure outcomes are related to the underlying pathology, with most favorable outcomes occurring in those with acquired and progressive etiologies as compared to developmental etiologies (such as cortical dysplasia, hemimegalencephaly, etc.). Based on the comparison between preoperative versus postoperative IQ scores, the decline and/or arrest of intellectual function occurred well before presentation for surgery, with both studies indicating improvements in long-term follow-up in only a small percentage of cases.^11,51

The relationship between etiology and hemispheric side of removal was pursued further in the report by Pulsifer et al. Whereas no differences were found between the means of the left and right hemispherectomized groups within the etiologic category of cortical dysplasia or vascular abnormality on any of the cognitive measures, including intelligence, significant effects of side of removal were indicated in the Rasmussen encephalitis groups.⁵¹ The right hemispherectomy group obtained significantly higher intelligence and language scores than the left hemispherectomy group. The absence of an effect of hemispheric side of removal in the group with cortical dysplasia is consistent with the ontogenetic specialization model, and the notion that in the face of extensive early brain abnormality hemispheric specialization (i.e., nonredundancy of function) is sacrificed to enable the development of cognitive abilities at a basic level.

The question of whether intellectual outcome after hemispherectomy is related not only to etiology and hemispheric side of removal but also to age at onset of pathology deserves attention. In contrast to adult studies,^42,74 previous reports based on large cohorts of pediatric patients with unilateral lesions, with or without epilepsy,^3,18,68,69 failed to reveal a significant difference between verbal and nonverbal intellectual outcome as a function of hemispheric side of lesion. However, in those with late-acquired pathology (i.e., >5 years) a pattern of IQ scores at least qualitatively similar to the one reported in adults has been found (i.e., verbal IQ [VIQ] < performance IQ [PIQ] after late left hemispheric lesions; PIQ < VIQ after late right hemispheric lesions).⁶⁹ It is not known whether this age-at-injury–dependent and side-of-injury–dependent pattern documented in pediatric patients with unilateral lesions is also found in hemispherectomized patients.

We examined the interaction between age at onset of pathology and surgical side of removal on IQ type in a subset (N = 36) of hemispherectomized patients drawn from the series operated at Great Ormond Street Hospital¹¹ and King’s College Hospital.^67,68 Admittedly, our postoperative hemispherectomized cohort is not yet large enough to permit assessment of the effects of three variables (i.e., hemispheric side of removal, age at onset of pathology, and IQ type) in the same analysis. Nevertheless, despite the small groups that this comparison inevitably entails, it is informative to examine the pattern of results in patients who have undergone hemispherectomy compared to patients who have not received surgery but have hemiplegia due to unilateral left- or right-sided lesions and seizures grouped by age at onset of pathology.⁵⁸

Hemispherectomized patients were selected for this study if they could undergo formal intelligence testing and if they had undergone surgery at least 2 years prior to testing. In each case, both verbal and nonverbal IQs were obtained using the age-appropriate Wechsler Scales of Intelligence. The left- and right-hemispherectomized patients were each divided into three groups on the basis of age at onset of pathology (congenital; early, ≤5 years; late, >5 years; see Table 1). Therefore we had six subgroups: left congenital (LC), left early (LE), left late (LL), right congenital (RC), right early (RE), and right late (RL). Relevant details regarding the etiology of the seizure disorder leading to hemispherectomy are presented in Table 2. As indicated in Table 1, the full-scale IQs of the six hemispherectomized groups ranged from 53 (RC group) to 73 (RL group). These mean IQ scores are between two to three standard deviations below normal and are consistent with results reported in previous studies.^51,72 The pronounced restriction in overall intellectual ability is the most striking cost of severe intractable epilepsy leading to hemi- spherectomy.

Prior to statistical analysis, the two variables of IQ type were reduced to one by subtracting verbal IQ from nonverbal (performance IQ), thus yielding a discrepancy score. This reduced the number of analyses carried out on a relatively small sample. A discrepancy score above zero indicated a verbal IQ score that was higher than nonverbal IQ, whereas a discrepancy score below zero signified the opposite. FIGURE 1 presents the mean discrepancy scores for each of the six hemispherectomy groups. As indicated, five groups show positive mean discrepancy scores and the LL group shows a strongly negative mean discrepancy score.