FULL-LENGTH ORIGINAL RESEARCH

Distinct functional and structural MRI abnormalities in mesial temporal lobe epilepsy with and without hippocampal sclerosis *1Ana C. Coan, *1Brunno M. Campos, †Guilherme C. Beltramini, *Clarissa L. Yasuda, †Roberto J. M. Covolan, and *Fernando Cendes Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

SUMMARY

Ana C. Coan is a postdoctoral researcher at the University of Campinas.

Brunno M. Campos is a PhD student at the University of Campinas.

Objective: We aimed to investigate patterns of electroencephalography-correlated functional MRI (EEG-fMRI) and subtle structural abnormalities in patients with mesial temporal lobe epilepsy (MTLE) with hippocampal sclerosis (MTLE-HS) or normal MRI (MTLE-NL). Methods: We evaluated EEG-fMRI acquisition of the 25 patients with diagnosis of MTLE who had interictal epileptiform discharges (IEDs) in the intra-MRI EEG: 13 MTLE-HS and 12 MTLE-NL. fMRI was performed using echo-planar images in a 3T MRI coupled with EEG acquired with 64 MRI-compatible electrodes. In the first level analyses, the time of the IEDs ipsilateral to the epileptogenic zone was used as the paradigm, and four contrasts maps were built according to the variation of the hemodynamic response function (HRF) peaks (0, +3, +5, and +7 s). Second level group analyses were performed combining the contrast maps of MTLE-HS or MTLE-NL patients with each different HRF obtained at the first level. Areas of gray matter atrophy were evaluated with voxel-based morphometry (VBM) in both groups. Results: MTLE-HS and MTLE-NL had IED-related positive BOLD (posBOLD) detected in the ipsilateral anterior temporal lobe and insula. However, only MTLE-HS had significant posBOLD on contralateral hippocampus and anterior cingulate, whereas MTLE-NL had areas of posBOLD on ipsilateral frontal lobe. Both groups had significant IED-related negBOLD responses in areas of the default mode network (DMN), such as posterior cingulate and precuneus. There was no overlap of both posBOLD and negBOLD and areas of atrophy detected by VBM. Significance: Similar IEDs have different patterns of hemodynamic responses in subgroups of MTLE. In both MTLE-HS and MTLE-NL, there is a possible suppression of the DMN related to IEDs, as demonstrated by the negBOLD in these areas. The brain areas involved in the interictal related hemodynamic network are not the regions with the most significant gray matter atrophy in MTLE with or without MRI signs of HS. KEY WORDS: Hippocampal sclerosis, Functional neuroimaging, EEG, Default mode network.

Accepted April 24, 2014; Early View publication June 5, 2014. *Department of Neurology, Neuroimaging Laboratory, University of Campinas, Campinas, S~ao Paulo, Brazil; and †Neurophysics Group, Gleb Wataghin Physics Institute, University of Campinas, Campinas, S~ao Paulo, Brazil 1 Both authors contributed equally to this work. Address correspondence to Fernando Cendes, Departamento de Neurologia, Faculdade de Ci^encias Medicas – UNICAMP, Cidade Universitaria Zeferino Vaz, Campinas, SP, CEP 13083-970, Brazil. E-mail: [email protected] Wiley Periodicals, Inc. © 2014 International League Against Epilepsy

Epilepsies are conditions with functionally and anatomically connected networks,1 and ictal and interictal phenomena may be explained by the dysfunction of these networks.2 The most studied epilepsy network is the one associated with mesial temporal lobe epilepsy (MTLE).3–5 Although patients may share a common clinical and electroencephalography (EEG) expression, MTLE has a constellation of different etiologies, which may also imply

1187

1188 A. C. Coan et al. diverse neuronal networks. Hippocampal sclerosis (HS) is the most common pathologic substrate in MTLE refractory to antiepileptic drugs (AEDs)6; however, a significant number of patients have normal magnetic resonance imaging (MRI).7–9 Better understanding of these neural networks may have important implications in the comprehension of the biology of MTLE and its associated comorbidities.10,11 Combined continuous EEG and functional MRI (EEGfMRI) has been used to investigate neural networks in patients with epilepsies. The combination of these techniques permits noninvasive simultaneous measurement of neural activity and hemodynamics and allows the study of neurovascular coupling through the variation of BOLD (blood oxygen level–dependent) signal.12 EEG-fMRI can reveal hemodynamic changes related to ictal or interictal epileptiform discharges (IEDs), giving insights for the determination of the seizure-onset zone,13–15 but it also provides concomitant pattern of hemodynamic activity of all other brain areas distant from the presumed epileptogenic zone (EZ).16 Previous reports of EEG-fMRI and TLE patients have demonstrated a consistent pattern of IED-related BOLD responses in areas such as bilateral mesial and neocortical temporal structures but also including extratemporal regions as insula and anterior cingulate.17–19 All these studies have combined individuals with TLE of different underlying pathologies as well as mesial and neocortical temporal presumed seizure origin. Moreover, IED-related BOLD responses have been detected in areas in which subtle structural damage has also been described, such as anterior cingulate and insula4; however, it has not been observed in other brain regions with consistent description of atrophy, such as thalamus.19 In the present study, we used EEG-fMRI in an attempt to infer the brain structures involved in the network related to IEDs of two groups of refractory MTLE: patients with MRI signs of HS and patients with normal MRI. Our hypothesis was that, despite the similar semiology and interictal EEG findings, the underlying cause of the epilepsy might also contribute to the pattern of metabolic changes observed during the IEDs. In addition, we tried to evaluate whether the functional network defined by EEG-fMRI is related to the network of structural abnormalities defined by MRI voxelbased morphometry (VBM).4,20

Methods Patients We included 29 patients (11 men; mean age 41 years, range 19–58 years) with clinical and video-EEG diagnosis of MTLE, followed at the Epilepsy Clinic, University of Campinas. All patients were submitted to EEG-fMRI recordings from March 2010 to January 2012. Patients were divided in two groups according to visual MRI analysis: TLE with MRI signs of HS (MTLE-HS, 14 patients) and MTLE with normal MRI (MTLE-NL, 15 Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

patients). Patients with MTLE secondary to other brain lesions or dual pathology were not selected. Only patients with well-defined ictal and interictal EEG and typical MTLE seizure semiology were initially screened. Final selection included only patients with refractory epilepsy and routine EEG studies with frequent IEDs restricted to the anterior and medium portion of the temporal lobes. Patients with polyspikes, bisynchronous discharges, or rhythmic epileptiform discharges were not selected. All patients signed informed consent form approved by the ethics committee of UNICAMP prior to acquisition of EEG-fMRI data. The EZ was defined by prolonged ictal and interictal video-EEG recordings. All patients had unilateral MTLE and those with MRI signs of HS had ipsilateral EZ. EEG-fMRI acquisition and pre-processing fMRI examinations were performed on a 3T-Achieva MRI (Philips Medical Systems, Best, The Netherlands) with EPI (echo-planar image) sequences of 24–48 min (voxel size = 3 9 3 9 3 mm³, 39 slices, no gap, FOV = 240 9 240 9 117 mm³, TE = 30 msec, TR = 2,000 msec, flip angle = 90 degrees). There was no significant difference in the variability of the EPI duration between the two groups of patients (MTLE-HS: median EPI duration of 37  10 min; MTLE-NL: median EPI duration of 41  7 min; Two-sample t-Test, p = 0.27). The patient’s head was immobilized with an air cushion. Patients were instructed to remain still and with the eyes closed during the acquisition of the exam. The concomitant EEG was recorded with 64 MR-compatible (Ag/AgCl) electrodes (BrainProducts, Munich, Germany). The signal was amplified with BrainAmp Amplifier (BrainProducts) and transmitted through optic fibers to a recording terminal outside the MRI room. The fMRI data were processed and analyzed using SPM8 (Welcome Trust Center for Neuroimaging, London, United Kingdom). The EPIs were realigned, slice-timing corrected, normalized according to Montreal Neurological Institute (MNI) template, and smoothed with a Gaussian kernel of 6 mm full width at half maximum (FWHM). The EEG postprocessing was performed with Brain Vision Analyzer 2.0 (BrainProducts), and gradient and ballistocardiogram artifacts were removed with Average Artifact Subtraction correction.21,22 Intra-MRI EEG An expert neurophysiologist (ACC) reviewed the filtered EEG studies acquired inside the scanner to mark the IEDs. IEDs were marked as single points and used as events in an fMRI paradigm to look for BOLD changes in the MR signal. Three patients (all from MTLE-NL group) did not have IEDs during the scan and were excluded from the analysis. No seizures occurred during the fMRI acquisitions. The number of IEDs during the fMRI varied from 11 to 281 (mean 77.75) for MTLE-NL and from 5 to 923 (141.30) for

1189 MTLE Functional and Structural Network Table 1. Clinical data of TLE patients included in functional and structural network analysis

Case

Age

Gender

TLE group

Age 1st seizure

1

24

F

TLE-HS

1 year

2

40

F

TLE-HS

1 year

3 4

38 56

M F

TLE-HS TLE-HS

6 months 7 years

5 6

36 38

F F

TLE-HS TLE-HS

5 years 1 year

7 8

50 58

M M

TLE-HS TLE-HS

9 months 16 years

9 10

46 47

F M

TLE-HS TLE-HS

17 years 14 years

11

28

M

TLE-HS

17 years

12 13

19 38

M F

TLE-HS TLE-HS

1 year 2 years

14 15

20 48

M F

TLE-NL TLE-NL

14 years

16

43

F

TLE-NL

13 years

17

51

F

TLE-NL

28 years

18

27

M

TLE-NL

3 years

19

47

F

TLE-NL

5 years

20 21

56 45

M F

TLE-NL TLE-NL

8 years

22

27

F

TLE-NL

27 years

23 24

37 28

F F

TLE-NL TLE-NL

13 years 16 years

25

48

M

TLE-NL

19 years

Seizure semiology Jamais vu, staring, dystonic posture of left arm, right hand automatisms Fear, loss of consciousness, staring, oral automatisms No aura. Staring, oral and bimanual automatisms Abdominal pain, loss of consciousness, staring, oral and manual automatisms Fear, loss of consciousness, staring Epigastric pain, nausea, loss of consciousness, chewing, dystonic posture of right arm Sudden holocranial headache, manual automatisms Jamais vu, speech arrest, lost of consciousness, staring Epigastric sensation, nausea, staring, chewing Strange and bad feeling, loss of consciousness, upward eye deviation, oral automatisms No aura, loss of consciousness, bimanual automatisms Abdominal sensation, starring, walk around No aura, sudden loss of consciousness, say meaningless sentences, manual automatisms Rising epigastric sensation, staring Feeling of discomfort and imminent death, fear, loss of consciousness, hypomotor Fear, loss of consciousness, ictal speech, manual automatisms Bad rising epigastric sensation, staring, chewing, hands automatisms Rising epigastric sensation, buzz in both years, staring/hypomotor No aura, sudden loss of consciousness, manual automatisms Jamais vu, loss of consciousness, hypomotor Epigastric sensation, fear, loss of consciousness, oral and bimanual automatisms Fear, loss of consciousness, swallow, walk with no purpose Epigastric sensation, loss of consciousness, chewing “Religious” feeling, loss of consciousness, bimanual automatisms Rising epigastric sensation, staring/hypomotor

Laterality EZ (interictal/ictal scalp EEG)

IEDs during EEG-fMRI (no. events)

Structural analysis (VBM)

RT

RT (10)

No

RT

RT (143)/LT (4)

No

RT LT

RT (32)/LT (203) LT (122)

Yes Yes

LT LT

RT (13)/LT (3) RT (3)/LT (15)

Yes Yes

LT LT

LT (17) LT (13)

Yes No

LT RT

LT (5) RT (187)

Yes Yes

RT

RT (16)/LT (40)

Yes

LT LT

LT (88) RT (260)/LT (663)

Yes Yes

LT LT

LT (147) RT (46)/LT (36)

Yes Yes

RT

RT (26)/LT (13)

Yes

RT

RT (11)/LT (12)

Yes

RT

RT (84)

Yes

LT

LT (85)

Yes

LT LT

LT(13) RT (24)/LT (18)

Yes Yes

RT

RT (241)/LT (40)

Yes

LT LT

RT (24)/LT (28) RT (12)/LT (62)

Yes Yes

RT

RT (11)

Yes

TLE, temporal lobe epilepsy; VBM, voxel-based morphometry; F, female; M, male; HS, hippocampal sclerosis; NL, normal magnetic resonance imaging; RT, right temporal; LT, left temporal; R, right; L, left; EZ, epileptogenic zone; IEDs, interictal epileptiform discharges.

MTLE-HS (two-sample t-test, p = 0.40). Fourteen patients had bilateral temporal IEDs (Table 1). All IEDs had similar morphology and location, restricted to the anterior and medium portions of temporal lobes (electrodes F8, T7, T8, P7, P8, TP9, TP10, FT7, FT8, TP7, TP8, FT9, and FT10), with minor difference in the fields between each patient. None had IEDs outside the temporal lobes or different from those observed in the routine EEG. One patient (MTLE-HS) was also excluded because of an artifact in the EPI image. Therefore, 25 patients were included in the analysis (13 MTLE-HS and 12 MTLE-NL). Detailed clinical data are described on Table 1.

EEG-fMRI statistical analysis The temporal series of IEDs were convolved with the canonical SPM8 hemodynamic response function (HRF) (peak at 5 s relative to onset, delay of undershoot 16 s, ratio of response to undershoot 6, and length of kernel 32 s). To increase the sensitivity to detect IED-related BOLD changes,23 for each subject, nine design matrices were created varying the beginning of the HRF from 10 to +10 s from the instant of the IEDs (HRFs time-to-peak at 5, 2, 0, +3, +5, +7, +9, +11, +14 s from the IEDs). The HRF derivatives (temporal and dispersion) were used as regressors. Six realignment regressors (three rotation and Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

1190 A. C. Coan et al. three translation parameters) were included in the design matrix in order to consider errors related to movement artifacts. All IEDs observed in the intra-MRI EEG were added to the design matrix; however, only the maps of the IEDs ipsilateral to the EZ were considered in the group analysis. Positive (posBOLD) and negative BOLD (negBOLD) contrast maps were built for each HRF. Subsequently, for each MTLE group we performed a second level (“random effect”) statistical analysis using the normalized contrast maps created for the IED ipsilateral to the EZ in the single subject analysis (MTLE-HS: five right and eight left temporal; MTLE-NL: five right and seven left temporal). The spatial coregistration of these maps was checked and a covariance test was performed. The maps built for the right IEDs were flipped (right–left orientation). Thus all the results are described as ipsilateral (left side) or contralateral (right side), referring to the IED marked on the intra-MRI EEG, and consequently the EZ. The duration of the EPI sequence of each patient was included in this second level analysis as a regressor. One-sample t-tests (p < 0.005, uncorrected) of posBOLD and negBOLD were performed for each group of individuals. The BOLD maps from all the different HRFs were visually checked, and for the final analysis we chose those with any detected BOLD response in all the four subgroups (posBOLD for MTLE-HS and MTLE-NL and negBOLD for MTLE-HS and MTLE-NL). Thus, the results are described for the maps with HRF at 0, +3, +5, and +7 s from the IEDs. To appropriately explore the differences between MTLEHS and MTLE-NL patients, second level analyses were also done with the BOLD contrast maps obtained from the variation of the HRF time-to-peak (0, +3, +5 and +7 s) in the first level: two-sample t-tests (p < 0.005, uncorrected) looking for what MTLE-HS have more than MTLE-NL and what MTLE-NL have more than MTLE-HS. This resulted in eight different posBOLD and eight negBOLD T-maps, and these detailed results are provided in table format in the Supporting Information. For the graphic visualization of these results, we performed the overlay (simultaneous visualization) of the four T-maps derived from each of the HRFs time-to-peak with the SPM8 software (Wellcome Trust Center for Neuroimaging, London, United Kingdom). Structural analysis (voxel-based morphometry–VBM) For VBM, we acquired three-dimensional (3D) T1weighted images (voxel size = 1 9 1 9 1 mm³, TR = 7 msec, TE = 3.2 msec, flip angle = 8°, matrix = 240 9 240) for patients and for a control group of 74 healthy subjects (age and sex matched). Three patients with no structural 3T-MRI previous to epilepsy surgery were excluded from the MTLE-HS group (Table 1). The MRI studies of patients with right EZ were flipped in left–right orientation. Pre-processing and statistical analysis were performed with VBM8/SPM8 (Wellcome Department of Cognitive Neurology, London, United Kingdom). Preprocess included Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

normalization and modulation (MNI template) using DARTEL (in order to increase the accuracy of the alignment between subjects) and segmentation of the images in gray matter (GM), white matter, and cerebrospinal fluid. The resultant GM images were smoothed (10 mm FWHM). A test of quality was performed to observe homogeneity and co-registration between the data, and no outliers were detected. Two-sample t-test (p < 0.001, uncorrected) was performed between each MTLE group and controls. Age and sex were used as covariates in the statistical model. Bilinear interpolation was used in SPM to combine positive and negBOLD maps from all the different HRFs, with structural maps for each MTLE group. This step was taken to adapt the difference in voxels size from EPI and T1 images. This grouping of the results of the functional and structural analysis was done with the overlay (simultaneous visualization) of the four posBOLD or negBOLD T-maps (HRFs time-to-peak at 0, +3, +5, and +7 s from the IEDs) and the GM T-maps for each patient group. The number of voxels from structural analysis superimposed to voxels from functional analysis was calculated in the resampled maps.

Results MTLE-HS MTLE-HS posBOLD In MTLE-HS, accordingly to T-scores and number of voxels, IED-related posBOLD was most prominently observed in the ipsilateral insula, superior and inferior temporal gyrus, postcentral gyrus, anterior cingulated, and parahippocampal gyrus (Fig. 1A–D). MTLE-HS negBOLD The IED-related negBOLD detected in MTLE-HS patients included areas overlapping with the default mode network (DMN), as precuneus, posterior cingulate, contralateral supramarginal gyrus, cuneus and middle frontal gyrus (Fig. 2A–D). NegBOLD was more prominently detected in the hemisphere contralateral to the EZ. MTLE-NL MTLE-NL posBOLD In MTLE-NL, accordingly to T-scores and number of voxels, IED-related posBOLD was most prominently observed in the ipsilateral superior temporal gyrus, medial and middle frontal gyrus, postcentral gyrus, parahippocampal gyrus, and insula and contralateral superior temporal gyrus, precuneus, and cerebellum (Fig. 1E–H). MTLE-NL negBOLD The IED-related negBOLD detected in MTLE-HS patients included areas overlapping with the DMN, as ipsi-

1191 MTLE Functional and Structural Network A

E

B

F

C

G

D

H

Figure 1. EEG-fMRI–positive BOLD in MTLE-HS and MTLE-NL. Group analysis of interictal epileptiform discharge–related positive hemodynamic responses (posBOLD) in patients with MTLE (t-test, p < 0.005, uncorrected). Boxes A–D show posBOLD of MTLE-HS group with the HRF peaks at 0 (A), +3 (B), +5 (C), and +7 (D) seconds from the EEG discharges. Boxes E–H show posBOLD of MTLE-NL group with the HRF peaks at 0 (E), +3 (F), +5 (G), and +7 (H) seconds from the EEG discharges. BOLD, blood oxygen level–dependent; MTLE, mesial temporal lobe epilepsy; posBOLD, positive BOLD; HS, hippocampal sclerosis; NL, normal MRI; HRF, hemodynamic response function; T, T-score. Epilepsia ILAE

lateral middle frontal gyrus, posterior cingulate, contralateral precuneus, and supramarginal gyrus (Fig. 2E–H). NegBOLD was more prominently detected in the hemisphere contralateral to the EZ. Detailed information of the different clusters of posBOLD and negBOLD, including the distribution of the different peaks of the HRFs, is described in Tables S1 and S2, respectively. Comparison of MTLE-HS and MTLE-NL BOLD maps The second level analysis of IED-related posBOLD maps was performed to explore the differences between patients with MTLE-HS and MTLE-NL. This analysis revealed that patients with MTLE-HS had more significant posBOLD changes in the ipsilateral insula and anterior cingulate and contralateral caudate, hippocampus, thalamus (pulvinar), and anterior cingulate than MTLE-NL. In contrast, MTLENL had more significant posBOLD changes in the ipsilateral inferior, middle, and superior frontal gyrus and posterior cingulate and contralateral precuneus and cerebellum than MTLE-HS. Patients with MTLE-HS had more robust

negBOLD changes in the areas of the DMN than patients with MTLE-NL, including bilateral frontal, temporal, precuneus, and posterior cingulate. MTLE-NL patients had more significant negBOLD changes in the ipsilateral parahippocampal than MTLE-HS. Table S3 shows the areas of significant difference, as it was observed with each of the four HRF time-to-peak for posBOLD and negBOLD. To simplify the visualization of these results, the Figure 3 shows the overlays (simultaneous visualization) of the BOLD maps derived from each of the four HRFs. Structural analysis and comparison with functional maps Structural analysis: MTLE-HS In MTLE-HS, GM volume reduction was observed in both hippocampi, temporal lobes, thalamus, occipital regions, and ipsilateral caudate (Fig. 4A,B; Table S4). IED-related pos/negBOLD responses did not show significant overlap with areas of GM atrophy. Combined structural plus posBOLD maps and structural plus negBOLD maps Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

1192 A. C. Coan et al. A

E

B

F

C

G

D

H

Figure 2. EEG-fMRI negative BOLD in MTLE-HS and MTLE-NL. Group analysis of interictal epileptiform discharge–related negative hemodynamic responses (negBOLD) in patients with MTLE (t-test, p < 0.005, uncorrected). Boxes A–D show negBOLD of MTLE-HS group with the HRF peaks at 0 (A), +3 (B), +5 (C), and +7 (D) seconds from the EEG discharges. Boxes E–H show negBOLD of MTLE-NL group with the HRF peaks at 0 (E), +3 (F), +5 (G), and +7 (H) seconds from the EEG discharges. BOLD, blood oxygen level–dependent; MTLE, mesial temporal lobe epilepsy; negBOLD, negative BOLD; HS, hippocampal sclerosis; NL, normal MRI; HRF, hemodynamic response function. Epilepsia ILAE

showed only 0.36% and 1.46% of superimposed voxels, respectively (Fig. 4A,B).

patients and a simultaneous comparison of structural abnormalities.

Structural analysis: MTLE-NL MTLE-NL patients showed subtle GM atrophy in the ipsilateral neocortical temporal region as shown in Figure 4C,D and Table S4. IED-related pos/negBOLD responses did not show significant overlap with areas of GM atrophy. Combined structural plus posBOLD and structural plus negBOLD maps showed 0% of superimposed voxels (Fig. 4C,D).

Methodologic considerations In this study we detected BOLD changes through the convolution of the IEDs timing with different canonical HRF distributed across time. Different methods of EEG-fMRI analysis have been used in the last years, and there is no consistent evidence of which is the best approach.24 The use of canonical HRF has been the most widely used and it has good sensitivity to detect the EZ in patients with refractory epilepsies.14,25 Other methods have been proposed, such as the combination of different HRFs26 or patient-specific HRF.27,28 These latter techniques result in an increase of extent and degree of BOLD detection.24 However, the significance of this increased sensitivity is not known. Indeed, it has been proposed that noncanonical responses may represent distinct phenomena, including artifacts and propagated epileptiform activity.29 In that sense, the best methodologic approach for EEG-fMRI analysis remains unknown. In our study, we described the IED-related BOLD changes of MTLE-HS and MTLE-NL in two different ways:

Discussion We demonstrated that similar IEDs have different patterns of hemodynamic responses in two subgroups of MTLE: MTLE associated with HS and MTLE with normal MRI. Moreover, we observed that there is no overlap of the functional and structural abnormalities in these groups. Although previous authors had already addressed functional analysis in TLE,17–19 our work adds new information because it includes more homogeneous subtypes of MTLE Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

1193 MTLE Functional and Structural Network A

B

C

D

Figure 3. Comparison of MTLE-HS and MTLE-NL BOLD maps. Second level analysis of posBOLD and negBOLD maps showing the differences between patients with MTLE-HS and MTLE-NL. Second level analyses were done with the BOLD contrast maps obtained from the variation of the HRF peaks (0, +3, +5, and +7 s) in the first level. To simplify the visualization of these results, this figure shows the overlays of the posBOLD and negBOLD maps derived from each of the four HRFs. (A) PosBOLD more significant in MTLE-HS than in MTLE-NL (two-sample t-test, p < 0.005, uncorrected). (B) NegBOLD more significant in MTLE-HS than in MTLE-NL (two-sample t-test, p < 0.005, uncorrected). (C) PosBOLD more significant in MTLE-NL than in MTLE-HS (two-sample t-test, p < 0.005, uncorrected). (D) NegBOLD more significant in MTLE-NL than in MTLE-HS (two-sample t-test, p < 0.005, uncorrected). BOLD, blood oxygen level– dependent; MTLE, mesial temporal lobe epilepsy; posBOLD, positive BOLD; HS, hippocampal sclerosis; NL, normal MRI; HRF, hemodynamic response function; T, T-score. Epilepsia ILAE

(1) the maps of pos/negBOLD in the MTLE-HS and MTLENL groups, which really shows the network of IED-related BOLD in these patients (Fig. 1); (2) the maps of comparison between MTLE-HS and MTLE-NL (Fig. 3), which shows the brain areas more “activated/deactivated” in MTLE-HS than MTLE-NL patients or in MTLE-NL than MTLE-HS patients. It is important to emphasize that this second analysis has the issues associated with the power of the multiple statistical comparison. MTLE-HS and MTLE-NL IED-related posBOLD Despite the similarity of morphology, field distribution and frequency of the IEDs in MTLE-HS and MTLE-NL, differences were observed in the posBOLD responses of each group. IED-related posBOLD responses have been addressed to reflect areas of discharge generation30; however, concomitant posBOLD is frequently observed in areas distant from the presumed EZ in individuals with refractory epilepsies.31 It remains to be confirmed whether this diffuse IED-related posBOLD reflects the propagation of the interictal activity.

The group analysis of IEDs in patients with MTLE-HS demonstrated posBOLD mostly in the ipsilateral neocortical temporal lobe and parahippocampal gyrus, insula, anterior cingulate, and putamen. Accordingly, previous studies including TLE patients with diverse etiology have demonstrated common areas of IED-related posBOLD, such as the ipsilateral mesial temporal structures,17,18 putamen/globus pallidus, bilateral superior temporal gyrus, and inferior insula.18 More recently, another study found posBOLD in areas concordant with our MTLE-HS group in TLE with diverse etiologies.19 Because the majority of individuals included in these previous studies had TLE with MRI findings of HS, it is possible that their results reflect, in fact, networks not common to any TLE but more specific to MTLE-HS. In addition, in our study, posBOLD in MTLE-HS was observed in other brain areas distant from the mesial temporal lobe and also with less direct connections with the hippocampus, such as postcentral gyrus and superior parietal lobule. The higher number of brain areas with detected posBOLD in our study is possibly due to the homogeneity of Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

1194 A. C. Coan et al. A

B

C

D

Figure 4. Combined structural (VBM) and functional (EEG-fMRI) analysis of MTLE-HS and MTLE-NL. These maps show the overlay (simultaneous visualization) of the four posBOLD (yellow) or negBOLD (blue) T-maps (HRF peaks at 0, +3, +5 and +7 s from the EEG discharge) with the T-maps of the areas of gray matter atrophy detected with VBM (purple). No overlap was observed between the functional and structural maps (VBM: t-test, p < 0.001, uncorrected; EEG-fMRI: t-test, p < 0.005, uncorrected). (A) Overlay of MTLE-HS posBOLD maps (HRF peaks at 0, +3, +5, and +7 s from the EEG discharge) and gray matter atrophy; (B) overlay of MTLE-HS negBOLD maps (HRF peaks at 0, +3, +5, and +7 s from the EEG discharge) and gray matter atrophy; (C) overlay of MTLE-NL posBOLD maps (HRF peaks at 0, +3, +5, and +7 s from the EEG discharge) and gray matter atrophy; (D) overlay of MTLE-NL negBOLD maps (HRF peaks at 0, +3, +5, and +7 s from the EEG discharge) and gray matter atrophy. VBM, voxel based morphometry; BOLD, blood oxygen level–dependent; MTLE, mesial temporal lobe epilepsy; posBOLD, positive BOLD; negBOLD, negative BOLD; HS, hippocampal sclerosis; NL, normal MRI; HRF, hemodynamic response function. Epilepsia ILAE

our group, which was composed of only MTLE patients with MRI signs of HS and no other pathologies. Similar to MTLE-HS, in our group with MTLE-NL, the maximum posBOLD was observed in the anterior region of the ipsilateral temporal lobe and insula. However, significant difference was observed by visual comparison of the posBOLD maps and the second level analysis between the two groups. The statistical comparison of the posBOLD maps revealed that positive hemodynamic responses in brain areas, such as ipsilateral insula and anterior cingulate and contralateral caudate and hippocampus, are more significant in patients with MTLE-HS, whereas the MTLE-NL group has more significant positive hemodynamic responses in the ipsilateral frontal regions. The results observed for one important structure, the ipsilateral insula, might, at first sight, have contradictory results: Despite the fact that this structure shows posBOLD in both groups, in MTLE-HS the ipsilateral insula is significantly more activated than in MTLE-NL. This can be seen in both Figures 1 and 3. With the results obtained in the present study, we can conclude that the ipsilateral insula has a stronger involvement in the IED-related network of Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

MTLE-HS than of MTLE-NL patients. Another important difference observed was the absence of BOLD changes in the anterior cingulate of MTLE-NL patients. The anterior cingulate has been repeatedly indicated as part of the of interictal hemodynamic network of patients with TLE,18,19 and its importance is associated with the connections of this structure with the limbic system.19 The absence of hemodynamic response observed in the anterior cingulate and the weaker involvement of the ipsilateral insula in MTLE-NL suggests that, although the semiology and scalp EEG findings of these individuals do not differ from the MTLE-HS group, these patients may have different IED generators and, consequently, different patterns of hemodynamic response propagation. In our study we did not detect IED-related BOLD changes in the ipsilateral hippocampus of both groups, although in MTLE-HS posBOLD changes were observed in the ipsilateral mesial temporal structures as parahippocampal gyrus and uncus. Previous EEG-fMRI studies of TLE patients have shown posBOLD in the ipsilateral hippocampus.17–19 Our results resemble, however, what has been demonstrated in a recent publication19: IED-related BOLD

1195 MTLE Functional and Structural Network changes in TLE show a mild involvement of the ipsilateral mesial temporal structures compared with the more robust involvement of the neocortex. The absence of a clear IEDrelated posBOLD in the hippocampus in our groups is supported by the complexity of the BOLD response in this structure in comparison with the BOLD response expected in the neocortex. For example, one study demonstrated that the increase in the hippocampal neuronal activity during seizures in rats is often followed by negative BOLD signal, suggesting that, in these cases, the increase in the oxygen metabolism exceeds the increase of blood flow.32 This is the opposite of what is observed in the neocortex. Other hippocampal peculiarity that might influence the detection of hippocampal BOLD changes is the decreased vascularization of this structure demonstrated by the increased risk of hypoxic damage33 and the elevated hippocampal activity during rest, which might compromise the baseline used for the statistical comparison of the fMRI paradigms as in our case.34 MTLE-HS and MTLE-NL IEDs related negBOLD In both groups, negBOLD was observed in areas related to the DMN,35 such as precuneus and posterior cingulum, although in MTLE-NL group the negative responses were subtler, as confirmed by the statistical comparison of the negBOLD maps. The meaning of negBOLD and its relation with blood flow and metabolism is not fully understood. Possible mechanisms of negBOLD response are decrease in blood flow concomitant to relative decrease of cortical neuronal activity from the baseline30; neuronal inhibition36; or a vascular origin (“vascular steal”).37 According to previous studies comparing negBOLD responses and perfusion changes related to IEDs, the mechanism of a relative decrease in the basal cortical activity is the most suitable to explain the BOLD changes observed in areas of the DMN.30 Suspension of the DMN during generalized spike-and-wave discharges has been demonstrated in idiopathic generalized epilepsies38 and also in MTLE.17 Recently, deactivation of the DMN in response to IEDs was observed in different groups of focal epilepsies.19 Our results and previous fMRI studies consistently demonstrated that the DMN is affected not only by generalized spike-and-waves but also by isolated IEDs in focal epilepsies, which are usually not accompanied by any apparent behavioral or cognitive change. We also observed that in both MTLE-HS and MTLE-NL, but more evident in the former group, there was a predominance of negBOLD in the hemisphere contralateral to the IEDs and the EZ. Abnormal connectivity of the DMN has been described in MTLE and reduced connectivity of the posterior cingulate cortex with the ipsilateral, but not with the contralateral hippocampus has been described.39 However, whether the epileptogenic hippocampus affects the function of the DMN in the ipsilateral hemisphere remains to be determined.

Comparison of subtle gray matter atrophy and IEDs related to pos/negBOLD We did not observe significant overlap of the areas of GM reduction and the areas with IEDs-related pos/negBOLD changes. This may indicate that the structures involved in interictal network in these two groups of MTLE do not sustain significant loss of volume. However, previous studies have reported structural abnormalities that also included the areas with posBOLD observed by our EEGfMRI group analysis4 in both MTLE-HS and MTLE-NL patients.39,40 One possibility is that the number of individuals included in each of our groups was too small to detect the subtle structural abnormalities of these brain regions. Yet we were able to show consistent GM reduction in other important areas in both groups; therefore, it is still possible that even if there are structural abnormalities in the regions related to IED-related pos/negBOLD, these are not as relevant as seen in other areas, such as thalamus, caudate, or occipital regions. Likewise, although morphometric and volumetric studies of patients with MTLE have demonstrated diverse results with a significant variability of the brain regions with detected atrophy, areas such as bilateral thalamus are consistently reported as atrophic in refractory MTLE-HS.40 In the present study, thalamic atrophy was detected in both MTLE-HS and MTLE-NL, but only a minor posBOLD in the pulvinar of the contralateral thalamus was observed in the comparison of MTLE-HS and MTLE-NL posBOLD maps. The absence of overlap between IED-related hemodynamic response and atrophy corroborates the complex interactions between functional and structural networks in MTLE. In fact, although the propagation of ictal activity has been implicated in progressive structural damage, paradoxically it could be expected that in brain regions with atrophy, and consequently a smaller concentration of neurons, a lesser amount of ictal or interictal activity would happen. In addition, the neurovascular coupling of brain damaged areas is not completely known, and although the atrophy in brain regions distant from the temporal lobe detected in MTLE patients are very subtle and seen only by refined MRI analysis, it is possible that also the perfusion of these areas are compromised. In summary, we were able to show that brain structures involved in the IED-related functional network differ in MTLE-HS and MTLE-NL patients, and also that the structures involved in these functional networks are not those with the most significant structural damage. These findings may direct future investigations about the interictal dysfunctions in patients with MTLE. Moreover, by revealing distinct patterns of BOLD response from similar IEDs, EEG-fMRI adds information to the scalp EEG studies of patients with MTLE and improves the understanding of the functional networks of sub-groups of patients with MTLE. Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

1196 A. C. Coan et al.

Acknowledgments This study was funded by S~ao Paulo Research Foundation (FAPESP), grants 2005/56578-4, 2009/54552-9, and 2011/03477-7, and by the Brazilian National Counsel for Scientific and Technological Development (CNPq), grants 140379/2008-8 and 305585/2009-6.

Disclosure None of the authors has any conflict of interest to disclose. 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.

References 1. Spencer SS. Neural networks in human epilepsy: evidence of and implications for treatment. Epilepsia 2002;43:219–227. 2. Laufs H. Functional imaging of seizures and epilepsy: evolution from zones to networks. Curr Opin Neurol 2012;25:194–200. 3. Blumenfeld H, McNally KA, Vanderhill SD, et al. Positive and negative network correlations in temporal lobe epilepsy. Cereb Cortex 2004;14:892–902. 4. Bonilha L, Rorden C, Castellano G, et al. Voxel-based morphometry reveals gray matter network atrophy in refractory medial temporal lobe epilepsy. Arch Neurol 2004;61:1379–1384. 5. King D, Spencer S. Invasive electroencephalography in mesial temporal lobe epilepsy. J Clin Neurophysiol 1995;12:32–45. 6. Semah F, Picot MC, Adam C, et al. Is the underlying cause of epilepsy a major prognostic factor for recurrence? Neurology 1998;51:1256–1262. 7. Kuzniecky R, de la Sayette V, Ethier R, et al. Magnetic resonance imaging in temporal lobe epilepsy: pathological correlations. Ann Neurol 1987;22:341–347. 8. Cascino GD. Surgical treatment for epilepsy. Epilepsy Res 2004;60:179–186. 9. Bell ML, Rao S, So EL, et al. Epilepsy surgery outcomes in temporal lobe epilepsy with a normal MRI. Epilepsia 2009;50:2053–2060. 10. Heuser K, Taubøll E, Nagelhus EA, et al. Phenotypic characteristics of temporal lobe epilepsy: the impact of hippocampal sclerosis. Acta Neurol Scand 2009;189:8–13. 11. Mueller SG, Laxer KD, Cashdollar N, et al. Voxel-based optimized morphometry (VBM) of gray and white matter in temporal lobe epilepsy (TLE) with and without mesial temporal sclerosis. Epilepsia 2006;47:900–907. 12. Ogawa S, Tank DW, Menon R, et al. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 1992;89:5951–5955. 13. Zijlmans M, Huiskamp G, Hersevoort M, et al. EEG-fMRI in the preoperative work-up for epilepsy surgery. Brain 2007;130:2343–2353. 14. Thornton R, Vulliemoz S, Rodionov R, et al. Epileptic networks in focal cortical dysplasia revealed using electroencephalography-functional magnetic resonance imaging. Ann Neurol 2011;70:822–837. 15. Sierra-Marcos A, Maestro I, Falcon C, et al. Ictal EEG-fMRI in localization of epileptogenic area in patients with refractory neocortical focal epilepsy. Epilepsia 2013;54:1688–1698. 16. Federico P, Archer JS, Abbott DF, et al. Cortical/subcortical BOLD changes associated with epileptic discharges: an EEG-fMRI study at 3 T. Neurology 2005;64:1125–1130. 17. Laufs H, Hamandi K, Salek-Haddadi A, et al. Temporal lobe interictal epileptic discharges affect cerebral activity in “default mode” brain regions. Hum Brain Mapp 2007;28:1023–1032. 18. Kobayashi E, Grova C, Tyvaert L, et al. Structures involved at the time of temporal lobe spikes revealed by interindividual group analysis of EEG/fMRI data. Epilepsia 2009;50:2549–2556. 19. Fahoum F, Lopes R, Pittau F, et al. Widespread epileptic networks in focal epilepsies: EEG-fMRI study. Epilepsia 2012;53:1618–1627. 20. Ashburner J, Friston KJ. Voxel-based morphometry – the methods. Neuroimage 2000;11:805–821. 21. Allen PJ, Polizzi G, Krakow K, et al. Identification of EEG events in the MR scanner: the problem of pulse artifact and a method for its subtraction. Neuroimage 1998;8:229–239. Epilepsia, 55(8):1187–1196, 2014 doi: 10.1111/epi.12670

22. Allen PJ, Josephs O, Turner R. A method for removing imaging artifact from continuous EEG recorded during functional MRI. Neuroimage 2000;12:230–239. 23. Bagshaw AP, Aghakhani Y, Benar CG, et al. EEG-fMRI of focal epileptic spikes: analysis with multiple haemodynamic functions and comparison with gadolinium-enhanced MR angiograms. Hum Brain Mapp 2004;22:179–192. 24. van Houdt P, Zijlmans M. Different ways to analyze EEG-fMRI in focal epilepsy: does it matter? Clin Neurophysiol 2013; 124:2070–2072. 25. Grouiller F, Thornton RC, Groening K, et al. With or without spikes: localization of focal epileptic activity by simultaneous electroencephalography and functional magnetic resonance imaging. Brain 2011;134:2867–2886. 26. Pittau F, Dubeau F, Gotman J. Contribution of EEG/fMRI to the definition of the epileptic focus. Neurology 2012;78:1479–1487. 27. Masterton RA, Harvey AS, Archer JS, et al. Focal epileptiform spikes do not show a canonical BOLD response in patients with benign rolandic epilepsy (BECTS). Neuroimage 2010;51:252–260. 28. van Houdt PJ, de Munck JC, Leijten FS, et al. EEG-fMRI correlation patterns in the presurgical evaluation of focal epilepsy: a comparison with electrocorticographic data and surgical outcome measures. Neuroimage 2013;75:238–248. 29. Lemieux L, Laufs H, Carmichael D, et al. Noncanonical spike-related BOLD responses in focal epilepsy. Hum Brain Mapp 2008;29:329– 345. 30. Hamandi K, Laufs H, N€oth U, et al. BOLD and perfusion changes during epileptic generalised spike wave activity. Neuroimage 2008;39:608–618. 31. Kobayashi E, Bagshaw AP, Benar CG, et al. Temporal and extratemporal BOLD responses to temporal lobe interictal spikes. Epilepsia 2006;47:343–354. 32. Schridde U, Khubchandani M, Motelow JE, et al. Negative BOLD with large increases in neuronal activity. Cereb Cortex 2008;18:1814– 1827. 33. Bartsch T, Alfke K, Deuschl G, et al. Evolution of hippocampal CA-1 diffusion lesions in transient global amnesia. Ann Neurol 2007;62:475–480. 34. Stark CE, Squire LR. When zero is not zero: the problem of ambiguous baseline conditions in fMRI. Proc Natl Acad Sci USA 2001;98:12760– 12766. 35. Raichle ME, MacLeod AM, Snyder AZ, et al. A default mode of brain function. Proc Natl Acad Sci USA 2001;98:676–682. 36. Czisch M, Wehrle R, Kaufmann C, et al. Functional MRI during sleep: BOLD signal decreases and their electrophysiological correlates. Eur J Neurosci 2004;20:566–574. 37. Harel N, Lee SP, Nagaoka T, et al. Origin of negative blood oxygenation level dependent fMRI signals. J Cereb Blood Flow Metab 2002;22:908–917. 38. Gotman J, Grova C, Bagshaw A, et al. Generalized epileptic discharges show thalamocortical activation and suspension of the default state of the brain. Proc Natl Acad Sci USA 2005;102:15236–15240. 39. McCormick C, Quraan M, Cohn M, et al. Default mode network connectivity indicates episodic memory capacity in mesial temporal lobe epilepsy. Epilepsia 2013;54:809–818. 40. Li J, Zhang Z, Shang H. A meta-analysis of voxel-based morphometry studies on unilateral refractory temporal lobe epilepsy. Epilepsy Res 2012;98:97–103.

Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1. BOLD—MTLE-HS. Table S2. BOLD—MTLE-NL. Table S3. BOLD—comparison of MTLE-HS and MTLE-NL. Table S4. Structural analysis.

Distinct functional and structural MRI abnormalities in mesial temporal lobe epilepsy with and without hippocampal sclerosis.

We aimed to investigate patterns of electroencephalography-correlated functional MRI (EEG-fMRI) and subtle structural abnormalities in patients with m...
1000KB Sizes 2 Downloads 6 Views