Voxel-based morphometry of unilateral temporal lobe epilepsy reveals abnormalities in cerebral white matter

www.elsevier.com/locate/ynimg

NeuroImage 23 (2004) 167–174

Voxel-based morphometry of unilateral temporal lobe epilepsy reveals

abnormalities in cerebral white matter

Alan B. McMillan,a,* Bruce P. Hermann,b Sterling C. Johnson,c Russ R. Hansen,b

Michael Seidenberg,d and Mary E. Meyerande

aDepartment of Medical Physics, University of Wisconsin-Madison, Madison, WI 53706, USAbDepartment of Neurology, University of Wisconsin-Madison, Madison, WI 53706, USAcVA Medical Center and Department of Medicine, University of Wisconsin-Madison, Madison, WI 53706, USAdDepartment of Psychology, Chicago Medical School, North Chicago, IL 60064, USAeDepartment of Medical Physics, University of Wisconsin-Madison, Madison, WI 53706, USA

Received 16 January 2004; revised 22 April 2004; accepted 5 May 2004

Voxel-based morphometric (VBM) investigations of temporal lobe

epilepsy have focused on the presence and distribution of gray matter

abnormalities. VBM studies to date have identified the expected

abnormalities in hippocampus and extrahippocampal temporal lobe, as

well as more diffuse abnormalities in the thalamus, cerebellum, and

extratemporal neocortical areas. To date, there has not been a

comprehensive VBM investigation of cerebral white matter in nonle-

sional temporal lobe epilepsy. This study examined 25 lateralized

temporal lobe epilepsy patients (13 left, 12 right) and 62 healthy

controls in regard to both temporal and extratemporal lobe gray and

white matter. Consistent with prior reports, gray matter abnormalities

were evident in ipsilateral hippocampus and ipsilateral thalamus.

Temporal and extratemporal white matter was affected ipsilateral to

the side of seizure onset, in both left and right temporal lobe epilepsy

groups. These findings indicate that chronic temporal lobe epilepsy is

associated not only with abnormalities in gray matter, but also with

concomitant abnormalities in cerebral white matter regions that may

affect connectivity both within and between the cerebral hemispheres.

D 2004 Elsevier Inc. All rights reserved.

Keywords: Voxel-based morphometry; Temporal lobe epilepsy; White

matter; Gray matter

Introduction

The majority of traditional region-of-interest-based quantitative

volumetric magnetic resonance (MR) imaging studies in temporal

lobe epilepsy have focused on neural regions involved in the

genesis and propagation of seizures. Volumetric abnormalities

(atrophy) are evident in hippocampus (Jack et al., 1992; Quigg et

al., 1997; Tasch et al., 1999; Woermann et al., 1998), associated

1053-8119/$ - see front matter D 2004 Elsevier Inc. All rights reserved.

doi:10.1016/j.neuroimage.2004.05.002

* Corresponding author. Department of Medical Physics, University of

Wisconsin-Madison, 1530 Medical Sciences Center, 1300 University

Avenue, Madison, WI 53706. Fax: +1-608-265-9840.

E-mail address: [email protected] (A.B. McMillan).

Available online on ScienceDirect (www.sciencedirect.com.)

mesial temporal lobe structures including amygdala (Kalviainen et

al., 1997; Martin et al., 1999), fornix (Kuzniecky et al., 1999;

Martin et al., 1999), and entorhinal cortex (Bernasconi et al.,

1999); as well as thalamus and basal ganglia (DeCarli et al.,

1998). In addition, atrophy has been reported in extrahippocampal

temporal lobe regions (Moran et al., 2001) and extratemporal areas

such as the cerebellum (Bohnen et al., 1998; Lawson et al.,

2000a,b; Sandok et al., 2000).

Considerably fewer quantitative MR studies of temporal lobe

epilepsy have examined whole brain volumes or volumes of

extratemporal gray or white matter, but the findings to date suggest

that abnormalities in brain structure extend well outside the

neuronal networks responsible for seizure generation and propa-

gation. Sisodiya et al. (1997) described widespread occult struc-

tural abnormalities occurring in visually normal appearing MRIs in

27 patients with hippocampal sclerosis. Marsh et al. (1997)

reported significant bilateral volumetric reductions in frontoparietal

regions in 14 males with temporal lobe epilepsy. Lee et al. (1998)

reported reduced whole brain volume in 27 patients with temporal

lobe epilepsy, and Theodore et al. (2003) recently described

reduced whole brain volume in patients with temporal lobe

epilepsy with a history of complex febrile convulsions.

Comparing patients with temporal lobe epilepsy (n = 58) to

healthy controls (n = 62), we recently reported that significant

volumetric reductions were particularly evident in cerebral white

matter, both ipsilateral and contralateral to the side of temporal

lobe seizure onset (Hermann et al., 2003a). Closer examination of

the corpus callosum in patients with chronic temporal lobe epilepsy

revealed significant volumetric reduction of this major white matter

tract (Hermann et al., 2003b), as well as lower diffusion anisotropy

and higher diffusivity in directions perpendicular to the axons on

DTI (Arfanakis et al., 2002). However, much remains to be

clarified regarding the nature and distribution of abnormalities

suggested by region-of-interest-based approaches. For instance,

the distribution of white matter abnormality within and between

lobar regions of interest remains unclear. In addition, many specific

and important gray matter structures (such as the thalamus) are not

routinely quantified in standard lobar-based segmentation pro-

Table 1

Demographic and clinical information of study groups

Group Mean SD

Age (years) Control 32.42 12.181

Epilepsy 32.23 11.198

Years of education Control 13.68 2.418

Epilepsy 12.96 2.289

IQ Control 107.67 13.812

Epilepsy 96.42 15.256

Onset age (years) Epilepsy 11.95 9.072

Duration (years) Epilepsy 19.09 12.341

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174168

grams, and potential differences in the patterns of white or gray

matter abnormality remain to be determined.

Voxel-based morphometry (VBM) is a technique used to

examine regional morphological differences in gray or white matter

between groups. Methods of VBM have been described that

include an automated approach to the distribution and localization

of whole-brain morphometric abnormalities that are less restricted

to the limitations associated with traditional region of interest

approaches (Good et al., 2001). To date, VBM studies of temporal

lobe epilepsy have focused on gray matter abnormalities (Keller et

al., 2002a,b; Woermann et al., 1999), and no investigation has

examined the presence or distribution of abnormalities in white

matter using VBM.

The purpose of this investigation is to comprehensively

characterize the distribution of abnormalities in gray and white

matter in patients with unilateral temporal lobe epilepsy. As will

be demonstrated, abnormalities in cerebral white matter in uni-

lateral temporal lobe epilepsy were significant, affected temporal

and extratemporal regions ipsilateral to the side of seizure onset,

and were equal in magnitude to abnormalities detected in gray

matter.

Methods

Subjects

Subjects were patients with temporal lobe epilepsy (n = 25, 13

unilateral left TLE, 12 unilateral right TLE) and healthy controls

(n = 62). Selection criteria for epilepsy patients included the

following: (a) chronological age from 14 to 60 years, (b) complex

partial seizures of unilateral temporal lobe origin demonstrated by

ictal EEG monitoring of spontaneous seizures, (c) absence of

MRI abnormalities other than atrophy on clinical reading, and (d)

no other neurological disorder. The majority of these patients

were candidates for anterior temporal lobectomy and as such

underwent a series of procedures including FDG-PET, Wada Test,

neuropsychological assessment, and extensive EEG monitoring of

spontaneous seizures with scalp or more invasive (e.g., subdural

strip or grid electrodes, depending on the details of the case as

decided by a multidisciplinary team). Patients with bilateral

independent left and right temporal lobe seizure onset were

excluded from the study. The subjects investigated here demon-

strated consistent unilateral temporal lobe onset of their typical

complex partial seizures.

Selection criteria for healthy controls included the following:

(a) chronological age from 14 to 60, (b) either a friend or family

member of the patient, (c) no current substance abuse, medical,

or acute psychiatric condition that could affect cognitive func-

tioning, and (d) no history of loss of consciousness >5 min or

developmental learning disorder. Table 1 provides sociodemo-

graphic and clinical features of the subjects. As can be seen, the

groups were equivalent in age and education, while the epilepsy

patients had significantly lower Full Scale IQ. The epilepsy

patients suffered from chronic epilepsy (mean duration = 19.1

years) of childhood onset (mean onset age = 11.9 years). The

left and right temporal lobe groups were not significantly

different in chronological age, duration of epilepsy, or Full Scale

IQ. The left temporal lobe group had significantly (P = 0.02)

less formal education than the right temporal lobe group (11.9

versus 14.2 years).

Image acquisition

Images were obtained on a 1.5-T GE Signa MR scanner. For

each subject, a T1-weighted, three-dimensional SPGR image was

acquired with the following parameters: TE = 5, TR = 24, NEX =

2, flip angle = 40j, slice plane = coronal, matrix size = 256 � 192,

FOV = 26 cm, slice thickness = 1.5 mm.

Voxel-based morphometry

Analysis was performed on a workstation running MATLAB

6.5 (The Mathworks, Inc., Natick, MA) and the statistical para-

metric mapping software SPM2 (Wellcome Department of Cogni-

tive Neurology, London, UK). The methodology used closely

parallels that of Good et al. (2001). Before the creation of a

study-specific template and morphometric analysis, each image

was visually inspected to ensure that its orientation was compatible

with SPM2, its origin centered on the anterior commissure, and

free of image artifacts. This was accomplished using the MRIcro

software package (Rorden and Brett, 2000).

Template creation

To allow for group comparison, voxel-based morphometry

registers each MR image to a standard template (spatial normaliza-

tion) before the image is automatically segmented into gray and

white matter components (Ashburner and Friston, 2000). While it

has been noted that the use of templates derived from the study

population has been shown to have insignificant consequences on

the quality of spatial normalization (Salmond et al., 2002) when

compared to the templates included with the SPM software, VBM

studies have used templates created from the study population or a

subset thereof to improve the quality of the segmentation step (e.g.,

Good et al., 2001; Karas et al., 2003; Rusch et al., 2003). Thus,

study-specific templates for whole brain volumes, gray matter, and

white matter were created. To accomplish this, images were first

segmented and spatial normalization parameters were calculated to

best match the segmented gray matter image to the SPM gray matter

template image. The whole brain image was then spatially normal-

ized using these calculated parameters and automatically segmented

into gray matter, white matter, and CSF. Note that the templates

were spatially normalized only to the default gray matter template

so that the resultant template images are complementary between

gray matter, white matter, and CSF compartments. Normalizing

each image to the respective gray and white matter template image

is performed in subsequent processing of the images, but at this

stage would have likely resulted in non-overlapping spatial nor-

malization between respective gray and white matter images. The

respective template images were formed from the average voxel

Table 2

Locations of local maximum of volume decreases in gray matter VBM

analysis significant at P < 0.05, corrected for multiple comparisons

Location Size

(voxels)

t P (FDR

corrected)

Left TLE �33 �3 �34 12,140 5.7 0.003

GM < controls �12 �35 9 1260 4.82 0.004

�22 �32 �20 105 3.8 0.019

27 50 �3 339 3.75 0.022

�28 52 �1 76 3.65 0.027

�29 55 �2 1 3.4 0.047

�22 �27 �21 1 3.39 0.048

Right TLE 41 2 �30 16,181 5.47 0.001

GM < controls 3 �24 35 262 4.06 0.008

5 37 9 475 3.83 0.014

43 �25 11 171 3.73 0.019

11 6 12 14 3.67 0.021

28 �35 13 141 3.65 0.022

11 2 14 2 3.64 0.023

11 8 10 2 3.5 0.032

57 �18 �10 2 3.45 0.036

10 7 8 5 3.42 0.039

57 �14 �12 3 3.41 0.039

55 �12 �14 1 3.34 0.046

4 �38 31 1 3.32 0.048

�10 �42 �55 1 3.31 0.049

8 4 8 1 3.31 0.049

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174 169

intensity for each segmented image for all subjects in the study and

smoothed with an 8-mm full width at half maximum (FWHM)

Gaussian kernel.

Image preprocessing

Similar to the template creation step, the images were spatially

normalized to the same coordinate system before group compari-

son. Each image was first segmented and spatial normalization

parameters were calculated for the segmented gray matter image

from the created gray matter template. Because accurate spatial

normalization is crucial for VBM, spatial normalization parameters

for the segmented white matter image were calculated from the

created white matter template. Each whole brain image was then

spatially normalized for each set of normalization parameters

obtained from the previous step and automatically segmented into

gray and white matter components, where the respective images

normalized to the specific template were used in analysis (i.e.,

gray-matter-template-normalized images were used for gray matter

analysis and white-matter-template-normalized images were used

for white matter analysis). To account for the different spatial

normalization parameters applied to each image, as individual

voxels were either shrunk or stretched, the voxel intensities were

multiplied by the Jacobian determinates from the spatial normal-

ization parameters (Ashburner and Friston, 2000). In previous

VBM methods, this process has been named modulation (Good

et al., 2001). The resulting gray and white matter images were

smoothed with a 12-mm FWHM Gaussian kernel.

Statistical analysis

To eliminate voxels outside of the volume of interest (i.e., gray

or white matter), the images were masked before analysis so that

voxels of noninterest (e.g., regions corresponding to white matter

and CSF in a gray matter mask) were not included in statistical

calculations. Masks were created from the study template images. A

gray matter mask was created where voxels in the complementary

image compartments (white matter and CSF) were removed from

the mask. Nonzero voxels from the white matter and CSF template

images with intensity greater than or equal to the mean intensity

plus one standard deviation from the respective white matter and

CSF templates were then removed from the gray matter template

image. Upon smoothing with a small diameter Gaussian kernel, the

statistical comparisons were confined to an area more inclusive of

the gray matter volume. A white matter mask was created and

applied in a similar manner. The mask images are shown in Fig. 1.

To compare groups, analysis of covariance (ANCOVA) was

performed using the total volume of each segmented image after

modulation, with age and gender as confounding covariates for the

Fig. 1. Example of template image used to mask input images. The gray matter m

overlap exists between gray and white matter masks, as the purpose is to restrict th

respective gray and white matter analyses. Total volume of each

segmented and modulated image before smoothing was used to

investigate regional volume changes beyond global gray or white

matter changes. Both left and right temporal lobe epilepsy groups

were incorporated as separate groups into the same statistical model

because it was expected that differences in temporal lobe epilepsy

laterality added more information to the statistical model. Using

SPM2, t-statistic maps were created for each voxel in the standard

atlas space to reflect differences in gray and white matter for the

lateralized (left and right) groups. Resultant t-statistic maps were

thresholded at aP-value of <0.05 corrected for multiple comparisons

using the False Discovery Rate approach (Genovese et al., 2002).

Results

Gray Matter

Table 2 details the regions of gray matter volume decrease that

were apparent at P < 0.05, corrected for multiple comparisons

ask is shown in green and the white matter mask is shown in red. Note that

e respective analysis to regions more representative of gray or white matter.

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174170

across the entire search volume. Cluster size is included for reader

convenience, but all inferences are drawn from voxelwise tests.

Additionally, results will be discussed in relation to their voxelwise

correspondence with anatomical structures. While presented in

tables for completeness, results indicating volume differences of

only a few voxels are difficult to interpret and will not be discussed

in detail. Fig. 2 depicts areas of decreased gray and white matter

volume in the left and right temporal lobe groups. As can be seen

in Fig. 3, left temporal lobe epilepsy patients exhibited significant

abnormalities relative to controls in the thalamus (both ipsilateral

and contralateral to side of seizure onset) and ipsilateral hippo-

campus. The left temporal lobe epilepsy group did not exhibit

significant regions of increased gray matter volume with respect to

the controls. The right temporal lobe epilepsy group exhibited

decreased gray matter volume in the ipsilateral thalamus and

abnormalities near the ipsilateral hippocampus. The right temporal

lobe epilepsy group also did not exhibit significant regions of

increased gray matter volume in comparison to the controls.

White matter

Table 3 details the white matter volume decreases for left and

right TLE groups apparent at P < 0.05, corrected for multiple

comparisons across the entire search volume. In addition to Fig. 2,

Fig. 4 provides a detailed depiction of decreased white matter

volume in the left and right TLE groups. As can be seen, temporal

lobe epilepsy patients exhibited a marked white matter volume

Fig. 2. VBM results at P < 0.05, corrected for multiple comparisons. Top left, regi

gray matter volume reduction for right TLE group; bottom left, regions of white

matter volume reduction for right TLE group.

decrease predominantly focused in ipsilateral temporal pole white

matter. This volume loss extends extratemporally to bilateral

prefrontal white matter in the left temporal lobe epilepsy group,

affecting voxels in the corpus callosum of both left and right TLE

groups, and the fornix of the right TLE group.

The left temporal lobe epilepsy group did not exhibit areas of

increased white matter volume compared to the controls. A

significant amount of white matter increase was found in the right

TLE group in the right hemisphere near the parietal–occipital

fissure. However, the voxels in this region are predominantly gray

matter, likely due to misregistration of tissue during the automated

segmentation, and therefore discounted from further inference.

Discussion

This report confirms and extends previous findings concerning

structural brain abnormalities observed in patients with unilateral

temporal lobe epilepsy. In addition, an extensive degree of tem-

poral lobe white matter abnormality is demonstrated. These find-

ings, their implications, potential significance, and the limitations

of this investigation are reviewed in the material to follow.

Gray matter abnormalities

Overall, the VBM gray matter findings observed in this study

complement previously reported abnormalities in patients with

ons of gray matter volume reduction for left TLE group; top right, regions of

matter volume reduction for left TLE group; bottom right, regions of white

Table 3

Locations of local maximum of volume decreases in white matter VBM

analysis significant at P < 0.05, corrected for multiple comparisons

Location Size

(voxels)

t P (FDR

corrected)

Left TLE �9 �25 9 8739 5.27 0.01

WM < controls 9 �14 5 724 3.94 0.027

�3 �24 17 2 3.7 0.041

46 �48 �61 1 3.61 0.048

Right TLE 11 �25 10 7957 5.63 0.003

WM < controls 54 �8 �16 41 3.75 0.039

32 �20 �13 8 3.72 0.042

33 �17 �16 10 3.69 0.044

24 �19 �8 1 3.69 0.044

25 �16 �8 1 3.67 0.047

26 �18 �8 1 3.67 0.047

32 �23 �10 1 3.66 0.048

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174 171

chronic temporal lobe epilepsy using either region-of-interest-

based quantitative MRI or VBM. Prior VBM investigations of

gray matter in temporal lobe epilepsy have shown slightly different

results, possibly due to variations in methodology used, mainly in

that other VBM methods in TLE have not included a modulation

step (Keller et al., 2002a,b; Woermann et al., 1999). In methods

including a modulation step, the statistical output represents a

region of significant volume change. Analysis of unmodulated data

yields regions of significant concentration differences (Ashburner

and Friston, 2000).

In a group comparison of 10 patients with left TLE and evident

hippocampal sclerosis versus controls, Woermann et al. (1999)

detected an increase in temporal lobe gray matter concentration

hypothetically implicated by disorganized gray matter and under-

lying white matter changes. Additionally, no significant differences

were detected in a comparison of a group of 10 patients with left

TLE and normal appearing MRI. Keller et al. (2002a) detected

gray matter concentration decreases in the ipsilateral hippocampus

of patients with measured left and right hippocampal atrophy with

dispersed effects throughout the cortex, most notably in the

prefrontal cortex. Regions of gray matter concentration increase

were reported in the cingulate gyrus, ipsilateral hippocampal

regions, and cerebellum; again, these increases were hypothesized

to be resultant of decreased distinction between gray and white

matter and more representative of white matter abnormalities.

Further studies demonstrated decreases in gray matter concentra-

tion in ipsilateral hippocampus, thalamus, and cerebellum, with

dispersed neocortical effects across frontal, parietal, and occipital

lobes in relation to left or right temporal lobe seizure origin (Keller

et al., 2002b).

These VBM reports of gray matter concentration decrease

complement previous region-of-interest-based quantitative volu-

metric MRI studies in which abnormalities have been reported in

the hippocampus (Jack et al., 1992; Quigg et al., 1997; Tasch et

al., 1999; Woermann et al., 1998), amygdala (Kalviainen et al.,

1997; Martin et al., 1999), fornix (Kuzniecky et al., 1999;

Martin et al., 1999), entorhinal cortex (Bernasconi et al.,

1999), thalamus (DeCarli et al., 1998; Natsume et al., 2003),

basal ganglia (Dreifuss et al., 2001), extrahippocampal temporal

lobe regions (Moran et al., 2001), and extratemporal areas such

as the cerebellum (Bohnen et al., 1998; Ney et al., 1994;

Sandok et al., 2000; Specht et al., 1997). Additionally, these

VBM reports acknowledge a possible white matter effect in

temporal lobe epilepsy, which will be discussed further in the

following section.

As demonstrated here, volume decreases in the hippocampus

are more defined for the left TLE group; however, voxels in

brain regions near the hippocampus of the right TLE group do

indicate a volume decrease. An analysis of the effect at an

uncorrected threshold (P < 0.001) indicated a larger underlying

effect throughout the spatial extent of the hippocampus similar to

the findings in the left TLE group. Furthermore, a large

hippocampal effect was not necessarily expected because the

epilepsy subjects were selected on the basis of EEG criteria (i.e.,

consistent unilateral temporal lobe onset of seizures), not neuro-

pathological criteria (i.e., hippocampal volume loss or sclerosis).

Recent VBM publications have also hypothesized that large

degrees of smoothing with respect to smaller structures like the

hippocampus may cause expected structural differences to remain

undetected by VBM (White et al., 2003). In addition, volume

increases in gray matter were not expected. The subjects chosen

for this study exhibited no lesions or MRI abnormalities resultant

from cortical dysplasias. Finally, especially striking was the

significant impact on the thalamus in both the left and right

TLE groups. As noted, decreased thalamic volume has been

reported previously in region-of-interest-based quantitative MRI

investigations (DeCarli et al., 1998; Natsume et al., 2003). The

robustness of the effect across imaging techniques and the

relative magnitude of the abnormality in the thalamus in this

investigation certainly suggests that further attention should be

devoted to the etiology and clinical consequences of comorbid

thalamic abnormality.

White matter abnormalities

An important finding of this investigation is the presence,

degree, and distribution of white matter abnormalities in unilateral

temporal lobe epilepsy (see Table 3; Figs. 2 and 4). White matter

abnormalities were marked in both the left and right temporal lobe

epilepsy groups. The distribution of these abnormalities included

not only the ipsilateral temporal lobe, but also bilaterally in the

frontal and parietal lobes in the left TLE group. Further, discrete

white matter tracks such as the corpus callosum were affected and

additionally the fornix in the right TLE group. These patterns of

white matter abnormality were seen in both the right and left

temporal lobe epilepsy groups.

Using traditional region-of-interest-based quantitative MRI, our

group has reported significant reductions in cerebral white matter in

TLE (Hermann et al., 2003a). Among patients with unilateral

temporal lobe onset, reductions in white matter volumes are detected

in temporal as well as extratemporal regions (frontal and parietal)

and are also evident contralaterally in the same regions (temporal,

frontal, and parietal), but the volume loss is significantly greater

ipsilateral to the side of seizure onset. In the future, we plan to

directly compare and contrast the abnormalities identified in the

same cohort of TLE patients using both VBM and traditional region-

of-interest quantitative MRI volumetrics.

Additionally, our group has applied diffusion tensor imaging

(DTI) to the corpus callosum in TLE. Increased diffusion per-

pendicular to the axons was apparent indicating compromised

axonal structure in the anterior corpus callosum, posterior corpus

callosum, and the external capsule (Arfanakis et al., 2002). Other

studies using DTI in temporal lobe epilepsy have indicated an

abnormality in temporal lobe white matter. For example, Rugg-

Fig. 3. VBM results at P < 0.05, corrected for multiple comparisons indicating gray matter volume decrease. (a) Left thalamus in left TLE group, (b) coronal

slice showing bilateral thalamic volume reduction and ipsilateral hippocampus in left TLE group, (c) hippocampus in left TLE group, (d), and (e) right thalamus

in right TLE group.

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174172

Gunn et al. (2001) reported decreased diffusion anisotropy in the

left temporal lobes of patients with left TLE. Similar results were

recently reported by Assaf et al. (2003), 12 TLE patients

exhibited decreased diffusion anisotropy in the hippocampal

region ipsilateral to side of seizure onset compared to the

contralateral side.

The findings reported here represent the first depiction of these

ipsilateral and contralateral abnormalities in gray and white matter

using modulated VBM and contrast the magnitude of white matter

abnormality to a similar degree of gray matter abnormality. Further

demonstrated in unilateral left and right temporal lobe epilepsy

Fig. 4. VBM results at P < 0.05, corrected for multiple comparisons indicating wh

group, (b) corpus callosum in left TLE group, (c) prefrontal white matter in left T

fornix in right TLE group.

groups is a distribution of white matter temporal and extratemporal

lobe abnormality observed ipsilateral to the side of seizure onset.

Of considerable interest is the degree to which significant changes

in both gray and white matter are evident among patients with

demonstrated localization-related temporal lobe epilepsy.

Limitations

The subjects investigated here were seeking care at a tertiary

university-based center and as such are not representative of the

broader population of patients with temporal lobe epilepsy. How-

ite matter volume decrease. (a) Left temporal pole white matter in left TLE

LE group, (d) right temporal pole white matter in right TLE group, and (e)

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174 173

ever, a recent population-based study of a heterogeneous group of

epilepsy patients also found substantial presence of both focal and

generalized neocortical atrophy among chronic epilepsy patients

(Liu et al., 2003).

While the automated approach of voxel-based morphometry

has distinct advantages over traditional region-of-interest-based

methods, it is not without limitations due to imperfect spatial

normalization, segmentation, and smoothing of the source images

(Good et al., 2001). Systematic misclassification of structures can

cause volume changes to appear in regions where none truly

exists, such as gray matter changes in brain regions that should

be white matter. This problem is potentially worsened in regions

without distinct gray–white matter boundaries (Ashburner and

Friston, 2000; Keller et al., 2002b) and poor boundaries due to

pathological reasons as previously described. Here, the search

volume for the VBM results was confined to an area more

inclusive of the respective gray or white matter. By eliminating

voxels outside of the tissue type of interest, the results are confined

to areas with greater probabilities of being correctly classified.

Even with a modest sample size, we present gray and white matter

results with findings that are well complemented between the left

and right TLE groups and previous volumetric findings. It is likely

that inclusion of larger numbers of patients would provide in-

creased statistical power resulting in the identification of additional

areas of abnormality.

Implications

The identified changes in gray and especially white matter

affected multiple brain regions and could be expected to adversely

affect cognitive status. Indeed, neuropsychological investigations

of patients with chronic temporal lobe epilepsy have demonstrated

diffuse cognitive dysfunction, the etiology of which has been

difficult to determine (Hermann et al., 1997). The clinical signif-

icance of identified extratemporal gray matter VBM abnormalities

has recently been demonstrated by Keller et al. (2003). They

reported that frontal lobe gray matter abnormalities were evident

in patients with temporal lobe epilepsy, and importantly that these

frontal lobe VBM abnormalities were associated with poorer

performance on measures of frontally dependent executive func-

tions. We have recently reported that whole brain white matter

brain volume abnormalities are also significantly associated with

neuropsychological dysfunction in TLE (Hermann et al., 2003a).

Thus, the presence of extratemporal lobe abnormalities in TLE

may contribute to the cumulative cognitive burden of the disorder.

Future research examining the relationship between VBM identi-

fied abnormalities in cerebral white matter and abnormalities in

specific mental functions and emotional–behavioral status could

provide further clarification of the etiologies underlying important

comorbidities of TLE.

The identified anomalies in cerebral white matter raise the

possibility of compromised connectivity within the temporal lobe

as well as within and between the left and right cerebral hemi-

spheres. The presence of significant disruption in cortical con-

nectivity secondary to abnormalities in cerebral white matter in a

disease traditionally considered a disorder of gray matter repre-

sents a new conceptualization of the etiology underlying some of

the cognitive pathology in epilepsy. Sophisticated functional

imaging research has led to the widely appreciated view that

many cognitive tasks dependent on the coordinated activity of

distributed neuronal network for efficient and successful perfor-

mance. Disruptions in cerebral connectivity could be expected to

adversely affect tasks dependent on such distributed neuronal

systems.

Finally, the pathophysiology and etiology of the identified

abnormalities in cerebral white matter remains to be determined.

Possible causes include the neurodevelopmental impact of epilepsy

and its treatment on white matter development and/or the potential

adverse effects of chronic epilepsy and its treatment with anti-

epilepsy medications on brain structure. Resolution of this issue

has considerable theoretical and treatment implications. Ultimately,

longitudinal studies employing multimodal neuroimaging techni-

ques will be necessary to provide data relevant to this issue.

Acknowledgments

This study was supported by NIH NS 2RO1-37738 and NIH

RO1 RR16591-02.

References

Arfanakis, K., Hermann, B.P., Rogers, B.P., Carew, J.D., Seidenberg, M.,

Meyerand, M.E., 2002. Diffusion tensor MRI in temporal lobe epilepsy.

Magn. Reson. Imaging 20, 511–519.

Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry—The

methods. NeuroImage 11, 805–821.

Assaf, B.A., Mohamed, F.B., Abou-Khaled, K.J., Williams, J.M., Yazeji,

M.S., Haselgrove, J., Faro, S.H., 2003. Diffusion tensor imaging of the

hippocampal formation in temporal lobe epilepsy. Am. J. Neuroradiol.

24, 1857–1862.

Bernasconi, N., Bernasconi, A., Andermann, F., Dubeau, F., Feindel, W.,

Reutens, D.C., 1999. Entorhinal cortex in temporal lobe epilepsy: a

quantitative MRI study. Neurology 52, 1870–1876.

Bohnen, N.I., O’Brien, T.J., Mullan, B.P., So, E.L., 1998. Cerebellar

changes in partial seizures: clinical correlation of quantitative SPECT

and MRI analysis. Epilepsia 39, 640–650.

DeCarli, C., Hatta, J., Fazilat, S., Fazilat, S., Gaillard, W.D., Theodore,

W.H., 1998. Extratemporal atrophy in patients with complex partial

seizures of left temporal origin. Ann. Neurol. 43, 41–45.

Dreifuss, S., Vingerhoets, J.G., Lazeyras, F., Andino, S.G., Spinelli, L.,

Delavelle, J., Seeck, M., 2001. Volumetric measurements of subcortical

nuclei in patents with temporal lobe epilepsy. Neurology 57, 1636–1641.

Genovese, C.R., Lazar, N.A., Nichols, T., 2002. Thresholding of statistical

maps in functional neuroimaging using the false discovery rate. Neuro-

Image 15 (4), 870–878.

Good, C.D., Johnsrude, I.S., Ashburner, J., Henson, R.N., Friston, K.J.,

Frackowiak, R.S., 2001. A voxel-based morphometric study of ageing

in 465 normal adult human brains. NeuroImage 14, 21–36.

Hermann, B.P., Seidenberg, M., Schoenfeld, J., Davies, K., 1997. Neuro-

psychological characteristics of the syndrome of mesial temporal lobe

epilepsy. Arch. Neurol. 54, 369–376.

Hermann, B.P., Seidenberg, M., Bell, B., Rutecki, P., Sheth, R., Sutula, T.,

Wendt, G., O’Leary, D., Magnotta, V., 2003a. Extratemporal quantita-

tive MRI volumetrics and neuropsychological function in temporal lobe

epilepsy. J. Int. Neuropsychol. Soc. 9, 353–362.

Hermann, B.P., Hansen, R., Seidenberg, M., Magnotta, V., O’Leary, D.,

2003b. Neurodevelopmental vulnerability of the corpus callosum to

childhood onset localization-related epilepsy. NeuroImage 18, 284–292.

Jack, C.R, Sharbrough, F.W., Cascino, G.D., Hirschorn, K.A., O’Brien, P.C.,

Marsh,W.R., 1992. Magnetic resonance image-based hippocampal volu-

metry: correlation with outcome after temporal lobectomy. Ann. Neurol.

31, 138–146.

Kalviainen, R., Salmenpera, T., Partanen, K., Vainio, P., Riekkinen, P.,

A.B. McMillan et al. / NeuroImage 23 (2004) 167–174174

Duncan, J.S., 1997. MRI volumetry and T2 relaxometry of the amyg-

dala in newly diagnosed and chronic temporal lobe epilepsy. Epilepsy

Res. 28, 39–50.

Karas, G.B., Burton, E.J., Rombouts, S.A., van Schijndel, R.A., O’Brien,

J.T., Scheltens, P., McKeith, I.G., Williams, D., Ballard, C., Barkhof,

F., 2003. A comprehensive study of gray matter loss in patients with

Alzheimer’s disease using optimized voxel-based morphometry. Neu-

roImage 18, 895–907.

Keller, S.S., Mackay, C.E., Barrick, T.R., Wieshmann, U.C., Howard, M.A.,

Roberts, N., 2002a. Voxel-based morphometric comparison of hippo-

campal and extrahippocampal abnormalities in patients with left and

right hippocampal atrophy. NeuroImage 16, 23–31.

Keller, S.S., Wieshmann, U.C., Mackay, C.E., Denby, C.E., Webb, J.,

Roberts, N., 2002b. Voxel based morphometry of grey matter abnor-

malities in patients with medically intractable temporal lobe epilepsy:

effects of side of seizure onset and epilepsy duration. J. Neurol., Neuro-

surg. Psychiatry 73, 648–656.

Keller, S.S., Baker, G., Weishmann, U.C., Sluming, V., Roberts, N., 2003.

Mapping volume atrophy in patients with temporal lobe epilepsy and

hippocampal atrophy using optimised voxel-based morphometry. Pro-

ceedings of the International Society for Magnetic Resonance in Med-

icine, Toronto, Canada, July 2003.

Kuzniecky, R., Bilir, E., Gilliam, F., Faught, E., Martin, R., Hugg, J., 1999.

Quantitative MRI in temporal lobe epilepsy: evidence for fornix atro-

phy. Neurology 53, 496–501.

Lawson, J.A., Vogrin, S., Bleasel, A.F., Cook, M.J., Bye, A.M., 2000a.

Cerebral and cerebellar volume reduction in children with temporal lobe

epilepsy. Epilepsia 41, 1456–1462.

Lawson, J.A., Vogrin, S., Bleasel, A.F., Cook, M.J., Burns, L., McAnally,

L., Pereira, J., Bye, A.M., 2000b. Predictors of hippocampal, cerebral,

and cerebellar volume reduction in childhood epilepsy. Epilepsia 41,

1540–1545.

Lee, J.W., Andermann, F., Dubeau, F., Bernasconi, A., MacDonald, D.,

Evans, A., Reutens, D.C., 1998. Morphometric analysis of the temporal

lobe in temporal lobe epilepsy. Epilepsia 39, 727–736.

Liu, R.S.N., Lemieux, L., Bell, G.S., Hammers, A., Sisodiya, S.M.,

Bartlett, P.A., Shorvon, S.D., Sander, J., Duncan, J.S., 2003. Progres-

sive neocortical damage in epilepsy. Ann. Neurol. 53, 312–324.

Marsh, L., Morrell, M.J., Shear, P.K., Sullivan, E.V., Freeman, H., Marie,

A., Lim, K.O., Pfefferbaum, A., 1997. Cortical and hippocampal vol-

ume deficits in temporal lobe epilepsy. Epilepsia 38, 576–587.

Martin, R., Hugg, J.W., Roth, D.L., Bilar, E., Gilliam, F.G., Faught, E.,

Kuzneicky, R., 1999. MRI extrahippocampal volumes and visual mem-

ory: correlations independent of MRI hippocampal volumes in temporal

lobe epilepsy. J. Int. Neuropsychol. Soc. 5, 540–548.

Moran, N.F., Lemieux, L., Kitchen, N.D., Fish, D.R., Shorvon, S.D., 2001.

Extrahippocampal temporal lobe atrophy in temporal lobe epilepsy and

mesial temporal sclerosis. Brain 124, 167–175.

Natsume, J., Bernasconi, N., Andermann, F., Bernasconi, A., 2003. MRI

volumetry of the thalamus in temporal, extratemporal, and idiopathic

generalized epilepsy. Neurology 60, 1296–1300.

Ney, G.C., Lantos, G., Barr, W.B., Schaul, N., 1994. Cerebellar atrophy in

patients with long-term phenytoin exposure and epilepsy. Arch. Neurol.

51, 767–771.

Quigg, M., Bertram, E.H., Jackson, T., Laws, E., 1997. Volumetric mag-

netic resonance imaging evidence of bilateral hippocampal atrophy in

mesial temporal lobe epilepsy. Epilepsia 38, 588–594.

Rorden, C., Brett, M., 2000. Stereotaxic display of brain lesions. Behav.

Neurol. 12, 191–200.

Rugg-Gunn, F.J., Eriksson, S.H., Symms, M.R., Barker, G.J., Duncan, J.S.,

2001. Diffusion tensor imaging of cryptogenic and acquired partial

epilepsies. Brain 124, 627–636.

Rusch, N., van Elst, L.T., Ludaescher, P., Wilke, M., Huppertz, H.J., Thiel,

T., Schmahl, C., Bohus, M., Lieb, K., Hesslinger, B., Hennig, J.,

Ebert, D., 2003. A voxel-based morphometric MRI study in female

patients with borderline personality disorder. NeuroImage 20, 385–392.

Salmond, C.H., Ashburner, J., Vargha-Khadem, F., Connelly, A., Gadian,

D.G., Friston, K.J., 2002. The precision of anatomical normalization in

the medial temporal lobe using spatial basis functions. NeuroImage 17,

507–512.

Sandok, E.K., O’Brien, T.J., Jack, C.R., So, E.L., 2000. Significance of

cerebellar atrophy in intractable temporal lobe epilepsy: a quantitative

MRI study. Epilepsia 41, 1315–1320.

Sisodiya, S.M., Moran, N., Free, S.L., Kitchen, N.D., Stevens, J.M.,

Harkness, W.F., Fish, D.R., Shorvon, S.D., 1997. Correlation of

widespread preoperative magnetic resonance imaging changes with

unsuccessful surgery for hippocampal sclerosis. Ann. Neurol. 41,

490–496.

Specht, U., May, T., Schulz, R., Rohde, M., Ebner, A., Schmidt, R.C.,

Schutz, M., Wolf, P., 1997. Cerebellar atrophy and prognosis after tem-

poral lobe resection. J. Neurol., Neurosurg. Psychiatry 62, 501–506.

Tasch, E., Cendes, F., Li, L.M., Dubeau, F., Andermann, F., Arnold, D.L.,

1999. Neuroimaging evidence of progressive neuronal loss and dys-

function in temporal lobe epilepsy. Ann. Neurol. 45, 568–576.

Theodore, W.H., DeCarli, C., Gaillard, W.D., 2003. Total cerebral volume

is reduced in patients with localization-related epilepsy and a history of

complex febrile seizures. Arch. Neurol. 60 (2), 250–252.

White, N.S., Alkire, M.T., Haier, R.J., 2003. A voxel-based morphometric

study of nondemented adults with Down syndrome. NeuroImage 20,

397–403.

Woermann, F.G., Barker, G.J., Birnie, K.D., Meencke, H.J., Duncan, J.S.,

1998. Regional changes in hippocampal T2 relaxation and volume: a

quantitative magnetic resonance imaging study of hippocampal sclero-

sis. J. Neurol., Neurosurg. Psychiatry 65, 656–664.

Woermann, F.G., Free, S.L., Koepp, M.J., Ashburner, J., Duncan, J.S.,

1999. Voxel-by-voxel comparison of automatically segmented cerebral

gray matter—A rater-independent comparison of structural MRI in

patients with epilepsy. NeuroImage 10, 373–384.