SEARCH

SEARCH BY CITATION

Keywords:

  • Resting functional connectivity;
  • Spike-wave seizure;
  • Functional magnetic resonance imaging;
  • Cortex;
  • Thalamus

Summary

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Purpose

Functional magnetic resonance imaging (fMRI)–based resting functional connectivity is well suited for measuring slow correlated activity throughout brain networks. Epilepsy involves chronic changes in normal brain networks, and recent work demonstrated enhanced resting fMRI connectivity between the hemispheres in childhood absence epilepsy. An animal model of this phenomenon would be valuable for investigating fundamental mechanisms and testing therapeutic interventions.

Methods

We used fMRI-based resting functional connectivity for studying brain networks involved in absence epilepsy. Wistar Albino Glaxo rats from Rijswijk (WAG/Rij) exhibit spontaneous episodes of staring and unresponsiveness accompanied by spike-wave discharges (SWDs) resembling human absence seizures in behavior and electroencephalography (EEG). Simultaneous EEG-fMRI data in epileptic WAG/Rij rats in comparison to nonepileptic Wistar controls were acquired at 9.4 T. Regions showing cortical fMRI increases during SWDs were used to define reference regions for connectivity analysis to investigate whether chronic seizure activity is associated with changes in network resting functional connectivity.

Key Findings

We observed high degrees of cortical–cortical correlations in all WAG/Rij rats at rest (when no SWDs were present), but not in nonepileptic controls. Strongest connectivity was seen between regions most intensely involved in seizures, mainly in the bilateral somatosensory and adjacent cortices. Group statistics revealed that resting interhemispheric cortical–cortical correlations were significantly higher in WAG/Rij rats compared to nonepileptic controls.

Significance

These findings suggest that activity-dependent plasticity may lead to long-term changes in epileptic networks even at rest. The results show a marked difference between the epileptic and nonepileptic animals in cortical–cortical connectivity, indicating that this may be a useful interictal biomarker associated with the epileptic state.

Absence seizures are brief staring spells, which occur most commonly in children and are accompanied by large amplitude 3–4 Hz spike-wave discharges (SWDs) on electroencephalography (EEG). SWD-associated blood oxygenation level–dependent (BOLD) functional magnetic resonance imaging (fMRI) changes are bilateral and symmetric in children (Archer et al., 2003; Salek-Haddadi et al., 2003; Gotman et al., 2005; Hamandi et al., 2006; Berman et al., 2010) as well as in animal models (Tenney et al., 2003; Nersesyan et al., 2004a; Mishra et al., 2011b). Absence epilepsy is increasingly recognized as a disorder characterized by impaired function both during and between seizures (Blumenfeld, 2005; Hughes, 2009). Evidence from animal models indicates that this bilateral synchrony during SWDs is mediated by the corpus callosum (Musgrave & Gloor, 1980; Vergnes et al., 1989; Chahboune et al., 2009).

Brain networks exhibit long- and short-range synchronous activity at multiple time scales (Sporns, 2009). Slow, <0.1 Hz, synchronous fluctuations in hemodynamic signals occur at rest without stimulation (Golanov et al., 1994; Biswal et al., 1995). fMRI-based resting functional connectivity is well suited for measuring slow correlated activity throughout the brain. Resting functional connectivity analysis has identified brain networks related to sensory, motor, attention, or other functions in humans (Biswal et al., 1995; Lowe et al., 1998; Cordes et al., 2001; Raichle et al., 2001) and also in animal models (Vincent et al., 2007; Pawela et al., 2008; Zhao et al., 2008; Wu et al., 2009). Resting functional connectivity has been studied in several human disease states including Alzheimer's disease, multiple sclerosis, depression, and others (Li et al., 2002; Lowe et al., 2002; Greicius, 2008), as well as in partial epilepsy (Waites et al., 2006; Bettus et al., 2009; Pereira et al., 2010) and generalized epilepsy (Bai et al., 2011; Zhang et al., 2011). We recently found in human patients with typical childhood absence epilepsy that there is abnormally increased resting connectivity between the two hemispheres even when seizures are not occurring (Bai et al., 2011). This increased bi-hemispheric synchrony in the interictal period was most evident in cortical areas intensely involved in seizures during the ictal period.

In the present study, we performed functional MRI in epileptic and nonepileptic control rat brain with the hypothesis that the bilateral synchronous nature of absence epilepsy pathophysiology leads to abnormally enhanced resting functional connectivity between the two hemispheres in specific brain regions involved in the disorder. The availability of an animal model with abnormal connectivity similar to humans would be highly valuable for experimental testing of therapeutic interventions in preventing chronic changes in generalized epilepsy (Blumenfeld et al., 2008). To examine this hypothesis, we first analyzed simultaneous EEG-fMRI of SWDs to identify brain regions involved in SWDs. We then used these regions of interest (ROIs) to compare resting functional connectivity between the hemispheres in the Wistar Albino Glaxo rats from Rijswijk (WAG/Rij) rat absence epilepsy model (Coenen & Van Luijtelaar, 1987, 2003) with nonepileptic Wistar control rats.

Materials and Methods

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

All experimental procedures were in full compliance with Yale University Institutional Animal Care and Use protocols approved in agreement with the National Institutes of Health. Data from 11 female WAG/Rij (Harlan, Indianapolis, IN, U.S.A.) aged 7.3 ± 1.2 months (mean ± standard deviation [SD], range 5.9–9.5 months) with an average weight of 222 ± 1 g (mean ± SD) and five Wistar nonepileptic control female rats (Charles River Laboratories, Raleigh, NC, U.S.A.) aged 7.3 ± 0.1 months (mean ± SD, range 7–7.5 months) with an average weight of 300 ± 5.6 g (mean ± SD) were included in this study.

Animal preparation

Animal preparation and anesthesia has been described in detail previously (Mishra et al., 2011b). Briefly, anesthesia was induced using 1.5% halothane and later switched to intravenous (i.v.) fentanyl (40 μg/kg) and intraperitoneal (i.p.) haloperidol (1 mg/kg). To prevent any movements during the fMRI experiments and to facilitate artificial breathing, rats were paralyzed by i.v. injections of 0.5 mg/kg D-tubocurarine chloride every 2 h, tracheotomized, and artificially ventilated. Level of anesthesia was assessed by heart rate, blood pressure, and continuous EEG monitoring. Anesthesia was maintained by injection of a similar dose of fentanyl and haloperidol every 45 min. One femoral vein was cannulated and an intraperitoneal line was placed for the injections of anesthetics. A femoral artery was cannulated for continuous arterial blood pressure monitoring and periodic blood sampling for measurements of blood gases and pH. All physiologic values [blood gases, mean arterial blood pressure (MABP), and pH] were maintained within physiologic range throughout the experiments in both groups by adjusting respiration and level of anesthesia throughout the experiments. The body temperature of the rat was monitored with a rectal probe and kept constant at 37°C using a water heating pad.

MRI experiments were acquired in experimental runs lasting for 30 min, with one to three runs obtained per experiment. Under these conditions epileptic animals typically have multiple episodes of spontaneous SWDs as described in our previous studies (Nersesyan et al., 2004a,b; Mishra et al., 2011b) as well as epochs lasting up to a few minutes without SWDs.

EEG-fMRI experiments

Data from the WAG/Rij rats were analyzed separately for experimental runs (and animals), with versus without SWD. Of the 11 WAG/Rij rats, 6 had SWDs during EEG-fMRI (seven runs). For this group (WAG/Rij Group I), we used only those experiments where WAG/Rij rats had SWDs, and part of the time they also had no SWDs (resting state) in the same experimental run. Experiments without epochs of resting fMRI data lasting at least 180 s were excluded from the analysis. Experimental runs from WAG/Rij rats without any SWD were analyzed as a separate group (WAG/Rij group II), and were obtained from one rat (two runs) who also had SWDs in other runs, as well as five WAG/Rij rats without any SWD (nine runs; total of 11 runs from WAG/Rij without SWD). Data were also obtained from five Wistar nonepileptic control rats (12 runs). Wistar control rats did not have any SWDs during fMRI experiments.

During MRI recordings, the rat was positioned prone in a specially designed plastic holder with the head fixed and bregma positioned at the center of the surface coil. The animal was then inserted into the magnet with its head positioned at the isocenter of the magnet. EEG signals were acquired simultaneously with fMRI as described in detail previously (Mishra et al., 2011b). Briefly, EEG signals were recorded using a pair of 1-mm in diameter carbon-filament electrodes (WPI, Sarasota, FL, U.S.A.), placed between the scalp and the upper surface of the skull in the frontal and occipital areas. EEG signals were amplified and filtered (1–30 Hz) using a model 79D Data Recording System (Grass Instruments Co., Quincy, MA, U.S.A.), digitized and recorded (sampling rate, 1,000 Hz) using a CED Micro 1401 and Spike 2 software (Cambridge Electronics Design, Cambridge, United Kingdom).

All fMRI experiments were acquired on a 9.4 T Bruker horizontal bore (16-cm internal diameter) spectrometer (Bruker Avance, Billerica, MA, U.S.A.), equipped with passively shielded shim/gradient coils (47.5 G/cm) operating at 400.5 MHz for protons. The transceiver system consisted of a proton resonator radio frequency surface coil (15.18 mm diameter) for transmission of radio frequency pulses and receiving. The system was shimmed before each experiment using global manual shimming to optimize the homogeneity of the static magnetic field.

Anatomic images for each animal were acquired with the fast low angle shot sequence, 11 interlaced slices in the coronal plane with repetition time (TR), 500 msec; echo time (TE), 6 msec; flip angle, 40–55 degrees; field of view (FOV), 25 × 25 mm; matrix size, 256 × 256; in-plane resolution, 98 × 98 μm; and slice thickness, 1,000 μm, without gap. BOLD fMRI data were obtained in the same planes as anatomic images. All data were acquired using single-shot spin echo, echo planar imaging (SE-EPI) with the following parameters: TR, 1,000 msec; TE, 25 msec; excitation flip angle, 90 degrees; inversion flip angle, 180 degrees; FOV, 25 × 25 mm; matrix size, 64 × 64; in-plane resolution, 390 × 390 μm; and slice thickness, 1,000 μm without gap. All 30 SE-EPI BOLD fMRI experimental runs were acquired with 11 slices. Eighteen of 30 experimental runs were acquired on WAG/Rij rats (n = 11), whereas 12 experimental runs were acquired on Wistar control rats (n = 5). Images in 25 of these experimental runs were acquired every 6 s over 1,800 s (NR = 300 images); images in five experimental runs were acquired every 3 s (over 1,800 s for three runs with NR = 600 images; over 1,350 s for two runs with NR = 450 images). Twenty-four dummy scans occurred before the receiver was turned on and were not included in the data. Dummy scans were used to ensure that the proton spin system was in a steady state before data were collected.

EEG analysis

All EEG signals acquired during fMRI experiments were first processed using Spike 2 software, and methods were used to reduce magnetic field-induced artifacts similar to those described previously (Englot et al., 2008; Mishra et al., 2011b). All EEG data acquired during fMRI experiments were marked for individual SWD onset and offset time in Spike 2 software manually using scripts provided by CED. SWDs were defined as large-amplitude (more than two times the baseline EEG peak-to-peak amplitude) rhythmic 5–8 Hz discharges with typical spike-wave morphology lasting >1.0 s (Coenen & Van Luijtelaar, 1987; Blumenfeld et al., 2008; Mishra et al., 2011b). The SWD onset and offset time marks were used for all subsequent analyses of fMRI data. Overall mean SWD duration for all experiments was 1.91 ± 0.42 s (mean ± SD; N = 240 SWDs total). Epochs of resting fMRI were defined as periods lasting at least 180 s without SWD (even with duration <1 s) seen on EEG. To further ensure that resting state epochs were not contaminated by SWD-induced fMRI changes, we discarded fMRI data from 40 s before until 60 s after any SWD adjacent to the selected resting fMRI epochs we used. Although human studies have shown prolonged fMRI changes after SWDs (Moeller et al., 2008; Bai et al., 2010), in the WAG/Rij rat the fMRI signal returns to baseline within ~10 s of SWD end (e.g., see Mishra et al., 2011b, Fig. 3) so that 60 s should be a reasonable safety margin in this model.

To facilitate mapping of fMRI data, SWD onset time was used as “time 0” for binning and analyzing time-series data across multiple SWDs. Temporally aligning the onset time of multiple SWDs allowed mean changes to be determined in specific time bins after SWD onset.

Analysis of SWD fMRI data

fMRI images were processed for SWD-related changes using an in-house program running on a MATLAB platform (The MathWorks, Inc., Natick, MA, U.S.A.) as described previously (Mishra et al., 2011b). Briefly, all fMRI series were first screened for movement to ensure that all runs exhibited movement of <20% of a pixel in either the X or Y direction. Analysis was then performed using 2-s time bins, temporally aligning all SWD episodes by setting SWD onset as time 0. Baseline was obtained for each SWD episode using any images acquired during the 6.5-s interval preceding each SWD, with the additional requirement that no images were included in baseline if a preceding SWD event continued or ended within 3 s of image onset. If more than one image occurred during the baseline interval for any given SWD, these were averaged to obtain a single baseline image. BOLD fMRI t-maps were calculated using a paired t-test where each pair consisted of one baseline image (defined above) and one seizure image acquired in the time bin 2–4 s after seizure onset, because this time bin showed maximum changes in fMRI signals (Mishra et al., 2011b). t-maps were obtained across multiple SWD episodes (n = 240 SWD in five WAG/Rij rats) and displayed using threshold t > 2. Based on our prior experience with this model and the signal-to-noise ratio of our data acquisition system, this threshold is reasonably conservative and provides physiologically interpretable results (Nersesyan et al., 2004a; Mishra et al., 2011b). These experimental runs are the same experiments in WAG/Rij rats where we had both seizure and resting state in the same animals (WAG/Rij Group I, see above). The group t-map was superimposed onto high-resolution anatomic images and regional changes were identified based on a standard rat atlas (Paxinos & Watson, 1998).

Resting fMRI correlation analysis

Two ROIs for resting functional connectivity analysis were constructed based on group t-maps of SWD-related changes (Fig. 1) using a growing region approach (MATLAB 7.1; MathWorks). For each hemisphere of the brain, a region was identified by selecting a voxel near the largest cortical (face somatosensory cortex) group t-map increase, and neighboring voxels were then added within the same hemisphere exceeding a t-value threshold >2. The two ROIs obtained in this manner are shown in Fig. S1. We then identified epochs of at least 180 s on EEG without SWDs (resting data) in the epileptic WAG/Rij rats (WAG/Rij Group I) and analyzed resting fMRI during these time periods. We analyzed all epochs of resting data in each run.

image

Figure 1. Mapping fMRI changes during spike-wave seizures. Prominent increases were seen in bilateral somatosensory cortex (S1BF), with additional increases seen in several other areas including thalamus (Thal). Decreases were most prominent in the caudate–putamen (CPu). Simultaneous EEG was used to identify images obtained 2–4 s after SWD onset for comparison with baseline images and calculation of t-maps (threshold t > 2) as reported previously (Mishra et al., 2011b). Results are displayed on high-resolution anatomic images. Slices are shown from anterior to posterior, with approximate coordinates relative to bregma (Paxinos & Watson, 1998). Color bars indicate t-values for fMRI increases (warm colors) and decreases (cold colors). N = 240 SWD in five rats.

Download figure to PowerPoint

The fMRI time-varying signals of interest for resting functional connectivity analysis are typically slow, so we applied a 0.08 Hz low-pass filter to the fMRI data as described previously (Biswal et al., 1995; Hampson et al., 2006; Lu et al., 2007; Zhao et al., 2008). Analysis was restricted to voxels within the brain boundary. Image coregistration was performed using automated point-based registration in Bioimage Suite (http://bioimagesuite.yale.edu/index.aspx). We coregistered all animal's images to one image set from within the WAG/Rij rat group. Nine of the 11 matching (centrally located) slices were used after coregistration for final analysis at the group level, since data were missing for the end slices (front or back) in some animals.

An ROI-based connectivity approach was used for resting functional connectivity analysis (Biswal et al., 1995; Lowe et al., 1998; Lu et al., 2007; Zhao et al., 2008; Bai et al., 2011; Killory et al., 2011) using in-house software written by one of the authors (XB) in MATLAB 7.1. For each experimental run from each WAG/Rij rat, a reference time course for each ROI was calculated by averaging the time courses of all voxels within this ROI during the epochs of resting data (defined above). We then calculated the Pearson's correlation coefficients between the reference time course for each ROI and all voxels within the brain. The correlation (r) maps were then converted to z-score maps by Fisher's z transform, using the transform equation z(r) = 0.5 ln [(1 + r)/(1 − r)] (Jenkins & Watts, 1968). To normalize for differences in number of resting images in different experimental runs, each z-score was divided by the square root of variance, calculated as 1/√(n − 3), where n is the degrees of freedom defined as the number of image acquisitions (used for analysis) within each run. For one WAG/Rij rat two experimental runs were obtained, so the z-score maps for the two runs were averaged. Therefore for the six WAG/Rij rats with SWD (WAG/Rij Group I) this resulted in a total of six z-score maps, with the two (left and right) ROIs yielding a total of 12 z-score maps.

To compare the WAG/Rij data to Wistar controls, we matched the duration and timing within the run of each WAG/Rij epoch to identical times across all Wistar control runs. We thus obtained time-matched data from the Wistar controls, again yielding six z-score maps with two ROIs resulting in a total of 12 control z-score maps. The same analysis was also performed on the six WAG/Rij rats that had runs without any SWD (WAG/Rij group II).

We next performed two types of connectivity analysis on the group data (Bai et al., 2011). First we examined connectivity between each ROI and all other brain regions by constructing t-maps in the WAG/Rij and Wistar control groups. This approach allowed us to examine connectivity between the mean fMRI time course in a single reference ROI (defined above) and the whole brain on a voxel-by-voxel basis. To accomplish this, the individual z-score maps of each group for each reference ROI were entered into a second-level, one sample t-test (p < 0.01 uncorrected; cluster extent threshold = 3 voxels) in statistical parametric mapping 2 (SPM 2) (http://www.fil.ion.ucl.ac.uk/spm/software/spm2/) on a MATLAB 7.1 platform.

Second, we calculated connectivity between each ROI and its corresponding ROI in the contralateral hemisphere. For this analysis, we computed the between-hemisphere connectivity as the average z-score in the ROI contralateral to the reference ROI for each subject. In this approach, if the left ROI was used as the reference, this yielded a single time course for the left ROI and a set of z-scores for the corresponding right ROI, which were averaged to calculate one connectivity value. When the right ROI was used as the reference, the z-scores for each voxel in the left ROI were averaged. Therefore, each interhemispheric pair of corresponding ROIs yielded two connectivity values, which were generally similar but not identical. For group statistical analyses, we then performed one-way analysis of variance (ANOVA) followed by Tukey's Honestly Significant Difference (HSD; Kirk, 1995) method for post hoc pairwise comparisons (significance threshold p < 0.05) to assess group differences in WAG/Rij versus nonepileptic Wistar control rats. HSD is the most conservative approach for post hoc pair-wise comparisons with ANOVA.

Results

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

BOLD fMRI signal changes in SWD epilepsy

BOLD fMRI mapping during a total of 240 SWDs (in five rats) showed both increases and decreases in multiple cortical and subcortical regions (Fig. 1). As reported previously (Nersesyan et al., 2004a; David et al., 2008; Mishra et al., 2011b), we found intense bilateral increased BOLD fMRI signals in face somatosensory cortex (S1BF). We also saw increases in other cortical regions adjacent to somatosensory cortex, in the thalamus, anterior and posterior cingulate, as well as prominent bilateral decreases in the BOLD fMRI signal in caudate–putamen (CPu; Fig. 1). Analysis of baseline versus baseline BOLD fMRI images did not show significant changes in any brain region (data not shown).

Increased connectivity between the hemispheres in WAG/Rij epileptic rats at rest

We found significantly increased resting connectivity in the epileptic rats (WAG/Rij group I) between cortical regions most intensely involved in seizures in the left and right hemispheres. Left and right mean fMRI signal time courses in reference regions at rest when SWDs are not occurring in WAG/Rij rats (Fig. 2A) showed high correlation between the left and right regions. In contrast, nonepileptic Wistar control rats (Fig. 2B) showed less interhemispheric correlation. We performed group analyses using these reference regions by correlating them both with the whole brain (Fig. 3), and more specifically with the contralateral reference region (Fig. 4). First, to determine which regions overall are most correlated at rest with the regions most intensely involved by seizures, we correlated each cortical SWD reference region with the time course of all other voxels in the brain at rest, when no SWD were present. In the WAG/Rij group I data, bilateral cortical areas including somatosensory and adjacent cortices, where seizures are most intense, showed high degrees of resting functional connectivity with the left cortical reference region (Fig. 3A). Similar results were obtained using the right cortical seizure region as reference (Fig. S2). In contrast, much less correlation with the reference region was observed when the same anatomic region was analyzed in nonepileptic Wistar control rats (Fig. 3B). In reviewing maps from individual animals (data not shown), we observed similar high degrees of cortical–cortical correlations between the hemispheres in 100% or six of six WAG/Rij group I rats studied compared to 0% or zero of five nonepileptic Wistar controls, suggesting this is a robust measure that may ultimately be useful for testing individual subjects. Basal ganglia interestingly did not show high positive or negative correlations with the reference regions at rest (Fig. 3), despite strong negative fMRI signals during SWDs (Fig. 1).

image

Figure 2. Examples of left and right mean fMRI signal time courses in reference regions (Fig. S1) at rest when SWDs are not occurring. WAG/Rij rat (A) shows high correlation between slow (<0.1 Hz) changes in left and right reference regions. Nonepileptic Wistar control rat (B) shows less interhemispheric correlation in left and right reference regions.

Download figure to PowerPoint

image

Figure 3. Resting functional connectivity differences in epileptic WAG/Rij rats and Wistar control rats. Left cortex reference region of interest (see Fig. S1) versus whole brain. (A) WAG/Rij rats (n = 6 rats, seven experimental runs) show increased connectivity in contralateral somatosensory (S1BF) and adjacent cortices. (B) Wistar control rats (n = 5 rats, 12 experimental runs) show only localized connectivity in the ipsilateral (left) somatosensory cortex. Increases are in warm colors and decreases are in cold colors. Similar results were obtained using the right cortex ROI as reference (see Fig. S2).

Download figure to PowerPoint

image

Figure 4. Interhemispheric resting state cortical connectivity. Increased interhemispheric resting cortical connectivity in WAG/Rij rats (mean ± standard deviation of z-scores, from left and right brain regions, 2.31 ± 1.06 and 2.47 ± 1.41, respectively) compared to nonepileptic Wistar controls (1.13 ± 0.23 and 1.11 ± 0.18). Epileptic WAG/Rij: n = 6 rats, seven experiments. Wistar controls: n = 5 rats, 12 experiments. (**p = 0.01, one-way ANOVA followed by Tukey's HSD method for post hoc pairwise comparisons). Results were similar using either the left ROI time course as reference (red) and calculating mean z-score for all voxels in the right ROI, or using the right ROI time course as reference (blue).

Download figure to PowerPoint

As another approach to obtain group statistics comparing WAG/Rij and control animals, we analyzed specific region-to-region connectivity between the left and right cortical reference functions (Fig. 4). Resting interhemispheric cortical–cortical correlations (Fig. 4) were significantly higher in group I WAG/Rij rat data at rest (when no SWDs were present) compared to Wistars (p < 0.01). Interhemispheric resting cortical connectivity in WAG/Rij rats was increased in left reference region to right cortex (mean ± SD of z-score, 2.31 ± 1.06) and right reference region to left cortex (mean ± SD of z-score, 2.47 ± 1.41) compared to left reference region to right cortex (mean ± SD of z-score, 1.13 ± 0.23) and right reference region to left cortex (mean ± SD of z-score, 1.11 ± 0.18) in nonepileptic Wistar controls.

For comparison, we also performed resting connectivity analysis in runs from WAG/Rij rats without any SWD (WAG/Rij group II). Connectivity values were intermediate between the WAG/Rij group I and Wistar control data, and also showed greater variability (larger group SD). For WAG/Rij group II runs without any SWDs (n = 6 rats, 11 runs) means z-score for left reference region to right cortex was 1.62 ± 0.42 (mean ± SD) and for right reference region to left cortex it was 1.27 ± 0.44. Neither was significantly different from WAG/Rij group I or from controls.

Discussion

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Our results show increased correlation in low frequency BOLD fMRI fluctuations in the seizure-onset regions between hemispheres of the WAG/Rij rat genetic model of absence epilepsy. Although most prior studies of brain disorders showed decreased connectivity (Auer, 2008), it is reasonable to expect increased between-hemisphere connectivity in regions involved in absence epilepsy because it emerges from bilateral abnormal enhanced excitability. We recently reported abnormal increased connectivity between the hemispheres in patients with childhood absence epilepsy in seizure-related regions (lateral orbitofrontal cortex), even when seizures were not occurring (Bai et al., 2011). These findings provide further evidence for bilateral network abnormalities in this disorder, which may contribute to abnormal synchrony and excitability during the interictal period. Absence epilepsy shows widespread, bilaterally symmetric BOLD fMRI changes during seizures in animal models (Tenney et al., 2003; Nersesyan et al., 2004a; Mishra et al., 2011b) as well as in children with absence epilepsy (Salek-Haddadi et al., 2003; Laufs et al., 2006; Moeller et al., 2008; Bai et al., 2010). Although our findings support the concept of abnormal bilateral network synchrony during ictal (Musgrave & Gloor, 1980; Vergnes et al., 1989) as well as in the interictal period, and preferentially involving the regions most intensely involved in seizures, we do not know what causes these increased interhemispheric correlations in WAG/Rij rat model of absence epilepsy. One could speculate that the abnormally enhanced interhemispheric connectivity leads to functional changes in cortico–cortical connections in the regions of seizure discharges we observed and it could represent a preseizure or subthreshold excitatory phenomenon not detectable on EEG. Of note, we did not see significantly enhanced resting interhemispheric connectivity in WAG/Rij rats during runs in which no SWD occurred (WAG/Rij group II). One possibility is that the putative subthreshold activity not detectable on EEG is enhanced during the interictal period in temporal proximity to SWD activity, which occurred in the WAG/Rij group I but not group II data. Another possibility is that the five WAG/Rij group II animals that had no SWD activity in any runs simply had a less severe epileptic phenotype and less connectivity at all times. To distinguish these possibilities, future work should more firmly establish the severity of the phenotype in each animal through chronic EEG recordings at other times from the EEG-fMRI. Abnormal connectivity could be a consequence of recurrent SWD or may be caused by some other aspect of the underlying disorder. However, chronic seizures cause network dysfunction that could contribute to some of the long-term attentional and other psychosocial problems seen in the generalized epilepsies (Wirrell et al., 1997; Camfield & Camfield, 2002). Abnormally increased resting functional connectivity may also reflect chronic, abnormal, functional integration resulting from anatomic abnormalities or plasticity (Luo et al., 2011). It will be useful to test if the regions of abnormal cortical connectivity are capable of driving one another. This can be accomplished by performing cortical stimulation experiments to see if one drives another or some other brain structure is responsible for driving bilateral cortex at the same time. In human brain, correlations in low-frequency BOLD fluctuations are considered to reflect cortico–cortical connections (Lowe et al., 2000). Therefore, enhanced connectivity could be a specific characteristic of absence epilepsy, which involves dissemination of bilateral slow-wave discharges and seizures, which might also lead to partial attention impairment and loss of consciousness during seizures (Killory et al., 2011).

Resting functional connectivity approaches offer a number of advantages in studying interictal changes in absence epilepsy. The relatively slow time scale of neurovascular events measured by resting fMRI connectivity provides a window into disease mechanisms over the longer time scale of the interictal period. In contrast to EEG, fMRI allows for detection of whole brain signals simultaneously. Other neuroimaging techniques used in investigating interictal changes in generalized epilepsy are fluoro-2-deoxy-d-glucose positron emission tomography (PET) (Hikima et al., 2004; Ciumas et al., 2010), magnetic resonance spectroscopy (Helms et al., 2006), and voxel-based morphometry in MRI (Caplan et al., 2009). These techniques have been useful in investigating interictal changes in idiopathic generalized epilepsy. Similar to the above-mentioned methods, the resting functional connectivity could potentially serve as a useful biomarker for absence epilepsy during the interictal period and this may have advantage for detecting abnormal network function.

If early and prolonged treatment is to be administered to prevent epileptogenesis in children with absence seizures (Blumenfeld et al., 2008), it will be crucial to detect beneficial effects of treatment without needing to wait until adulthood. Therefore, safe and noninvasive methods are needed to monitor whether biologic changes associated with epileptogenesis (Blumenfeld et al., 2008; Chahboune et al., 2009; Mishra et al., 2011a) are effectively blocked by early treatment. Neuroimaging can detect brain biologic changes noninvasively, and animal models such as the WAG/Rij rat are ideal for initial testing of neuroimaging biomarkers of epileptogenesis. One possible concern is that this study was performed under anesthesia, whereas fMRI in children to monitor therapy would be done in the awake state. Use of fentanyl and haloperidol anesthesia greatly enhances our ability to obtain stable and reliable MRI measurements in the rodent model, without blocking SWD. Previous studies demonstrate that SWD under fentanyl and haloperidol anesthesia (Pinault et al., 1998; Nersesyan et al., 2004a) are similar to the awake state (Meeren et al., 2002; Blumenfeld et al., 2008; Chahboune et al., 2009) in terms of spatial and temporal distribution of seizures. However, anesthesia depth is known to modulate resting functional connectivity (Lu et al., 2007). It should be noted that anesthesia tends to reduce resting functional connectivity, but in our study we observed very significantly increased connectivity in the epileptic animals despite anesthesia. The large increase in connectivity we observed in the somatosensory and adjacent cortices of epileptic animals compared to our controls, and compared to the lower values observed by others (Lu et al., 2007) in normal animals in this region, suggest that the differences between groups are unlikely to be caused entirely by anesthesia. Another potential limitation of the current model is that unlike human patients, the rat cortex shows mainly fMRI increases in relation to SWD (Nersesyan et al., 2004a; Tenney et al., 2004; Mishra et al., 2011b). fMRI decreases in the rat model were confined mostly to the basal ganglia, which did not show significant connectivity to other seizure regions.

Conclusions

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Even when seizures are not occurring, the regions in which seizures are most intense show markedly increased resting functional connectivity in epileptic WAG/Rij rats compared to Wistar control rats. This finding can be a promising biomarker for monitoring SWD epileptogenesis. Our findings can also be useful in monitoring treatment effects in absence epilepsy and help us understand the role of brain plasticity in this disease.

Acknowledgments

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

This work was supported by NIH R01 NS049307 and R01 NS066974 (HB), Epilepsy Foundation Postdoctoral Research and Training Award ID: 123505 (AMM), P30 NS052519 (FH), and by the Betsy and Jonathan Blattmachr family. We thank Dr. Bei Wang and Xiaoxian Ma for their technical assistance in animal preparation, the engineering staff of the Magnetic Resonance Research Center including Peter Brown and Scott McIntyre and Quantitative Neuroscience with Magnetic Resonance (http://qnmr.yale.edu) for hardware maintenance and radio frequency probe construction.

Disclosure

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

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. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgments
  8. Disclosure
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
epi12227-sup-0001-FigureS1.tifimage/tif25864KFigure S1. Regions of interest (ROIs) used as reference regions for resting functional connectivity analysis.
epi12227-sup-0002-FigureS2.tifimage/tif44389KFigure S2. Resting functional connectivity differences in epileptic WAG/Rij rats and Wistar control rats.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.