Termination Patterns of Complex Partial Seizures: An Intracranial EEG Study

Abstract

1. Introduction

Continuous video-EEG monitoring in the presurgical evaluation of patients with intractable partial epilepsy appropriately focuses on patterns and localization of seizure onset zones. Accurate determination of seizure termination requires intracranial recordings. There are few studies of seizure termination, in part because this has little impact on patient selection for resective surgery. The few published analyses of patterns of seizure termination focus on relationships to surgical outcomes [1, 2, 3].

Unifocal partial seizures often have a stereotyped pattern of initiation, whether assessed by visual inspection of the EEG or by more sophisticated methods of time-frequency decompositions [4, 5, 6]. Some patients with either mesial temporal or neocortical onset seizures interestingly appear to have simultaneous or near simultaneous seizure termination in all intracranial contacts.

An early report by Spencer and Spencer [1] examined intracranial ictal recordings from patients exclusively with temporal lobe epilepsy with bilateral subdural and/or depth arrays and discussed seizure termination classified as localized to onset, focal elsewhere than onset, or diffuse. Another study of temporal lobe seizure offset termination patterns by Brekelmans et al. [2] again looked at seizure offset in terms of ipsilateral, contralateral, or diffuse/bilateral. Interestingly the Spencer and Spencer study [1] found a correlation of seizure freedom with offset in the seizure origin zone while the Breckelmans study [2] did not. These studies examined seizure offset based on localization, and timing of seizure offset; the concept of synchronous termination was not the focus of discussion. Indeed the illustration of a “diffuse” offset in the Breckelmans study [2] may reveal just such a multichannel synchronous termination pattern. Alternatively, however, a diffuse offset may also exhibit an asynchronous termination pattern. The report by Verma et al. [3] studied 13 patients with bilateral depth arrays and examined the side of termination in the context of outcome. A recent study of scalp patterns in children discusses an abruptly attenuated terminal ictal pattern which is synchronous termination, but this was more common in idiopathic and generalized seizures, a different population than has been studied elsewhere [7].

In the study here we report a group of patients with partial seizures from mesial temporal (MT) or neocortical (NC) onset to determine how often synchronous termination is seen, how often asynchronous termination is present, the frequency of mixed patterns, and the relationship of these patterns to seizure duration. Such synchronous patterns of termination may reflect properties of the seizure focus or networks involved in propagation. Some of these results have been presented in abstract form [8].

2. Methods

3. Results

Tables 1 and ​and22 list the characteristics of the two patient groups (MT and NC). The tables list patient’s sex and age at the time of monitoring, MRI abnormalities, identified histopathology, pattern of termination, location of termination, surgical outcome, number, side and type of electrodes.

Figure 1 illustrates an example of synchronous termination. Cessation of seizure activity is seen in all of the recorded electrodes nearly simultaneously. On the right is an illustration of the GAD plot for the respective channels of the same seizure. As mentioned in the methods, GAD is a composite measure of signal complexity derived from the matching pursuit time-frequency decompositions of individual channels. Multichannel GAD plots are assembled from the individual channel analyses. This increase in GAD correlates with seizure duration [4, 5, 6]. GAD can be used to measure the level of complexity, patterns of propagation, and seizure termination. Here the multichannel GAD plots reveal not only the channels involved with seizure propagation, but the abrupt drop in signal complexity (to blue scaling) that correlates with decreased signal complexity, seizure termination, and the post-ictal period.

Open in a separate window

Figure 1

ICEEG (left) and color-coded GAD propagation map (right) for a mesial temporal onset seizure with a synchronous pattern of termination. Time zero indicates the onset of the seizure. Yellow shading on the ICEEG marks the duration of the seizure. Channels are in the same order for the ICEEG and the GAD propagation map and reflect consecutive numbering of intracranial contacts, which may not be contiguous. Red indicates the highest level of signal complexity. Blue indicates lowest level of signal complexity and orange and yellow are intermediate levels. The black arrow shows the electrodes with earliest sustained ictal activity.

Figure 2 illustrates one pattern of asynchronous termination where the seizure activity at the focus stops while seizure activity continues in some other channels. In some other instances asynchronous termination occurred after more regional onset or after transient non-focal participation (not shown). All asynchronous terminations for a given patient were grouped together for analysis.

Open in a separate window

Figure 2

ICEEG (left) and color-coded GAD propagation map (right) for a mesial temporal onset seizure with an asynchronous pattern of termination, in this case with termination earlier in the seizure onset zone. Time zero indicates the onset of the seizure. Yellow shading on the ICEEG marks the duration of the seizure. Channels are in the same order for the ICEEG and the GAD propagation map and reflect consecutive numbering of intracranial contacts, which may not be contiguous. Red indicates the highest level of signal complexity. Blue indicates lowest level of signal complexity and orange and yellow are intermediate levels. The black arrow shows the electrodes with earliest sustained ictal activity.

There were 26 patients in the MT group and 33 in the NC groups in our study. In the MT group, ten of 26 patients (38%) of patients had seizures with only synchronous offset and three of 26 (12%) of patients had only asynchronous termination. Thirteen of 26 patients (50%) of the patients had mixed termination patterns with some synchronous and others with asynchronous patterns of termination (Figure 3).

Open in a separate window

Figure 3

Number of patients with each pattern of termination, either S_o synchronous only, A_o asynchronous only, or S/A both synchronous and asynchronous termination patterns. MT: mesial temporal group. NC: neocortical group.

There were 33 patients in the NC group. Seventeen of 33 (52%) had seizures with only synchronous pattern of termination and six of 33 patients (18%) had only seizures with asynchronous pattern. Ten of 33 patients (30%) had mixed patterns of termination (Figure 3).

Table 3 lists the patterns of termination for both MT and NC groups noted whether bilateral arrays (either depth electrodes or subdural strips) or unilateral arrays (combinations of subdural strips and grids). If all seizures ended synchronously (S_o) or asynchronously (A_o) they were placed in these respective groups. If patients had mixed patterns of termination, both synchronous and asynchronous then they were grouped into the S/A group. In the MT group only there is a suggestion that bilateral arrays (7 depth only; 5 with subdural arrays only) might detect a few more patients with asynchronous termination. In three of the seven patients with bilateral depth arrays, all seizures had synchronous terminations with two patients in the A group and two in the S/A. Because intracranial arrays are determined based on individual patient needs, not all patients had bilateral recording arrays.

Figures 4a and 4b illustrate the scatterplots of seizure duration for the patients with MT or NC seizures that had exclusively synchronous (S_o) or asynchronous (A_o) termination patterns. Most (7 of 10) MT patients and (16 of 19) NC patients with exclusively synchronous termination patterns had only unilateral recording arrays (Table 3). In the MT patients with exclusively synchronous termination patterns, the seizure durations were usually (9 of 10) closely spaced with low variability of the range of seizure durations for each patient. The comparison of the range of seizure durations did not reveal any significant difference between the A_o and S_o groups (Mann-Whitney U=6.0; p=0.128). Since the A_o group contains only three patients this severely limits the robustness of this assessment.

Open in a separate window

Figure 4

a) Scatterplot of the seizure durations for patients with mesial temporal lobe onset seizures and only synchronous patterns of termination (S_o: closed circles) or only asynchronous patterns of termination (A_o: open circles). Patient numbers are on the x-axis; b) Scatterplot of the seizure duration for patients with necortical onset seizures and only synchronous patterns of termination (S_o: closed circles) or only asynchronous patterns of termination (A_o: open circles). Patient numbers are on the x-axis.

In contrast, the NC patients with exclusively synchronous termination patterns (S_o) had a significant difference of the range of seizure durations (Mann-Whitney U=19.0; p=0.02) with the A_o group exhibiting a larger range of seizures compared to S_o. As was seen with the MT group the range of seizure durations for the S_o group in NC had low variability whereas the A_o often exhibit higher variability.

Figures 5a and 5b illustrate those patients with mixed asynchronous and synchronous termination patterns. We tried to test the hypothesis that within the same patient that the seizure with asynchronous termination pattern (open circles) had longer seizure duration than synchronous termination pattern (closed circles). The data is relatively complex due to the repeated measures design (several seizures) with unbalanced group (unequal number of seizures) for each patient. Although ANOVA analysis can be corrected for unbalanced groups, the comparison of variance to allow for ANOVA testing was not satisfied. This was likely due to the observed increase variability of the asynchronous seizures compared to the synchronous seizures. Individual comparison of seizure variability within a given patient remains possible.

Open in a separate window

Figure 5

a) Scatterplot of the seizure duration for patients with mesial temporal lobe onset seizures and mixed patterns of seizure termination with both synchronous patterns of termination (S_o: closed circles) and asynchronous patterns of termination (A_o: open circles). Patient numbers are on the x-axis; b) Scatterplot of the seizure duration for neocortical patients with mixed patterns of seizure termination with both synchronous patterns of termination (S_o: closed circles) and asynchronous patterns of termination (A_o: open circles). Patient numbers are on the x-axis.

Analysis of the range of seizure durations between groups S and A for MT and NC patients with mixed seizures (i.e. both types) was performed on patients where the range could be computed (i.e. more than 2 seizures in each group). The remaining patients did not show a significant difference between S and A seizure duration spread.

In the MT group with only synchronous termination 8 of 10 patients (80%) had Engel Class I outcomes; the other two had Class III or IV outcomes. In the small group of MT patients with only asynchronous offset, one of three had Class I outcome and two had Class II or III outcomes. The patients with MT seizures and mixed termination patterns nine of 13 (69%) had Engel Class I outcomes, four had Engel Class III or IV outcomes.

In the patients with NC seizures and only synchronous termination 6 of 17 (35 %) had Engel Class I outcomes, 10 of 17 (59 %) had Class III and IV outcomes, and one patients was lost to follow-up. Of the six patients with only asymmetric termination patterns, four were Engel Class III or IV, and two did not have surgery. In the NC patients with mixed patterns of termination, six of 10 (60 %) had Engel Class I outcomes, and four had Class IV outcomes.

4. Discussion

Previous studies of seizure termination patterns from intracranial recordings focused on correlation of outcome [1, 2, 3]. While undoubtedly synchronous termination patterns were observed in these studies, this is not specifically mentioned although diffuse termination patterns probably incorporate synchronous termination patterns but can also include asynchronous termination patterns as well.

The goal of this study was to quantify the occurrence of synchronous termination patterns in patients with mesial temporal and neocortical (temporal and extratemporal) seizures who had intracranial recording arrays as a part of their surgical evaluation. Interestingly 38% and 52% of MT and NC respectively had exclusively synchronous termination patterns and an additional 50% of the MT and 30% of the NC group had mixed termination patterns with at least some seizures terminating synchronously. One might have speculated that neocortical onset seizures because of their rapid spread might have been less likely to have synchronous termination patterns but this was not observed.

We hypothesized that seizures with asynchronous patterns of termination are longer than seizures with synchronous pattern of termination. This was demonstrated in the NC group, but could not be demonstrated in the MT group, due to limited patients with only asynchronous terminations and the range of seizure durations observed in these few patients.

Classifications into synchronous and asynchronous patterns of termination are descriptive terms based on timing of termination rather than localization. These do not imply degree of propagation. Synchronous termination can occur in seizures with offset areas that are limited to the region of focal onset. It can also involve a broad region of spread (Figure 1). Asynchronous termination can occur either earliest in the seizure onset region and later in remote regions (the most common pattern) or the remote involvement may end earlier than the onset region. Not all of our patients had bilateral recording arrays. While such arrays might have theoretically increased the number of asynchronous terminations, many of the synchronous terminations involved many electrodes over broad regions.

It was not the goal of this study to correlate surgical outcome with patterns of seizure termination. Previous studies have addressed this with variable results, one [2] suggesting seizure-offset pattern does not correlate with outcome after surgery, another [1] reporting that localization of seizure termination to the seizure onset region correlates with a better outcome, and a third [3], exclusive depth electrode study of 13 patients find that final termination in the contralateral temporal lobe correlates with a poorer outcome. All of these studies are of patients with temporal lobe epilepsy. Studies of neocortical onset seizures can be confounded by lesional or nonlesional causes, the extent of the resection, and the proximity of eloquent cortex. No conclusions regarding seizure termination patterns and outcome can be made from our study.

Interestingly those patients with seizures that ended synchronously (either exclusively or mixed) had seizure durations of low variability. It is already known that multiple seizures from a single region in a given patient typically have very similar patterns of onset [4, 5, 6].

How seizures terminate remains one of the unanswered questions in epilepsy. Several pathophysiologic mechanisms of seizure termination have been studied at the cellular/metabolic level including glutamate depletion [12], changes in ionic concentration [13, 14, 15], PH [14, 16]. Additionally modulatory effects from subcortical brain nuclei and cerebellum may be involved in seizure termination [16, 17]. At the network level multiple characteristics of neurophysiologic recordings have been quantitatively studied [17]. Several proposed mechanisms suggest hypersynchronicity of epileptic activity. Kramer et al. reconstructed dynamic network representations of electrocorticogram data and reported dramatic changes in topological organization throughout evolution of seizure, while the overall synchronization changed only weakly, although at seizure termination there was increased synchronization [18]. They concluded that epilepsy was manifestation of network reorganization (and not hypersynchrony).

In patch clamps studies of rat CA1 neurons asynchrony for sustained periods of time is necessary to maintain a high level of activity (during seizure-like events) and that synchrony may disrupt such activity [19]. Increase correlation that may be casually related to seizure termination has been studied in intracranial recorded seizures [20]. Another study describes increase in synchronization of neural activity prior to seizure termination in patients with status epilepticus both spontaneously and following administration of anticonvulsant drugs. Therefore increasing synchronization of neuronal activity was suggested as an emergent self-regulatory mechanism of seizure termination [21]. Additionally network dynamic studies suggest that a) synchornizability of networks decrease during seizures but increase prior to termination [22] and b) assortativity of networks increase during evolution of a seizure and decrease prior to termination [23].

Yet another computational study proposes that the seizing brain is signature of an impending critical transition [24]. The secondary generalized seizure termination is consistent with crossing a critical transition or bifurcation in which the system (i.e. the ictal state) traverses a critical threshold and shifts suddenly to an alternative, attracting dynamical regime (i.e. the postictal state).

Many of the above mentioned studies point to presence of increased synchronization upon seizure termination. The importance of the two types of seizure termination pattern is not known. One can suggest that seizures with asynchronous termination may consist of less synchronized components at the cellular or network level. This may at least in some seizures result in longer duration (which was demonstrated in the NC group). We think that the first step may be to classify the termination pattern before further studies could be directed at different pathophysiologic mechanisms underlying the two termination patterns.

5. Conclusion

Synchronous seizure termination is a commonly observed phenomenon either exclusively or associated with other termination patterns. It can be seen in both seizures of neocortical or mesial temporal origin. These patterns of termination may reflect stereotyped network involvement or dynamics at the seizure focus.

Acknowledgments

Footnotes

References