Alteration of global workspace during loss of consciousness: A study of parietal seizures

Authors

  • Isabelle Lambert,

    1. CHU Timone, Service de Neurophysiologie Clinique, Assistance Publique des Hôpitaux de Marseille, Marseille, France
    Search for more papers by this author
  • Marie Arthuis,

    1. CHU Timone, Service de Neurophysiologie Clinique, Assistance Publique des Hôpitaux de Marseille, Marseille, France
    2. Institut de Neurosciences des Systèmes, Inserm UMR, Marseille, France
    Search for more papers by this author
  • Aileen McGonigal,

    1. CHU Timone, Service de Neurophysiologie Clinique, Assistance Publique des Hôpitaux de Marseille, Marseille, France
    2. Institut de Neurosciences des Systèmes, Inserm UMR, Marseille, France
    Search for more papers by this author
  • Fabrice Wendling,

    1. Inserm, Université Rennes 1, Rennes, France
    Search for more papers by this author
  • Fabrice Bartolomei

    1. CHU Timone, Service de Neurophysiologie Clinique, Assistance Publique des Hôpitaux de Marseille, Marseille, France
    2. Institut de Neurosciences des Systèmes, Inserm UMR, Marseille, France
    3. Hôpital Henri Gastaut/Centre Saint-Paul, Cinapse, 300 Bd Sainte-Marguerite, Marseille, France
    Search for more papers by this author

Address correspondence to Fabrice Bartolomei, Service de Neurophysiologie Clinique, CHU Timone-264 Rue st Pierre, F-13005-Marseille, France. E-mail: fabrice.bartolomei@ap-hm.fr

Summary

Purpose:  Loss of consciousness (LOC) in epileptic seizures has a strongly negative impact on quality of life. Recently, we showed that LOC occurring during temporal lobe seizures was correlated with a nonlinear increase of neural synchrony in associative—and particularly parietal—cortices. Whether these mechanisms might be observed in other types of seizures is unknown. This study aimed at investigating the relationship between changes in synchrony and degree of LOC during parietal lobe epilepsy (PLE), a form of epilepsy in which seizures directly involve the parietal associative cortices.

Methods:  Ten patients undergoing stereoelectroencephalography (SEEG) during presurgical evaluation of PLE were studied. The LOC intensity was scored using the Conscious Seizure Scale (CSS). For each studied seizure (n = 29), interdependencies between signals recorded from six brain regions were estimated as a function of time by using nonlinear regression analysis (h2 coefficient).

Key Findings:  Seizures were divided into three groups according to the CSS scale: group A (no LOC) with a score ≤1, group B (intermediate or partial LOC) with a score ranging from 2 to 5, and group C (maximal LOC) with a score ≥6. The majority of seizures in patients with PLE disclosed significant LOC (17/29, group C). Mean h2 values were significantly different between the three groups (p = 0.008), the maximal values of synchrony being observed in group C. In addition, a statistically significant nonlinear relationship (p = 0.0021) was found between the h2 values and the CSS scores, suggesting a threshold effect.

Significance:  This study indicates that excess of EEG signal synchrony within associative cortices is likely to be a crucial phenomenon associated with LOC.

Loss of consciousness (LOC) in epileptic seizures is a severe clinical manifestation affecting the quality of life of patients with epilepsy (Cavanna & Monaco, 2009; Blumenfeld, 2011).

In partial seizures, LOC occurs in 70% of patients with temporal lobe epilepsy (Maillard et al., 2004). LOC is also an important feature of seizures in patients with frontal lobe (Williamson et al., 1985) or parietal lobe seizures (Salanova et al., 1995).

Numerous theories dealing with the mechanism of LOC during partial seizures have been proposed. These theories stem mainly from investigations in the context of temporal lobe seizures. Although such seizures may arise from different parts of the temporal lobe, there is considerable evidence that LOC is linked to the extension of the discharges outside the temporal lobe in particular the thalamic nuclei (Lee et al., 2002; Guye et al., 2006) and parietofrontal associative cortices (Munari et al., 1980; Blumenfeld et al., 2004a; Englot & Blumenfeld, 2009).

Recently, we showed that LOC occurring during temporal lobe seizures was correlated with a nonlinear increase of neural synchrony within distant corticocortical and corticothalamic networks (Arthuis et al., 2009). We interpreted these results in light of the global workspace (GW) theory (Dehaene & Naccache, 2001; Baars, 2002; Dehaene & Changeux, 2011). In this theory, conscious processing results from a coherent neuronal activity between widely distributed brain regions including frontoparietal associative cortices. We then proposed that excessive synchrony as observed in the course of temporal lobe seizures with LOC prevents this distributed network from encoding conscious representations. In our previous studies the nonlinear relationship between level of synchrony and LOC followed a sigmoid curve suggesting a threshold effect (Arthuis et al., 2009).

Whether such mechanisms are involved in LOC occurring during extratemporal seizures is unknown. In particular the parietal cortex has been shown to be crucial in both externally and internally oriented conscious representations (Laureys et al., 2006; Dehaene & Changeux, 2011). It thus appears particularly pertinent to explore the mechanisms by which LOC occurs during parietal lobe seizures.

This study thus aimed at investigating the relationship between changes in synchrony and degree of LOC during parietal lobe seizures, a subtype of seizures directly involving the parietal associative regions (Bartolomei et al., 2011). The study of correlations between signals from different brain regions without any reference to physical connections (often referred to as “functional connectivity”) has been proposed as an approach to measure brain function in an individual subject and a potentially useful marker of brain disease (Bartolomei et al., 2006; Ponten et al., 2007).

Methods

Patients and SEEG recordings

Ten patients undergoing presurgical evaluation of drug-resistant parietal lobe epilepsy (PLE) were selected from a series of patients in whom intracerebral recordings had been performed between 2001 and 2008 (Table 1).

Table 1.   Clinical and electrophysiologic features of the patients
 GenderAgeAge at onsetEtiologySideCSSNumber of analyzed seizuresEZ networkPropagation network
  1. M, male; F, female; FCD, focal cortical dysplasia; R, right, L, left; SPL, superior parietal lobule, IPL, inferior parietal lobule; PHG, parahippocampal gyrus, pCG, posterior cingulate region; SMA, supplementary motor area, PoP, parietal operculum, AG, angular gyrus, SMG, supramarginalis gyrus; BA, Brodmann area; CSS, consciousness seizure scale: scores obtained from the analyzed seizures.

P1F2613CryptogenicL0;0;03SPL (BA 5)Postcentral rolandic, Premotor
P2F3610FCD precuneusR5;5;53Precuneus, pCGAmygdala, SMA
P3F3012FCD SPLR3;5;33SPL (BA 7 lateral)Premotor (SMA and BA6)
P4F3412FCD SPL (BA5)R1;02SPL (BA5)Postcentral rolandic, SMA
P5F5014FCD (IPL)R7;7;83IPL (SMG)SMA, mesial temporal
P6M387FCD (IPL)R8;8;93IPL (AG)SPL, Premotor BA6, Occipital lateral
P7M136Antenatal strokeL6;6;63PrecuneusParahippocampal Gyrus
P8F3513FCD (pOP)L7;6;83pOPSPL (BA7)
P9M258Antenatal strokeL8;6;93SPL (BA7 lateral), IPL (SMG)Lateral Temporal, Premotor BA6
P10M4012FCDR9;8;93SPL and SMAPrefrontal

All patients underwent comprehensive evaluation including detailed history and neurologic examination, neuropsychological testing, routine magnetic resonance imaging (MRI), surface electroencephalography (EEG), and stereoelectroencephalography (SEEG, depth electrodes).

SEEG exploration was carried out during long-term video-EEG monitoring. Recordings were undertaken with use of intracerebral multiple contact electrodes (10–15 contacts, length: 2 mm, diameter: 0.8 mm, 1.5 mm apart) placed according to Talairach’s stereotactic method (Bartolomei et al., 2011) as illustrated in Fig. 1.

Figure 1.


(A) Schematic representation of stereoelectroencephalographic (SEEG) exploration in the 10 studied patients. A lateral view of all depth electrodes superimposed on a three-dimensional reconstruction of the neocortical surface of the brain is shown. For simplicity, all electrodes are located on a left profile. Each electrode is orthogonally implanted and represented by a specific color for each patient. These are grouped into three regions for studying interactions (F, frontal lobe; P, parietal regions, T, temporal lobe). (B) MRI (3D T1 sequence) showing electrodes in the parietal region (internal leads in precuneus and external leads in the superior parietal lobule, SPL) and the prefrontal cortex (P8). (C) Upper part: SEEG recordings of a parietal seizure (P5). Each line corresponds to a bipolar recording between two adjacent contacts of the electrode. F8-9: lateral prefrontal cortex; PA1-2: Precuneus; PA8-9: Superior parietal lobule BA 7; GC1-2: posterior cingulate gyrus, B1-2: Anterior hippocampus, PI11-12: Inferior parietal lobule. Seizure starts from the inferior parietal lobule (PI11-12) and secondarily spread to the other explored regions. Involvement of the hippocampus (B1-2) is delayed. Lower part: Bivariate estimation of cross-correlations (computation of the h2 non linear correlation) between the six selected traces. The two periods of analysis (SO, seizure onset; MS/ES, middle seizure, End of seizure) are represented.

The anatomic targeting of electrodes was established in each patient according to available noninvasive information and hypotheses about the localization of the epileptogenic zone (for details see previous reports (Bartolomei et al., 2011). Fusion of computed tomography (CT)/MRI data was undertaken to accurately confirm the anatomic location of each contact along the electrode trajectory using MEDINRIA software (http://gforge.inria.fr/projects/medinria). In select patients, several distinct functional regions of the parietal lobe were explored. The schematic position of electrodes and the terminology used in our group is indicated in Fig. 1. However the electrode map may differ from one patient to another, since electrode placement depended on the details of each clinical case. Signals were recorded on a 128-channel Deltamed System (Paris, France), and they were sampled at 512 Hz and recorded on a hard disk (16 bits/sample) using no digital filter. A high-pass filter (cut-off frequency equal to 0.16 Hz at −3 dB) was used to remove very slow variations that sometimes contaminate the baseline. Table 1 provides clinical information about the patients selected for the purpose of this study.

Determination of LOC during seizures: the consciousness seizure scale (CSS)

Two to three representative seizures from each patient were analyzed. Video stereotactic-EEG (SEEG) recordings were reviewed and LOC intensity was scored by two of the authors (IL and FB) using an eight criteria scale—Consciousness Seizure Scale, CSS (Arthuis et al., 2009). The mean of the two scores was retained for each analyzed seizure.

This scale takes into account different aspects of consciousness in humans: (1) unresponsiveness (criteria 1 and 2), (2) visual attention (criterion 3), (3) consciousness of the seizure (criterion 4), (4) adapted behavior (criterion 5), and (5) amnesia (criteria 6 and 7). The last criteria is a “global appreciation criteria” scored between 0 and 2 representing the global appreciation of LOC made by the epileptologist about the conscious state of the patient. Each seizure has been scored from 0 (no alteration of consciousness) to 9 (complete alteration of consciousness).

Finally, 29 video-recorded seizures from those 10 patients were analyzed.

SEEG signal analysis

Definition of regions of interest

In this study, we specifically analyzed the relationship between bipolar intracerebral EEG signals (derived from two contiguous leads of the same electrode) recorded from regions of interest. These regions could be variable from one patient to the other but included at least one frontal and one parietal region. The detailed anatomic regions analyzed in each patient are presented in Table S1. For each patient we chose bipolar derivations representative of at least one frontal region, 4–5 parietal subregions, and in the majority of patients, one temporal region. Signal correlation (“synchronization”) was evaluated between SEEG signals recorded from these selected regions. Therefore, the estimation of correlations involved both intralobar synchronization (for interactions in the parietal lobe) and long-distance interlobar synchronization (parietotemporal, frontotemporal, or parietofrontal interactions).

Periods of interest

To compare the 29 seizures previously selected, three periods of interest were defined. Each period was defined according to the dynamic properties of intracerebral signals during seizures. These periods have therefore been defined independent of the clinical analysis of the seizures. They reflect the neural activity during the seizure (two periods) as compared to a background one.

  • Background (BKG): this period of 30 s was selected at least 1 min apart from the onset of the ictal discharge, during a period of awake quiet rest and of normal consciousness state. When the seizure was elicited by electrical stimulation, the BKG period was selected one minute before the onset of the first stimulation.

  • Seizure Onset (SO): we have arbitrarily chosen a duration of 10 s including 5 s before and 5 s after the appearance of a rapid discharge in parietal structures. This rapid discharge was delimited by visual inspection. A time-frequency representation of signals was also used to accurately determine the beginning of the rapid activity (Guye et al., 2006).

  • Middle/End part of the seizure (MS/ES): this period follows the previous one (SO) and covers the rest of the seizure period to the end of the discharge in the explored brain regions (Fig. 1B).

Estimation of long-distance synchronization during seizures

The synchronization of EEG signals between two distant regions may be estimated by several different methods (Stam, 2005) (Ansari-Asl et al., 2006) aimed at evaluating the degree of mathematical relationship between signals. In the present paper, interdependencies between signals recorded from six structures were estimated as a function of time by using nonlinear regression analysis (Pijn & Lopes Da Silva, 1993). Details of the method can be found in our previous studies (Bartolomei et al., 2004; Guye et al., 2006; Arthuis et al., 2009). Nonlinear regression analysis provides a parameter, referred to as the nonlinear correlation coefficient h2, which takes values in the range [0, 1]. Low values of h2 denote that signals X and Y are independent. In contrast, high values of h2 mean that signal Y may be explained by a transformation (possibly nonlinear) of signal X, that is, signals X and Y are dependent. The analysis was performed over a sliding window (duration 2 s) by steps of 0.25 s. The h2 values were averaged over each period of interest defined above, for each of the 10 considered pairs of signals (see below) and for each of the 29 seizures recordings. We particularly chose to study the values obtained between each individual bipolar derivation representative of one brain region and all the other selected bipolar channels.

In the present study, h2 values were computed on broadband signals (0.5–90 Hz), providing a global estimation of nonlinear interdependencies.

Statistical analysis

For each seizure, correlation values (h2 values) computed from each ictal period (BKG, SO, MS/ES) were compared using a nonparametric Wilcoxon test.

Differences in the distribution of h2 values between the three groups of seizures defined with the CSS were compared using the Kruskal-Wallis analysis.

A logarithmic regression was performed to better capture the nonlinear association observed between h2 values and CSS scores. A Spearman correlation test was applied to correlate the regional changes in correlations and the CSS scores. A p-value < 0.05 was considered significant in all the statistical analyses.

Results

A summary of clinical and SEEG data obtained in the 10 patients is provided in Table 1.

Alteration of consciousness in parietal seizures

Seizures were divided into three groups according to the consciousness alteration scored using the CSS scale: group A (no LOC) with a score ≤1, group B (intermediate or partial LOC) with a score ranging from 2 to 5, and group C (maximal LOC) with a score ≥6. Group A included 4 seizures, group B 6 seizures, and group C 17 seizures. Therefore, the majority of seizures in patients with parietal seizures disclosed important LOC (17/29, group C). These three groups are represented in Fig. 2A.

Figure 2.


(A) Distribution of the consciousness scores using the CSS scores according to the three defined groups of seizures (A, B, C). Mean (bar) and median (cross) are represented. (B) Global changes in h2 values (averaged from all interactions in all the analyzed seizures) during background period (BKG), seizure onset (OS), and middle/end of seizure (MES) periods. Increased values are observed in the SO and MES periods. Differences are significant with the BKG values (p < 0.0001). (C) Box plot representation of the correlation values (h2 values averaged from all interactions) in the three groups of seizures. Values are maximal in group C and minimal in group A, and disclosed intermediary levels in group B.

Changes in synchrony associated with parietal seizures

We have first estimated the global changes in signal synchronization by averaging for each analyzed seizure the values from all the six selected regions (15 interactions). As indicated in Fig. 2B, we observed an increase of h2 values. Increased correlation in comparison with the reference “background region” (mean 0.08 ± 0.03) was significant in the OS period (mean h2 0.165 ± 0.09) as well as for the ME/ES period (0.173 ± 0.1) (p < 0.0001). No significant difference was found between SO and ME/ES values (p > 0.05).

Relationship between altered consciousness and synchronization changes

Because no difference was found between SO and ME/ES values, we used the ME/ES values in the following analysis.

We looked for differences in the values of averaged interactions between the six regions (15 interactions) between group A (with no LOC), B (intermediate LOC), and group C (maximal LOC).

As illustrated in Fig. 2C, the values were significantly different between the three groups (p = 0.008), the maximal values of synchrony being observed in the group C. Post hoc analysis (Wilcoxon) showed that differences were significant for B versus C (p = 0.009), A versus C (p = 0.002), and almost reached significance for B versus A (p = 0.055).

Examples of three different seizures disclosing different values of LOC are depicted in Fig. 3A. As illustrated, pattern of synchrony was different in the three seizures, with maximal synchrony level and distribution observed in group C and B seizures.

Figure 3.


(A) Example of correlation changes in three examples of seizures: a group A seizure, a group B seizure, and a group C seizures during MES period. Figure represents graphs of correlation and synchronization matrices. The synchronization matrix is a 6 × 6 square matrix, where the x axis and the y axis correspond with the channel numbers (each representing one brain region), and where the entries indicate the mean strength of the h2 between specific pairs of channels. The strength of the h2 is indicated with a color scale, from blue (h2 = 0) to red (h2 = 1). The diagonal running from the upper left to the lower right is intentionally left red. Slight differences can be observed along the diagonal because of the asymmetrical nature of the nonlinear correlation coefficient h2 (values of the h2 coefficient are different when the computation is performed from signal X to signal Y vs. Y to X). The graphs of correlations represent the mean values of hxy2 and hyx2 and are thresholded with the same value (0.1) for each case. Only link with values above this threshold are represented. More links and links with increased values are observed in the group C seizure with regard to the group A and B seizures. (B) Nonlinear (logarithmic regression) between h2 values and CS scores (MES period). This relation follows a sigmoid curve suggesting a nonlinear bi-stable function for consciousness.

Therefore, the hypothesis of more synchronized brain network in the group with major LOC is in agreement with these results.

We then looked at the relationship between h2 values and consciousness scores. This relation is shown in Fig. 3B and appears to be nonlinear. A logarithmic regression was used to better capture the nonlinear relationship between CSS scores and h2 vales and found a significant relationship (p = 0.0021).

Relation between correlations values in different anatomic regions and consciousness

We then investigated whether CSS scores were differently correlated with values averaged for interactions affecting frontal electrodes (Fr), parietal lateral regions (ParL), parietal medial regions (Pcu), and temporal regions (Temp).

To this aim, we have computed the average synchronization (h2) between each channel belonging to a given region (Fr, ParL, Pcu, Temp) and all other channels. We then correlated these averaged values to the CSS values.

A significant relationship was found for parietal lateral cortex (Spearman correlation, p = 0.01), frontal cortex (p = 0.0009), and to a lesser extent for medial parietal cortex (p = 0.04). No significant relationship was found for temporal lobe regions (p = 0.08).

Discussion

This study investigated the hypothesis that excess synchronization affecting brain regions essential for consciousness processing is a potential mechanism of the alteration of consciousness during extratemporal lobe seizures.

Our results are in agreement with this hypothesis. Indeed we found that seizures associated with LOC were associated with a more marked synchronization of SEEG signals, and that a significant nonlinear relationship was found between the level of consciousness alteration and the level of synchrony. Maximal correlation was found between consciousness scores and alterations occurring in parietal and frontal regions. The curve of the relationship between altered synchronization and consciousness alteration is suggestive of a nonlinear process. It appears that above a certain value of synchrony, most of the patients have an important loss of consciousness.

These results are close to those we found in temporal lobe seizures (Arthuis et al., 2009; Bartolomei & Naccache, 2011) and are also in agreement with current hypotheses linking LOC with the involvement of parietofrontal associative cortices (Blumenfeld et al., 2004a,b). In temporal lobe seizures, the LOC is not linked to the alteration of temporal lobe function alone but has been correlated to a slowing of the EEG activity (Blumenfeld et al., 2004b) and is correlated with excess synchronization between thalamus and parietal cortex (Arthuis et al., 2009).

In TLE seizure we also found that above a certain level of synchronization most of the seizures were associated with an alteration of consciousness (Arthuis et al., 2009).

Under physiologic conditions, consciousness representations have been associated with transient synchrony between engaged brain regions (Dehaene & Changeux, 2011). For example, visual consciousness is associated with an increased synchrony between areas extending outside the limits of the visual system (Rodriguez et al., 1999; Gaillard et al., 2009). Finally, parietofrontal associative cortices have been found to be crucial in the consciousness access and representation in different sensorial modalities (review in Dehaene & Changeux, 2011; Dehaene & Naccache, 2001; Rees et al., 2002). It is noteworthy that the involvement of these cortices is a plausible common mechanism of different states of consciousness alterations without vigilance alteration (Laureys et al., 2005, 2007). Because parietal seizures directly involve these associative cortices, this could be a plausible explanation of the significant and highly prevalent LOC that we observed in our patients.

The role of these cortices can also been interpreted in light of the “default mode network model” (DMN) (Buckner et al., 2008). Indeed parietal (in particular posterior cingulate cortex and precuneus) and prefrontal cortices are thought to be the core of this network. These regions maintain a great level of activity at rest, whereas they showed decreased activity during different cognitive tasks (Raichle et al., 2001). The defenders of this still-controversial concept hypothesized that this deactivation corresponds to the arrest of the “default state of the brain” during which the subject is not performing any specific task, but is awake and conscious. The role of DMN in LOC during seizure is however not well known and understood, in particular because DMN regions has been found to decrease their activity in epileptic seizures with LOC (see review in Danielson et al., 2011).

Our study has several limitations. In particular we are unable to evaluate the importance of the contralateral involvement, since most included patients had only unilateral depth electrode placement. In addition, the coverage of the frontal regions was generally limited to one electrode. We cannot therefore precisely determine which structures in the frontal regions are the most importantly affected in LOC.

In conclusion, the present study confirms that increased synchrony between distant cortical regions could be an important link between LOC and changes in EEG activity. In parietal seizures we found that a majority of patients presented LOC, the degree of which was greater in patients with more synchronized cortical regions. We postulate that this phenomenon is not contingent but is an important potential mechanism leading to LOC.

Better understanding of the mechanisms underlying LOC in epileptic seizures could eventually lead to development of specific therapeutic approaches aimed at decreasing this oversynchrony. Interesting perspectives emerge from the results of experimental data showing that direct or indirect brain stimulation is able under certain condition to decrease brain synchrony (Ananda et al., 2009; Besio et al., 2011; Nichols et al., 2011). This type of research merits future development, since means of decreasing consciousness alteration in seizures could improve patients’ quality of life.

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.’

Ancillary