Abstract

Although the hippocampus is not considered a key structure in semantic memory, patients with medial-temporal lobe epilepsy (mTLE) have deficits in semantic access on some word retrieval tasks. We hypothesized that these deficits reflect the negative impact of focal epilepsy on remote cerebral structures. Thus, we expected that the networks that support word retrieval tasks would be altered in left mTLE patients. We measured brain activity with fMRI while participants (13 controls, 13 left mTLE, and 13 right mTLE) performed a verb generation task. We examined functional connectivity during this task in relation to language performance on an off-line clinical test of lexical access (Boston Naming Test, BNT). Using task–seed–behavior partial least squares, we identified a canonical language network that was more active during verb generation than the baseline condition, but this network did not correlate with variability in BNT performance in either controls or patients. Instead, additional networks were identified for each group, with more anterior temporal and prefrontal regions recruited for controls and more posterior temporal regions for both left and right mTLE patients. Our findings go beyond the literature emphasizing differences in laterality of language processes in mTLE patients and, critically, highlight how network changes can be used to account for performance variation among patients on clinically relevant measures. This strategy of correlating network changes and off-line behavior may provide a powerful tool for predicting a postoperative decline in language performance.

INTRODUCTION

Although it is well established that individuals with left medial-temporal lobe epilepsy (mTLE) are impaired on tests of verbal episodic memory (Hermann et al., 1994; Miller, Munoz, & Finmore, 1993; Frisk & Milner, 1990; Milner, 1971), there also are indications of subtle deficits in semantic access on word retrieval tasks such as category fluency (Gleissner & Elger, 2001; N'Kaoua, Lespinet, Barsse, Rougier, & Claverie, 2001; Troster et al., 1995) and naming to description or confrontation (Busch, Frazier, Iampietro, Chapin, & Kubu, 2009; Loring et al., 2008; Messas, Mansur, & Castro, 2008). How might we account for this, given that the medial-temporal region is not a central node of the canonical language/word retrieval network? Although some studies have shown a direct correlation between language impairment and structural integrity of the hippocampus (Alessio et al., 2006; Davies et al., 1998), it is likely that depressed function on semantic verbal tasks reflects diaschisis, such that focal epileptogenic tissue adversely influences remote cerebral structures leading to additional cognitive deficits (Engel, Bandler, Griffith, & Caldecott-Hazard, 1991).

There is now considerable evidence for widespread structural and functional abnormalities associated with mTLE, including atrophy and sclerosis of associated temporal neocortex (Bonilha & Halford, 2009; Riederer et al., 2008), and more remote physiologic disturbance, including the contralesional hemisphere on PET and MRS (Cendes, Caramanos, Andermann, Dubeau, & Arnold, 1997; Savic, Altshuler, Baxter, & Engel, 1997; Henry, Mazziotta, & Engel, 1993; Theodore et al., 1983). Recent studies using diffusion tensor imaging have documented abnormalities in major fiber tracts throughout the affected hemisphere, indicating that connectivity between regions may be compromised (Focke et al., 2008; Thivard et al., 2005). Furthermore, several studies suggest that these abnormalities affect verbal performance. Arnold et al. (1996) showed that in patients with left mesiotemporal seizure origin, hypometabolic thalamic and lateral temporal regions were related to impairments in word fluency. Also with left mTLE patients, Trebuchon-Da Fonseca et al. (2009) showed that the rate of anomic states increases with left posterior and basal temporal hypometabolism. Jokeit et al. (1997) showed that left mTLE patients were more likely than right mTLE patients to have prefrontal metabolic asymmetry and that this asymmetry was correlated with verbal intelligence. Structural compromise of several major fiber pathways (e.g., arcuate fasciculus, uncinate fasciculus) that demonstrate on diffusion tensor imaging has been shown to be correlated with naming deficits (McDonald et al., 2008). These widespread anatomical and physiological abnormalities in language-relevant regions likely account for enhanced recruitment of additional or alternate regions during language tasks by patients with dominant mTLE as shown by fMRI (Powell et al., 2007; Janszky, Mertens, Janszky, Ebner, & Woermann, 2006; Billingsley, McAndrews, Crawley, & Mikulis, 2001) and cortical stimulation mapping (Hamberger et al., 2007).

Although it is clear that there is a distributed network of regions supporting lexical and semantic processing and that there exists a disturbance in brain processing beyond the medial-temporal region in mTLE patients, there has been little systematic exploration of the abnormalities in brain activation patterns at a network level for language performance. In particular, studies of functional connectivity in language networks that are crucial to understanding how medial-temporal dysfunction impacts these networks in mTLE patients are rare. To our knowledge, no one has examined task-related functional connectivity for language networks in mTLE. Two studies have examined resting state functional connectivity between language regions (e.g., correlation of low frequency oscillations during rest among brain regions). Waites, Briellmann, Saling, Abbott, and Jackson (2006) used fMRI to examine both task-related (lexical retrieval) regional activation and resting state functional connectivity between language areas in left mTLE patients and healthy controls. They found that regional activation on the task was similar in patients and controls but that patients had reduced connectivity in resting state language networks. Azari et al. (1999) examined resting state functional connectivity for language in mTLE patients using PCA and identified differentiating patterns of brain metabolic interactions. Left mTLE patients had enhanced connectivity between left inferior frontal cortex, left superior temporal cortex, right medial-temporal cortex, and right thalamus, and this abnormal connectivity pattern predicted verbal intelligence deficits. Clearly, there is some ambiguity on the nature of connectivity changes, and importantly, neither study examined functional connectivity while participants performed a language task. We argue that networks identified during language performance could provide a basis for stronger inferences regarding activation patterns and behavior as networks identified in “resting state” analyses may not be directly relevant to systems engaged by coordinated functioning (cf. Calhoun, Kiehl, & Pearlson, 2008). Thus, for the current study, we characterize the relevant functional system to define language-related connectivity and its relationship to clinically relevant behavior.

We were interested in network alterations that demonstrate compensation (i.e., those regions that are more engaged in relation to better clinical performance). In the current study, we used partial least squares (PLS) analysis to examine task-related functional connectivity for a semantic word retrieval task (verb generation) that is one of the most frequently used paradigms in clinical settings to provide optimal language lateralization profiles (Szaflarski et al., 2008; Powell et al., 2007; Benson et al., 1999). This analysis provided us with a pattern of brain activity that operationally defined a “canonical” word retrieval network. Importantly, this network did not include the medial-temporal regions in our data. We then asked whether variations in this functional network across patients and groups explained variation in a clinical indicator of word retrieval sensitive to mTLE damage, specifically on the Boston Naming Test (BNT; Kaplan, Goodglass, & Weintraub, 1983). By using independent clinically relevant tasks to define the functional network and the outcome variables, we could enable more general inferences regarding network organization and neuropsychological performance. Thus, we hypothesized that language-related difficulties in mTLE patients would reflect alterations in networks supporting semantic access.

METHODS

Participants

Twenty-six individuals with pharmacologically intractable mTLE were recruited for this study through the Neuropsychology and Epilepsy program at Toronto Western Hospital. Thirteen patients had seizure originating in the left medial-temporal lobe (MTL; 6 men, 7 women; mean age = 38.5 years, range = 20–53 years), and 13 patients had seizures originating in the right MTL (4 men, 9 women; mean age = 38.2 years, range = 18–62 years). In each case, MTL seizure onset was confirmed by scalp or (rarely) intracranial EEG recordings. Among those patients who underwent surgery (9 of 13 left mTLE patients and 10 of 13 right mTLE patients) and have at least 6 months of follow-up, all but one (right mTLE) have been seizure-free. The patient who is not seizure-free will have a second surgery because it was determined that not enough MTL tissue was removed during the first surgery. All patients had language lateralized to the left hemisphere based on lateralization indices derived from a panel of fMRI activation tasks (verb generation, sentence completion, and category fluency). Finally, all were on a combination of two to four antiepileptic drugs at the time of fMRI and neuropsychological testing, but none had good seizure control, which was the reason for consideration of surgery. Exclusion criteria included foci or lesions outside of the MTL (e.g., frontal or temporal neocortical focus) and evidence of psychiatric or neurological disorders. Demographic, clinical, and neuropsychological characteristics are presented in Table 1.

Table 1. 

Group Demographics, Clinical, and Neuropsychological Characteristics

Group
Left mTLE
Right mTLE
Control
Age at testing (years) 38.5 (20–53) 38.2 (18–62) 35.9 (22–59) 
Education 14 (10–19) 14.4 (12–22) 18.6 (16–22) 
Handedness 12 right, 1 left 13 right 12 right 
Age at onset 16.7 (0.5–40) 20.1 (3–46)  
Disease duration 21.7 (2–45.5) 18.2 (2–42)  
MTS on MRI 8 of 13 4 of 13  
Verbal IQ 96.3 (79–123) 108.7 (94–128)  
Performance IQ 98.4 (88–123) 104 (89–114)  
WRAT 94.9 (82–122) 103.6 (92–109)  
Group
Left mTLE
Right mTLE
Control
Age at testing (years) 38.5 (20–53) 38.2 (18–62) 35.9 (22–59) 
Education 14 (10–19) 14.4 (12–22) 18.6 (16–22) 
Handedness 12 right, 1 left 13 right 12 right 
Age at onset 16.7 (0.5–40) 20.1 (3–46)  
Disease duration 21.7 (2–45.5) 18.2 (2–42)  
MTS on MRI 8 of 13 4 of 13  
Verbal IQ 96.3 (79–123) 108.7 (94–128)  
Performance IQ 98.4 (88–123) 104 (89–114)  
WRAT 94.9 (82–122) 103.6 (92–109)  

Values in parentheses represent the range. MTS = medial-temporal sclerosis; WRAT = Wide Range Achievement Test.

The control group comprised four men and eight women (mean age = 36.25 years, range = 22–59 years) who were recruited from local hospital and university communities. They had no prior history of neurological or psychiatric impairment. On the basis of lateralization indices from an fMRI panel of language tasks, all controls had language lateralized to the left hemisphere. The control group did not differ significantly from the patient groups in age (see Table 1), but they attained a higher mean level of education (18.5 years vs. 14.0 and 14.4 years in the left and right mTLE groups, respectively).

Informed consent was obtained from all participants, in accordance with a protocol approved by University Health Network Research Ethics Board.

Language Task

During a blocked design paradigm, participants saw either nouns (e.g., chairs), which they covertly generated verbs, or symbols (e.g., *#*#*), which required only visual attention. Both verb generation and visual attention (i.e., baseline) blocks lasted 27 sec and were comprised of 5 stimuli, presented for 5 sec, with a 500-msec ISI; there were 10 repetitions of the paradigm during the scan. A practice session before scanning ensured that participants were able to comply with task instructions. As part of our standard clinical practice for presurgical language localization, univariate analyses were conducted with AFNI (http://afni.nimh.nih.gov/; Cox, 1996) to confirm task engagement. These demonstrated language-related activation in frontal and temporal-parietal regions, with left dominance in each TLE participant included in the study.

Data Acquisition

Anatomical and functional data were acquired on a 3-T Signa MR System (GE Medical Systems, Milwaukee WI). Functional data were acquired in an interleaved order (repetition time = 2 sec; 28 slices or 32 slices determined by head size,1 field of view = 440 mm, 64 × 64 matrix, resulting in a voxel size of 3.75 × 3.75 × 5.0). Scans were taken in an oblique orientation to maximize signal intensity and to minimize partial volume effects in the MTL. Three-dimensional anatomical scans were acquired with higher spatial resolution (T1-weighted sequence, 120 slices, field of view = 220 mm, 256 × 256 matrix, resulting in a voxel size of 0.78125 × 0.78125 × 1.0).

Data Processing and Statistical Analyses

Normalization to the Montreal Neurological Institute (MNI) EPI template and smoothing were performed using SPM5 (Statistical Parametric Mapping 5; Wellcome Department of Imaging Neuroscience, London). The transformation of each participant to the EPI template was achieved using a 12-parameter affine transform with sinc interpolation. Images were smoothed with an 8-mm isotropic Gaussian filter before analysis. For each participant, “brain” voxels in a specific image were defined as voxels with intensities greater than 15% of the maximum value in that image. The union of masks was used for group analyses as described below.

Data Analysis

The primary image analysis was done with PLS, which is a multivariate technique that enables the identification of the optimal spatial patterns that differentiate tasks or groups in terms of activation or in terms of functional connectivity and brain–behavior correlations (http://rotman-baycrest.on.ca/index.php?section=84; for a detailed description of PLS's application to blocked design fMRI data, see McIntosh, Bookstein, Haxby, & Grady, 1996). Two forms of PLS were performed. The first, task PLS, identified distributed activity patterns, or latent variables (LVs), that show similarities or differences between participant groups and experimental conditions. The task PLS also allowed us to identify an ROI (i.e., seed voxel) whose activity reliably differentiated verb generation from baseline tasks. The second, task–seed–behavior PLS, decomposes a correlation matrix that describes brain–behavior (covariance between voxel activity and measures of behavior across participants), brain–seed (covariance between activity in a seed voxel and activity in the rest of the brain), and brain–task (changes in voxel activity as a function of task conditions) relationships (referred to as “multiblock PLS”; Caplan, Luks, Simpson, Glaholt, & McIntosh, 2006; McIntosh, Cabeza, & Lobaugh, 1998; Schreurs et al., 1997; McIntosh et al., 1996) This analysis allowed us to identify cortical regions that were functionally connected with our seed voxel, correlated with a clinical measure of language performance, and also assessed how these relationships changed as a function of task condition and patient group. The motivation of this multivariate analysis was to uncover functional networks that support language and to characterize their dependence on task-related activity as well as their relationship to a clinical test of language function, the Boston Naming Test. We were particularly interested in whether these functional networks would distinguish between our participant populations. A single multivariate analysis included all these factors, enabling us to concisely summarize the multiple functional networks. These analyses are explained further below.

Task PLS

Task PLS uses singular value decomposition to identify distributed activity patterns or LVs that show similarities or differences between participant groups and experimental conditions. Each LV contains three vectors. The first vector contains a singular value, which indicates the strength of the effect expressed by the LV. The remaining two vectors relate experimental design and brain activity. The experimental design vector contains task saliences, which indicate the degree to which each task is related to the brain activity pattern identified in the LV. These task saliences can be interpreted as the optimal contrast that codes the effect depicted in the LV. The brain activity vector contains voxel saliences. These are numerical voxel weights that identify the collection of voxels that as a whole are most related to the effects expressed in the LV. Note that for each LV, there is one salience per voxel that applies for all groups and tasks. Multiplying the BOLD signal value in each brain voxel for each subject by the salience for that voxel and summing across all voxels gives a brain score for each subject on a given LV. Brain scores indicate the degree to which each subject shows the spatial pattern of voxels expressed in the LV. PLS is similar to other multivariate techniques, such as PCA, in that contrasts across conditions or groups typically are not specified by the experimenter. Instead, the algorithm extracts LVs explaining the covariance between groups, conditions, and brain activity in order of the amount of covariance explained, with the LV accounting for the most covariance extracted first.

Statistical assessment for PLS is done using permutation tests for the LVs and bootstrap estimation of standard errors for the voxel saliences. The permutation test assesses whether the effect represented in a given LV, captured by the singular value, is sufficiently strong to be different from random noise (McIntosh et al., 1996). This was accomplished using sampling without replacement to reassign the order of conditions for each subject. PLS was recalculated for each sample, and the number of times the permuted singular values exceeded the observed singular values was calculated. Exact probabilities are presented for all LVs.

The standard error estimates of the voxel weights/saliences from the bootstrap tests (i.e., bootstrap ratios) are used to assess the reliability of the nonzero voxel saliences in significant LVs. Bootstrap tests were generated using sampling with replacement, keeping the assignment of experimental conditions fixed for all subjects. PLS was recalculated for each bootstrap test. A salience whose value depends greatly on which subjects are in the sample is less precise than one that remains stable regardless of the sample chosen (Sampson, Streissguth, Barr, & Bookstein, 1989). No corrections for multiple comparisons are necessary because the voxel saliences are calculated in a single mathematical step on the whole brain. The bootstrap ratio is proportional to a z score but should be interpreted as a confidence interval. For the current article, we designated a threshold of at least 3.8 corresponding roughly to a 99.99% confidence interval or a p value less than .0001. We used different bootstrap thresholds to capture visually all the stable voxel clusters but kept these clusters small enough that they are easy to differentiate.

We additionally used task PLS to identify an ROI (i.e., seed voxel) whose activity most reliably differentiated verb generation from the baseline task. Specifically, from the LV that differentiated brain activation during verb generation versus baseline, we chose the voxel within canonical language regions with the highest bootstrap ratio.

Task–Seed–Behavior PLS

Task–seed–behavior PLS combines three uses of PLS into one analysis. As described above, task PLS analyzes mean changes in brain activity as a function of conditions and groups. Seed PLS examines group- and task-dependent correlations between a seed ROI (or seed voxel) and the rest of the brain and identifies distributed patterns of functional connectivity. Behavior PLS examines group- and task-dependent correlations between the behavior measure and the rest of the brain and identifies distributed patterns of brain–behavior correlations.

For the current article, only the verb generation condition was included in the seed portion of the task–seed–behavior analysis because we were interested specifically in the networks that support task-related functional connectivity for language and also vary with word retrieval performance.

Singular value decomposition of the task–behavior–seed–brain correlation matrix produces a set of distributed activity patterns or LVs that show similarities or differences between participant groups, experimental conditions, and brain–seed–behavior correlations. In the task PLS portion of the task–seed–behavior analysis, the group and the task saliences indicate the degree to which each group and task is related to the identified pattern of BOLD amplitude differences. For the seed–behavior part of the task–seed–behavior analysis, the variation across the group saliences and seed/behavior saliences indicates whether a given LV represents a similarity or a difference in the behavior–seed–brain correlations across groups. This also can be shown by calculation of correlation between the brain scores (dot product of the voxel salience and fMRI data) and seed fMRI signal or behavior for each task. The voxel saliences give the corresponding spatio-temporal activity pattern.

Our three-group task–seed–behavior PLS identified the cortical regions that were functionally connected with our seed voxel, correlated with our clinical measure of language performance, and also assessed how these relationships changed as a function of task condition and patient group. Statistical assessment is similar to that used for task PLS.

RESULTS

A three-group ANOVA yielded a significant group difference in mean BNT scores, F(2, 42) = 10.90, p < .001 (control mean = 55.31, SE = 1.24; right mTLE mean = 52.81, SE = 1.16; and left mTLE mean = 44.69, SE = 2.25). Planned comparisons indicate that there was a significant difference between controls and left mTLE patients, t(27) = 3.87, p < .001, and left and right mTLE patients, t(27) = 3.21, p < .005, but not between controls and right mTLE patients, t(27) = 1.46, p > .05. Importantly, there was sufficient variability in performance in all three groups that could be exploited to identify networks that facilitate clinically relevant language performance in the context of MTL damage.

Our activity analysis (three-group task PLS) identified one significant LV, which differentiated brain activation during verb generation versus baseline; control and mTLE groups demonstrated the same pattern of activation (p < .001, explained covariance = 88%; Figure 1; for a list of local maxima, see Table 2). Dominant positive saliences (related to increased activity during verb generation) were seen in the left inferior pFC (Broca's area), left middle frontal cortex, left temporal pole, left SMA, and left anterior cingulate. Dominant negative saliences (related to increased activity during baseline) were seen in the medial pFC, right posterior cingulate, and right inferior parietal lobule. We used this LV to identify the canonical language regions that best differentiated between verb generation and baseline tasks. The voxel within these regions with the highest bootstrap ratio was located in the left inferior frontal cortex (Brodmann's area [BA] 44; MNI coordinates −48 16 12); of note, this is well outside the focal region of epileptogenicity in our patients.

Figure 1. 

(A) Brain scores bar graph and (B) singular image for LV1 in the three-group task PLS. The brain scores bar graph captures the mean brain score for each condition in each group. The error bars show the 95% confidence intervals derived from bootstrap estimation. This bar graph indicates that there are activity differences between verb generation and baseline, but not between control and mTLE groups. The relationship is embodied in the singular image, which identifies regions with maximal differentiation between verb generation and baseline conditions for control and mTLE groups (this is the same for all three subject groups), displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Yellow regions represent increased verb generation-related activity, and blue regions represent increased baseline-related activity. White circle indicates the approximate location of the BA 44 seed voxel used in the task–seed–behavior PLS analysis.

Figure 1. 

(A) Brain scores bar graph and (B) singular image for LV1 in the three-group task PLS. The brain scores bar graph captures the mean brain score for each condition in each group. The error bars show the 95% confidence intervals derived from bootstrap estimation. This bar graph indicates that there are activity differences between verb generation and baseline, but not between control and mTLE groups. The relationship is embodied in the singular image, which identifies regions with maximal differentiation between verb generation and baseline conditions for control and mTLE groups (this is the same for all three subject groups), displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Yellow regions represent increased verb generation-related activity, and blue regions represent increased baseline-related activity. White circle indicates the approximate location of the BA 44 seed voxel used in the task–seed–behavior PLS analysis.

Table 2. 

Local Maxima for the Three-Group Task PLS LV1

Region
x (mm)
y (mm)
z (mm)
BSR
Left inferior prefrontal (Broca's area) −48 16 12 13.58 
Left middle frontal cortex (BA 9) −48 40 12.39 
Left temporal pole (BA 38) −48 20 −16 11.6 
Left SMA −4 12 52 16.25 
Left anterior cingulate −8 20 40 12.53 
Bilateral medial prefrontal (BA 11) 56 −12 −8.12 
Right posterior cingulate (BA 31) 12 −52 28 −9.87 
Right inferior parietal (precuneus BA 7) −56 32 −9.62 
Region
x (mm)
y (mm)
z (mm)
BSR
Left inferior prefrontal (Broca's area) −48 16 12 13.58 
Left middle frontal cortex (BA 9) −48 40 12.39 
Left temporal pole (BA 38) −48 20 −16 11.6 
Left SMA −4 12 52 16.25 
Left anterior cingulate −8 20 40 12.53 
Bilateral medial prefrontal (BA 11) 56 −12 −8.12 
Right posterior cingulate (BA 31) 12 −52 28 −9.87 
Right inferior parietal (precuneus BA 7) −56 32 −9.62 

Regions indicate the gyral locations and BA of the cluster peak. The MNI coordinates were converted into Talairach coordinates using the mni2tal script (http://eeg.sourceforge.net/mridoc/mri_toolbox/mni2tal.html). Gyral locations and BA were then determined by reference to Talairach and Tournoux (1988). x, y, and z indicate voxel coordinates in the MNI space. BSR represents each voxel's PLS parameter estimate divided by its standard error.

We next looked for networks (functionally connected regions) that might be related to off-line variability in naming performance. The three-group task–seed–behavior PLS revealed four significant LVs. The first significant LV identified the subset of brain regions whose activity distinguished between verb generation and baseline conditions and also functionally connected with BA 44 during verb generation (p < .001, explained covariance = 49%; Figure 2; for a list of local maxima, see Table 3). This pattern did not vary by group and showed no stable correlation with BNT scores. Dominant positive saliences (related to increased activation during verb generation and positive correlations with the BA 44 voxel) were located in the left inferior pFC (Broca's area), left middle frontal cortex, left temporal pole, left anterior cingulate, left SMA, and left thalamus. Dominant negative saliences (related to increased activation during baseline and negative correlations with BA 44) were located in the bilateral medial pFC, right posterior cingulate, and right inferior parietal lobule.

Figure 2. 

(A) Brain scores bar graph, (B) correlation bar graph, (C) and singular image for LV1 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph captures the mean brain score for each condition in each group and indicates that there are activity differences between verb generation and baseline, but not between control and mTLE groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations during the verb generation task. The error bars show the 95% confidence interval derived from bootstrap estimation. The error bar crosses zero for BNT scores in all three groups, indicating that the correlation between BNT scores and the areas identified in the singular image is nonreliable. (C) The singular image shows brain–seed correlations for control and mTLE groups, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). For all groups, regions in yellow demonstrate increased activation during the verb generation task and reliable positive correlations with the BA 44 voxel. Regions in blue are more active during baseline and anticorrelated to the verb generation network. This network does not account for variation in performance on the clinical BNT measure, as there is no reliable correlation with BNT scores.

Figure 2. 

(A) Brain scores bar graph, (B) correlation bar graph, (C) and singular image for LV1 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph captures the mean brain score for each condition in each group and indicates that there are activity differences between verb generation and baseline, but not between control and mTLE groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations during the verb generation task. The error bars show the 95% confidence interval derived from bootstrap estimation. The error bar crosses zero for BNT scores in all three groups, indicating that the correlation between BNT scores and the areas identified in the singular image is nonreliable. (C) The singular image shows brain–seed correlations for control and mTLE groups, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). For all groups, regions in yellow demonstrate increased activation during the verb generation task and reliable positive correlations with the BA 44 voxel. Regions in blue are more active during baseline and anticorrelated to the verb generation network. This network does not account for variation in performance on the clinical BNT measure, as there is no reliable correlation with BNT scores.

Table 3. 

Local Maxima for the Three-Group Task–Seed–Behavior PLS LV1

Region
x (mm)
y (mm)
z (mm)
BSR
Left inferior pFC (Broca's area) −48 12 12 15.44 
Left middle frontal cortex (BA 9) −48 40 14.02 
Left temporal pole (BA 38) −52 24 −16 16.08 
Left anterior cingulate (BA 32) −4 16 44 19.87 
Left SMA  −4 64 20.99 
Left thalamus −16 13.5 
Bilateral medial pFC (BA 11) 52 −12 −7.33 
Right posterior cingulate (BA 31) −56 28 −5.69 
Right inferior parietal lobule (BA 40) 64 −52 24 −4.98 
Region
x (mm)
y (mm)
z (mm)
BSR
Left inferior pFC (Broca's area) −48 12 12 15.44 
Left middle frontal cortex (BA 9) −48 40 14.02 
Left temporal pole (BA 38) −52 24 −16 16.08 
Left anterior cingulate (BA 32) −4 16 44 19.87 
Left SMA  −4 64 20.99 
Left thalamus −16 13.5 
Bilateral medial pFC (BA 11) 52 −12 −7.33 
Right posterior cingulate (BA 31) −56 28 −5.69 
Right inferior parietal lobule (BA 40) 64 −52 24 −4.98 

Regions indicate the gyral locations and BA of the cluster peak. The MNI coordinates were converted into Talairach coordinates using the mni2tal script (http://eeg.sourceforge.net/mridoc/mri_toolbox/mni2tal.html). Gyral locations and BA were then determined by reference to Talairach and Tournoux (1988). x, y, and z indicate voxel coordinates in the MNI space. BSR represents each voxel's PLS parameter estimate divided by its standard error.

Although the foregoing LV describes a “canonical” network underlying verb generation, it does not account for variations in performance on the clinical BNT measure. Each of the following LVs identified group differences with respect to regions in which activation was correlated with BNT performance. Thus, we identified three distinct patterns of brain activity that accounted for variation in naming within a cohort of participants. Regions that are positively correlated with BNT scores in all the following LVs will be highlighted in blue in the corresponding figures.

In the second significant LV (p < .001, explained covariance = 21%; Figure 3; for a list of local maxima, see Table 4), regions highlighted in blue were more active during verb generation in controls and right mTLE patients and positively correlated with BNT scores in controls. These dominant saliences were located in the left inferior pFC (Broca's area, anterior to the region identified in LV1), left dorsolateral pFC, left premotor cortex, left temporal pole, and bilateral cerebellum. The regions highlighted in yellow are more active during baseline in controls and right mTLE patients and positively correlated with BA 44 activity in all groups. This involved the right posterior cingulate and the precuneus; when the activity in the BA 44 network increased, so did the activity in the posterior cingulate/precuneus. This seems counterintuitive as one would expect this region (part of the “default network”) to be suppressed in relation to task performance (as the task relationship demonstrates). As discussed below, this may reflect a “maladaptive” pattern in that for controls, individuals showing less suppression during verb generation had poorer scores on the BNT.

Figure 3. 

(A) Brain scores bar graph, (B) correlation bar graph, and (C) singular image for LV2 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph captures the mean brain score for each condition in each group and indicates reliable activity differences between verb generation and baseline for control and right mTLE groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations. The error bar crosses zero for BNT scores in both mTLE groups, indicating that the correlation between BNT scores and the areas identified in the singular image is nonreliable for these groups. (C) The singular image shows brain–seed correlations for control and mTLE groups, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Regions in blue are more active during verb generation in controls and right mTLE patients and positively correlated with BNT scores in controls. Regions in yellow are more active during baseline in controls and right mTLE patients and positively correlated with BA 44 activity in all groups.

Figure 3. 

(A) Brain scores bar graph, (B) correlation bar graph, and (C) singular image for LV2 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph captures the mean brain score for each condition in each group and indicates reliable activity differences between verb generation and baseline for control and right mTLE groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations. The error bar crosses zero for BNT scores in both mTLE groups, indicating that the correlation between BNT scores and the areas identified in the singular image is nonreliable for these groups. (C) The singular image shows brain–seed correlations for control and mTLE groups, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Regions in blue are more active during verb generation in controls and right mTLE patients and positively correlated with BNT scores in controls. Regions in yellow are more active during baseline in controls and right mTLE patients and positively correlated with BA 44 activity in all groups.

Table 4. 

Local Maxima for the Three-Group Task–Seed–Behavior PLS LV2

Region
x (mm)
y (mm)
z (mm)
BSR
Right posterior cingulate (BA 31) 12 −44 28 13.14 
Precuneus (BA 7) −52 48 11.57 
Left inferior pFC (BA 45) −52 40 −6.58 
Left dorsolateral pFC (BA 9) −52 40 −5.51 
Left premotor cortex (BA 6) −44 36 −5.08 
Left temporal pole (BA 38) −56 20 −12 −6.5 
Left cerebellum −40 −84 −24 −3.9 
Right cerebellum 36 −72 −40 −6.74 
Region
x (mm)
y (mm)
z (mm)
BSR
Right posterior cingulate (BA 31) 12 −44 28 13.14 
Precuneus (BA 7) −52 48 11.57 
Left inferior pFC (BA 45) −52 40 −6.58 
Left dorsolateral pFC (BA 9) −52 40 −5.51 
Left premotor cortex (BA 6) −44 36 −5.08 
Left temporal pole (BA 38) −56 20 −12 −6.5 
Left cerebellum −40 −84 −24 −3.9 
Right cerebellum 36 −72 −40 −6.74 

Regions indicate the gyral locations and BA of the cluster peak. MNI coordinates were converted into Talairach coordinates using the mni2tal script (http://eeg.sourceforge.net/mridoc/mri_toolbox/mni2tal.html). Gyral locations and BA were then determined by reference to Talairach and Tournoux (1988). x, y, and z indicate voxel coordinates in MNI space. BSR represents each voxel's PLS parameter estimate divided by its standard error.

The third significant LV identified a subset of brain regions that were positively correlated with BNT scores during verb generation only in the right mTLE group (p < .005, explained covariance = 8%; Figure 4; for a list of local maxima, see Table 5). These regions were not functionally connected with BA 44 and were not related to activity differences between verb generation and baseline in any group. Thus, they relate specifically to residual variance in BNT performance in the right mTLE group. Dominant negative saliences (related to positive correlations with BNT scores in the right mTLE group) were located in the left ventrolateral pFC, right caudate, right superior temporal cortex, and left posterior cingulate. The only significant salience related to negative correlations with BNT scores in the right mTLE group was in the left cerebellum.

Figure 4. 

(A) Brain scores bar graph, (B) correlation bar graph, and (C) singular image for LV3 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph indicates that this LV shows no reliable activity differences between conditions or groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations. The error bar crosses zero for all measures except BNT scores in the right mTLE group, indicating that the areas identified in the singular image depict only BNT scores–brain correlations for the right mTLE group. (C) The singular image shows brain–seed correlations for the right mTLE group, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Regions in blue are positively correlated with BNT scores in right mTLE patients, and regions in yellow are anticorrelated with BNT scores for this group.

Figure 4. 

(A) Brain scores bar graph, (B) correlation bar graph, and (C) singular image for LV3 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph indicates that this LV shows no reliable activity differences between conditions or groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations. The error bar crosses zero for all measures except BNT scores in the right mTLE group, indicating that the areas identified in the singular image depict only BNT scores–brain correlations for the right mTLE group. (C) The singular image shows brain–seed correlations for the right mTLE group, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Regions in blue are positively correlated with BNT scores in right mTLE patients, and regions in yellow are anticorrelated with BNT scores for this group.

Table 5. 

Local Maxima for the Three-group Task–Seed–Behavior PLS LV3

Region
x (mm)
y (mm)
z (mm)
BSR
Left cerebellum −36 −76 −24 7.94 
Left ventrolateral pFC (BA 47) −52 28 −8 −5.4 
Right caudate 12 −9.53 
Right superior temporal cortex (BA 22) 68 −44 12 −8.48 
Left posterior cingulate (BA 32) −16 −56 20 −7.46 
Region
x (mm)
y (mm)
z (mm)
BSR
Left cerebellum −36 −76 −24 7.94 
Left ventrolateral pFC (BA 47) −52 28 −8 −5.4 
Right caudate 12 −9.53 
Right superior temporal cortex (BA 22) 68 −44 12 −8.48 
Left posterior cingulate (BA 32) −16 −56 20 −7.46 

Regions indicate the gyral locations and BA of the cluster peak. MNI coordinates were converted into Talairach coordinates using the mni2tal script (http://eeg.sourceforge.net/mridoc/mri_toolbox/mni2tal.html). Gyral locations and BA were then determined by reference to Talairach and Tournoux (1988). x, y, and z indicate voxel coordinates in MNI space. BSR represents each voxel's PLS parameter estimate divided by its standard error.

The fourth significant LV identified a subset of brain regions that were positively correlated with BNT scores during verb generation only in the left mTLE group (p < .001, explained covariance = 7%; Figure 5; for a list of local maxima, see Table 6). These regions were not functionally connected with BA 44 and were not related to activity differences between verb generation and baseline in any group. Rather, like LV3 above, these regions are related to variance in the left mTLE group's BNT performance. Regions highlighted in blue that were positively correlated with BNT scores in the left mTLE group were located in the left inferior temporal gyrus, left lingual gyrus, left occipital gyrus, and right middle temporal gyrus. Regions highlighted in yellow were negatively correlated with BNT scores in this patient group, and these were located in the left inferior frontal gyrus (BA 44, superior to the seed) and right medial pFC.

Figure 5. 

(A) Brain scores bar graph, (B) correlation bar graph, and (C) singular image for LV4 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph indicates that this LV shows no reliable activity differences between conditions or groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations. The error bar crosses zero for all measures except BNT scores in the left mTLE group, indicating that the areas identified in the singular image depict only BNT scores–brain correlations for the left mTLE group. (C) The singular image shows brain–seed correlations for the left mTLE group, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Regions in blue are positively correlated with BNT scores in left mTLE patients, and regions in yellow are anticorrelated with BNT scores for this group.

Figure 5. 

(A) Brain scores bar graph, (B) correlation bar graph, and (C) singular image for LV4 from the three-group task–seed–behavior PLS. (A) The brain scores bar graph indicates that this LV shows no reliable activity differences between conditions or groups. (B) The correlation bar graph captures the group-dependent changes in the BA 44 voxel–brain and BNT scores–brain correlations. The error bar crosses zero for all measures except BNT scores in the left mTLE group, indicating that the areas identified in the singular image depict only BNT scores–brain correlations for the left mTLE group. (C) The singular image shows brain–seed correlations for the left mTLE group, displayed on axial slices from −28 to 40 in the MNI atlas space. The brain is displayed according to neurological convention (L = L). Regions in blue are positively correlated with BNT scores in left mTLE patients, and regions in yellow are anticorrelated with BNT scores for this group.

Table 6. 

Local Maxima for the Three-group Task–Seed–Behavior PLS LV4

Region
x (mm)
y (mm)
z (mm)
BSR
Left inferior frontal gyrus (BA 44) −60 32 4.63 
Right medial pFC (BA 10) 12 48 5.22 
Left inferior temporal gyrus (BA 21) −56 −16 −16 −5.99 
Left lingual gyrus (BA 18) −20 −80 −4 −5.51 
Left occipital gyrus (BA 19) −44 −84 −8 −5.21 
Right middle temporal gyrus (BA 39) 48 −78 −5.82 
Region
x (mm)
y (mm)
z (mm)
BSR
Left inferior frontal gyrus (BA 44) −60 32 4.63 
Right medial pFC (BA 10) 12 48 5.22 
Left inferior temporal gyrus (BA 21) −56 −16 −16 −5.99 
Left lingual gyrus (BA 18) −20 −80 −4 −5.51 
Left occipital gyrus (BA 19) −44 −84 −8 −5.21 
Right middle temporal gyrus (BA 39) 48 −78 −5.82 

Regions indicate the gyral locations and BA of the cluster peak. MNI coordinates were converted into Talairach coordinates using the mni2tal script (http://eeg.sourceforge.net/mridoc/mri_toolbox/mni2tal.html). Gyral locations and BA were then determined by reference to Talairach and Tournoux (1988). x, y, and z indicate voxel coordinates in MNI space. BSR represents each voxel's PLS parameter estimate divided by its standard error.

The three-group task–seed–behavior PLS revealed group differences in three significant LVs (LV2, LV3, and LV4). To ensure that these group differences were not due to differing levels of education between patients and controls, we computed Pearson correlations between education and brain scores (recall that brain scores indicate the degree to which each subject shows the effect expressed in the LV) (Grady et al., 2010). None of these correlations were statistically significant. In addition, to ensure that group differences were not due to differing sex distributions between patients and controls, we grouped brain scores from LV2, LV3, and LV4 on the basis of sex. We performed t tests comparing these groups and found no significant differences.

DISCUSSION

We hypothesized that word retrieval difficulties in left mTLE patients occur because localized malfunctioning epileptogenic tissue adversely influences remote cerebral structures and expected that the networks that support these tasks would be altered in left mTLE patients. We measured brain activity with fMRI while participants performed a language task (verb generation) that did not activate the MTL, and we examined language-related functional connectivity during verb generation in relation to language performance on a clinical test of lexical access (BNT). We did not intend for BNT to stand as a “proxy” for verb generation because it is clear that the two tasks draw upon some different linguistic processes. Rather, we attempted to use the “gold standard” in each domain. BNT is considered as a “canonical” language competency task in epilepsy clinical research, and verb generation has been shown to be the most reliable fMRI paradigm for localizing and lateralizing language-related activation (Szaflarski et al., 2008; Fernandes, Smith, Logan, Crawley, & McAndrews, 2006; Rutten, Ramsey, van Rijen, Alpherts, & van Veelen, 2002; Benson et al., 1999).

Using task–seed–behavior PLS, we identified a canonical language network that was more active during verb generation than the baseline condition and was functionally connected with BA 44 (the brain region with the strongest task-relevant activity). This network, which is similar to that shown in univariate analyses in other functional imaging studies of lexical access during verb generation (Szaflarski et al., 2008; Powell et al., 2007; Holland et al., 2001), did not vary by group. Engagement of this network is undoubtedly crucial for word retrieval as focal lesions in most of the components is associated with aphasia. However, in the current study, variability in naming performance within and across participant groups cannot be accounted for by this network alone. Rather, our data demonstrate that recruitment of additional regions, which vary by group, accounts for variation in BNT performance. In controls, these regions were mainly anterior (prefrontal and anterior temporal; LV2 from the task–seed–behavior PLS). In the left mTLE group, patients with higher BNT scores recruited more posterior regions (e.g., posterior temporal and occipital; LV4 from the task–seed–behavior PLS). Finally, in the right mTLE patients, patients with higher BNT scores recruited some anterior regions like controls (e.g., prefrontal) but also more posterior regions (e.g., posterior temporal and posterior cingulate; LV3 from the task–seed–behavior PLS).

Verb Generation Activity and BA 44 Functional Connectivity

There are a few previous studies of functional connectivity in mTLE patients in relation to language performance. Consistent with the findings of Waites et al. (2006), we found no significant differences in brain activation during verb generation versus baseline between our patient and control groups. However, contrary to the findings of Waites et al. and Azari et al. (1999) of differential functional connectivity for language networks between patients and controls, we did not identify a major group difference in our data set. That is, LV1 in the task–seed–behavior analysis revealed similar connectivity patterns with the BA 44 seed. These discrepancies may be due to the fact that we examined language-related functional connectivity during verb generation in relation to language performance, whereas both the previous functional connectivity studies examined language networks identified by resting state connectivity. Where “baseline” connectivity may be weak, our findings indicate that patients can engage the canonical network appropriately under task demands. In relating language-relevant activation to off-line behavior, we found that the primary difference between patients and controls was that the recruitment of additional, more posterior regions in mTLE patients accounted for performance variation.

Regional Activation and Naming Performance

The second LV from task–seed–behavior PLS indicates that several regions are recruited in addition to the BA 44 network associated with verb generation by those control participants with high BNT scores. These regions include the left inferior pFC (BA 45), the left dorsolateral pFC (BA 9), and the left temporal pole. These regions fit with what one would expect from both functional imaging and lesion data. For example, a recent meta-analysis of functional imaging studies shows these regions to be part of a general semantic system (Binder, Desai, Graves, & Conant, 2009). Importantly, in patient populations (e.g., semantic dementia, mTLE), there is clear evidence for a relationship between word retrieval deficits and structural damage or metabolic insufficiency in temporal polar and lateral prefrontal regions (Rami et al., 2008; Brambati et al., 2006; Hodges & McCarthy, 1995). Furthermore, regional white matter changes in these areas are correlated with naming deficits in patients with mTLE (McDonald et al., 2008).

The primary difference seen in the mTLE groups relative to controls was the additional recruitment of posterior regions in the networks supporting naming performance. Specifically, variation in recruitment of posterior temporal regions during our language activation paradigm was related to successful BNT performance in both left and right mTLE. Patients with right mTLE recruited right posterior temporal regions in addition to the more anterior ones also shown in controls. Patients with left mTLE recruited both right posterior temporal and a more caudal and inferior left temporal region in comparison to controls. Of interest, hypometabolism in left posterior temporal cortex in resting state has been linked to anomic states in these patients (Bonilha & Halford, 2009; Riederer et al., 2008), which parallels our finding that patients who can show increased recruitment of this region perform better on word retrieval tasks. These results also converge with a study by Hamberger et al. (2007), which showed that increased left MTL damage in mTLE (i.e., patients with left hippocampal sclerosis as compared with patients with structurally normal hippocampi) had a more posterior distribution of language sites on stimulation mapping with a naming task. It is possible that the reason that the direction of reorganization is posterior links back to diaschisis. Seizure activity in mTLE is more likely to arise from anterior than posterior hippocampus (King, Bronen, Spencer, & Spencer, 1997) and interictal discharges more frequently propagate anteriorly than posteriorly (Emerson, Turner, Pedley, Walczak, & Forgione, 1995), likely in relation to strong anatomical connectivity through pathways such as the uncinate fasciculus. Therefore, posterior reorganization may be a consequence of the MTL–cortical connections and the patterns of EEG discharges propagating from an epileptogenic medial-temporal region. Posterior reorganization may also reflect a reliance on different task performance strategies. For example, patients may have used visual imagery rather than performing the task with strictly linguistic resources. Additional studies would be required to be confident of either explanation.

Functional Connectivity of Task and Default Mode Networks

One seemingly counterintuitive finding is the positive correlation between posterior parts of the default mode network (cingulate and precuneus) and task-relevant activation in controls (second LV from task–seed–behavior PLS). Consistent with previous research on default mode activity, these regions were more active during baseline than verb generation (for a review, see Buckner, Andrews-Hanna, & Schacter, 2008). However, contrary to expectation for the task-independent default network, these regions were functionally connected with the BA 44 seed, and thus a part of a network identified during verb generation. Of importance, this LV also identified an anticorrelation between functional coupling with BA 44 and off-line naming performance. Thus, the more engaged cingulate region is in the BA 44 network, and the worst is in the off-line naming performance.

We speculate that this result may reflect a “maladaptive” network, expressed here only in controls, such that individuals who are less effective at suppressing default mode activity during expressive language tasks have generally poorer performance in word retrieval. This is consistent with recent findings indicating that the efficiency with which activation “toggles” between active and default states can be an index of “healthy” of brain function (Dickerson & Sperling, 2009; Frings, Schulze-Bonhage, Spreer, & Wagner, 2009; Pihlajamaki & Sperling, 2009). For example, Pihlajamaki and Sperling (2009) reported that the pattern of task-induced deactivation is progressively disrupted along a continuum from normal aging through mild cognitive impairment to Alzheimer's disease. However, we do not have direct evidence of how this hypothesized reduction in task-related deactivation may have affected the on-line language task (i.e., verb generation), and so this interpretation of our data remains speculative.

Conclusions

Our findings suggest that in the context of longstanding seizures in both dominant and nondominant temporal lobe, focal epileptogenic tissue adversely influences remote cerebral structures, causing language networks in mTLE patients to undergo functional reorganization. This general conclusion is quite compatible with findings of previous functional imaging studies demonstrating abnormal patterns of task-related activation in mTLE patients (for reviews, see Szaflarski et al., 2008; Powell & Duncan, 2005). Here, we show widespread differences in intrahemispheric organization supporting language, a key feature that is overlooked in articles focusing on atypical lateralization in mTLE (Briellmann et al., 2006; Sveller et al., 2006; Yuan et al., 2006). Importantly, our data confirm that these network abnormalities account not only for group differences in word retrieval during the activation task but also for performance variation between mTLE patients on clinically relevant measures of lexical access. Tying clinically relevant measures to network reorganization provides us with a powerful tool to investigate individual variability in the context of similar focal damage. Eventually, it might provide a more powerful tool for predicting postoperative changes in language performance.

Acknowledgments

The authors thank Filomeno Cortese for his help with data illustration. This work was supported by a James S. McDonnell Foundation grant to M. P. M.

Reprint requests should be sent to Andrea B. Protzner, Department of Neuropsychology, Krembil Neuroscience Centre, University Health Network, 399 Bathurst St., Room 4F-409, Toronto, ON, Canada M5T 2S8, or via e-mail: andrea.protzner@gmail.com.

Note

1. 

TR and voxel size remained unchanged, regardless of the number of slices acquired (i.e., 28 or 32). When acquiring 32 slices, less time was spent acquiring each slice than when acquiring 28 slices.

REFERENCES

Alessio
,
A.
,
Bonilha
,
L.
,
Rorden
,
C.
,
Kobayashi
,
E.
,
Min
,
L. L.
,
Damasceno
,
B. P.
,
et al
(
2006
).
Memory and language impairments and their relationships to hippocampal and perirhinal cortex damage in patients with medial temporal lobe epilepsy.
Epilepsy and Behavior
,
8
,
593
600
.
Arnold
,
S.
,
Schlaug
,
G.
,
Niemann
,
H.
,
Ebner
,
A.
,
Luders
,
H.
,
Witte
,
O. W.
,
et al
(
1996
).
Topography of interictal glucose hypometabolism in unilateral mesiotemporal epilepsy.
Neurology
,
46
,
1422
1430
.
Azari
,
N. P.
,
Knorr
,
U.
,
Arnold
,
S.
,
Antke
,
C.
,
Ebner
,
A.
,
Niemann
,
H.
,
et al
(
1999
).
Reorganized cerebral metabolic interactions in temporal lobe epilepsy.
Neuropsychologia
,
37
,
625
636
.
Benson
,
R. R.
,
FitzGerald
,
D. B.
,
LeSueur
,
L. L.
,
Kennedy
,
D. N.
,
Kwong
,
K. K.
,
Buchbinder
,
B. R.
,
et al
(
1999
).
Language dominance determined by whole brain functional MRI in patients with brain lesions.
Neurology
,
52
,
798
809
.
Billingsley
,
R. L.
,
McAndrews
,
M. P.
,
Crawley
,
A. P.
, &
Mikulis
,
D. J.
(
2001
).
Functional MRI of phonological and semantic processing in temporal lobe epilepsy.
Brain
,
124
,
1218
1227
.
Binder
,
J. R.
,
Desai
,
R. H.
,
Graves
,
W. W.
, &
Conant
,
L. L.
(
2009
).
Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies.
Cerebral Cortex
,
19
,
2767
2796
.
Bonilha
,
L.
, &
Halford
,
J. J.
(
2009
).
Network atrophy in temporal lobe epilepsy: A voxel-based morphometry study [Author reply 2052].
Neurology
,
72
,
2052
.
Brambati
,
S. M.
,
Myers
,
D.
,
Wilson
,
A.
,
Rankin
,
K. P.
,
Allison
,
S. C.
,
Rosen
,
H. J.
,
et al
(
2006
).
The anatomy of category-specific object naming in neurodegenerative diseases.
Journal of Cognitive Neuroscience
,
18
,
1644
1653
.
Briellmann
,
R. S.
,
Labate
,
A.
,
Harvey
,
A. S.
,
Saling
,
M. M.
,
Sveller
,
C.
,
Lillywhite
,
L.
,
et al
(
2006
).
Is language lateralization in temporal lobe epilepsy patients related to the nature of the epileptogenic lesion?
Epilepsia
,
47
,
916
920
.
Buckner
,
R. L.
,
Andrews-Hanna
,
J. R.
, &
Schacter
,
D. L.
(
2008
).
The brain's default network: Anatomy, function, and relevance to disease.
Annals of the New York Academy of Sciences
,
1124
,
1
38
.
Busch
,
R. M.
,
Frazier
,
T. W.
,
Iampietro
,
M. C.
,
Chapin
,
J. S.
, &
Kubu
,
C. S.
(
2009
).
Clinical utility of the Boston Naming Test in predicting ultimate side of surgery in patients with medically intractable temporal lobe epilepsy: A double cross-validation study.
Epilepsia
,
50
,
1270
1273
.
Calhoun
,
V. D.
,
Kiehl
,
K. A.
, &
Pearlson
,
G. D.
(
2008
).
Modulation of temporally coherent brain networks estimated using ICA at rest and during cognitive tasks.
Human Brain Mapping
,
29
,
828
838
.
Caplan
,
J. B.
,
Luks
,
T. L.
,
Simpson
,
G. V.
,
Glaholt
,
M.
, &
McIntosh
,
A. R.
(
2006
).
Parallel networks operating across attentional deployment and motion processing: A multi-seed partial least squares fMRI study.
Neuroimage
,
29
,
1192
1202
.
Cendes
,
F.
,
Caramanos
,
Z.
,
Andermann
,
F.
,
Dubeau
,
F.
, &
Arnold
,
D. L.
(
1997
).
Proton magnetic resonance spectroscopic imaging and magnetic resonance imaging volumetry in the lateralization of temporal lobe epilepsy: A series of 100 patients.
Annals of Neurology
,
42
,
737
746
.
Cox
,
R. W.
(
1996
).
AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages.
Computers in Biomedical Research
,
29
,
162
173
.
Davies
,
K. G.
,
Bell
,
B. D.
,
Bush
,
A. J.
,
Hermann
,
B. P.
,
Dohan
,
F. C.
, Jr., &
Jaap
,
A. S.
(
1998
).
Naming decline after left anterior temporal lobectomy correlates with pathological status of resected hippocampus.
Epilepsia
,
39
,
407
419
.
Dickerson
,
B. C.
, &
Sperling
,
R. A.
(
2009
).
Large-scale functional brain network abnormalities in Alzheimer's disease: Insights from functional neuroimaging.
Behavioural Neurology
,
21
,
63
75
.
Emerson
,
R. G.
,
Turner
,
C. A.
,
Pedley
,
T. A.
,
Walczak
,
T. S.
, &
Forgione
,
M.
(
1995
).
Propagation patterns of temporal spikes.
Electroencephalography and Clinical Neurophysiology
,
94
,
338
348
.
Engel
,
J.
, Jr.,
Bandler
,
R.
,
Griffith
,
N. C.
, &
Caldecott-Hazard
,
S.
(
1991
).
Neurobiological evidence for epilepsy-induced interictal disturbances.
Advances in Neurology
,
55
,
97
111
.
Fernandes
,
M. A.
,
Smith
,
M. L.
,
Logan
,
W.
,
Crawley
,
A.
, &
McAndrews
,
M. P.
(
2006
).
Comparing language lateralization determined by dichotic listening and fMRI activation in frontal and temporal lobes in children with epilepsy.
Brain and Language
,
96
,
106
114
.
Focke
,
N. K.
,
Yogarajah
,
M.
,
Bonelli
,
S. B.
,
Bartlett
,
P. A.
,
Symms
,
M. R.
, &
Duncan
,
J. S.
(
2008
).
Voxel-based diffusion tensor imaging in patients with mesial temporal lobe epilepsy and hippocampal sclerosis.
Neuroimage
,
40
,
728
737
.
Frings
,
L.
,
Schulze-Bonhage
,
A.
,
Spreer
,
J.
, &
Wagner
,
K.
(
2009
).
Remote effects of hippocampal damage on default network connectivity in the human brain.
Journal of Neurology
,
256
,
2021
2029
.
Frisk
,
V.
, &
Milner
,
B.
(
1990
).
The role of the left hippocampal region in the acquisition and retention of story content.
Neuropsychologia
,
28
,
349
359
.
Gleissner
,
U.
, &
Elger
,
C. E.
(
2001
).
The hippocampal contribution to verbal fluency in patients with temporal lobe epilepsy.
Cortex
,
37
,
55
63
.
Grady
,
C. L.
,
Protzner
,
A. B.
,
Kovacevic
,
N.
,
Strother
,
S. C.
,
Afshin-Pour
,
B.
,
Wojtowicz
,
M.
,
et al
(
2010
).
A multivariate analysis of age-related differences in default mode and task-positive networks across multiple cognitive domains.
Cerebral Cortex
,
20
,
1432
1447
.
Hamberger
,
M. J.
,
Seidel
,
W. T.
,
Goodman
,
R. R.
,
Williams
,
A.
,
Perrine
,
K.
,
Devinsky
,
O.
,
et al
(
2007
).
Evidence for cortical reorganization of language in patients with hippocampal sclerosis.
Brain
,
130
,
2942
2950
.
Henry
,
T. R.
,
Mazziotta
,
J. C.
,
Engel
,
J.
, Jr.
(
1993
).
Interictal metabolic anatomy of mesial temporal lobe epilepsy.
Archives of Neurology
,
50
,
582
589
.
Hermann
,
B. P.
,
Wyler
,
A. R.
,
Somes
,
G.
,
Dohan
,
F. C.
, Jr.,
Berry
,
A. D.
, III, &
Clement
,
L.
(
1994
).
Declarative memory following anterior temporal lobectomy in humans.
Behavioral Neuroscience
,
108
,
3
10
.
Hodges
,
J. R.
, &
McCarthy
,
R. A.
(
1995
).
Loss of remote memory: A cognitive neuropsychological perspective.
Current Opinion in Neurobiology
,
5
,
178
183
.
Holland
,
S. K.
,
Plante
,
E.
,
Weber Byars
,
A.
,
Strawsburg
,
R. H.
,
Schmithorst
,
V. J.
,
Ball
,
W. S.
, Jr.
(
2001
).
Normal fMRI brain activation patterns in children performing a verb generation task.
Neuroimage
,
14
,
837
843
.
Janszky
,
J.
,
Mertens
,
M.
,
Janszky
,
I.
,
Ebner
,
A.
, &
Woermann
,
F. G.
(
2006
).
Left-sided interictal epileptic activity induces shift of language lateralization in temporal lobe epilepsy: An fMRI study.
Epilepsia
,
47
,
921
927
.
Jokeit
,
H.
,
Seitz
,
R. J.
,
Markowitsch
,
H. J.
,
Neumann
,
N.
,
Witte
,
O. W.
, &
Ebner
,
A.
(
1997
).
Prefrontal asymmetric interictal glucose hypometabolism and cognitive impairment in patients with temporal lobe epilepsy.
Brain
,
120
,
2283
2294
.
Kaplan
,
E. F.
,
Goodglass
,
H.
, &
Weintraub
,
S.
(
1983
).
The Boston Naming Test
(2nd ed.).
Philadelphia, PA
:
Lea & Febiger
.
King
,
D.
,
Bronen
,
R. A.
,
Spencer
,
D. D.
, &
Spencer
,
S. S.
(
1997
).
Topographic distribution of seizure onset and hippocampal atrophy: Relationship between MRI and depth EEG.
Electroencephalography and Clinical Neurophysiology
,
103
,
692
697
.
Loring
,
D. W.
,
Strauss
,
E.
,
Hermann
,
B. P.
,
Barr
,
W. B.
,
Perrine
,
K.
,
Trenerry
,
M. R.
,
et al
(
2008
).
Differential neuropsychological test sensitivity to left temporal lobe epilepsy.
Journal of the International Neuropsychological Society
,
14
,
394
400
.
McDonald
,
C. R.
,
Ahmadi
,
M. E.
,
Hagler
,
D. J.
,
Tecoma
,
E. S.
,
Iragui
,
V. J.
,
Gharapetian
,
L.
,
et al
(
2008
).
Diffusion tensor imaging correlates of memory and language impairments in temporal lobe epilepsy.
Neurology
,
71
,
1869
1876
.
McIntosh
,
A. R.
,
Bookstein
,
F. L.
,
Haxby
,
J. V.
, &
Grady
,
C. L.
(
1996
).
Spatial pattern analysis of functional brain images using partial least squares.
Neuroimage
,
3
,
143
157
.
McIntosh
,
A. R.
,
Cabeza
,
R. E.
, &
Lobaugh
,
N. J.
(
1998
).
Analysis of neural interactions explains the activation of occipital cortex by an auditory stimulus.
Journal of Neurophysiology
,
80
,
2790
2796
.
Messas
,
C. S.
,
Mansur
,
L. L.
, &
Castro
,
L. H.
(
2008
).
Semantic memory impairment in temporal lobe epilepsy associated with hippocampal sclerosis.
Epilepsy and Behavior
,
12
,
311
316
.
Miller
,
L. A.
,
Munoz
,
D. G.
, &
Finmore
,
M.
(
1993
).
Hippocampal sclerosis and human memory.
Archives of Neurology
,
50
,
391
394
.
Milner
,
B.
(
1971
).
Interhemispheric differences in the localization of psychological processes in man.
British Medical Bulletin
,
27
,
272
277
.
N'Kaoua
,
B.
,
Lespinet
,
V.
,
Barsse
,
A.
,
Rougier
,
A.
, &
Claverie
,
B.
(
2001
).
Exploration of hemispheric specialization and lexico-semantic processing in unilateral temporal lobe epilepsy with verbal fluency tasks.
Neuropsychologia
,
39
,
635
642
.
Pihlajamaki
,
M.
, &
Sperling
,
R. A.
(
2009
).
Functional MRI assessment of task-induced deactivation of the default mode network in Alzheimer's disease and at-risk older individuals.
Behavioural Neurology
,
21
,
77
91
.
Powell
,
H. W.
, &
Duncan
,
J. S.
(
2005
).
Functional magnetic resonance imaging for assessment of language and memory in clinical practice.
Current Opinion in Neurology
,
18
,
161
166
.
Powell
,
H. W.
,
Parker
,
G. J.
,
Alexander
,
D. C.
,
Symms
,
M. R.
,
Boulby
,
P. A.
,
Wheeler-Kingshott
,
C. A.
,
et al
(
2007
).
Abnormalities of language networks in temporal lobe epilepsy.
Neuroimage
,
36
,
209
221
.
Rami
,
L.
,
Caprile
,
C.
,
Gomez-Anson
,
B.
,
Sanchez-Valle
,
R.
,
Monte
,
G. C.
,
Bosch
,
B.
,
et al
(
2008
).
Naming is associated with left temporal pole metabolite levels in neurodegenerative diseases.
Dementia and Geriatric Cognitive Disorders
,
25
,
212
217
.
Riederer
,
F.
,
Lanzenberger
,
R.
,
Kaya
,
M.
,
Prayer
,
D.
,
Serles
,
W.
, &
Baumgartner
,
C.
(
2008
).
Network atrophy in temporal lobe epilepsy: A voxel-based morphometry study.
Neurology
,
71
,
419
425
.
Rutten
,
G. J.
,
Ramsey
,
N. F.
,
van Rijen
,
P. C.
,
Alpherts
,
W. C.
, &
van Veelen
,
C. W.
(
2002
).
fMRI-determined language lateralization in patients with unilateral or mixed language dominance according to the Wada test.
Neuroimage
,
17
,
447
460
.
Sampson
,
P. D.
,
Streissguth
,
A. P.
,
Barr
,
H. M.
, &
Bookstein
,
F. L.
(
1989
).
Neurobehavioral effects of prenatal alcohol: Part II. Partial least squares analysis.
Neurotoxicology and Teratology
,
11
,
477
491
.
Savic
,
I.
,
Altshuler
,
L.
,
Baxter
,
L.
,
Engel
,
J.
, Jr.
(
1997
).
Pattern of interictal hypometabolism in PET scans with fludeoxyglucose F 18 reflects prior seizure types in patients with mesial temporal lobe seizures.
Archives of Neurology
,
54
,
129
136
.
Schreurs
,
B. G.
,
McIntosh
,
A. R.
,
Bahro
,
M.
,
Herscovitch
,
P.
,
Sunderland
,
T.
, &
Molchan
,
S. E.
(
1997
).
Lateralization and behavioral correlation of changes in regional cerebral blood flow with classical conditioning of the human eyeblink response.
Journal of Neurophysiology
,
77
,
2153
2163
.
Sveller
,
C.
,
Briellmann
,
R. S.
,
Saling
,
M. M.
,
Lillywhite
,
L.
,
Abbott
,
D. F.
,
Masterton
,
R. A.
,
et al
(
2006
).
Relationship between language lateralization and handedness in left-hemispheric partial epilepsy.
Neurology
,
67
,
1813
1817
.
Szaflarski
,
J. P.
,
Holland
,
S. K.
,
Jacola
,
L. M.
,
Lindsell
,
C.
,
Privitera
,
M. D.
, &
Szaflarski
,
M.
(
2008
).
Comprehensive presurgical functional MRI language evaluation in adult patients with epilepsy.
Epilepsy and Behavior
,
12
,
74
83
.
Talairach
,
J.
, &
Tournoux
,
P.
(
1988
).
Co-planar sterotaxic atlas of the human brain.
New York
:
Thieme
.
Theodore
,
W. H.
,
Newmark
,
M. E.
,
Sato
,
S.
,
Brooks
,
R.
,
Patronas
,
N.
,
De La Paz
,
R.
,
et al
(
1983
).
[18F]fluorodeoxyglucose positron emission tomography in refractory complex partial seizures.
Annals of Neurology
,
14
,
429
437
.
Thivard
,
L.
,
Lehericy
,
S.
,
Krainik
,
A.
,
Adam
,
C.
,
Dormont
,
D.
,
Chiras
,
J.
,
et al
(
2005
).
Diffusion tensor imaging in medial temporal lobe epilepsy with hippocampal sclerosis.
Neuroimage
,
28
,
682
690
.
Trebuchon-Da Fonseca
,
A.
,
Guedj
,
E.
,
Alario
,
F. X.
,
Laguitton
,
V.
,
Mundler
,
O.
,
Chauvel
,
P.
,
et al
(
2009
).
Brain regions underlying word finding difficulties in temporal lobe epilepsy.
Brain
,
132
,
2772
2784
.
Troster
,
A. I.
,
Warmflash
,
V.
,
Osorio
,
I.
,
Paolo
,
A. M.
,
Alexander
,
L. J.
, &
Barr
,
W. B.
(
1995
).
The roles of semantic networks and search efficiency in verbal fluency performance in intractable temporal lobe epilepsy.
Epilepsy Res
,
21
,
19
26
.
Waites
,
A. B.
,
Briellmann
,
R. S.
,
Saling
,
M. M.
,
Abbott
,
D. F.
, &
Jackson
,
G. D.
(
2006
).
Functional connectivity networks are disrupted in left temporal lobe epilepsy.
Annals of Neurology
,
59
,
335
343
.
Yuan
,
W.
,
Szaflarski
,
J. P.
,
Schmithorst
,
V. J.
,
Schapiro
,
M.
,
Byars
,
A. W.
,
Strawsburg
,
R. H.
,
et al
(
2006
).
fMRI shows atypical language lateralization in pediatric epilepsy patients.
Epilepsia
,
47
,
593
600
.