SEARCH

SEARCH BY CITATION

Keywords:

  • Temporal lobe epilepsy;
  • Mesial temporal sclerosis;
  • Functional magnetic resonance imaging;
  • Diffusion tensor imaging;
  • Probabilistic tractography

Summary

  1. Top of page
  2. Materials and Methods
  3. Results
  4. Discussion
  5. References

Purpose: Language functional magnetic resonance imaging (fMRI) is used to noninvasively assess hemispheric language specialization as part of the presurgical work-up in temporal lobe epilepsy (TLE). White matter asymmetries on diffusion tensor imaging (DTI) may be related to language specialization as shown in controls and TLE. To refine our understanding of the effect of epilepsy on the structure–function relationships, we focused on the arcuate fasciculus (ArcF) and the inferior occipitofrontal fasciculus (IOF) and tested the relationship between DTI- and fMRI-based lateralization indices in TLE.

Methods: fMRI with three language tasks and DTI were obtained in 20 patients (12 right and 8 left TLE). The ArcF, a major language-related tract, and the IOF were segmented bilaterally using probabilistic tractography to obtain fractional anisotropy (FA) lateralization indices. These were correlated with fMRI-based lateralization indices computed in the inferior frontal gyrus (Pearson's correlation coefficient).

Results: fMRI indices were left-lateralized in 16 patients and bilateral or right-lateralized in four. In the ArcF, FA was higher on the left than on the right side, reaching significance in right but not in left TLE. We found a positive correlation between ArcF anisotropy and fMRI-based lateralization indices in right TLE (p < 0.009), but not in left TLE patients. No correlation was observed for the IOF.

Conclusions: Right TLE patients with more left-lateralized functional activations also showed a leftward-lateralized arcuate fasciculus. The decoupling between the functional and structural indices of the ArcF underlines the complexity of the language network in left TLE patients.

In the presurgical work-up of patients with intractable temporal lobe epilepsy (TLE), functional magnetic resonance imaging (fMRI) is increasingly used to lateralize language function and to identify eloquent cortex to be spared during epilepsy surgery (Springer et al., 1999; Lehericy et al., 2000; Adcock et al., 2003; Sabbah et al., 2003). Several studies have shown a higher proportion of atypical language representation in TLE patients, especially in those with left-sided intractable seizures (Springer et al., 1999; Sabbah et al., 2003; Thivard et al., 2005; Sveller et al., 2006). This supports a reorganization of language circuits with greater involvement of the right hemisphere, which may represent an adaptive process following a brain insult early in life (Thivard et al., 2005). Hemispheric language dominance is associated with structural brain asymmetries in the size of regions of the cerebral cortex (Geschwind & Levitsky, 1968; Josse et al., 2006). Larger white matter volumes in the left hemisphere could provide a structural basis for greater intrahemispheric connectivity (Golestani et al., 2002; Josse et al., 2006). The role of white matter in hemispheric language specialization is further supported by the use of diffusion tensor imaging (DTI), (Pierpaoli et al., 1996; Le Bihan et al., 2001). Anisotropy measures such as fractional anisotropy (FA) provided by DTI, could reflect the organization of white matter fiber tracts (Beaulieu, 2002). Axonal white matter fibers connecting cortical regions involved in speech, language, and reading may demonstrate asymmetry in their microstructural organization, the degree of axon myelination, or in the integrity of axonal cell membranes (Golestani et al., 2002; Pujol et al., 2002). In line with this, a DTI study revealed a leftwards FA asymmetry of the arcuate fasciculus (ArcF) (Buchel et al., 2004), a white matter tract connecting dorsally the anterior and posterior language poles. Besides the ArcF, the subinsular white matter, a region that contains connections between anterior and posterior cortical language regions, has also been the focus of interest. Leftward structural asymmetries have been reported in the subinsular region in right-handed volunteers (Cao et al., 2003). By segmenting the subinsular area using tractography, it has been shown that two tracts, namely the uncinate fasciculus and the inferior occipitofrontal fasciculus (IOF), also display a leftward structural asymmetry in right-handers (Rodrigo et al., 2007). The IOF connects Wernicke's area and Broca's area ventrally and is likely implicated in the language network (Duffau et al., 2005). Diffusion MR tractography has shown that the IOF also has stronger connections in the left hemisphere in right-handed volunteers (Rodrigo et al., 2007a). Overall, interhemispheric white matter asymmetries of tracts connecting Wernicke's area to Broca's area, namely the ArcF and IOF, may be important anatomical substrates for hemispheric language specialization.

Imaging the white matter tracts using diffusion tractography can be combined with fMRI using blood-oxygenated level dependent signal changes, to noninvasively explore the in vivo structure–function relationship (Werring et al., 1998; Powell et al., 2006). Recently, a study combining DTI and fMRI reported that right-handed subjects had more extensive frontotemporal connections between Broca's area and Wernicke's area on the left than on the right side (Powell et al., 2006). Interestingly, this study found a correlation between measures of structure and function, with subjects with more lateralized fMRI activation having a more lateralized anisotropy of their frontotemporal connections. These correlations were obtained in a group of strongly right-handed patients, raising the question of the effect of atypical language dominance upon the structure–function relationship. Atypical language representation, defined as right hemisphere dominance or bilateral representation, is more frequent in left-handers, and in patients with left hemispheric lesions (Staudt et al., 2001; Hertz-Pannier et al., 2002). Powell et al. were the first to combine fMRI and diffusion tractography, in a study involving a group of 14 patients with TLE (Powell et al., 2007). They showed that controls and right TLE patients had a left-lateralized pattern of both language-related activations and the associated structural connections. Left TLE patients showed more symmetrical language activations, along with reduced left hemisphere and increased right hemisphere structural connections. If replicated, these findings would have important clinical implications. Indeed, the study of white matter structure using DTI could become a useful adjunct to fMRI in noninvasively predicting postoperative language deficits in TLE patients (Sabsevitz et al., 2003). Recently, another group found unexpected results in a group of 7 right-handed and 13 left-handed volunteers (Vernooij et al., 2007). Replicating the correlation between structural measurements in the ArcF and fMRI language lateralization indices in right-handers, this group found no correlation in left-handed volunteers. They showed an overall leftward asymmetry of the ArcF, regardless of functional hemispheric language lateralization. In light of these conflicting data, we combined fMRI and diffusion probabilistic tractography to search for additional evidence supporting the structure–function relationship in 20 TLE patients. We specifically focused our study on the ArcF and IOF, since these are both structurally asymmetric (Buchel et al., 2004; Rodrigo et al., 2007), connecting Broca's area to Wernicke's area, and therefore potential candidates to reflect functional hemispheric language lateralization.

Materials and Methods

  1. Top of page
  2. Materials and Methods
  3. Results
  4. Discussion
  5. References

Patients

We studied 20 native French-speaking right-handed subjects with intractable TLE. The population consisted of 12 patients with right TLE (hippocampal sclerosis, n = 9; cryptogenic type, n = 3), aged from 16 to 46 years (median, 33 years) and 8 patients with left TLE (hippocampal sclerosis, n = 6; cryptogenic type, n = 2), aged from 14 to 44 years (median, 31 years). All patients were referred for a presurgical evaluation of refractory epilepsy of temporal lobe origin. The temporal origin of epilepsy was established on the basis of noninvasive data, including medical history, early event, standard EEG and video-EEG recording, and a brain MRI. The full scale intelligence quotient (IQ), the verbal IQ, the performance IQ (Wechsler, 1997) and the handedness according to the Edinburgh Hand Preference Inventory are given in Table 1. No significant differences between left and right TLE patients were found for the demographic data or any of the neuropsychological tests (Mann–Whitney test). The duration of epilepsy was not significantly different between right (median, 21 years; range 5–36 years) and left (median, 14 years; range 9–37 years) TLE patients (see Table 1). The age at seizure onset was not significantly different between right (median, 8 years; range 1–26 years) and left (median, 16 years; range 4–27 years) TLE patients. The study was approved by the local Ethical Committee and all patients gave written informed consent.

Table 1.  Clinical data
PatientsAgeGenderSide of TLEEpilepsy duration (years)HandednessFIQVIQPIQ
  1. TLE, temporal lobe epilepsy; L, left; R, right; FIQ, full scale intelligence quotient (IQ); VIQ, verbal IQ; PIQ, performance IQ.

 116MR15Right 9980135 
 238FR20Right 807985
 331MR 5Right 808085
 438MR26Right 808974
 518FR12Right 8975112 
 638MR36Right10997126
 724FR22Right 778868
 835FR30Right 706086
 928MR14Right 97100 96
1030MR10Right 969696
1138MR34Right 707174
1246MR36Right102103 99
1344FL37Right100100 99
1435ML15Right 746391
1523ML 9Right114115 110 
1630ML12Right 837498
1722ML17Right10389122
1814FL10Right 878888
1932FL14Right10097112 
2041ML14Right 807196

MRI acquisition

All patients were investigated on a 1.5 T Signa MR scanner (Excite 1.5T, General Electrics Healthcare, Milwaukee, WI, U.S.A.). To prevent head movements during the procedure, the patient's head was restrained on either side with foam pads. The brain MRI included three functional runs, an anatomical and a diffusion-weighted sequence for a total acquisition time of 28 min.

Patients performed three language tasks during a single fMRI session. Audio stimuli were delivered via a 30-dB noise attenuation audio headset (Resonance Technology, Inc., Northridge, CA, U.S.A.). The language tasks consisted of a blocked experimental design beginning with a 30-s rest period where patients were asked to fix their attention on MRI gradient noise, alternating with a 30-s activation period. For the verbal fluency task, patients had to generate covertly as many words as possible beginning with a given letter (L, B or R), which was provided at the beginning of each activation period. For the semantic fluency task, patients had to generate covertly as many words as possible from a given semantic category (colors, fruits, and animals) (Yetkin et al., 1995; Gaillard et al., 2004). The name of a different category was presented aurally at the beginning of each activation period. For the semantic association task, patients heard a single verb every 3 s during activation periods and were asked to generate covertly a semantically appropriate noun for each verb. The three fMRI acquisitions included 24 axial gradient-echo echoplanar (EPI) images (repetition time/echo time/flip angle: 5000/60 ms/90°, 5-mm slice thickness, no gap, in-plane resolution: 3.75 mm × 3.75 mm, acquisition time: 3 min 50 s) for each task and an inversion recovery three-dimensional T1-weighted fast-spoiled gradient recalled acquisition for anatomic localization (repetition time/echo time/flip angle: 10/2 ms/15°, 1.2 mm slice thickness, no gap, in-plane resolution: 0.93 mm × 0.93 mm, acquisition time: 6 min 14 s).

Diffusion-weighted data were acquired by using 50 axial single-shot spin echo EPI images (repetition time/echo time/flip angle: 12600/81 ms/90°, 2.5 mm slice thickness, no gap, in-plane resolution: 2.5 mm × 2.5 mm, total acquisition time: 7 min) along 15 diffusion gradient directions (b = 1000 s/mm2) as well as one image without diffusion weighting (b = 0 s/mm2, called b0 image).

fMRI analysis

Image preprocessing and statistical analyses were carried out using the FMRI expert analysis tool (FEAT) version 3.3.7, part of FMRIB's software library (FSL, http://www.fmrib.ox.uk/fsl) (Smith et al., 2004). Each run was motion-corrected with the middle volume serving as the registration reference using motion correction FMRIB's linear image registration tool (MCFLIRT)) and nonbrain removal was done by using brain extraction tool (BET). Spatial (5 mm FWHM) and high-pass temporal (sigma = 60 s) filtering were used. Time-series statistical analysis was carried out with a general linear modeling approach (Worsley & Friston, 1995), using FMRIB's improved linear model (FILM) with local autocorrelation correction (Smith et al., 2004). Z-statistic images were thresholded using clusters determined by a Z threshold adjusted for each task (Fernandez et al., 2003) and a corrected cluster significance threshold of p = 0.05. Additionally, we conducted group analyses to compare the left and right TLE groups for each functional task. This was carried out by means of mixed effect analysis using FMRIB's local analysis of mixed effects (FLAME) (Beckmann et al., 2003); Z-statistic images were thresholded using clusters determined by Z > 2.3 and a corrected cluster significance threshold of p = 0.05 (Worsley & Friston, 1995).

Three volume of interest (VOI) were defined bilaterally on the standard Montreal Neurological Institute T1-weighted template (MNI 152 brain, http://www.bic.mni.mcgill.ca/software) starting from the regions proposed by the automatic anatomic labeling toolbox (Tzourio-Mazoyer et al., 2002): VOIF3 included the lateral aspect of the inferior frontal gyrus (Broca's area), VOIFr included the lateral aspect of the remaining prefrontal regions outside Broca's area and VOITmpP included the posterior part of the middle and superior temporal gyri and the inferior parietal lobule (Fernandez et al., 2003). These VOIs were realigned and resliced using FMRIB's linear registration tool (FLIRT) (Smith et al., 2004) from the MNI template to each patient's functional run. This was done by applying a single transformation from the MNI template to the patient's anatomical image and the latter to the corresponding EPI image. Each VOI was used to compute a weighted lateralization index (LI) (Fernandez et al., 2003):

  • image

where V is the set of activated voxels within the specified VOI, XL is the Z-value of left hemispheric voxels and XR is the Z-value of right hemispheric voxels (Z-statistic images from the first level analysis). In each VOI, the LIs were computed for each of the three tasks and then a global LI corresponding to the mean LI across the three tasks was computed. Functional LIs within VOIF3, VOIFr, and VOITmpP were compared between TLE groups for the global LI, by means of an unpaired Student's t-test (p < 0.05). Based on a study of reproducibility of presurgical language lateralization using fMRI, we chose the mean LI within VOIF3 to decide whether the language processes were left dominant, symmetric, or right dominant (Fernandez et al., 2003). A positive index within VOIF3 corresponded to a left hemispheric dominance (LI > 0.1) and a negative index to a right hemispheric dominance (LI < −0.1). Symmetric/bilateral activations corresponded to LIs from −0.1 to +0.1. Global functional LIs were compared to the results of the intracarotid amobarbital procedure (Wada test), which was available in five patients.

Diffusion imaging analysis

Using FMRIB's diffusion toolbox (FDT) (Behrens et al., 2003; Smith et al., 2004), Eddy current distortions were corrected by affine registration to the reference b0, a b0-derived brain mask was applied to each diffusion-weighted volume (Smith et al., 2004), diffusion tensors were fitted at each voxel, and FA and mean diffusivity (MD) maps were generated. After partial volume diffusion model fitting, probabilistic diffusion tractography generated a connectivity distribution to segment the ArcF and the IOF, bilaterally (Behrens et al., 2003). This was done using a two regions of interest (ROIs)-based approach (Fig. 1); these ROIs were first drawn on the MNI T1-weighted template, and then sent to each patient's diffusion image. For the ArcF (Fig. 1), we chose a coronal ROI perpendicular to this pathway at the level of the rolandic operculum (1.6 cm3) and an axial ROI adjacent to the ventricular trigone (1.6 cm3). For the IOF (Fig. 1), we chose a coronal ROI perpendicular to this pathway at the level of the ventricular trigone (2.3 cm3) and an axial ROI covering the external capsule (5.7 cm3) (Kier et al., 2004). After nonbrain removal using BET, the b0 image was coregistered to the MNI template using FLIRT with an affine transformation (Smith et al., 2004) for each patient. The tracking ROIs were then resliced to each patient's b0 image and used for the probabilistic diffusion tractography algorithm with the following options: two masks symmetric tracking mode, 5000 samples, curvature threshold set to 0.2, loopchecks and modified Euler streamlining, exclusion mask (cerebrospinal fluid and midline). The probabilistic diffusion tractography was considered successful when the resulting fascicles matched the known anatomical shape of ArcF and IOF (Dejerine, 1895; Kier et al., 2004). The fascicles were then thresholded to remove low connectivity probability pixels. This threshold was set to 20 samples and the resulting paths were binarized and used as a mask to extract FA and MD values. Finally, we computed diffusion LI as follows: [(L − R)/(L + R)]× 100 for MD and [(L − R) × 100] for FA, where L is the left value and R the right value of MD or FA. For the ArcF and IOF, diffusion parameters (FA and MD) were compared side to side within each TLE group. Diffusion parameters measured in homologous sides and diffusion LIs were also compared between TLE groups. These comparisons were done using Student's t-test (p < 0.05). Diffusion LIs of the ArcF and IOF were correlated with the duration of epilepsy, age at the onset of seizures, and neuropsychological tests using Pearson's product moment correlation coefficient.

image

Figure 1. Axial and coronal MNI 152 T1-weighted magnetic resonance imaging and the sets of the ROIs used for tractography purposes. For the ArcF tractography, an axial ROI adjacent to the ventricular trigone (A) and a coronal ROI perpendicular to this pathway at the level of the rolandic operculum (B) were chosen bilaterally. For the IOF tractography, a coronal ROI perpendicular to this pathway at the level of the ventricular trigone (C) and an ROI covering the external capsule (D) were chosen bilaterally.

Download figure to PowerPoint

Combined fMRI and diffusion MRI analysis

Functional global LIs measured in VOIF3, VOIFr, and VOITmpP were correlated with diffusion-based LIs for the ArcF and IOF using Pearson's product moment correlation coefficient. All statistical analyses and graphics were done using the R stats package (R Development Core Team, http://www.R-project.org).

Results

  1. Top of page
  2. Materials and Methods
  3. Results
  4. Discussion
  5. References

Functional MRI

The mixed effect fMRI group analysis showed significant activation in both hemispheres for each language task (see Table 2). In both groups, the fluency tasks were associated with strong activations in the left anterior language network (inferior frontal gyrus, rolandic operculum, insula) along with an activation of the left supplementary motor area. Right TLE patients also showed activations within the right inferior frontal gyrus and the insula. The semantic association task showed strong activations in both anterior and posterior language epicenters in left and right TLE patients. On direct comparison between right and left TLE groups, no brain regions demonstrated either significantly greater or lesser activation for any of the three tasks, at p < 0.05 corrected for multiple comparisons. Global functional LIs from VOIF3, VOIFr, and VOITmpP showed a trend toward left-lateralization in the right TLE patients (mean ± SD: VOIF3= 37 ± 19, VOIFr= 21 ± 21, VOITmpP= 27 ± 36) compared to left TLE patients (mean ± SD: VOIF3= 30 ± 45, VOIFr= 8 ± 19, VOITmpP= 24 ± 39), although the difference did not reach significance. Global functional LI from VOIF3 indicated left-sided language dominance in all but one patient with right TLE and in six patients with left TLE. Activations were bilateral in one patient of each TLE group, and right-sided in one left TLE patient. Wada test, obtained in two right TLE, and in three left TLE patients confirmed the hemispheric dominance as determined by fMRI.

Table 2.  Anatomical localization and max Z-score for the three functional tasks in each TLE patient group
Anatomical region according to fMRI taskRight TLE MNI coordinates of max Z-scoreMax Z-scoreLeft TLE MNI coordinates of max Z-scoreMax Z-score
xyzxyz
  1. Suppl., supplementary.

Verbal fluency task
 Left hemisphere
 Inferior frontal−4832104.12−4020−144.84
 Rolandic operculum−484162.88−581023.12
 Suppl. motor area−44503.97−424463.39
 Insula−362442.90−3818−123.89
 Superior temporal −546−43.05
 Right hemisphere
 Suppl. motor area66523.64 
 Thalamus−16−28142.9514−8162.75
Semantic fluency task
 Left hemisphere
 Inferior frontal−3822−63.61−488243.87
 Rolandic operculum−48802.87−50623.03
 Suppl. motor area210504.24012443.35
 Insula−46823.47−361064.04
 Thalamus−12−10145.32−20−1603.69
 Right hemisphere
 Suppl. motor area410504.25614502.34
 Thalamus4−1823.5210−1022.54
 Cerebellum14−62−244.44 
Semantic association task
 Left hemisphere
 Inferior frontal−4416−104.89−488225.09
 Rolandic operculum−561023.71−446163.08
 Suppl. motor area−102644.76−104643.75
 Insula−3416−84.87−4414−84.60
 Parietal inferior−58−50363.40 
 Thalamus−4−1005.21−12−884.13
 Superior temporal−52−36104.26−54−6−83.38
 Right hemisphere
 Inferior frontal3824−85.123824−84.83
 Rolandic operculum581403.085414−23.01
 Suppl. motor area412504.501410643.17
 Insula4222−105.353624−64.67
 Thalamus4−1445.3918−12123.33
 Superior temporal62−2845.0760−10−84.91
 Cerebellum32−56−344.6412−38−262.77

Diffusion MRI

Probabilistic tracking of the ArcF and IOF was successful for each side in all cases (see Figs. 2, 3 and 4), except in two left TLE patients where susceptibility artifacts led to distortions at the level of the temporal stem and prevented bilateral IOF tracking. In right TLE patients, left FA (mean ± SD: 0.49 ± 0.04) was significantly (p = 0.02) higher than right FA (mean ± SD: 0.44 ± 0.05) measured along the ArcF. This asymmetry showed a trend toward left-lateralization in the left TLE patients (left FA, mean ± SD: 0.53 ± 0.04; right FA, mean ± SD: 0.48 ± 0.06, p = 0.27). The MD values did not differ significantly between sides within either the right or the left TLE group.

image

Figure 2. Coronal (A), axial (B), and sagittal (C, D) T1-weighted magnetic resonance imaging slices (radiological convention) along with the tractography results of the ArcF (red to yellow coding low to high connectivity).

Download figure to PowerPoint

image

Figure 3. Axial (A, B), coronal (C), and sagittal (D) T1-weighted magnetic resonance imaging slices (radiological convention) with the tractography results of the IOF (red to yellow coding low to high connectivity).

Download figure to PowerPoint

image

Figure 4. Whole tractography results of the ArcF superimposed on perisylvian sagittal T1-weighted images for right (red) and left (blue) TLE patients (color code: bright color means maximum overlap of tracts between subjects and dark color means minimum overlap).

Download figure to PowerPoint

For the IOF in right TLE patients, left FA (mean ± SD: 0.49 ± 0.07) was higher than right FA (mean ± SD: 0.47 ± 0.04) although the difference did not reach significance (p = 0.33). This asymmetry was not significant in the left TLE patients (left FA, mean ± SD: 0.48 ± 0.05; right FA, mean ± SD: 0.49 ± 0.04; p = 0.59). The MD values did not differ significantly between sides within either the right or the left TLE group.

Overall, comparisons between the left and the right TLE patients indicated no significant differences for diffusion parameters measured in homologous sides or for diffusion LIs for the ArcF and IOF (see Fig. 5). FA diffusion LIs for the ArcF and IOF were not significantly correlated with age at the onset of seizures, duration of epilepsy, or neuropsychological data.

image

Figure 5. Diffusion LIs from the ArcF and the IOF. A positive diffusion LI (%, y-coordinate) indicated a left-greater-than-right FA and a negative diffusion LI a right-greater-than-left FA. Each box has a line at the first quartile, median, and third quartile values. The whiskers are lines extending from each end of the box to show the extent of the rest of the data. Outliers (○) are data with values beyond the ends of the whiskers.

Download figure to PowerPoint

Combined analysis of fMRI and diffusion LIs

For the ArcF, right TLE patients with greater left anisotropy tended to have greater left functional lateralization. There was a positive correlation between the global functional LIs from VOIF3 and the FA-derived diffusion LIs in right TLE patients (Pearson's correlation coefficient = 0.71, p < 0.009). Interestingly, this correlation was no longer observed in left TLE patients (Pearson's correlation coefficient =−0.26, N.S.). Figure 6 illustrates the relationship between VOIF3-functional and FA-derived diffusion LIs from the ArcF. No significant correlation was found between the global functional LIs from VOIFr or VOITmpP and the FA-derived diffusion LI in right or left TLE patients. MD-derived diffusion LIs were not correlated with any of the functional LIs for either the right TLE patients or the left TLE patients. For the IOF, no significant correlation was found between functional and diffusion LIs for either the right TLE patients or the left TLE patients.

image

Figure 6. LI of the FA of the ArcF (%, y-coordinate) and fMRI LI of the inferior frontal gyrus (F3) across the three tasks (%, x-coordinate) in right TLE patients (red dots) and left TLE patients (yellow dots). The line shows the linear regression between the diffusion and the fMRI laterality indices in the right temporal lobe epilepsy patients.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Materials and Methods
  3. Results
  4. Discussion
  5. References

The combination of fMRI, to identify cortical regions involved in a given function, and diffusion tractography, to visualize pathways connecting these regions, provides an opportunity to study the relationship between brain structure and function. As previously done for healthy controls (Powell et al., 2006; Vernooij et al., 2007), we combined these two imaging tools to search for relationships between functional hemispheric language lateralization and structural white matter asymmetries in TLE patients. We found a correlation between the ArcF structural lateralization and the leftward asymmetry in functional activation in right TLE patients, with patients with a more lateralized language function having a more lateralized FA measurements. This correlation was no longer observed in left TLE patients, suggesting a decoupling between structural- and functional-based lateralization indices in patients with epilepsy of left temporal origin.

The lateralization of language is a striking feature of human brain function. Left-sided hemispheric language lateralization has been shown, using fMRI, in up to 95% of right-handed controls (Springer et al., 1999). Recent DTI studies showed white matter structural asymmetries in tracts involved in the language network (Cao et al., 2003; Catani et al., 2005; Parker et al., 2005; Rodrigo et al., 2007). Several studies have reported higher anisotropy in the left than in the right ArcF in right-handed volunteers (Buchel et al., 2004; Rodrigo et al., 2007). These findings were reported in subjects whose language lateralization was not known, although it was presumably left-sided in most right-handed participants. Recently, in a series of 10 right-handed controls with confirmed left language representation on fMRI, Powell et al. found greater frontotemporal connectivity in the left hemisphere than in the right (Powell et al., 2006). The putative role of such structural asymmetries in the language network was further strengthened by the correlation between measures of structure and function, with right-handed controls with more lateralized fMRI language activation having a more highly lateralized anisotropy of their connections (Powell et al., 2006). The same group repeated the analysis in TLE patients (Powell et al., 2007). As for right-handed controls, they found that right TLE patients showed greater frontotemporal connectivity in the left hemisphere than in the right. Despite numerous methodological differences, we also found greater leftward anisotropy of the ArcF in right TLE patients, along with a correlation between functional and structural indices in this group of patients. We did, however, use a different approach. We were interested in measuring diffusion anisotropy in two particular bundles, the IOF and the ArcF, which are likely implicated in the language network. Powell et al., on the other hand, analyzed a larger volume of white matter using fMRI-active regions as starting points for fiber tracking (Powell et al., 2007). Our analysis thus complements that of Powell et al. by directly demonstrating that the ArcF likely participates in the structure–function correlation that they found in a large temporal area in right TLE patients (Powell et al., 2007).

In left TLE patients, we observed a loss of the significantly greater leftward anisotropy within the ArcF, along with the lack of correlation between the ArcF structural lateralization and the leftward asymmetry in functional activation. Atypical language representation is observed more frequently in TLE patients with a left-sided epileptic focus, suggesting that the side of the epileptogenic process plays a role in language lateralization (Adcock et al., 2003; Sabbah et al., 2003; Thivard et al., 2005). Accordingly, we found atypical fMRI language representation more frequently in left TLE than in right TLE patients. In left TLE patients, Powell et al. recently found a correlation between functional lateralization and lateralization of the mean FA of frontal connections (Powell et al., 2007). In that study, left TLE patients showed more symmetrical language activations than controls and right TLE patients, along with reduced left frontal and increased right frontal structural connections. However, in that study, anisotropy was measured in a relatively large “connectivity volume,” which provides a global measurement over the frontal white matter but does not allow the pathways that most contribute to this structure–function relationship to be identified. In our study, the relationship between brain structure and function that was observed in right TLE patients for the ArcF was not maintained in left TLE patients. This could simply be due to the large variability in fMRI and diffusion-based indices between patients and the relatively small sample of left TLE patients in our study. It could also indicate that the functional reorganization that presumably occurs as a result of chronic ictal discharge arising from the left temporal lobe is not associated with a shift from the left towards the right ArcF. The fact that the asymmetry of the ArcF connectivity in left TLE patients was not driven by hemispheric language specialization can be compared with recent DTI findings that showed a leftward asymmetry of the ArcF in volunteers, irrespective of handedness or hemispheric specialization (Vernooij et al., 2007). The question arises about the functional interpretation of the leftward ArcF asymmetry. It seems likely that the ArcF is not exclusively involved in language processing but also plays a role in other functions that could be lateralized to the left hemisphere irrespective of the side of language lateralization, such as acoustic processing (Zatorre & Belin, 2001). Thus, the leftward asymmetry of the ArcF connectivity found in controls, irrespective of their manual preference (Vernooij et al., 2007), in most right TLE, and in some left TLE patients could be driven by the summation of these functional roles and not solely reflect hemispheric language specialization. This suggests that careful consideration should be given to the ArcF during surgery of the left hemisphere, irrespective of the side of functional language dominance (Vernooij et al., 2007).

Besides the dorsal ArcF, Parker et al. identified a ventral pathway in humans, the existence of which has been suggested by nonhuman primate studies (Parker et al., 2005). This ventral route, which connects the posterior superior temporal region with the orbitofrontal and dorsolateral prefontal regions, might anatomically correspond, at least in part, to the IOF (Catani et al., 2002). Its putative role for the semantic processing system is supported by corticosubcortical electrostimulation data, especially in the left dominant hemisphere (Duffau et al., 2005). Using DTI, Parker et al. were able to identify this ventral pathway only in the left hemisphere, implying more extensive anatomical connectivity on the left than on the right, in right-handed volunteers (Parker et al., 2005). In TLE patients, our results did not confirm the leftward FA asymmetry of the IOF that was found in right-handed controls (Rodrigo et al., 2007). We are unaware of any imaging study that has attempted to compare the diffusion-based measurement in the IOF with lateralization indices derived from fMRI language tasks in controls or in epilepsy patients. Given that we did not find any correlation between functional and structural indices measured in the IOF, this bundle might not constitute a major component of the language reorganization network in epilepsy.

Finally, the neuronal reorganization of language in epilepsy may involve other paths than the ArcF and IOF, and rely on a widespread neuronal network. In left TLE patients, this distributed network could be hosted by the right frontal white matter, within the large “connectivity volume” where the mean anisotropy was computed by Powell et al. Although speculative, this would reconcile our negative finding in the ArcF with the positive findings of Powell et al. in a large “connectivity volume” in the frontal lobe in left TLE patients (Powell et al., 2007).

Our study has several limitations. First, we did not include a control group. However, leftward structural asymmetry of white matter pathways in relation to functional language lateralization has previously been reported in right-handed volunteers, using nonprobabilistic and probabilistic tractography (Catani et al., 2005; Parker et al., 2005; Vernooij et al., 2007). Second, we chose to use a two ROI-based approach for fiber tracking to specifically study the ArcF and the IOF, the course of which have been documented using DTI (Catani et al., 2002). To avoid observer bias inherent in manual ROI definition in each patient, we decided to draw on the MNI template a set of predefined ROIs, which were then used for all patients after having transformed them into each patient's space. This procedure avoids repeating freehand ROI definition in each patient but might induce false-negative or false-positive tracking because of misregistration. We initially chose relatively large ROIs to minimize the risk of missing the ArcF and the IOF. This proved rather successful since the ArcF tracking was obtained bilaterally in all cases and the IOF could not be segmented in only two patients. Each of the tracts was checked for consistency with known anatomy and no aberrant tracts were observed. It is unlikely that misregistrations had a great impact on our DTI measurements given that our findings in right TLE patients are in line with reports of leftward asymmetry of the ArcF using a different methodology (Powell et al., 2007). Last, to be clinically compatible when combined with fMRI, DTI acquisition time was limited to 7 min and we aimed at covering the whole brain with isotropic voxels. We chose to maximize the spatial resolution (2.5 mm), which allowed for 15 diffusion gradient directions within the time constraints. Increasing the number of directions would most likely have improved the accuracy of connectivity measurements (Behrens et al., 2007).

We have combined probabilistic tractography and fMRI to study the effect of TLE upon the structure—function relationship. We found a correlation between the ArcF structural lateralization and the leftward asymmetry in functional activation in right TLE patients, with patients with a more lateralized language function having a more lateralized FA measurements. This suggests that probabilistic tractography and fMRI can be combined to confirm a left hemispheric language specialization in patients with right TLE. The decoupling between structure and function in left TLE suggests that neuronal reorganization of language may involve other paths than the ArcF and IOF, and rely on a widespread neuronal network. Before fMRI and diffusion combined analysis becomes clinically applicable for predicting postoperative deficits, additional work needs to be done to understand the structural bases of the language network and how it is affected by chronic seizures.

Conflict of interest: We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. None of the authors have any conflicts of interest.

References

  1. Top of page
  2. Materials and Methods
  3. Results
  4. Discussion
  5. References
  • Adcock JE, Wise RG, Oxbury JM, Oxbury SM, Matthews PM. (2003) Quantitative fMRI assessment of the differences in lateralization of language-related brain activation in patients with temporal lobe epilepsy. Neuroimage 18:423438.
  • Beaulieu C. (2002) The basis of anisotropic water diffusion in the nervous system—a technical review. NMR Biomed 15:435455.
  • Beckmann CF, Jenkinson M, Smith SM. (2003) General multilevel linear modeling for group analysis in FMRI. Neuroimage 20:10521063.
  • Behrens TE, Woolrich MW, Jenkinson M, Johansen-Berg H, Nunes RG, Clare S, Matthews PM, Brady JM, Smith SM. (2003) Characterization and propagation of uncertainty in diffusion-weighted MR imaging. Magn Reson Med 50:10771088.
  • Behrens TE, Berg HJ, Jbabdi S, Rushworth MF, Woolrich MW. (2007) Probabilistic diffusion tractography with multiple fibre orientations: What can we gain? Neuroimage 34:144155.
  • Buchel C, Raedler T, Sommer M, Sach M, Weiller C, Koch MA. (2004) White matter asymmetry in the human brain: a diffusion tensor MRI study. Cereb Cortex 14:945951.
  • Cao Y, Whalen S, Huang J, Berger KL, DeLano MC. (2003) Asymmetry of subinsular anisotropy by in vivo diffusion tensor imaging. Hum Brain Mapp 20:8290.
  • Catani M, Howard RJ, Pajevic S, Jones DK. (2002) Virtual in vivo interactive dissection of white matter fasciculi in the human brain. Neuroimage 17:7794.
  • Catani M, Jones DK, Ffytche DH. (2005) Perisylvian language networks of the human brain. Ann Neurol 57:816.
  • Dejerine J. (1895) Anatomie des Centres Nerveux. Rueff et Cie, Paris .
  • Duffau H, Gatignol P, Mandonnet E, Peruzzi P, Tzourio-Mazoyer N, Capelle L. (2005) New insights into the anatomo-functional connectivity of the semantic system: a study using cortico-subcortical electrostimulations. Brain 128:797810.
  • Fernandez G, Specht K, Weis S, Tendolkar I, Reuber M, Fell J, Klaver P, Ruhlmann J, Reul J, Elger CE. (2003) Intrasubject reproducibility of presurgical language lateralization and mapping using fMRI. Neurology 60:969975.
  • Gaillard WD, Balsamo L, Xu B, McKinney C, Papero PH, Weinstein S, Conry J, Pearl PL, Sachs B, Sato S, Vezina LG, Frattali C, Theodore WH. (2004) fMRI language task panel improves determination of language dominance. Neurology 63:14031408.
  • Geschwind N, Levitsky W. (1968) Human brain: left-right asymmetries in temporal speech region. Science 161:186187.
  • Golestani N, Paus T, Zatorre RJ. (2002) Anatomical correlates of learning novel speech sounds. Neuron 35:9971010.
  • Hertz-Pannier L, Chiron C, Jambaque I, Renaux-Kieffer V, Van de Moortele PF, Delalande O, Fohlen M, Brunelle F, Le Bihan D. (2002) Late plasticity for language in a child's non-dominant hemisphere: a pre- and post-surgery fMRI study. Brain 125:361372.
  • Josse G, Herve PY, Crivello F, Mazoyer B, Tzourio-Mazoyer N. (2006) Hemispheric specialization for language: Brain volume matters. Brain Res 1068:184193.
  • Kier EL, Staib LH, Davis LM, Bronen RA. (2004) MR imaging of the temporal stem: anatomic dissection tractography of the uncinate fasciculus, inferior occipitofrontal fasciculus, and Meyer's loop of the optic radiation. AJNR Am J Neuroradiol 25:677691.
  • Le Bihan D, Mangin JF, Poupon C, Clark CA, Pappata S, Molko N, Chabriat H. (2001) Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging 13:534546.
  • Lehericy S, Cohen L, Bazin B, Samson S, Giacomini E, Rougetet R, Hertz-Pannier L, Le Bihan D, Marsault C, Baulac M. (2000) Functional MR evaluation of temporal and frontal language dominance compared with the Wada test. Neurology 54:16251633.
  • Parker GJ, Luzzi S, Alexander DC, Wheeler-Kingshott CA, Ciccarelli O, Lambon Ralph MA. (2005) Lateralization of ventral and dorsal auditory-language pathways in the human brain. Neuroimage 24:656666.
  • Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G. (1996) Diffusion tensor MR imaging of the human brain. Radiology 201:637648.
  • Powell HW, Parker GJ, Alexander DC, Symms MR, Boulby PA, Wheeler-Kingshott CA, Barker GJ, Noppeney U, Koepp MJ, Duncan JS. (2006) Hemispheric asymmetries in language-related pathways: a combined functional MRI and tractography study. Neuroimage 32:388399.
  • Powell HW, Parker GJ, Alexander DC, Symms MR, Boulby PA, Wheeler-Kingshott CA, Barker GJ, Koepp MJ, Duncan JS. (2007) Abnormalities of language networks in temporal lobe epilepsy. Neuroimage 36:209221.
  • Pujol J, Lopez-Sala A, Deus J, Cardoner N, Sebastian-Galles N, Conesa G, Capdevila A. (2002) The lateral asymmetry of the human brain studied by volumetric magnetic resonance imaging. Neuroimage 17:670679.
  • Rodrigo S, Naggara O, Oppenheim C, Golestani N, Poupon C, Cointepas Y, Mangin JF, Le Bihan D, Meder JF. (2007) Human subinsular asymmetry studied by diffusion tensor imaging and fiber tracking. AJNR Am J Neuroradiol 28:15261531.
  • Sabbah P, Chassoux F, Leveque C, Landre E, Baudoin-Chial S, Devaux B, Mann M, Godon-Hardy S, Nioche C, Ait-Ameur A, Sarrazin JL, Chodkiewicz JP, Cordoliani YS. (2003) Functional MR imaging in assessment of language dominance in epileptic patients. Neuroimage 18:460467.
  • Sabsevitz DS, Swanson SJ, Hammeke TA, Spanaki MV, Possing ET, Morris GL, 3rd, Mueller WM, Binder JR. (2003) Use of preoperative functional neuroimaging to predict language deficits from epilepsy surgery. Neurology 60:17881792.
  • Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TE, Johansen-Berg H, Bannister PR, De Luca M, Drobnjak I, Flitney DE, Niazy RK, Saunders J, Vickers J, Zhang Y, De Stefano N, Brady JM, Matthews PM. (2004) Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 23(Suppl 1):S208219.
  • Springer JA, Binder JR, Hammeke TA, Swanson SJ, Frost JA, Bellgowan PS, Brewer CC, Perry HM, Morris GL, Mueller WM. (1999) Language dominance in neurologically normal and epilepsy subjects: a functional MRI study. Brain 122(Pt 11):20332046.
  • Staudt M, Grodd W, Niemann G, Wildgruber D, Erb M, Krageloh-Mann I. (2001) Early left periventricular brain lesions induce right hemispheric organization of speech. Neurology 57:122125.
  • Sveller C, Briellmann RS, Saling MM, Lillywhite L, Abbott DF, Masterton RA, Jackson GD. (2006) Relationship between language lateralization and handedness in left-hemispheric partial epilepsy. Neurology 67:18131817.
  • Thivard L, Hombrouck J, Du Montcel ST, Delmaire C, Cohen L, Samson S, Dupont S, Chiras J, Baulac M, Lehericy S. (2005) Productive and perceptive language reorganization in temporal lobe epilepsy. Neuroimage 24:841851.
  • Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, Mazoyer B, Joliot M. (2002) Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15:273289.
  • Vernooij MW, Smits M, Wielopolski PA, Houston GC, Krestin GP, Van Der Lugt A. (2007) Fiber density asymmetry of the arcuate fasciculus in relation to functional hemispheric language lateralization in both right- and left-handed healthy subjects: a combined fMRI and DTI study. Neuroimage 35:10641076.
  • Wechsler D. (1997) WAIS-III: Administration and scoring manual. Psychological Corporation, San Antonio .
  • Werring DJ, Clark CA, Barker GJ, Miller DH, Parker GJ, Brammer MJ, Bullmore ET, Giampietro VP, Thompson AJ. (1998) The structural and functional mechanisms of motor recovery: complementary use of diffusion tensor and functional magnetic resonance imaging in a traumatic injury of the internal capsule. J Neurol Neurosurg Psychiatry 65:863869.
  • Worsley KJ, Friston KJ. (1995) Analysis of fMRI time-series revisited–again. Neuroimage 2:173181.
  • Yetkin FZ, Hammeke TA, Swanson SJ, Morris GL, Mueller WM, McAuliffe TL, Haughton VM. (1995) A comparison of functional MR activation patterns during silent and audible language tasks. AJNR Am J Neuroradiol 16:10871092.
  • Zatorre RJ, Belin P. (2001) Spectral and temporal processing in human auditory cortex. Cereb Cortex 11:946953.