We studied 46 patients with medically refractory TLE [26 left (14 females); median age 41.5 years, range 17–63; 20 right (13 females); median age 34.5 years, range 23–52]. All underwent left or right ATLR at the National Hospital for Neurology and Neurosurgery, London. Preoperatively, all patients had undergone detailed presurgical evaluation including structural MRI at 3 T with qualitative assessment and quantification of hippocampal volumes and T2
relaxation times by expert neuroradiologists (Woermann et al., 1998
; Bartlett et al., 2007
), prolonged interictal and ictal video-EEG monitoring and standardized neuropsychological and psychiatric assessment. Structural MRI showed unilateral hippocampal sclerosis in 25 patients with left TLE and 16 patients with right TLE; one patient had a left and one patient a right medial dysembryoplastic neuroepithelial tumour, one had a right anterior temporal cavernoma, one had right anterior temporal focal cortical dysplasia and one patient showed a right anterior temporal ganglioglioma. All patients had normal contralateral medial temporal lobe structures on qualitative and quantitative MRI. Video-EEG confirmed ipsilateral seizure onset in all patients. All patients’ first language was English; handedness was determined using a standardized questionnaire (Oldfield, 1971
). All patients underwent language and memory functional MRI and standard neuropsychological assessment preoperatively and again 4 months after ATLR. Language dominance was assessed using a range of functional MRI tasks (Bonelli et al., 2011
) revealing left hemisphere dominance in 20 patients with left TLE and 13 patients with right TLE, and atypical (bilateral or right) language representation in six left and seven right patients with TLE. As described previously (Bonelli et al., 2010
), we additionally calculated pre- and postoperative lateralization indices for the contrast ‘verbal fluency’ using the Bootstrap method of the SPM toolbox (Wilke and Lidzba, 2007
) for each subject in the middle and inferior frontal gyri, which were used as covariates for the second level analysis. Mean verbal IQ as measured using the Wechsler Adult Intelligence Scale–III was 91.62 (SD 13.4) in left TLE and 93.7 (SD 14.7) in right TLE. Mean performance IQ was 97.9 (SD 13.4) in left and 92.0 (SD 13.0) in right TLE. As memory may be affected by anxiety and depression, all patients in this study were tested for co-morbid anxiety and depression preoperatively and again at the time of their postoperative assessment (4 months after ATLR) using the Hospital Anxiety and Depression Scale (HADS) as a measure of self reported symptoms of anxiety and depression (Zigmond and Snaith, 1983
The scale is a user-friendly, compact questionnaire comprising 14 items that assess current levels of anxiety and depression. The score is derived from responses on a four-point Likert-type scale. A score of 7 or above is considered positive. A clinically significant change was defined by a change in category; the different categories are defined as follows: normal (0–6), mild (7–10), moderate (11–13), severe (14 and above).
Preoperatively (data missing in two cases), there was no significant difference in anxiety scores between left and right TLE patients, but a significant difference in depression scores with patients with left TLE showing higher scores (anxiety median: left TLE, 7; range, 1–18; right TLE, 7.5; range, 4–15; depression median: left TLE, 5.5; range, 0–10; right TLE, 4.5; range, 0–15; P = 0.03). In right TLE, scores were within the pathological range in 11 patients for anxiety (eight mild, two moderate, one severe), and in two patients for depression (one mild, one moderate); in left TLE, scores were considered as positive in 13 patients for anxiety (five mild, five moderate, three severe) and eight patients for depression (eight mild).
Postoperatively, there was no significant difference in anxiety and depression scores between left and right TLE patients (anxiety median: left TLE, 6; range, 0–19; right TLE, 6.5; range, 0–16; depression median: left TLE, 3; range, 0–15; right TLE, 3; range, 0–13). In right TLE, postoperative scores were within the pathological range in 10 patients for anxiety (five mild, one moderate, four severe) and in five patients for depression (four mild, one severe); in left TLE, scores were considered as positive in 10 patients for anxiety (five mild, two moderate, three severe) and six patients for depression (five mild, one severe).
In right TLE there was no significant difference between pre- and postoperative anxiety or depression scores; in left TLE depression scores improved significantly after left ATLR compared to preoperatively (paired t-test; P = 0.0023), whereas there was no significant difference between pre- and postoperative anxiety scores.
There were no statistically significant correlations between pre- and postoperative anxiety or depression ratings and pre- and postoperative performance on verbal and visual memory tests (verbal learning and design learning) in left or right TLE patients and therefore pre- and postoperative anxiety and depression scores were not considered a factor in performance of verbal and visual memory tests in our patients and therefore not included as additional covariates.
All patients were treated with anti-epileptic medication, which mostly remained unchanged at the time of their postoperative assessment. The standard neurosurgical procedure was removal of the temporal pole, opening of the temporal horn, followed by en bloc
resection of the hippocampus with a posterior resection margin at the mid brainstem level. Postoperative seizure outcome was classified according to the International League Against Epilepsy classification (Wieser et al., 2001
) showing a seizure outcome grade 1 or 2 in 21 patients with left and 14 with right TLE and a seizure outcome grade 3 to 5 in five patients with left and six patients with right TLE. Seizure outcome is given at 1 year following surgery for all subjects.
The study was approved by the National Hospital for Neurology and Neurosurgery and the Institute of Neurology Joint Research Ethics Committee, and written informed consent was obtained from all subjects.
As described previously (Bonelli et al., 2010
) two learning tests, one verbal and one visual, which have been demonstrated to be good indicators of postoperative memory decline/performance (Baxendale et al., 2006
) were selected from our standard memory tests and were repeated 4 months after left or right ATLR.
In brief, subjects are read a list of 15 words five times during the verbal learning task and on each presentation they recall as many words as possible with the total number of correct words expressed as a percentage. Similarly, for the design learning test, patients are presented with a design five times with recall being tested after each presentation. Again, percentage of correct responses over the five trials was used as the measure of performance.
Patients completed neuropsychological tests before and 4 months after ATLR. Measures of change in verbal and design learning following surgery were calculated as postoperative − preoperative scores. Changes in scores and postoperative scores alone were correlated with pre- versus postoperative change in/ postoperative functional MRI activation patterns for left and right TLE patients. At the second level of analysis, patients were additionally divided into groups who suffered a clinically significant decline in verbal or visual memory using reliable change indices (Baxendale and Thompson, 2005
). These were defined as a change of 16% for verbal learning and 28% change for design learning (90% confidence interval).
Magnetic resonance data acquisition
MRI data acquisition was the same for all pre- and postoperative MRI studies, which were performed on a 3 T General Electric Excite HDx
scanner. Standard imaging gradients with a maximum strength of 40 m/Tm and slew rate 150 Tm/s were used. All data were acquired using an 8-channel array head coil for reception and the body coil for transmission. In addition to the functional MRI data, we acquired a high resolution echo planar image covering the whole brain with the following parameters for each subject: two shots, echo time 30 ms, repetition time 4500 ms, matrix 256 × 256, 88 contiguous 1.5 mm slices; the geometric distortions were matched by introducing an additional delay to increase the echo spacing (Boulby et al., 2005
For the functional MRI task, gradient-echo planar T2*-weighted images were acquired, providing blood oxygenation level-dependent contrast. Each volume comprised 44 contiguous 1.5 mm oblique axial slices through the temporal and frontal lobes, with a 24 cm field of view, 128 × 128 matrix and in-plane resolution of 1.88 × 1.88 mm. Echo time was 30 ms and repetition time 4.5 s. The field of view was positioned to cover the temporal lobe with the anterior–posterior axis aligned with the long axis of the hippocampus on sagittal views, and with the body of the hippocampus in the centre.
Memory functional magnetic resonance imaging paradigm and data analysis
In this postoperative follow-up study we applied the same memory paradigm as preoperatively (Bonelli et al., 2010
) using a parallel set of stimuli containing three different material types [Pictures (P), black and white nameable line drawn objects; Words (W), single concrete nouns; and Faces (F), partly black and white, partly coloured photographs unfamiliar to the subjects] in order to investigate postoperative verbal and visual memory encoding. In brief, a total of 210 stimuli were visually presented to the subjects during a single scanning session, one every 4 s in seven cycles. Each cycle consisted of a block of 10 pictures, 10 words and 10 faces (each lasting 40 s) followed by 20 s of crosshair fixation as a resting period. Subjects performed a deep encoding task which involved making a judgement on whether a stimulus was pleasant or unpleasant in order to encourage stimulus encoding, but the response type was not used in any subsequent parts of the functional MRI analysis. Sixty minutes after scanning subjects performed a recognition test outside the scanner comprising three blocks, one for each of the three material types. During the recognition tests the 70 stimuli for each material type were randomly mixed with additional 35 foils each and presented in a manner identical to that used during scanning. Subjects were instructed to indicate whether they could remember seeing each stimulus during scanning (R response) or whether it was new (N response).
The 210 encoding stimuli that had been presented during scanning were then classified according to the responses made during the recognition test. A correct (R) response indicated the stimulus was subsequently remembered while an incorrect (N) response indicated the stimulus was subsequently forgotten. For each of the three stimulus types (P, W, F), R and N responses were identified, giving a total of six event types: PR, PF, WR, WF, FR and FF. These were then entered as regressors in the design matrix.
Imaging data were analysed using Statistical Parametric Mapping (SPM5) (Friston et al., 1995
) (Wellcome Trust Centre for Imaging Neuroscience; http://www.fil.ion.ucl.ac.uk/spm
). The postoperative imaging time series of each subject was realigned using the mean image as a reference. Rigid body coregistration was used to coregister postoperative scans to the preoperative mean image; scans were then spatially normalized into standard space applying each subject’s preoperative spatial normalization parameters to the subject’s postoperative realigned and coregistered scans. Preoperatively, a scanner specific template created from 30 healthy control subjects, 15 patients with left and 15 patients with right hippocampal sclerosis was used for normalization. All scans were then smoothed with a Gaussian kernel of 10 mm full-width at half-maximum. Coregistration of postoperative scans was checked visually for each subject, three patients with unsatisfactory coregistration were excluded from further analysis.
To test for subsequent memory effects, an event-related analysis was used to compare encoding-related responses to individual stimuli that were subsequently remembered versus stimuli that were forgotten (Friston et al., 1998
; Mechelli et al., 2003
; Richardson et al., 2004
; Powell et al., 2005
; Seghier et al., 2012
). A two-level event–related random-effect analysis was employed. At the first level, for each subject trial-specific responses were modelled by convolving a delta function that indicated each event onset with the canonical haemodynamic response function to create regressors of interest, one for each of the six event types (PR, PF, WR, WF, FR and FF). Each subject’s movement parameters were included as confounds and parameter estimates pertaining to the height of the haemodynamic response function for each regressor of interest were calculated for each voxel. Three contrast images were created for each subject corresponding to the subsequent memory effect for each material type (picture encoding defined by PR-PF, word encoding defined by WR-WF and face encoding defined by FR-FF). Recognition accuracy for each event type was calculated in all our patients as follows: Stimuli seen in the recognition test were classified as ‘hits’ (stimuli correctly remembered) and ‘false alarms’ (foils incorrectly tagged as remembered). Recognition accuracy was then calculated for each stimulus type as: hit rate minus false alarm rate. All subjects with rates of <20% or >80% for the two possible responses ‘remembered’ and ‘forgotten’ were not included in this study as there were not enough responses in the different categories to ensure sufficient contrast. These contrast images were used for the second-level analysis.
At the second level of the random effects analysis, subjects were divided into two groups, patients with left and right TLE. Each subject’s contrast images were entered into a second level one sample t-test, which modelled the group effect (left and right TLE patients, pre- and postoperatively) on the various contrasts. In order to test for correlations between areas of postoperative functional MRI activation and subjects’ performance on postoperative verbal learning and design learning, simple and multiple regression analyses were performed over the whole brain. For each subject verbal learning and design learning scores were entered as covariates separately for patients with left TLE and patients with right TLE. Encoding pictures usually gave more bilateral activations and was not considered for further postoperative correlational analyses. The language lateralization index derived from postoperative language functional MRI was used as an additional covariate.
Difference image analysis
In order to investigate the relationship between pre- and postoperative change in memory functional MRI activation and change in scores for verbal and visual memory from before to 4 months after ATLR, we created ‘difference images’ by subtracting the coregistered, normalized postoperative contrast images from the original preoperative contrast images and vice versa. The created images represent activation changes for each contrast, highlighting areas of greater/lesser pre- than postoperative activation for the contrasts ‘word and face encoding’ over the whole brain. At the second level of the random effects analysis, we looked for brain regions showing correlations between greater/lesser pre- than postoperative or post- than preoperative activation and change (postoperative minus preoperative) scores for verbal and visual memory after ATLR.
Second level of analysis
At the second level we investigated: (i) effects of ATLR on the functional anatomy of verbal and visual memory encoding by comparing pre- versus postoperative main effects in patients with left and patients with right TLE; and (ii) efficiency of reorganization of postoperative verbal and visual memory functions by correlating postoperative activations on encoding words and faces with postoperative verbal learning and design learning scores in patients with left and right TLE.
Unless otherwise stated we report all medial temporal lobe activations at a threshold of P < 0.01, corrected for multiple comparisons [family-wise error (FWE) in a small volume of interest]. In view of our a priori hypothesis we performed the small volume correction using a sphere of 10 mm diameter for the left and right hippocampi based on the group peak activation.