Citation
The Effects of valproate monotherapy with carnitine or carnitine placebo on attention and memory in a pediatric idiopathic epilepsy sample

Material Information

Title:
The Effects of valproate monotherapy with carnitine or carnitine placebo on attention and memory in a pediatric idiopathic epilepsy sample
Creator:
Booth-Jones, Margaret, 1965-
Publication Date:
Language:
English
Physical Description:
iv, 125 leaves : ; 29 cm.

Subjects

Subjects / Keywords:
Anticonvulsants ( jstor )
Control groups ( jstor )
Dosage ( jstor )
Epilepsy ( jstor )
Medications ( jstor )
Memory ( jstor )
Migraine ( jstor )
Pediatrics ( jstor )
Seizures ( jstor )
Side effects ( jstor )
Attention -- Adolescent ( mesh )
Attention -- Child ( mesh )
Attention -- Infant ( mesh )
Attention -- drug effects ( mesh )
Carnitine -- pharmacology ( mesh )
Carnitine -- therapeutic use ( mesh )
Department of Clinical and Health Psychology thesis Ph.D ( mesh )
Dissertations, Academic -- College of Health Related Professions -- Department of Clinical and Health Psychology -- UF ( mesh )
Epilepsy -- Adolescent ( mesh )
Epilepsy -- Child ( mesh )
Epilepsy -- Infant ( mesh )
Epilepsy -- drug therapy ( mesh )
Memory -- Adolescent ( mesh )
Memory -- Child ( mesh )
Memory -- Infant ( mesh )
Memory -- drug effects ( mesh )
Research ( mesh )
Valproic Acid -- pharmacology ( mesh )
Valproic Acid -- therapeutic use ( mesh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1995.
Bibliography:
Bibliography: leaves 115-124.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Margaret Booth-Jones.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
028475105 ( ALEPH )
50189689 ( OCLC )

Downloads

This item has the following downloads:


Full Text











THE EFFECTS OF VALPROATE MONOTHERAPY WITH CARNITINE OR
CARNITINE PLACEBO ON ATTENTION AND MEMORY IN A
PEDIATRIC IDIOPATHIC EPILEPSY SAMPLE
















By

MARGARET BOOTH-JONES


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1995














ACKNOWLEDGEMENTS

Numerous people were involved in this project. The subjects and their families, and the Departments of Pediatric Neurology, Pharmacy, Biostatistics, and Clinical and Health Psychology were integral parts. To name every member of this extensive team would be impossible; however, the author will not forget the commitment to this project.

The author would like to extend her deepest gratitude to Eileen B. Fennell, Ph.D., her chair and mentor, for her inexhaustible knowledge and support. The past six years of warmth and encouragement have profoundly influenced the author and will never be forgotten. Sincere thanks goes to the members of my committee, Russell Bauer, Cynthia Belar, James Rodrigue, and Bernard Maria, without whose support and guidance this study could not have been a success.

The author would like to acknowledge the support and prodding of her parents, Pete and Carolyn Booth. Special appreciation goes to the author's sister, Veronica, who provided very much needed distraction and joy. Finally, the author extends her warmest thanks to her husband, Graeme Jones, who patiently understood and listened to the trials and tribulations of completing this project.

ii












TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS. ....................................... ii

ABSTRACT................................................ iv

CHAPTERS

1 INTRODUCTION.......................... ........ 1

Epilepsy: General Description................... 2
Migraine: Description and Proposed Mechanism.... 15 Relationship Between Epilepsy and Migraine...... 19 Cognitive and Behavioral Deficits............... 21
Antiepileptic Therapy and Cognitive Function.... 27 Valproate: Pharmacology and Action.............. 31
Valproate: Neuropsychological Effects........... 53
Rationale for Further Study............. ... 62
Specific Aims................................... 63

2 METHODS......................................... 64

Subjects........................................ 67
Materials and Apparatus......................... 63
Procedures and Design............................ 76
Hypotheses...................................... 78

3 RESULTS......................................... 82

Hypothesis One .................................. 83
Hypothesis Two.................................. 84
Hypothesis Three.................................. 86

4 DISCUSSION....... ..................... 93

Interpretation of Findings......................102
Attrition and Non-compliance..................109
Limitations of Study............................110
Conclusions.....................................112

REFERENCES..............................................115

BIOGRAPHICAL SKETCH.....................................125


iii












Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE EFFECTS OF VALPROATE MONOTHERAPY WITH CARNITINE OR
CARNITINE PLACEBO ON ATTENTION AND MEMORY IN A PEDIATRIC
IDIOPATHIC EPILEPSY SAMPLE

By

MARGARET BOOTH-JONES

May 1995

Chair: Eileen B. Fennell
Major Department: Clinical and Health Psychology

Twelve children ranging in age from 6 to 16 years, of

average intelligence, diagnosed with idiopathic epilepsy and on valproate monotherapy were the subjects for this study. The effect of the valproate on memory and attention was assessed at three intervals over a three-month period, with plasma levels of valproate monitored and maintained throughout the study. After four weeks of the valproate monotherapy, half of the children received carnitine and half received carnitine placebo in a double-blind paradigm. All subjects remained on their regimen for eight weeks and were re-evaluated. The performance of the epileptic children was compared with that of 14 children with migraine also prescribed valproate with carnitine or carnitine placebo, and with that of 20 age-matched normal controls.


iii







At baseline, the epileptic children demonstrated mild

impairment with attention and verbal memory as compared with the migraine and normal control groups. A similar pattern of comparative test performance was observed at the 4-week and 12-week intervals. There were no appreciable cognitive changes in the subgroup of children taking carnitine. Most children were in the therapeutic range of valproate and experienced excellent seizure control and minimal negative side effects. A high rate of attrition occurred in the migraine sample (50 percent), greatly limiting further comparison; however, for the most part they maintained an intermediate position on neuropsychological measures between the epileptic and control groups. In this study with this sample, valproate monotherapy did not lead to change in performance on tasks assessing memory and attention.


iv

















CHAPTER 1

INTRODUCTION

Epilepsy is a common neurological disorder which has been described in the literature for centuries with attributions to mystical forces, deities, and demons for its etiology. In 400 B.C., Hippocrates argued that epilepsy was some form of brain disorder (Bennett, 1992). This line of thinking was not supported until the work of John Hughlings Jackson in the 1800s when the acceptance of epilepsy as a disease process emerged. The differing seizure types were elucidated by several researchers in the late nineteenth century and early twentieth century with great assistance from the development of the electroencephalogram (EEG) by Hans Berger in 1929 (Kolb & Wishaw, 1990).

For well over one hundred years, cognitive

abnormalities were documented in the epilepsy population and moral and intellectual deterioration were thought to be inevitable (Trimble, 1987). Emil Kraeplin labeled epilepsy as a psychiatric disorder and, though he wrongly diagnosed epilepsy as a form of insanity, he proposed psychiatric

1








2

treatment for sufferers (Bennett, 1992). Henri Gastaut used EEG to further study seizures and probable neuroanatomical regions of causation (Bennett, 1992). Idiosyncratic personality, and cognitive and behavioral characteristics of adult and child epileptics have been observed and reported throughout the decades of epilepsy research. Pharmacological intervention, which has been utilized since the early twentieth century, has been beneficial to this population for the most part, but not without consequence. Description of the clinical presentation of epileptics on and off antiepileptic medications is an area of intense and essential investigation.

Epilepsy: General Description

Epilepsy is a chronic convulsive disorder characterized by persistent, recurrent seizures associated with acute and/or chronic psychological changes and a pathological EEG (Rapin, 1982; Kolb & Wishaw, 1990; Engel, 1989). Seizures in the human population are relatively common as 1 in 20 will experience some form of seizure in their lifetime (Rapin, 1982). However, recurrent seizures warranting the diagnosis of epilepsy are less common, affecting an estimated 1 in 200 people (Fenichel,1988). Epilepsy usually develops before adulthood with an estimated 90 percent presenting before age 20 (Spreen et al., 1984). Up to one million children in the United States alone have some form








3


of seizure experience or disorder (Hartlage & Hartlage, 1989). An estimated fifteen percent of occasional convulsions develop into epilepsy (Spreen et al., 1984). The specific functional and psychological characteristics of epilepsy appear related to age of onset and duration of the disorder, type of seizure and EEG abnormality, frequency of the seizures, and etiology (Trimble, 1990; Rapin, 1982; Engel, 1989). Childhood epilepsy, defined by onset under age 15 years, is generally considered different from adult onset epilepsy in etiology, types of seizures, and frequency of seizures (Spreen et al., 1984). A genetic continuum for susceptibility to seizure disorder has been proposed with a significantly greater chance of a seizure occurrence when a close relative has a seizure disorder (Spreen et al., 1984; Fenichel, 1988; Engel, 1989). Epilepsy: Proposed Mechanism

An epileptic seizure is the result of particular

changes in the membrane conductance and neurotransmitter activity in the brain, however, the specific physiological and chemical changes continue to be under investigation (Rapin, 1982; Bennett, 1992; Eadie & Tyrer, 1989). In a normal brain, the excitatory and inhibitory synaptic responses and activity of neurons are well modulated; however, in the epileptic the balance of this complex system is disturbed, forming an epileptogenic region in the gray







4


matter. At the beginning of a seizure, a large hyperpolarization occurs and spreads to other areas of the brain. With epileptogenesis, it has been theorized that either some of the neuronal cells are discharging more readily as a result of membrane changes or there is failure of neurotransmitters to inhibit cellular discharge (Rapin, 1982; Eadie & Tyrer, 1989). When the epileptogenic region fails to be inhibited, there is rapid recruitment of excitability of neighboring cells which increase in excitability leading to seizure presentation. The area of seizure activity may develop with a structural or cellular abnormality resulting in seizure susceptibility, and subsequent seizure activity can further damage the area. Site of origin can be cortical or subcortical and seizure activity can remain relatively focal or involve widespread areas of the brain (Rapin, 1982; Eadie & Tyrer, 1989; Engel, 1989).

The type of electrochemical activity, its location, and its pattern of spread define the type of seizure. Briefly, a partial seizure is characterized by slower, localized recruitment of synchronously discharging neurons (Rapin, 1982; Engel, 1989). With generalized tonic-clonic seizures, the tonic phase is characterized by maximum excitation and the clonic phase is the reassertion of inhibition. (Rapin, 1982; Engel, 1989). Without the








5


inhibition phase, status epilepticus occurs which can lead to cell death and, potentially, death of the patient. With absence seizures, inhibition predominates halting the epileptic's activity.

Kindling has been used to explain several phenomena in epilepsy. Kindling is defined as a permanent alteration of postsynaptic membranes of neighboring neurons and efferents of epileptogenic foci resulting from subclinical to clinical seizure activity which leaves the neuron more susceptible to future seizure activity (Rapin, 1982; Engel, 1989). Kindling has been suggested as the mechanism to explain development of mirror foci and secondary generalization of seizure activity.

Use of EEG has allowed illustration of the electrical activity in the epileptic's brain during and between seizures. The interictal spike is the marker for epileptogenesis. There is a "depolarization shift" of a given neuronal region leading to synchronous discharge of neighboring neuronal regions which, if severe enough, will lead to the seizure activity (Rapin, 1982; Engel, 1989). To confound the diagnostic issue, EEG abnormalities are sometimes not observed in patients presenting with seizures and EEG abnormalities of various types, including spike waves, have been observed in patients who do not manifest overt behavioral signs of seizure activity.








6


Epilepsy is not a disease, but rather it is a symptom (seizures) of recurring brain dysfunction which can be caused by trauma, toxicity, drugs, fever, or congenital anomaly. Epilepsy is diagnosed by type and frequency of seizures, EEG abnormality, ruling out of other external causes (e.g., transient metabolic toxicity), and presence of particular behavior patterns associated with epilepsy.

Epilepsy is usually treated with antiepileptic drug

therapy, and the type of drug prescribed depends on seizure type and physician preference. Pharmacological control may be achieved with one antiepileptic drug (monotherapy) or a combination of antiepileptic drugs (polytherapy), and drug type and dosage may be changed throughout the course of treatment in response to seizure control and side effects. Seizure Classification

The classification of seizure type within the epilepsy population has changed in specificity and name in the past decade. The International Classification of Epileptic Seizures provides three broad classes: partial seizures, generalized seizures, and unclassified epileptic seizures (Commission on Classification and Terminology of the International League Against Epilepsy, 1981). Some seizures observed in children do not fit neatly into one of the common diagnostic categories and are labeled unclassified, and some children present a combination of seizures types.








7


To further characterize the epileptic condition, seizures leading to alteration of consciousness are labeled 'complex' while those that manifest in motor or sensory changes only, without alteration of consciousness, are 'simple'. See Table 1-1.

Partial seizures are localized in origin and most often include specific motor, sensory, or psychic changes (Fenichel, 1988; Rapin, 1982; Bennett, 1992). Partial seizures are sub-classified as simple partial, complex partial, and partial with secondary generalization. In simple partial seizures, the seizure manifestation is restricted to motor (e.g., limb movement), sensory (e.g., tingling or numbness), or psychic phenomena (e.g., disruption of ongoing activity). These three categories of partial seizures are further delineated by the presence and progression of the above mentioned symptoms. Previously called psychomotor seizures, partial seizures with psychic alterations are characterized by automatisms such as lip smacking and blinking. Strong emotional sensations and sensory hallucination have been reported with some partial seizures. Interictal EEG for patients with simple partial seizures most commonly demonstrates focal spike-and-wave discharges in the involved cortical region, but these are not necessarily consistently present (Engel, 1989). Ictal EEG activity with simple partial seizures varies greatly in








8


wave type, localization, and spreading, and there is no single characteristic pattern (Engel, 1989).

The most common partial seizure disorder in children is Benign Rolandic Epilepsy with the age of onset between 3 and 13 years and the greatest rate between 5 and 10 years of age. This disorder often spontaneously resolves in the teen years (Fenichel, 1988). Most children who have this disorder experience relatively few seizures, most of which occur during sleep. The seizures usually last between 1 and

2 minutes, and when the child wakens he or she may experience paresthesias around the mouth, facial twitching, and speech cessation. There is no loss of consciousness unless the seizure generalizes. The interictal EEG demonstrates unilateral or bilateral high voltage spike discharges in the central or centrotemporal region.

Benign Occipital Epilepsy is a relatively uncommon partial seizure disorder observed in children. It is a familial disorder with onset usually occurring in children below 9 years of age (Fenichel, 1988; Engel, 1989). The first symptoms include visual hallucination, transient blindness, hemianopia, and visual illusions. After the paroxysmal episode, symptoms can include a migraine-like headache, nausea, and possible vomiting. This partial seizure can generalize as well. The characteristic EEG is a unilateral or bilateral, high amplitude, spike and wave







9


discharge in the occipital region unilaterally or bilaterally. The EEG activity can spread during seizure activity to the central and temporal regions (Engel, 1989). This form of epilepsy is often difficult to differentiate from basilar migraine (Fenichel, 1988).

Complex partial seizures, focal seizures that spread and lead to alterations of consciousness, are more frequently observed in the older child and adult populations than in younger children. The etiology for complex partial seizures is heterogeneous and can include trauma, neoplasm, and disease (Fenichel, 1988; Engel, 1989; Rapin, 1982). These seizures originate in cortical regions, most often the temporal lobe, but epileptogenesis in the frontal and parietal lobes has been observed (Fenichel, 1989; Bennett, 1992). Many authors interchange the terms complex partial seizures and temporal lobe epilepsy; however, complex partial epilepsy may not necessarily originate in the temporal lobe, but rather affect that brain region or neighboring limbic structures (Eadie & Tyrer, 1989).

Complex partial seizures can arise from simple partial seizures or present independently of simple partial seizures, and many patients experience both types of seizures. The complex partial seizure can occur spontaneously or be associated with sleep transitions, and the frequency can vary from one or more per day to one per








10


year. The duration of complex partial seizures is usually no greater than two minutes.

The initial symptom for a complex partial seizure is

often the same as for the simple seizure in a given patient, but many report additional sensations including a general, nondescript unpleasant feeling, auditory hallucinations, or abdominal discomfort. Staring, automatisms (e.g., grimacing, finger movements), tonic extension of the limbs, decrease in postural tone, and resistance to restraint have been observed during complex partial seizures. A postictal confusion, amnesia for the event, anterograde amnesia lasting minutes to hours, and post-seizure lethargy occur in most patients (Eadie & Tyrer, 1989; Fenichel, 1988). The EEG pattern for complex partial seizures is usually a single spike or slow wave focus in the temporal lobe or frontal lobe, or multifocal throughout cortical regions (Fenichel, 1988). During ictus there is an EEG progression from single spike discharges in the cortical area of involvement to spike slow wave complexes involving a greater cortical area followed by slow waves with varying amplitude (Fenichel, 1988). Interictal EEGs of patients with complex partial seizures most often demonstrate unilateral or bilateral spikes in the anterior temporal lobe with the similar variation seen in simple partial epilepsy.








11


Generalized seizures, defined as non-localizing seizures, are classified as absence, atypical absence, myoclonic, clonic, tonic, tonic-clonic, and atonic seizures. The EEG demonstrates bilateral involvement, with epileptogenic origins in both cortical and subcortical regions. Extensive involvement of both hemispheres occurs.

Absence epilepsy is a genetic disorder which presents

only in children and rarely persists into adulthood. Age of onset is most commonly between 4 and 8 years of age, and more girls than boys are affected in a 60 percent to 40 percent ratio (Fenichel, 1988; Engel, 1989). Absence epilepsy is rarely secondary to neurologic disease or trauma, and children with absence epilepsy are usually healthy otherwise. A typical absence seizure lasts a few seconds to one minute and the frequency varies from occasional seizures to hundreds per day. During an absence attack all ongoing activity ceases and staring and possibly rhythmic eyelid movement occur. When the seizure ends, activity recommences with no postictal confusion. Additional symptoms can include myoclonus, increases or decreases in postural tone, and automatic movements such as picking at clothes or turning of the head.

The characteristic ictal EEG pattern for absence

seizures is the 3 Hz spike-wave per second bilaterally which is synchronous and symmetric with the greatest amplitude in






12


the frontal and central regions of the brain. Interictal EEG patterns appear normal in most cases. Absence seizures can be induced with hyperventilation and are more frequent during transition of sleep and wake. Nearly half of all children with absence epilepsy have experienced at least one generalized tonic-clonic seizure (Fenichel, 1988).

Myoclonic epilepsy is a form of epilepsy that has numerous sub-types primarily observed with infants, degenerative neurologic syndromes, and following toxic and anoxic events (Rapin, 1982; Engel, 1989; Fenichel, 1988). Myoclonus is involuntary muscle contractions that are brief and repetitive. One type of myoclonic epilepsy is myoclonic absence which has an age of onset between 2 and 12 years. A large proportion, upwards of forty percent, of patients presenting with myoclonic absence have mental impairment prior to seizure onset (Fenichel, 1988). With this form of epilepsy, there is a combination of absence seizures and severe myoclonus in the limbs in the patient. No loss of consciousness occurs. The EEG is most often characteristic of absence with a 3 Hz spike-wave pattern (Fenichel, 1988; Engel, 1989).

Generalized tonic-clonic seizures are the most common form of pediatric seizure disorder with onset typically after the neonatal period. Generalized motor seizures include tonic, clonic, tonic-clonic and atonic phases








13


TABLE 1-1 CLASSIFICATION OF SEIZURE


1. Generalized seizures (bilaterally symmetrical and without
local onset)
A. Tonic or clonic or tonic-clonic (grand mal seizures)
B. Absence (petit mal)
1. Simple (loss of consciousness only)
2. complex with brief tonic, clonic, or automatic
movements
C. Lennox-Gastaut syndrome
D. Juvenile myoclonic epilepsy
E. Infantile spasms (West syndrome)
F. Atonic seizures (sometimes w/ myoclonic jerks)

2. Partial (focal) seizures
A. Simple partial seizures (consciousness not impaired)
1. Motor (including Jacksonian)
2. Sensory or somatosensory
3. Autonomic
4. Psychic
B. Complex partial seizures (w/impairment of
consciousness). Includes:cognitive, affective,
psychosensory, and psychomotor symptomatology
i.e. "temporal lobe" or "psychomotor seizures"
C. Partial seizures (simple or complex) with secondary
generalization

3. Unilateral or predominantly unilateral seizures (tonic,
clonic, or tonic-clonic, with or without impaired
consciousness)

4. Unclassifiable (due to inadequate data)


Source: R.D. Adams and M. Victor (1989). Principles of Neurology (Fourth Edition). Page 250.







14


depending on the progression of motor activity. The first sign of a tonic-clonic seizure is a sudden loss of consciousness often preceded by a sharp cry. The tonic phase is characterized by convulsions affecting the whole body. Brief periods of clonus, jerking caused by relaxation of the muscles, begin to rhythmically interrupt the tonic spasm with increasingly greater duration until the seizure ends. During the seizure, the patient's eyes roll backward, breathing is rapid and deep resulting in increased saliva in the mouth (e.g., frothing), and bowel and bladder incontinence may occur. The increased saliva and changes in the muscle tone of the airway can lead to obstruction of respiration. Tonic-clonic seizures typically last one minute or less, and may be followed by a tonic phase (Engel, 1989). In the postictal phase, muscle flaccidity, disorientation, deep sleep, and headache most often occur.

The ictal EEG of the tonic phase is associated with a

generation of widespread low-voltage rapid activity referred to as the "recruitment rhythm" (Engel, 1989). This buildup most often is followed by generalized high amplitude polyspike or spike-and-wave phenomena. The clonic phase is represented on EEG by widespread suppression of activity. These patterns alternate throughout the seizure and either dissipate with decreasing amplitude and rapidity or end abruptly (Engel, 1989; Fenichel, 1988). The postictal EEG







15


of the generalized tonic-clonic seizure patient is characterized by generalized slowing. Partial seizures with secondary generalization to tonic-clonic seizures may show focal activity on the postictal EEG in the area of the focal involvement (Engel, 1989; Fenichel, 1988).

Migraine: Description and Proposed Mechanism

Migraine is classified as a paroxysmal neurological

disorder involving alterations in cerebral blood flow that may occur alone or secondary to another disease or disorder. Migraine is reportedly the most common neurological disorder occurring in 5 to 20 percent of the population and the most common cause of severe headache in children (Kolb & Wishaw, 1990; Fenichel, 1988). Migraine is most often a hereditary disorder, with a prevalence of 2.5 percent for children under 7 years of age, 5 percent for age 7 to pubescent children, 5 percent of post-pubescent males, and 10 percent of post-pubescent females (Fenichel, 1988). Prevalence by gender appears nearly equal in younger children, but females with migraine outnumber males 3 to 2 in the range from age 7 to pubescence (Fenichel, 1988). Migraine occurrence can be triggered by stress, exercise, head trauma, allergy, diet and estrogen levels (Fenichel, 1988; Sacks, 1985).

The exact mechanism of migraine is not known. Vascular changes have been reported during and between migraine attacks (Wolff, 1985). Between migraines, asymmetric







16

vascular reactivity has been observed. There is an apparent decrease in cerebral blood flow during the prodromal phase and an increase in extracranial blood flow during the migraine. The most common hypothesis involves vasoconstriction beginning in one or more cerebral arteries, usually in the posterior region of the brain (Sacks, 1985; Wolff, 1985). This decrease in blood flow progresses anteriorly, though not along a particular pathway, and usually affects both hemispheres. The severe pain of migraine begins with the onset of vasodilation.

Differences in platelet aggregation have been observed in association with migraine (Wolff, 1985). Specifically, platelet aggregation has been observed to be more rapid in migraineurs than normal controls. More platelet aggregation has been observed in the prodromal phase of the migraine with a decrease in the headache phase. Serotonin in the plasma, which is related to platelet concentration, appears to increase before the migraine attack and decrease during the headache phase. Serotonin associated with platelet concentration can lead to constriction of arteries or large vessels and dilation of arterioles and capillaries (Theisler, 1990). Plasma serotonin levels may reflect changes in central nervous system serotonin which affect two pathways associated with migraine.







17


Migraine can vary in presentation, intensity,

frequency, and duration and can occur unilaterally or bilaterally. Migraine is frequently associated with anorexia, nausea and vomiting, and neurological and mood disturbances. Migraine is classified as: classic, common, cluster headaches, hemiplegic, and ophthalmoplegic. Classic migraine affects 12 percent of the migraine population and has a distinctive aura, most likely related to the ischemia in the occipital cortex (Fenichel, 1988). The classic migraine occurs in two phases, the first of which involves the progression of excitation of brain function from posterior brain regions to anterior brain regions with decreases of regional cerebral blood flow and transitory neurologic disturbances (Fenichel, 1988; Wolff, 1985; Sacks, 1985). Common neurologic symptoms in this first phase, also called the prodromal phase, are visual abnormalities, dysthesias of the mouth area and limbs, and occasionally motor changes. The second phase is characterized by headache and nausea and sometimes vomiting. The frequency of classic migraines is most often several times per month with a duration of several hours (Fenichel, 1988; Wolff, 1987).

Common migraine does not have two distinct phases and an aura is not associated with this classification of migraine. Visual changes may occur, but the more frequent








18

symptoms include changes in personality, dizziness, general malaise, and nausea and vomiting in conjunction with headache (Fenichel, 1988; Wolff, 1987).

There is a high familial incidence for cluster

headaches and they are more prevalent in males than females with onset most often by age 10 years (Fenichel, 1988). The headache most often originates behind one eye and extends along the same hemicranium. These tend to last 30 minutes to 2 hours (Fenichel, 1988; Wolff, 1987).

Hemiplegic and ophthalmoplegic migraines are classified as complicated migraines, defined as presentation of focal motor deficits during a classic migraine (Fenichel, 1988). Hemiplegic migraines are characterized by hemiparesis followed by a throbbing frontotemporal headache located contralateral to the hemiparesis. Nausea and vomiting usually occur. Ophthalmoplegic migraines present with unilateral eye pain and ipsilateral ocular motor palsy usually preceded by ptosis which can last for days or weeks (Fenichel, 1988).

There is a higher rate of EEG abnormality in patients with migraine than normal population, but not all patients with migraine have EEG abnormalities (Fenichel, 1988; Wolff, 1987). Unilateral slow wave focus in the temporal lobe has been observed in patients with classic migraine. Between migraine attacks, EEG is usually normal (Fenichel, 1988).







19


Relationship Between Epilepsy and Migraine

The relationship between migraine and epilepsy has been the focus of numerous medical arguments and the subject of relatively little research. Some suggest that the two may be related and others believe the two are completely distinct. Regardless, there appear to be several common features which have posed difficulty for diagnosticians in certain cases. Fenichel (1988) lists the following four features which most strongly suggest a relationship between epilepsy and migraine. (1) Both disorders can have a strong familial component, are paroxysmal and are associated with transient neurologic disturbances. (2) There is an increased incidence of migraine in epileptics and an increased incidence of epilepsy in migraineurs. (3) Headaches are associated with some forms of epilepsy and in most forms of migraine. (4) Abnormal EEG activity has been observed in both populations and some specific EEG patterns are common to both.

Additionally, both neurologic disorders can occur in individuals with no apparent cause such as a lesion or traumatic event or be part of the neurologic sequela of head trauma. Occurrence of epileptic seizures and migraines can be spontaneous or be triggered by external factors and internal factors. Both epileptic and migrainous events occur in stages during which time the individual experiences







20


changes in mental status with or without physical changes. For example, the epileptic and migraineur can experience an aura, alterations in consciousness, changes in postural tone, motor changes, hallucinations, changes in mood and changes in higher cortical functions (Sacks, 1985).

There are differences between these two disorders. On the whole, the auras experienced by the migraine population are more visual and often include the occurrence of scotoma. Convulsions and loss of consciousness are rare in migraine. The intensity of alterations of mood and higher cortical function is generally greater with epilepsy than migraine.

Focal cerebral symptoms and EEG abnormalities have been observed in children with migraine with and without seizure activity which has lead to some difficulty with differential diagnosis (Fenichel, 1988). The greatest similarity is reported between temporal lobe epilepsy and classic migraine especially in children (Engel, 1989). Migraine and epilepsy can occur in the same individual and are not mutually exclusive. Benign occipital epilepsy and benign rolandic epilepsy are associated with migraine. Basilar artery migraine can lead to alterations and loss of consciousness and seizure. Partial-complex seizures have been noted to be followed by vascular headaches. The relationship between epilepsy and migraine is tenuous at best; however, there is some overlap in the symptoms, neuroanatomy, and treatment.







21


Cognitive and Behavioral Deficits

Trimble (1987) stated that the relationship between abnormal cognition and epilepsy has been commented on for well over one hundred years with the consensus that moral and intellectual deterioration in the epileptic were inevitable. Today, there is general agreement in the field of epileptology that cognitive deficits are over-represented in children with epilepsy in comparison with neurologically healthy children (Trimble, 1990). These cognitive deficits have been attributed to gross structural abnormalities, subtle structural abnormalities, drug effects, interference from the actual seizure activity in the brain, social and emotional difficulties, and loss of time in school (Stores, 1971; Trimble, 1990). Risk factors for determining cognitive impairment include age of onset, duration of seizure disorder, seizure type, level of seizure control, medication, and related metabolic changes (Rapin, 1982; Trimble, 1990, Engel, 1989). For individual patients, some or all of these factors may play a role in their relative cognitive abilities.

The research examining cognition and pediatric epilepsy is often difficult to interpret and to generalize due to several crucial study design flaws (Stores, 1971). The groups of children studied have often been heterogenous covering wide age ranges, multiple seizure types, great








22

variance in the severity of seizures, numerous combinations of seizure medications, and wide ranges of intellectual and physical disabilities. Many of the early studies focused on intelligence quotients (IQs) as the sole measure of cognitive ability.

The prevalent messages from these studies indicate that, as a patient population, children with generalized epilepsy have attentional deficits, but are relatively free of memory impairment (Trimble, 1987). Involvement of subcortical structures and their connections with higher cortical regions have been implicated in the attentional difficulty. Children with partial epilepsy, specifically with temporal lobe involvement, demonstrate a higher rate of verbal memory deficits with left hemisphere foci and nonverbal memory disability with right hemisphere foci (Trimble, 1987).

Children with epilepsy have been observed to have a variety of school difficulties; learning and behavioral problems are over-represented in the pediatric epilepsy population (Trimble, 1990). There is a greater variability in the intellectual level (IQs) in this patient group (Seidenberg, 1986). Teachers have described these children as having poorer concentration, slower mental processing and reduced alertness as compared with non-epileptic children. Attentional problems, poor memory, poor social skills, and







23

emotional problems have been attributed to be both the cause and the result of epilepsy and interaction with medication. Reading difficulty and reading impairment are more frequently observed with these children (Trimble, 1990).

Hermann (1982) studied 50 children with epilepsy aged 8 to 12 in regular school placement. The children were administered the Luria Nebraska Neuropsychological Battery Children's Version by Golden (1981) comprised of 149 items in the following 11 scales: Motor, Rhythm, Tactile, Visual, Receptive Language, Expressive Speech, Writing, Reading, Arithmetic, Memory, and Intellectual Procedures. The Child Behavior Checklist by Achenbach (1986) was given to the parents to assess behavior. According to their test performance, the children where separated into a "good neuropsychological" and "poor neuropsychological" group along the median number of scaled scores below a T score of 60. Those in the "poor neuropsychological" group were associated with reports of increased psychopathology, more aggression, and decreased social competence. Interpretation and generalizability of these results are limited as this study did not describe the types of epilepsy, control for or even mention the antiepileptic medication, describe the pattern of deficits, or employ a normal control group.

Farwell, Dodrill, and Batzel (1985) studied 118

epileptic children between the ages of 6 and 15 with a







24


variety of seizure types to determine which aspects of the seizure disorder were associated with cognitive impairment. Children with IQs below 50 and serious behavioral disturbance were excluded. Though all children were receiving some form of antiepileptic medication, medication was not a variable in the study. The WISC-R and the ageappropriate version of the Halstead Reitan Battery were the measures employed. One hundred normal children acted as controls.

Farwell and her colleagues found significant

differences in Full Scale IQ with the mean for epileptics at 93.16 and the mean for normals at 100.00. Epileptic children with later onset of their seizure disorder had higher IQs. Longer duration of seizure disorder was associated with lower IQ scores. Children with better seizure control had a mean IQ of 99.56, and those with less seizure control had a mean IQ of 86.96, which is significantly lower. The children were grouped by seizure type: classic absence (n=8), both classic absence and generalized tonic clonic (n=8), generalized tonic clonic (n=31), partial seizures only (n=31), partial and generalized seizures (n=20), atypical absence (n=15), and minor motor seizures (n=5). Significant differences in IQ between these seizures groups were found with minor motor, IQ=70, and atypical absence, IQ=74. The remaining seizure








25


groups had mean IQs between 96 and 100 (average range).

The neuropsychological impairments were reported as a combined rating along the continuum of neuropsychological impairment rather than specific deficits. The rating was assigned by one of the examiners according to the median number of impairments with "good" rating above the median and "bad" rating below the median. The minor motor and atypical absence children demonstrated the most severe neuropsychological impairment. The other groups demonstrated relatively equivalent proportions of neuropsychological impairment, most of which was in the mild and very mild range. The classic absence group demonstrated the least amount of neuropsychological impairment.

This study provides strong evidence for IQ differences by seizure type. This study also provides convincing evidence against combining numerous seizure types when exploring specific aspects of childhood epilepsy. This study adds strength to the premise that significant variations among children with epilepsy exist.

Farwell et al. (1985) excluded children below an IQ of 50, therefore including many children with IQ scores below the average level which is problematic when comparing performance to a normal control group in attempts to discern specific cognitive patterns of an epileptic sample. The primary weakness of this study is the failure to report and








26


describe neuropsychological results in any form of detail. As a result, group differences in specific motor, visual, and language domains, for example, can not be discerned from the data provided. It is not surprising that the epileptic group demonstrated more impairment on the Halstead Reitan as intelligence and this measure are strongly correlated and the sample was biased to the below-average range of intelligence. Memory measures were not employed. Another limitation of this study is the lack of information about anticonvulsant medication.

Binnie, Channon, and Marston (1990) investigated the relationship of EEG activity and learning difficulties through literature review of studies in which neuropsychological tests (not specified) were administered during EEG recordings. All EEG readings were reported to be subclinical spike-wave activity. Generalized subclinical EEG activity was associated with impairments of attention; left focal activity was associated with poor performance on verbal tasks, and right focal activity was associated with poor performance on nonverbal tasks. Binnie et al. (1990) suggest that the EEG activity in the epileptic interferes with learning at an acute level and at a chronic level. Acutely, with loss or alteration of consciousness ictally with postictal confusion, seizures impair learning with disruptions in the processes of elaboration, storage, and







27

retrieval of information. Chronic focal EEG activity may go undetected and cause specific areas of brain damage and subsequent learning problems. Chronic subclinical EEG activity is theorized to reduce the total capacity for learning.

Binnie et al. (1990) further suggest that epileptics with gross lesions and generalized seizures have more impaired learning than epileptics with focal seizures, and deficits are greater for patients who demonstrate multiple seizures types. The five major findings of this study regarding the effects of subclinical EEG activity on learning suggest that this can: 1) result in the loss of immediate information; 2) disrupt consolidation; 3) cause permanent damage to neural tissue; 4) alter brain function; and 5) be reduced with antiepileptic drug treatment which also may damage neural tissue.

Antiepileptic Therapy and Cognitive Function

Most forms of epilepsy are treated with antiepileptic medications designed to reduce the number and severity of seizures; however, side effects have been reported with all antiepileptic drugs and the specific nature varies according to the drug, dosage, combination with other medication, and individual patient characteristics.

Phenobarbital, a barbiturate, has been widely used in the pediatric epilepsy population (over 50 years) with some







28

deleterious side effects including slower reaction times and difficulty sustaining attention (Trimble, 1987; Smith, 1991). Drowsiness and dysarthria have been reported, as well as hyperactivity and excitation in children (Engel, 1989; Fenichel, 1988). Phenobarbital is prescribed for tonic-clonic and simple partial seizures (Fenichel, 1988).

Phenytoin has been associated with a general decline in mental ability. Hirsutism, facial rashes, gum hypertrophy, and decreased attention span have been observed with phenytoin therapy (Fenichel, 1988). Drowsiness, dizziness, anorexia, hyperactivity, personality changes, and progressive encephalopathy have also been reported (Engel, 1989). Impairments in memory, motor speed, and mental speed have been observed as well (Trimble, 1987; Smith, 1991). More severe impairments are associated with higher serum levels of the drug. Phenytoin is most frequently used in patients with partial seizures, tonic-clonic seizures, and status epilepticus (Engel, 1989; Fenichel, 1988).

Primidone is used in the management of generalized

tonic clonic seizures, complex partial and partial seizures (Fenichel, 1988). Its central nervous system side effects include drowsiness, vertigo, ataxia, lethargy and some behavior changes. Rash, nausea and vomiting, hepatic disorder, and hematologic changes have been reported. Primidone has two active metabolites, one of which is







29

phenobarbital. Intolerable sedation from the first dose is one of the major adverse side effects (Smith, 1991).

Carbamazepine has been available for nearly thirty

years, and several studies have shown that many epileptics perform better on cognitive tests when they are switched from the other antiepileptic drugs or polytherapy to carbamazepine monotherapy, though whether the improvements are related to discontinuation of the sedative antiepileptic drugs or to the carbamazepine is not fully elucidated. Improvements in attention and speed have been observed (Trimble, 1987). Carbamazepine appears to be the drug of choice for partial and complex partial seizures. Deleterious side effects include depression, drowsiness, irritability, anorexia, and personality change (Engel, 1989). High levels can lead to ataxia, nystagmus, cognitive impairment, hyperactivity, and excessive salivation (Fenichel, 1988).

Succinimides and benzodiazepines are other classes of drugs used to control seizure disorder. Ethosuximide is used to treat absence seizures, and nausea and vomiting are the most common side effects. Diazepam and clonazepam are benzodiazepines used to treat status epilepticus, myoclonic seizures, and other forms of epilepsy (Fenichel, 1988). Sedation, changes in behavior, and memory dysfunction have been reported (Smith, 1991).










TABLE 1-2 PHYSICAL SIDE EFFECTS OF


Medication


1.
2.


COMMON ANTIEPILEPTIC MEDICATIONS


Physical Side Effects


Phenobarbital Phenytoin


3. Carbamazepine

4. Primidone

5. Ethosuximide

6. Valproate

7. Clonazepam
8. Diazepam.


Source: I. Rapin (1982). Page 192.


Drowsiness, gastric pain, rash Rash, gum hypertrophy, ataxia, hirsutism, coarsening of facial features, peripheral neuropathy Rash, nausea & vomiting (n&v), neutropenia,liver involvement Drowsiness, ataxia, behavior disorder, anemia, rash, n & v Sedation, dizziness, behavior changes, n & v, anorexia, Drowsiness, alopecia, liver enzyme increase, n & v Drowsiness, rash, n & v, anemia Drowsiness

Children with Brain Dysfunction.


TABLE 1-3
COGNITIVE SIDE EFFECTS OF COMMON ANTIEPILEPTIC MEDICATION


Medication


Cognitive Side Effects


1. Phenobarbital sedation; hyperactivity; impaired
vigilance and verbal learning 2. Phenytoin memory impairment; decreased
attention span; personality change; psychomotor slowing
3. Carbamazepine little effect; less sedation;
mild psychomotor slowing 4. Primidone similar to phenobarbital
5. Valproate no sedation; little effect
5. Clonazepam sedation
6. Diazepam sedation


Source: D. B. Smith (1991). Cognitive Effects of Antiepileptic Drugs. Advances in Neurology, Vol 55. edited by D. Smith, D. Treiman, and M. Trimble. Raven Press, Ltd. New York.


30







31

New antiepileptic drugs are being developed, including gabapentin, felbamate, lamotrigine, and vigabatrin (Fisher, 1993). Gabapentin is currently used predominantly in conjunction with other medications in the treatment of complex partial seizures and partial seizures with secondary generalization. Side effects include somnolence, fatigue, ataxia, and dizziness. Long-term effects on physical and cognitive function are not known. The mechanism of gabapentin is not known though it is similar in structure to the neurotransmitter GABA (Goa and Sorkin, 1993).

Felbamate is used in monotherapy in adults with partial seizures and in children with Lennox-Gastaut syndrome (myoclonic-astatic epilepsy associated with mental retardation) (Dodson, 1993). Adverse side effects include mild gastrointestinal complaints, weight loss, insomnia, and non-specific nervous system complaints. However, these side effects have been reported as mild to moderate in most cases (Graves, 1993). No studies involving the effects of these new antiepileptic drugs on cognition of children or adults are available at this time.

Valproate: Pharmacology and Action

Sodium valproate (sodium di-n-propylacetate) is one of the most recent additions to the antiepileptic drug arsenal and has been the subject of relatively few studies. It was identified as an anticonvulsant in 1963 and was allowed for







32


use with seizure patients in 1978. Some clinical trials suggest that it is most effective with absence, generalized motor, and myoclonic seizure patients and less effective with partial and partial complex seizure patients. However, in adults and older adolescents, it has shown significant promise for partial seizures with secondary generalization as well as being an adjunct to other antiepileptic drugs (Wilder, 1987).

Valproate is water soluble, unlike most other

antiepileptic drugs, and it is rapidly absorbed after oral administration and reaches a peak plasma level within two hours. Valproate has a half-life of six to ten hours and the efficacy range is 50 to 100 micrograms per milliliter. There is evidence of long lived metabolites of valproate which enhance seizure control and they are not easily detected with monitoring of valproate plasma levels (Gram et al., 1979).

The dose plasma level ratio of valproate, critical

information for assessing therapeutic range, has been the focus of several studies. Levy (1984) studied the leveldose ratio of valproate in monotherapy and polytherapy conditions. Levy reported previous findings that valproate does not have a characteristic level dose ratio leading to great inter-patient variability with the dose explaining less than 40 percent of the variability in the rate of







33

clearance. The plasma level of valproate increases with the dose ingested; however, the relationship appears to be curvilinear. This finding was replicated by Gram, Flachs, Wurtz-Jorgensen, Parnas, and Andersen (1979).

Similarly, Armijo, Herranz, Arteaga, and Valiente (1986) report a poor correlation coefficient for plasma concentration and dose ratios. The plasma level is dependent on dose absorbed, dosing interval, and rate of clearance. Valproate provides nearly 100 percent bioavailability to the patient with most absorption in less than 2 hours, though meals, enteric coating, and compliance are crucial factors for measuring absorption. The time intervals for dosing and sampling provide a great proportion of the inter-patient variance. Polytherapy shortens the half-life of the valproate. Levy (1984) reports that the age of the patient and polytherapy are the most important factors for determining the clearance rate. Children have a higher clearance than adults, with most of the variance occurring in patients below 7.5 years. Especially in children, valproate monotherapy is recommended for better monitoring of drug plasma level.

Johnston (1984) analyzed available research literature and reported three theoretical mechanisms for valproate's effect on the nervous system. The first proposed mechanism suggests that valproate increases the level of the







34


inhibitory neurotransmitter gamma-aminobutyric acid (GABA) in the brain. The second theory proposes that valproate potentiates the postsynaptic GABA response. The third mechanism described in this paper suggests that valproate has a direct effect on the neuronal membranes in the brain. Johnston repeatedly stated that none of these theories have been consistently supported by research and several studies using animal subjects directly refute crucial aspects of these proposed mechanisms. However, there does appear to be a relationship between valproate and the inhibitory neurotransmitter GABA, the specific nature of which is unknown.

Gamma-aminobutyric acid, an amino acid derived from glutamate, is a ubiquitous inhibitory neurotransmitter released at the inhibitory synapses. The full role of GABA is not known; however, GABA is purported to help maintain the electrical balance of the brain in conjunction with other neurotransmitters. Chemicals which increase GABA appear to increase neural inhibition and drugs that decrease GABA decrease neural inhibition, altering the neurochemical stability of the brain (Engel, 1989; Enna & Beutler, 1985; Eadie & Tyrer, 1989).

GABA-ergic neurons in the human brain are found in the granule cells of the olfactory bulb, amacrine cells of the retina, Purkinje and basket cells of the cerebellum, basket







35

cells of the hippocampus, many striatal cells, and numerous interneurons. The GABA-ergic neurons of the caudate nucleus and putamen (striatum) project to the substantia nigra and the globus pallidus (Noback, Strominger, & Demarest, 1991). GABA appears to have an integral role in the function of cortical/ subcortical pathway function, the extent of which is not fully elucidated.

GABA has been implicated in having a role in epilepsy, with the vast majority of data coming from animal research (Eadie & Tyrer, 1989, Enna & Beutler, 1985). By inducing seizure activity in mice, rats, cats, and baboons through biochemical, pharmacological, and neuroanatomical techniques, alterations in GABA and its related metabolites have been observed. The current findings suggest that lower levels of GABA are related to seizure activity. A study of GABA levels in the cerebral spinal fluid (CSF) in humans found lower levels in patients with generalized and secondarily generalized seizures in contrast to partial seizures and lower GABA levels in children with febrile convulsions when compared to normal children (Schmidt and

Loscher, 1981).

Generalizations from the animal and CSF studies to the activity of GABA in the epileptic brain warrant strong caution. For instance, GABA levels in the CSF may not directly reflect GABA levels in the brain. Drawing direct







36


correlations from animal models to human brain function is not possible due inherent anatomical and physiological differences. However, the relationship between GABA levels and seizure activity appears to have modest scientific foundation.

Several classes of antiepileptic medications appear to be indirect GABA agonists (Enna & Beutler, 1985). Barbiturates affect the chloride ion channels by prolonging the availability of GABA. Benzodiazepines bind at specific neuron sites and increase GABA receptor sensitivity. Valproate inhibits two enzymes in the metabolism of GABA, specifically GABA transaminase and succinic semialdehyde dehydrogenase, leading to prolonged availability of GABA. Several researchers posit that though certain classes of antiepileptic drugs control seizure activity presumably through the GABA system, the effects on the CNS are not limited or contained within the epileptogenic region. If the mode of seizure control is to alter GABA levels, these neurotransmitter changes should occur throughout the brain. Again, the extent of influence at the biological and at the cognitive levels requires further study.

Valproate has been observed, again through animal

models, to influence other neurotransmitters. Increases in the amino acid neurotransmitters taurine and glycine and a reduction of aspartate have been recorded following








37


administration of valproate (Eadie & Tyrer, 1989). Valproate appears to have no direct influence on serotonin levels; however, tryptophan, a precursor of serotonin appears to be increased by valproate (Hwang & Van Woert, 1979; MacMillan, 1979).

Valproate: Studies on Normals

An early study by Boxer, Herzberg, and Scott (1976)

assessed the behavior of 8 normal males using a single dose double-blind placebo controlled design, with the subjects receiving 400mg valproate, 60mg phenobarbital, the 2 drugs combined or placebo. The subjects were observed behaviorally and with EEG. Boxer et al. reported that the single dose of valproate tended to cause drowsiness which worsened with the addition of phenobarbital; the subjects on valproate and valproate plus phenobarbital tended to fall asleep if they were left alone. Neither standardized data nor empirical results were reported.

Thompson and Trimble (1981) studied 10 normal adults (mean age 26 years) in a double blind cross-over paradigm. The subjects were given valproate in a therapeutic range or a placebo. The following cognitive domains were assessed: verbal memory, nonverbal memory, concentration (Stroop), perceptual speed (minimum duration of exposure of targets flashed on a screen), simple decision making (40 yes/no trials for color and category), motor speed (finger







38

tapping), and mood. Verbal memory and nonverbal memory were assessed for immediate and delayed memory for 20 words and 20 pictures, respectively. Recognition memory was assessed with the addition of 20 distractor stimuli to each array. Mood was assessed with the Mood Adjective Check List by Lishman (1972).

Using multiple paired t-tests, the one significant finding between the valproate and placebo condition was slower decision making in the valproate condition. These researchers noted that improvements in cognitive ability would be difficult to observe in normal adults given their pre-drug normal range of intellectual functioning.

This study has very small sample size and the duration of drug exposure was relatively short (2 weeks). Given the large number of dependent variables and the small sample size, use of multiple paired t-tests is susceptible to a Type I statistical error.

Harding, Alford, and Powell (1985) studied 10 normal adult volunteers, mean age 24 years, in three valproate conditions: low dose, high dose, and after withdrawal of valproate. Simple reaction time ("yes" button to light flash), visual evoked potentials, and EEG's during sleep were assessed in all three conditions. A decrease in REM sleep and an increase in delta activity were observed in the EEG's after valproate withdrawal.







39

This study has a small sample size (n=10), and data for the simple reaction time task was obtained for only 6 of the 10 volunteers due to technical difficulties. Duration of the valproate was 4 days for the low dose and 14 days for the high dose, which may limit generalizability to the epileptic population who are receiving the medication over significantly longer periods of time.

Studies of valproate on normal children have not been

conducted for obvious ethical reasons. The results of these three studies involving normal adult subjects do strongly suggest that valproate has some physiological effect on the human nervous system, the full extent of which is not clear. The specific findings from brief drug trials of drowsiness, slower decision making, and sleep EEG changes are not directly generalizable to the pediatric epilepsy population due to age and developmental differences, confounding factors of illness, polytherapy, lesions and trauma, and length of exposure to valproate. Valproate: Side Effects

As with any drug, both the benefits and the risks to the intended patient need to be examined and weighed when deciding its usage. The Physician's Desk Reference (1992) reports that hepatic failure is the most significant and problematic risk with valproate monotherapy; however, overall incidence of hepatic failure is rare (Fenichel,







40

1988). Tremor and sedation have been observed. Additional central nervous system disturbances include ataxia, headache, nystagmus, diplopia, and incoordination. These symptoms are relatively infrequent and they can be reduced or eliminated with dose alteration.

Several studies have focussed on defining the side

effects of valproate. Herranz, Arteaga, and Armijo (1982) studied 88 pediatric epileptic patients receiving valproate monotherapy. The plasma level of the valproate was monitored and maintained in a therapeutic range defined as 40 to 90 micrograms per milliliter for the minimum range and 75 to 150 micrograms per milliliter for the maximum range. These researchers found the incidence of valproate toxicity to be lower than other antiepileptic drugs. Of the 88 patients, 42 percent reported some form of side effect through patient, parent, or physician report. However, with the use of a very detailed questionnaire, 80.7 percent reported the occurrence of side effects. The most frequent complaints were anorexia (8.0%), vomiting (10.2%), and sleep alterations including sleeping too much and difficulty sleeping (13.6%). Lassitude and drowsiness were associated with higher plasma levels, but the other side effects did not show a positive correlation with higher plasma levels. Neurological alterations, though relatively rare, included paresthesias (2.3%), tremor (1.1%), and ataxia (1.1%), and







41


were not associated with higher doses. Other rare side effects included polydipsia, polyuria, diaphoresis, and enuresis. To counteract the gastrointestinal side effects (e.g., anorexia, abdominal pain, vomiting, nausea, diarrhea, and constipation), the pharmaceutical preparation was altered without alteration of the dosage. Nine patients were discontinued from the valproate therapy due to negative side effects which did not resolve with pharmaceutical alteration. The use of the questionnaire detected significantly more behavioral side effects, with irritability and hyperactivity having the greatest frequency.

Herranz, Armijo, and Arteaga (1988) studied 392

epileptic and febrile convulsant patients under 15 years of age. They compared the rate of onset of side effects of phenobarbital, primidone, carbamazepine, and valproate monotherapy. Good tolerance for all drugs in the normal therapeutic range was reported. The rate of side effects for patients treated with valproate was 43 percent which is considerably lower than rates for phenobarbital and primidone, and equal to the rate for carbamazepine. As in their earlier study, digestive tract problems, such as nausea and vomiting, appetite changes, and abdominal pain, were the most commonly reported side effects for valproate, followed by behavioral changes including excitability,







42


depression, and hyperactivity. One child demonstrated ataxia and tremor.

Schmidt (1984) affirmed that many of the side effects associated with valproate therapy do not appear to be dose dependent. He reported the summary of 16 valproate trials involving 1140 epileptic patients. The side effects reported at the various sites were described as generally mild and transient, and appearing in the early stage of the valproate treatment. A significant increase in drowsiness was reported for patients receiving valproate in conjunction with phenobarbital. Valproate appears to increase the plasma concentration of phenobarbital. Schmidt reported a significant reduction of gastrointestinal complaints with the use of enteric coated valproate preparations. A small number of patients reported a reversible tremor (1.0%), increased coarseness of hair (1.0%), and blood disorders (0.1%).

Covanis, Gupta, and Jeavons (1982) studied the relative effectiveness and side effects of valproate on different seizure types. They reported 80 percent of the generalized seizure patients (absence, myoclonic, and primarily generalized) were seizure free. Forty-seven percent of the partial seizure patients were free of seizures over a three year period. Covanis et al. (1982) stated that valproate is most effective and least problematic to the epileptic







43


patient when given as monotherapy. Approximately three percent of these patients developed changes in their liver function determined by abnormal levels of certain hepatic enzymes, specifically aspartate, alkaline phosphatase, and bilirubin. They further stated that, though quite rare, severe hepatic and pancreatic disorders were reported.

Dreifuss, Santilli, Langer, Sweeney, Moline, and

Menander (1987) reported a retrospective study of all known hepatic fatalities secondary to valproate treatment from 1978 to 1984 in the United States. They analyzed all reports made to the drug company (Abbott Laboratories) which came from medical personnel, pharmacists, and patients. Thirty-seven of the 47 reports of liver fatalities were determined to be coincident to valproate therapy. The patients who died ranged in age from 5 months to 71 years, and all but one patient had significant medical disorders in addition to seizure disorder (e.g., other neurologic disease and congenital abnormalities). Five patients were receiving valproate monotherapy and 32 patients were receiving valproate as an adjunct to other antiepileptic therapy, most frequently phenytoin and phenobarbital. During 1978 and 1984, an estimated 400,000 people in the United States were taking valproate, 47 percent as monotherapy and 53 percent in polytherapy.







44

Dreifuss et al. (1987) described the mechanism of liver failure associated with valproate as liver necrosis caused by the production of aberrant, toxic metabolites in susceptible individuals. They distinguished dose related hepatotoxicity from idiosyncratic hepatotoxicity. They described the dose related hepatic changes as relatively harmless and characterized by an increased concentration of the enzyme serum transaminase. They stated that this abnormality is usually asymptomatic and it is reversible with drug discontinuation. The idiosyncratic hepatotoxicity was described as unpredictable and was observed in both the early and later stages of valproate therapy with symptoms of fever, rash, nausea, vomiting, edema, jaundice, and lethargy. They reported great variation in the liver abnormalities among the cases analyzed.

Risk factors associated with fatal liver toxicity from valproate therapy are age, with children under 2 years most vulnerable; pre-existing medical conditions; and valproate as an adjunct in polytherapy. The incidence of fatal liver failure associated with valproate therapy was estimated to be 1:500 for children under age 2 years on polytherapy; 1:7,000 for children under age 2 years on monotherapy; 1:12,000 for people above the age of 2 years on polytherapy; and, 1:37,000 for people over 2 years on monotherapy.







45


Carnitine Supplementation with Valproate Monotherapy

One of the physiological side effects of valproate

therapy is the reduction of free carnitine concentrations in the plasma. Carnitine is an essential nutrient found in the diet (meat and dairy products). It is essential for fatty acid metabolism (Stedman's Medical Dictionary, 1982) and mitochondrial function (Coulter, 1991). Some research indicates a possible relationship between low carnitine levels and liver disease caused by increased concentration of liver enzymes and fatty deposits in the liver cells; however, this evidence is limited and inconclusive and the abnormalities are at subclinical levels in most patients (Fenichel, 1988; Coulter, 1991; Beghi, Bissi, Codegoni, Trevisan, & Torri, 1990). Carnitine deficiency in children also presents as a primary inborn metabolic disorder, and carnitine supplementation is the accepted treatment (Fenichel, 1988; Coulter, 1991). Melegh, Kerner, Acsadi, Lakatos, and Sandor (1990) determined that the plasma level of valproate and overall seizure control were not affected by carnitine supplementation in a sample of 10 epileptic children.

Limited research has been conducted to discern the full effect of carnitine deficiency observed during valproate therapy. Stumpf, Parker, and Angelini (1985) described a three-step hypothesis for carnitine changes associated with







46


valproate therapy. (1) Carnitine acts as a buffer against toxic acyl CoA (a metabolic intermediate needed for fat oxidation); (2) with decreased levels of carnitine, toxic levels of acyl CoA occur which impair the citrate cycle, gluconeogenesis, the urea cycle, and fat oxidation; and (3) carnitine supplementation leads to greater excretion of the acyl CoA through urine excretion.

Polytherapy for epilepsy treatment, pre-existing liver damage, age below 5 years, poor nutrition and significant neurologic disabilities other than the epilepsy appear to increase the risk for problems thought to be related to decreased carnitine levels (Coulter, 1991; Sugimoto, Nishida, Murakami, Woo, Sakane, Yasuhara, Shuto, Hatanaka, & Kobayashi, 1990; Opala, Winter, Vance, Vance, Hutchinson, & Linn, 1991).

To date, no study reporting the interactive

relationship of valproate and carnitine on the cognitive and psychological status of children has been reported in the literature. Two adult patient groups, Alzheimer's dementia and chronic alcoholism, have been the subjects of a small number of studies in which carnitine supplementation was administered as a "cognition enhancer". Due to the limited nature of these studies, the age and medical confounds, nutritional variables, and lack of consensus, these studies will not be dealt with in this paper.







47


Freeman, Vining, Cost and Singhi (1994) studied the effect of carnitine on the symptoms associated with antiepileptic medication. In a double-blind, cross-over, placebo-controlled study, they studied the "well-being" of 47 children with seizures prescribed valproic acid or carbamazepine via parent report in person or over the telephone at the beginning and end of each 4 week cycle. They determined that there were no significant findings as the parents consistently reported improved "well-being" (a term not quantified or qualified) with carnitine and with carnitine placebo. They concluded that the cost of the carnitine ($.30/kilogram of body weight) is not justified by the non-significant improvements. Studies of Valproate Therapy with Epileptic Patients

An early study by Gram et al. (1979) studied 13

inpatients with generalized and partial epilepsy. A triple blind multiple cross-over design of varying valproate levels was implemented. Gram et al. (1979) reported a positive correlation between a decrease in the number of seizures and dose level. They further reported mild, transient side effects including drowsiness, anorexia, and hypersalivation which were not correlated with increased dose levels. They concluded that there is a positive correlation between higher valproate levels and better seizure control. They admit difficulty drawing conclusions about valproate







48

efficacy within a specific seizure type because of the small sample size.

Sommerbeck, Theilgaard, Rasmussen, Lohren, Gram and

Wulff (1977) studied 20 epileptics, age 13 to 63 years, with numerous seizure types and varying etiologies. Five patients had known brain damage and most patients demonstrated a mixed seizure profile. Ten had "major seizures", 11 had "minor seizures", 1 had focal seizures, and 13 had psychomotor seizures. All patients were receiving other anticonvulsant therapy which was maintained at a constant level throughout the valproate trial. A triple blind cross over design was employed. The purpose of the study was to assess changes in psychometric ability on tests of attention, visual perception, decision making, reaction time, verbal learning, and other general cognitive tests. The test battery consisted of Bourdon's Test (cancellation of 4 targets); Visual Gestalt (nonverbal learning and memory); 100 Minus 7 Test (verbal subtraction); Stroop's Color Naming (attention); Simple Reaction Time; Paired Associate Learning from the Wechsler Memory Scale; Continuous Line Patterns Test (learning and nonverbal reproduction); Digit Span from the Wechsler Adult Intelligence Scale modified to 3 trials per length (immediate span); Tapping Test for one minute trials (motor speed); Hidden Patterns (visual perception); and Time







49


Estimation Test. Behavior was assessed through use of the Objective Rating Scale (20 behaviors to rate) by the examiner after the neuropsychological testing.

Sommerbeck et al. (1977) found that the valproate

slowed decision making, reaction time, and motor tapping. Changes in the EEG's were observed. They concluded that valproate therapy for epileptics can lead to significant decreases in psychomotor speed as assessed by motor speed, simple reaction time, and timed tasks. A non-significant trend for impaired visuospatial analytic and synthetic functioning as assessed by performance on Hidden Pictures task, Continuous Line task, and the Visual Gestalt Test was observed. The psychometric changes were not correlated with seizure frequency, EEG changes and valproate serum level.

This study has several serious design flaws. First, a small sample (n=20) with a wide age range (50 years) was utilized without use of a normal control group. The different seizure types were employed, and duration and severity of the epilepsy were not controlled and were not included in the analysis of the results. Many subjects were receiving polytherapy for seizure control in addition to the valproate, confounding the specific effects of valproate.

The Committee on Drugs from the American Academy of Pediatrics (1985) reviewed numerous studies comparing antiepileptic drugs on the following factors: monotherapy







50


versus polytherapy; drug and medical history; seizure etiology, type and frequency; tests employed by the researchers; and design and standardization of the studies. Concerning valproate, they reported drowsiness when prescribed in conjunction with barbiturates. They reported that valproate can lead to negative mood and behavior changes which appear more severe for patients with central nervous system damage; however, these two categories of side effects were not further described. The Committee strongly suggested careful monitoring, physician education regarding seizures and antiepileptic drugs, and development of consistent, sensitive, brief, easily administered screening tests.

A recent review article by Trimble (1990) suggested that minimal cognitive side effects occur with administration of valproate. He suggested that the two main cognitive changes associated with antiepileptic drugs are general cognitive functioning and cognitive speed, and valproate mainly affects cognitive speed.

As the preceding review suggests, there are a number of limitations in the available literature concerning the effects of valproate on the cognitive status of epileptic patients. Most significantly, the subject samples have not been well defined. Given the numerous factors that appear to influence the cognitive status of epileptics (e.g. type








51


of seizure, etiology, duration of disorder, severity of disorder, antiepileptic drug therapy, etc.), the previously reviewed studies fall short in controlling for variables that may interfere with interpretation of valproate's effects on cognition. Small sample sizes and lack of normal controls are additional factors that hinder generalization of the existing data.

A study by Gallassi, Morreale, Lorusso, Procaccianti, Lugaresi, and Baruzzi (1990) studied the effects of valproate in a sample of 20 seizure free epileptics during monotherapy and after its withdrawal. The subjects had a mean age of 20.0 years (sd=4.7) and had an average of 10.8 years of education. A control group comprised of 20 normals matched for age, education, and social level were employed. All subjects were assessed with a baseline battery consisting of the following: Raven's progressive matrices (intelligence); auditory reaction time simple (vigilance); auditory reaction time choice (attention); verbal digits and spatial span (immediate memory); verbal learning (learning); spatial learning and finger tapping test (manual dexterity); trail making test (visuomotor performance); and fingertip number writing (sensory discrimination). The epileptics were assessed at four intervals: baseline, at 3 months after half dose reduction; 3 months later; and at one year. Those who relapsed during the study were excluded.







52

Only 11 of the 20 subjects completed the study. Raw scores were converted to z-scores and then to t-scores. All tests were combined by creating a mean of the t-scores to create a 'global performance score' (GPS). ANOVA's were performed and significance was set at 0.01.

Gallassi et al. (1990) reported significant adverse

effects of valproate on cognition. At baseline, significant group differences were observed on measures of attention, visuomotor task performance, and GPS. At time 2 with the half-drug dose, GPS was still impaired. No impairments were observed at times 3 and 4. Gallassi et al. interpreted these findings as subtle, subclinical aversive effects on cognition caused by valproate monotherapy.

This study is important in that it used a variety of neuropsychological measures and employed a normal control group. The area of attention was found to be vulnerable to valproate. Performance on the Trail Making Test, defined to be a measure of visuospatial function, is also considered to be a measure of attention. Several design flaws were noted. The normal control group did not participate in the longitudinal testing; all comparisons were made to their baseline performance, therefore, practice effects were not controlled. Alternative forms of the tasks were employed, but without a control group, rendering interpretation of differential test performance beyond baseline limited.







53


Furthermore, loss of 9 of 20 subjects due to relapse of seizures suggests the possibility of a differential subgroup of epileptics remaining in the study. Possibly those who did not relapse were neurologically in better condition and therefore more cognitively intact confounding the interpretation of improved test scores.

Valproate: Neuropsychological Effects

Forsythe, Butler, Berg, and McGuire (1991) examined the relative effectiveness and side effects, physical and cognitive, of carbamazepine, phenytoin, and valproate. Sixty-four children ranging in age from 5 to 14 years with newly diagnosed epilepsy were the subjects of this study. They were diagnosed with one of the following seizures type: tonic-clonic, complex partial, or complex partial with secondary generalization. All children were reportedly within the average range of intelligence. Thirty-one enuretic patients and 9 migraine patients were employed as controls. The epilepsy patients were placed on carbamazepine, phenytoin, or valproate in a random assignment.

Forsythe et al. (1991) used a cognitive test battery, fashioned after the work of Trimble, consisting of the following: visual recall of pictures of objects (immediate and 30 minute delay), immediate recall of designs, auditory recall (digits forward and backward, immediate span), visual







54


scanning (attention), the Stroop test (concentration), and speed of information processing. The subjects were tested at the initiation of their treatment and at 1, 6, and 12 months into their treatment. The children were also given the WISC-R, and the Neale Analysis of Reading Ability at 1 month and 12 months after initiation of therapy.

Forsythe et al. (1991) reported that all the

antiepileptic drug groups had good, equally effective seizure control. The fewest adverse side effects were reported with valproate, and the most with carbamazepine. The combined score of all the memory measures was found to be the most sensitive measure for drugs effects. A positive correlation between the serum level of valproate and memory scores was reported. Impaired memory recall was observed in the carbamazepine group after 6 months of treatment. Slower speeds of information processing were observed in the carbamazepine and phenytoin groups after one month of treatment.

Several problems are evident in this longitudinal

study. Use of a sample with heterogenous seizure types from a wide age range treated with three different drugs makes generalization of results difficult. The purpose of enuretics as a control group is uncertain. The test procedures employed lack verbal learning measures and prose recall, sensitive indicators of memory function in children.







55

All memory measures were summarized across span, short term recall, and long term recall, losing information concerning specific memory differences among groups and over time. A principal components analysis was employed for statistical interpretation. This procedure requires more subjects per measure than employed in this study causing the results to be questionable in their strength for interpretation. Results comparing the epileptics and the controls were not reported. Baseline intelligence and reading scores were not reported, and it is unclear if initial differences were adjusted for in the subsequent analyses.

Vining, Mellits, Dorsen, Cataldo, Quaskey, Spielberg, and Freeman (1987) studied the psychologic and behavioral effects of valproate and phenobarbital in a double-blind, counterbalanced, crossover study with 21 children of normal intelligence with mild seizure disorder. These subjects ranged in age from 6 to 14.5 years. Nine children were diagnosed with tonic-clonic seizures, 11 with partialcomplex seizures, and 1 child demonstrated both types of seizure. The EEG's prior to initiation of the study were interpreted as 5 normal, 9 mildly abnormal, and 7 very abnormal patterns. Seizure etiology included perinatal trauma, postnatal trauma, and idiopathic onset. Subjects were randomly assigned to one of the antiepileptic drugs for a 6 month period followed by a 6 month cross over period.







56


The neuropsychological battery employed by Vining et

al. included the WISC-R, subtests from the Detroit Tests of Learning Aptitude (attention and memory), the Symbol Digit Modalities Test (sequencing and processing efficiency), Bender-Gestalt (perception accuracy and constructional skills), Berkeley Paired Association Learning Test (attention and short-term learning), and Seashore Rhythm Test (auditory discrimination and short-term memory). Other tests included Reading, Spelling and Arithmetic from the Wide Range Achievement Test, the Gray Oral Reading Test (fluency and accuracy for oral reading), and a Continuous Performance Reaction Test (vigilance and visual concentration). Measures of gross motor skills, fine motor skills, dexterity and tracking included Finger Tapping, Ambulation Backwards, Mazes, and a video arcade tracking game.

Vining and her group reported comparable seizure

control for both drug groups. Using paired t-tests on all dependent variables, 7 of 35 neuropsychological measures and

9 of 48 behavioral items were reported to be significantly different and in favor of the subjects on valproate therapy. Specifically, 4 neuropsychological measures were found to be significantly different at the p<.01 level, including Block Design (WISC-R subtest), Performance IQ, Full Scale IQ, and the Berkeley Paired Association Learning Test II.







57


Vocabulary (WISC-R subtest), Picture Completion (WISC-R subtest), and the Berkeley Paired Association Learning Test I were different at the p<.05 level of significance. Three behavior items were found to be significantly different at the p<.Ol level, including: "fails to finish", "basically unhappy", and "unable to stop". "Disobedient", "other aches", "excitable", "worries", "problems with friends", and "problems with sleep" were reported significantly different at the p<.05 level. Vining et al. concluded that valproate therapy for children with generalized tonic-clonic seizures and partial complex seizures is as effective as phenobarbital and less deleterious to cognition and behavior.

A number of problems are evident in the design of this study and the interpretation of the findings. First, the intensity of seizure disorder was defined by seizure frequency; however, the "mild" distinction included a range of seizure frequency from two per day to one per year. Heterogeneous seizure etiologies, as well as multiple seizure types, were included in the subject sample. The sample size was small and a normal control group was not employed to help interpret practice effects. Full Scale IQ'S were purported to be within the normal, or "average" range; however, with a mean and standard deviation of 94.0 + 14.4 many subjects appear to be below the average range of







58

intellectual function. Though a lengthy neuropsychological evaluation was performed, all measures, and their subtests when present, were statistically analyzed with paired ttests. A total of 35 neuropsychological measures were compared with this method. The first three of the above listed significantly different neuropsychological measures are highly correlated, as Block Design is a part of the Performance IQ which is half of the computed Full Scale IQ. In addition, the specific scores from the WISC-R labelled as significantly different are not clinically different as they all fall within the average range.

Forty-eight behavior items were compared with paired ttests. The behavioral reporting was along a four-point scale from "not very much = 1" to "very much = 4". The significant behavior problem scores fell between 1.19 and

2.20 suggesting a non-clinical problem level.

The statistical analysis by Vining et al., multiple

paired t-tests, is highly vulnerable to Type I errors. The findings reported to be significant are at a rate so low that chance occurrences may have explained their findings. Finally, the results from both the neuropsychological and behavioral domains with the exception of the Berkeley Paired Association Learning Test II, are not within a clinically problematic range.





59


Aldenkamp, Alpherts, Blennow, Elmqvist, Heijbel,

Nilsson, Sandstedt, Tonnby, Wahlander, and Wosse (1993) studied the effect of withdrawal of valproate from a pediatric epilepsy sample as part of the multicenter Holmfrid study. One hundred children with epilepsy who had been seizure free for at least one year were withdrawn from monotherapy of valproate, carbamazepine, or phenytoin. Seventeen children diagnosed with absence, tonic-clonic, partial, and rolandic epilepsy were withdrawn from valproate monotherapy. They were matched to a healthy classmate. The subjects were tested at baseline, prior to drug withdrawal, at 3 months at which time they were totally off the drug, and at 3 to 4 months after that time. They employed the Finger Tapping Test, Simple and Binary Choice Reaction Time Tests, Computerized Visual Searching Test, and Word and Figure Recognition in simultaneous and serial presentation paradigms.

Aldenkamp and his colleagues reported no significant improvements on the motor, attention, and memory measures after accounting for practice effects in the epilepsy group as a whole. Significant differences in the absence seizure group (n=7) at baseline and after withdrawal were observed and were interpreted as continued, subclinical EEG activity which interfered with cognition. They hypothesized that the lack of improvement or deterioration with valproate







60

withdrawal resulted from the combined effect of removing the beneficial, protective effect of valproate and the recurrence of subclinical EEG abnormalities. No data was presented to confirm this hypothesis.

Aman, Weary, Paton, and Turbott (1987) studied the

effect of valproate on psychomotor performance in children with epilepsy focussing on the variables of dose, fluctuation of concentration of valproate and the diagnosis of the children. Forty-six epileptic children with normal IQs ranging in age from 4.4 to 15.4 years were the subjects. They were diagnosed with partial epilepsy, most of them with secondary generalization (n=10), generalized epilepsy half with absence and half with tonic-clonic (n=34), or unclassified epilepsy (n=2). The children were previously treated with phenytoin, carbamazepine, or polytherapy and were placed on valproate monotherapy.

The neuropsychological test battery employed in this study included Matching Familiar Figures (impulsivity), Auditory-Visual Integration Task (cross modal matching), a non-verbal recognition test using cartoon pictures, Rosvold's Continuous Performance Task (vigilance and attention), Seat Movements measured by duration on the chair, Maze Task from Klove's Motor Steadiness Battery (impulsivity and motor coordination), Klove's Graduated Holes Task (resting tremor), Pursuit Rotor Task (hand-eye







61


coordination), and Porteus Mazes (intelligence and planning). The subjects were tested at baseline (pre-drug), in a delayed medication condition (low serum level), and shortly after given medication (high serum level). The children were divided into to a high dose and a low dose group (27.14 and 15.84 mg/kg/day median, respectively).

Aman and his group reported effects according to diagnosis, dose, and time of medication. Data was statistically analyzed with a Repeated General Linear Analysis. Four significant differences by diagnosis were reported. Subjects with partial epilepsy demonstrated a longer response time on the Auditory-Visual Integration Task (F=5.95, p<.05), longer response time on the non-verbal recognition task (F=5.23, p<.05), more errors of omission on the Continuous Performance Task (F=4.88, p<.05), and a higher ratio of contact time to travel time on the Maze Task (F=9.86, p<.005).

Four significant findings by dose were reported. The high dose group demonstrated more seat movement during the Matching Familiar Figures Test (F=8.02, p<.01), lower accuracy and longer response time on the Auditory-Visual Integration Task (F=10.38, p.005, F=6.86, p<.01, respectively), and a lower test quotient on the Porteus Mazes Task (F=4.99, p<.01). Only one significant finding for effect of plasma concentration level, a factor







62


determined by time of valproate administration, was reported. The high concentration level group demonstrated greater contact time on the Maze Task (F=4.65, p<.05). On the whole, Aman and his group reported strong effects for diagnosis and dosage and weak effects for time of medication.

Aman et al. (1987) designed a study loaded with motor measures and did not effectively address memory, verbal learning, and attention. Their subject group had numerous seizure types and a control group was not employed to control for practice effects. The subject group had a wide age range (11 years) and age corrections were not employed. All but one of the procedures were electronically administered which may limit generalization. The key advantage for this method is reduction of examiner variation or examiner effect; however, electronic administration is not a direct reflection of most daily interactions. Finally, the majority of the differences reported were significant at the p<.05 level which is more representative of a trend rather than a significant difference in studies with small samples and multiple variables.

Rationale for Further Study

Review of the current literature regarding the relationship between valproate therapy and the pediatric epilepsy population reveals a relative paucity of well-controlled







63


studies of memory and attention. Narrowing subject inclusion to specific seizure types shown to be responsive to valproate, limiting the age range, and excluding children of below average intelligence would address weaknesses in previous research studies. Examining only one or two aspects of cognitive functioning will allow a more thorough analysis of valproate's effect. Memory and attention have been described to be crucial to learning in children (Cooley & Morris, 1990) and the extent to which these abilities are affected by idiopathic epilepsy, valproate monotherapy, and carnitine supplementation is not known. This information is of significant importance for thorough medical treatment, school performance, and behavioral management.

Specific Aims

The specific aims of this study were: (1) to evaluate for the presence of attentional and memory disorder in children diagnosed with idiopathic generalized epilepsy (accomplished at baseline testing); (2) to determine the effect of therapeutic valproate levels on memory and attention in epileptic children; (3) to determine if differences exist in memory and attention between children with epilepsy on valproate, children with migraine on valproate, and a non-medicated, neurologically normal control group; and, (4) to determine the cognitive effect of valproate plus carnitine in a pediatric epilepsy sample.

















CHAPTER 2

METHODS

Subjects

The subjects of this study were 12 children with

idiopathic epilepsy referred to this study by the physicians at the Pediatric Neurology Clinic at Shands Hospital. The subjects had a seizure profile (type and frequency) appropriate for treatment with valproate, i.e. generalized motor, absence, or complex partial with secondary generalization. Subjects were between the ages of 6 and 16 years for both pharmacologic constancy and design control. All subjects underwent an EEG to obtain a baseline and to support diagnosis and an MRI was performed to rule out the presence of a structural lesion or abnormality. An Internal Review Board (IRB) protocol was presented and explained to the parents or guardians of the prospective subjects and informed consent was obtained. Accrual of the subjects began in January 1993 and continued throughout the year until December 1993. An IQ screening was performed at the time of inclusion in the study using four WISC-R subtests.

64







65


Only those subjects with a mean scaled score of 8 or above and those with individual scales scores of 7 and above were included.

The twelve children with epilepsy recruited for this study and were diagnosed with the following seizure types: absence, partial complex with secondary generalization, primary generalized, and Rolandic. The sample was comprised of five males and 7 females. The mean age was 11.1 years with a standard deviation of 2.8 years with a range of 7.0 to 15.3 years. Ten of the children were right hand dominant, one was left hand dominant, and one was ambidextrous. Age of onset, duration of illness, presence of significant headaches, diagnosis of attention deficit disorder with hyperactivity and family history of epilepsy were recorded (See Table 2-1). Ten children remained in the study at the four-week mark and nine children completed the study.

A control group comprised of twenty non-neurologically involved children was recruited from a local public school system by written requests through teachers to parents. Following parental consent, the control group participated in all the neuropsychological testing procedures but received no medication. The control group was screened by parent report for history of neurological disease and injury, learning disabilities, and significant behavior and







66


medical problems and were excluded if any of these conditions were present. Eighteen of these children remained through the three testing sessions. This normal control sample consisted of 14 males and 6 females with an average age of 12.1 years with a standard deviation of 3.2 years, ranging from 7.2 to 16.9 years. Nineteen were righthanded and one was left-handed.

In addition, the pediatric epilepsy group was compared to pediatric migraine patients being treated with valproate carnitine, valproate with carnitine placebo, and valproate placebo and carnitine placebo. The migraine patients were referred to the study from the Pediatric Neurology Clinic and were screened for appropriateness for this study by history, physical examination, EEG, and MRI (where appropriate) by their referring physician. The migraine subjects participated in the identical testing procedures as the epileptic sample. Age restrictions were similar to those for the epilepsy group (6 to 16 years). There were 14 migraine patients at baseline testing, with eleven males and 3 females and seven completed the study. Their mean age was 12.9 years with a standard deviation of 1.9 years, ranging from 9.5 to 15.8 years.

At baseline, there was no significant difference

between age, handedness, or intellectual (IQ) level among the three groups (See Table 2.2).







67


Materials and Apparatus

The following is a brief description of the

neuropsychological and behavioral measures used in this study.

Intellectual Screening

Four subtests from the Wechsler Intelligence Scale for Children Revised (WISC-R) were employed to assess intelligence. Use of the WISC-R has been chosen over the WISC-III for comparability to previous research. The WISC-R has well established norms for ages 6-16 years. Two subtests from the Verbal domain, Similarities and Comprehension, were selected because they provide strong reliability for intelligence estimation with .81 for Similarities and .77 for Comprehension. These subtests were chosen over the other subtests from the verbal domain of the WISC-R because they rely less of specific acquired facts and recall of previously learned information (Sattler, 1988). Similarities, a test which requires the subject to determine how two things are alike, measures abstraction ability and verbal knowledge. Comprehension evaluates the subject's practical reasoning ability. Two non-verbal subtests from the Performance domain, Object Assembly and Block Design, were employed. Both are timed tests which require visual analysis, synthesis, and construction performance. Both subtests offer good split-half reliability with .70 for







68


Object Assembly and .85 Block Design (Sattler, 1988). Furthermore, there is strong correlation of each subtest to the Performance IQ with .68 for Block Design and .60 for Object Assembly.

These four subtests of the WISC-R provided a reliable estimate of general intellect and served as a screening instrument for level of intelligence for all subject groups and as a population descriptor for the epilepsy and migraine groups. Intelligence quotients are thought to be relatively stable measures and the performance specific subtests are prone to practice effects. Therefore, these subtests were only administered during the baseline assessment. Memory Measures

Verbal memory was assessed at each of the three testing intervals with several measures. A modified version of the Logical Memory subtest from the Wechsler Memory Scales (WMS) Form I and Form II was employed to assess both immediate and delayed memory. The Logical Memory subtest was modified by reading only one story at a given administration as three forms are necessary for this testing paradigm. Both stories from the WMS Form I ("Anna Thompson" and "The American Liner") and the first story from the WMS Form II ("Dogs are trained") were chosen. To best interpret performance on the three versions of this task, scores were determined with established scoring criteria (Russell, 1975).







69


Wechsler Memory Scales Form I and Form II are not directly interchangeable; however, the Logical Memory subtest of each version is considered adequately comparable (Spreen and Strauss, 1991). Extensive research of the reliability and validity of the WMS Logical Memory subtest has been conducted. Norms are available for ages 10 through adulthood, using the standard two story administration. Raw scores were converted to z-scores by doubling the recall of the single story and subtracting the age appropriate mean and dividing by the standard deviation. As no normative data are available for this modification, comparison to the normal control group was necessitated.

The Buschke-Levin Selective Reminding Test (SRT) is

designed to measure verbal learning and verbal memory in a multiple trial list learning format. In this task, the subject is presented with a 12 item list of non-related words and is asked to recall as many words as possible in any order. The remainder of the words from the list not recalled is then repeated. This process continues for 8 trials or until the subject has learned all 12 words. This procedure allows the examiner to distinguish between immediate memory, long term memory and rate of learning. Though this task provides several scores, only two dependent measures were analyzed: learning (total number of words recalled over trials 3 through 8) and delayed recall.







70


The Buschke-Levin SRT has multiple forms that are relatively equivalent and interchangeable allowing for repeat testing (Clodfelter, Dickson, Newton Wilkes, & Johnson, 1987) and three forms were required for this paradigm. Test-retest reliability was found by Masur, Fuld, Blau, Thal, Levin, and Aronson, (1989) to be .92 for retrieval with low reliability (.42) for intrusion rate. Some practice effect is expected, and a normal control group was employed to account for this. Normative data is available for age 5 through adult.

Digit Span, taken from the WISC-R, was the last measure of verbal memory. In this task, the examiner reads aloud, at a rate on 1 per second, series of digits which the subject immediately repeats in the same order. Digits Forward begins with 3 digits and proceeds to 9 digits with 2 series of each length. Digits Backward, a task requiring the subject to reverse the order of the numbers presented, begins with 2 digits in a series and increases to 8 digits. The task is discontinued when the subject fails both attempts in a given trial.

Digit Span is a measure of immediate memory and

attention and has adequate reliability of .78 (Sattler, 1988). Practice effects have not been reported and it is appropriate for repeat testing. Normative data are available for age 5 through adult. Normative z-scores were







71


obtained in the identical manner as described above.

Nonverbal Memory was assessed with two measures, the Corsi Cube Test (Milner, 1971) and the Visual Learning subtest from the Wide Range Assessment of Memory and Learning (WRAML) (Sheslow & Adams, 1990). Corsi Cubes is considered to be the nonverbal counterpart of Digit Span as it requires the subject to repeat series of taps of increasing length both forward and backward. The examiner taps the sequence at a rate of 1 per second on the fixed Corsi array of 9 blocks which are 1.5 inch cubes. Because of the 2 dimensional aspect of the stimuli, both spatial ability and nonverbal memory are required. This task has not shown to be susceptible to practice effects and is acceptable for repeat testing. The normative performance is thought to be one less than the normative span for digits for a given age. Normative z-scores were obtained in a similar manner as above.

The Visual Learning subtest from the WRAML requires the subject to recall the spatial location of visual designs presented in specific positions on a four by four array. After presentation, the stimuli are covered with sponge shields. The child is required to recall targets over 4 trials. Sheslow and Adams (1990) report reliability coefficients the median of which is .88 with test-retest reliability at .81. As the stimuli are non-meaningful and







72


the reliability is good, significant practice effects are not expected. Normative data are available for age 5 through adult in the WRAML manual. Attention Measures

Attention was assessed with a computerized Continuous Performance Task (CPT) and the Paced Auditory Serial Addition Test (PASAT). The CPT can assess both vigilance and reaction time. The single-choice reaction time requires the subject to press one of two keys on a computer keyboard immediately following a selected cue. The cue, which appears at random points on the screen, signals to the participant with which hand to respond. This reaction time task is a measure of speeded decision making and attention. Designing the task to be a forced choice increases its difficulty, and therefore its sensitivity to possible impairments in decision making and attention. This measure was shown to be a sensitive measure in discerning differences between a pediatric renal disease sample and normal controls (Fennell, Fennell, Carter, Mings, Klausner, & Hurst, 1990).

The CPT is IBM computer administered and can be altered for multiple administrations (Fennell, et.al., 1990). In this study either the Leading Edge portable computer or a standard IBM computer was employed. Some practice effects were expected due to increasing familiarity with the







73


stimulus, but use of normal controls allowed for clearer interpretation of the results. The computerized format is engaging to the child and the motor requirements are minimal, thus allowing for assessment of attention and speeded decision making. Mean reaction time across all of the trials of the various conditions was employed as the sole dependent variable. As there is no normative data on this task, the use of the control group data was necessitated.

The PASAT (Gronwall & Wrightson, 1974) is a serial

addition task and the rate of presentation of the stimuli is controlled by audio tape administration. A total of 61 digits are presented at 2.4, 2.0., 1.6 or 1.2 seconds per digit for varying degrees of difficulty; however, only the two slowest speeds of presentation were employed in this study. The subject adds the first two digits and states aloud the sum and then adds the third digit to the second digit and so on. Gronwall and Wrightson (1981) report that the PASAT is a sensitive measure of information processing and sustained attention. It has been described as a task which assesses central information processing similar to that of reaction time and divided attention tasks (Gronwall and Sampson, 1974).

Two short-comings of the PASAT for this study are lack of normative data for children under the age of 8 and strong







74


practice effects (Gronwall, 1977) for the second presentation, though not at later presentations (Spreen and Strauss, 1991). Strong split-half reliability (.96) has been reported for the PASAT (Spreen and Strauss, 1991).

The two slowest trials were administered and scores (total correct responses per trial) were converted to zscores using available normative data. Only subjects eight years of age or older were included in the analysis due to limited normative data.

Motor Measures

Graphomotor constructional skill was assessed with the Beery Test of Visual Motor Integration (VMI) while motor speed was assessed with the Finger Tapping Test. The VMI (Beery, 1967) presents the subject with 24 geometric figures of increasing difficulty which the subject is to copy to the best of his perceptual motor ability. Cosden (1985) reports a mean interrater reliability of .93 and retest reliability ranges from .63 to .92 over various time spans and subject groups. Sattler (1988) reports good validity from a sample of 3090 subjects. Normative data is available for ages 2:10 through 17:11 and no gender differences have been noted. The VMI is thought to be relatively unaffected by environmental factors (i.e. medication, time of testing) and offered a comparison to more state oriented measures. This measure was employed only at baseline and final testing.







75


Motor speed was assessed with the Finger Tapping Test

(Reitan, 1969). In this task, the subject is presented with a tapping board with a metal key and an automated counter. The subject places his index finger on the key and while maintaining his hand in a relaxed position, taps the key as quickly as possible. After familiarizing the subject with the tapping board, four trials each of ten seconds duration for each hand were conducted, alternating between the hands after 2 trials. The mean numbers of taps with both dominant and the non-dominant hand were obtained. Normative data are available for age 6 through adult and z-scores were calculated. Moderate reliability has been reported, ranging from .58 to .93 (Spreen and Strauss, 1991). Behavioral Measures

All dependent measures of behavior were obtained

through parental report on the Child Behavior Checklist by Achenbach (CBCL) (Achenbach and Edelbrock, 1986). The CBCL is a parent report measure commonly used to assess behavior problems and psychopathology. This questionnaire was selfadministered to the subject's parent or guardian. The two broad factors determined with the CBCL are the Externalizing and Internalizing T-scores. Children who fall into the externalizing range tend to be overactive and to display conduct problems. The internalizing child tends to be more withdrawn and dependent on the parent. The CBCL also







76

provides several more specific behavior scales which load on the Internalizing and Externalizing scales. DuPaul, Guevremont, and Barkley (1991) state that the CBCL is the best standardized child behavior rating scale available and it has very good reliability and validity.

Given the large number of variables and the limited

sample size, the CBCL was described in the data as T-scores for the Internalizing and the Externalizing scales, measures which summarize the parental response.

Procedures and Design

The procedure for the baseline testing session for all participants in the study involved an initial screening procedure comprised of the four subtests of the WISC-R followed by the neuropsychological tests described above. Testing was completed prior to initiation of medication.

Once inclusion to the study was determined, the

neuropsychological battery began with the presentation and immediate recall of the modified WMS Logical Memory Story. The PASAT was the second measure so as to minimize fatigue. The Tapping Test followed for similar reasons. The CPT was administered next. Approximately 30 minutes had elapsed and the delayed recall of Logical Memory was assessed. The Buschke-Levin SRT was next administered, followed by Digit Span, the Beery VMI, the Corsi Cube Test and Visual Learning from the WRAML. The delayed recall trials for the Buschke-







77


Levin SRT and the WRAML Visual Learning were administered last. Parents or guardians completed the Child Behavior Checklist while their child was tested.

This study is a short-term follow-up design measuring

attention and memory in a pediatric epilepsy sample at three points over 3 months: before initiation of valproate monotherapy (baseline); after 4 weeks of valproate monotherapy; and after 8 weeks of valproate monotherapy with double blind randomization to a carnitine or carnitine placebo condition. The examiners were blind to both the diagnosis and to the drug regimen. The patient and his family were blind to the drug regimen at the time of randomization to carnitine or carnitine placebo. Randomization to drug groups was performed by the Pharmacy Department through a randomization computer program.

The diagnosis of the subjects was performed by

pediatric neurologists at Shands Hospital. Once diagnosed with pediatric idiopathic epilepsy, medical appropriateness for this study was determined by the neurologists through physical examination, EEG, and MRI. Once parental consent was obtained, the intellectual screening for inclusion was performed. Baseline neuropsychological testing was performed. To address practice effects in the test-retest paradigm, alternate forms of the measures, where applicable, and a normal age matched control group were employed.







78


The valproate monotherapy was monitored by the neurologists and the laboratory personnel at Shands Hospital. The subjects received therapeutic dosages resulting in plasma levels in the range of 50 and 100 micrograms per milliliter. The addition of the carnitine and carnitine placebo was monitored by patient report of compliance to the prescribed regimen.

Parallel to the epileptic subject group, a pediatric migraine sample was treated with valproate with carnitine, valproate with carnitine placebo, or valproate placebo with carnitine placebo. Identical criteria for inclusion and testing procedures were implemented. Attempts to age match this group to the epilepsy group were made. All patient information remained confidential.

The normal control group participated solely in the

neuropsychological testing and they were rewarded with gift certificates to a local fast-food restaurant. Their materials also remained confidential.

Hypotheses

In light of the current available literature on

childhood epilepsy, childhood migraine, valproate, and carnitine, the following hypotheses were made regarding this study. (1) The pediatric epilepsy sample screened for intelligence in the average range would demonstrate mild to moderate impairment on attention and memory at baseline







79


given that memory and attention impairments have been observed in children with generalized epilepsy, the type of epilepsy best treated with valproate monotherapy. The pediatric migraine sample would not demonstrate any cognitive deficits. The overall cognitive performance for both samples will be biased toward the average range as all subjects are screened for average intelligence. (2) Though the valproate will help normalize EEG activity in the pediatric epilepsy sample and control seizure activity, the current literature suggests that valproate may lead to poorer performance on timed decision making tests and tests requiring sustained attention. We predicted that the pediatric epilepsy sample would continue to demonstrate mild attention and memory impairment while on the valproate monotherapy. (3) The addition of carnitine to the valproate monotherapy would reduce the negative somatic side effects and would not lead to negative cognitive effects on either group. Those subjects in both the pediatric epilepsy sample and the pediatric migraine sample would perform as well or better on the neuropsychological battery following the addition of carnitine. (4) The necessity of repeated measures for this study's design would lead to improvement in performance across measures which are sensitive to repeated exposure. (5) The medical involvement of the epileptic sample would result in somatic complaints at baseline.







80


TABLE 2.1
CHARACTERISTICS OF THE EPILEPSY SAMPLE

Seizure Type 2 Rolandic, 6 Absence,
2 Partial Complex, 2 Mixed Age of Onset 1 year to 12 years

Range in years 0.5 years to 12 years

EEG findings 6 abnormal; 6 normal

Family history 2 positive; 10 negative

Significant headaches 7 positive; 5 negative

Attention Deficit Disorder 2 positive; 10 negative


w/ Hyperactivity

Level of Seizure Control
on VPA

Reported Negative Side
Effects of VPA Table 2.2 DEMOGRAPHIC INFORMATION

Epileptic


8 excellent; 2 moderate; 1 intolerant; 1 unknown

4 w/ adverse effects
8 w/ no adverse effects


Migraine


Control


Baseline sample 12 14 20
size
Gender Ratio M/F 5/7 11/3 14/6

Mean Age (years) 11.1 (2.8) 12.9 (1.9) 12.1 (3.2)

Age Range (years) 7.0 15.3 9.5 15.8 7.2 16.9

Handedness:R/L/Amb. 10 /1 /1 13 /1 /0 19 /1 /0

WISC-R IQ subtest means
Similarities 10.7 (2.5) 11.8 (2.3) 12.4 (3.2)
Comprehension 11.9 (3.4) 10.0 (2.2)* 12.8 (2.6)
Block Design 9.7 (1.8) 9.3 (2.6)* 11.4 (2.2)
Object Assembly 10.6 (1.3) 10.5 (4.2) 11.5 (3.5)

* significant difference from control group at p<.05 level. Note: Standard deviations are provided in parentheses.







81


TABLE 2.3 NEUROPSYCHOLOGICAL DOMAINS

FACTOR


Immediate Verbal Memory


Verbal learning

Delayed Verbal Memory


Non-Verbal Memory


Non-Verbal Learning

Attention


Motor Function


Behavior


TESTS


Digit Span Story Recall -Immediate Buschke-Levin learning Buschke-Levin retrieval Story Recall delayed Corsi Cubes Test Visual Learning delayed Visual Learning PASAT
CPT

Beery VMI Finger Tapping CBCL
















CHAPTER 3

RESULTS

All subjects were determined to be within the average range of general intelligence. Group differences on two of the WISC-R subtests, Block Design and Comprehension, were observed between the migraine group and control group, with epileptics holding a middle position. Given these differences in conjunction with the wide range for a normative "average score" on the IQ measure, IQ was employed as a covariate across the remaining neuropsychological measures to ensure its role in the group differences. An IQ factor was determined by deriving a mean of the four subtest scores.

An analysis of variance (ANOVA) was first used to

compare the epileptics and control group means at Baseline for each measure employed. Pairwise Tukey comparisons were performed to determine direction of difference. Tukey pairwise comparisons were chosen to be conservative (Howell, 1992). An analysis of covariance (ANCOVA) was used to compare the groups at baseline adjusting for IQ as described 82







83


above. Multiple analyses of variance (MANOVA's) by cognitive domain were not performed because of low outcome variable intercorrelations and small design cell frequency (Huberty & Morris, 1989). Furthermore, MANOVA's often necessitate subsequent ANOVA's to determine specificity of group mean differences leading to inflation of the Type I error rate (Dar, Serlin and Omer, 1994).

The neuropsychological measures described previously were employed at baseline. Table 3.1 provides the means, standard deviations, and significant p-values for the baseline neuropsychological scores for the epileptic and normal control groups.

Hypothesis One

The first hypothesis of this study was that children

with idiopathic epilepsy would perform at a level below the normal control group on measures of attention and memory before initiation of the valproate monotherapy. Please refer to Table 3.1.

At baseline, there were several significant differences between the epileptic group and the normal control group. Performance on the PASAT in both the slow and fast conditions by the epileptic group was significantly below that of the control group (p=0.0026 and p=0.0070 respectively). Migraineurs maintained an intermediate position which was not significantly different from the







84


other groups. Digits Forward was significant at p=0.0416; however, pairwise comparisons indicated no specific difference between any two groups. The Buschke Selective Reminding Test Learning score was significantly different (p=0.0062) with both the control group and the migraineurs performing significantly better than the epileptics.

As expected, a significant difference among group means was observed for the CBCL Internalizing T Score. Tukey pairwise comparisons indicated that the epileptic and the migraine group means were both significantly greater than the control group mean and they did not significantly differ from each other (p=.0041).

The ANCOVA adjusting for IQ provided comparable results on the both trials of the PASAT and the Buschke SRT learning score. In addition the CPT reaction time measure was significantly slower for the epileptic group compared to the control group (p=.031). However, the difference in Digits Forward was no longer significant when covaried with IQ.

Hypothesis Two

The second hypothesis stated that the epileptic group would continue to demonstrate impaired attention and memory and would have difficulty on timed decision making tasks (CPT) following the initiation of valproate monotherapy. Again, ANOVA's were performed and pairwise Tukey comparisons were employed. Results are presented in Table 3.2. The







85


ANCOVA with the IQ covariate was also performed.

At Time 2, four weeks after initiation of valproate, the epileptic group continued to display significant differences from the control group in performance on the PASAT (slow and fast conditions), Digits Forward, and Buschke Selective Reminding Learning score. On the PASAT slow condition, the epileptic group performed below the level of the control group at the p=.035 level. For the PASAT fast condition, the epileptic group performed below the control group at the p=.022 level. Again for both these measures the migraine group performed at an intermediate level which was not statistically different from the other groups. On Digits Forward, the ANOVA revealed a significant difference among the groups (p=.024); however, the Tukey pairwise comparisons failed to reveal a specific difference between any two groups. Again, similar to baseline data, the epileptic group performed at a significantly lower level than the control group on the Buschke SRT Learning score (p=.017), with the migraine group in an intermediate position. No other measures were significantly different at Time 2. These results support the second hypothesis that epileptics would continue to display lower performances on tasks tapping memory and attention while on valproate monotherapy.







86


Hypothesis Three

The third hypothesis stated that the epileptic children taking the carnitine supplementation in addition to the valproate monotherapy would perform better than those taking carnitine placebo. Using ANOVA's and pairwise Tukey comparisons, there were no significant differences on any of the measures between the epileptics receiving carnitine and those receiving carnitine supplementation. See Table 3.3 for means and standard deviations. The third hypothesis was not supported by these findings; however, the sample sizes were very small (5 epileptics in the carnitine condition and

4 in the carnitine placebo condition).

Given these findings, the epileptic group was collapsed at Time 3 and scores were reanalyzed using a repeated measures ANOVA with one between subject factor (control versus epileptic) and one within subject factor (time). Paired t-tests were performed to compare within subject responses between Baseline and Time 3 for the Achenbach Child Behavior Checklist. The migraine group was not included in this analysis because of their high rate of attrition and significant non-compliance with valproate monotherapy. Results are presented in Table 3.4.

At Time 3, the epileptic group performed at a

significantly lower level on PASAT fast condition (p=.0044). On the Buschke SRT delay trial, the epileptic group







87


performed at a lower rate than the control group (p=.0247). There were no other group differences at Time 3.

On further analysis of the PASAT slow and fast

condition across the groups and across time, there is a significant main effect for group (p=.0183 and p=.0075 respectively). There was no significant main effect for time nor was there a significant interaction of group by time.

Digits Backward showed a main effect for group (p=.0348); however, no main effect for time nor an interaction of group by time were observed. Corsi Forward showed a significant effect for time (p=.0293), but not for group or a group by time interaction. Analysis of Digits Forward and Corsi Backward revealed no significant differences for group or time or a group by time interaction.

A significant main group effect (p=.0236) on the Buschke SRT Learning measure was observed, but no significant time or time by group interaction was observed. The Buschke SRT Delay measure approached significance for a main group effect (p=.0589).

A significant main effect for time was observed on the Logical Memory immediate and delayed recalls (p=.0141 and p=.0102, respectively). Analysis of responses suggests that the version of the test administered at the second testing







88

was significantly more difficult for all groups compared to the versions at Baseline and Time 3.

A significant main effect for time was observed for the WRAML Visual Learning scaled score (p=.0097) and indicates that all groups improved their performance over time at a similar rate. Again, the WRAML Visual Learning delayed recall score showed a main effect for time with all groups improving in a similar fashion with repeated exposures. There were no significant differences on the Achenbach CBCL measures at Time 3.

To further elucidate the role of valproate, the blood plasma level of valproate was covaried with the test scores using an ANCOVA procedure. Most of the epileptic sample were in the therapeutic range (i.e., 50 or higher), and several were only slightly below that level (i.e., 30 or higher), which is consistent with the peak / trough fluctuations of acceptable valproate therapy compliance. At Time 3, there was a significant difference on the PASAT fast condition, in that the score increased with blood level of valproate. Interpretation of this finding is limited as the groups compared were the epileptics in the carnitine condition. No other scores differed by group when valproate level was covaried.












TABLE 3.1
BASELINE NEUROPSYCHOLOGICAL DATA EPILEPSY vs.


Epileptics


Controls


p-value


(The following are presented in z-scores)


PASAT Slow PASAT Fast Digits Forward Digits Backward Corsi Forward Corsi Backward Logical Memory
Immediate
Delayed Motor Tapping
Dominant
Non-dominant


-2.15
-1.69
-0.89
-0.88
-0.74
0.17


(0.93) (0.76)
(0.45) (0.60) (0.69) (0.67)


0.00 (1.79) 0.12 (1.81)

-0.36 (1.42)
-0.34 (1.49)


-0.43
-0.49
-0.50
-0.72
-0.38
0.29


(0.98) (0.89) (0.68) (0.61)
(0.54) (0.60)


0.93 (1.35) 0.75 (1.21)

-0.36 (0.87)
0.05 (0.98)


(The following are presented in obtained scores)


Selective Reminding
Learning
Delayed
Visual Learning
Scaled score
Delayed


45.50 (13.02) 7.08 (2.77)


56.65 (7.23)
7.75 (2.53)


10.33 (3.68) 10.70 (2.90) 9.00 (3.57) 9.20 (2.76)


CPT: Reaction time 49.38 (14.57)


Visual Motor Integration
Standard Score 98.45

CBCL T-scores
Internalizing 65.09 Externalizing 61.27


(7.83)


(10.88) (11.59)


39.20 (8.81) 102.00 (10.00) 46.43 (11.91) 56.57 (11.21)


.0062** NS NS NS NS NS


.0041** NS


NS = not significant; = approaching significance
** = significant with p< .01


Measure


89


CONTROL


.0026** .0070**
.0416* NS NS NS

NS NS

NS NS








90


TABLE 3.2
TIME 2 NEUROPSYCHOLOGICAL DATA EPILEPSY vs. CONTROL


Epileptics


Controls


p-value


(the following are presented in z-scores)


PASAT Slow PASAT Fast Digits Forward Digits Backward Corsi Forward Corsi Backward Logical Memory
Immediate
Delayed Motor Tapping
Dominant
Non-dominant


-2.33
-2.02
-0.93
-0.73
-0.38
0.40


(1.47) (0.89)
(0.43) (0.78) (0.71) (0.93)


-0.98 (0.99)
-1.25 (1.15)

-0.19 (1.28)
0.18 (0.94)


-0.82
-0.74
-0.37
-0.42
-0.08
0.55


(1.09) (0.97) (0.68) (0.56) (0.68) (0.79)


-0.50 (1.17)
-0.41 (1.13)

-0.43 (1.23)
0.10 (1.00)


(the following are presented in obtained scores)


Selective Reminding
Learning
Delayed

Visual Learning
Scaled score
Delayed


46.40 (14.30) 6.20 (3.46)


12.00 (4.11) 9.40 (3.50)


57.65 (7.55)
8.15 (2.89)


11.95 (3.14) 10.55 (3.02)


CPT: Reaction time 49.81 (12.94) 48.22 (10.76)


NS = Not Significant; ** = significant with p < .05.


Measure


.035**
.022** .024** NS NS NS

NS NS

NS NS


.012** NS


NS NS


NS








91


TABLE 3.3
TIME 3 NEUROPSYCHOLOGICAL DATA EPILEPSY W/ CARNITINE vs.


EPILEPSY W/ CARNITINE


PLACEBO vs. CONTROLS


Epi+C


Epi+P


Controls


(The following are presented in z-scores)


-1.92 (0.45)
-2.52 (0.39)


Forward -0.82 (0.13) Backward -0.68 (0.81) Corsi


Forward -0.34 Backward 0.11 Logical Memory
Immediate -1.17 Delayed -1.99 Motor Tapping
Dominant -0.23 Non-dom. 0.31


(0.83) (0.97)

(0.59)
(1.45)

(1.26)
(1.48)


-1.35 (1.93)
-1.64 (1.14)

-0.40 (0.18)
-1.09 (0.42)

-0.20 (0.66)
0.27 (0.44)

0.79 (1.99) 1.21 (3.30)


-0.85 (1.34)
-0.60 (0.96)

-0.30 (0.76)
-0.49 (0.58)

-0.28 (0.71)
0.50 (0.73)

0.43 (1.60) 0.58 (1.70)


0.00 (0.87) -0.28 (1.01) 0.18 (1.30) -0.20 (0.72)


(the following are presented in obtained scores)


Selective Reminding
Learning 51.25 (12.61) Delayed 5.00 (1.83)

Visual Learning
Scale sc. 12.25 (4.03) Delayed 8.25 (3.20)

CPT: Rxn time 43.62 (8.08)

Visual Motor Integration
Stnd sc. 97.50 (15.02)


48.80 (13.20) 5.80 (3.77)


12.80 (1.92) 10.60 (2.61)

47.94 (3.41)


56.24 (12.00) 8.35 (2.96)


12.59 (2.09) 10.41 (2.72) 44.15 (9.21)


106.00 (23.31) 108.2 (9.64)


CBCL T-scores
Int.
Ext.


65.00 (13.08) 63.00 (15.39)


57.40 (10.29) 57.40 (14.38)


Measure


PASAT Slow PASAT Fast Digits


NA NA







92


TABLE 3.4
MEMORY AND ATTENTION DATA EPILEPSY vs. CONTROL


Measure


Time 1


Time 2


Time 3


(The following are presented in z-scores)


PASAT Slow EP
CN
PASAT Fast EP
CN
Digits
Forward EP
CN
Backward EP
CN
Corsi
Forward EP
CN
Backward EP
CN
Logical Memory
Immed. EP
CN


Delayed EP
CN


-2.04
-0.49
-1.63
-0.58

-0.83
-0.54
-0.98
-0.81

-0.77
-0.37
0.01 0.36

0.27 0.91 0.56 0.68


(1.02) (0.97) (0.79) (0.91)

(0.44) (0.73) (0.61) (0.59)

(0.76) (0.55) (0.70) (0.53)

(1.84) (1.44) (1.49) (1.28)


-1.99
-0.73
-1.89
-0.71

-0.89
-0.36
-0.92
-0.31

-0.41
-0.06
0.42 0.64

-0.75
-0.41
-1.08
-0.35


(1.20) *
(1.02) (0.88) (0.95)

(0.44) (0.74) (0.54) (0.51)

(0.74) (0.68) (0.99) (0.79)

(0.80) (1.23)
(1.12) (1.20)


-1.59
-0.85
-2.02
-0.60

-0.59
-0.30
-0.91
-0.49

-0.26
-0.28
0.20 0.50

-0.05
0.43
-1.16
0.53


(1.42) MG
(1.14) (0. 96) *MG (0.96)


(The following are presented in obtained scores)


Selective Reminding
Learning EP 44.77
CN 56.53
Delayed EP 7.11
CN 8.12
Visual Learning
Scaled EP 10.55
CN 11.35 Delayed EP 8.78 CN 9.65
CPT EP 43.02
CN 40.28


(14.46)* (7.87)
(3.14) (2.57)

(4.00) (2.57) (3.60) (2.67)
(6.64) (9.45)


45.56 57.82 5.89 8.35

12.44 12.65
9.78
11.41 45.79 48.27


(14.90)* (8.03) (3.52) (2.96)

(4.10) (2.71) (3.50)
(1.94) (7.30) (11.38)


49.89 56.24 5.44 8.35

12.56 12.59
9.56
10.41 46.21 44.15


(12.18) MG (12.05) (2.92)* (2.96)

(2.83) MT (2.09) (2.96) MT (2.72) (5.27) (9.21)


NOTE: EP = Epileptic group; CN = Control group
denotes significant group differences p<.05.
MG = Main effect for group (p<.05)
MT = main effect for time (p<.05)


(0.26) (0.76) (0.62) (0.58)

(0.69) (0.71) (0.68) (0.73)

(1.79) (1.60) (3.01) (1.70)


MG


MT




MT MT

















CHAPTER 4

DISCUSSION

Review of available literature revealed few well

controlled studies of the effect of valproate on cognition, specifically memory and attention, in the pediatric epilepsy population. The relative recency of valproate in the role of seizure management, the increasing awareness of specificity of cognitive domains, and the ever growing realization of differences between child and adult cognition likely contribute to the limited findings. Studies of other antiepileptic drugs such as phenobarbital and phenytoin, have suggested that antiepileptic medications may in some instances directly affect cognitive studies. The work by Trimble (1987) describes slower reaction time and difficulty with sustained attention with the use of phenobarbital. Trimble (1987) also reports impairments in memory, motor speed, and mental speed with phenytoin.

Cognitive deficits characterized by poor attention,

impaired learning, and behavior problems have been reported in the epilepsy population at a greater frequency than in 93







94


the neurologically normal population. Attentional difficulties are over-represented in epileptic children with generalized seizures. Verbal memory deficits have been seen in children with partial epilepsy with a left temporal lobe focus and non-verbal memory deficits have been reported in children with partial epilepsy with a right hemisphere focus (Trimble, 1987). Seidenberg (1986) reported a wider range of IQ in the epilepsy population compared to normals.

Often group differences and general deficits are

observed, but the characteristics of the deficits (i.e. type of memory problem or clinical significance of the attentional deficits) are not well described, thus limiting generalizability. In other cases, specific information is lost in the attempt to define a patient's global functioning. For instance, the study by Farwell et al. (1985) which studied 118 children with epilepsy reported IQ scores and a neuropsychological rating of "good" or "bad". Hermann (1982) studied 50 children with epilepsy and grouped their subjects according to scores on the Luria Nebraska Neuropsychological Battery into a "poor neuropsychological" or a "good neuropsychological" group. In neither study were cognitive deficits described nor were the antiepileptic drug protocols reported or described.

Review of studies attempting to qualify and to quantify the pattern of abilities have revealed serious research







95


confounds including the heterogeneity of epilepsy for etiology, age of onset, duration of symptoms, previous and ongoing pharmacological intervention and the idiosyncratic characteristics of the patients. These factors confound interpretations of the findings and limit generalizability of findings

Results to date suggest that valproate monotherapy is very effective in the treatment of general and complex partial seizures in both adults and children, though not without some cost. The most concerning side effect identified with valproate treatment is liver failure. Upon close examination, the preponderance of cases are reported in the very young, those with other serious disorders, those with previous liver disease, and patients receiving polytherapy (Dreifuss, et al., 1987). Adverse physical side effects related to valproate do occur; however, with careful monitoring they appear to be at a rate and an intensity which is favorable in comparison to alternative medications (Herranz, et al., 1988). Recent literature suggests

that the most adverse central nervous system effects attributed to valproate (i.e. sedation and somnolence) occur when it interacts with other antiepileptic medications, polytherapy.

A few studies involving neurologically normal adults have been performed. Boxer et al. (1976) reported




Full Text

PAGE 1

THE EFFECTS OF VALPROATE MONOTHERAPY WITH CARNITINE OR CARNITINE PLACEBO ON ATTENTION AND MEMORY IN A PEDIATRIC IDIOPATHIC EPILEPSY SAMPLE By MARGARET BOOTH-JONES A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1995

PAGE 2

ACKNOWLEDGEMENTS Numerous people were involved in this project. The subjects and their families, and the Departments of Pediatric Neurology, Pharmacy, Biostatistics, and Clinical and Health Psychology were integral parts. To name every member of this extensive team would be impossible; however, the author will not forget the commitment to this project. The author would like to extend her deepest gratitude to Eileen B. Fennell, Ph.D., her chair and mentor, for her inexhaustible knowledge and support. The past six years of warmth and encouragement have profoundly influenced the author and will never be forgotten. Sincere thanks goes to the members of my committee, Russell Bauer, Cynthia Belar, James Rodrigue, and Bernard Maria, without whose support and guidance this study could not have been a success. The author would like to acknowledge the support and prodding of her parents, Pete and Carolyn Booth. Special appreciation goes to the author's sister, Veronica, who provided very much needed distraction and joy. Finally, the author extends her warmest thanks to her husband, Graeme Jones, who patiently understood and listened to the trials and tribulations of completing this project. ii

PAGE 3

TABLE OF CONTENTS ACKNOWLEDGEMENTS. ii ABSTRACT . . . . . . . . . . . . iv CHAPTERS 1 2 3 4 INTRODUCTION. Epilepsy: General Description ............... Migraine: Description and Proposed Mechanism. Relationship Between Epilepsy and Migraine .... Cognitive and Behavioral Deficits .......... Antiepileptic Therapy and Cognitive Function .. Valproate: Pharmacology and Action ... Valproate: Neuropsychological Effects .. Rationale for Further study ............ Specific Aims ....................... METHODS .. Subjects ................ Materials and Apparatus .. Procedures and Design. Hypotheses ................... RESULTS .... Hypothesis One. . . . .. . Hypothesis Two. . .... . . . . . . . Hypothesis Three . . . . .... DISCUSSION 1 2 15 19 21 27 31 53 62 63 64 67 63 76 78 82 83 84 86 93 Interpretation of Findings. Attrition and Non-compliance .. Limitations of Study ...... Conclusions .................. 102 109 ..110 112 REFERENCES ...... .115 BIOGRAPHICAL SKETCH. .125 iii

PAGE 4

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE EFFECTS OF VALPROATE MONOTHERAPY WITH CARNITINE OR CARNITINE PLACEBO ON ATTENTION AND MEMORY IN A PEDIATRIC IDIOPATHIC EPILEPSY SAMPLE By MARGARET BOOTH-JONES May 1995 Chair: Eileen B. Fennell Major Department: Clinical and Health Psychology Twelve children ranging in age from 6 to 16 years, of average intelligence, diagnosed with idiopathic epilepsy and on valproate monotherapy were the subjects for this study. The effect of the valproate on memory and attention was assessed at three intervals over a three-month period, with plasma levels of valproate monitored and maintained throughout the study. After four weeks of the valproate monotherapy, half of the children received carnitine and half received carnitine placebo in a double-blind paradigm. All subjects remained on their regimen for eight weeks and were re-evaluated. The performance of the epileptic children was compared with that of 14 children with migraine also prescribed valproate with carnitine or carnitine placebo, and with that of 20 age-matched normal controls. iii

PAGE 5

At baseline, the epileptic children demonstrated mild impairment with attention and verbal memory as compared with the migraine and normal control groups. A similar pattern of comparative test performance was observed at the 4-week and 12-week intervals. There were no appreciable cognitive changes in the subgroup of children taking carnitine. Most children were in the therapeutic range of valproate and experienced excellent seizure control and minimal negative side effects. A high rate of attrition occurred in the migraine sample (50 percent), greatly limiting further comparison; however, for the most part they maintained an intermediate position on neuropsychological measures between the epileptic and control groups. In this study with this sample, valproate monotherapy did not lead to change in performance on tasks assessing memory and attention. iv

PAGE 6

CHAPTER 1 INTRODUCTION Epilepsy is a common neurological disorder which has been described in the literature for centuries with attributions to mystical forces, deities, and demons for its etiology. In 400 B.C., Hippocrates argued that epilepsy was some form of brain disorder (Bennett, 1992). This line of thinking was not supported until the work of John Hughlings Jackson in the 1800s when the acceptance of epilepsy as a disease process emerged. The differing seizure types were elucidated by several researchers in the late nineteenth century and ~arly twentieth century with great assistance from the development of the electroencephalogram (EEG) by Hans Berger in 1929 (Kolb & Wishaw, 1990). For well over one hundred years, cognitive abnormalities were documented in the epilepsy population and moral and intellectual deterioration were thought to be inevitable (Trimble, 1987). Emil Kraeplin labeled epilepsy as a psychiatric disorder and, though he wrongly diagnosed epilepsy as a form of insanity, he proposed psychiatric 1

PAGE 7

2 treatment for sufferers (Bennett, 1992). Henri Gastaut used EEG to further study seizures and probable neuroanatomical regions of causation (Bennett, 1992). Idiosyncratic personality, and cognitive and behavioral characteristics of adult and child epileptics have been observed and reported throughout the decades of epilepsy research. Pharmacological intervention, which has been utilized since the early twentieth century, has been beneficial to this population for the most part, but not without consequence. Description of the clinical presentation of epileptics on and off antiepileptic medications is an area of intense and essential investigation. Epilepsy: General Description Epilepsy is a chronic convulsive disorder characterized by persistent, recurrent seizures associated with acute and/or chronic psychological changes and a pathological EEG (Rapin, 1982; Kolb & Wishaw, 1990; Engel, 1989). Seizures in the human population are relatively common as 1 in 20 will experience some form of seizure in their lifetime (Rapin, 1982). However, recurrent seizures warranting the diagnosis of epilepsy are less common, affecting an estimated 1 in 200 people (Fenichel,1988). Epilepsy usually develops before adulthood with an estimated 90 percent presenting before age 20 (Spreen et al., 1984). Up to one million children in the United States alone have some form

PAGE 8

3 of seizure experience or disorder (Hartlage & Hartlage, 1989). An estimated fifteen percent of occasional convulsions develop into epilepsy (Spreen et al., 1984). The specific functional and psychological characteristics of epilepsy appear related to age of onset and duration of the disorder, type of seizure and EEG abnormality, frequency of the seizures, and etiology (Trimble, 1990; Rapin, 1982; Engel, 1989). Childhood epilepsy, defined by onset under age 15 years, is generally considered different from adult onset epilepsy in etiology, types of seizures, and frequency of seizures (Spreen et al., 1984). A genetic continuum for susceptibility to seizure disorder has been proposed with a significantly greater chance of a seizure occurrence when a close relative has a seizure disorder (Spreen et al., 1984; Fenichel, 1988; Engel, 1989). Epilepsy: Proposed Mechanism An epileptic seizure is the result of particular changes in the membrane conductance and neurotransmitter activity in the brain, however, the specific physiological and chemical changes continue to be under investigation (Rapin, 1982; Bennett, 1992; Eadie & Tyrer, 1989). In a normal brain, the excitatory and inhibitory synaptic responses and activity of neurons are well modulated; however, in the epileptic the balance of this complex system is disturbed, forming an epileptogenic region in the gray

PAGE 9

4 matter. At the beginning of a seizure, a large hyperpolarization occurs and spreads to other areas of the brain. With epileptogenesis, it has been theorized that either some of the neuronal cells are discharging more readily as a result of membrane changes or there is failure of neurotransmitters to inhibit cellular discharge (Rapin, 1982; Eadie & Tyrer, 1989). When the epileptogenic region fails to be inhibited, there is rapid recruitment of excitability of neighboring cells which increase in excitability leading to seizure presentation. The area of seizure activity may develop with a structural or cellular abnormality resulting in seizure susceptibility, and subsequent seizure activity can further damage the area. Site of origin can be cortical or subcortical and seizure activity can remain relatively focal or involve widespread areas of the brain (Rapin, 1982; Eadie & Tyrer, 1989; Engel, 1989) The type of electrochemical activity, its location, and its pattern of spread define the type of seizure. Briefly, a partial seizure is characterized by slower, localized recruitment of synchronously discharging neurons (Rapin, 1982; Engel, 1989). With generalized tonic-clonic seizures, the tonic phase is characterized by maximum excitation and the clonic phase is the reassertion of inhibition. (Rapin, 1982; Engel, 1989). Without the

PAGE 10

5 inhibition phase, status epilepticus occurs which can lead to cell death and, potentially, death of the patient. With absence seizures, inhibition predominates halting the epileptic's activity. Kindling has been used to explain several phenomena in epilepsy. Kindling is defined as a permanent alteration of postsynaptic membranes of neighboring neurons and efferents of epileptogenic foci resulting from subclinical to clinical seizure activity which leaves the neuron more susceptible to future seizure activity (Rapin, 1982; Engel, 1989). Kindling has been suggested as the mechanism to explain development of mirror foci and secondary generalization of seizure activity. Use of EEG has allowed illustration of the electrical activity in the epileptic's brain during and between seizures. The interictal spike is the marker for epileptogenesis. There is a "depolarization shift" of a given neuronal region leading to synchronous discharge of neighboring neuronal regions which, if severe enough, will lead to the seizure activity (Rapin, 1982; Engel, 1989). To confound the diagnostic issue, EEG abnormalities are sometimes not observed in patients presenting with seizures and EEG abnormalities of various types, including spike waves, have been observed in patients who do not manifest overt behavioral signs of seizure activity.

PAGE 11

6 Epilepsy is not a disease, but rather it is a symptom (seizures) of recurring brain dysfunction which can be caused by trauma, toxicity, drugs, fever, or congenital anomaly. Epilepsy is diagnosed by type and frequency of seizures, EEG abnormality, ruling out of other external causes (e.g., transient metabolic toxicity), and presence of particular behavior patterns associated with epilepsy. Epilepsy is usually treated with antiepileptic drug therapy, and the type of drug prescribed depends on seizure type and physician preference. Pharmacological control may be achieved with one antiepileptic drug (monotherapy) or a combination of antiepileptic drugs (polytherapy), and drug type and dosage may be changed throughout the course of treatment in response to seizure control and side effects. Seizure Classification The classification of seizure type within the epilepsy population has changed in specificity and name in the past decade. The International Classification of Epileptic Seizures provides three broad classes: partial seizures, generalized seizures, and unclassified epileptic seizures (Commission on Classification and Terminology of the International League Against Epilepsy, 1981). Some seizures observed in children do not fit neatly into one of the common diagnostic categories and are labeled unclassified, and some children present a combination of seizures types.

PAGE 12

7 To further characterize the epileptic condition, seizures leading to alteration of consciousness are labeled 'complex' while those that manifest in motor or sensory changes only, without alteration of consciousness, are 'simple'. See Table 1-1. Partial seizures are localized in origin and most often include specific motor, sensory, or psychic changes (Fenichel, 1988; Rapin, 1982; Bennett, 1992). Partial seizures are sub-classified as simple partial, complex partial, and partial with secondary generalization. In simple partial seizures, the seizure manifestation is restricted to motor (e.g., limb movement), sensory (e.g., tingling or numbness), or psychic phenomena (e.g., disruption of ongoing activity). These three categories of partial seizures are further delineated by the presence and progression of the above mentioned symptoms. Previously called psychomotor seizures, partial seizures with psychic alterations are characterized by automatisms such as lip smacking and blinking. strong emotional sensations and sensory hallucination have been reported with some partial seizures. Interictal EEG for patients with simple partial seizures most commonly demonstrates focal spike-and-wave discharges in the involved cortical region, but these are not necessarily consistently present (Engel, 1989). Ictal EEG activity with simple partial seizures varies greatly in

PAGE 13

8 wave type, localization, and spreading, and there is no single characteristic pattern (Engel, 1989). The most common partial seizure disorder in children is Benign Rolandic Epilepsy with the age of onset between 3 and 13 years and the greatest rate between 5 and 10 years of age. This disorder often spontaneously resolves in the teen years (Fenichel, 1988). Most children who have this disorder experience relatively few seizures, most of which occur during sleep. The seizures usually last between 1 and 2 minutes, and when the child wakens he or she may experience paresthesias around the mouth, facial twitching, and speech cessation. There is no loss of consciousness unless the seizure generalizes. The interictal EEG demonstrates unilateral or bilateral high voltage spike discharges in the central or centrotemporal region. Benign Occipital Epilepsy is a relatively uncommon partial seizure disorder observed in children. It is a familial disorder with onset usually occurring in children below 9 years of age (Fenichel, 1988; Engel, 1989). The first symptoms include visual hallucination, transient blindness, hemianopia, and visual illusions. After the paroxysmal episode, symptoms can include a migraine-like headache, nausea, and possible vomiting. This partial seizure can generalize as well. The characteristic EEG is a unilateral or bilateral, high amplitude, spike and wave

PAGE 14

9 discharge in the occipital region unilaterally or bilaterally. The EEG activity can spread during seizure activity to the central and temporal regions (Engel, 1989). This form of epilepsy is often difficult to differentiate from basilar migraine (Fenichel, 1988). Complex partial seizures, focal seizures that spread and lead to alterations of consciousness, are more frequently observed in the older child and adult populations than in younger children. The etiology for complex partial seizures is heterogeneous and can include trauma, neoplasm, and disease (Fenichel, 1988; Engel, 1989; Rapin, 1982). These seizures originate in cortical regions, most often the temporal lobe, but epileptogenesis in the frontal and parietal lobes has been observed (Fenichel, 1989; Bennett, 1992). Many authors interchange the terms complex partial seizures and temporal lobe epilepsy; however, complex partial epilepsy may not necessarily originate in the temporal lobe, but rather affect that brain region or neighboring limbic structures (Eadie & Tyrer, 1989). Complex partial seizures can arise from simple partial seizures or present independently of simple partial seizures, and many patients experience both types of seizures. The complex partial seizure can occur spontaneously or be associated with sleep transitions, and the frequency can vary from one or more per day to one per

PAGE 15

10 year. The duration of complex partial seizures is usually no greater than two minutes. The initial symptom for a complex partial seizure is often the same as for the simple seizure in a given patient, but many report additional sensations including a general, nondescript unpleasant feeling, auditory hallucinations, or abdominal discomfort. Staring, automatisms (e.g., grimacing, finger movements), tonic extension of the limbs, decrease in postural tone, and resistance to restraint have been observed during complex partial seizures. A postictal confusion, amnesia for the event, anterograde amnesia lasting minutes to hours, and post-seizure lethargy occur in most patients (Eadie & Tyrer, 1989; Fenichel, 1988). The EEG pattern for complex partial seizures is usually a single spike or slow wave focus in the temporal lobe or frontal lobe, or multifocal throughout cortical regions (Fenichel, 1988). During ictus there is an EEG progression from single spike discharges in the cortical area of involvement to spike slow wave complexes involving a greater cortical area followed by slow waves with varying amplitude (Fenichel, 1988). Interictal EEGs of patients with complex partial seizures most often demonstrate unilateral or bilateral spikes in the anterior temporal lobe with the similar variation seen in simple partial epilepsy.

PAGE 16

11 Generalized seizures, defined as non-localizing seizures, are classified as absence, atypical absence, myoclonic, clonic, tonic, tonic-clonic, and atonic seizures. The EEG demonstrates bilateral involvement, with epileptogenic origins in both cortical and subcortical regions. Extensive involvement of both hemispheres occurs. Absence epilepsy is a genetic disorder which presents only in children and rarely persists into adulthood. Age of onset is most commonly between 4 and 8 years of age, and more girls than boys are affected in a 60 percent to 40 percent ratio (Fenichel, 1988; Engel, 1989). Absence epilepsy is rarely secondary to neurologic disease or trauma, and children with absence epilepsy are usually healthy otherwise. A typical absence seizure lasts a few seconds to one minute and the frequency varies from occasional seizures to hundreds per day. During an absence attack all ongoing activity ceases and staring and possibly rhythmic eyelid movement occur. When the seizure ends, activity recommences with no postictal confusion. Additional symptoms can include myoclonus, increases or decreases in postural tone, and automatic movements such as picking at clothes or turning of the head. The characteristic ictal EEG pattern for absence seizures is the 3 Hz spike-wave per second bilaterally which is synchronous and symmetric with the greatest amplitude in

PAGE 17

12 the frontal and central regions of the brain. Interictal EEG patterns appear normal in most cases. Absence seizures can be induced with hyperventilation and are more frequent during transition of sleep and wake. Nearly half of all children with absence epilepsy have experienced at least one generalized tonic-clonic seizure (Fenichel, 1988). Myoclonic epilepsy is a form of epilepsy that has numerous sub-types primarily observed with infants, degenerative neurologic syndromes, and following toxic and anoxic events (Rapin, 1982; Engel, 1989; Fenichel, 1988). Myoclonus is involuntary muscle contractions that are brief and repetitive. One type of myoclonic epilepsy is myoclonic absence which has an age of onset between 2 and 12 years. A large proportion, upwards of forty percent, of patients presenting with myoclonic absence have mental impairment prior to seizure onset (Fenichel, 1988). With this form of epilepsy, there is a combination of absence seizures and severe myoclonus in the limbs in the patient. No loss of consciousness occurs. The EEG is most often characteristic of absence with a 3 Hz spike-wave pattern (Fenichel, 1988; Engel, 1989). Generalized tonic-clonic seizures are the most common form of pediatric seizure disorder with onset typically after the neonatal period. Generalized motor seizures include tonic, clonic, tonic-clonic and atonic phases

PAGE 18

13 TABLE 1-1 CLASSIFICATION OF SEIZURE 1. Generalized seizures (bilaterally symmetrical and without local onset) A. Tonic or clonic or tonic-clonic (grand mal seizures) B. Absence (petit mal) 1. Simple (loss of consciousness only) 2. complex -with brief tonic, clonic, or automatic movements c. Lennox-Gastaut syndrome D. Juvenile myoclonic epilepsy E. Infantile spasms (West syndrome) F. Atonic seizures (sometimes w/ myoclonic jerks) 2. Partial (focal) seizures A. Simple partial seizures (consciousness not impaired) 1. Motor (including Jacksonian) 2. Sensory or somatosensory 3. Autonomic 4. Psychic B. Complex partial seizures (w/impairment of consciousness). Includes:cognitive, affective, psychosensory, and psychomotor symptomatology -i.e. "temporal lobe" or "psychomotor seizures" C. Partial seizures (simple or complex) with secondary generalization 3. Unilateral or predominantly unilateral seizures (tonic, clonic, or tonic-clonic, with or without impaired consciousness) 4. Unclassifiable (due to inadequate data) Source: R.D. Adams and M. Victor (1989). Principles of Neurology (Fourth Edition). Page 250.

PAGE 19

14 depending on the progression of motor activity. The first sign of a tonic-clonic seizure is a sudden loss of consciousness often preceded by a sharp cry. The tonic phase is characterized by convulsions affecting the whole body. Brief periods of clonus, jerking caused by relaxation of the muscles, begin to rhythmically interrupt the tonic spasm with increasingly greater duration until the seizure ends. During the seizure, the patient's eyes roll backward, breathing is rapid and deep resulting in increased saliva in the mouth (e.g., frothing), and bowel and bladder incontinence may occur. The increased saliva and changes in the muscle tone of the airway can lead to obstruction of respiration. Tonic-clonic seizures typically last one minute or less, and may be followed by a tonic phase (Engel, 1989). In the postictal phase, muscle flaccidity, disorientation, deep sleep, and headache most often occur. The ictal EEG of the tonic phase is associated with a generation of widespread low-voltage rapid activity referred to as the "recruitment rhythm" (Engel, 1989). This buildup most often is followed by generalized high amplitude polyspike or spike-and-wave phenomena. The clonic phase is represented on EEG by widespread suppression of activity. These patterns alternate throughout the seizure and either dissipate with decreasing amplitude and rapidity or end abruptly (Engel, 1989; Fenichel, 1988). The postictal EEG

PAGE 20

15 of the generalized tonic-clonic seizure patient is characterized by generalized slowing. Partial seizures with secondary generalization to tonic-clonic seizures may show focal activity on the postictal EEG in the area of the focal involvement (Engel, 1989; Fenichel,' 1988). Migraine: Description and Proposed Mechanism Migraine is classified as a paroxysmal neurological disorder involving alterations in cerebral blood flow that may occur alone or secondary to another disease or disorder. Migraine is reportedly the most common neurological disorder occurring in 5 to 20 percent of the population and the most common cause of severe headache in children (Kolb & Wishaw, 1990; Fenichel, 1988). Migraine is most often a hereditary disorder, with a prevalence of 2.5 percent for children under 7 years of age, 5 percent for age 7 to pubescent children, 5 percent of post-pubescent males, and 10 percent of post-pubescent females (Fenichel, 1988). Prevalence by gender appears nearly equal in younger children, but females with migraine outnumber males 3 to 2 in the range from age 7 to pubescence (Fenichel, 1988). Migraine occurrence can be triggered by stress, exercise, head trauma, allergy, diet and estrogen levels (Fenichel, 1988; Sacks, 1985). The exact mechanism of migraine is not known. Vascular changes have been reported during and between migraine attacks (Wolff, 1985). Between migraines, asymmetric

PAGE 21

16 vascular reactivity has been observed. There is an apparent decrease in cerebral blood flow during the prodromal phase and an increase in extracranial blood flow during the migraine. The most common hypothesis involves vasoconstriction beginning in one or more cerebral arteries, usually in the posterior region of the brain (Sacks, 1985; Wolff, 1985). This decrease in blood flow progresses anteriorly, though not along a particular pathway, and usually affects both hemispheres. The severe pain of migraine begins with the onset of vasodilation. Differences in platelet aggregation have been observed in association with migraine (Wolff, 1985). Specifically, platelet aggregation has been observed to be more rapid in migraineurs than normal controls. More platelet aggregation has been observed in the prodromal phase of the migraine with a decrease in the headache phase. Serotonin in the plasma, which is related to platelet concentration, appears to increase before the migraine attack and decrease during the headache phase. Serotonin associated with platelet concentration can lead to constriction of arteries or large vessels and dilation of arterioles and capillaries (Theisler, 1990). Plasma serotonin levels may reflect changes in central nervous system serotonin which affect two pathways associated with migraine.

PAGE 22

17 Migraine can vary in presentation, intensity, frequency, and duration and can occur unilaterally or bilaterally. Migraine is frequently associated with anorexia, nausea and vomiting, and neurological and mood disturbances. Migraine is classified as: classic, common, cluster headaches, hemiplegic, and ophthalmoplegic. Classic migraine affects 12 percent of the migraine population and has a distinctive aura, most likely related to the ischemia in the occipital cortex (Fenichel, 1988). The classic migraine occurs in two phases, the first of which involves the progression of excitation of brain function from posterior brain regions to anterior brain regions with decreases of regional cerebral blood flow and transitory neurologic disturbances (Fenichel, 1988; Wolff, 1985; Sacks, 1985). Common neurologic symptoms in this first phase, also called the prodromal phase, are visual abnormalities, dysthesias of the mouth area and limbs, and occasionally motor changes. The second phase is characterized by headache and nausea and sometimes vomiting. The frequency of classic migraines is most often several times per month with a duration of several hours (Fenichel, 1988; Wolff, 1987). Common migraine does not have two distinct phases and an aura is not associated with this classification of migraine. Visual changes may occur, but the more frequent

PAGE 23

18 symptoms include changes in personality, dizziness, general malaise, and nausea and vomiting in conjunction with headache (Fenichel, 1988; Wolff, 1987). There is a high familial incidence for cluster headaches and they are more prevalent in males than females with onset most often by age 10 years (Fenichel, 1988). The headache most often originates behind one eye and extends along the same hemicranium. These tend to last 30 minutes to 2 hours (Fenichel, 1988; Wolff, 1987). Hemiplegic and ophthalmoplegic migraines are classified as complicated migraines, defined as presentation of focal motor deficits during a classic migraine (Fenichel, 1988). Hemiplegic migraines are characterized by hemiparesis followed by a throbbing frontotemporal headache located contralateral to the hemiparesis. Nausea and vomiting usually occur. Ophthalmoplegic migraines present with unilateral eye pain and ipsilateral ocular motor palsy usually preceded by ptosis which can last for days or weeks (Fenichel, 1988). There is a higher rate of EEG abnormality in patients with migraine than normal population, but not all patients with migraine have EEG abnormalities (Fenichel, 1988; Wolff, 1987). Unilateral slow wave focus in the temporal lobe has been observed in patients with classic migraine. Between migraine attacks, EEG is usually normal (Fenichel, 1988).

PAGE 24

19 Relationship Between Epilepsy and Migraine The relationship between migraine and epilepsy has been the focus of numerous medical arguments and the subject of relatively little research. Some suggest that the two may be related and others believe the two are completely distinct. Regardless, there appear to be several common features which have posed difficulty for diagnosticians in certain cases. Fenichel (1988) lists the following four features which most strongly suggest a relationship between epilepsy and migraine. (1) Both disorders can have a strong familial component, are paroxysmal and are associated with transient neurologic disturbances. (2) There is an increased incidence of migraine in epileptics and an increased incidence of epilepsy in migraineurs. (3) Headaches are associated with some forms of epilepsy and in most forms of migraine. (4) Abnormal EEG activity has been observed in both populations and some specific EEG patterns are common to both. Additionally, both neurologic disorders can occur in individuals with no apparent cause such as a lesion or traumatic event or be part of the neurologic sequela of head trauma. Occurrence of epileptic seizures and migraines can be spontaneous or be triggered by external factors and internal factors. Both epileptic and migrainous events occur in stages during which time the individual experiences

PAGE 25

20 changes in mental status with or without physical changes. For example, the epileptic and migraineur can experience an aura, alterations in consciousness, changes in postural tone, motor changes, hallucinations, changes in mood and changes in higher cortical functions {Sacks, 1985). There are differences between these two disorders. On the whole, the auras experienced by the migraine population are more visual and often include the occurrence of scotoma. Convulsions and loss of consciousness are rare in migraine. The intensity of alterations of mood and higher cortical function is generally greater with epilepsy than migraine. Focal cerebral symptoms and EEG abnormalities have been observed in children with migraine with and without seizure activity which has lead to some difficulty with differential diagnosis (Fenichel, 1988). The greatest similarity is reported between temporal lobe epilepsy and classic migraine especially in children (Engel, 1989). Migraine and epilepsy can occur in the same individual and are not mutually exclusive. Benign occipital epilepsy and benign rolandic epilepsy are associated with migraine. Basilar artery migraine can lead to alterations and loss of consciousness and seizure. Partial-complex seizures have been noted to be followed by vascular headaches. The relationship between epilepsy and migraine is tenuous at best; however, there is some overlap in the symptoms, neuroanatomy, and treatment.

PAGE 26

21 Cognitive and Behavioral Deficits Trimble (1987) stated that the relationship between abnormal cognition and epilepsy has been commented on for well over one hundred years with the consensus that moral and intellectual deterioration in the epileptic were inevitable. Today, there is general agreement in the field of epileptology that cognitive deficits are over-represented in children with epilepsy in comparison with neurologically healthy children (Trimble, 1990). These cognitive deficits have been attributed to gross structural abnormalities, subtle structural abnormalities, drug effects, interference from the actual seizure activity in the brain, social and emotional difficulties, and loss of time in school (Stores, 1971; Trimble, 1990). Risk factors for determining cognitive impairment include age of onset, duration of seizure disorder, seizure type, level of seizure control, medication, and related metabolic changes (Rapin, 1982; Trimble, 1990, Engel, 1989). For individual patients, some or all of these factors may play a role in their relative cognitive abilities. The research examining cognition and pediatric epilepsy is often difficult to interpret and to generalize due to several crucial study design flaws (Stores, 1971). The groups of children studied have often been heterogenous covering wide age ranges, multiple seizure types, great

PAGE 27

22 variance in the severity of seizures, numerous combinations of seizure medications, and wide ranges of intellectual and physical disabilities. Many of the early studies focused on intelligence quotients (IQs) as the sole measure of cognitive ability. The prevalent messages from these studies indicate that, as a patient population, children with generalized epilepsy have attentional deficits, but are relatively free of memory impairment (Trimble, 1987). Involvement of subcortical structures and their connections with higher cortical regions have been implicated in the attentional difficulty. Children with partial epilepsy, specifically with temporal lobe involvement, demonstrate a higher rate of verbal memory deficits with left hemisphere foci and nonverbal memory disability with right hemisphere foci (Trimble, 1987). Children with epilepsy have been observed to have a variety of school difficulties; learning and behavioral problems are over-represented in the pediatric epilepsy population (Trimble, 1990). There is a greater variability in the intellectual level (IQs) in this patient group (Seidenberg, 1986). Teachers have described these children as having poorer concentration, slower mental processing and reduced alertness as compared with non-epileptic children. Attentional problems, poor memory, poor social skills, and

PAGE 28

23 emotional problems have been attributed to be both the cause and the result of epilepsy and interaction with medication. Reading difficulty and reading impairment are more frequently observed with these children (Trimble, 1990). Hermann (1982) studied 50 children with epilepsy aged 8 to 12 in regular school placement. The children were administered the Luria Nebraska Neuropsychological Battery -Children's Version by Golden (1981) comprised of 149 items in the following 11 scales: Motor, Rhythm, Tactile, Visual, Receptive Language, Expressive Speech, Writing, Reading, Arithmetic, Memory, and Intellectual Procedures. The Child Behavior Checklist by Achenbach (1986) was given to the parents to assess behavior. According to their test performance, the children where separated into a "good neuropsychological" and "poor neuropsychological" group along the median number of scaled scores below a T score of 60. Those in the "poor neuropsychological" group were associated with reports of increased psychopathology, more aggression, and decreased social competence. Interpretation and generalizability of these results are limited as this study did not describe the types of epilepsy, control for or even mention the antiepileptic medication, describe the pattern of deficits, or employ a normal control group. Farwell, Dodrill, and Batzel (1985) studied 118 epileptic children between the ages of 6 and 15 with a

PAGE 29

24 variety of seizure types to determine which aspects of the seizure disorder were associated with cognitive impairment. Children with IQs below 50 and serious behavioral disturbance were excluded. Though all children were receiving some form of antiepileptic medication, medication was not a variable in the study. The WISC-Rand the ageappropriate version of the Halstead Reitan Battery were the measures employed. One hundred normal children acted as controls. Farwell and her colleagues found significant differences in Full Scale IQ with the mean for epileptics at 93.16 and the mean for normals at 100.00. Epileptic children with later onset of their seizure disorder had higher IQs. Longer duration of seizure disorder was associated with lower IQ scores. Children with better seizure control had a mean IQ of 99.56, and those with less seizure control had a mean IQ of 86.96, which is significantly lower. The children were grouped by seizure type: classic absence (n=S), both classic absence and generalized tonic clonic (n=S), generalized tonic clonic (n=31), partial seizures only {n=31), partial and generalized seizures (n=20), atypical absence (n=l5), and minor motor seizures (n=5). Significant differences in IQ between these seizures groups were found with minor motor, IQ=70, and atypical absence, IQ=74. The remaining seizure

PAGE 30

25 groups had mean IQs between 96 and 100 (average range). The neuropsychological impairments were reported as a combined rating along the continuum of neuropsychological impairment rather than specific deficits. The rating was assigned by one of the examiners according to the median number of impairments with "good" rating above the median and "bad" rating below the median. The minor motor and atypical absence children demonstrated the most severe neuropsychological impairment. The other groups demonstrated relatively equivalent proportions of neuropsychological impairment, most of which was in the mild and very mild range. The classic absence group demonstrated the least amount of neuropsychological impairment. This study provides strong evidence for IQ differences by seizure type. This study also provides convincing evidence against combining numerous seizure types when exploring specific aspects of childhood epilepsy. This study adds strength to the premise that significant variations among children with epilepsy exist. Farwell et al. (1985) excluded children below an IQ of 50, therefore including many children with IQ scores below the average level which is problematic when comparing performance to a normal control group in attempts to discern specific cognitive patterns of an epileptic sample. The primary weakness of this study is the failure to report and

PAGE 31

26 describe neuropsychological results in any form of detail. As a result, group differences in specific motor, visual, and language domains, for example, can not be discerned from the data provided. It is not surprising that the epileptic group demonstrated more impairment on the Halstead Reitan as intelligence and this measure are strongly correlated and the sample was biased to the below-average range of intelligence. Memory measures were not employed. Another limitation of this study is the lack of information about anticonvulsant medication. Binnie, Channon, and Marston (1990) investigated the relationship of EEG activity and learning difficulties through literature review of studies in which neuropsychological tests (not specified) were administered during EEG recordings. All EEG readings were reported to be subclinical spike-wave activity. Generalized subclinical EEG activity was associated with impairments of attention; left focal activity was associated with poor performance on verbal tasks, and right focal activity was associated with poor performance on nonverbal tasks. Binnie et al. (1990) suggest that the EEG activity in the epileptic interferes with learning at an acute level and at a chronic level. Acutely, with loss or alteration of consciousness ictally with postictal confusion, seizures impair learning with disruptions in the processes of elaboration, storage, and

PAGE 32

27 retrieval of information. Chronic focal EEG activity may go undetected and cause specific areas of brain damage and subsequent learning problems. Chronic subclinical EEG activity is theorized to reduce the total capacity for learning. Binnie et al. (1990) further suggest that epileptics with gross lesions and generalized seizures have more impaired learning than epileptics with focal seizures, and deficits are greater for patients who demonstrate multiple seizures types. The five major findings of this study regarding the effects of subclinical EEG activity on learning suggest that this can: 1) result in the loss of immediate information; 2) disrupt consolidation; 3) cause permanent damage to neural tissue; 4) alter brain function; and 5) be reduced with antiepileptic drug treatment which also may damage neural tissue. Antiepileptic Therapy and Cognitive Function Most forms of epilepsy are treated with antiepileptic medications designed to reduce the number and severity of seizures; however, side effects have been reported with all antiepileptic drugs and the specific nature varies according to the drug, dosage, combination with other medication, and individual patient characteristics. Phenobarbital, a barbiturate, has been widely used in the pediatric epilepsy population (over 50 years) with some

PAGE 33

28 deleterious side effects including slower reaction times and difficulty sustaining attention (Trimble, 1987; Smith, 1991). Drowsiness and dysarthria have been reported, as well as hyperactivity and excitation in children (Engel, 1989; Fenichel, 1988). Phenobarbital is prescribed for tonic-clonic and simple partial seizures (Fenichel, 1988). Phenytoin has been associated with a general decline in mental ability. Hirsutism, facial rashes, gum hypertrophy, and decreased attention span have been observed with phenytoin therapy (Fenichel, 1988). Drowsiness, dizziness, anorexia, hyperactivity, personality changes, and progressive encephalopathy have also been reported (Engel, 1989). Impairments in memory, motor speed, and mental speed have been observed as well (Trimble, 1987; Smith, 1991). More severe impairments are associated with higher serum levels of the drug. Phenytoin is most frequently used in patients with partial seizures, tonic-clonic seizures, and status epilepticus (Engel, 1989; Fenichel, 1988). Primidone is used in the management of generalized tonic clonic seizures, complex partial and partial seizures (Fenichel, 1988). Its central nervous system side effects include drowsiness, vertigo, ataxia, lethargy and some behavior changes. Rash, nausea and vomiting, hepatic disorder, and hematologic changes have been reported. Primidone has two active metabolites, one of which is

PAGE 34

29 phenobarbital. Intolerable sedation from the first dose is one of the major adverse side effects (Smith, 1991). Carbamazepine has been available for nearly thirty years, and several studies have shown that many epileptics perform better on cognitive tests when they are switched from the other antiepileptic drugs or polytherapy to carbamazepine monotherapy, though whether the improvements are related to discontinuation of the sedative antiepileptic drugs or to the carbamazepine is not fully elucidated. Improvements in attention and speed have been observed (Trimble, 1987). Carbamazepine appears to be the drug of choice for partial and complex partial seizures. Deleterious side effects include depression, drowsiness, irritability, anorexia, and personality change (Engel, 1989). High levels can lead to ataxia, nystagmus, cognitive impairment, hyperactivity, and excessive salivation (Fenichel, 1988). Succinimides and benzodiazepines are other classes of drugs used to control seizure disorder. Ethosuximide is used to treat absence seizures, and nausea and vomiting are the most common side effects. Diazepam and clonazepam are benzodiazepines used to treat status epilepticus, myoclonic seizures, and other forms of epilepsy (Fenichel, 1988). Sedation, changes in behavior, and memory dysfunction have been reported (Smith, 1991).

PAGE 35

30 TABLE 1-2 PHYSICAL SIDE EFFECTS OF COMMON ANTIEPILEPTIC MEDICATIONS Medication 1. Phenobarbital 2. Phenytoin 3. Carbamazepine 4. Primidone 5. Ethosuximide 6. Valproate 7. Clonazepam 8. Diazepam Physical Side Effects Drowsiness, gastric pain, rash Rash, gum hypertrophy, ataxia, hirsutism, coarsening of facial features, peripheral neuropathy Rash, nausea & vomiting (n&v), neutropenia,liver involvement Drowsiness, ataxia, behavior disorder, anemia, rash, n & v Sedation, dizziness, behavior changes, n & v, anorexia, Drowsiness, alopecia, liver enzyme increase, n & v Drowsiness, rash, n & v, anemia Drowsiness Source: I. Rapin (1982). Children with Brain Dysfunction. Page 192. TABLE 1-3 COGNITIVE SIDE EFFECTS OF COMMON ANTIEPILEPTIC MEDICATION Medication 1. Phenobarbital 2. Phenytoin 3. Carbamazepine 4. Primidone 5. Valproate 5. Clonazepam 6. Diazepam Cognitive Side Effects sedation; hyperactivity; impaired vigilance and verbal learning memory impairment; decreased attention span; personality change; psychomotor slowing little effect; less sedation; mild psychomotor slowing similar to phenobarbital no sedation; little effect sedation sedation Source: D. B. Smith (1991). Cognitive Effects of Antiepileptic Drugs. Advances in Neurology, Vol 55. edited by D. Smith, D. Treiman, and M. Trimble. Raven Press, Ltd. New York.

PAGE 36

31 New antiepileptic drugs are being developed, including gabapentin, felbamate, lamotrigine, and vigabatrin (Fisher, 1993). Gabapentin is currently used predominantly in conjunction with other medications in the treatment of complex partial seizures and partial seizures with secondary generalization. Side effects include somnolence, fatigue, ataxia, and dizziness. Long-term effects on physical and cognitive function are not known. The mechanism of gabapentin is not known though it is similar in structure to the neurotransmitter GABA (Goa and Sorkin, 1993). Felbamate is used in monotherapy in adults with partial seizures and in children with Lennox-Gastaut syndrome (myoclonic-astatic epilepsy associated with mental retardation) (Dodson, 1993). Adverse side effects include mild gastrointestinal complaints, weight loss, insomnia, and non-specific nervous system complaints. However, these side effects have been reported as mild to moderate in most cases (Graves, 1993). No studies involving the effects of these new antiepileptic drugs on cognition of children or adults are available at this time. Valproate: Pharmacology and Action Sodium valproate (sodium di-n-propylacetate) is one of the most recent additions to the antiepileptic drug arsenal and has been the subject of relatively few studies. It was identified as an anticonvulsant in 1963 and was allowed for

PAGE 37

32 use with seizure patients in 1978. Some clinical trials suggest that it is most effective with absence, generalized motor, and myoclonic seizure patients and less effective with partial and partial complex seizure patients. However, in adults and older adolescents, it has shown significant promise for partial seizures with secondary generalization as well as being an adjunct to other antiepileptic drugs (Wilder, 1987). Valproate is water soluble, unlike most other antiepileptic drugs, and it is rapidly absorbed after oral administration and reaches a peak plasma level within two hours. Valproate has a half-life of six to ten hours and the efficacy range is 50 to 100 micrograms per milliliter. There is evidence of long lived metabolites of valproate which enhance seizure control and they are not easily detected with monitoring of valproate plasma levels (Gram et al., 1979). The dose plasma level ratio of valproate, critical information for assessing therapeutic range, has been the focus of several studies. Levy (1984) studied the leveldose ratio of valproate in monotherapy and polytherapy conditions. Levy reported previous findings that valproate does not have a characteristic level dose ratio leading to great inter-patient variability with the dose explaining less than 40 percent of the variability in the rate of

PAGE 38

33 clearance. The plasma level of valproate increases with the dose ingested; however, the relationship appears to be curvilinear. This finding was replicated by Gram, Flachs, Wurtz-Jorgensen, Parnas, and Andersen (1979). Similarly, Armijo, Herranz, Arteaga, and Valiente (1986) report a poor correlation coefficient for plasma concentration and dose ratios. The plasma level is dependent on dose absorbed, dosing interval, and rate of clearance. Valproate provides nearly 100 percent bioavailability to the patient with most absorption in less than 2 hours, though meals, enteric coating, and compliance are crucial factors for measuring absorption. The time intervals for dosing and sampling provide a great proportion of the inter-patient variance. Polytherapy shortens the half-life of the valproate. Levy (1984) reports that the age of the patient and polytherapy are the most important factors for determining the clearance rate. Children have a higher clearance than adults, with most of the variance occurring in patients below 7.5 years. Especially in children, valproate monotherapy is recommended for better monitoring of drug plasma level. Johnston (1984) analyzed available research literature and reported three theoretical mechanisms for valproate's effect on the nervous system. The first proposed mechanism suggests that valproate increases the level of the

PAGE 39

34 inhibitory neurotransmitter gamma-aminobutyric acid (GABA) in the brain. The second theory proposes that valproate potentiates the postsynaptic GABA response. The third mechanism described in this paper suggests that valproate has a direct effect on the neuronal membranes in the brain. Johnston repeatedly stated that none of these theories have been consistently supported by research and several studies using animal subjects directly refute crucial aspects of these proposed mechanisms. However, there does appear to be a relationship between valproate and the inhibitory neurotransmitter GABA, the specific nature of which is unknown. Gamma-aminobutyric acid, an amino acid derived from glutamate, is a ubiquitous inhibitory neurotransmitter released at the inhibitory synapses. The full role of GABA is not known; however, GABA is purported to help maintain the electrical balance of the brain in conjunction with other neurotransmitters. Chemicals which increase GABA appear to increase neural inhibition and drugs that decrease GABA decrease neural inhibition, altering the neurochemical stability of the brain (Engel, 1989; Enna & Beutler, 1985; Eadie & Tyrer, 1989). GABA-ergic neurons in the human brain are found in the granule cells of the olfactory bulb, amacrine cells of the retina, Purkinje and basket cells of the cerebellum, basket

PAGE 40

35 cells of the hippocampus, many striatal cells, and numerous interneurons. The GABA-ergic neurons of the caudate nucleus and putamen (striatum) project to the substantia nigra and the globus pallidus (Noback, strominger, & Demarest, 1991). GABA appears to have an integral role in the function of cortical/ subcortical pathway function, the extent of which is not fully elucidated. GABA has been implicated in having a role in epilepsy, with the vast majority of data coming from animal research (Eadie & Tyrer, 1989, Enna & Beutler, 1985). By inducing seizure activity in mice, rats, cats, and baboons through biochemical, pharmacological, and neuroanatomical techniques, alterations in GABA and its related metabolites have been observed. The current findings suggest that lower levels of GABA are related to seizure activity. A study of GABA levels in the cerebral spinal fluid (CSF) in humans found lower levels in patients with generalized and secondarily generalized seizures in contrast to partial seizures and lower GABA levels in children with febrile convulsions when compared to normal children (Schmidt and Loscher, 1981). Generalizations from the animal and CSF studies to the activity of GABA in the epileptic brain warrant strong caution. For instance, GABA levels in the CSF may not directly reflect GABA levels in the brain. Drawing direct

PAGE 41

36 correlations from animal models to human brain function is not possible due inherent anatomical and physiological differences. However, the relationship between GABA levels and seizure activity appears to have modest scientific foundation. Several classes of antiepileptic medications appear to be indirect GABA agonists (Enna & Beutler, 1985). Barbiturates affect the chloride ion channels by prolonging the availability of GABA. Benzodiazepines bind at specific neuron sites and increase GABA receptor sensitivity. Valproate inhibits two enzymes in the metabolism of GABA, specifically GABA transaminase and succinic semialdehyde dehydrogenase, leading to prolonged availability of GABA. Several researchers posit that though certain classes of antiepileptic drugs control seizure activity presumably through the GABA system, the effects on the CNS are not limited or contained within the epileptogenic region. If the mode of seizure control is to alter GABA levels, these neurotransmitter changes should occur throughout the brain. Again, the extent of influence at the biological and at the cognitive levels requires further study. Valproate has been observed, again through animal models, to influence other neurotransmitters. Increases in the amino acid neurotransmitters taurine and glycine and a reduction of aspartate have been recorded following

PAGE 42

administration of valproate (Eadie & Tyrer, 1989). Valproate appears to have no direct influence on serotonin levels; however, tryptophan, a precursor of serotonin appears to be increased by valproate (Hwang & Van Woert, 1979; MacMillan, 1979). Valproate: Studies on Normals 37 An early study by Boxer, Herzberg, and Scott (1976) assessed the behavior of 8 normal males using a single dose double-blind placebo controlled design, with the subjects receiving 400mg valproate, 60mg phenobarbital, the 2 drugs combined or placebo. The subjects were observed behaviorally and with EEG. Boxer et al. reported that the single dose of valproate tended to cause drowsiness which worsened with the addition of phenobarbital; the subjects on valproate and valproate plus phenobarbital tended to fall asleep if they were left alone. Neither standardized data nor empirical results were reported. Thompson and Trimble (1981) studied 10 normal adults (mean age 26 years) in a double blind cross-over paradigm. The subjects were given valproate in a therapeutic range or a placebo. The following cognitive domains were assessed: verbal memory, nonverbal memory, concentration (Stroop), perceptual speed (minimum duration of exposure of targets flashed on a screen), simple decision making (40 yes/no trials for color and category), motor speed (finger

PAGE 43

38 tapping), and mood. Verbal memory and nonverbal memory were assessed for immediate and delayed memory for 20 words and 20 pictures, respectively. Recognition memory was assessed with the addition of 20 distractor stimuli to each array. Mood was assessed with the Mood Adjective Check List by Lishman (1972). Using multiple paired t-tests, the one significant finding between the valproate and placebo condition was slower decision making in the valproate condition. These researchers noted that improvements in cognitive ability would be difficult to observe in normal adults given their pre-drug normal range of intellectual functioning. This study has very small sample size and the duration of drug exposure was relatively short (2 weeks). Given the large number of dependent variables and the small sample size, use of multiple paired t-tests is susceptible to a Type I statistical error. Harding, Alford, and Powell (1985) studied 10 normal adult volunteers, mean age 24 years, in three valproate conditions: low dose, high dose, and after withdrawal of valproate. Simple reaction time ("yes" button to light flash), visual evoked potentials, and EEG's during sleep were assessed in all three conditions. A decrease in REM sleep and an increase in delta activity were observed in the EEG's after valproate withdrawal.

PAGE 44

39 This study has a small sample size (n=lO), and data for the simple reaction time task was obtained for only 6 of the 10 volunteers due to technical difficulties. Duration of the valproate was 4 days for the low dose and 14 days for the high dose, which may limit generalizability to the epileptic population who are receiving the medication over significantly longer periods of time. studies of valproate on normal children have not been conducted for obvious ethical reasons. The results of these three studies involving normal adult subjects do strongly suggest that valproate has some physiological effect on the human nervous system, the full extent of which is not clear. The specific findings from brief drug trials of drowsiness, slower decision making, and sleep EEG changes are not directly generalizable to the pediatric epilepsy population due to age and developmental differences, confounding factors of illness, polytherapy, lesions and trauma, and length of exposure to valproate. Valproate: Side Effects As with any drug, both the benefits and the risks to the intended patient need to be examined and weighed when deciding its usage. The Physician's Desk Reference (1992) reports that hepatic failure is the most significant and problematic risk with valproate monotherapy; however, overall incidence of hepatic failure is rare (Fenichel,

PAGE 45

40 1988). Tremor and sedation have been observed. Additional central nervous system disturbances include ataxia, headache, nystagmus, diplopia, and incoordination. These symptoms are relatively infrequent and they can be reduced or eliminated with dose alteration. Several studies have focussed on defining the side effects of valproate. Herranz, Arteaga, and Armijo (1982) studied 88 pediatric epileptic patients receiving valproate monotherapy. The plasma level of the valproate was monitored and maintained in a therapeutic range defined as 40 to 90 micrograms per milliliter for the minimum range and 75 to 150 micrograms per milliliter for the maximum range. These researchers found the incidence of valproate toxicity to be lower than other antiepileptic drugs. Of the 88 patients, 42 percent reported some form of side effect through patient, parent, or physician report. However, with the use of a very detailed questionnaire, 80.7 percent reported the occurrence of side effects. The most frequent complaints were anorexia (8.0%), vomiting (10.2%), and sleep alterations including sleeping too much and difficulty sleeping (13.6%). Lassitude and drowsiness were associated with higher plasma levels, but the other side effects did not show a positive correlation with higher plasma levels. Neurological alterations, though relatively rare, included paresthesias (2.3%), tremor (1.1%), and ataxia (1.1%), and

PAGE 46

41 were not associated with higher doses. Other rare side effects included polydipsia, polyuria, diaphoresis, and enuresis. To counteract the gastrointestinal side effects (e.g., anorexia, abdominal pain, vomiting, nausea, diarrhea, and constipation), the pharmaceutical preparation was altered without alteration of the dosage. Nine patients were discontinued from the valproate therapy due to negative side effects which did not resolve with pharmaceutical alteration. The use of the questionnaire detected significantly more behavioral side effects, with irritability and hyperactivity having the greatest frequency. Herranz, Armijo, and Arteaga (1988) studied 392 epileptic and febrile convulsant patients under 15 years of age. They compared the rate of onset of side effects of phenobarbital, primidone, carbamazepine, and valproate monotherapy. Good tolerance for all drugs in the normal therapeutic range was reported. The rate of side effects for patients treated with valproate was 43 percent which is considerably lower than rates for phenobarbital and primidone, and equal to the rate for carbamazepine. As in their earlier study, digestive tract problems, such as nausea and vomiting, appetite changes, and abdominal pain, were the most commonly reported side effects for valproate, followed by behavioral changes including excitability,

PAGE 47

depression, and hyperactivity. One child demonstrated ataxia and tremor. 42 Schmidt (1984) affirmed that many of the side effects associated with valproate therapy do not appear to be dose dependent. He reported the summary of 16 valproate trials involving 1140 epileptic patients. The side effects reported at the various sites were described as generally mild and transient, and appearing in the early stage of the valproate treatment. A significant increase in drowsiness was reported for patients receiving valproate in conjunction with phenobarbital. Valproate appears to increase the plasma concentration of phenobarbital. Schmidt reported a significant reduction of gastrointestinal complaints with the use of enteric coated valproate preparations. A small number of patients reported a reversible tremor (1.0%), increased coarseness of hair (1.0%), and blood disorders (0.1%). Covanis, Gupta, and Jeavons (1982) studied the relative effectiveness and side effects of valproate on different seizure types. They reported 80 percent of the generalized seizure patients (absence, myoclonic, and primarily generalized) were seizure free. Forty-seven percent of the partial seizure patients were free of seizures over a three year period. Covanis et al. (1982) stated that valproate is most effective and least problematic to the epileptic

PAGE 48

43 patient when given as monotherapy. Approximately three percent of these patients developed changes in their liver function determined by abnormal levels of certain hepatic enzymes, specifically aspartate, alkaline phosphatase, and bilirubin. They further stated that, though quite rare, severe hepatic and pancreatic disorders were reported. Dreifuss, Santilli, Langer, Sweeney, Moline, and Menander (1987) reported a retrospective study of all known hepatic fatalities secondary to valproate treatment from 1978 to 1984 in the United States. They analyzed all reports made to the drug company (Abbott Laboratories) which came from medical personnel, pharmacists, and patients. Thirty-seven of the 47 reports of liver fatalities were determined to be coincident to valproate therapy. The patients who died ranged in age from 5 months to 71 years, and all but one patient had significant medical disorders in addition to seizure disorder (e.g., other neurologic disease and congenital abnormalities). Five patients were receiving valproate monotherapy and 32 patients were receiving valproate as an adjunct to other antiepileptic therapy, most frequently phenytoin and phenobarbital. During 1978 and 1984, an estimated 400,000 people in the United States were taking valproate, 47 percent as monotherapy and 53 percent in polytherapy.

PAGE 49

44 Dreifuss et al. (1987) described the mechanism of liver failure associated with valproate as liver necrosis caused by the production of aberrant, toxic metabolites in susceptible individuals. They distinguished dose related hepatotoxicity from idiosyncratic hepatotoxicity. They described the dose related hepatic changes as relatively harmless and characterized by an increased concentration of the enzyme serum transaminase. They stated that this abnormality is usually asymptomatic and it is reversible with drug discontinuation. The idiosyncratic hepatotoxicity was described as unpredictable and was observed in both the early and later stages of valproate therapy with symptoms of fever, rash, nausea, vomiting, edema, jaundice, and lethargy. They reported great variation in the liver abnormalities among the cases analyzed. Risk factors associated with fatal liver toxicity from valproate therapy are age, with children under 2 years most vulnerable; pre-existing medical conditions; and valproate as an adjunct in polytherapy. The incidence of fatal liver failure associated with valproate therapy was estimated to be 1:500 for children under age 2 years on polytherapy; 1:7,000 for children under age 2 years on monotherapy; 1:12,000 for people above the age of 2 years on polytherapy; and, 1:37,000 for people over 2 years on monotherapy.

PAGE 50

45 Carnitine Supplementation with Valproate Monotherapy One of the physiological side effects of valproate therapy is the reduction of free carnitine concentrations in the plasma. Carnitine is an essential nutrient found in the diet (meat and dairy products). It is essential for fatty acid metabolism (Stedman's Medical Dictionary, 1982) and mitochondrial function (Coulter, 1991). Some research indicates a possible relationship between low carnitine levels and liver disease caused by increased concentration of liver enzymes and fatty deposits in the liver cells; however, this evidence is limited and inconclusive and the abnormalities are at subclinical levels in most patients (Fenichel, 1988; Coulter, 1991; Beghi, Bissi, Codegoni, Trevisan, & Torri, 1990). Carnitine deficiency in children also presents as a primary inborn metabolic disorder, and carnitine supplementation is the accepted treatment (Fenichel, 1988; Coulter, 1991). Melegh, Kerner, Acsadi, Lakatos, and Sandor (1990) determined that the plasma level of valproate and overall seizure control were not affected by carnitine supplementation in a sample of 10 epileptic children. Limited research has been conducted to discern the full effect of carnitine deficiency observed during valproate therapy. Stumpf, Parker, and Angelini (1985) described a three-step hypothesis for carnitine changes associated with

PAGE 51

46 valproate therapy. (1) Carnitine acts as a buffer against toxic acyl CoA (a metabolic intermediate needed for fat oxidation); (2) with decreased levels of carnitine, toxic levels of acyl CoA occur which impair the citrate cycle, gluconeogenesis, the urea cycle, and fat oxidation; and (3) carnitine supplementation leads to greater excretion of the acyl CoA through urine excretion. Polytherapy for epilepsy treatment, pre-existing liver damage, age below 5 years, poor nutrition and significant neurologic disabilities other than the epilepsy appear to increase the risk for problems thought to be related to decreased carnitine levels (Coulter, 1991; Sugimoto, Nishida, Murakami, Woo, Sakane, Yasuhara, Shuto, Hatanaka, & Kobayashi, 1990; Opala, Winter, Vance, Vance, Hutchinson, & Linn, 1991) To date, no study reporting the interactive relationship of valproate and carnitine on the cognitive and psychological status of children has been reported in the literature. Two adult patient groups, Alzheimer's dementia and chronic alcoholism, have been the subjects of a small number of studies in which carnitine supplementation was administered as a "cognition enhancer". Due to the limited nature of these studies, the age and medical confounds, nutritional variables, and lack of consensus, these studies will not be dealt with in this paper.

PAGE 52

47 Freeman, Vining, Cost and Singhi (1994) studied the effect of carnitine on the symptoms associated with antiepileptic medication. In a double-blind, cross-over, placebo-controlled study, they studied the "well-being" of 47 children with seizures prescribed valproic acid or carbamazepine via parent report in person or over the telephone at the beginning and end of each 4 week cycle. They determined that there were no significant findings as the parents consistently reported improved "well-being" (a term not quantified or qualified) with carnitine and with carnitine placebo. They concluded that the cost of the carnitine ($.JO/kilogram of body weight) is not justified by the non-significant improvements. Studies of Valproate Therapy with Epileptic Patients An early study by Gram et al. (1979) studied 13 inpatients with generalized and partial epilepsy. A triple blind multiple cross-over design of varying valproate levels was implemented. Gram et al. (1979) reported a positive correlation between a decrease in the number of seizures and dose level. They further reported mild, transient side effects including drowsiness, anorexia, and hypersalivation which were not correlated with increased dose levels. They concluded that there is a positive correlation between higher valproate levels and better seizure control. They admit difficulty drawing conclusions about valproate

PAGE 53

48 efficacy within a specific seizure type because of the small sample size. Sommerbeck, Theilgaard, Rasmussen, Lohren, Gram and Wulff (1977) studied 20 epileptics, age 13 to 63 years, with numerous seizure types and varying etiologies. Five patients had known brain damage and most patients demonstrated a mixed seizure profile. Ten had "major seizures", 11 had "minor seizures", 1 had focal seizures, and 13 had psychomotor seizures. All patients were receiving other anticonvulsant therapy which was maintained at a constant level throughout the valproate trial. A triple blind cross over design was employed. The purpose of the study was to assess changes in psychometric ability on tests of attention, visual perception, decision making, reaction time, verbal learning, and other general cognitive tests. The test battery consisted of Bourdon's Test (cancellation of 4 targets); Visual Gestalt (nonverbal learning and memory); 100 Minus 7 Test (verbal subtraction); Stroop's Color Naming (attention); Simple Reaction Time; Paired Associate Learning from the Wechsler Memory Scale; Continuous Line Patterns Test (learning and nonverbal reproduction); Digit Span from the Wechsler Adult Intelligence Scale modified to 3 trials per length (immediate span); Tapping Test for one minute trials (motor speed); Hidden Patterns (visual perception); and Time

PAGE 54

Estimation Test. Behavior was assessed through use of the Objective Rating Scale (20 behaviors to rate) by the examiner after the neuropsychological testing. 49 Sommerbeck et al. (1977) found that the valproate slowed decision making, reaction time, and motor tapping. Changes in the EEG's were observed. They concluded that valproate therapy for epileptics can lead to significant decreases in psychomotor speed as assessed by motor speed, simple reaction time, and timed tasks. A non-significant trend for impaired visuospatial analytic and synthetic functioning as assessed by performance on Hidden Pictures task, Continuous Line task, and the Visual Gestalt Test was observed. The psychometric changes were not correlated with seizure frequency, EEG changes and valproate serum level. This study has several serious design flaws. First, a small sample (n=20) with a wide age range (50 years) was utilized without use of a normal control group. The different seizure types were employed, and duration and severity of the epilepsy were not controlled and were not included in the analysis of the results. Many subjects were receiving polytherapy for seizure control in addition to the valproate, confounding the specific effects of valproate. The Committee on Drugs from the American Academy of Pediatrics (1985) reviewed numerous studies comparing antiepileptic drugs on the following factors: monotherapy

PAGE 55

50 versus polytherapy; drug and medical history; seizure etiology, type and frequency; tests employed by the researchers; and design and standardization of the studies. Concerning valproate, they reported drowsiness when prescribed in conjunction with barbiturates. They reported that valproate can lead to negative mood and behavior changes which appear more severe for patients with central nervous system damage; however, these two categories of side effects were not further described. The Committee strongly suggested careful monitoring, physician education regarding seizures and antiepileptic drugs, and development of consistent, sensitive, brief, easily administered screening tests. A recent review article by Trimble (1990) suggested that minimal cognitive side effects occur with administration of valproate. He suggested that the two main cognitive changes associated with antiepileptic drugs are general cognitive functioning and cognitive speed, and valproate mainly affects cognitive speed. As the preceding review suggests, there are a number of limitations in the available literature concerning the effects of valproate on the cognitive status of epileptic patients. Most significantly, the subject samples have not been well defined. Given the numerous factors that appear to influence the cognitive status of epileptics (e.g. type

PAGE 56

51 of seizure, etiology, duration of disorder, severity of disorder, antiepileptic drug therapy, etc.), the previously reviewed studies fall short in controlling for variables that may interfere with interpretation of valproate's effects on cognition. Small sample sizes and lack of normal controls are additional factors that hinder generalization of the existing data. A study by Gallassi, Morreale, Lorusso, Procaccianti, Lugaresi, and Baruzzi (1990) studied the effects of valproate in a sample of 20 seizure free epileptics during monotherapy and after its withdrawal. The subjects had a mean age of 20.0 years (sd=4.7) and had an average of 10.8 years of education. A control group comprised of 20 normals matched for age, education, and social level were employed. All subjects were assessed with a baseline battery consisting of the following: Raven's progressive matrices (intelligence); auditory reaction time -simple (vigilance); auditory reaction time -choice (attention); verbal digits and spatial span (immediate memory); verbal learning (learning); spatial learning and finger tapping test (manual dexterity); trail making test (visuomotor performance); and fingertip number writing (sensory discrimination). The epileptics were assessed at four intervals: baseline, at 3 months after half dose reduction; 3 months later; and at one year. Those who relapsed during the study were excluded.

PAGE 57

52 Only 11 of the 20 subjects completed the study. Raw scores were converted to z-scores and then tot-scores. All tests were combined by creating a mean of the t-scores to create a 'global performance score' (GPS). ANOVA's were performed and significance was set at 0.01. Gallassi et al. (1990) reported significant adverse effects of valproate on cognition. At baseline, significant group differences were observed on measures of attention, visuomotor task performance, and GPS. At time 2 with the half-drug dose, GPS was still impaired. No impairments were observed at times 3 and 4. Gallassi et al. interpreted these findings as subtle, subclinical aversive effects on cognition caused by valproate monotherapy. This study is important in that it used a variety of neuropsychological measures and employed a normal control group. The area of attention was found to be vulnerable to valproate. Performance on the Trail Making Test, defined to be a measure of visuospatial function, is also considered to be a measure of attention. Several design flaws were noted. The normal control group did not participate in the longitudinal testing; all comparisons were made to their baseline performance, therefore, practice effects were not controlled. Alternative forms of the tasks were employed, but without a control group, rendering interpretation of differential test performance beyond baseline limited.

PAGE 58

53 Furthermore, loss of 9 of 20 subjects due to relapse of seizures suggests the possibility of a differential subgroup of epileptics remaining in the study. Possibly those who did not relapse were neurologically in better condition and therefore more cognitively intact confounding the interpretation of improved test scores. Valproate: Neuropsychological Effects Forsythe, Butler, Berg, and McGuire {1991) examined the relative effectiveness and side effects, physical and cognitive, of carbamazepine, phenytoin, and valproate. Sixty-four children ranging in age from 5 to 14 years with newly diagnosed epilepsy were the subjects of this study. They were diagnosed with one of the following seizures type: tonic-clonic, complex partial, or complex partial with secondary generalization. All children were reportedly within the average range of intelligence. Thirty-one enuretic patients and 9 migraine patients were employed as controls. The epilepsy patients were placed on carbamazepine, phenytoin, or valproate in a random assignment. Forsythe et al. (1991) used a cognitive test battery, fashioned after the work of Trimble, consisting of the following: visual recall of pictures of objects (immediate and 30 minute delay), immediate recall of designs, auditory recall (digits forward and backward, immediate span), visual

PAGE 59

54 scanning (attention), the Stroop test (concentration), and speed of information processing. The subjects were tested at the initiation of their treatment and at 1, 6, and 12 months into their treatment. The children were also given the WISC-R, and the Neale Analysis of Reading Ability at 1 month and 12 months after initiation of therapy. Forsythe et al. (1991) reported that all the antiepileptic drug groups had good, equally effective seizure control. The fewest adverse side effects were reported with valproate, and the most with carbamazepine. The combined score of all the memory measures was found to be the most sensitive measure for drugs effects. A positive correlation between the serum level of valproate and memory scores was reported. Impaired memory recall was observed in the carbamazepine group after 6 months of treatment. Slower speeds of information processing were observed in the carbamazepine and phenytoin groups after one month of treatment. Several problems are evident in this longitudinal study. Use of a sample with heterogenous seizure types from a wide age range treated with three different drugs makes generalization of results difficult. The purpose of enuretics as a control group is uncertain. The test procedures employed lack verbal learning measures and prose recall, sensitive indicators of memory function in children.

PAGE 60

55 All memory measures were summarized across span, short term recall, and long term recall, losing information concerning specific memory differences among groups and over time. A principal components analysis was employed for statistical interpretation. This procedure requires more subjects per measure than employed in this study causing the results to be questionable in their strength for interpretation. Results comparing the epileptics and the controls were not reported. Baseline intelligence and reading scores were not reported, and it is unclear if initial differences were adjusted for in the subsequent analyses. Vining, Mellits, Dorsen, Cataldo, Quaskey, Spielberg, and Freeman (1987) studied the psychologic and behavioral effects of valproate and phenobarbital in a double-blind, counterbalanced, crossover study with 21 children of normal intelligence with mild seizure disorder. These subjects ranged in age from 6 to 14.5 years. Nine children were diagnosed with tonic-clonic seizures, 11 with partialcomplex seizures, and 1 child demonstrated both types of seizure. The EEG's prior to initiation of the study were interpreted as 5 normal, 9 mildly abnormal, and 7 very abnormal patterns. Seizure etiology included perinatal trauma, postnatal trauma, and idiopathic onset. Subjects were randomly assigned to one of the antiepileptic drugs for a 6 month period followed by a 6 month cross over period.

PAGE 61

56 The neuropsychological battery employed by Vining et al. included the WISC-R, subtests from the Detroit Tests of Learning Aptitude (attention and memory), the Symbol Digit Modalities Test (sequencing and processing efficiency), Bender-Gestalt (perception accuracy and constructional skills), Berkeley Paired Association Learning Test (attention and short-term learning), and Seashore Rhythm Test (auditory discrimination and short-term memory). Other tests included Reading, Spelling and Arithmetic from the Wide Range Achievement Test, the Gray Oral Reading Test (fluency and accuracy for oral reading), and a Continuous Performance Reaction Test (vigilance and visual concentration). Measures of gross motor skills, fine motor skills, dexterity and tracking included Finger Tapping, Ambulation Backwards, Mazes, and a video arcade tracking game. Vining and her group reported comparable seizure control for both drug groups. Using paired t-tests on all dependent variables, 7 of 35 neuropsychological measures and 9 of 48 behavioral items were reported to be significantly different and in favor of the subjects on valproate therapy. Specifically, 4 neuropsychological measures were found to be significantly different at the p<.01 level, including Block Design (WISC-R subtest), Performance IQ, Full Scale IQ, and the Berkeley Paired Association Learning Test II.

PAGE 62

57 Vocabulary (WISC-R subtest), Picture Completion (WISC-R subtest), and the Berkeley Paired Association Learning Test I were different at the p<.05 level of significance. Three behavior items were found to be significantly different at the p<.01 level, including: "fails to finish", "basically unhappy", and "unable to stop". "Disobedient", "other aches", "excitable", "worries", "problems with friends", and "problems with sleep" were reported significantly different at the p<.05 level. Vining et al. concluded that valproate therapy for children with generalized tonic-clonic seizures and partial complex seizures is as effective as phenobarbital and less deleterious to cognition and behavior. A number of problems are evident in the design of this study and the interpretation of the findings. First, the intensity of seizure disorder was defined by seizure frequency; however, the "mild" distinction included a range of seizure frequency from two per day to one per year. Heterogeneous seizure etiologies, as well as multiple seizure types, were included in the subject sample. The sample size was small and a normal control group was not employed to help interpret practice effects. Full scale IQ'S were purported to be within the normal, or "average" range; however, with a mean and standard deviation of 94.0 + 14.4 many subjects appear to be below the average range of

PAGE 63

58 intellectual function. Though a lengthy neuropsychological evaluation was performed, all measures, and their subtests when present, were statistically analyzed with paired ttests. A total of 35 neuropsychological measures were compared with this method. The first three of the above listed significantly different neuropsychological measures are highly correlated, as Block Design is a part of the Performance IQ which is half of the computed Full Scale IQ. In addition, the specific scores from the WISC-R labelled as significantly different are not clinically different as they all fall within the average range. Forty-eight behavior items were compared with paired ttests. The behavioral reporting was along a four-point scale from "not very much= 111 to "very much= 411 The significant behavior problem scores fell between 1.19 and 2.20 suggesting a non-clinical problem level. The statistical analysis by Vining et al., multiple paired t-tests, is highly vulnerable to Type I errors. The findings reported to be significant are at a rate so low that chance occurrences may have explained their findings. Finally, the results from both the neuropsychological and behavioral domains with the exception of the Berkeley Paired Association Learning Test II, are not within a clinically problematic range.

PAGE 64

59 Aldenkamp, Alpherts, Blennow, Elmqvist, Heijbel, Nilsson, Sandstedt, Tonnby, Wahlander, and Wosse (1993) studied the effect of withdrawal of valproate from a pediatric epilepsy sample as part of the multicenter Holmfrid study. One hundred children with epilepsy who had been seizure free for at least one year were withdrawn from monotherapy of valproate, carbamazepine, or phenytoin. Seventeen children diagnosed with absence, tonic-clonic, partial, and rolandic epilepsy were withdrawn from valproate monotherapy. They were matched to a healthy classmate. The subjects were tested at baseline, prior to drug withdrawal, at 3 months at which time they were totally off the drug, and at 3 to 4 months after that time. They employed the Finger Tapping Test, Simple and Binary Choice Reaction Time Tests, Computerized Visual Searching Test, and Word and Figure Recognition in simultaneous and serial presentation paradigms. Aldenkamp and his colleagues reported no significant improvements on the motor, attention, and memory measures after accounting for practice effects in the epilepsy group as a whole. Significant differences in the absence seizure group (n=7) at baseline and after withdrawal were observed and were interpreted as continued, subclinical EEG activity which interfered with cognition. They hypothesized that the lack of improvement or deterioration with valproate

PAGE 65

60 withdrawal resulted from the combined effect of removing the beneficial, protective effect of valproate and the recurrence of subclinical EEG abnormalities. No data was presented to confirm this hypothesis. Aman, Weary, Paton, and Turbott (1987) studied the effect of valproate on psychomotor performance in children with epilepsy focussing on the variables of dose, fluctuation of concentration of valproate and the diagnosis of the children. Forty-six epileptic children with normal IQs ranging in age from 4.4 to 15.4 years were the subjects. They were diagnosed with partial epilepsy, most of them with secondary generalization (n=lO), generalized epilepsy half with absence and half with tonic-clonic (n=34), or unclassified epilepsy (n=2). The children were previously treated with phenytoin, carbamazepine, or polytherapy and were placed on valproate monotherapy. The neuropsychological test battery employed in this study included Matching Familiar Figures (impulsivity), Auditory-Visual Integration Task (cross modal matching), a non-verbal recognition test using cartoon pictures, Rosvold's Continuous Performance Task (vigilance and attention), Seat Movements measured by duration on the chair, Maze Task from Klove's Motor Steadiness Battery (impulsivity and motor coordination), Klove's Graduated Holes Task (resting tremor), Pursuit Rotor Task (hand-eye

PAGE 66

61 coordination), and Porteus Mazes (intelligence and planning). The subjects were tested at baseline (pre-drug), in a delayed medication condition (low serum level), and shortly after given medication (high serum level). The children were divided into to a high dose and a low dose group (27.14 and 15.84 mg/kg/day median, respectively). Aman and his group reported effects according to diagnosis, dose, and time of medication. Data was statistically analyzed with a Repeated General Linear Analysis. Four significant differences by diagnosis were reported. Subjects with partial epilepsy demonstrated a longer response time on the Auditory-Visual Integration Task (F=S.95, p<.05), longer response time on the non-verbal recognition task (F=S.23, p<.05), more errors of omission on the Continuous Performance Task (F=4.88, p<.05), and a higher ratio of contact time to travel time on the Maze Task (F=9.86, p<.005). Four significant findings by dose were reported. The high dose group demonstrated more seat movement during the Matching Familiar Figures Test (F=S.02, p<.01), lower accuracy and longer response time on the Auditory-Visual Integration Task (F=l0.38, p.oos, F=6.86, p<.01, respectively), and a lower test quotient on the Porteu. s Mazes Task (F=4.99, p<.01). Only one significant finding for effect of plasma concentration level, a factor

PAGE 67

determined by time of valproate administration, was reported. The high concentration level group demonstrated greater contact time on the Maze Task (F=4.65, p<.05). On the whole, Aman and his group reported strong effects for diagnosis and dosage and weak effects for time of medication. 62 Aman et al. (1987) designed a study loaded with motor measures and did not effectively address memory, verbal learning, and attention. Their subject group had numerous seizure types and a control group was not employed to control for practice effects. The subject group had a wide age range (11 years) and age corrections were not employed. All but one of the procedures were electronically administered which may limit generalization. The key advantage for this method is reduction of examiner variation or examiner effect; however, electronic administration is not a direct reflection of most daily interactions. Finally, the majority of the differences reported were significant at the p<.05 level which is more representative of a trend rather than a significant difference in studies with small samples and multiple variables. Rationale for Further Study Review of the current literature regarding the relationship between valproate therapy and the pediatric epilepsy population reveals a relative paucity of well-controlled

PAGE 68

63 studies of memory and attention. Narrowing subject inclusion to specific seizure types shown to be responsive to valproate, limiting the age range, and excluding children of below average intelligence would address weaknesses in previous research studies. Examining only one or two aspects of cognitive functioning will allow a more thorough analysis of valproate's effect. Memory and attention have been described to be crucial to learning in children (Cooley & Morris, 1990) and the extent to which these abilities are affected by idiopathic epilepsy, valproate monotherapy, and carnitine supplementation is not known. This information is of significant importance for thorough medical treatment, school performance, and behavioral management. Specific Aims The specific aims of this study were: (1) to evaluate for the presence of attentional and memory disorder in children diagnosed with idiopathic generalized epilepsy (accomplished at baseline testing); (2) to determine the effect of therapeutic valproate levels on memory and attention in epileptic children; (3) to determine if differences exist in memory and attention between children with epilepsy on valproate, children with migraine on valproate, and a non-medicated, neurologically normal control group; and, (4) to determine the cognitive effect of valproate plus carnitine in a pediatric epilepsy sample.

PAGE 69

CHAPTER 2 METHODS Subjects The subjects of this study were 12 children with idiopathic epilepsy referred to this study by the physicians at the Pediatric Neurology Clinic at Shands Hospital. The subjects had a seizure profile (type and frequency) appropriate for treatment with valproate, i.e. generalized motor, absence, or complex partial with secondary generalization. Subjects were between the ages of 6 and 16 years for both pharmacologic constancy and design control. All subjects underwent an EEG to obtain a baseline and to support diagnosis and an MRI was performed to rule out the presence of a structural lesion or abnormality. An Internal Review Board (IRB) protocol was presented and explained to the parents or guardians of the prospective subjects and informed consent was obtained. Accrual of the subjects began in January 1993 and continued throughout the year until December 1993. An IQ screening was performed at the time of inclusion in the study using four WISC-R subtests. 64

PAGE 70

65 Only those subjects with a mean scaled score of 8 or above and those with individual scales scores of 7 and above were included. The twelve children with epilepsy recruited for this study and were diagnosed with the following seizure types: absence, partial complex with secondary generalization, primary generalized, and Rolandic. The sample was comprised of five males and 7 females. The mean age was 11.1 years with a standard deviation of 2.8 years with a range of 7.0 to 15.3 years. Ten of the children were right hand dominant, one was left hand dominant, and one was ambidextrous. Age of onset, duration of illness, presence of significant headaches, diagnosis of attention deficit disorder with hyperactivity and family history of epilepsy were recorded (See Table 2-1). Ten children remained in the study at the four-week mark and nine children completed the study. A control group comprised of twenty non-neurologically involved children was recruited from a local public school system by written requests through teachers to parents. Following parental consent, the control group participated in all the neuropsychological testing procedures but received no medication. The control group was screened by parent report for history of neurological disease and injury, learning disabilities, and significant behavior and

PAGE 71

66 medical problems and were excluded if any of these conditions were present. Eighteen of these children remained through the three testing sessions. This normal control sample consisted of 14 males and 6 females with an average age of 12.1 years with a standard deviation of 3.2 years, ranging from 7.2 to 16.9 years. Nineteen were righthanded and one was left-handed. In addition, the pediatric epilepsy group was compared to pediatric migraine patients being treated with valproate carnitine, valproate with carnitine placebo, and valproate placebo and carnitine placebo. The migraine patients were referred to the study from the Pediatric Neurology Clinic and were screened for appropriateness for this study by history, physical examination, EEG, and MRI (where appropriate) by their referring physician. The migraine subjects participated in the identical testing procedures as the epileptic sample. Age restrictions were similar to those for the epilepsy group (6 to 16 years). There were 14 migraine patients at baseline testing, with eleven males and 3 females and seven completed the study. Their mean age was 12.9 years with a standard deviation of 1.9 years, ranging from 9.5 to 15.8 years. At baseline, there was no significant difference between age, handedness, or intellectual (IQ) level among the three groups (See Table 2.2).

PAGE 72

Materials and Apparatus The following is a brief description of the neuropsychological and behavioral measures used in this study. Intellectual Screening 67 Four subtests from the Wechsler Intelligence Scale for Children -Revised (WISC-R) were employed to assess intelligence. Use of the WISC-R has been chosen over the WISC-III for comparability to previous research. The WISC-R has well established norms for ages 6-16 years. Two subtests from the Verbal domain, Similarities and Comprehension, were selected because they provide strong reliability for intelligence estimation with .81 for Similarities and .77 for Comprehension. These subtests were chosen over the other subtests from the verbal domain of the WISC-R because they rely less of specific acquired facts and recall of previously learned information (Sattler, 1988). Similarities, a test which requires the subject to determine how two things are alike, measures abstraction ability and verbal knowledge. Comprehension evaluates the subject's practical reasoning ability. Two non-verbal subtests from the Performance domain, Object Assembly and Block Design, were employed. Both are timed tests which require visual analysis, synthesis, and construction performance. Both subtests offer good split-half reliability with .70 for

PAGE 73

68 Object Assembly and .85 Block Design (Sattler, 1988). Furthermore, there is stron9 correlation of each subtest to the Performance IQ with .68 for Block Design and .60 for Object Assembly. These four subtests of the WISC-R provided a reliable estimate of general intellect and served as a screening instrument for level of intelligence for all subject groups and as a population descriptor for the epilepsy and migraine groups. Intelligence quotients are thought to be relatively stable measures and the performance specific subtests are prone to practice effects. Therefore, these subtests were only administered during the baseline assessment. Memory Measures Verbal memory was assessed at each of the three testing intervals with several measures. A modified version of the Logical Memory subtest from the Wechsler Memory Scales (WMS) Form I and Form II was employed to assess both immediate and delayed memory. The Logical Memory subtest was modified by reading only one story at a given administration as three forms are necessary for this testing paradigm. Both stories from the WMS Form I ("Anna Thompson" and "The American Liner") and the first story from the WMS Form II ("Dogs are trained") were chosen. To best interpret performance on the three versions of this task, scores were determined with established scoring criteria (Russell, 1975).

PAGE 74

69 Wechsler Memory Scales Form I and Form II are not directly interchangeable; however, the Logical Memory subtest of each version is considered adequately comparable (Spreen and Strauss, 1991). Extensive research of the reliability and validity of the WMS Logical Memory subtest has been conducted. Norms are available for ages 10 through adulthood, using the standard two story administration. Raw scores were converted to z-scores by doubling the recall of the single story and subtracting the age appropriate mean and dividing by the standard deviation. As no normative data are available for this modification, comparison to the normal control group was necessitated. The Buschke-Levin Selective Reminding Test (SRT) is designed to measure verbal learning and verbal memory in a multiple trial list learning format. In this task, the subject is presented with a 12 item list of non-related words and is asked to recall as many words as possible in any order. The remainder of the words from the list not recalled is then repeated. This process continues for 8 trials or until the subject has learned all 12 words. This procedure allows the examiner to distinguish between immediate memory, long term memory and rate of learning. Though this task provides several scores, only two dependent measures were analyzed: learning (total number of words recalled over trials 3 through 8) and delayed recall.

PAGE 75

70 The Buschke-Levin SRT has multiple forms that are relatively equivalent and interchangeable allowing for repeat testing (Clodfelter, Dickson, Newton Wilkes, & Johnson, 1987) and three forms were required for this paradigm. Test-retest reliability was found by Masur, Fuld, Blau, Thal, Levin, and Aronson, (1989) to be .92 for retrieval with low reliability (.42) for intrusion rate. Some practice effect is expected, and a normal control group was employed to account for this. Normative data is available for age 5 through adult. Digit Span, taken from the WISC-R, was the last measure of verbal memory. In this task, the examiner reads aloud, at a rate on 1 per second, series of digits which the subject immediately repeats in the same order. Digits Forward begins with 3 digits and proceeds to 9 digits with 2 series of each length. Digits Backward, a task requiring the subject to reverse the order of the numbers presented, begins with 2 digits in a series and increases to 8 digits. The task is discontinued when the subject fails both attempts in a given trial. Digit Span is a measure of immediate memory and attention and has adequate reliability of .78 (Sattler, 1988). Practice effects have not been reported and it is appropriate for repeat testing. Normative data are available for age 5 through adult. Normative z-scores were

PAGE 76

71 obtained in the identical manner as described above. Nonverbal Memory was assessed with two measures, the Corsi Cube Test (Milner, 1971) and the Visual Learning subtest from the Wide Range Assessment of Memory and Learning (WRAML) (Sheslow & Adams, 1990). Corsi Cubes is considered to be the nonverbal counterpart of Digit Span as it requires the subject to repeat series of taps of increasing length both forward and backward. The examiner taps the sequence at a rate of 1 per second on the fixed Corsi array of 9 blocks which are 1.5 inch cubes. Because of the 2 dimensional aspect of the stimuli, both spatial ability and nonverbal memory are required. This task has not shown to be susceptible to practice effects and is acceptable for repeat testing. The normative performance is thought to be one less than the normative span for digits for a given age. Normative z-scores were obtained in a similar manner as above. The Visual Learning subtest from the WRAML requires the subject to recall the spatial location of visual designs presented in specific positions on a four by four array. After presentation, the stimuli are covered with sponge shields. The child is required to recall targets over 4 trials. Sheslow and Adams (1990) report reliability coefficients the median of which is .88 with test-retest reliability at .81. As the stimuli are non-meaningful and

PAGE 77

the reliability is good, significant practice effects are not expected. Normative data are available for age 5 through adult in the WRAML manual. Attention Measures 72 Attention was assessed with a computerized Continuous Performance Task (CPT) and the Paced Auditory Serial Addition Test (PASAT). The CPT can assess both vigilance and reaction time. The single-choice reaction time requires the subject to press one of two keys on a computer keyboard immediately following a selected cue. The cue, which appears at random points on the screen, signals to the participant with which hand to respond. This reaction time task is a measure of speeded decision making and attention. Designing the task to be a forced choice increases its difficulty, and therefore its sensitivity to possible impairments in decision making and attention. This measure was shown to be a sensitive measure in discerning differences between a pediatric renal disease sample and normal controls (Fennell, Fennell, Carter, Mings, Klausner, & Hurst, 1990). The CPT is IBM computer administered and can be altered for multiple administrations (Fennell, et.al., 1990). In this study either the Leading Edge portable computer or a standard IBM computer was employed. Some practice effects were expected due to increasing familiarity with the

PAGE 78

stimulus, but use of normal controls allowed for clearer interpretation of the results. The computerized format is engaging to the child and the motor requirements are minimal, thus allowing for assessment of attention and speeded decision making. Mean reaction time across all of the trials of the various conditions was employed as the sole dependent variable. As there is no normative data on this task, the use of the control group data was necessitated. 73 The PASAT (Gronwall & Wrightson, 1974) is a serial addition task and the rate of presentation of the stimuli is controlled by audio tape administration. A total of 61 digits are presented at 2.4, 2.0., 1.6 or 1.2 seconds per digit for varying degrees of difficulty; however, only the two slowest speeds of presentation were employed in this study. The subject adds the first two digits and states aloud the sum and then adds the third digit to the second digit and so on. Gronwall and Wrightson (1981) report that the PASAT is a sensitive measure of information processing and sustained at~ention. It has been described as a task which assesses central information processing similar to that of reaction time and divided attention tasks (Gronwall and Sampson, 1974). Two short-comings of the PASAT for this study are lack of normative data for children under the age of 8 and strong

PAGE 79

74 practice effects (Gronwall, 1977) for the second presentation, though not at later presentations (Spreen and Strauss, 1991). Strong split-half reliability (.96) has been reported for the PASAT (Spreen and Strauss, 1991). The two slowest trials were administered and scores (total correct responses per trial) were converted to zscores using available normative data. Only subjects eight years of age or older were included in the analysis due to limited normative data. Motor Measures Graphomotor constructional skill was assessed with the Beery Test of Visual Motor Integration (VMI) while motor speed was assessed with the Finger Tapping Test. The VMI (Beery, 1967) presents the subject with 24 geometric figures of increasing difficulty which the subject is to copy to the best of his perceptual motor ability. Cosden (1985) reports a mean interrater reliability of .93 and retest reliability ranges from .63 to .92 over various time spans and subject groups. Sattler (1988) reports good validity from a sample of 3090 subjects. Normative data is available for ages 2:10 through 17:11 and no gender differences have been noted. The VMI is thought to be relatively unaffected by environmental factors (i.e. medication, time of testing) and offered a comparison to more state oriented measures. This measure was employed only at baseline and final testing.

PAGE 80

75 Motor speed was assessed with the Finger Tapping Test (Reitan, 1969). In this task, the subject is presented with a tapping board with a metal key and an automated counter. The subject places his index finger on the key and while maintaining his hand in a relaxed position, taps the key as quickly as possible. After familiarizing the subject with the tapping board, four trials each of ten seconds duration for each hand were conducted, alternating between the hands after 2 trials. The mean numbers of taps with both dominant and the non-dominant hand were obtained. Normative data are available for age 6 through adult and z-scores were calculated. Moderate reliability has been reported, ranging from .58 to .93 (Spreen and Strauss, 1991). Behavioral Measures All dependent measures of behavior were obtained through parental report on the Child Behavior Checklist by Achenbach (CBCL) (Achenbach and Edelbrock, 1986). The CBCL is a parent report measure commonly used to assess behavior problems and psychopathology. This questionnaire was selfadministered to the subject's parent or guardian. The two broad factors determined with the CBCL are the Externalizing and Internalizing T-scores. Children who fall into the externalizing range tend to be overactive and to display conduct problems. The internalizing child tends to be more withdrawn and dependent on the parent. The CBCL also

PAGE 81

76 provides several more specific behavior scales which load on the Internalizing and Externalizing scales. DuPaul, Guevremont, and Barkley (1991) state that the CBCL is the best standardized child behavior rating scale available and it has very good reliability and validity. Given the large number of variables and the limited sample size, the CBCL was described in the data as T-scores for the Internalizing and the Externalizing scales, measures which summarize the parental response. Procedures and Design The procedure for the baseline testing session for all participants in the study involved an initial screening procedure comprised of the four subtests of the WISC-R followed by the neuropsychological tests described above. Testing was completed prior to initiation of medication. Once inclusion to the study was determined, the neuropsychological battery began with the presentation and immediate recall of the modified WMS Logical Memory Story. The PASAT was the second measure so as to minimize fatigue. The Tapping Test followed for similar reasons. The CPT was administered next. Approximately 30 minutes had elapsed and the delayed recall of Logical Memory was assessed. The Buschke-Levin SRT was next administered, followed by Digit Span, the Beery VMI, the Corsi Cube Test and Visual Learning from the WRAML. The delayed recall trials for the Buschke-

PAGE 82

77 Levin SRT and the WRAML Visual Learning were administered last. Parents or guardians completed the Child Behavior Checklist while their child was tested. This study is a short-term follow-up design measuring attention and memory in a pediatric epilepsy sample at three points over 3 months: before initiation of valproate monotherapy (baseline); after 4 weeks of valproate monotherapy; and after a weeks of valproate monotherapy with double blind randomization to a carnitine or carnitine placebo condition. The examiners were blind to both the diagnosis and to the drug regimen. The patient and his family were blind to the drug regimen at the time of randomization to carnitine or carnitine placebo. Randomization to drug groups was performed by the Pharmacy Department through a randomization computer program. The diagnosis of the subjects was performed by pediatric neurologists at Shands Hospital. Once diagnosed with pediatric idiopathic epilepsy, medical appropriateness for this study was determined by the neurologists through physical examination, EEG, and MRI. Once parental consent was obtained, the intellectual screening for inclusion was performed. Baseline neuropsychological testing was performed. To address practice effects in the test-retest paradigm, alternate forms of the measures, where applicable, and a normal age matched control group were employed.

PAGE 83

The valproate monotherapy was monitored by the neurologists and the laboratory personnel at Shands Hospital. The subjects received therapeutic dosages resulting in plasma levels in the range of 50 and 100 micrograms per milliliter. The addition of the carnitine and carnitine placebo was monitored by patient report of compliance to the prescribed regimen. 78 Parallel to the epileptic subject group, a pediatric migraine sample was treated with valproate with carnitine, valproate with carnitine placebo, or valproate placebo with carnitine placebo. Identical criteria for inclusion and testing procedures were implemented. Attempts to age match this group to the epilepsy group were made. All patient information remained confidential. The normal control group participated solely in the neuropsychological testing and they were rewarded with gift certificates to a local fast-food restaurant. Their materials also remained confidential. Hypotheses In light of the current available literature on childhood epilepsy, childhood migraine, valproate, and carnitine, the following hypotheses were made regarding this study. (1) The pediatric epilepsy sample screened for intelligence in the average range would demonstrate mild to moderate impairment on attention and memory at baseline

PAGE 84

79 given that memory and attention impairments have been observed in children with generalized epilepsy, the type of epilepsy best treated with valproate monotherapy. The pediatric migraine sample would not demonstrate any cognitive deficits. The overall cognitive performance for both samples will be biased toward the average range as all subjects are screened for average intelligence. (2) Though the valproate will help normalize EEG activity in the pediatric epilepsy sample and control seizure activity, the current literature suggests that valproate may lead to poorer performance on timed decision making tests and tests requiring sustained attention. We predicted that the pediatric epilepsy sample would continue to demonstrate mild attention and memory impairment while on the valproate monotherapy. (3) The addition of carnitine to the valproate monotherapy would reduce the negative somatic side effects and would not lead to negative cognitive effects on either group. Those subjects in both the pediatric epilepsy sample and the pediatric migraine sample would perform as well or better on the neuropsychological battery following the addition of carnitine. (4) The necessity of repeated measures for this study's design would lead to improvement in performance across measures which are sensitive to repeated exposure. (5) The medical involvement of the epileptic sample would result in somatic complaints at baseline.

PAGE 85

80 TABLE 2.1 CHARACTERISTICS OF THE EPILEPSY SAMPLE Seizure Type Age of Onset Range in years EEG findings Family history Significant headaches Attention Deficit Disorder w/ Hyperactivity Level of Seizure Control on VPA Reported Negative Side Effects of VPA Table 2.2 DEMOGRAPHIC INFORMATION Epileptic Baseline sample size Gender Ratio M/F Mean Age (years) Age Range (years) 12 5/7 11.1 (2.8) 7.0 -15.3 Handedness:R/L/Amb. 10 /1 /1 WISC-R IQ subtest Similarities Comprehension Block Design Object Assembly means 10.7 11.9 9.7 10.6 (2.5) (3.4) (1.8) ( 1. 3) 2 Rolandic, 6 Absence, 2 Partial Complex, 2 Mixed 1 year to 12 years 0.5 years to 12 years 6 abnormal; 6 normal 2 positive; 10 negative 7 positive; 5 negative 2 positive; 10 negative 8 excellent; 2 moderate; 1 intolerant; 1 unknown 4 w/ adverse effects 8 w/ no adverse effects Migraine Control 14 20 11/3 14/6 12.9 ( 1. 9) 12.1 (3.2) 9.5 -15.8 7.2 -16.9 13 /1 /0 19 /1 /0 11. 8 (2.3) 12.4 (3.2) 10.0 (2.2)* 12.8 ( 2. 6) 9.3 (2.6)* 11. 4 ( 2. 2) 10.5 (4.2) 11.5 (3.5) significant difference from control group at p~.05 level. Note: Standard deviations are provided in parentheses.

PAGE 86

TABLE 2.3 NEUROPSYCHOLOGICAL DOMAINS FACTOR Immediate Verbal Memory Verbal learning Delayed Verbal Memory Non-Verbal Memory Non-Verbal Learning Attention Motor Function Behavior TESTS Digit Span Story Recall -Immediate Buschke-Levin learning Buschke-Levin retrieval story Recall -delayed Corsi Cubes Test 81 Visual Learning -delayed Visual Learning PASAT CPT Beery VMI Finger Tapping CBCL

PAGE 87

CHAPTER 3 RESULTS All subjects were determined to be within the average range of general intelligence. Group differences on two of the WISC-R subtests, Block Design and Comprehension, were observed between the migraine group and control group, with epileptics holding a middle position. Given these differences in conjunction with the wide range for a normative "average score" on the IQ measure, IQ was employed as a covariate across the remaining neuropsychological measures to ensure its role in the group differences. An IQ factor was determined by deriving a mean of the four subtest scores. An analysis of variance (ANOVA) was first used to compare the epileptics and control group means at Baseline for each measure employed. Pairwise Tukey comparisons were performed to determine direction of difference. Tukey pairwise comparisons were chosen to be conservative (Howell, 1992). An analysis of covariance (ANCOVA) was used to compare the groups at baseline adjusting for IQ as described 82

PAGE 88

above. Multiple analyses of variance (MANOVA's) by cognitive domain were not performed because of low outcome variable intercorrelations and small design cell frequency (Huberty & Morris, 1989). Furthermore, MANOVA's often necessitate subsequent ANOVA's to determine specificity of group mean differences leading to inflation of the Type I error rate (Dar, Serlin and Omer, 1994). The neuropsychological measures described previously were employed at baseline. Table 3.1 provides the means, standard deviations, and significant p-values for the baseline neuropsychological scores for the epileptic and normal control groups. Hypothesis One 83 The first hypothesis of this study was that children with idiopathic epilepsy would perform at a level below the normal control group on measures of attention and memory before initiation of the valproate monotherapy. Please refer to Table 3.1. At baseline, there were several significant differences between the epileptic group and the normal control group. Performance on the PASAT in both the slow and fast conditions by the epileptic group was significantly below that of the control group (p=0.0026 and p=0.0070 respectively). Migraineurs maintained an intermediate position which was not significantly different from the

PAGE 89

84 other groups. Digits Forward was significant at p=0.0416; however, pairwise comparisons indicated no specific difference between any two groups. The Buschke Selective Reminding Test Learning score was significantly different (p=0.0062) with both the control group and the migraineurs performing significantly better than the epileptics. As expected, a significant difference among group means was observed for the CBCL Internalizing T Score. Tukey pairwise comparisons indicated that the epileptic and the migraine group means were both significantly greater than the control group mean and they did not significantly differ from each other (p=.0041). The ANCOVA adjusting for IQ provided comparable results on the both trials of the PASAT and the Buschke SRT learning score. In addition the CPT reaction time measure was significantly slower for the epileptic group compared to the control group (p=.031). However, the difference in Digits Forward was no longer significant when covaried with IQ. Hypothesis Two The second hypothesis stated that the epileptic group would continue to demonstrate impaired attention and memory and would have difficulty on timed decision making tasks (CPT) following the initiation of valproate monotherapy. Again, ANOVA's were performed and pairwise Tukey comparisons were employed. Results are presented in Table 3.2. The

PAGE 90

85 ANCOVA with the IQ covariate was also performed. At Time 2, four weeks after initiation of valproate, the epileptic group continued to display significant differences from the control group in performance on the PASAT (slow and fast conditions), Digits Forward, and Buschke Selective Reminding Learning score. On the PASAT slow condition, the epileptic group performed below the level of the control group at the p=.035 level. For the PASAT fast condition, the epileptic group performed below the control group at the p=.022 level. Again for both these measures the migraine group performed at an intermediate level which was not statistically different from the other groups. On Digits Forward, the ANOVA revealed a significant difference among the groups (p=.024); however, the Tukey pairwise comparisons failed to reveal a specific difference between any two groups. Again, similar to baseline data, the epileptic group performed at a significantly lower. level than the control group on the Buschke SRT Learning score (p=.017), with the migraine group in an intermediate position. No other measures were significantly different at Time 2. These results support the second hypothesis that epileptics would continue to display lower performances on tasks tapping memory and attention while on valproate monotherapy.

PAGE 91

86 Hypothesis Three The third hypothesis stated that the epileptic children taking the carnitine supplementation in addition to the valproate monotherapy would perform better than those taking carnitine placebo. Using ANOVA's and pairwise Tukey comparisons, there were no significant differences on any of the measures between the epileptics receiving carnitine and those receiving carnitine supplementation. See Table 3.3 for means and standard deviations. The third hypothesis was not supported by these findings; however, the sample sizes were very small (5 epileptics in the carnitine condition and 4 in the carnitine placebo condition). Given these findings, the epileptic group was collapsed at Time 3 and scores were reanalyzed using a repeated measures ANOVA with one between subject factor (control versus epileptic) and one within subject factor (time). Paired t-tests were performed to compare within subject responses between Baseline and Time 3 for the Achenbach Child Behavior Checklist. The migraine group was not included in this analysis because of their high rate of attrition and significant non-compliance with valproate monotherapy. Results are presented in Table 3.4. At Time 3, the epileptic group performed at a significantly lower level on PASAT fast condition (p=.0044). On the Buschke SRT delay trial, the epileptic group

PAGE 92

87 performed at a lower rate than the control group (p=.0247). There were no other group differences at Time 3. On further analysis of the PASAT slow and fast condition across the groups and across time, there is a significant main effect for group (p=.0183 and p=.0075 respectively). There was no significant main effect for time nor was there a significant interaction of group by time. Digits Backward showed a main effect for group (p=.0348); however, no main effect for time nor an interaction of group by time were observed. Corsi Forward showed a significant effect for time (p=.0293), but not for group or a group by time interaction. Analysis of Digits Forward and Corsi Backward revealed no significant differences for group or time or a group by time interaction. A significant main group effect (p=.0236) on the Buschke SRT Learning measure was observed, but no significant time or time by group interaction was observed. The Buschke SRT Delay measure approached significance for a main group effect (p=.0589). A significant main effect for time was observed on the Logical Memory immediate and delayed recalls (p=.0141 and p=.0102, respectively). Analysis of responses suggests that the version of the test administered at the second testing

PAGE 93

88 was significantly more difficult for all groups compared to the versions at Baseline and Time 3. A significant main effect for time was observed for the WRAML Visual Learning scaled score (p=.0097) and indicates that all groups improved their performance over time at a similar rate. Again, the WRAML Visual Learning delayed recall score showed a main effect for time with all groups improving in a similar fashion with repeated exposures. There were no significant differences on the Achenbach CBCL measures at Time 3. To further elucidate the role of valproate, the blood plasma level of valproate was covaried with the test scores using an ANCOVA procedure. Most of the epileptic sample were in the therapeutic range (i.e., 50 or higher), and several were only slightly below that level (i.e., 30 or higher), which is consistent with the peak/ trough fluctuations of acceptable valproate therapy compliance. At Time 3, there was a significant difference on the PASAT fast condition, in that the score increased with blood level of valproate. Interpretation of this finding is limited as the groups compared were the epileptics in the carnitine condition. No other scores differed by group when valproate level was covaried.

PAGE 94

TABLE 3.1 BASELINE NEUROPSYCHOLOGICAL DATA -EPILEPSY vs. CONTROL Measure Epileptics Controls p-value (The following are presented in z-scores) PASAT Slow PASAT Fast Digits Forward Digits Backward Corsi Forward Corsi Backward Logical Memory Immediate Delayed Motor Tapping Dominant Non-dominant -2.15 (0.93) -1.69 (0.76) -0.89 (0.45) -0.88 (0.60) -0.74 (0.69) 0.17 (0.67) 0.00 (1.79) 0.12 (1.81) -0.36 (1.42) -0.34 (1.49) -0.43 (0.98) -0.49 (0.89) -0.50 (0.68) -0.72 (0.61) -0.38 (0.54) 0.29 (0.60) 0.93 (1.35) 0.75 (1.21) -0.36 (0.87) 0.05 (0.98) (The following are presented in obtained scores) Selective Reminding Learning Delayed Visual Learning Scaled score Delayed CPT: Reaction time 45.50 (13.02) 7.08 (2.77) 10.33 (3.68) 9.00 (3.57) 49.38 (14.57) Visual Motor Integration 56.65 (7.23) 7.75 (2.53) 10.70 (2.90) 9.20 (2.76) 39.20 (8.81) .0026** .0070** .0416* NS NS NS NS NS NS NS .0062** NS NS NS NS Standard Score 98.45 (7.83) 102.00 (10.00) NS CBCL T-scores Internalizing Externalizing 65.09 (10.88) 61.27 (11.59) 46.43 (11.91) 56.57 (11.21) NS= not significant; *=approaching significance **=significant with p~ .01 .0041** NS 89

PAGE 95

90 TABLE 3.2 TIME 2 NEUROPSYCHOLOGICAL DATA EPILEPSY vs. CONTROL Measure Epileptics Controls p-value (the following are presented in z-scores) PASAT Slow -2.33 (1.47) -0.82 ( 1. 09) .035** PASAT Fast -2.02 (0.89) -0.74 (0.97) .022** Digits Forward -0.93 (0.43) -0.37 (0.68) .024** Digits Backward -0.73 (0.78) -0.42 (0.56) NS Corsi Forward -0.38 (0.71) -0.08 (0.68) NS Corsi Backward 0.40 (0.93) 0.55 (0.79) NS Logical Memory Immediate -0.98 (0.99) -0.50 (1.17) NS Delayed -1. 25 ( 1. 15) -0.41 ( 1. 13) NS Motor Tapping Dominant -0.19 ( 1. 28) -0.43 (1.23) NS Non-dominant 0.18 (0.94) 0.10 (1.00) NS (the following are presented in obtained scores) Selective Reminding Learning 46.40 (14.30) 57.65 (7.55) .012** Delayed 6.20 (3.46) 8.15 (2.89) NS Visual Learning Scaled score 12.00 (4.11) 11.95 (3.14) NS Delayed 9.40 (3.50) 10.55 (3.02) NS CPT: Reaction time 49.81 (12.94) 48.22 (10.76) NS NS= Not Significant;**= significant with p .05.

PAGE 96

91 TABLE 3.3 TIME 3 NEUROPSYCHOLOGICAL DATA EPILEPSY WL CARNITINE vs. EPILEPSY WL CARNITINE PLACEBO vs. CONTROLS Measure Epi+C Epi+P Controls (The following are presented in z-scores) PASAT Slow -1.92 (0.45) -1. 35 (1.93) -0.85 ( 1. 34) PASAT Fast -2.52 (0.39) -1. 64 ( 1. 14) -0.60 (0.96) Digits Forward -0.82 (0.13) -0.40 (0.18) -0.30 (0.76) Backward -0.68 (0.81) -1. 09 (0.42) -0.49 (0.58) Corsi Forward -0.34 (0.83) -0.20 (0.66) -0.28 (0.71) Backward 0.11 (0.97) 0.27 (0.44) 0.50 (0.73) Logical Memory Immediate -1.17 (0.59) 0.79 (1.99) 0.43 (1.60) Delayed -1.99 ( 1. 45) 1.21 (3.30) 0.58 (1. 70) Motor Tapping Dominant -0.23 (1.26) 0.00 (0.87) -0.28 ( 1. 01) Non-dom. 0.31 ( 1. 48) 0.18 (1.30) -0.20 (0.72) (the following are presented in obtained scores) Selective Reminding Learning 51. 25 (12.61) 48.80 (13.20) 56.24 (12.00) Delayed 5.00 (1.83) 5.80 (3.77) 8.35 (2.96) Visual Learning Scale sc. 12.25 (4.03) 12.80 ( 1. 92) 12.59 ( 2. 09) Delayed 8.25 (3.20) 10.60 (2.61) 10.41 (2.72) CPT: Rxn time 43.62 (8.08) 47.94 (3.41) 44.15 (9.21) Visual Motor Integration Stnd sc. 97.50 (15.02) 106.00 (23.31) 108.2 (9.64) CBCL T-scores Int. 65.00 (13.08) 57.40 (10.29) NA Ext. 63.00 (15.39) 57.40 (14.38) NA

PAGE 97

TABLE 3.4 MEMORY AND ATTENTION DATA EPILEPSY vs. CONTROL Measure Time 1 Time 2 (The following are presented in z-scores) PASAT Slow EP CN PASAT Fast EP CN Digits Forward EP CN Backward EP CN Corsi Forward EP CN Backward EP CN Logical Memory Immed. EP CN Delayed EP CN -2.04 -0.49 -1.63 -0.58 -0.83 -0.54 -0.98 -0.81 -0.77 -0.37 0.01 0.36 0.27 0.91 0.56 0.68 (1.02)* (0.97) (0.79)* (0.91) (0.44) (0.73) (0.61) (0.59) (0.76) (0.55) (0.70) (0.53) ( 1. 84) ( 1. 44) ( 1. 49) ( 1. 28) -1.99 -0.73 -1.89 -0.71 -0.89 -0.36 -0.92 -0.31 -0.41 -0.06 0.42 0.64 -0.75 -0.41 -1. 08 -0.35 (1.20)* ( 1. 02) (0.88)* (0.95) (0.44) (0.74) (0.54)* (0.51) (0.74) (0.68) (0.99) (0.79) (0.80) (1.23) ( 1.12) (1.20) Time 3 -1.59 -0.85 -2.02 -0.60 -0.59 -0.30 -0.91 -0.49 -0.26 -0.28 0.20 0.50 -0.05 0.43 -1.16 0.53 (The following are presented in obtained scores) Selective Reminding Learning EP 44.77 CN 56.53 Delayed EP 7.11 CN 8.12 Visual Learning Scaled EP 10.55 CN 11. 35 Delayed EP 8.78 CN 9.65 CPT EP 43.02 CN 40.28 (14.46)* (7.87) (3.14) (2.57) (4.00) (2.57) (3.60) (2.67) (6.64) (9.45) 45.56 57.82 5.89 8.35 12.44 12.65 9.78 11. 41 45.79 48.27 (14.90)* (8.03) (3.52) (2.96) ( 4. 10) (2.71) (3.50) ( 1. 94) (7.30) (11.38) 49.89 56.24 5.44 8.35 12.56 12.59 9.56 10.41 46.21 44.15 NOTE: EP = Epileptic group; CN = Control group ( 1. 42) MG (1.14) (0.96)*MG (0.96) (0.26) (0.76) (0.62) MG (0.58) (0.69) MT (0.71) (0.68) (0.73) (1. 79) MT ( 1. 60) (3.01) MT (1.70) 92 (12.18) MG (12.05) (2.92)* (2.96) (2.83) MT (2.09) (2.96) MT (2.72) (5.27) (9.21) denotes significant group differences p<.05. MG= Main effect for group (p<.05) MT= main effect for time (p<.05)

PAGE 98

CHAPTER 4 DISCUSSION Review of available literature revealed few well controlled studies of the effect of valproate on cognition, specifically memory and attention, in the pediatric epilepsy population. The relative recency of valproate in the role of seizure management, the increasing awareness of specificity of cognitive domains, and the ever growing realization of differences between child and adult cognition likely contribute to the limited findings. studies of other antiepileptic drugs such as phenobarbital and phenytoin, have suggested that antiepileptic medications may in some instances directly affect cognitive studies. The work by Trimble (1987) describes slower reaction time and difficulty with sustained attention with the use of phenobarbital. Trimble (1987) also reports impairments in memory, motor speed, and mental speed with phenytoin. Cognitive deficits characterized by poor attention, impaired learning, and behavior problems have been reported in the epilepsy population at a greater frequency than in 93

PAGE 99

94 the neurologically normal population. Attentional difficulties are over-represented in epileptic children with generalized seizures. Verbal memory deficits have been seen in children with partial epilepsy with a left temporal lobe focus and non-verbal memory deficits have been reported in children with partial epilepsy with a right hemisphere focus (Trimble, 1987). Seidenberg (1986) reported a wider range of IQ in the epilepsy population compared to normals. Often group differences and general deficits are observed, but the characteristics of the deficits (i.e. type of memory problem or clinical significance of the attentional deficits) are not well described, thus limiting generalizability. In other cases, specific information is lost in the attempt to define a patient's global functioning. For instance, the study by Farwell et al. (1985) which studied 118 children with epilepsy reported IQ scores and a neuropsychological rating of "good" or "bad". Hermann (1982) studied 50 children with epilepsy and grouped their subjects according to scores on the Luria Nebraska Neuropsychological Battery into a "poor neuropsychological" or a "good neuropsychological" group. In neither study were cognitive deficits described nor were the antiepileptic drug protocols reported or described. Review of studies attempting to qualify and to quantify the pattern of abilities have revealed serious research

PAGE 100

95 confounds including the heterogeneity of epilepsy for etiology, age of onset, duration of symptoms, previous and ongoing pharmacological intervention and the idiosyncratic characteristics of the patients. These factors confound interpretations of the findings and limit generalizability of findings Results to date suggest that valproate monotherapy is very effective in the treatment of general and complex partial seizures in both adults and children, though not without some cost. The most concerning side effect identified with valproate treatment is liver failure. Upon close examination, the preponderance of cases are reported in the very young, those with other serious disorders, those with previous liver disease, and patients receiving polytherapy (Dreifuss, et al., 1987). Adverse physical side effects related to valproate do occur; however, with careful monitoring they appear to be at a rate and an intensity which is favorable in comparison to alternative medications (Herranz, et al., 1988). Recent literature suggests that the most adverse central nervous system effects attributed to valproate (i.e. sedation and somnolence) occur when it interacts with other antiepileptic medications, polytherapy. A few studies involving neurologically normal adults have been performed. Boxer et al. (1976) reported

PAGE 101

96 drowsiness with monotherapy which worsened with addition of phenobarbital. Thompson and Trimble (1981) reported that valproate could slow decision making, but found no significant effect on measures of concentration, perceptual speed, verbal and non-verbal memory, motor speed, and mood. Harding et al. (1985) did not find significant differences with simple reaction time after administering valproate at two levels or after withdrawal. These studies assessing the effects of valproate in normal adults were of short term effects and did not address the possibility of adjustment to valproate. Smith (1991) in a review article discussed the issue of a tolerance to the effects of valproate which occurs in the first few weeks of drug therapy and this tolerance is characterized by reduction in transient side effects. Use of neurologically normal children to assess valproate monotherapy in a nonepileptic condition is not possible for obvious ethical reasons. The studies of normal adults provide evidence that valproate is not overtly deleterious to cognition as a whole, but specific interpretations regarding children, particularly those with epilepsy cannot be drawn. A powerful confound to understanding the role of valproate on cognition occurs when patients are maintained on another antiepileptic drug such as phenobarbital, primidone, or phenytoin, and valproate is added as

PAGE 102

97 polytherapy or when cognitive ability is compared between valproate and another antiepileptic medication in a cross over design. A study by Sommerbeck et al. (1977) reported that valproate slowed decision making, reaction and tapping speed in a group of 20 epileptics ranging in age from 13 to 63 years who were receiving at least one other antiepileptic medication. Vining et al. (1987) compared the effects of valproate to the effects of phenobarbital on numerous neuropsychological measures in a sample of 21 children with epilepsy with normal IQs, and found significant differences in WISC-R Block Design and the Berkeley Paired Association Learning, with those on phenobarbital performing below the level of those on valproate. Forsythe et al. (1991) compared cognitive performances of 64 children with epilepsy receiving carbamazepine, phenytoin, or valproate. In the presence of equivalent seizure control, those receiving carbamazepine and phenytoin had slower information processing, and memory problems were observed in those taking carbamazepine. Results of these studies report cognitive performances relative to other medications, and do not clearly state the effect of valproate on cognition. The first goal of this study was to describe the memory and attention abilities of children with idiopathic epilepsy who were within the average range of intelligence prior to the initiation of valproate monotherapy. The second goal

PAGE 103

98 was to assess attention and memory performance after one month of valproate monotherapy and compare it to baseline performance to determine the effect of valproate on these abilities. The third goal was to compare performance within the epilepsy group on these same measures in a carnitine and carnitine placebo condition at 12 weeks after baseline testing to determine the effect of adding carnitine, a dietary suppleme~t found to be depleted in many individuals taking valproate. Disentangling the effect of epilepsy from the potential effects of treatment with anticonvulsant medications necessitated the use of a patient group which did not have epilepsy, but was medically and ethically able to participate in valproate therapy. Several studies have provided evidence that valproate is an effective prophylactic medication for migraine (Sorensen, 1988; Hering and Kuritzky, 1992). A relationship between epilepsy and migraine exists; however, epilepsy is considered to more likely involve some form of cerebral dysfunction and is more likely to lead to cognitive deficits. Pediatric migraine does not present with cognitive or neurological deficits in most cases, and there is no specific neuropsychological pattern of deficits associated with pediatric migraine. Therefore, group differences observed following comparison at baseline could be attributed to valproate therapy.

PAGE 104

99 Parallel to the epilepsy group, age matched group of migraineurs randomized to a valproate or valproate placebo condition underwent similar testing and medical evaluation. This group was to provide a control group which had a different neurological substrate, but an identical pharmacological substrate to allow comparisons which could assist disentangling the role of epilepsy from the role of valproate in cognitive performance. A normal control group was added to provide information on the effects of repeated measurement in this design. These normal children were age matched to the two treatment groups and were in the average IQ range. The use of neuropsychological tests in a longitudinal study poses a second set of problems. Determining the contribution of a specific treatment with respect to practice effects, familiarity with the testing process, and time, is difficult and warrants the use of a normal control group. Caution and care designing a battery addressing these issues are imperative. Focussing the assessment on a specific cognitive domain and narrowing the scope of a study are beneficial in that fewer demands are placed on the subject and a more detailed exploration of the predetermined cognitive domain is possible. Use of tests which are relatively stable over time and insensitive to state changes, in conjunction with tests thought to be sensitive

PAGE 105

100 to a given condition, are of benefit to allow comparison of performance consistency. Employment of tests with alternate forms or which are not sensitive to practice effects is essential. Furthermore, neuropsychological tests are not consistently sensitive to subtle changes in cognition and they do not always have a practical generalizability (i.e. to classroom or work environments). Utilization of tasks which have some equivalence to school demands (i.e. attention and learning) could facilitate understanding of some of the difficulties presented to the professional community by the pediatric epileptic population. This study was a short-term follow-up treatment study. All epileptic and migraine subjects were evaluated by members of the Pediatric Neurology Department and these physicians determined medical qualification. Average performance on four selected subtests of the WISC-R, in conjunction with parent report of no psychiatric disorder and parent informed consent were required for inclusion. Baseline testing consisted of measures in the following domains: attention, immediate and delayed verbal memory, immediate and delayed non-verbal memory, verbal learning, non-verbal learning, visual motor integration, and motor speed. Behavior and emotional function were assessed through parent report. The repeat testings were at 4 weeks following initiation of valproate monotherapy and at 8 weeks

PAGE 106

101 following blind randomization to valproate plus carnitine or valproate plus carnitine placebo. The subjects were not funded or rewarded directly. Valproate was supplied to the parents without cost. The subjects underwent monthly neurological evaluations and valproate plasma levels were monitored. The age matched normal control group underwent identical neuropsychological testing, but received no medical assessment or intervention. The hypotheses were as follows. (1) At baseline, the epileptic sample would display mild to moderate deficits in attention and memory given the nature of seizures (i.e. primary or secondary generalization). The migraine sample would have no cognitive impairments at baseline. (2) Following 4 weeks of valproate monotherapy, the epileptic sample would demonstrate more severe deficits in memory and attention. (3) After 12 weeks of valproate monotherapy and 8 weeks of carnitine or carnitine placebo, those subjects receiving carnitine supplementation would perform at a less impaired level than those in the placebo condition and they would report fewer negative side effects. (4) Improvement across most measures sensitive to practice effects would be observed in all the groups and these changes could be controlled for by use of the normal control group. (5) Somatic complaints would be present for the epilepsy and migraine groups at a significant rate at baseline.

PAGE 107

102 Interpretation of Findings In general, the pediatric epilepsy sample demonstrated excellent seizure control with the valproate monotherapy; however, one child was intolerant of the drug and was released from the study prior to the second and third testings. The pediatric migraine sample reported a mixture of efficacy in prophylaxis of migraine, and this matter is dealt with in another study (Fiano, 1994). Though one third of epileptic subjects reported some form of adverse side effect attributed to the valproate monotherapy, the severity was minimal to moderate range and treatment was not discontinued for any other subject. The decision to remain in the study and on valproate monotherapy was made by neurologist and the subject's parent or guardian. At each point in testing, a pattern of results emerged which provided important information. In the Baseline intellectual screening, the pediatric migraine group performed at a significantly lower level on two WISC-R subtests compared to the performance of the normal controls. This was an unexpected finding, and though all the scores were well within the average range as necessitated by the inclusion criteria, this illustrated the need to use an IQ covariate for accurate interpretation of the neuropsychological data.

PAGE 108

103 The performance of the epileptic group at Baseline was significant for deficits in attention and verbal learning. This pattern of deficient functioning in epileptics with generalized seizures has been previously reported in the literature (Trimble, 1987). Specifically, the epileptics' scores on the PASAT in both time conditions and on the Buschke Verbal Learning score were well outside of the normative average range. As expected, given the somatic complaint component, the baseline score in the parent report measure CBCL Internalizing T-score was at the low end of the clinically significant range. This measure further differentiates the epileptic group from the control group according to their behavior. The scores for all the groups at Baseline provided useful clinical information. First, the control group was well within the normative average range across procedures, and most of the scores obtained by the migraine group were within normal limits. The epileptic group performed well on measures of motor speed, visual spatial integration, visual learning and memory, and story memory. This was an expected finding, given the literature review and the strict inclusion criteria of average intellectual functioning and no known brain lesion. The results from study by Vining et al. (1987) demonstrated clinically normal scores on measures of visual spatial perception, achievement, verbal fluency,

PAGE 109

104 vigilance, gross motor and fine motor skills when IQ was in the average range. Also at Baseline, mild clinical differences were observed on the CPT reaction time and Digit Span, with the epileptic group performing below the level of the controls and in the low end of the normative average range. These findings of mildly deficient attentional abilities in a sample of children with generalized epilepsy are consistent with the literature, specifically Trimble (1990) who attributes attention problems to epilepsy as a syndrome and Binnie et al. (1990) who attributes attentional difficulties to subclinical EEG abnormalities. After one month of valproate monotherapy, the epileptic and control groups provided a pattern of test results very similar to that obtained at Baseline. As a group, the epileptics were well within the therapeutic valproate blood plasma range. The epileptic group continued to perform at a significantly lower level on the PASAT in both time condition, digits forward, and Buschke Verbal Learning. These scores remained in the clinically impaired range as well; however, no other group differences were observed. The addition of valproate as a variable did not alter the neuropsychological presentation of the groups; measures that were in the average range remained there and those that were below the average range remained there as

PAGE 110

105 well. Psychomotor slowing and slower decision making, as assessed with the CPT were not significantly different in the epileptic group compared to the normal controls. This is somewhat unexpected given the results of Aman et al. (1987) and Gallassi et al. (1990). Aman and his group found slower response time and more errors of omission on the CPT. Gallassi et al. (1990) reported impaired attention and reaction time in a group of young adults on valproate monotherapy. The lack of significant group differences on the CPT in our study may be due to the small sample sizes, or the CPT may not have been a sufficiently sensitive measure. At the final testing, after 12 weeks on valproate monotherapy and 8 weeks of carnitine or carnitine placebo, nine children in the epilepsy group remained and they were well within the therapeutic range of valproate; five were in the carnitine condition and four were in the carnitine placebo condition. Of note, compliance with the carnitine supplementation cannot be monitored with blood levels, further limiting the analysis of carnitine's effect. Good valproate levels, in conjunction with parent and patient report of compliance, were estimates of carnitine compliance. Given the small sample size, within group comparisons were limited and the effect of carnitine supplementation on the neuropsychological performance of

PAGE 111

epileptics could not be fully explored. However, ANOVA's were performed, but no significant group differences appeared. 106 At the final testing, the combined epilepsy group performed at a significantly lower level on the PASAT Fast condition and on the delayed recall of the Buschke Selective Reminding Test. No other significant group differences were observed. Analysis of the neuropsychological performance of those remaining in the epileptic group and in the control group across the three testing sessions revealed several measures sensitive to a main effect by group: the PASAT Slow and Fast; Digits Backward; and Buschke Verbal Learning. The epileptic group consistently performed better with each exposure on the PASAT Slow condition and on the Buschke Verbal Learning, while the control group's performance remained constant. However, the epileptic group's performance on the PASAT Fast condition worsened with each exposure and the control group remained constant. The epileptic group's performance on Digits Backward remained constant while the control group improved its performance. Main effects for time were observed for Digits Forward, Corsi Forward, and Visual Learning Scaled Score and Delayed Recall with a trend for improvement over the multiple presentations. This is an expected finding and appears to

PAGE 112

107 represent practice/ exposure effects. The time effect observed in our version of the Logical Memory Immediate and Delayed Recall is readily interpretable as consequences of test variation, with the second story more difficult than the others. In the epileptic sample, analyses conducted after the administration of valproate indicated that valproate has little impact on neuropsychological function. The findings of Hwang and Van Woert (1979) suggest that valproate has a greater effect on striatal and hippocampal areas provides some rationale for potential changes in attention due to connections of the striatum to the frontal lobes. Improved seizure control is correlated with improved EEG activity which Binnie et al. (1990) which has been linked with improved attention and concentration. The positive effect of seizure management attributed to valproate monotherapy may have negated the possible negative effect of valproate on attention. In summary, the first hypothesis was supported by the findings of this study. At baseline testing, when IQ was in the average range and covaried, the children in the epilepsy sample demonstrated statistically and clinically significant impairments in some measures of attention and verbal learning. After 4 weeks of valproate monotherapy, the epileptic sample continued to perform at a significantly

PAGE 113

108 lower level than the control group on some measures of attention and learning; however, not at a level significantly different from their baseline performance suggesting that valproate did not impair the attention and memory function in the epilepsy sample. The effect of carnitine on the cognition of those receiving valproate monotherapy could not be fully explored because of the small sample size. Significant differences were not observed in the analyses conducted. Practice effects were observed on several of the measures and across the groups, illustrating the need for a normal control group. At Baseline the epileptic group had a statistically and clinically significant T-score on the Internalizing Scale from the CBCL on which somatic complaints highly load. At the end of the study, the epileptic group was within the clinically normal range on this measure. Pearson Correlation Coefficients were performed on the changes in scores from Baseline to Time 3 on the Pasat variables and the Buschke Verbal Learning variable versus the Internalizing and Externalizing T-scores of the CBCL. No correlations reached significance (p<.05) with scores ranging from 0.16 to 0.45. The lack of significant correlations suggests that changes in behavior and emotional well-being were not significantly related to changes in cognitive function on the above mentioned variables.

PAGE 114

109 Attrition and Non-compliance In a repeated testing, small group design such as was the case for the present study, the effects of noncompliance with the therapy and attrition from the study group needed to be examined. The loss of three subjects from the epilepsy group over a 3 month period equaled 25 percent attrition. Only one of the three subjects was removed from the study for medical reasons. However, valproate blood levels suggests good compliance for those remaining in the study. Attrition was higher in the migraine group (50%). Noncompliance to valproate therapy for those who remained in the study was apparent with monitoring of blood levels. The blood levels of carnitine could not be monitored. Proposed comparison of the effects of valproate maintenance therapy on a pediatric epileptic sample versus a pediatric migraine sample was not possible due to a high rate of attrition in the migraine sample. The length of the study and the length of time between test administrations, one month and two months, may have been too great for patients and their families to feel attached to the study. Several people interacted with the subjects and their families during the test administrations, and in many cases different people met with the subject for repeat testing reducing continuity and the sense of relationship between the subject and the examiner.

PAGE 115

110 Assigning one person to a subject and his family and having more frequent interactions such as weekly phone calls or reminder post-cards may have been beneficial. Approximately 2 hours were required for each testing session, and this may have been aversive to the subjects. Shortening the assessment battery may have also increased the commitment to the study by reducing the demand on the subject and the family members. The subjects in the study did not receive monetary reinforcement directly. The drug was distributed free of charge, providing some incentive for the parents or guardians but no clear incentive for the child participants. Medication compliance was relatively good for the epileptic sample as compared to the migraine sample. The severity of the symptoms of the disorder (i.e. seizures versus migraines) and the pragmatic approach to medicating seizure disorder as compared to prophylactically medicating migraine in an experimental format possibly distinguished the compliance rates for both groups. Further attention to the non-compliance demonstrated by the migraine group is dealt with in another study (Fiano, 1994). Limitations of the Study This study had several limitations. Attempts were made to reduce the subject variables identified as problematic in other studies. Variation in age of onset, duration of seizure disorder, and seizure type were represented in this

PAGE 116

111 epilepsy sample. Two of the children had previous diagnoses of ADDH and one had been treated with Ritalin. Five of the 12 subjects at baseline reported history of significant headaches. This finding is consistent with the literature regarding the relationship between epilepsy and migraine, however, the meaningfulness to the results of this study are unknown. As previously stated, the small sample size, the attrition in both patient groups, and the non-compliance with valproate monotherapy observed in the migraine group created statistical confounds with prevented addressing the hypotheses involving the effect of carnitine and the differential response of valproate in a neurologically different sample. The small sample size at baseline (46 total subjects) and the use of many dependent variables will likely raise the issue of Type I errors. P-values of less than .OS and Tukey pairwise comparisons were chosen because of their conservative nature. With the exception of Digit Span, the differences observed at baseline were very significant, consistent with previous literature, and replicated throughout the repeated measures analyses. The issue of statistical power is also of importance due to the small sample size and level of test sensitivity. Confidence intervals at the 95 percent confidence level were performed on the Pasat variables, Digit Spans, and Buschke

PAGE 117

112 variables for the time interval of Baseline to Time 2 and Baseline to Time 3. These confidence intervals demonstrated an overlap between the epileptic group and the normal control group which suggests inadequate power. This raises the possibility of a Type II error. Attention and verbal learning were domains sensitive to group differences. However, differences between verbal memory and non-verbal learning and memory never reached statistical significance. From a clinical perspective, the epileptics and controls did not differ in these areas. The use of the Logical Memory stories from the WMS Form I and II was based on assumed equivalence. However, the out-dated language of the prose, particularly on the second story, deflated scores in all groups; the level of difficulty may have obscured group differences. Only one form of the WRAML Visual Learning is available and significant practice effects were observed and this may have obscured group differences. The version of the CPT used in this study was approximately 20 minutes in duration which several subjects reported as aversive. Furthermore, computer malfunctions reduced the number of successful administrations which may have confounded determination of group measures. Conclusions The main goal of this study was to assess the effect of valproate on attention and memory in a pediatric epilepsy

PAGE 118

113 population. A secondary goal was to describe the neuropsychological characteristics in a subgroup of this population, i.e. those with minimal neurological insults as assessed by normal MRI and average intelligence. A third goal was to determine the effect of carnitine supplementation on memory and attention. Despite the limitations presented previously in this paper, several conclusions have been drawn. After controlling for normal intelligence, children with epilepsy present with mild deficits in verbal learning and attention. This finding is consistent with the literature to this point; however, this study attempted to limit the sources for these deficits to the presence of seizure disorder. The results of this study suggest good compliance with and response to valproate monotherapy with children with epilepsy. Adverse physical side effects were minimal. Valproate monotherapy did not worsen or improve the attention and memory functions of the epilepsy group. One explanation for this finding may be that a balance was achieved between the positive effects of having well controlled seizures and any attention and memory deficits resulting from the valproate therapy. Adequate comparison to the migraine sample might have provided better resolution of this question, however, attrition and non-compliance from this patient group precluded these analyses.

PAGE 119

114 Future research in the study of the neuropsychology pediatric seizure disorder and its treatment may benefit from use of a limited age range; better, more sensitive measures of attention and memory; larger sample sizes; and address the issues of attention and memory in those below the average range of intelligence.

PAGE 120

REFERENCES Abikoff, H., Alvir, J., Hong, G., Sukoff, R., Orazio, J., Solomon, s., and Saravay, s. (1987). Logical memory subtest of the Wechsler Memory Scale: age and education norms and alternate-form reliability of two scoring systems. Journal of Clinical and Experimental Neuropsychology, .2.1...4.l:435-448. Achenbach, T.M., & Edelbrock, c.s. (1986). Child Behavior Checklist and Youth Self-Report. Burlington, VT: Author. Adman, M.G., Weary, J.S., Paton, J.W. and Turbott, S.H. (1987). Effect of sodium valproate on psychomotor performance in children as a function of dose, fluctuations in concentration, and diagnosis. Epilepsia, 28(2) :115-124. Aldenkamp, A.P., Alpherts, W.C.J., Blennow, G., Elmqvist, D., Heijbel, J., Nilsson, H.L., Sandstedt, P., Tonnby, B., Wahlander, L., Wosse, E. (1993). Withdrawal of antiepileptic medication in children -effects on cognitive function: The multicenter Holmfrid study. Neurology. 43(1):41-50. Alpherts, w.c.J. and Aldenkamp, A.P. (1990). Computerized neuropsychological assessment of cognitive functioning in children with epilepsy. Epilepsia, 3l(Suppl.4):S35-S40. American Academy of Pediatrics Committee on Drugs (1985). Behavioral and cognitive effects of anticonvulsant therapy. Pediatrics, 76(4):644-647. Andrasik, F., Kabela, E., Quinn, s. Attanasio, V., Blanchard, E.B., and Rosenblum, E.L. (1988). Psychological functioning of children who have recurrent migraine. Pain, 34:43-52. 115

PAGE 121

116 Armijo, J.A., Herranz, J.L., Arteaga, R., and Valiente, R. (1986). Poor correlation between single-dose data and steady state kinetics for phenobarbitone, primidone, carbamazepine and sodium valproate in children during monotherapy. Clinical Pharmacokinetics, 11:323-335. Beery, K.E. (1967). Developmental Test of Visual-Motor Integration. Administration and Scoring Manual. Follett Publishing, Chicago. Beghi, E., Bizzi, A., Codegoni, A.M., Trevisan, D., and Torri, w. (1990). Valproate, carnitine metabolism, and biochemical indicators of liver function. Collaborative group for the study of epilepsy. Epilepsia, 31(3} :346-352. Bennett, T.L. (ed.} (1992}. The Neuropsychology of Epilepsy. Plenum Press, New York. Bennett-Levy, J. and Stores, G. (1984}. The nature of cognitive dysfunction in school children with epilepsy. Acta Neurology Scandinavia, 69(Suppl. 99):79-82. Binnie, C.D., Channon, s., and Marston, D. (1990). Learning disabilities in epilepsy: neurophysiological aspects. Epilepsia, 3l(Suppl.4):S2-S8. Boxer, c., Herzberg, J.L., and Scott, D.F. (1976). Has sodium valproate hypnotic effects? Epilepsia, 17:367-370. Clodfelter, C.J., Dickson, A.L., Newton Wilkes, c., and Johnson, R.B. (1987). Alternate forms of selective reminding for children. Clinical Neuropsychologist, 1:243-249. Commission on Classification and Terminology of the International League Against Epilepsy (1981}. Proposal for revised clinical and electroencephalographic classification of epileptic seizures. Epilepsia, 14:741-751. Cooley, E.L. and Morris, R.D. (1990). Attention in children: a neuropsychologically based model for assessment. Developmental Neuropsychology, .__(]_)_:239-274. Corbett, J.A., Trimble, M.R., and Nichol, T.C. (1985}. Behavioral and cognitive impairments in children with epilepsy: The long-term effects of anticonvulsant

PAGE 122

117 therapy. Journal of the American Academy of Child Psychiatry, 24:17-23. Cosden, M. (1985). Developmental Test of Visual-Motor Integration. In D.J. Keyser & R.C. Sweetland, (Eds.), Test Critiques (Vol. 4, pp. 229-237). Kansas City, Missouri: Test Corporation of America. Coulter, D.L. (1991). Carnitine, valproate, and toxicity. Journal of Child Neurology, .Q....(.ll_:7-14. Covanis, c., Gupta, A.K., and Jeavons, P.M. (1982). Sodium Valproate: monotherapy and polytherapy. Epilepsia, 23:693-720. Crosson, B., Hughes, c.w., Roth, D.L., and Monkowski, P.G. (1984). Review of Russell's (1975) norms for the Logical Memory and Visual Reproduction subtests of the Wechsler Memory Scale. Journal of Consulting and Clinical Psychology,52(4):635-641. Dar, R., Serlin, R.C., and Omer, H. (1994). Misuse of statistical tests in three decades of psychotherapy research. Journal of Consulting and Clinical Psychology. 62 (1): 75-82. Dodson, W.E. (1993). Felbamate in the treatment of LennoxGastaut syndrome: results of a 12-month open label study following a randomized clinical trial. Epilepsia, 34(suppl.7): S18-24. Dreifuss, F.E., Santilli, N., Langer, D.H., Sweeney, Moline, and Menander (1987). Valproic acid: hepatic fatalities. A retrospective review. Neurology, 37:379-380. DuPaul, G.J., Guevremont, D.C., and Barkley, R.A. (1991). Attention Deficit-Hyperactivity Disorder in Adolescence: Critical assessment parameters. Clinical Psychology Review, 11:231-245. Eadie, M.J. and Tyrer, J.H. (1989). Anticonvulsant Therapy -Pharmacological Basis and Practice, Third Edition. Churchill Livingstone, New York. Ellenberg, J.H., Hirtz, D.G., and Nelson, K.B. (1986). Do seizures in children cause intellectual deterioration? The New England Journal of Medicine, 314(17):1085-1088.

PAGE 123

118 Engel, J. Jr. (1989). Seizures and Epilepsy. F. A. Davis Company, Philadelphia. Enna, S.J. and Beutler, J.A. (1985). GABA receptor as a site for antiepileptic drug action. In: Epilepsy and GABA Receptor Agonists. Editors: Bartholini, Bossi, Lloyd, and Morselli. Raven Press, New York. Farwell, J.R., Dodrill, C.B., and Batzel, L.W. (1985). Neuropsychological abilities of children with epilepsy. Epilepsia, 26(5):395-400. Fenichel, G.M. (1988). Clinical Pediatric Neurology. W.B. Saunders Company, Philadelphia. Fennell, R.S., Fennell, E.B., Carter, R.L., Mings, E.L., Klausner, A.B., and Hurst, J.R. (1990). Association between renal function and cognition in childhood chronic renal failure. Pediatric Nephrology, ~:16-20. Fiano, K. (1994). The neuropsychological effects of valproate with and without carnitine supplement in a pediatric migraine sample. A dissertation presented at the University of Florida. Unpublished manuscript. Fisher, R.S. (1993). Emerging antiepileptic drugs. Neurology. 43(11 Suppl.5): s12-s20. Forsythe, I., Butler, R., Berg, I. and McGuire, R. (1991). Cognitive impairment in new case of epilepsy randomly assigned to carbamazepine, phenytoin and sodium valproate. Developmental Medicine and Child Neurology.ld.:524-534. Freeman, J.M., Vining, E.P., Cost, s. and Singhi, P. (1994). Does carnitine administration improve the symptoms attributed to anticonvulsant medications?: a doubleblinded, cross-over study. Pediatrics. 96:893-895. Gallassi, R., Morreale, A., Lorusso, s., Procaccianti, G., Lugaresi, E. and Baruzzi, A. (1990). Cognitive effects of valproate. Epilepsy Research, ~:160-164. Giordani, B., Berent, s., Sackellares, J.C., Rourke, D., Seidenberg, M., O'Leary, D.S., Dreifus, F.E., and Boll, T.J. (1985). Intelligence test performance of patients with partial and generalized seizures. Epilepsia, 26(1) :37-42.

PAGE 124

119 Giovannini, M., Agostoni, c., and Salari, P.C. (1991). Is carnitine essential in children? Journal of International Medical Research, 19(2):88-102. Goa, K.L. and Sorkin, E. M. (1993). Gabapentin. A review of its pharmacological properties and clinical potential in epilepsy. Drugs, 46(3): 409-427. Golden C.J. (1981). The Luria-Nebraska children's battery: Theory and initial formulation. In: Hynd G. Obrzut J. eds. Neuropsychological assessment and the school-age child: Issues and procedures. New York: Grune and Stratton. Gram, L., Flachs, H., Wurtz-Jorgensen, A., Parnas, J., and Andersen B. (1979). Sodium valproate, serum level and clinical effect in epilepsy: a controlled study. Epilepsia, 20:303-312. Graves, N.M. (1993). Felbamate. Annals of Pharmacotherapy, 27(9): 1073-1081. Gronwall, D.M.A., and Samson, H. (1974). The Psychological Effects of Concussion. Auckland, N.Z.: Auckland University Press. Gronwall, D.M.A., and Wrightson, P. (1974). Delayed recovery of intellectual function after minor head injury. Lancet, ~:605-609. Gronwall, D.M.A., and Wrightson, P. (1981). Memory and information processing capacity after closed head injury. Journal of Neurology, Neurosurgery, and Psychiatry, 44:889-895. Harding, G.F.A., Alford, C.A., and Powell, T.E. (1985). The effect of sodium valproate on sleep, reaction times, and visual evoked potential in normal subjects. Epilepsia, 26(6):597-601. Hartlage, L.C., and Hartlage, P.L.(1989). Neuropsychological aspects of epilepsy. Chapter 22 in Handbook of Clinical Child Psychology. C.R. Reynolds and E. Fletcher-Janzen, eds. Plenum Press, New York. Hering, R. & Kuritzky, A. (1992). Sodium valproate in the prophylactic treatment of migraine: a double-blind study versus placebo. Cephalagia, 12, 81-84.

PAGE 125

Hermann, B.P. (1982). Neuropsychological functioning and psychopathology in children with epilepsy. Epilepsia, 23(4):545-554. Herranz, J.L., Armijo, J.A., and Arteaga, R. (1988). Clinical side effects of phenobarbital, primidone, phenytoin, carbamazepine, and valproate during monotherapy in children. Epilepsia, 29(4):794-804. 120 Herranz, J.L., Arteaga, R., and Armijo, J.A. (1982). Side effects of sodium valproate in monotherapy controlled by plasma levels: a study in 88 pediatric patients. Epilepsia, 21_:203-214. Howell, D. (1992). Statistical Methods for Psychology -Third Edition. PWS-Kent Publishing Company. Boston Massachusetts. Huberty, C.J. and Morris, J.D. (1989). Multivariate analysis versus multiple univariate analyses. Psychological Bulletin, 105:302-308. Hwang, E.C., and Van Woert, M.H. (1979). Effect of valproic acid on serotonin metabolism. Neuropharmacology, 18:391-397. Jensen, P.K. (1993). Felbamate in the treatment of refractory partial-onset seizures. Epilepsia, 34(suppl. 1.l: S25-S29. Johnston, D. (1984). Valproic acid: Update on its mechanisms of action. Epilepsia, 25(Suppl.):Sl-S4. Kolb, B., & Whishaw, I.Q. (1990). Fundamentals of Human Neuropsychology -Third Edition. W.H. Freeman and Company, New York. Levy, R.H. (1984). Variability in level-dose ration of valproate: monotherapy versus polytherapy. Epilepsia, 25 (Suppl.l): Sl0-S13. Lezak, M. (1983). Neuropsychological Assessment Second Edition. Oxford University Press, New York. Lloyd, K.G., Bossi, L., Morselli, P.L., Rougier, M., Loiseau, P., and Munari, c. (1985). Biochemical evidence for dysfunction of GABA neurons in human epilepsy. In: Epilepsy and GABA Receptor Agonists. Editors: Bartholini, Bossi, Lloyd, and Morselli. Raven Press, New York.

PAGE 126

121 MacMillan, v. (1979). The effect of the anticonvulsant valproic acid on cerebral indole amine metabolism. Canadian Journal of Physiology and Pharmacology, 57:843-847. Masur, D.M., Fuld, P.A., Blau, A.O., Thal, L.J., Levin, H.S., and Aronson, M.K. (1989). Distinguishing normal and demented elderly with the Selective Reminding Test. Neuropsychology, 40:391-394. Melegh, B., Kerner, J., Acsadi, G., Lakaros, J., and Sandor, A. (1990). L-carnitine replacement therapy in chronic valproate therapy. Neuropediatrics, 21(1):40-43. Milner, B. (1971). Interhemispheric differences in the localization of psychological processes in man. British Medical Bulletin, 27:272-277. Noback, C.R., Strominger, N.L., and Demarest, R.J. (1991). The Human Nervous System -Fourth Edition. Lea & Febiger, Philadelphia. Opala, G., Winter, s., Vance, c., Vance, H., Hutchinson, H.T., and Linn, L.S. (1991). The effect of valproic acid on plasma carnitine levels. American Journal of Diseases of Children. 145(9):999-1001. Physician's Desk Reference (1990). Barnhart, E.R. (ed.). Oradell, NJ: Medical Economics Company, Inc. Prevey, M.L., Mattson, R.H., and Cramer, J.A. (1989). Improvement in cognitive functioning and mood state after conversion to valproate monotherapy. Neurology, 39:1640-1641. Rapin, I. (1982). Children With Brain Dysfunction. Raven Press, New York. Reitan, R.M. and Wolfson, D. (1985). The Halstead-Reitan Neuropsychological Test Battery: Theory and Interpretation. Neuropsychology Press, Tucson. Reynolds, E.H. and Trimble, M.R. (1985). Adverse neuropsychiatric effects of anticonvulsant drugs. Drugs, 2.2_:570-581. Rugland, A.L. (1990). Neuropsychological assessment of cognitive functioning in children with epilepsy. Epilepsia, 3l(Suppl.4):S4l-S44.

PAGE 127

122 Russell, E.W. (1975). A multiple scoring method for assessment of complex memory functions. Journal of Consulting and Clinical Psychology, 43:800-809. Sacks, o. (1985). Migraine. University of California Press, Berkeley. Sattler, J.M. (1988). Assessment of Children. San Diego, California: Sattler. Scarpa, P., and Carassini, B. (1982). Partial epilepsy in childhood: clinical and EEG study of 261 cases. Epilepsia, 23:333-341. Schmidt, D. (1984). Adverse effects of valproate. Epilepsia, 25(Suppl.1):S44-S49. Schmidt, D. (1993). Felbamate: successful development of a new compound for the treatment of epilepsy. Epilepsia, 34(suppl.7): S30-S33. Schmidt, D. and Loscher, w. (1981). GABA concentrations in cerebrospinal fluid and plasma of patients with epileptic seizures. In: Neurotransmitters. Seizures. and Epilepsy. Editors: Morselli, Lloyd, Loscher, Meldrum, and Reynolds, Raven Press, New York. Seidenberg, M., Beck, N., and Geisser, M. (1986). Academic achievement of children with epilepsy. Epilepsia, 27:753-759. Sheslow, D. and Adams, w. (1990). WRAML-Wide Range Assessment of Memory and Learning Administration Manual. Wilmington, DE: Jastak Associates, Inc. Smith, D. E. (1991). Cognitive effects of antiepileptic drugs. Advances in Neurology. Editors: Smith, Treiman, and Trimble, Raven Press, Ltd. New York. Sommerbeck, K.W., Theilgaard, A., Rasmussen, K.E., Lohren, v., Gram, L., and Wulff, K. (1977). Valproate sodium: evaluation of so-called psychotropic effect. A controlled study. Epilepsia, 18:159-167. Sorensen, K.V. (1988). Valproate: A new drug in migraine prophylaxis. Acta Neurologia Scandinavica, 78, 346-348.

PAGE 128

123 Spreen, o., and Strauss, E. (1991). A Compendium of Neuropsychological Tests. Oxford University Press: New York. Spreen, o., Tupper, D., Risser, A., Tuokko, H., and Edgell, D. (1984). Human Developmental Neuropsychology. Oxford University Press, New York. Stedman's Medical Dictionary. (1982). Williams & Wilkes, Baltimore. Stores, G. (1971) Cognitive function in children with epilepsy. Developmental Medicine and Child Neurology, 13:390-393. Stumpf, D.A., Parker, W.D., and Angelini, c. (1985). Carnitine deficiency, organic acidemias, and Reye's syndrome. Neurology, 35:1041-1045. Sugimoto, T., Nishida, N., Murakami, K., Woo, M., Sakane, Y., Yasuhara, A., Shuto, H., Hatanaka, T., and Kobayashi, Y. (1990). Valproate-induced hepatotoxicity: protective effect of L-carnitine supplementation. The Japanese Journal of Psychiatry and Neurology, 44(2):387-388. Theisler, c.w. (1990). Migraine Headache Disease. Aspen Publishers, Inc., Gaithersburg, Maryland. Thom, H., Carter, P.E., Cole, G.F., and Stevenson, K.L. (1991). Ammonia and carnitine concentrations in children treated with sodium valproate compared with other anticonvulsant drugs. Developmental Medicine and Child Neurology, 33(9):795-802. Thompson, P.J. and Trimble, M.R. (1981). Sodium valproate and cognitive functioning in normal volunteers. Journal of Clinical Pharmacy, 12:819-924. Thompson, P.J. and Trimble, M.R. (1982). Anticonvulsant drugs and cognitive functions. Epilepsia, 23:531-544. Trimble, M.R. (1987). Anticonvulsant drugs and cognitive function: a review of the literature. Epilepsia, 28(Suppl.3):S37-S45. Trimble, M.R. (1990). Antiepileptic drugs, cognitive function, and behavior in children: evidence from recent studies. Epilepsia, 3l(Suppl.4):S30-S34.

PAGE 129

124 Vining, E.P.G., Mellits, E.D., Cataldo, M.F., Dorsen, M.M., Spielberg, S.P., and Freeman, J.M. (1983). Effects of phenobarbital and sodium valproate on neuropsychological function and behavior. Annals of Neurology, 14(3) :360. Vining, E.P.G., Mellits, E.D., Dorsen, M.M., Cataldo, M.F., Quaskey, S.A., Spielberg, S.P., and Freeman, J.M. (1987). Psychologic and behavioral effects of antiepileptic drugs in children: a double-blind comparison between phenobarbital and valproic acid. Pediatrics, 80(2):165-174. Wilder, B.J. (1987). Treatment considerations in anticonvulsant monotherapy. Epilepsia, 28(Suppl.2):SlS7. Wolff's Headache and Other Head Pain -Fifth Edition. (1985) D.J. Dalesio, Editor. Oxford University Press, New York. Wood, J.H., Hare, T.A., Glaeser, B.S., Ballenger, J.C., and Post, R.M. (1979). Low cerebral spinal fluid gammaaminobutyric acid content in seizure patients. Neurology, 29:1203-1208.

PAGE 130

BIOGRAPHICAL SKETCH Margaret Booth-Jones was born in Norfolk, Virginia, on May 21, 1965, while her father was stationed there with the U.S. Navy. She lived throughout the northeast until her teen years when her family moved to Nokomis, Florida. She enrolled at the University of Florida on an early acceptance program, and completed her Bachelor of Science degree in Neurobiology in December 1986. After two years of working for the Department of Neuroscience and the Department of Hyperbaric Medicine at the University of Florida, Margaret was accepted to the graduate program in the Department of Clinical and Health Psychology in the Fall of 1988. She earned her Master of Science degree in 1992. She was accepted at the University of Chicago's neuropsychology internship program for 1993 through 1994. Margaret is currently completing a clinical fellowship at the Department of Clinical and Health Psychology at the University of Florida and has accepted a position there as a Visiting Assistant Professor for 1995-1996. 125

PAGE 131

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Eileen B. Fennell, Chair Professor of Clinical and Health Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the deg ee of Doctor of Philosophy. us sell auer Associate Professor of Clinical and Health Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. / ~Cf!~ '-'cynt~i~ D. Belar Prof~sor of Clinical and Health Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, n scope and quality, as a dissertation for the degree c of Philosophy. es R. Rodi ue s ociate Pro ssor of Clinical Health Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree Doctor of Philosophy. ard Maria ociate Professor of Neuroscience

PAGE 132

This dissertation was submitted to the Graduate Faculty of the College of Health Related Professions and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. May 1995 ~~,-k:_ Dean, College of Health Related Professions Dean, Graduate School