Callosal morphology in epilepsy and neuropsychological test performance

MISSING IMAGE

Material Information

Title:
Callosal morphology in epilepsy and neuropsychological test performance influence of early versus late seizure onset
Alternate title:
Influence of early versus late seizure onset
Physical Description:
ix, 121 leaves : ill. ; 29 cm.
Language:
English
Creator:
Flynn, Thomas Bernard, 1963-
Publication Date:

Subjects

Subjects / Keywords:
Corpus Callosum -- anatomy & histology   ( mesh )
Corpus Callosum -- physiology   ( mesh )
Epilepsy   ( mesh )
Neuropsychological Tests   ( mesh )
Cognition   ( mesh )
Magnetic Resonance Imaging   ( mesh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Statement of Responsibility:
by Thomas Bernard Flynn.
Thesis:
Thesis (Ph. D.)--University of Florida, 1996.
Bibliography:
Includes bibliographical references (leaves 102-120).
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
Permission granted to the University of Florida to digitize, archive and distributed this item for non-profit and educational purposes only. Any reuse of this item in excess of fair use requires permission of the copyright holder.
Resource Identifier:
oclc - 49349343
ocm49349343
System ID:
AA00020007:00001

Full Text









CALLOSAL MORPHOLOGY IN EPILEPSY
AND NEUROPSYCHOLOGICAL TEST PERFORMANCE:
INFLUENCE OF EARLY VERSUS LATE SEIZURE ONSET











By

THOMAS BERNARD FLYNN


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


1996













ACKNOWLEDGMENTS


It would not have been possible to complete this dissertation without the

encouragement and generous support of my committee members. I would like to express

my gratitude to Dr. Gilmore for her time as part of my committee, and for providing

access to Epilepsy Monitoring Unit records. Dr. Crosson's thoughtful comments were

always appreciated, and contributed to my conceptualization of the research. I owe a

particular debt to Dr. Leonard who made the resources or her laboratory available to me

without hesitation, and gave generously of her time, attention, and counsel. Dr. Bauer's

support has always been important to me, and I am grateful for his willingness to work

with me on this project. To Dr. Fennell, I owe thanks and appreciation for seeing me

through not only this dissertation, but a graduate career as well.













TABLE OF CONTENTS


page
ACKNOW LEDGMENTS ............................................... 11ii

LIST OF TABLES ................................................... vi

LIST OF FIGURES .................................................. viii

AB STRA CT ........................................................ ix

IN TROD U CTION .................................................... 1

Background ....................... ........................... 1
Normal Anatomy and Callosal Development ...................... 1
Abnormalities Involving the Corpus Callosum ..................... 7
Agenesis of the corpus callosum ......................... 7
Sectioning of the corpus callosum ....................... 10
Influences on Development and Morphology of the Corpus Callosum .. 13
Effects of early events and experiences on callosal development 13
Effect of gender on callosal morphology .................. 17
Effects ofhandedness and age on callosal morphology ....... 18
Relationships between Callosal Morphology and Cognitive Functioning 19
IQ and other global measures of cognitive performance ....... 19
Other neuropsychological measures ...................... 24
Attention-deficit/hyperactivity disorder ................... 26
Callosal Morphology and Intellectual Functioning ................. 28
Study Rationale and Predictions .................................... 34
Initial Study Objectives ..................................... 34
Anterior Callosal Morphology and Wisconsin Card Sorting Tes
Perform ance ....................................... 35
Posterior Callosal Morphology and Facial Recognition Test
Perform ance ....................................... 42








page
M E TH O D .......................................................... 45

Subjects ...................................................... 45
Neuropsychological M measures ..................................... 47
M orphometric M measure .......................................... 48
Statistical A analysis .............................................. 52
R liability ............................................... 52
Approach to Data Analysis .................................. 52

R E SU L T S .......................................................... 55

Neuropsychological M measures ..................................... 55
V erbal Scale IQ .......................................... 55
Performance Scale IQ ...................................... 55
Full Scale IQ ............................................ 56
Freedom from Distractibility Index ........................... 56
WCST Perseverative Error Score ............................. 56
Facial Recognition Test Score ............................... 58
Grouping Based on Age at Seizure Onset ....................... 58
M orphom etric D ata ............................................. 59
Inter-rater Reliability ...................................... 59
Intra-rater Reliability ...................................... 60
M easurements of Brain Size ................................. 60
Brain Size and Age at Seizure Onset ........................... 61
Measurements of Corpus Callosum ............................ 62
Relationship Between Morphometric and Neuropsychological Measures ..... 63
Full Scale IQ ............................................ 63
Correlation of FSIQ with area of callosal regions (R1-R7) ..... 63
Comparison of correlations for subgroups based on age at
seizure onset, sex and laterality of seizure focus ....... 65
V erbal Scale IQ .......................................... 67
Correlation of VIQ with area of callosal regions (R1-R7) ..... 67
Comparison of correlations for subgroups based on age at
seizure onset, sex and laterality of seizure focus ....... 68
Performance Scale IQ ...................................... 72
Correlations with area of callosal regions (R1-R7)........... 72
Comparison of correlations for subgroups based on age at
seizure onset, sex and laterality of seizure focus ....... 73
Freedom from Distractibility Index ............................ 74
Correlations with area of callosal regions (R1-R7)........... 74







page
Comparison of correlations for subgroups based on age at
seizure onset, sex and laterality of seizure focus ....... 75
Correlations of Anterior and Posterior Callosal Areas with Performance
on Wisconsin Card Sorting and Facial Recognition Tests ..... 76

D ISCU SSIO N ....................................................... 80

Overview and Summary .......................................... 80
Patient Sam ple ........................................... 83
Neuropsychological Measures ............................... 84
"Early" versus "Late" Seizure Onset ........................... 85
M orphom etric D ata ............................................. 85
B rain Size ............................................... 85
Area of the Corpus Callosum ................................ 86
Relationships Between Morphometric and Neuropsychological Measures ..... 87
W AIS-R M measures ........................................ 87
Results Related to Age at Seizure Onset ........................ 88
Gender Differences ........................................ 89
Other Neuropsychological Measures ........................... 90
"What Does the Corpus Callosum do for a Living?" ............... 91
Role in Attentional Regulation ............................... 91
Potential Advantages of Interhemispheric Processing .............. 92
Potential Mechanisms of Callosal Action ....................... 93
Study Lim stations ............................................... 99
Closing Sum m ary .............................................. 100

LIST OF REFERENCES ............................................. 102

BIOGRAPHICAL SKETCH ........................................... 121













LIST OF TABLES


Table page
1 Demographic Features Patient Sample (N=62) ......................... 47

2 Descriptive statistics for neuropsychological measures ................... 57

3 Demographic features and neuropsychological test performance of subjects
with "early" and "late" seizure onset ........................... 59

4 Measurements of brain size in the total sample and separately for males and
fem ales ................................................. 6 1

5 Midsagittal area (cm2) of the corpus callosum (N = 62) ................. 62

6 Correlations of Full Scale IQ with area (cm2) ofcallosal subregions (N = 62) .. 64

7 Correlations of Full Scale IQ and area (cm2) ofcallosal subregions for
subjects with "early" (n = 20) and "late" (n = 42) seizure onset ....... 66

8 Correlations of Verbal Scale IQ with area (cm2) of each callosal subregion .... 68

9 Correlations of Verbal Scale IQ and area (cm2) of callosal subregions for
subjects with "early" (n = 20) and "late" (n = 42) seizure onset ....... 69

10 Correlations of Performance Scale IQ with each of the callosal subregions .... 72

11 Correlations of Performance Scale IQ and each of the callosal subregions
for subjects with "early" (n = 20) and "late" (n = 42) seizure onset .... 74

12 Correlations between FDI scores and area of callosal regions .............. 75

13 Correlations between callosal area (anterior vs. posterior in cm2) and test
score (WCST vs. Facial Recognition) .......................... 77

14 Correlations between callosal area (anterior vs. posterior) and test score
(WCST vs. Facial Recognition) for all subjects and separately for
subjects with "early" and "late" seizure onset .................... 78












LIST OF FIGURES

Figure page
1 Diagram depicting a midsagittal brain image highlighting division of the
corpus callosum into seven subregions.......................... 51

2 Scatterplot of VIQ and area of the splenium (R7) with regression lines
plotted separately for subjects with "early" and late" seizure onset ... 71

3 Scatterplot of anterior area cm2 of the corpus callosum and Facial
Recognition Test score in subjects (n=l 1) with and "early" onset of
seizures ................................................ 79













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



CALLOSAL MORPHOLOGY IN EPILEPSY
AND NEUROPSYCHOLOGICAL TEST PERFORMANCE:
INFLUENCE OF EARLY VERSUS LATE SEIZURE ONSET

By

Thomas Bernard Flynn

December, 1996

Chairperson: Eileen Fennell
Cochairperson: Russell Bauer
Major Department: Clinical and Health Psychology

The corpus callosum is a prominent cerebral structure linking most cortical areas in

both cerebral hemispheres. Multiple lines of evidence suggest that callosal projections

may play a role in the highest-order and latest maturing neural networks in the brain. In

humans, individual differences in area of the corpus callosum have been linked to

performance on neuropsychological measures of cognitive functioning. Sixty-two right-

handed people with intractable epilepsy, left hemisphere language dominance, and a

unilateral temporal lobe seizure focus were tested with the WAIS-R, the Wisconsin Card

Sorting, and Facial Recognition Tests. Morphometric measures of the corpus callosum

and brain size were made from midsagittal magnetic resonance images, and correlations







with measures of cognitive performance were explored. A significant positive correlation

was found between size of the callosal region including fibers linking posterior association

cortex and Verbal and Full Scale, but not Performance IQ. This result replicated a finding

previously reported by Strauss, Wada, and Hunter in a similar patient sample. The finding

did not differ depending upon subject's sex or laterality of seizure focus, and were not

attributable to differences in overall brain size. However, larger area of the posterior

callosum was significantly associated with higher IQs only among subjects with an onset

of chronic seizures after age five. Among subjects with an earlier seizure onset, the

direction of the correlation appeared to be reversed. A significant sex difference was also

found, with larger area of an anterior callosal region significantly correlated with higher IQ

in females and lower scores in males. The study also demonstrated a positive correlation

between a WAIS-R measure of Freedom from Distractibility and area of the callosal

region containing fibers from posterior parietal cortex. Predicted relationships between

areas of anterior and posterior callosum and WCST Perseverative Error and Facial

Recognition Test scores were not supported. The findings are discussed in the context of

evidence that interhemispheric communication contributes to the dynamic allocation of

attentional and information processing resources. The study is limited by the fact that all

subjects had epilepsy, a chronic condition affecting the central nervous system, and by the

correlational nature of the research.













INTRODUCTION


Background


The corpus callosum is the main fiber tract interconnecting the cerebral

hemispheres. In mature brain it is heavily myelinated and distinct in appearance from

surrounding brain structures. Its size, appearance, and central location in the brain have

drawn attention since antiquity (Bogen, 1993). Harris (1995) reviewed the long history of

attempts to understand the structure and function of the corpus callosum highlighting the

importance of interactions among anatomical, physiological, and behavioral evidence.



Normal Anatomy and Callosal Development

Based on light microscopy of the human corpus callosum, Tomasch (1954)

estimated that more than 200 million fibers pass through the callosum. This figure has

been frequently cited since its publication, and continues to be referenced by some

authors. However, Koppel and Innocenti (1983) reported that in comparison to electron

microscopy, light microscopy underestimates the number of fibers in the corpus callosum

of the cat by more than four times. Reasoning by analogy, they estimated that the number

of axons in the human corpus callosum may be more than 800 million. This is consistent

with the number of axons estimated by Aboitiz, Scheibel, Fisher and Zaidel (1992) based









on examination of a human corpus callosum using electron microscopy. As noted by

Bogen (1993), the corpus callosum is larger than all the descending and ascending tracts

taken together which connect the cerebrum with the outside world.

Growth of the corpus callosum has been found to continue in humans up to the

middle twenties, making it one of the latest-maturing cerebral structures (Pujol, Vendrell,

Junque, & Marti Vilalta, 1993). Pujol et al. (1993) reported that development of the

corpus callosum parallels cognitive maturation, and suggested that the corpus callosum

may be part of the highest order, latest maturing neural network of the brain. Substantial

variability in the cross sectional area of the corpus callosum has been described in normal

human subjects (Demeter, Ringo, & Doty, 1988). Demeter et al. (1988) reported that the

variability is not simply accounted for by sex differences or variation in brain weight, and

suggested that it may reflect individual differences in cortical organization and structure.

Both genetic and non-genetic factors appear to influence individual variability

(Oppenheim, Skerry, Tramo, & Gazzaniga, 1989).

The corpus callosum contains interhemispheric fibers which connect overlying

cortex in both hemispheres. The topography of callosal connections has been established

using a variety of approaches, including the study of Wallerian degeneration of the corpus

callosum following cortical lesions in humans (DeLacoste, Kirkpatrick, & Ross, 1985) and

in monkeys (Pandya, Karol, & Heilbronn, 1971; Seltzer & Pandya, 1983); studies of the

retrograde transport of labeled tracers (Pandya & Rosene, 1985); and electrical

stimulation of the callosum in humans during surgery (Schaltenbrand, Spuler, & Wahren,

1972; Tan et al., 1991). Tan et al. (1991) studied 30 people with intractable generalized









epilepsy during surgery prior to commissurotomy. The surface of the corpus callosum

was electrically stimulated in one centimeter increments while evoked potentials were

simultaneously recorded from the cortex. Stimulation of the most anterior portion of the

corpus callosum resulted in an evoked potential only in the frontal lobe. Slightly posterior

to that point, stimulation resulted in evoked potentials in frontal and temporal lobes, but

rarely in parietal or occipital lobes. Stimulation of posterior callosal regions resulted in

evoked potentials from parietal and occipital cortices (Tan et al., 1991).

Although uniform in appearance to the naked eye, the corpus callosum has been

divided into subregions using a variety of methods. Regions differ not only in terms of the

cortical regions they interconnect, but also in terms of their axonal composition (LaMantia

& Rakic, 1990). Witelson (1989) proposed an anatomically grounded system resulting in

seven callosal segments based on the cortical regions interconnected by commissural fibers

passing through each segment. She summarized the topography of callosal fibers as

established by the primate and human anatomical literature, indicating that the rostrum

contains fibers from the caudal and orbitoprefrontal and inferior premotor cortex. The

genu includes fibers from prefrontal cortex, with fibers from premotor and supplementary

motor cortex in the rostral body (Witelson, 1989). Considering the expansion of the

frontal pole in humans, Steere and Arnsten (1995) highlighted the possibility that the

rostral body also includes prefrontal fibers. Zaidel, Aboitiz, Clark, Kaiser, and Matteson

(1995) noted that, in humans, the anterior third of the corpus callosum might include

fibers from the anterior inferior parietal lobe. Witelson (1989) described the anterior

midbody as including fibers from motor cortex, with fibers from somatosensory and









posterior parietal cortex passing through the posterior midbody. DeLacoste, Kirkpatrick,

and Ross (1985) indicated that the midbody region may also include fibers from

midtemporal cortex. Witelson (1989) reported that the isthmus contains fibers from

superior temporal and posterior parietal cortex, with the splenium including fibers from

occipital and inferior temporal cortex. Highlighting differences between rhesus monkey

and human anatomy, DeLacoste et al. (1985) indicated that, in humans, the splenium and

possibly the isthmus, contains fibers from posterior association cortex at the temporal-

parietal-occipital junction.

Small diameter fibers in anterior and posterior corpus callosum interconnect

cortical association areas. A higher percentage of small diameter fibers are found in the

callosum of macaque than in mice, and in humans than in macaque (Aboitiz, Scheibel,

Fisher, & Zaidel, 1992). In humans, small diameter fibers are unevenly distributed in the

callosum, with the highest percentage found in anterior (rostrum and genu) and posterior

(isthmus and splenium) callosal regions (Aboitiz, Scheibel, Fisher, & Zaidel, 1992). Small

diameter fibers in the anterior region of the callosum primarily interconnect prefrontal

cortex, while small diameter fibers in the posterior region interconnect posterior

association cortex. Aboitiz et al. (1992) suggested that the increased proportion of small

diameter axons in anterior and posterior callosal regions reflects the relative growth of

association areas in the human brain, areas believed responsible for the most sophisticated

behaviors (Luria, 1980).

During gestation, the fibers of the corpus callosum develop in parallel with the

lobes of the cerebral cortex (Hewitt, 1962). The earliest callosal fibers found in embryos







5

of 10-11 weeks, with growth progressing along a rostral to caudal gradient (Gilles, 1983).

Transvaginal ultrasonography of the normal human fetus suggests that growth of the

corpus callosum occurs at a relatively constant ratio to the size of the developing brain

(Malinger & Zakut, 1993). The number ofcallosal axons in rhesus monkey increases

dramatically during fetal development, with the number of axons in newborn monkeys

exceeding the number present in the adult by at least three and a half times (LaMantia &

Rakic, 1990). About 70% of those callosal axons are eliminated in two phases after birth,

a rapid phase during the first three postnatal weeks in which 50 axons a second are lost,

and a slower phase extending over a three-month period in which 5 axons per second are

lost (LaMantia & Rakic, 1990). This same phenomenon has been described in other

animals, including the cat (Koppel & Innocenti, 1983). LaMantia and Rakic (1990) stated

that axons are eliminated because they either fail to establish synaptic contacts with target

neurons or because the cortical neurons giving rise to them are lost.

Innocenti (1986) suggested that mechanisms underlying the elimination of juvenile

projections may play a role in establishing the topography ofintercortical connections

found at maturity. However, Olavarria and Van Sluyters (1986) found that neonatal rats

show the same broad topographic pattern as adult rats, with axons from restricted regions

of cortex grouped in tight bundles within the corpus callosum. The location of these

bundles along the anterior to posterior dimension of the callosum depends upon the fibers'

region of origin in the cortex. Based on studies of axon overproduction and elimination in

the developing rhesus monkey, LaMantia and Rakic (1990) argued it is unlikely that

postnatal elimination of axons plays a significant role in determining the broad pattern of









callosal terminal fields or projection zones in primates, although it may play a role in fine-

tuning the pattern of connections. LaMantia and Rakic noted a temporal relationship

between the elimination of excess callosal axons and the process of synaptogenesis in

primate cortex. They suggested that the process of axon elimination may influence the

final number and pattern of cortical synaptic connections. In association, limbic, and

motor cortex of rhesus monkey cortico-cortico fibers passing through the callosum

innervate spatially segregated columns of cortex (Goldman & Nauta, 1977).

Witelson and Kigar (1988) proposed a model of callosal growth based on

published anatomic studies of fetal, infant, and children's brains. The model posits three

stages: (1) a fetal stage with rapidly accelerating growth; (2) a stage from birth to age two

with extensive growth (approximately 13 mm2 per month); and (3) a stage beginning

around age two or three marked by a slower rate of growth. Witelson and Kigar reported

that the callosum approaches minimal adult size by age two, and suggested that its growth

rate probably slows dramatically around that age, with continued slow growth throughout

the remainder of brain maturation. The subsequent work of Pujol et al. (1993)

demonstrating continued growth of the callosum into adulthood supports Witelson and

Kigar's model.









Abnormalities Involving the Corpus Callosum
Genesis of the corpus callosum

Anomalies in the course of embryologic development can result in abnormalities or

even frank absence of the corpus callosum. Complete or partial agenesis of the corpus

callosum is a relatively common malformation of the central nervous system, with an

incidence of one to three per 100,000 (Myrianthropoulos, 1974). It is a feature of at least

12 genetic syndromes, including Aicardi's syndrome, Andermann's syndrome, Shapiro's

syndrome, and Arocallosal syndrome (Aicardi, 1992). In recent years, genes and

individual chromosomes related to callosal agenesis have been identified (Casaubon et al.,

1996; Fransen et al., 1995; Genuardi, Calvieri, Tozzi, Coslovi, & Neri, 1994; Nyberg,

Karhu, Karikoski, & Simola, 1994). Agenesis of the corpus callosum has also been

associated with a number of metabolic diseases, including hyperglycemia (Kolodny, 1989)

and hypercalcemia (Johnson & Jones, 1985), and can result from intrauterine exposure to

toxins. Riley et al. (1995) studied 13 children with histories of significant prenatal alcohol

exposure and 12 normal control children. Of the 13 alcohol-exposed children, two had

agenesis of the corpus callosum. The remaining 11 alcohol-exposed children had a smaller

total callosal area than controls, with four of five subregions significantly reduced in size.

The difference in total callosal area was no longer significant when controlling for brain

size, although three of five callosal regions were significantly smaller in the alcohol

exposed children (Riley et al., 1995).

Callosal agenesis has traditionally been classified as either partial or total, although

that dichotomy may not adequately represent observed abnormalities of callosal







8

development. Jinkins, Whittemore, and Bradley (1989) retrospectively reviewed 15 cases

involving absence or dysgenesis of callosum, and identified three distinct categories:

agenesis, hypogenesis, and hypoplasia. They suggested an embryologic basis for the

distinction, with agenesis associated with disruption of neural tube closure, hypogenesis

associated with disruption of the formation and/or maintenance of the inductive plate of

the massa commissuralis, and hypoplasia associated with abnormal migration of neuronal

elements of the cerebral cortex responsible for the projection of commissural fibers

(Jinkins, Whittemore, & Bradley, 1989). Three anatomic abnormalities are considered

hallmarks of true callosal agenesis: presence of Probst's bundle, a large ipsilateral fiber

track running anterior to posterior; a distinct radial pattern of sulci found in the medial

portion of both cerebral hemispheres; and abnormally shaped lateral ventricles (Smith &

Rourke, 1995).

Callosal agenesis is frequently associated with mental retardation, particularly

when it occurs in the context of a genetic syndrome or in association with other central

nervous system anomalies. Lacey (1985) reported on a sample of 40 children described as

having partial agenesis of the corpus callosum, concluding that neurodevelopmental

prognosis is strongly related to the age at which the initial diagnosis of callosal agenesis is

made. When early neuroimaging was prompted by macrocephaly, multiple congenital

anomalies, or seizures, agenesis was strongly associated with mental retardation or IQs in

the Borderline range. Among children with callosal agenesis identified after age four, IQs

were typically in the Average to Low Average range (Lacey, 1985). In characterizing the

neurodevelopmental correlates of callosal agenesis it appears important to distinguish









between cases identified early in life in the context of other anomalies and clinical

symptoms, and cases in which callosal agenesis is a relatively isolated anomaly. Gupta and

Lilford (1995) reviewed 70 cases in which agenesis of the corpus callosum had been

identified in utero by ultrasonography. When agenesis was not associated with other

sonographically detectable abnormalities, 85% had a "normal" developmental outcome. In

cases in which multiple anomalies were detected the prognosis was poor. As

neuroimaging has become more widely used, the incidence of "asymptomatic" callosal

agenesis has risen, with absence of the corpus callosum sometimes identified on

neuroimaging done for unrelated clinical reasons (e.g., headache, head injury).

Smith and Rourke (1995) summarized neuropsychological findings in acallosal

subjects, noting limitations imposed by the small number of cases studies and by the fact of

mental retardation or low IQ in many cases complicating neuropsychological assessment.

Despite extensive investigations of cerebral lateralization in acallosal subjects with normal

or near-normal intelligence, relatively few have received comprehensive

neuropsychological evaluations. Smith and Rourke highlighted the need for such

neuropsychological evaluations, warning that the absence of findings indicative of

disconnection on neurological examination of should not be taken as evidence of a lack of

behavioral pathology.

Fine motor and tactile perceptual deficits have been noted with some consistency

in acallosals (Smith & Rourke, 1995), along with deficits in aspects of language and verbal

functioning. Language functioning has been extensively investigated in a small number of

subjects with callosal agenesis (Dennis, 1981; Jeeves & Temple, 1987; Temple, Jeeves, &







10

Vilarroya, 1989). A common finding has across cases has been difficulty rhyming (Smith

& Rourke, 1995), which Temple, Jeeves, and Vilarroya (1989) interpreted as either

reflecting a deficit of explicit phonological processing or a defect of auditory matching.

Temple and Ilsley (1993) demonstrated impaired auditory discrimination in three acallosal

children, and suggested that impaired phonemic discrimination may have a profound effect

on language development in acallosals. In reading, impaired phonological decoding has

been described in acallosals (Temple, Jeeves, & Vilarroya, 1990), along with defects of

syntactic and pragmatic comprehension (McCardle & Wilson, 1993; Sanders, 1989).


Sectioning of the corpus callosum

In addition to acallosals, people who have sustained sectioning of the corpus

callosum, either surgically or traumatically, have been studied in an effort to understand

the role the callosum plays in brain functioning. The largest and most carefully studied

group has been patients undergoing callosotomy as a palliative treatment for certain forms

of epilepsy (Bogen, 1995). Other groups have included patients with surgical division of

portions of the callosum as an approach to tumors of the lateral or third ventricles

(Bellotti, Pappada, Sani, Oliveri, & Stangalino, 1991; Misra, Rout, Padamadan, &

Radhakrishnan, 1993; Oepen et al., 1988; Petrucci, Buchheit, Woodruff, Karian, &

DeFilipp, 1987; Woiciechowsky, Vogel, Lehmann, & Staudt, 1995), or in the process of

removing arteriovenous malformations or treating aneurysms (Dickey, Bloomgarden,

Arkins, & Spencer, 1992; Guidetti & Spallone, 1982; Kosary, Braham, Shacked, &

Kronenberg, 1978). Traumatic sectioning of the corpus callosum has also been reported









as an effect of head injury or cerebrovascular hemorrhage (Cukiert, Haddad, Mussi, &

Marino Junior, 1992; Gentry, Thompson, & Godersky, 1988; Komatsu, Sato, Kagawa,

Mori, & Namiki, 1979; Opeskin, 1995; Rubens, Geschwind, Mahowald, & Mastri, 1977;

Senegor, 1991; Vuilleumier & Assal, 1995).

Originally described in the 1930s, division of the corpus callosum is recognized as

a viable treatment for people with medically refractory epilepsy not amenable to surgical

resection (Williamson, 1995). In well selected cases, callosotomy offers the prospect of a

substantial reduction in seizure frequency with relatively few lasting neurologic or

neuropsychologic deficits. Early commissurotomies involved sectioning the entire

callosum along with other commissural pathways (Bogen, 1995). However, evidence that

partial sectioning sparing posterior callosum is effective in controlling certain seizure types

with less potential for neurobehavioral consequences than complete sectioning has lead to

the current practice in most centers of staged callosotomy (Roberts, Reeves, & Nordgren,

1995). The anterior half to two-thirds of the callosum is sectioned in an initial surgery,

and the section is completed in a second operation in patients with an unsatisfactory

seizure response.

Bogen (1993) outlined the acute effects of complete commissurotomy, including

left-arm hypotonia, abnormal grasping with the left hand, bilateral Babinski response,

mutism and akinesia, particularly acute on the left side in response to verbal commands.

The extent to which the symptoms reflect acute effects of interhemispheric disconnection

versus effects related to the trauma of surgery (e.g., traction injury, edema) remains

unclear, although the disability is usually temporary (Reeves & Risse, 1995). After









resolution of the acute syndrome, there is a striking absence of frank neurologic or

neuropsychologic consequences in most cases (Reeves & Risse, 1995). In everyday social

situations, or on cursory examination, subjects appear to function normally. However,

under laboratory conditions with lateralized presentation of stimuli, features of the "split

brain syndrome" can be elicited (Bogen, 1993). These include a lack of interhemispheric

transfer and effects related to hemispheric specialization in humans (e.g., inability of right-

handers to name objects placed in the left hand). Adults seem to progressively acquire

compensatory strategies circumventing deficits (e.g., cross cuing) (Bogen, 1993). There is

evidence of substantial functional reorganization in children, with absence the

disconnection syndrome when callosotomy is performed before puberty (Lassonde,

Sauerwein, Geoffroy, & Mercier, 1995).

Although most studies have reported that callosotomy is not associated with frank

neuropsychological deficits (Ferrell, Culver, & Tucker, 1983; Mamelak, Barbaro, Walker,

& Laxer, 1993; Oepen et al., 1988), Williamson (1995) and Reeves and Risse (1995)

outlined a growing body of evidence that, in some cases, commissurotomy causes lasting

deficits resulting in significant handicaps. Rayport, Ferguson, and Corrie (1983) reported

that complete sectioning of the corpus callosum may result in prolonged behavioral

disturbances in the areas of language, hemispheric competition, and attention/memory

significant enough to impact daily living. Sass et al. (1988) reviewed data from 18

patients tested before and after partial or total callosotomy. They found that people with

crossed language dominance on Wada testing were at increased risk of developing lasting

post-operative language deficits regardless of the extent of surgery.









Zaidel and Sperry (1974) reported significant memory impairments after

commissurotomy, although other factors including extra-callosal brain lesions, seizures,

and antiepileptic drug burden may have contributed to their findings (LeDoux, Risse,

Springer, Wilson, & Gazzaniga, 1977). In a review, Clark and Geffen (1989) failed to find

evidence of memory deficits in the absence of extra-callosal damage. Ferguson, Rayport,

and Corrie (1985) reported other persisting deficits after callosotomy, including

interhemispheric antagonism resulting in significant disability. With the advent of MRI it

became possible to verify the extent of callosal sections and identify preserved

interhemispheric connections, making it possible to study the neuropsychological

correlates of particular spared callosal pathways (Gazzaniga, Holtzman, Deck, & Lee,

1985; Gazzaniga, Kutas, Van Petten, & Fendrich, 1989; Tramo et al., 1995).



Influences on Development and Morphology of the Corpus Callosum

Effects of early events and experiences on callosal development

Juraksa and Kopcik (1988) reported that environmental experiences can have an

effect on callosal morphology in rats. They studied male and female rats raised in an

"enriched" environment including group housing, regularly changed play toys, and a 30-

minute "outing" each day in an open field. Matched litter mates were housed individually

in standard stainless steel laboratory cages. Rats raised in the "enriched" environment had

a larger cross-sectional area of the corpus callosum than rats raised in standard cages. Sex

differences were also reported, with females having more unmyelinated axons regardless

of environment, and females from the enriched environment having more myelinated axons







14

than comparably housed males. Males in the "enriched" environment demonstrated larger

myelinated axons than females raised in the same environment (Juraska & Kopcik, 1988).

There is recent evidence that environmental experience may influence callosal

morphology in humans. Schlaug, Jancke, Huang, Staiger, and Steinmetz (1995) reported

that midsagittal area of the anterior half of the corpus callosum is larger in musicians who

began their musical training before age seven than in musicians who began training after

age seven or normal controls. Schlaug et al. concluded that the early, intensive, bimanual

training required of musicians plays a role in establishing the size and fiber composition of

the corpus callosum. They interpreted the larger anterior area of the corpus callosum as a

morphometric substrate of increased interhemispheric communication. It remains

possible, however, that larger anterior callosal area is a preexisting feature of children

identified as musically talented and selected for early training.

Miller et al. (1993) also sought to study the impact of neonatal events on brain

development. Ischemia was induced in two week old kittens by bilateral surgical

occlusion of the carotid arteries. At one and two months after surgery, MRI imaging was

used to assess global brain structures. At three months post-surgery the animals were

killed and their brains examined histologically, including examination of the distribution of

visual callosal projections labeled with a retrograde tracer. The major findings of the

study concerned the gross morphology of the corpus callosum and number of surviving

callosal projections. The callosum was atrophied throughout its extent in ischemic

animals, yet the number of callosal axons interconnecting contralateral areas 17 and 18

was much higher than normal. Although apparently paradoxical, Miller et al. suggested







15

that ischemia-related dysmyelination resulted in callosal atrophy which disguised an excess

number of surviving callosal axons. Ventricular dilatation of varying degrees was also

demonstrated by MRI, and histologic examination at three months showed lower neuron

density in motor but not in occipital cortex of animals with a history of ischemia. Miller et

al. claimed their study as the first to show that connectivity of the developing brain can be

altered by ischemic insult, and speculated that neonatal ischemia may interfere with the

process of eliminating excess callosal axons.

Much of the literature concerning the effects of postnatal experience on callosal

development has involved studies of the visual system. The effects of dark rearing (Frost

& Moy, 1989), visual deprivation versus enucleation (Innocenti & Frost, 1980), and

comparisons among normal-eyed, congenitally anopthalamic, and neonatally enucleated

mice (Olavarria & Van Sluyters, 1984) have highlighted the importance of stimulation

after birth in refining and stabilizing the normal complement of callosal projections.

Based on that literature, Grigonis and Murphy (1994) started from the premise that

appropriate projections are selected for survival by correlated pre- and post-synaptic

activity, and that inappropriate projections are eliminated when that correlated stimulation

is absent. Grigonis and Murphy suggested that if epileptic activity was induced in one

cortex, the resulting hypersynchronous discharge of cells with axons through the callosum

would summate and activate their post-synaptic targets in the contralateral cortex. This

would have the effect of mimicking normal correlated activity resulting in the sparing of an

inappropriate projection which otherwise would have been eliminated. Grigonis and

Murphy described a two-millimeter band at the border of Broadman areas 17 and 18 in the









cortex of adult rabbits where interhemispheric fibers passing through the callosum

terminate. This band is reportedly twice as large in newborn rabbits, but is "fine tuned"

and attains its adult dimension through the loss of excess callosal projections in the first

two weeks after birth. Grigonis and Murphy hypothesized that induced epileptic activity

in one cortex would result in the stabilization of immature callosal projections allowing

their survival into adulthood. This would have the effect of maintaining the immature,

enlarged projection zone at the border of areas 17 and 18.

To test this hypothesis, Grigonis and Murphy (1994) implanted penicillin-

containing beads on the dorsal surface of visual cortex unilaterally in neonatal rabbits.

Sustained release of penicillin from the beads resulted in epileptic activity throughout the

cortex. This was verified in some animals with electroencephalographic recordings.

Approximately one month post-implantation, a retrograde tracer was injected in some

animals, and all the animals were subsequently killed and their brains examined.

Grigonis and Murphy (1994) found that the width of the callosal projection zone at

the border of areas 17 and 18 was two times normal size in the animals in which a

penicillin bead had been implanted. This finding supported Grigonis and Murphy's

hypothesis that epileptic activity interferes with the normal process selecting callosal axons

for survival into adulthood. The authors acknowledged the possibility that their results

could be due to delayed elimination rather than a stabilization of excess callosal

projections, but believed this an unlikely explanation since the axons were extant long after

the period in which exuberant axons are typically eliminated. Grigonis and Murphy

concluded that the cognitive deficits observed in children with epilepsy may be attributable







17

to a widespread failure in the refinement of callosal projections. Pruning of callosal axons

is temporally, and may be causally, related to cortical synaptic synaptogenesis (LaMantia

& Rakic, 1990). As Keshavan, Sanders, Pettegrew, and Dombrowsky (1993) suggested,

failure of callosal pruning may interfere with the development of cortical neurons altering

neurodevelopment and eventual neurobehavioral functioning.


Effect of gender on callosal morphology

Sex differences in callosal morphology have been reported, but the precise nature

of the differences remains controversial. Differences in the nature of the study (e.g.,

postmortem versus MRI) and dependent measures (e.g., total callosum versus some

number of regions; absolute area of callosum versus measures adjusted in some way for

brain size) have made it difficult to compare results and attempt to draw conclusions

across studies. Driesen and Raz (in press) reported results of a meta-analysis including 36

studies examining sex differences in the size of the corpus callosum. Their results

indicated that men tend to have larger absolute callosal areas, with the median corpus

callosum area for men falling at approximately the 62nd percentile for women. However,

the effect size was relatively small, with only 21% of the combined distribution of corpus

callosum areas for men and women non-overlapping (Driesen & Raz, in press). The mean

effect size did not differ between postmortem and MRI studies, and no effect was found

for age or the percentage of right-handed subjects included in the samples studied.

In contrast, across 11 studies in which corpus callosum measurements were

adjusted for brain size, women's callosal areas were found to be larger than men's (Driesen









& Raz, in press). Relative to intracranial volume, the corpus callosum was also found to

be significantly larger in females than males in a large scale MRI study of 200 healthy

normal volunteers (Johnson, Farnworth, Pinkston, Bigler, & Blatter, 1994). In a

methodological critique, Amrndt, Cohen, Alliger, Swayze and Andreasen (1991) criticized

the use of such ratio and proportion measures, like those used by Johnson et al. (1994)

and in the studies reviewed by Driesen and Raz (in press), suggesting that they are less

reliable than absolute volume measures. In situations in which it is meaningful to control

for brain size, Arndt et al. recommended using multivariate statistical methods rather than

ratio or proportion measures.

DeLacoste-Utamsing and Holloway (1982) reported that splenium of female's

corpus callosum is larger and more bulbous than male's, and suggested that the difference

might be related to sex differences in the degree of cerebral laterality. However, Driesen

and Raz (in press) found the effect size in DeLacoste-Utamsing and Holloway's study to

be an outlier in a meta-analysis of 21 studies comparing area of the splenium in men and

women. Excluding DeLacoste-Utamsing and Holloway's report, they found no significant

sex difference in size of the splenium (Driesen & Raz, in press).


Effects of handedness and age on callosal morphology

Differences in morphology of the corpus callosum have also been linked to hand

preference. Research in this area is complicated by many of the same factors which

complicate interpretation of studies examining sex differences in callosal morphology.

Witelson (1985), Witelson and Goldsmith (1991), and Habib et al. (1991) have reported









that the callosum is larger in non-right-handed than in right-handed people. However,

other investigators have failed to find differences related to hand preference (Kertesz,

Polk, Howell, & Black, 1987; O'Kusky et al., 1988; Steinmetz et al., 1992; Yoshii et al.,

1986). Based on a meta-analysis of seven studies which examined the effects of

handedness on corpus callosum area, Driesen and Raz (in press) concluded that the corpus

callosum is larger in non-right handers than in right-handers. Cowell, Kertesz, and

Dennenberg (1993) reported that consistency of hand preference, and not direction of

hand preference alone, is a significant factor in understanding differences in callosal

morphology.

Witelson (1989), Denenberg, Kertesz, and Cowell (1991), and Burke and Yeo

(1994) have reported significant interactions between sex and handedness. Age-related

changes have also been described. Weiss, Kimbacher, Wenger, and Neuhold (1993)

reported that the genu and anterior body of the corpus callosum decreases in size as a

function of normal aging. Sex, handedness, and age-related differences in callosal

morphology have all been reported, although the existence, nature, and relationship among

their effects on callosal morphology remains controversial.



Relationships between Callosal Morphology and Cognitive Functioning

Q10 and other global measures of cognitive performance

The relationship between callosal morphology and measures of cognitive

functioning has been studied in a number of pathologic conditions, including diseases

which primarily affect white matter (e.g., multiple sclerosis) or gray matter (e.g.,









Alzheimer's disease). Barnard and Triggs (1974) observed that mental deterioration and

atrophy of the corpus callosum are common findings in people with multiple sclerosis, but

noted that little effort had been made to link callosal atrophy with clinical features of the

disorder. Barnard and Triggs studied at autopsy the brains of 20 people who had multiple

sclerosis. Marked mental deterioration was a common feature in the clinical histories of

cases with severe atrophy of the callosum (Barnard & Triggs, 1974). The authors

acknowledged limitations imposed on their conclusions by the lack of formal testing of

intellectual performance or callosal functioning.

Brodensteiner, Schaefer, Breeding, and Cowan (1994) studied 445 children less

than 17 years of age who presented consecutively for MRI scans for a variety of clinical

reasons. Fifty children were identified as having a small corpus callosum based on visual

inspection of the scans. Seven cases ofhypoplasia of the corpus callosum, defined as

having a callosum more than two standard deviations smaller than age-appropriate

normative values, were identified. Seventy-one percent of the children with a hypoplastic

corpus callosum versus 29% of a control sample were found to have an "impaired

functional level" as inferred from medical records. An impaired functional level was

defined as failing in school, demonstrating mild-to-moderate developmental delay or

mental retardation.

With the development of neuroimaging, it became possible to study the

relationship in living subjects between callosal morphometry and performance on

psychometric and neurobehavioral tests. Aboitiz et al. (1992) noted that callosal area

provides a good estimate of the number if small diameter fibers passing through the







21

callosum, although the relationship between callosal area and large diameter fibers is less

clear. Coffinan, Bornstein, Olson, Schwarzkopf, and Nasrallah (1990) explored the

relationship between callosal morphology and neuropsychological test performance in 30

outpatients diagnosed with bipolar affective disorder and a group of 52 controls. Both

groups had comparable IQ levels, although the bipolar group had, on average, a smaller

corpus callosum. Across groups, area of the corpus callosum had a significant positive

correlation with neuropsychological summary indices of left-hemisphere, right-hemisphere,

and generalized cognitive functioning.

Huber et al. (1987) examined MRI correlates of dementia in a sample of 32

patients diagnosed clinically with multiple sclerosis. Comparisons were made to the

performance of 12 normal controls, many of whom were spouses of the patients. A 30-

minute neuropsychological battery was administered, including mental status examination

and tests of language, memory, apraxia, and visuospatial ability. "Dementia" was defined

as impairment two standard deviations or more below normal controls' performance on at

least three measures, including the Zung Depression inventory administered as a measure

of personality change sometimes accompanying dementia. Huber et al. (1987) found that

atrophy of the corpus callosum, rated on a scale from one to five, was more extreme in

"demented" than non-demented patients.

Huber, Bornstein, Rammohan, and Christy (1992) subsequently studied 35 people

with clinically probable or definitive multiple sclerosis. Subjects received MRI imaging

and neuropsychological testing, including tests of executive functioning, memory,

interhemispheric transfer, and psychomotor speed. Summary scores of







22

neuropsychological test performance were calculated and patients classified as "normal to

mildly impaired," "moderately impaired," or "severely impaired." Total lesion area in the

left- and right-hemisphere, surface area of the cortical lobes, and midsagittal surface area

of the corpus callosum were measured. Huber et al. (1992) reported that total lesion area,

regardless its distribution, correlated with most neuropsychological measures. Patients

with "severe impairment" demonstrated a greater total lesion area and had significantly

smaller callosal areas than other subjects. Reduced area of the corpus callosum was found

to be highly correlated with slower performance on tests involving speed of information

processing (e.g., Trails B, Tactual Performance Test) and poorer performance on tasks

requiring interhemispheric transfer of information (e.g., Tactile Forms). Huber et al.

(1992) interpreted the findings as evidence that atrophy of the corpus callosum reflects

global disease processes and provides a focal morphologic marker of severe cognitive

impairment in multiple sclerosis.

Rao, Leo, Haughton, St. Aubin-Faubert, and Bernadin (1989) studied the

relationship between neuropsychological test performance and MRI findings in 53 people

with multiple sclerosis. The neuropsychological battery included measures of verbal

intelligence, memory, abstract reasoning, attention/concentration, language, and

visuospatial skills. Three morphometric variables were derived from the MRI images:

total lesion area, ventricle to brain ratio, and midsagittal area of the corpus callosum.

Using stepwise multiple regression, they found area of the callosum to be a significant

predictor beyond age and education of performance on measures of information

processing speed and rapid problem solving. Total lesion area was a robust predictor of









cognitive dysfunction, while ventricle to brain ratio did not independently predict any

cognitive test finding. Based on the relationship between callosal area and test

performance, Rao et al. suggested that specific cognitive processes may be disrupted by

demyelinating lesions involving relatively focal morphological structures.

Fletcher et al. (1992) reported a significant relationship between size of the corpus

callosum measured from MRI and cognitive skills in 35 children with meningomyelocele,

meningocele, or aqueductal stenosis and a control group of 12 normal children. Total area

of the corpus callosum was measured, and a verbal and nonverbal composite of

neuropsychological measures was constructed. Both verbal and nonverbal measures

correlated positively with area of the corpus callosum, with a higher correlation reported

for nonverbal measures. Pathology outside the corpus callosum was observed in the

patient group, including hydrocephalus-related changes in the lateral ventricles and

cerebral white matter tracts.

In a study of 39 healthy, right-handed subjects Schultz and Staib (1995) reported a

positive correlation between the size of the splenium and Full Scale and Verbal Scale IQs,

but the direction of the relationship was mediated by sex. For females, a larger area of

the splenium was correlated with a higher FSIQ (r = .50) and higher VIQ (r = .55). For

males the direction of the relationship was reversed, with a larger area of the splenium

correlated with a lower FSIQ (r = -.62) and lower VIQ (r = -.66).

In a significant contribution, Yamauchi, Fukuyama, Harada, and Nabatame (1993)

demonstrated parallels between callosal area measured from MRI, cortical oxygen

metabolism, and intellectual impairment in people with Alzheimer's disease. Ten right-









handed male patients with early Alzheimer's disease and 14 age- and sex-matched right-

handed control subjects were tested. Using a midsagittal MRI image, the callosum was

divided into four segments from front to back, and "resting" blood flow was measured

with positron emission tomography. The Wechsler Adult Intelligence Scale was used to

estimate IQ. Yamauchi et al. (1993) reported that, relative to normal controls,

Alzheimer's patients had significantly smaller callosi, particularly in the posterior region.

Total area of the corpus callosum was significantly correlated with Full and Verbal Scale

IQs, but not Performance Scale IQ. Area of the anterior half of the callosum correlated

significantly with oxygen metabolism in frontal cortex. Area of the posterior half of the

callosum correlated significantly with oxygen metabolism in posterior association cortex.

In a second study Yamauchi, Fukuyama, Ogawa, Ouchi, and Kimura (1994) demonstrated

a positive correlation between callosal area and IQ in 11 right-handed male subjects with

lacunar infarcts and evidence of diffuse white matter abnormalities. In eight of the 11

patients with dementia, area of the corpus callosum was significantly correlated with the

mean level of oxygen metabolism in cerebral white matter. Yamauchi et al. (1994)

suggested that callosal atrophy may reflect the extent of white matter damage which plays

a role in determining the severity of dementia.


Other neuropsychological measures

Pozzili, Bastianello, Padovani, and Passafiume (1991) reported that anterior

callosal area was strongly correlated with performance on a measure of verbal fluency,

even when the effect of total area ofdemyelinating lesions was controlled for, in a group









of 18 women with multiple sclerosis. Other cognitive measures did not show a specific

relationship with anterior or posterior callosal atrophy. Pozzilli et al. (1991) suggested

that the cognitive mechanisms underlying performance on verbal fluency depend upon

interhemispheric transfer of information accomplished through anterior portions of the

corpus callosum.

Hines, Chiu, McAdams, and Bentler (1992) examined the relationship between the

midsagittal surface area of anterior and posterior segments of the corpus callosum and

measures of verbal fluency, visuospatial ability, and language lateralization. They studied

28 healthy women and found that verbal fluency correlated significantly with area of the

posterior callosum, particularly the region of the splenium. No other consistent

relationships were found between callosal morphology and test performance. The

correlation of verbal fluency and posterior callosal area appears at odds with the

relationship between anterior corpus callosum and verbal fluency reported by Pozzilli,

Bastianello, Padovani and Passafiume (1991). This discrepancy may be accounted for by

differences in the "verbal fluency" measure studied. Pozzilli et al. (1991) used a word

fluency task in which subjects were asked to generate words to target letters of the

alphabet. Hines et al. (1992) constructed a verbal fluency index representing performance

on a semantically guided word generation task, a sentence generation task, and word

fluency to a target letter of the alphabet. There were also differences in the populations

studied. Hines et al. tested healthy females while Pozzilli et al. (1991) tested women with

multiple sclerosis.









Wang, Doherty, Hesselink, and Bellugi (1992) described relationships among

callosal morphology, neurobehavioral functioning, and neuropathology in children with

Down and Williams syndromes. The callosi of children with Down syndrome were found

to be rounded with a smaller anterior region than in children with Williams syndrome. The

authors noted a convergence between this finding and previously reported hypocellularity

of frontal lobes in Down syndrome and neuropsychological profiles including prominent

frontal lobe dysfunction. They also cite unpublished findings showing that, relative to age-

and IQ-matched subjects with Williams syndrome, children with Down syndrome

demonstrate deficits including poor verbal memory, a tendency toward perseveration on a

test of verbal memory, and greater difficulty on a task requiring flexible problem solving

(Wang, Doherty, Hesselink, & Bellugi, 1992).


Attention-deficit/hyperactivity disorder

Hynd et al. (1991) reported a relationship between callosal morphology measured

from MRI images and behavioral ratings of children with attention-deficit/hyperactivity

disorder (ADHD) and normal controls. They compared seven children with a primary

diagnosis of ADHD to a sample often nondisabled controls. Although all MRI scans

were clinically read as normal, ADHD children had smaller corpus callosi, particularly in

the anterior (genu) and posterior (isthmus and splenium) regions compared to controls

(Hynd, Semrud-Clikeman, Lorys, & Novey, 1991). The authors interpreted their findings

with reference to interhemispheric fibers which pass through anterior and posterior

portions of the callosum interconnecting frontal and posterior association cortex.









Following Hynd et al. (1991), Geidd et al. (1994) examined the relationship

between callosal morphology and ADHD. They measured the area of the corpus callosum

on midsagittal MRI sections in 18 boys with ADHD and 18 age-, height-, handedness- and

Tanner stage-matched controls. They found that two anterior regions of the callosum, the

rostrum and the rostral body, were significantly smaller in the ADHD group. The size of

these regions also correlated significantly with teacher and parent ratings of hyperactivity/

impulsivity, with smaller anterior callosal areas associated with poorer behavioral ratings

(Giedd et al., 1994). Geidd et al. interpreted the relationship between area of the rostral

body of the corpus callosum and hyperactivity/impulsivity in terms of premotor and

supplementary motor commissural connections passing through that callosal region.

However, citing anatomical work done in primates, Steere and Amsten (1995) noted that

the rostral body of the corpus callosum is likely to also contain fibers from the anterior

cingulate and prefrontal cortices, regions they suggested are likely involved with

attentional regulation.

Semrud-Clikeman et al. (1994) reported significantly smaller posterior but not

anterior callosal area in a sample of 15 right-handed males diagnosed with ADHD in

comparison to age and IQ-matched controls. Citing methodological differences in subject

selection, imaging, and mensuration technique, Semrud-Clikeman et al. attempted to

reconcile their failure to find differences in anterior callosal regions with the previously

reported findings. They suggested that posterior cortex may be part of a functional system

involved with sustained attention, and that smaller area of posterior callosum reflects









fewer callosal connections in that region of brain negatively affecting children's ability to

sustain attention.

It is important to note that many attempts to correlate morphologic features of the

corpus callosum with behavior have not found a significant relationship. For example,

Johnson, Bigler, Burr, and Blatter (1994) failed to find a significant relationship between

midsagittal surface area of the corpus callosum and IQ among brain injury survivors.

Although the callosum was significantly atrophied, callosal area was not significantly

related to any of the IQ variables studied. The degree of atrophy was related to trauma

severity and the extent of ventricular enlargement. They described the lack of a systematic

relationship between area of the corpus callosum and performance on intellectual testing

as an "enigma." The failure may be related to the use of total area as the sole measure of

callosal morphology.



Callosal Morphology and Intellectual Functioning

Strauss, Wada, and Hunter (1994) studied the relationship between midsagittal

surface area of the callosum measured from MRI and performance on intelligence testing.

They tested 47 patients with medically refractory seizures undergoing evaluation as

candidates for neurosurgical treatment. Intelligence was assessed with the WAIS-R,

although Full Scale IQ scores were prorated from two administered subtests (Vocabulary

and Block Design) in approximately 28% of the subjects. The IQs ranged from 40 to 114,

with a mean of 86 and a standard deviation of 14. The callosum was measured from the

tip of the genu to the end of the splenium and divided into five equal segments. The area







29

of each segment, and aggregate total callosal area, was measured from a midsagittal MRI

image. Strauss et al. (1994) found that the area of the fifth segment, corresponding to the

region of the splenium, had a significant correlation with Full Scale IQ. The partial

correlation between area of the splenium and Full Scale IQ remained significant when

controlling for cerebral speech pattern and attempting to control for language

lateralization.

Following the report by Strauss et al. (1994) of a positive correlation between area

of the splenium and Full Scale IQ, the current study examined the relationship between

area of subregions of the corpus callosum and performance on measures of intellectual

functioning. Aside from the significance of potential replication, the proposed study

improves upon the work of Strauss et al. (1994) in several important ways.

The sample studied by Strauss et al. (1994) was identified as patients with

medically refractory epilepsy. This presumably included people with temporal or frontal

foci, multiple foci, as well as patients with partial complex or primary generalized epilepsy.

They acknowledged the limitation their heterogeneous sample imposed on interpretation

of their results, and called for replication with a more homogeneous group of patients with

well-defined focal lesions (Strauss, Wada, & Hunter, 1994). In addition to differences in

epileptogenic focus, Strauss et al. (1994) included subjects differing on handedness and

language laterality. Speech dominance and handedness have both been identified as

factors potentially affecting callosal morphology. Driesen and Raz (in press) concluded

that the callosum is larger in non-right handers, and O'Kusky et al. (1988) found that







30

people with right hemisphere speech dominance have a larger corpus callosum than people

with left hemisphere language dominance.

The current study included only right-handed subjects with left-hemisphere speech

dominance and a unilateral anterior temporal lobe epileptogenic focus. Limiting the study

to people with an epileptic focus in anterior temporal lobe lessened the possibility that

subject's epileptogenic lesion exerted a direct effect on callosum. The region of the

temporal pole sends interhemispheric fibers primarily through the anterior commissure,

and not through the corpus callosum (Pandya & Rosene, 1985).

Strauss et al. (1994) were also limited by the intelligence test data available to

them. While the majority of their subjects received the full Wechsler intelligence test

appropriate for their age, approximately 28% were given only the Vocabulary and Block

Design subtests. Full Scale IQs were estimated based on their performance on these

subtests. This is problematic because estimating IQ from scores on the Vocabulary and

Block Design subtests may systematically overestimate IQ (Thompson, Howard, &

Anderson, 1986). Evidence from research with other short forms of the WAIS-R suggests

that short forms may also lead to a biased estimate of IQ based on lesion laterality

(Massad, Bobbitt, Kelly, & Beasley, 1988). Because they are based on fewer administered

items, prorated scores are also a less reliable estimate of the underlying construct

(Nunnally, 1978), weakening observed correlations relative to the strength of the

underlying association. Strauss et al. (1994) reported a range of scores including a FSIQ

of 40. That extreme low score presumably reflects an artifact of estimating FSIQ scores

because an IQ score of 40 is not valid on the WAIS-R (Wechsler, 1981, p. 89). The







31

existence of extreme values extending the range of scores may have artificially inflated the

correlation coefficient reported by Strauss et al. (1994). The current study used data from

the full WAIS-R, avoiding problems associated with estimated Full Scale IQ scores, and

making it possible to explore the relationship between area of the callosum and

performance on the WAIS-R Verbal and Performance Scales.

Bomrnstein, Drake, and Pakalnis (1988) conducted a factor analysis of the WAIS-R

in a sample of 107 patients with a variety of seizure disorders, the majority having

intractable epilepsy. Using oblique rotation, the results demonstrated a three-factor

solution with Verbal/Comprehension, Perceptual/Organizational, and Freedom from

Distractibility factors extracted. Together these factors accounted for 73% of the total

variance in FSIQ. The most prominent difference between the factor structure of the

WAIS-R in the standardization sample and epileptics was the greater prominence of the

Freedom from Distractibility factor in the epileptic group. Bornstein et al. (1988) reported

that the Freedom from Distractibility factor accounted for more variance than the

Perceptual/Organization factor. Leckliter, Matzrazzo, and Silverstein (1986) reviewed

factor analytic studies of the WAIS-R. Most of the studies were done with data from the

original normative sample, but some included patient groups. They concluded that the

Freedom from Distractibility factor may be more salient in clinical populations, and

suggested that a three-factor solution provides a richer set of hypotheses in patient groups

than interpretations based on a two-factor solution.

The impact of early brain damage on subsequent cognitive and behavioral

development has been studied in a range of conditions, including people with epilepsy.







32

Dikeman, Matthews, and Harley (1975) reported a study of the differential effect of early

versus late onset of major motor epilepsy on IQ and neuropsychological test performance.

People with an early onset of major motor epilepsy were found to have significantly more

impairment on psychometric testing as adults than people with later onset of major motor

seizures. People with an early onset of seizures also demonstrated significantly greater

impairment than comparison groups with early and late onset of brain damage without

seizures.

Saykin, Gur, Sussman, O'Connor, and Gur (1989) examined the effects of age at

the onset of seizures and laterality of seizure focus on memory deficits after temporal

lobectomy. They found that seizure onset before five years of age was associated with

lower IQ and poorer memory scores. Regardless of the age at onset, right temporal

lobectomy patients demonstrated improved semantic and figural memory after surgery,

while left temporal lobectomy patients showed a worsening of semantic memory.

Unexpectedly, the early onset left hemisphere patients also showed a marked decline of

figural memory after surgery. Saykin et al. (1989) discussed their findings in terms of the

impact early brain damage has on cortical development and hemispheric specialization.

Stafiniak et al. (1988) reported that age at first risk for seizures was a significant

predictor of anomia in the acute period following dominant anterior temporal lobectomy.

Early risk for seizures was defined as including a history of febrile convulsions, tumors,

perinatal distress, encephalitis, meningitis or head trauma. Although there was no

difference between the early and late onset groups on confrontation naming performance

prior to surgery, significant differences emerged in the period immediately after surgery.









None of the patient's with an early risk for seizures showed anomia, defined as

performance more than 25% below pre-surgical baseline. In contrast, 60% of patients

without early risk for seizures demonstrated post-surgical anomia. Stafiniak et al.

interpreted this finding as evidence that the cerebral representation of naming may be

atypical in patients with a history including early risk factors for seizures.

O'Leary et al. (1983) studied the effect of age at onset of partial or generalized

seizures on intellectual and neuropsychological test performance in children. Children

with seizures beginning before five years of age performed significantly worse on

measures of Verbal and Performance IQ and on tests of attention and mental flexibility

than children with later seizure onset. Similar patterns of performance were found for

children with partial and generalized seizures. The authors interpreted their findings in the

context of reports that brain damage can result in widespread and permanent changes in

brain structures if the damage occurs early in the course of development. Early seizure

activity has been shown to alter the normal process of pruning excess callosal axons

(Grigonis & Murphy, 1994), a process which appears to be related to cortical

synaptogenesis (LaMantia & Rakic, 1990). It has been speculated that a failure of callosal

pruning alters the course of neurodevelopment (Keshavan, Sanders, Pettegrew, &

Dombrowsky, 1993), and may play a role in the cognitive and neurobehavioral deficits

observed in children with early onset of epilepsy (Grigonis & Murphy, 1994).









Study Rationale and Predictions


Initial Study Objectives

The study's first goal was to determine if the significant correlation between

splenial area and FSIQ reported by Strauss et al. (1994) was replicable. Positive

correlations between VIQ, PIQ, and FSIQ and area of the callosal regions were predicted,

with the strongest correlation expected with area of the splenium.

Following the report of a prominent freedom from distractibility factor in the

WAIS-R data of subjects with epilepsy (Bornstein, Drake, & Pakalnis, 1988), and reports

of significant correlation between ADHD symptomatology and anterior (Hynd et al.,

1991) and posterior callosal areas (Semrud-Clikeman et al., 1994), the study planned to

examine the correlation between freedom from distractibility factor scores and callosal

morphology. Based on findings in children with ADHD, significant positive correlations

were predicted with anterior and posterior callosal areas.

Because early seizures interfere with the process of pruning callosal projections, it

was predicted that the relationship between callosal area and WAIS-R IQ scores and

Freedom from Distractibility Index score would differ for people with seizures beginning

before age five and people with later seizure onset. Early epileptic activity may result in

the survival of an excess number of callosal projections (Grigonis & Murphy, 1994). The

excess number oftranscallosal fibers may result in larger cross-sectional areas, particularly

in the anterior and posterior regions where small diameter fibers predominate (Aboitiz,







35

Scheibel, Fisher, & Zaidel, 1992), although decreased area can disguise an excess number

of axons.



Anterior Callosal Morphology and Wisconsin Card Sorting Test Performance

The Wisconsin Card Sorting Test (WCST) has traditionally been among the most

commonly used measures for assessing frontal lobe functioning (Spreen & Strauss, 1991).

Introduced by Berg (1948), and originally formalized by Grant and Berg (1948), the test

requires the subject to sort a deck of response cards according to various stimulus

dimensions. The subject is not informed of the sorting principle, which changes in the

course of testing and must be inferred on the basis of feedback from the examiner

(Damasio & Anderson, 1993). Heaton (1981) and Heaton, Chelune, Talley, Kay, and

Curtiss (1993) formalized administration and scoring criteria, and reported normative,

reliability, and validity data for a version of the WCST.

Milner (1963) used a form of the WCST with people who had unilateral cortical

excisions for the relief of focal epilepsy. She concluded that impaired performance on the

WCST was related to damage of the frontal lobes, and identified the fundamental deficit as

a failure to shift from one category to another due to perseverative interference from prior

response sets. The finding was specific to people with dorsolateral prefrontal lesions;

people with orbitofrontal lesions did not show the same impairment. Milner's total sample

of 94 subjects was divided into two groups: 71 people tested before and approximately

18-days after surgery; and 23 people tested only postoperatively, from two weeks to 15-

years after surgery. The pattern of results for both groups was the same. The role of the







36

frontal lobes, particularly the left frontal lobe, in WCST performance was also highlighted

by Drewe (1974) in a study of 91 people with either unilateral frontal or non-frontal

lesions.

Robinson, Heaton, Lehman, and Stilson (1980) studied 107 people with structural

cerebral lesions and 123 normal controls. Sixty-nine of the patients were judged to have

focal lesions in either the left- or right-frontal, non-frontal, or frontal plus other cortical

areas. They reported findings consistent with those of Milner (1963) and Drewe (1974),

concluding that people with frontal lesions are significantly impaired on the WCST in

comparison with people who have non-frontal lesions.

Data from functional imaging has also highlighted the role of prefrontal cortex in

performance on the WCST. Weinberger, Bernam, and Zee (1986) tested 20 medication-

free schizophrenics and 25 normal controls using an automated version of the WCST

while measuring regional cerebral blood flow. The authors reported that normals showed

an increase in flow to the dorsolateral prefrontal cortex during WCST performance

relative to flow during a number matching control task. The reported change in blood

flow was specific to the dorsolateral prefrontal cortex, and was not observed in the

schizophrenic group. In the patients, but not the controls, the level of blood flow in the

dorsolateral prefrontal cortex was significantly correlated with level of performance on the

WCST. As summarized by Weinberger, Berman, and Daniel (1992), similar results have

been reported by Weinberger, Berman and Illowsky (1988) and Berman, Torrey, Daniel,

and Weinberger (1989).









Berman et al. (1995) reported that WCST performance results in significant

physiological activation of dorsolateral prefrontal cortex even after training and practice

on the test. The authors highlighted the role of working memory in WCST performance,

emphasizing the bilateral activation of a network including dorsolateral prefrontal cortex

and inferior parietal lobule (Berman et al., 1995).

Heaton (1981) and Heaton et al. (1993) reported on the performance of patients

with structural brain lesions and normal controls on the WCST. Heaton (1981) and

Heaton et al. (1993) found that all brain damaged groups performed worse than normal

controls on the test, and that subjects with lesions involving the frontal lobes performed

worse than subjects with lesions not involving the frontal lobes. Heaton et al. cautioned,

however, that the WCST is not sufficiently specific to frontal lobe lesions to warrant

interpreting poor performance on the test as diagnostic of frontal lobe damage

independent of other evidence.

This caution is given prominence by reports that, in some instances, people with

frontal and non-frontal lesions can not be differentiated on the basis of WCST

performance. Graffminan, Jones, and Salazar (1990) failed to find differences in WCST

performance between open head injured Vietnam veterans with frontal lobe or non-frontal

damage. Anderson, Damasio, Jones, and Tranel (1991) studied the specificity of the

WCST as a measure of frontal lobe damage in patients with stable, focal brain lesions.

Their primary comparison was between 49 subjects with frontal lobe lesions and 24

subjects with lesions outside the frontal lobes, including some with lesions of the thalamus

and basal ganglia. They reported no significant differences between the frontal and non-







38

frontal groups on any of the WCST variables examined. Anderson et al. (1991) concluded

that performance on the WCST likely reflects coordinated interaction among multiple

brain regions, and that poor performance alone is not an adequate basis on which to

diagnose frontal versus non-frontal damage.

The performance of people with epilepsy on the WCST has also been studied.

Hermann, Wyler, and Richey (1988) compared the performance of 35 people with

complex partial seizures and an established temporal lobe focus to an epilepsy control

group of six people with primary generalized seizures. Relative to the epilepsy controls,

patients with a non-dominant temporal lobe focus showed significantly more total errors

and perseverative responses. Patients with a dominant temporal lobe focus made

significantly more perseverative errors than epilepsy controls. Seventeen consecutive

cases were followed and tested after surgery; a significant decrease in perseverative

responses was found. Hermann et al. (1988) cautioned interpretation of the WCST in

patients with temporal lobe epilepsy, and suggested the possibility that "neural noise" from

temporal lobe/hippocampus might propagate along pathways linking temporal lobe with

frontal cortex leading to frontal dysfunction and impaired WCST performance. The

functional connection between hippocampus and frontal lobe was highlighted by the report

ofMeador et al. (1991) that forebrain mechanisms involved in the organization and

initiation of voluntary responses play a role in generating hippocampal theta activity.

Corcoran and Upton (1993) examined performance on a modified version of the

WCST in three groups of people with epilepsy: 16 with hippocampal sclerosis; 13 with a

unilateral temporal lobe focus; and 18 people with a unilateral focus in various frontal lobe







39

regions. Patients with hippocampal sclerosis performed worse than either the temporal or

frontal lobe groups. They took longer to complete the task, completed fewer categories,

and made more perseverative errors. The group with hippocampal sclerosis was not

significantly impaired relative to the other patient groups on other tests typically

associated with frontal lobe functioning (e.g., verbal fluency and Stroop testing).

Corcoran and Upton criticized Hermann, Wyler, and Richey's (1988) "neural noise"

hypothesis, and suggested that hippocampal damage results in poor card sorting

performance because of heavy working memory demands of the task. However, Corcoran

and Upton fail to address the potential impact their use of the modified WCST (Nelson,

1976) may have had on their findings. This version of the test reduces the ambiguity of

the task by providing subjects with considerably more structure. It uses response cards

which share one and only one attribute with each of the stimulus cards, and calls for

informing the subject when the sorting principle is changed. These differences may have

diminished the sensitivity of the task to frontal lobe dysfunction, and contributed to

Corcoran and Upton's results and conclusions.

Strauss, Hunter, and Wada (1993) studied 77 people with complex partial seizures

and clear evidence of a unilateral right- or left-temporal lobe focus. Twenty-six of the

patients had a history of damage before one year of age and 51 had a history of damage

after that time. It is unclear from the report if "damage" refers to the onset of seizures, a

risk factor for seizures, or some combination of both. Strauss et al. (1993) reported that

people with early damage and a left sided focus demonstrated significantly more

perseverative errors and perseverative responses on the WCST than people with a left









sided focus and a later onset of damage. The effects were still observed when the

contribution of language laterality was partialled out. Age at time of damage did not

predict the performance of people with right temporal lobe foci. The group with a right

sided focus also demonstrated impaired performance, but was not as severely impaired as

the early left sided patients. Strauss et al. (1993) suggested that the cognitive

consequences of brain damage may depend upon laterality and age at the time of damage.

Demonstrations that WCST performance can be impaired in people with non-

frontal seizure foci reinforce Heaton, Chelune, Talley, Kay, and Curtiss' (1993) caution

against reflexive interpretation of impaired performance as indicative of frontal lobe

damage. Impaired WCST performance is not specific to frontal lobe damage, particularly

with chronic lesions. As suggested by research with epileptic subjects, dysfunction in

other parts of the brain can interfere with the ability of the frontal lobes to perform the

task. However, this vulnerability to interference from other brain regions, including the

temporal lobes, does not mitigate the importance of the frontal lobes in performance on

the WCST. While it is clear that performance likely involves the integrated activity of

multiple brain regions, a significant contribution is made by the frontal lobes, particularly

the dorsolateral region. The WCST can be viewed as a sensitive, but not specific, measure

of frontal lobe functioning.

Anatomical studies have demonstrated that interhemispheric fibers from prefrontal

cortex pass through the anterior portion of the corpus callosum, including the rostrum and

genu (DeLacoste, Kirkpatrick, & Ross, 1985; Pandya & Rosene, 1985). Although she

acknowledged that the evidence was not unambiguous, Witelson (1989) indicated that the







41

rostrum likely contains fibers from the orbital prefrontal and inferior premotor areas while

the genu contains fibers from prefrontal cortex, including the dorsolateral prefrontal area.

Based on the important role of prefrontal cortex in performance on the WCST, and the

anatomy of its interhemispheric connections through anterior callosum, the anterior

callosum may be considered part of the integrated brain system subtending WCST

performance. The morphology of anterior callosum has not been directly studied with

respect to WCST performance, but Kapur (1985) reported a case which bears on this

issue. Kapur described a double dissociation between perseverative responses on a short-

term verbal memory test and perseveration on a modified version of the WCST. A patient

with a craniopharyngioma and dilatation of the third ventricle showed perseveration on a

test of short-term verbal memory, but not on the card sorting task. In contrast, a patient

with a midline subfrontal tumor described as an astrocytoma of the anterior corpus

callosum perseverated on the card sorting task and not on the test of verbal short-term

memory. Although the tumor likely affected more than anterior callosum, this finding is

consistent with the suggestions that the interhemispheric connections between prefrontal

cortices may play a role in performance on the WCST.

Based on the role of frontal cortex on WCST performance, it was predicted that

WCST perseverative error score would be more strongly correlated with anterior area of

the corpus callosum than posterior callosal area. The correlation was expected to be

negative, with higher perseverative error scores associated with smaller callosal areas.









Posterior Callosal Morphology and Facial Recognition Test Performance

The Facial Recognition Test (Benton, des. Hamsher, Varney, & Spreen, 1983) is a

test involving matching photographs of faces. Some of the items involve matching

identical images, while others involve matching to faces seen in side view or in shadow

under varying lighting conditions. Originally developed by Benton and Van Allen (1968),

Hamsher, Levin, and Benton (1979) studied the performance of 145 right-handed people

with unilateral focal cerebral lesions on the Facial Recognition Test. They concluded that

the performance of patients with right hemisphere lesions is more frequently impaired than

the performance of patients with left hemisphere lesions, and that people with posterior

damage have a higher frequency of defective performance than those with anterior lesions.

Among patients with left hemisphere disease, only those with substantial impairment of

oral language comprehension showed defective levels of performance on the Facial

Recognition test. Among this subgroup, the majority had posterior lesions (Benton, des.

Hamsher, Varney, & Spreen, 1983). Dricker, Butters, Samuels, and Carey (1978)

reported that people with right hemisphere lesions made more errors on a face-matching

task than controls, perhaps relying on "paraphernalia and expressions" rather than

underlying visuospatial features of the faces to guide their responses. Egelko et al. (1988)

reported that performance on the Facial Recognition test was uncorrelated with frontal

lobe damage, and closely associated with parietal lobe damage in a sample of right

hemisphere stroke patients. Impaired performance has also been reported with diffuse

damage following closed head injury (Levin, Grossman, & Kelly, 1977).









Sergent and Corballis (1989) suggested that both the left and right hemisphere

have the capacity to process facial stimuli, but that the right hemisphere has an advantage

in processing spatially rotated or disoriented faces. Based on the faster response times of

normal subjects than commissurotomized patients, Sergent and Corballis suggested that

corpus callosum allows for the joint participation of the hemispheres in the processing of

facial information initially received by one hemisphere, facilitating performance in normals.

Reviewing data from lesion and PET research with prosopagnosics, and visual field study

with normal subjects, Sergent (1995) suggested that the essential question is not which

hemisphere is better at carrying out a particular function, but how the two hemispheres

coordinate their separate operations to accomplish normal performance (Sergent, 1995).

Based on work with rhesus monkeys, Pandya and Rosene (1985) reported that

interhemispheric fibers from posterior parietal association areas pass through the posterior

body of the corpus callosum. Although there is some overlap, they indicated that fibers

from the superior parietal lobule generally pass through the dorsal portion of the callosum

while fibers from the inferior parietal lobule pass through the ventral portion of the

callosum. In humans, DeLacoste, Kirkpatrick and Ross (1985) reported that fibers from

the temporal-parietal-occipital region pass through posterior regions of the callosum.

Based on its relationship with posterior regions of cortex, it was predicted that

Facial Recognition Test scores would be more strongly correlated with posterior area of

the corpus callosum than anterior callosal area. The correlation was expected to be

positive, with higher Facial Recognition Test scores associated with larger callosal areas.








44

Differences in the strength of correlations were predicted between subjects with early and

late onset of seizures.













METHOD


Subjects


Two-hundred thirty-three individuals treated at Shands Teaching Hospital at the

University of Florida for intractable seizures were screened through record review as

potential study participants. Because the study relied on existing clinical records and data,

and did not impose any expense, risk, or time demands upon study participants, the Health

Center Institutional Review Board granted an exemption from the need for participant's

informed consent (F. N. Tompson, personal communication, February 10, 1995).

Records were reviewed with the goal of identifying right-handed subjects, 16-years

or older, with a unilateral left- or right-temporal lobe seizure focus and left-hemisphere

language dominance as demonstrated by Wada testing. A number of criteria were used

with the goal of excluding subjects with acquired pathologic changes in the central

nervous system which might affect callosum. Potential participants were excluded if they

had a history of head injury resulting in a loss of consciousness >30-minutes or if their

MRI revealed intracranial masses larger than 3-cm3 or evidence of other gross pathology

not including mesial temporal sclerosis. Potential participants with a history of cranial

irradiation or intracranial neurosurgery, not including grid or depth electrode placement as

part of the process of evaluation as candidates for epilepsy surgery, were excluded. Valid







46

data from presurgical neuropsychological testing and a presurgical MRI acquired using a

protocol that permitted subsequent volumetric analysis were also required for inclusion.

One-hundred thirty-seven subjects were excluded based on the criteria outlined

above. Some subjects might have met more than one exclusionary criterion, with the

reported reason for exclusion simply the first criterion identified. Twenty percent of the

excluded subjects were eliminated because either no epileptogenic focus had been found

or because an extra-temporal or bilateral foci were identified. Fourteen percent were

excluded because they were left-handed, with an additional 14% excluded because of

significant structural abnormalities noted on MRI. Thirteen percent had prior

neurosurgery, with 10% excluded for a history of head injury. Ten percent were younger

than 16-years of age at the time of their evaluation, with the remainder excluded because

an MRI image was not available for analysis (7%), because they were found to have non-

epileptic convulsive episodes (7%), because results of neuropsychological testing were

considered invalid (3%), or because right-hemisphere or mixed language dominance was

demonstrated on Wada testing (2%). An additional 34 subjects were excluded because

inadequate information was available to determine if they met criterion for inclusion as

study participants.

The final sample included 62 subjects evaluated as candidates for epilepsy surgery.

Descriptive information about the subjects is presented in Table 1. The total sample

included more females than males. Subjects had, on average, a high school education,

with a range from eight to 16-years of schooling. The sample included comparable

percentages of left- and right-temporal lobe epileptics. All subjects had a history of








47

chronic seizures, with a mean age at the onset of chronic seizures in the early teens. The

distribution of age at the onset of chronic seizures had a marked positive skew (z-score =

6.40, p < .001). Chronic seizures began before age ten for 50% of subjects, with the

remaining 50% beginning between age 11 and 62-years. The sample was equally divided

into subjects with a history of febrile seizures (44%, n = 27) and no history of febrile

seizures (45%, n = 28); medical records for 11% of subjects (n = 7) were inconclusive

regarding the presence or absence of febrile seizures.

Table 1
Demographic Features Patient Sample (N=62)


Mean or Percentage Standard Deviation or Percentage
Gender 58% Female (n = 36) 42% Male, (n = 26)
Age (years) Mean =36 SD = 10.9
Education (years) Mean = 12.3 SD = 02.0
Laterality of Seizure Focus 55% Left-temporal (n = 34) 45% Right-temporal (n = 28)
Age at Onset of Chronic Seizures Mean = 12.8 SD = 12.0


Neuropsychological Measures


All subjects had completed a comprehensive neuropsychological assessment as

part of the process of evaluation as candidates for epilepsy surgery. Most subjects had

been evaluated through the Neuropsychology Service of the Psychology Clinic, Shands

Teaching Hospital. Subjects' clinic charts were reviewed, with data recorded directly from

test protocols. In two cases, the subject's neuropsychological evaluation had been

completed elsewhere. In those cases, scores contained in the report of the









neuropsychological evaluation were entered as data. Results from testing with the

Wechsler Adult Intelligence Scale-Revised (WAIS-R) (Wechsler, 1981), the Wisconsin

Card Sorting Test (WCST) (Heaton, Chelune, Talley, Kay, & Curtiss, 1993), and the

Facial Recognition Test (Benton, des. Hamsher, Vamrney, & Spreen, 1983) were used as

data.

Based on testing with the WAIS-R, each subject's Verbal, Performance, and Full

Scale Intelligence Quotients had been determined (Wechsler, 1981). Freedom from

Distractibility index scores were calculated for each subject using the factor loadings

reported by Bornstein, Drake, and Pakalnis (1988). The Wisconsin Card Sorting Test was

administered and scored by a standardized method (Heaton, Chelune, Talley, Kay, &

Curtiss, 1993). The WCST perseverative error score was entered as data. The Facial

Recognition Test (FRT) was administered and scored by the standardized method

(Benton, des. Hamsher, Vamrney, & Spreen, 1983). Age and education corrected FRT raw

scores were used as data in this study.



Morphometric Measures


Magnetic resonance images were collected using a quadrature head coil in a

Seimens 1-T Magnetom. The scans were acquired using a volumetric MP-RAGE

sequence (Mugler & Bookeman, 1991) (repetition time, 10ms; echo time, 4ms; 10 flip

angle; FOV = 25.5; 160-mm excited volume; 130 x 256 matrix) as described by Leonard

et al. (1993). The procedure generated a series of artifact-free, high-contrast, gapless







49

sagittal sections. For most subjects (n = 50) images were collected with a 1.25-mm slice

thickness. Images for the remainder of the subjects (n = 12) had been collected using the

same machine with slightly different acquisition parameters resulting in 1.406-mm thick

slices. Image files were identified with a unique subject number. Morphometric

measurements were made blindly with respect to data on neuropsychological measures.

MRI data was downloaded from archival storage and transferred to a SUN-

workstation for image analysis. All images remained as digital data throughout the

analysis process. Consecutive sagittal slices were inspected to identify the slice

corresponding most closely to a midsagittal image based principally on patency of the

cerebral aqueduct. Comparisons were made among adjacent slices in the process of

selecting the best midsagittal image. All midsagittal slices were chosen by a single rater.

Using the MRI slice identified as the midsagittal image, several estimates of total

brain size (brain length, total midsagittal intracranial area, midsagittal area of cortex) and a

measure of total area of the corpus callosum were made by a single rater using software

written for this purpose. To determine the length of brain, a cursor was placed at its most

anterior and posterior points. The distance between the points was calculated, and the

result reported as a measure of the brain's length in centimeters.

A manual tracing routine was used to estimate the total midsagittal intracranial

area of the brain. The rater began tracing the dorsal surface, marking the contours of the

dura with cursor points. Tracing followed the dura from anterior to posterior, tracing

above the cortex and cerebellum. A straight line was followed at the level of the foramen

magnum, with tracing continuing along the dura on the ventral surface of the brain stem







50

and midbrain to the level of the infundibular stalk and pituitary. Tracing continued along

the ventral surface of the frontal lobe eventually connecting with the initial point on the

dorsal surface of the brain. Software calculated the area enclosed by the tracing and

reported the total area in square centimeters.

A similar procedure was used to measure the midsagittal area of cortex by tracing

its dorsal surface with a line parallel and slightly inferior to the surface of the dura. At the

tip of the occipital lobe, tracing proceeded rostrally following the ventral surface of cortex

at the superior margin of the straight sinus. Tracing continued rostrally along the superior

margin of the corpus callosum following the contours of its genu and rostrum. Tracing

continued along the ventral surface of the frontal lobe eventually connecting with the

starting point on the dorsal surface of the cortex. The total area enclosed by the tracing

was calculated and reported as the midsagittal area of cortex in square centimeters.

The total area of the corpus callosum on midsagittal section was measured by

manually tracing the surface of the callosum beginning at an anterior point on the dorsal

surface tracing posteriorly, continuing along the ventral surface anteriorly eventually

connecting with the starting point. The total area of the corpus callosum was calculated

and reported in square centimeters.

Using a separate program, the corpus callosum was divided into seven regions

following the method originally described by Witelson (1989). The rater selected the most

anterior and posterior points of the callosum, as well as at the most anterior point on the

inner convexity of the anterior callosum. Using those landmarks, and following the

algorithm reported by Witelson (1989), the program automatically generated lines marking










the borders between callosal regions. The rater manually traced the margin of each

segment resulting in a measure of the area of each region in square centimeters. A second

estimate of total midsagittal area of the corpus callosum was made by adding the area of

all segments. The corpus callosum of each subject was measured by the primary rater

using this program. A second rater measured the seven subregions of the corpus callosum

in 85% of the total sample. Figure 1 depicts a midsagittal section of brain highlighting

division of the corpus callosum into seven subregions as described by Witelson (1989).


Figure 1. Diagram depicting a midsagittal brain image highlighting division of the corpus callosum into seven
subregions.









Statistical Analysis


Neuropsychological and morphometric data were summarized using descriptive

statistics. The total sample was divided into two groups based on age at the onset of

chronic seizures. Group means on neuropsychological and morphometric measures were

compared using independent sample t-tests.


Reliability

The reliability of morphometric measurements was assessed in several ways. For a

sample including 85% (n = 53) of subjects, measurements of callosal area were made

separately by the primary and a secondary rater. A difference score between the primary

and secondary rater's measurement of total callosal area was calculated and tested against

the null hypothesis that the mean difference score was equal to zero. Pearson correlation

coefficients were calculated for the primary and secondary rater's measurements of each

callosal region. Intra-rater reliability was assessed by correlating separate measures of

total callosal area made by the primary rater using separate programs. Measurements

made by the primary rater were used in all subsequent analyses.


Approach to Data Analysis

Univariate distributions of neuropsychological and morphometric data were

screened for outliers using standardized scores and inspection of box plots of the data.

The assumption of normality was checked by inspection of histograms of data and by

testing for the significance of statistics for skewness and kurtosis. A conservative alpha







53

level of .001 was used to evaluate the significance of skewness and kurtosis (Tabachnick

& Fidell, 1996, p. 73). Unless otherwise indicated, distributions were free ofunivariate

outliers and did not deviate significantly from a normal distribution. The strength of

association between morphometric and neuropsychological measures was assessed using

Pearson correlation coefficients with significance tests at the a = .05-level (two-tailed)

unless otherwise indicated. In the total sample, power was approximately .80 to detect an

effect comparable in magnitude to that reported by Strauss, Wada, and Hunter (1994).

Calculations of power were made using a computer program written for the purpose

(Borenstein & Cohen, 1988). Screening for bivariate outliers was done using Mahalanobis

distances as recommended by Tabachnick and Fidell (1996, p. 67-68). Mahalanobis

distances represent the distance of each case from the centroid of remaining cases where

the centroid is the point created by the means of all the variables (Tabachnick & Fidell,

1996, p. 67). Data were screened for outliers in solution by examination of standardized

residuals. Significance testing for differences between correlation coefficients were

conducted using r to z transformations and procedures appropriate for either independent

or non-independent correlation coefficients.

Correlations between morphometric and neuropsychological measures were

calculated for the entire sample and separately for subjects with "early" and "late" seizure

onset. When results suggested a difference between groups in terms of the relationship

between morphometric and neuropsychological measures, the possibility of other between

group differences (e.g., males vs. females, L-temporal vs. R-temporal seizure focus) were








54

evaluated by testing the significance of differences between independent correlation

coefficients and by examination of scatterplots of data graphed separately by subgroup.













RESULTS


Neuropsychological Measures


Table 2 summarizes data from neuropsychological measures for the total sample,

and separately based on the laterality of subjects' seizure focus. Data from every measure

was not available for all subjects because of differences in practice among

neuropsychologists supervising subjects' clinical evaluations.



Verbal Scale Q10

Subjects' mean VIQ was at the low end of the Average range (M = 90.2, SD =

10.5). A single extreme score was identified (VIQ = 116, z-score = 2.5), although the

case was not a true univariate outlier as defined by Tabachnick and Fidel (1996, p. 67), p

>.001. Subjects with left-temporal lobe epilepsy had significantly lower VIQ scores than

subjects with a right sided focus, t (60) = -.3.53, p < .01.



Performance Scale IQ

Subjects' mean PIQ was at the low end of the Average range (M = 89.7, SD =

11.1). A single extreme score was identified (PIQ = 118, z-score = 2.6), although the case

was not a true univariate outlier (p > .001). The difference in PIQ scores between







56

subjects with left- and right-temporal lobe epilepsy was not significant, t (60) = -1.48, P >

.05.



Full Scale IQ

Subjects' mean FSIQ was in the Low Average range (M = 88.9, SD = 10.1). A

single extreme score was identified (FSIQ = 117, z-score = 2.8), although the case was

not a true univariate outlier (p > .001). Subjects with left-temporal lobe epilepsy had

significantly lower FSIQ scores than subjects with a right sided seizure focus, t (60) = -

3.01, p <.01.



Freedom from Distractibility Index

Freedom from distractibility factor scores were approximated for each subject

using a method to compute factor scores described by Comrey and Lee (1992). Scaled

scores from all WAIS-R subtests were multiplied by the factor loadings reported by

Bornstein, Drake, Pakalnis (1988) and summed. Digit Span (.76) and Arithmetic (.59)

had the heaviest loadings, with factor loading below .50 for other subtests. Mean FDI

scores are presented in Table 2. Subjects with left temporal lobe epilepsy had significantly

lower FDI scores than subjects with a right sided seizure focus, t (60) = -2.53, p < .05.



WCST Perseverative Error Score

A perseverative error score for the full 128-card administration of the Wisconsin

Card Sorting Test was available for 47 subjects. Five additional subjects had received an











Table 2
Descriptive statistics for neuropsvchological measures


WAIS-R VIQ
All Subjects
L-temporal
R-temporal
WAIS-R PIQ
All Subjects
L-temporal
R-temporal
WAIS-R FSIQ
All Subjects
L-temporal
R-temporal
FDI Factor Score
All Subjects
L-temporal
R-temporal
WCST Perseverative Errors
All Subjects
L-temporal
R-temporal
FRT Long Form Score
All Subjects
L-temporal
R-temporal


Mean


90.2
86.3
95.0

89.7
87.8
91.9

88.9
85.6
92.9

13.1
12.1
14.2

18.0
20.0
15.4

44.4
43.9
45.0


Standard Deviation Minimum


10.5 75
08.8 74
10.5 78

11.1 64
11.1 64
10.7 75

10.1 71
09.0 71
10.1 78

03.4 6.1
03.1 6.1
03.4 7.8

10.5 05
11.5 05
08.5 05

03.6 36
03.4 36
03.8 38


abbreviated 64-card administration of the WCST. A perseverative error score for a full

128-card administration was prorated for those five subjects by doubling the number of

perseverative errors made over 64-trials. Subjects had a mean perseverative error score of

18.0 (SD = 10.5; range 5 42). As a point of reference, this is a perseverative error score

approximately one standard deviation higher than average for adults less than 40 years of

age (Heaton, 1981). The difference in WCST perseverative error score between subjects

with L- and R-temporal lobe epilepsy was not significant, t (50) = 1.58, p > .05.


Maximum


116
109
116

118
110
118

117
105
117

20.9
17.6
20.9

42
42
32

50
49
50









Facial Recognition Test Score

Facial Recognition Test scores were available for 36 subjects. Twenty-eight

subjects had received the short form of the test. Short form scores were converted to long

form equivalents using standardized procedures (Benton, des. Hamsher, Vamrney, &

Spreen, 1983). Age and education corrections were made as described by Benton (1983),

although in most cases no correction applied. The mean score for all subjects on the

Facial Recognition Test was 44.4 (SD = 3.63; range 36 50). Although normally

distributed, the scores fell within a markedly restricted range. Only two subjects obtained

scores in the "defective" or "severely defective" ranges as defined by Benton (1983).

However, comparison of the score equivalents of percentile ranks reported by Benton

(1983) and those achieved by subjects in this study suggested a similar distribution of

scores. The difference in FRT score between subjects with L-ATL and R-ATL epilepsy

was not significant, t (34) = -0.92, p > .05.



Grouping Based on Age at Seizure Onset

The total sample (N = 62) was divided into two groups based on age at the onset

of chronic seizures. The "early" group included subjects with chronic seizures beginning

before age five (n = 20); the "late" group (n = 42) included subjects with a later seizure

onset. Subjects with isolated febrile seizures at a young age were not included in the

"early" group unless their chronic seizures began before age five. As shown in Table 3,

the groups did not differ in terms of age at the time of neuropsychological evaluation.

However, the "early" group had completed significantly fewer years of schooling, and had








59

significantly lower scores on all WAIS-R IQ measures. The difference between groups in

terms of FDI score was not significant. There were no significant between-group

differences in performance on the Facial Recognition Test, or in perseverative errors on

the Wisconsin Card Sorting Test. The "early" and "late" groups did not differ significantly

in terms of the percentage of included males and females (x2 (1, N = 62) = 1.73, p > .05).

Although there were a small number of subjects with a right sided seizure focus in the

"early" group, there was no significant difference in terms of the laterality of subjects'

seizure focus (x2 (1, N = 62) = 2.74, p > .05).


Table 3
Demographic features and neuropsychological test performance of subjects with "early"
and "late" seizure onset


Age (years)

Education (years)

WAIS-R VIQ
WAIS-R PIQ

WAIS-R FSIQ

FDI Score

WCST PE Score

FRT Long Form Score


Early

M = 36.0, SD= 09.0

M = 11.6, SD= 01.5

M = 84.5, SD = 08.3

M = 85.7, SD= 09.0

M= 83.8, SD = 07.7

M=12.1, SD = 03.3

M = 20.6, SD= 10.6

M = 44.7, SD= 04.2


Late

M=36.0, SD= 11.8

M = 12.6, SD= 02.1

M = 93.0, SD= 10.4
M=91.6, SD= 11.5

M=91.3, SD = 10.3

M = 13.5, SD= 03.4

M = 16.8, SD= 10.4

M = 44.3, SD= 03.5


Significance Test

t (60) = -.002, p > .05

t (52.1) = -2.11, p < .05"

t (60) = -3.19, p < .01*
t (60) = -2.02, p < .05"

t (60) = -2.89, p < .01*

t (60) = -1.52, p> .05

t (50) = 1.23, p > .05

t (34) = 0.31, p > .05


Morphometric Data


Inter-rater Reliability

A primary rater measured the callosum of all subjects, including measurement of

the seven callosal regions (R1-R7). Measurements made by a secondary rater were









available for 85% (n = 53) of the total sample. A difference score between the primary

and secondary rater's measurements of total area of the corpus callosum (sum of all

subregions) was calculated for each subject. The mean difference score (M = 0.02, SD =

.39) did not differ significantly from zero, (t (52) = 0.371, p = .712).

Pearson correlation coefficients were calculated between the primary and

secondary rater's measurements of total area of the callosum, and separately for each of

the seven subregions. The primary and secondary rater's measurements of total callosal

area were highly correlated (r = 0.92, p < .0001). Correlations between measurements of

the seven subregions were all significant at the p < .0001 level, (RI: r = .91; R2: r = .92;

R3: r = .86; R4: r = .86; R5: r = .86; R6: r = .90; R7: r = .93).



Intra-rater Reliability

The primary rater measured the corpus callosum of all subjects twice, once

measuring total area of the corpus callosum and once measuring the area of the seven

subregions. The correlation between the measurement of total area and the sum of area

measurements RI through R7 was high (r = .94, p < .0001).



Measurements of Brain Size

Table 4 summarizes measurements of subjects' brain size, including brain length,

total intracranial area, and total area of cortex on the midsagittal MRI slice. Significant

sex differences were found, with males having greater brain length (t (60) = 5.7, p <.001),








61

greater midsagittal intracranial area (t (60) = 6.3, p <.001, and greater midsagittal area of

cortex (t (60) = 5.5, p <.001) than females.


Table 4
Measurements of brain size in the total sample and separately for males and females


Mean Standard Deviation Minimum Maximum
Brain Length (cm)
All Subjects 16.5 0.93 14.5 18.8
Males (n=26) 17.1 0.82 15.6 18.8
Females (n=36) 16.0 0.71 14.5 17.8

Midsagittal Intracranial Area (cm2)
All Subjects 153.6 13.51 126.8 186.7
Males (n=26) 163.5 10.7 142.8 186.7
Females (n=36) 146.4 10.6 126.8 175.5

Midsagittal Area of Cortex (cm2)
All Subjects 90.0 11.1 61.5 115.0
Males (n=26) 97.5 9.7 83.6 115.0
Females (n=36) 84.6 8.6 61.5 101.0


Brain Size and Age at Seizure Onset

There was no significant between-group difference in midsagittal area of cortex (t

(60) = -1.43, p > .05), although there was a trend toward a larger midsagittal intracranial

area in the "late" (M = 155.86, SD = 14.07) versus "early" group (M = 148.86, SD =

11.14), (t (60)= -1.95, p = .06). Brain length was significantly greater in the "late" group,

(t (60) = -2.14, p < .05). However, the absolute magnitude of the difference was

extremely small: "early" (M = 16.63, SD = 0.94) vs. "late" (M = 16.11, SD = 0.83).

Measurements of Corpus Callosum

Table 5 summarizes measurements of the corpus callosum, including total area and

area of subregions (R1-R7) for all subjects.










Table 5
Midsagittal area (cm2) of the corpus callosum (N = 62)


Mean Standard Deviation Minimum Maximum
Total Area 6.20 0.96 4.14 8.67
RI (Rostrum) 0.27 0.11 0.05 0.63
R2 (Genu) 1.30 0.29 0.81 2.01
R3 (Rostral Body) 0.87 0.19 0.50 1.31
R4 (Ant. Midbody) 0.77 0.15 0.50 1.11
R5 (Post. Midbody) 0.67 0.16 0.38 1.06
R6 (Isthmus) 0.59 0.15 0.30 1.00
R7 (Splenium) 1.73 0.28 1.02 2.36


Standard scores were calculated representing each subject's total callosal area; no

univariate outliers were identified. The distribution of measurements appeared normally

distributed, and was not marked by excessive skewness or kurtosis.

Examination of standardized scores revealed an outlier among measurements of

the rostrum (RI). The value was a true univariate outlier (z-score = 3.35, P < .001). With

the outlier included the distribution of RI measurements was marked by a substantial

positive skew skewnesss = 0.98, z-score = 3.22). The midsagittal image for the subject in

question was reviewed. The subject's rostrum appeared larger than other subject's, with

comparable area measurements made by both the primary and secondary rater. Because

extreme scores can distort correlation coefficients, the data point was dropped from

subsequent analyses. Removal of the outlier reduced the distribution's positive skew

skewnesss = 0.65, z-score = 2.13, p > .01)

The remaining distributions of area measurements for callosal regions (R2-R7)

were checked for the presence of univariate outliers and violations of the assumption of









normality. No univariate outliers were identified. The distributions appeared normally

distributed; none were marked by excessive skewness or kurtosis.

Subjects' age at the onset of chronic seizures ("early" versus "late") did not affect

area of callosal regions (R1-R7) in a one-way MANOVA, F (7,53) = 0.273, p > .05.

There was a significant main effect of sex on total callosal area in a one-way

MANOVA, F (7,53) =4.1, p < .01. Each of the seven callosal regions was significantly

larger in males than females, pg < .01. However, the main effect of sex was no longer

significant in a one-way MANCOVA entering midsagittal intracranial area (F (7,52) =

1.5, p > .05) or midsagittal area of cortex as covariates (F (7,52) = 2.0, p> .05).

Laterality of subjects' seizure focus (L-temporal versus R-temporal lobe) did not

affect area of callosal regions (R1-R7) in a one-way MANOVA, F (7,53) = 0.345, p > .05.



Relationship Between Morphometric and Neuropsychological Measures


Full Scale 10

Correlation of FSIO with area of callosal regions (R1-R7)

Significant positive correlations were predicted between FSIQ and area of callosal

subregions, particularly area of the splenium, based on Strauss, Wada, and Hunter's (1994)

findings in a comparable patient sample. To test that prediction, correlations were

calculated between FSIQ and area of callosal regions (R1-R7), and are presented in Table

6.










Table 6
Correlations of Full Scale IQ with area (cm2) of callosal subregions (N = 62)


RI R2 R3 R4 R5 R6 R7
FSIQ r=.16 r=.13 r=.15 r=.19 r=.13 r=.15 r=.23
*
ns ns ns ns ns ns ns
Two-tailed


Because bivariate outliers can distort correlation coefficients, data were screened

for the presence of bivariate outliers using Mahalanobis distances for each case as

recommended by Tabachnick and Fidel (1996, p. 57). The Mahalanobis distances were

evaluated using the chi-square distribution with two degrees of freedom at the 1 < .001

level (Tabachnick & Fidell, 1996, p. 94). Using that criteria, no bivariate outliers were

found among data for the relationship between FSIQ and area of each of the seven callosal

subregions.

Although the correlation between FSIQ and R7 area was not significant at the a =

.05 level as a two-tailed test, Strauss, Wada, and Hunter's (1994) report of a significant

positive correlation between area of the splenium and Full Scale IQ provided a basis for

one-tailed significance testing. As a one-tailed test, the null hypothesis of no relationship

between FSIQ and midsagittal area of R7 was rejected, p < .05.

A potential outlier in solution was identified among standardized residuals for the

correlation of FSIQ and areas R3, R5, and R7. The residuals were not significant at the a

= .001 level, although they were extreme (p < .01). A single case accounted for each

extreme residual. With data from that case excluded, the following correlations of FSIQ

and callosal area were obtained: R3 r = .26, p = .05); R5 r = .23, p > .05); and R7 r = .35,









p < .01). The correlation between FSIQ and area of the splenium remained significant

when controlling for brain size using midsagittal area of cortex, and did not differ

significantly between males and females or between subjects with L-temporal and R-

temporal lobe epilepsy.

The prediction that area of the splenium would have a significant positive

correlation with FSIQ was supported. The correlation of FSIQ and area of the splenium

was significant as a one-tailed test, and as a two-tailed test after data from a potential

outlier in solution was excluded. However, the correlation between FSIQ and R7 area

was not significantly different (p < .05) from the correlations between FSIQ and callosal

areas R1-R6 using the technique proposed by Meng, Rosenthal, and Rubin (1992) to

compare non-independent correlation coefficients.


Comparison of correlations for subgroups based on age at seizure onset, sex and laterality

of seizure focus

Influence of age at seizure onset. The question of whether age at seizure onset

influences correlations between morphometric and neuropsychological measures was

approached by comparing correlation coefficients calculated separately for subjects with

"early" and "late" seizure onset. Table 7 presents correlation coefficients between FSIQ

and area of callosal regions (R1-R7) calculated separately for subjects with "early" and

"late" seizure onset. Mahalanobis distances were examined separately for each group; no

bivariate outliers were identified.










Table 7
Correlations of Full Scale IQ and area (cm2) of callosal subregions for subjects with
"early" (n = 20) and "late" (n = 42) seizure onset


RI R2 R3 R4 R5 R6 R7
"Early"
FSIQ r = -.25 r = -.23 r = -.01 r=.17 r =-.12 r=.004 r=-.16
S=ns 2 = ns = ns = ns = ns p2 = ns I = ns
"Late"
FSIQ r=.28 r=.22 r=.22 r=.18 r=.21 r=.15 r=.29
12 = n1s p = ns = ns p = ns = ns = ns = ns
Two-tailed


None of the differences between correlation coefficients for the "early" and "late"

groups were statistically significant. However, there was a trend toward between group

differences for regions RI (p = .06), R2 (p =. 12), and R7 (p =. 11). For each of these

regions, larger callosal areas in the "early" group were associated with lower FSIQ scores

while in the "late" group larger areas were associated with higher FSIQ scores. The

difference between correlation coefficients did not approach significance for regions R3 (p

= .42), R4 (p = .97), R5 (p = .25) and R6 (p = .61).

A single case in the "late" group with an extreme standardized residual (p < .01)

from the regression line relating FSIQ and R7 area was identified. Excluding that case,

the correlation between FSIQ and R7 area in the "late" group was significant (r = .45, p <

.01), with variation in area of the splenium accounting from approximately 20% of the

variance in FSIQ. The difference between that correlation and the correlation for the

"early" group (r = -. 16) was significant, p < .05.









The prediction that age at seizure onset would influence correlations between

FSIQ and callosal areas was supported. With data from a potential outlier in solution

excluded, there was a significant positive correlation in the "late" group between area of

the splenium and FSIQ, and a negative correlation in the "early" group. The difference

between correlations for the "early" and "late" groups was significant. Including data from

all subjects, there was a trend toward negative weak-moderate correlations between FSIQ

and area of anterior (rostrum, genu) and posterior (splenium) callosum in subjects with

"early" seizure onset, and positive weak-moderate positive correlations between FSIQ and

anterior and posterior callosal areas in subjects with "late" seizure onset.

Effects of sex and laterality of seizure focus. There were no significant differences

in correlations of FSIQ and area of callosal regions R1-R7 between subjects with L-

temporal and R-temporal lobe epilepsy. However, there was a significant difference

between males and females in the partial correlations of FSIQ and area of the rostrum

controlling for midsagittal intracranial area. Among females there was a significant

positive correlation (r = .46, p < .01), while for males there was a non-significant negative

correlation (r = -.33, > .05). The difference between correlations for males and females

was significant, p < .01.



Verbal Scale 10

Correlation of VIO with area of callosal regions (R1-R7)

Significant positive correlations were predicted between VIQ and area of callosal

regions, particularly area of the splenium. Table 8 presents correlation coefficient between

Verbal Scale IQ and midsagittal area of each callosal subregion.










Table 8
Correlations of Verbal Scale IQ with area (cm2) of each callosal subregion (N = 62)

RI R2 R3 R4 R5 R6 R7
VIQ r=.22 r=.16 r=.21 r=.19 r=.19 r=.17 r=.32
P ns ns ns ns ns ns <.05
* Two-tailed

A significant correlation was identified between VIQ and area of the splenium,

with larger splenial area associated with higher VIQ. The correlation remained significant

when controlling for brain size using midsagittal area of the cortex (r = .27, p < .05).

However, the correlation between VIQ and R7 area was not significantly different from

the correlations between VIQ and callosal areas R1-R6, p > .05.


Comparison of correlations for subgroups based on age at seizure onset, sex and laterality

of seizure focus

Effect of laterality of seizure focus. The relationship between VIQ and splenial

area did not differ depending upon the laterality of subject's seizure focus. For subjects

with a left temporal lobe focus the correlation was r = .44, p < .01. Among subjects with

a right temporal lobe focus, two cases were identified with extreme residuals (P < .01).

Excluding those cases, the correlation between VIQ and splenial area was significant (r =

.51, p < .01), although with data from the two cases included the correlation was not

significant, r = .24, p > .05.

Influence of gender. The relationship between VIQ and splenial area did differ

depending upon subject's sex. For males the correlation was significant, (r = .41, P < .05),

with a comparable correlation for women (r = .35, p < .05) after data from two cases with










extreme residuals (p < .01) were removed. The cases are the same individuals identified

above. With the cases included, the correlation between VIQ and splenial area was not

significant, (r = 18, p > .05).

Effect of age at seizure onset. Correlation coefficients were calculated separately

for subjects with "early" and "late" seizure onset. Mahalanobis distances were examined

separately for each group; no bivariate outliers were identified. Table 9 presents

correlations between VIQ and area of callosal regions (R1-R7) separately for subjects with

"early" and "late" seizure onset.


Table 9
Correlations of Verbal Scale IQ and area (cm2) ofcallosal subregions for subjects with
"early" (n = 20) and "late" (n = 42) seizure onset

RI R2 R3 R4 R5 R6 R7
"Early"
VIQ r=-.14 r=-.25 r=.16 r=.001 r=-.16 r = -.09 r = -.04
S= ns p = ns = ns = ns = ns = ns I2 = ns
"Late"
VIQ r=.35 r=.27 r=.27 r=.24 r=.31 r=.22 r =.39
p <.05 p = ns p = ns 1 = ns P <.05 p = ns p <.05
* Two-tailed


None of the differences between correlation coefficients for the "early" and "late"

groups were statistically significant. However, there was a trend toward between group

differences for RI (p = .08), R2 (p = .07), R5 (p =. 10), and R7 (p =. 12). Larger areas

for each subregion in the "early" group were associated with lower VIQ scores, while in

the "late" group larger areas were associated with higher VIQ scores. The difference









between correlation coefficients did not approach significance for R3 (p = .69), R4 (p =

.40), or R6 (p =.28).

For subjects with a "late" onset of seizures, larger midsagittal areas of RI (genu),

R5 (posterior midbody), and R7 (splenium) were significantly correlated with higher VIQ

scores. The correlation between R5 (posterior midbody) and VIQ appeared related to

brain size; the partial correlation controlling for midsagittal intracranial area (r = .25, p =

.13) was not significant. The correlations between VIQ and area of the rostrum (RI) and

splenium (R7) were reduced by controlling for total midsagittal area, but generally

remained significant. The partial correlation of Ri and VIQ controlling for midsagittal

area was marginally significant (r = .32, p = .05), while the partial correlation of R7 and

VIQ controlling for midsagittal area was significant (r = .36, p < .05).

Area of the rostrum (R1) and splenium (R7) were intercorrelated (r = .45), and

were not independent predictors of VIQ. In stepwise multiple regression modeling VIQ

entering Rl- and R7-areas as potential predictors, area of the rostrum was eliminated from

the model. The model including area of the splenium (R7) was significant (F (1,39) =

7.24, p = .01), accounting for 15% of the variance in VIQ.

Figure 2 is a scatterplot of VIQ and area of the splenium (R7) with linear

regression lines plotted separately for subjects with "early" and "late" seizure onset.









71



120




110'
o3 0
o 0

E3
100' a a -E"
0' 0 -[



DO O Seiur O se
El

0 0 E3t (
0)
90. 3 0 0
E3


0 0 --- ezure Onset
80 C3 0 Late (>5-yrs)
o 0 Rsq = 0.1535
0 00 E ~~3
0 Early (<5-yrs)
70 Rsq = 0.0012
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

Area of Splenium (R7) cm/sq

Figure 2. Scatterplot of VIQ and area of the splenium (R7) with regression lines plotted
separately for subjects with "early" and "late" seizure onset.



As suggested by Figure 2, data contributed by two subjects ("late" onset with VIQ

> 110) may have had a disproportionate influence on the correlation coefficient for the

"late" subgroup. It is important to note that the cases were not identified as bivariate

outliers based on Mahalanobis distances, and do not represent true outliers in solution (p <

.001) (Tabachnick & Fidell, 1996, p. 139). However, the two cases did have extreme

residuals, p2 < .01. With data from the two cases excluded, the correlation for the "late"

group between area of the splenium and VIQ was strengthened (r = .56, p < .001). The








72

difference between that correlations for the "late" group and the correlation for the "early"

group (r = -.04, > .05) was significant, p < .05.

There was a significant difference between partial correlations of VIQ and area of

the rostrum controlling for midsagittal intracranial between males and females. Among

females there was a significant positive correlation (r = .47, p < .01), while for males

there was a non-significant negative correlation (r =-. 21, p > .05). The difference

between correlations for males and females was significant, p < .01.



Performance Scale IQ

Correlations with area of callosal regions (R1-R7)

Significant positive correlations were predicted between PIQ and area of callosal

regions, although the results did not support that conclusion. Table 10 presents

correlation coefficients between Performance Scale IQ and midsagittal area of each

callosal subregion. None of the correlations between PIQ and area of callosal regions

were significant.

Table 10
Correlations of Performance Scale Q10 with each of the callosal subregions


RI R2 R3 R4 R5 R6 R7
VIQ r=.04 r=.10 r=.01 r=.17 r=.06 r=.16 r=.10
*
p* ns ns ns ns ns ns ns
Two-tailed







73

Comparison of correlations for subgroups based on age at seizure onset, sex and laterality

of seizure focus

Influence of age at seizure onset. The prediction that age at seizure onset would

influence correlations between PIQ and area of callosal regions was not directly

supported. Correlation coefficients were calculated separately for subjects with an onset

of chronic seizures before age five ("early") and subjects with later onset ("late") and are

presented in Table 11. None of the differences between correlation coefficients for the

"early" and "late" groups were statistically significant. However, there was a trend toward

between group differences for regions RI (p = .05), R2 (p =. 13), R3 (p =. 12) and R7 (P

=.11). Larger areas for each of these regions in the "early" group was associated with

lower PIQ scores, while in the "late" group larger areas were associated with higher PIQ

scores. The difference between correlation coefficients did not approach significance for

regions R4 (p = .57), R5 (p = .47), or R6 (p = .89).

Effects of sex and laterality of seizure focus. There were no significant differences

in correlations of PIQ and area ofcallosal regions R1-R7 between subjects with L-

temporal and R-temporal lobe epilepsy. However, there was a significant difference

between males and females in the partial correlation of FSIQ and area of the rostrum

controlling for midsagittal intracranial area. Among females there was a significant

positive correlation (r = .35, p < .05), while for males there was a marginally significant

negative correlation (r = -.41, p = .05). The difference between correlations for males

and females was significant, p < .01.










Table 11
Correlations of Performance Scale Q10 and each of the callosal subregions for subjects
with "early" (n = 20) and "late" (n = 42) seizure onset

RI R2 R3 R4 R5 R6 R7
"Early"
PIQ r=-.37 r=-.24 r=-.23 r=.28 r= -.10 r=.10 r= -.29
p=ns ==ns l=ns =nfs p =ns p =ns p =ns
"Late"
PIQ K=.18 r=.19 r=.21 r=.12 r=.ll r=.14 r=.16
P = ns p = ns = ns = ns = ns p = ns p=ns
Two-tailed


Effects of sex and laterality of seizure focus. There were no significant differences

in correlations of PIQ and area of callosal regions R1-R7 between subjects with L-

temporal and R-temporal lobe epilepsy. However, there was a significant difference

between males and females in the partial correlation of FSIQ and area of the rostrum

controlling for midsagittal intracranial area. Among females there was a significant

positive correlation (r = .35, p < .05), while for males there was a marginally significant

negative correlation (r = -.41, p = .05). The difference between correlations for males

and females was significant, p < .01.


Freedom from Distractibility Index

Correlations with area of callosal regions (RI-R7)

Based on research in children with ADHD, it was predicted that anterior and

posterior regions of callosum would have significant positive correlations with FDI score.

Table 12 presents the correlations between FDI score and area of callosal regions (R1-R7)

for the total sample, and separately for subjects with "early" and "late" seizure onset. For








75

the total sample, area of region R5 had a significant positive correlation with FDI score, (r

= .27, p < .05). The correlation between area of region R4 (anterior midbody) and FDI

score approached significance.

Table 12
Correlations between FDI scores and area ofcallosal regions


RI R2 R3 R4 R5 R6 R7
All Subjects (n= 62) r=.18 r=.17 r=.17 r=.24 r=.27 r=.18 r=.19
p = ns = ns P = ns p =.06 12 <.05 p = ns p = ns
"Early" (n = 20) r=-.ll r=-.ll r=-.10 r =-.05 r=-.32 r=-.12 r =-.44
p = ns p = ns p = ns p = ns p = ns p = ns p =.06
"Late" (n= 42) r =.31 r=.27 r =.28 r=.36 r=.50 r=.29 r=.34
p <.05 p = ns p=ns p <.05 P <.O1 p= .07 p <.05
Two-tailed


Comparison of correlations for subgroups based on age at seizure onset, sex and laterality
of seizure focus

Influence of age at seizure onset. Among subjects with "early" seizure onset, none

of the correlations between FDI score and area of callosal regions was significant (p >

.05), although the correlation between FDI score and area of the splenium did approach

significance, (r = -.44, p = .06). In the "early" group, there was a trend toward larger

splenial areas being associated with lower FDI scores.

Among subjects with "late" seizure onset, correlations between FDI scores and the

area ofcallosal regions RI (r = .31, p < .05), R4 (r = .36, p < .05), R5 (r = .50, p < .01)

and R7 (r = .34, p < .05) were significant, and the correlation with R6 (r = .29, p = .07)

approached significance. In a stepwise multiple regression modeling FDI score and

entering the area of each callosal region as a potential predictor, only the area of the

posterior midbody was retained. The equation relating area of the posterior midbody (R5)









and FDI score was significant, F (1, 39) = 11.89, p < .001. The area of other callosal

regions were excluded as nonsignificant predictors from the regression model including R5

area.

The difference between correlations for the "early" and "late" groups were

significant for regions R5, posterior midbody (p < .01) and R7, splenium (p < .01). For

both regions, larger callosal regions in the "early" group were associated with lower FDI

scores, while in the "late" group larger areas were associated with higher FDI scores.

Effects of sex and laterality of seizure focus. There were no significant differences

between males and females or based on the laterality of subject's seizure focus in terms of

correlations of FDI score and area of callosal regions (R1-R7).



Correlations of Anterior and Posterior Callosal Areas with Performance on Wisconsin

Card Sorting and Facial Recognition Tests

It was predicted that WCST perseverative error score would be more strongly

correlated with area of the anterior portion of callosum which contains fibers

interconnecting frontal lobes than with area of posterior callosum. To test this

hypothesis, a variable reflecting area of the anterior portion of the callosum was created by

adding the areas of the rostrum (RI) and genu (R2). A variable reflecting area of the

posterior callosum was created by adding the areas of the isthmus (R6) and splenium (R7).

The distributions of anterior and posterior callosal areas appeared normally distributed,

without significant skewness or kurtosis. No univariate outliers were identified. Table 13

presents correlations of anterior and posterior callosal areas with WCST perseverative

error and FRT score.










Table 13
Correlations between callosal area (anterior vs. posterior in cm2) and test score (WCST
vs. Facial Recognition)

Anterior Callosal Area Posterior Callosal Area
(Rostrum + Genu) (Isthmus + Splenium)
WCST Perseverative Error Score r =-.20 r = -.15

p =. 16 p =.30

n=51 n=52
Facial Recognition Test r = -.21 r=. 14
Corrected Long Form Score
2 =.24 12 =.42

n=35 n=36
STwo-tailed


The correlation of anterior callosal area with WCST perseverative error score (r =

-.20, p > .05) was not significantly different from the correlation between posterior

callosal area and WCST performance (r = -. 15, p > .05), p > .05. The correlation between

posterior callosal area and FRT score (r =. 14, p > .05) was not significantly different from

the correlation between anterior callosal area and FRT score (r = -.21, p > .05), p >.05.

Table 14 presents correlations of anterior and posterior callosal areas with WCST

perseverative error and FRT separately for subjects with "early" and "late" seizure onset.

In the subsample of subjects with an "early" onset of seizures, the correlation between

anterior callosal area and WCST perseverative error score (r = -.01, 1 > .05) was not

significantly different from the correlation between posterior callosal area and WCST

performance (r = -.36, p > .05), p > .05. The correlation of posterior callosal area with

FRT score (r = -.47, p > .05) was not significantly different from the correlation between

anterior callosal area and FRT score (r = -.83, p < .01), p > .05.











Table 14
Correlations between callosal area (anterior vs. posterior) and test score (WCST vs.
Facial Recognition) for all subjects and separately for subjects with "early" and "late"
seizure onset


Anterior Callosal Area Posterior Callosal Area
(Rostrum + Genu) (Isthmus + Splenium)

WCST Perseverative Error "Early" r= -.01 r = .36
Score p =.97 =.16
n=17 n=17
"Late" r = -.23 r = -.08
p =.20 12 =.67
n=34 Dn=35
Facial Recognition Test "Early" r = -.83 r =.-.47
Corrected Long Form Score P = <.01 p =.15
n=ll n=ll

"Late" r=.16 r =.31
p =.45 p =.13
n 24 nR=25
* Two-tailed


As suggested by Figure 3, the very strong correlation between anterior callosal

area and FRT score may have been influenced by data from two cases with relatively large

anterior callosal areas. Excluding those two cases, the correlation of FRT performance

with area of anterior callosum fell from r = -.83 to r = -.63 (p = .05). Among subjects

with "late" seizure onset, the correlation of WCST perseverative error score with anterior

callosal area (r = -.23, p > .05) was not significantly different from the correlation of

WCST score with posterior callosal area (r = -.08, p > .05), p > .05. The correlation

between FRT score and posterior callosal area (r = .31, p > .05) was not significantly

different from the correlation between FRT score and anterior callosal area (r =. 16, 1 >

.05), > .05.










79


There were no significant differences between correlations ofWCST perseverative


error score and FRT score with anterior and posterior callosal areas based on either


subjects' sex or the laterality of their seizure focus.


50,


48"


S46-


I-4 44-


I 42-

41
400


1.0 1.2 1.4 1.6 1.8 2. .2 2.4 2.6

Anterior Callosal Area (cm/sq)

Figure 3. Scatterplot of anterior area (cm/sq) of the corpus callosum and Facial
Recognition Test score in subjects (n=l 1) with an "early" onset of seizures.


a
a


D a a


a a
a
a





a



D













DISCUSSION


Overview and Summary


Because the corpus callosum appears late in phylogeny, becomes apparent

relatively late in the course of ontogeny, and does not become fully myelinated, or

presumably fully functional, until relatively late in life, Hoptman and Davidson (1994)

advanced the hypothesis that callosal connections have cognitive consequences, and that

the functional significant of the corpus callosum may be in terms of the highest order, most

complex, and recently evolved cognitive capabilities. This view is bolstered by findings

that growth of the callosum continues beyond the point at which most other cerebral

structures have reached maturity (Pujol, Vendrell, Junque, Marti Vilalta, 1993), and can

be influenced by environmental experiences (Juraksa & Kopcik, 1988; Schlaug, Jancke,

Huang, Staiger & Steinmetz, 1995).

Relationships between the size of the corpus callosum and global measures of

cognitive functioning have been demonstrated in a number of clinical samples, with the

typical finding that atrophy of the corpus callosum is associated with impaired cognitive

functioning. Decreased callosal size in some patient groups may be attributable to disease

processes primarily affecting white matter tracts (e.g., multiple sclerosis) while in other

groups, like those with dementia, decreased callosal size may reflect damage to the cortex.

The extent to which callosal size and cognitive functioning are associated in normals is

80









unsettled. In a heterogenous group of patients with intractable epilepsy, Strauss et al.

(1994) demonstrated a positive correlation between area of the posterior region of the

callosum corresponding to the splenium and Full Scale IQ. They suggested that area of

posterior callosum may serve as a marker for the number of interhemispheric connections

linking posterior brain regions particularly important for the type of problem solving

required in intelligence tests (Strauss et al., 1994).

This study demonstrated a significant positive correlation between area of the

splenium and measures of intellectual ability in epileptics, replicating the result previously

reported by Strauss et al. (1994). The relationship was not accounted for by differences in

brain size as indexed by midsagittal area of cortex, and did not differ depending upon

subject's sex or the laterality of their temporal lobe seizure focus. However, the age at

which subjects began to experience chronic seizures did affect the nature of the

relationship. The significant positive correlation between area of the splenium and

measures of Verbal and Full Scale IQ was only true for subjects with chronic seizures

beginning after age five. In contrast, for subjects with earlier seizure onset, correlations

between measured IQ and splenial area were in the opposite direction (i.e., negative),

although not statistically different from zero. The difference between correlations for the

"early and "late" groups was significant, consistent with predictions based animal research

indicating that early seizure activity may result in the sparing of inappropriate callosal

projections. Correlations between measures of callosal morphology and PIQ were not

significant.

Although not predicted, a significant sex difference was found in the relationship

between area of the callosum's rostrum and measured IQ. For females, larger rostral areas









were significantly associated with higher Verbal, Performance, and Full Scale IQs, while

for males the correlations were negative. This finding is discussed in the context of a

previously reported sex difference in the relationship of callosal morphology and measured

IQ, and suggestions that cognitive processes may be less highly lateralized in females and

more dependent upon interhemispheric communication.

Based on reports that children with attention-deficit/hyperactivity disorder have

smaller anterior and posterior callosal areas than normal controls (Hynd et al., 1991;

Geidd et al., 1994; Semrud-Clikeman, et al. 1994), and that the Freedom from

Distractibility factor is especially prominent in WAIS-R data of people with epilepsy

(Bomrnstein & Cohen, 1988), positive correlations were predicted between the area of

anterior and posterior callosal regions and FDI score. Examining data from all subjects,

the predicted relationships between FDI score and measures of anterior and posterior

callosal morphology were not found. However, a larger area of the callosal region

including fibers from posterior parietal cortex was found to have a significant association

with higher FDI scores.

As with other correlations between callosal morphology and IQ measures, the

significant positive correlation in the total sample appeared to only reflect the nature of the

relationship for the subgroup of subjects with chronic seizures beginning after age five.

The direction of the relationship appeared to be reversed for subjects with earlier seizure

onset. The difference between the significant positive correlation of area of the posterior

midbody and FDI score for subjects in the "late" seizure onset group and the negative

correlation for subjects in the "early" group was significant. For subjects with "late"

seizure onset, significant positive correlations were also found, as had been predicted for









the total sample, between FDI score and area of the callosal regions interconnecting

anterior and posterior association cortex. The results did not support predictions made

about differential relationships between area of anterior and posterior callosum and WCST

Perseverative Error and Facial Recognition Test scores.

The discussion begins with a comparison of demographic features,

neuropsychological test performance, and size of the corpus callosum between this study's

sample and other groups with intractable epilepsy described in the literature. The

discussion proceeds, attempting to place the study's findings in the context of empirical

and theoretical accounts of the role and mechanism of interhemispheric communication in

cognitive processing. Finally, limits to external validity imposed by the nature of the

population studied, correlational design, reliance on archival test data, and limited power

to test some potential associations will be discussed.



Patient Sample

On basic demographic features, the subjects included in this study were similar to

other samples of patients with intractable epilepsy described in the literature. The subjects'

average age and number of years of education was comparable to that reported for other

groups tested in the process of evaluation as candidates for surgical treatment (Bornstein,

Drake, & Pakalnis, 1988; Chelune, Naugle, Luders, & Awad, 1991). Their average age at

seizure onset was nearly identical to those reported by Chelune, Naugle, Luders, and Iwad

(1991) for left and right temporal lobectomy patients. In comparison to the patients tested

by Strauss et al. (1994), the subjects included in this study were older, with a difference of

approximately six years between the groups' mean age. Other demographic comparisons









were not possible because Strauss et al. (1994) did not report the gender composition,

number of years of education, age at seizure onset, or other information concerning the

nature of their subjects' medically refractory seizures.



Neuropsychological Measures

Subjects' average FSIQ on the WAIS-R was in the Low Average range of

intellectual functioning, in the same range as scores reported for other samples of patients

with intractable epilepsy (Bornstein, Drake, & Pakalnis, 1988; Chelune, Naugle, LUiders &

Awad, 1991; Strauss, Wada, & Hunter, 1994). Subjects with right temporal lobe epilepsy

had a significantly higher FSIQ, and had completed more years of schooling on average

than the left temporal lobe group. These same differences were also reported by Chelune

et al. (1991). As anticipated (Kaufman, 1990), subjects with a left temporal lobe seizure

focus scored significantly lower on the Verbal than Performance Scale of the WAIS-R.

However, there was no significant difference in PIQ based on laterality of seizure focus.

Subjects with left temporal lobe epilepsy also scored significantly lower on the Freedom

from Distractibility Index than subjects with a right sided focus. This likely reflects the

fact that the heaviest loadings on the index came on two Verbal Scale subtests (i.e., Digit

Span and Arithmetic).

No significant difference in WCST Perseverative Error score was found between

subjects with left and right temporal lobe epilepsy. This is not surprising in light of the

somewhat contradictory reports in the literature. Hermann, Wyler, and Richey (1988)

reported the highest rate of perseverative errors in patients with non-dominant temporal

lobe epilepsy, while Strauss et al. (1993) reported that subjects with dominant temporal









lobe epilepsy and damage before 12 months of age showed the greatest level of

impairment.

Virtually all subjects performed within normal limits on the Facial Recognition

Test, with only two subjects scoring within the impaired range. As discussed below, this

ceiling-effect limited the study's potential to demonstrate correlations.


"Early" versus "Late" Seizure Onset

Subjects with an onset of chronic seizures before age five obtained significantly

lower VIQ, PIQ, and consequently FSIQ scores, than subjects with a later seizure onset.

Subjects in the "early" group had also completed significantly fewer years of education

than subjects in the "late" group. There was no significant difference between groups in

terms of performance on the WAIS-R Freedom from Distractibility index, WCST

Perseverative Error, or Facial Recognition Test scores.



Morphometric Data


Brain Size

On all measures of brain size (e.g., brain length, total intracranial area, total area of

cortex), males were found to have significantly larger brains than females. Similar sex

differences in brain size have been frequently reported, for examples see Johnson et al.

(1994) or Rauch and Jinkins (1994). Although the magnitude of the difference was not

nearly as large, the results suggested that subjects with an "early" onset of seizures had

slightly shorter and smaller brains than subjects in the "late" group. This could reflect a

subtle morphologic effect of seizure activity beginning at an earlier point in the course of









development, or an effect of the presumably longer duration of epilepsy and history of

treatment with antiepileptic drugs. It is important to note that the magnitude of the effect

was slight, with grouping based on age at seizure onset accounting for only 6% of the

variance in brain length.



Area of the Corpus Callosum

In this study, the average total midsagittal area of the corpus callosum was 620-

mm2 (SD = 96). Strauss et al. (1994) reported an average midsagittal area of the corpus

callosum for their epilepsy subjects of634-mm2 (SD = 119), which they stated was not

significantly different from the average area for their 55 normal controls (M = 644-mm2,

SD = 93). The average total callosal area for subjects in this study did not differ

significantly in one-sample t-tests from the mean values reported by Strauss et al. (1994)

for their patient or control subjects. Johnson et al. (1994) summarized data from 26 MRI

and post-mortem studies of callosal morphology in normal subjects, and reported a grand

mean of midsagittal areas of 667-mm2, with a standard deviation for sample means of 81.

Although the mean for this study's subjects was significantly smaller than that average, it is

well within the broad range (545-mm2 to 890-mm2) of average callosal areas reported for

normals in the studies summarized by Johnson et al. (1994).

The area of each of the seven callosal regions was found to be significantly larger

in males than females, a finding consistent with previous reports in the literature (Johnson,

Bigler, Burr, & Blatter, 1994). However, after accounting for variance associated with

brain size using an approach consistent with recommendations made by Arndt et al.

(1991), differences between males and females in terms of callosal areas were no longer









significant. Parashos, Wilkinson, and Coffey (1995) reported that, beyond age and an

estimate of brain size, gender was not a significant predictor of midsagittal callosal area.

No between group differences were found for the area of any callosal region based the

laterality of subject's seizure focus, or with grouping based on age at seizure onset.

The fact that the callosum was not significantly larger in subjects with "early"

seizure onset raises questions about the nature of the impact seizure activity beginning

before age five has on callosal development. The lack of a difference suggests that it does

not result in an excess number of surviving callosal projections, unless, for some reason,

the number of surviving axons is poorly indexed by midsagittal area. It remains possible

that early seizures may result in the fixation of inappropriate callosal projections

potentially supplanting more functional connections. It is also possible that seizures

during the period of rapid callosal growth before age two identified by Witelson and Kigar

(1988) may not alter the number of axons, but rather, may alter the nature of connections

they establish.



Relationships Between Morphometric and Neuropsychological Measures


WAIS-R Measures


This study demonstrated a significant positive correlation between area of the

posterior callosal region corresponding to the splenium and both Verbal and Full Scale IQ

in a sample of 62 right-handed subjects with intractable epilepsy. The significant

correlation between splenial area and Full Scale IQ replicates the previously reported

finding of Strauss et al (1994). In fact, after excluding data from a potential outlier, a







88

correlation identical to that reported by Strauss et al. (1994) was found. This replication

was obtained despite a number of methodological differences, including differences in

subject selection criteria, MRI acquisition parameters (e.g., 8.5-mm versus 1.25 or 1.406-

mm slice thickness), technique used to subdivide the callosum, and available test data

(e.g., prorated IQ scores versus scores based on the complete WAIS-R administration).

However, unlike the work of Strauss et al. (1994), in this study the correlation between

splenial area and FSIQ was not significantly stronger than correlations between FSIQ and

area of the remaining callosal regions.



Results Related to Age at Seizure Onset

Consistent with the prediction that early seizure activity would affect the

relationship between callosal morphology and measured IQ, a significant difference was

found between the positive correlation of splenial area and FSIQ for subjects with seizure

onset after age five and the negative correlation for subjects with an earlier age at seizure

onset. The "early" and "late" groups did not differ in terms of total callosal area or area of

any of the callosum's regions. However, the "early" group did have significantly lower IQ

scores, and demonstrated significantly less variation in area of the splenium but not other

callosal regions. These differences may have influenced the finding, although both

truncation of range and restricted variability of scores would be expected to effect the

magnitude, but not direction, of the correlation.

No significant differences were found in the correlation of IQ scores and area of

callosal regions based on the laterality of subjects' temporal lobe seizure focus. Strauss et

al. (1994) called for replication of their study with a more homogeneous sample to address









potential effects related to the site and laterality of subjects' epileptogenic disturbance.

Limiting this study to subjects with temporal lobe epilepsy was a step toward addressing

the first concern, while knowledge of subjects' epileptogenic focus made it possible to test

and exclude an effect of laterality on the relationship between callosal morphology and

measured IQ. Unlike subjects with epileptogenic lesions in extratemporal areas which

could exert a direct effect on callosal morphology, the seizure focus of all subjects in this

study was in anterior temporal lobe, a region relatively free of callosal projections.



Gender Differences

Although the correlation between splenial area and measured IQ did not differ

depending upon subject's gender, a consistent sex significant difference involving anterior

callosal area was found. For females, larger area of the rostrum was significantly

correlated with higher Verbal, Performance, and Full Scale IQ scores. While not

significant, correlations for males were in the opposite direction, with larger rostral area

associated with lower measured IQs. Schultz and Staib (1995) have reported similar sex

differences with respect to the relationship between callosal area and measured IQ.

In a study of 23 females and 16 males, largely comprised of twin pairs, Schultz and

Staib (1995) reported that area of the splenium was positively correlated with Full and

Verbal Scale IQ in females, but negatively correlated in males. Although the region of

callosum involved in the sex difference is not the same, this study's finding and Schultz and

Staib's (1995) report both involve positive correlations between regions of callosum

interconnecting association cortex and IQ for females and negative correlations for males.

These findings are consistent with literature suggesting that cognitive functions are more









highly lateralized in males, making interhemispheric communication more critical for

females (Clarke & Zaidel, 1994; Witelson, 1989; Witelson & Nowakowski, 1991; Zaidel,

Aboitiz, Clarke, Kaiser, & Matteson, 1995).



Other Neuropsychological Measures

Predictions made about associations among anterior and posterior callosal areas

and performance on the Wisconsin Card Sorting and Facial Recognition Tests were not

supported. No significant correlation was found between WCST Perseverative Error

score and either anterior or posterior callosal areas in the total sample, or separately for

subjects with "early" or "late" seizure onset. Assuming a correlation equal to that reported

by Strauss et al. (1994), with WCST data available for 52 subjects, power was .73 to

detect a significant relationship between WCST perseverative error score and callosal

area.

Making the same assumption about the strength of the underlying association, with

data available for fewer subjects, power to detect a significant correlation between FRT

score and callosal area was lower. With power equal to .53, there was a 44% chance that

if a significant correlation did exist between callosal area and FRT performance this study

would have been unable to detect it. Inadequate power is not the only factor which might

account for the failure to find a significant association between callosal morphology and

FRT performance. Aside from the obvious possibility that area of the callosum is not a

good marker for interhemispheric processes involved in face matching, the truncated range

and lack of variability in FRT scores would have made it difficult it difficult to

demonstrate potentially significant associations if they had existed. The significant







91

negative correlation between anterior callosal area and FRT score for subjects with "early"

seizure onset should be interpreted with caution. The correlation was based on data from

a small number of subjects (n = 11), and appeared to be unduly influenced by data from

two subjects. Their presence in the data may have inflated the observed correlations by

artificially stretching the range of observed values.



"What Does the Corpus Callosum do for a Living?"

After reporting the empirical observation that aspects of callosal morphology are

significantly associated with performance on behavioral measures of cognitive functioning,

the question quickly comes to focus on the functional significance of interhemispheric

pathways coursing through the callosum. As Hoptman and Davidson (1994, p. 205)

posed the question, "What does the corpus callosum do for a living?"



Role in Attentional Regulation

Hoptman and Davidson (1994) reviewed evidence concerning the role of the

corpus callosum in attentional systems, including results demonstrating that split-brain

patients show deficits in selective, sustained, divided, and focused attention which may be

due to disruption of an interhemispheric mechanism responsible for maintaining bilateral

cerebral arousal (Hoptman & Davidson, 1994). Citing work by Levy, Wagner, and Luh

(1990), Posner, Walker, Friedrich, and Rafal (1984), and Heilman and Van Den Abell

(1979), Hoptman and Davidson (1994) suggested that the parietal lobes, and especially

the right parietal lobe, may be critical to processes involved with the meta-control and

allocation of attentional resources. Callosal projections from posterior cortex pass