Duration of complex partial seizures: An intracranial EEG study

Transcription

1 FULL-LENGTH ORIGINAL RESEARCH Duration of complex partial seizures: An intracranial EEG study Pegah Afra, Christophe C. Jouny, and Gregory K. Bergey Department of Neurology, Johns Hopkins Epilepsy Center, Johns Hopkins University School of Medicine, 600 North Wolfe Street, Meyer 2 147, Baltimore, MD, USA SUMMARY Purpose: The dynamics of partial seizures originating from neocortical and mesial temporal regions are thought to differ, yet there are no quantitative comparative studies. The studies reported here investigate the duration of complex partial seizures in these populations using analyses of seizures recorded from intracranial arrays. Methods: Data were collected from patients undergoing presurgical evaluation with intracranial electrodes. Seizure duration was defined as the time of earliest sustained ictal activity until the termination either in all electrodes (global duration, GD), or at the onset area (focal duration, FD). Patients were divided into three groups: mesial temporal lobe epilepsy (MTLE), neocortical temporal lobe epilepsy (NCTLE), and neocortical extratemporal lobe epilepsy (NCXTLE). Results: Complex partial seizure durations were significantly longer in the MTLE group compared to the NCXTLE group. Median GD for MTLE was 106 s, and for NCXTLE was 78 s. There were no significant differences between seizure durations when comparing MTLE group to NCTLE group, or comparing NCTLE group to NCXTLE group. In the MTLE group, patients with bilateral recording arrays had significantly longer median seizure durations (GD and FD) than those sampled with unilateral arrays. Conclusions: In this select group of patients there is a significant difference between the duration of complex partial seizures of mesial temporal and neocortical extratemporal origin with mesial temporal complex partial seizures being longer. This may result from a number of possibilities including the bilateral propagation of some mesial temporal seizures and differences in ictal generators of the underlying networks. KEY WORDS: Seizure duration, Seizure termination, Seizure dynamics, Complex partial seizure, Intracranial EEG. Epileptic seizures are transient paroxysmal electroclinical phenomenon resulting from abnormal synchronous discharges of neurons. Seizures can be either focal with variable regional spread (partial seizures) or bilateral (primary or secondarily generalized seizures). The hallmark of an epileptic seizure is the temporary nature of the event. It has been generally thought that complex partial seizures (CPS) originating from neocortex, particularly those originating from frontal lobe regions, are shorter in duration and have briefer postictal periods than do seizures of mesial tem- Accepted September 28, 2007; Online Early publication November 20, Address correspondence to Pegah Afra, M.D., Department of Neurology, University of Utah Health Sciences Center, 30 North 1900 East Room 3R210 SOM, Salt Lake City, UT 84132, U.S.A. Pegah.Afra@hsc.utah.edu Blackwell Publishing, Inc. C 2008 International League Against Epilepsy poral origin; however, there are no published systematic studies comparing mesial temporal and neocortical seizure durations. There are only a few studies of the electroclinical duration of partial seizures. A recent report (Jenssen et al., 2006) addressed this in the context of defining status epilepticus and compared complex partial seizure duration with generalized tonic clonic seizures (GTCS). In this study from continuous scalp recordings, the median complex partial seizure duration was 78 s; no comparisons were made between regions of onset. A few earlier studies of seizure duration exist. These include a videotape analysis of complex partial seizure durations (Theodore et al., 1983), and a video-eeg telemetry study of GTCS duration (Theodore et al., 1994). The presence of a postictal period may confound clinical analyses and ictal scalp recordings may not accurately record seizure onset and termination. There are no reports of comparative quantitative analysis 677

2 678 P. Afra et al. of complex partial seizure duration. Here we address this issue by comparing the intracranial electrographic durations of CPS from mesial temporal and neocortical foci. METHODS ICEEG monitoring, duration definition and patient selection Intracranial recordings of 140 patients who underwent invasive intracranial EEG (ICEEG) monitoring with grids, strips, or depth electrodes as a part of their presurgical evaluation, from July 2000 to July 2006 at Johns Hopkins Epilepsy Monitoring Unit (EMU) were reviewed. Data were collected using a 128-channel video-eeg Telefactor acquisition system with a sampling rate of 200 Hz (July 2000 spring 2004) or Stellate acquisition system with a sampling rate of 1000 Hz (spring 2004 July 2006). The recordings were stripped from the patient identification data, converted to EDF format and stored for further analysis in a HIPAA compliant database. The duration of CPS were determined by visual analysis of ICEEG recordings by two observers (PA and GKB). The duration was defined as the time of earliest sustained local or regional ictal activity until the termination of ictal activity either in all electrodes (global duration, GD), or at the onset area (focal duration, FD). Some seizures terminate synchronously or near synchronously in all electrodes; others may have asynchronous terminations. To be included in the analyses, patients had to have two or more CPS. CPS was defined as any partial seizure with alteration of consciousness as assessed and documented by nursing staff. For practical purposes in the EMU, altered consciousness is defined as either decreased responsiveness or loss of awareness. Decreased responsiveness is tested by assessing the patient s ability to respond appropriately to external stimuli (either by answering simple questions or following simple commands). Awareness is patient s memory during the event and is tested by asking the patient to remember three objects. In instances of short frontal CPS, patients subjective reports or visitor assessment were used. Because of concern that closely spaced seizures could influence the duration of subsequent seizures, seizures that occurred within 60 min of the previous seizure were excluded from the study. This did not cause any of the patients to be excluded from the study. There were no seizures that exhibited start-stop-start phenomena. One patient had a prolonged seizure that waxed and waned without stopping; this did not affect the median seizure duration. Secondarily GTCS or simple partial seizures were not considered in this study. From 140 patients, 33 patients were excluded from the study due to bilateral or multifocal seizures (22 patients), ill-defined or no distinguishable seizure pattern (ten patients). Another patient was excluded because of no recorded seizures. Table 1 lists the numbers of patients excluded and included in the respective groups. Forty-five patients were in the mesial temporal lobe epilepsy (MTLE) group. Twenty-six patients had two or more CPS originating in the mesial temporal region and were included (Table 2 lists the patient s characteristics including presence of temporal lobe pathology). Sixteen patients were excluded because they had less than two CPS, with only secondarily GTCS or only simple partial seizures. In three other patients it was not possible to exclude a neocortical origin with certainty (i.e., it could not be determined whether the seizure onset was from mesial or lateral temporal lobe regions) and they were excluded. Sixty-two patients had neocortical epilepsy (NCE). We included any patient who had two or more CPS, had a diagnosis of lesional epilepsy by MRI or histopathology, or had normal MRI and the histopathological diagnosis of nonlesional epilepsy. Twenty-seven patients were included and divided into a neocortical temporal lobe epilepsy group (NCTLE, 10 patients, Table 3) and a neocortical extratemporal epilepsy group (NCXTLE, 17 patients, Table 4). Lesional causes (based on pathology or MRI) for all three groups are listed in the respective tables for reference. Nonlesional (negative pathology and MRI) and lesional patients were combined in these analyses. Twenty-eight patients had less than two CPS, with only secondary GTCS or only simple partial seizures and were excluded. Another seven patients were excluded because they had normal MRI and they either did not undergo surgery or did not have adequate sampling for Table 1. Patient inclusion and exclusion Total patients MTLE NCTLE + NCXTLE Other Included patients 27 patients N/A 10 with NCTLE 17 with NCXTLE Excluded patients 35 patients 33 patients Less than 2 CPS: 16 Less than 2 CPS: 28 Bilateral or multifocal: 22 Not possible to differentiate between Normal MRI and no histopathology Ill-defined or no distinguishable mesial and neocortical onset: 3 or no surgery: 7 ictal pattern: 10 No seizures: 1 MTLE, mesial temporal lobe epilepsy; NCTLE, neocortical temporal lobe epilepsy; NCXTLE, neocortical extratemporal lobe epilepsy; CPS, complex partial seizure.

3 679 Duration of Complex Partial Seizures Table 2. Mesial temporal lobe epilepsy summary Patient Age/sex Side MRI Path No Sz GD (s) FD (s) Electrodes No Cont MT 1 43 Fe R MTS, DV, IS NAH Bi D 48 F, AG, HC MT 2 35 Fe R MTS, DV, IS MTS Bi D 48 F, AG, HC MT 3 42 Fe R MTS, DV, IS MTS Bi D 48 F, AG, HC MT 4 30 M R MTS, DV, IS MTS Bi S 96 F, T MT 5 22 M R MTS, DV, IS MTS Bi S 86 F, T, P MT 6 16 Fe R MTS, DV, IS MTS R G, S, D 97 F, T, P MT 7 32 Fe R MTS, DV, IS MTS R G, S 112 R, T, P, O MT 8 16 M R MTS, DV, IS NAH R G 81 F, T, P MT 9 41 M R Normal MTS Bi S 74 F, T, P MT Fe L MTS, DV, IS MTS Bi D 48 F, AG, HC MT M L MTS, DV, IS MTS Bi D 40 F, AG, HC MT M L MTS, DV, IS MTS Bi D 48 F, AG, HC MT Fe L MTS, DV, IS NAH Bi (L G, S; R S) 120 L F, T, B, P & R F, P MT Fe L MTS, DV, IS NAH L G 30 T, B MT Fe L MTS, DV, IS NAH L G, S 102 F, T, P, O MT M L MTS, DV, IS MTS L G, S 68 F, T, B MT Fe L MTS, DV, IS NAH L S 96 F, T, P, O MT18 56M L LMTmass AST LG 64F,T,B MT M L Normal AHNP L G 122 F, T, P, B MT M L Normal AHNP L G 79 F, TB MT Fe L Normal AHNP L G, S 69 F, T, P, B MT22 25Fe R ROAbnSig NAH RS,RD(O) 49F,T,P,B,O MT Fe L No MRI NAH L G, S 80 F, T, P, B MT M L Normal NAH Bi (L G; R S) 95 L F, T & R T MT Fe L Normal NAH Bi D 47 F, AG, HC MT M L Normal NAH L G 112 F, T MT, mesial temporal; Fe, female; M, male; R, right; L, left; MTS, mesial temporal sclerosis; DV, decreased volume; IS, increased signal; Abn Sig, abnormal signal; AST, astrocytoma; Path, pathology; GD, median global duration; FD, median focal duration; No Sz, number of seizures; Bi, bilateral; G, grid electrodes; S, strip electrodes; D, depth electrodes; No Cont, number of contacts; F, frontal; P, parietal; T, temporal; O, occipital; B, basal; AG, amygdala; HC, hippocampus; AHNP, adequate hippocampus no pathology identified; NAH, No adequate hippocampus available for pathologic examination. not a candidate for MRI due to bilateral stapedial implants (patient had normal CT scan). Table 3. Neocortical temporal lobe epilepsy summary Patient Age/sex Loc MRI Path No SZ GD (s) FD (s) Electrodes No Cont NT 1 20 Fe R T Mig Abn CD R G and S 94 F, T, P, O, B NT 2 34 Fe R T Normal MD Bi S 107 F, T, IH NT 3 29 Fe R T GM HT GM HT R g 118 F,T,P,O NT 4 45 M R T IS in R H NLO Bi S and Bi D 121 F, T, P, AG, HC NT 5 26 Fe R T IS COR & SC WM Glioma RG 107 F, T, P NT 6 20 Fe L T Tu PXA L G 76 F, T, B, P NT 7 18 Fe L T Gliosis HyC L G 104 F, T, B NT 8 32 M L T T2 IS: L T & F WM CaVM L G and S 128 F, T, P, B, IH NT 9 37 Fe L T FA (STG, MTG, IC) None L G and S 96 F, T, O NT M L T Min MT Asy normal R G and S 92 F, T, P, O NT, Neocortical temporal; Fe, female; M, male; Loc, localization; R, right; L, left; Mig Abn, migrational abnormality; GM, gray matter; HT, heterotopia; WM, white matter; IS, increased signal; COR, cortical, SC, subcortical; Tu, tumor; FA, focal atrophy; STG, superior temporal gyrus; MTG, middle temporal gyrus; IC, Insular cortex; Min, minimal; MT, mesial temporal; Asy, asymmetry; Path, pathology; CD, cortical dysplasia; MD, microdysgenesis; NLO, no laminar organization; PXA, pleomorphic xanthoastrocytoma; HyC, hypercellularity; Ca VM, calcified vascular malformation; No Sz, number of seizures; GD, median global duration; FD, median focal duration; No Cont, number of contacts; Bi, bilateral; G, grid electrodes; S, strip electrodes; D, depth electrodes; F, frontal; P, parietal; T, temproal; O, occipital; B, basal; IH, interhemispheric; AG, amygdala; HC, hippocampus. Patient was not a surgical candidate.

4 680 P. Afra et al. Table 4. Neocortical extratemporal lobe epilepsy summary Patient Age/sex Loc MRI Path No Sz GD(s) FD(s) Electrodes No Cont NE 1 23 Fe R Fe CD CD R G 128 F, T, P, O, IH NE 2 29 Fe R Fe T2 IS: SCWM CD R G 67 F, P NE 3 16 M R Fe Mig Abn MD Bi (R G/ L S) 126 R F, T, P & L F NE4 19M RO BiO(R>L) IS LS R G 126 T, P, O NE 5 19 Fe R PO-IH Ca WD Tu R G 96 P, O, IH NE 6 50 M R Fe Tu Glioma R G, S and D 89 F, T, P NE 7 22 M R P-IH Tu Muc Tu R D 24 AG, HC, CG NE 8 5 Fe R Fe Tu Glioma R G 59 F, P NE 9 33 M R P Enceph NP R G 64 F, T, P NE Fe L Fe Enceph NP L G and S 110 F, T, P, O NE M R PO Enceph Enceph R G 58 F, T, P, O, B NE M L Fe HC Asy NL L G 100 F, T, IH NE M L F-IH Normal NL Bi (L G, and R S) 128 L F, T, P, IH & R F NE Fe L Fe Normal NL L G and S 66 F, C, P, IH NE Fe L Fe Normal NL Bi (L G and R S) 115 L F, T, P & R F, T, P NE 16 6 Fe R MO Normal NL R G 91 T, P, O NE 17 4 Fe R Fe Normal NL R G and S 68 F, IH NE, Neocortical extratemporal; Fe, female; M, male; Loc, localization; R, right; L, left; F, frontal; P, parietal; T, temporal; O, occipital; MO, mesial occipital; B, basal; IH, interhemispheric; CD, cortical dysplasia; IS, increased signal; SCWM, subcortical white matter; Mig Abn, migrational abnormality; Ca, calcification; Tu, tumor; Enceph, encephalomalacia; HC, Asy hippocampal asymmetry; Path, pathology; MD, microdysgenesis; LS, laminar scarring; WD, well differentiated; Muc, mucinous; NP, no pathology; NL, nonlesional; No Sz, number of seizures; GD median, global duration; FD median, focal duration; Bi, bilateral; G, grid electrodes; S, strip electrodes; D, depth electrodes; No Cont, number of contacts; AG, amygdala; HC, hippocampus; CG, cingulate. posterior cingulate. Depth electrode passing through grid. histopathological diagnosis (to establish that they were nonlesional). Statistical analysis Median is a robust and nonparametric estimate of the center of a distribution (Small 1990). It is more resistant to outliers and for nonsymmetrical distribution is a better representation of the typical value of the observations than the mean. In studies of epileptic seizures, the occurrence of long seizures renders the mean an inadequate estimator to properly describe the typical seizure duration. Therefore GD and FD were computed by using the median as a nonparametric measure of the duration of analyzed seizures for each patient. The presence of an overall group effect (MTLE, NC- TLE, NCXTLE) was determined with the one-way nonparametric Kruskal Wallis ANOVA. Following the determination of the group effect, the comparison of groups was done with the Mann Whitney U test. Both methods use average rank of the pooled dataset as a measure. Statistics for the repeated Mann Whitney U tests were corrected by the Bonferroni method. Repeated two-group comparisons were performed with a threshold p of RESULTS Tables 2, 3, and 4 list the patient summaries for the three groups. The tables list patient sex and age at the time of monitoring, MRI abnormalities, identified histopathology, number of seizures for each patient, focal and global seizure durations, and other characteristics. The MTLE group includes seven patients with recordings with depth electrodes and 19 patients with recordings with various combinations of subdural grid and strip electrodes (see Table 2). To assess the potential influence of the type of recording arrays on observed seizure durations, a Mann Whitney U test for two independent samples was done. There was no significant difference in GD (p = 0.188) or FD (p = 0.651) of the seizures observed in the seven patients with depth electrodes compared to the 19 patients with unilateral or bilateral subdural electrodes. When bilateral recordings from subdural and depth electrodes (n = 12) were compared to unilateral recordings from subdural electrodes (n = 14), there was a significant differences in FD (p = 0.008) and GD (p < 0.001). In recordings from unilateral arrays both the FD and GD had median durations of 78 s. Bilateral recordings had median FD of 108 s and median GD of 138 s. The number of seizures was not different for unilateral versus bilateral groups (p = 0.85). There were limited numbers of bilateral recordings for NCTLE (n = 2) and NCXTLE (n = 3). These numbers preclude any robust statistical analysis. Tables 2, 3, and 4 list the median focal and global seizure durations for each patient in each of the three groups.

5 681 Duration of Complex Partial Seizures Figure 1. Distribution of global seizure durations in the 26 MTLE patients. The crosses represent individual seizure duration and squares represent the median of the seizure durations for each patient. Patient 4 had two seizures that were outliers. The durations of these two seizures are in parenthesis. MTS, mesial temporal sclerosis. Epilepsia C ILAE Figs. 1, 2, and 3 are scatterplots of global partial seizure durations for each analyzed seizure of the respective patients as well as the median seizure duration for each patient (squares). A number of patients, but certainly not all, had seizures of similar duration. Neocortical onset seizures, particularly those of extratemporal origin, tended to more frequently exhibit this pattern of similar seizure durations than did seizures of mesial temporal origin. Effect of origin on seizure duration We compared the three groups (MTLE, NCTLE, and NCXTLE) using the nonparametric Kruskal Wallis ANOVA test. This test uses the rank of the seizure durations for each patient when all data are pooled (Fisher & van Belle, 1993). For global seizure duration there was a significant effect of seizure origin (p = 0.019) as well as for focal seizure duration (p = 0.043). Because of these demonstrated differences, we could now appropriately perform the Mann Whitney U tests for nonparametric group comparisons. There was no significant difference between MTLE and NCTLE groups for GD (p = 0.155). Ranges of both groups overlapped with GD ranging from 66 s to 319 s for MTLE and from 33 s to 551 s for NCTLE. Similarly FD ranged from 64 s to 203 s for MTLE and from 33 s to 551 s for NCTLE, the groups were not significantly different (p = 0.32). Comparison of MTLE versus NCXTLE, however, revealed significant changes in GD (p = 0.007) with GD ranging from 66 s to 319 s for MTLE versus 16 s to 341 s for NCXTLE. Analyzing seizure duration using FD also yields a significant difference between MTLE and NCXTLE (p = 0.013). FD ranges from 64 s to 203 s for MTLE and from 16 s to 341 s for NCXTLE. Seizure duration for NCTLE and NCXTLE did not differ significantly whether GD (p = 0.334) or FD (p = 0.334) were analyzed. Fig. 4 shows a summary presentation of these datasets using box-and-whisker diagram (Tukey, 1977). The figure shows that median seizure durations are consistent whether FD or GD is considered. Median seizure durations are less susceptible to outliers than mean durations. In the MTLE group there is a slight median duration difference between GD and FD: 106 s for GD and 102 s for FD. The GD and

6 682 P. Afra et al. Figure 2. Distribution of global seizure durations in the 10 NC- TLE patients. The crosses represent individual seizure duration and the squares represent the median of the seizure durations for each patient. Patient 9 had one seizure that was an outlier. The duration of this seizure is in parenthesis. L, lesional; NL, nonlesional. Epilepsia C ILAE Figure 3. Distribution of global seizure durations in the 17 NCX- TLE patients. The crosses represent individual seizure durations and squares represent the median of the seizure durations for each patient. L, lesional; NL, nonlesional. Epilepsia C ILAE FD were 82 s for NCTLE group and 78 s for the NCXTLE group. As mentioned before, there is a significant difference in the median seizure duration between the MTLE and the NCXTLE groups, (i.e., MTLE exhibiting a higher mean rank of seizure duration when compared to NCX- TLE). DISCUSSION These results indicate that mesial temporal lobe CPS are of significantly longer duration than are CPS from neocortical extratemporal regions. There was no statistically significant difference in the duration of CPS between the two groups of mesial temporal and neocortical temporal onset seizures. The NCTLE group is the smallest group; larger numbers of patients in this group could produce a significant difference (versus MTLE) not seen here. Frontal lobe CPS are generally considered to be of shorter duration than mesial temporal onset CPS; how- ever, no quantitative comparative studies have been published previously. The few published studies on CPS do not compare regions of seizure onset. The results here reflect a select group of patients with highly refractory unifocal epilepsy undergoing presurgical evaluations with intracranial electrodes. Whether intractability affects seizure duration is not known. Early studies of video-eeg recordings of CPS (Delgado-Escueta et al., 1982) did not differentiate between ictal and postictal durations. In the study of Theodore et al. (1983) the average duration (determined by clinical assessment) of CPS from all patients (n = 40) was 128 s; in the 37 patients where the ictal period could be differentiated from the postictal period the average duration was 54 s. All of the seizures in the Theodore et al. study came from the temporal lobe; no comment was made regarding mesial temporal or neocortical onset in this pre- MRI era study. The average seizure duration was determined by averaging all seizures from all patients. Three patients had over 12 seizures; 21 patients had only one or

7 683 Duration of Complex Partial Seizures Figure 4. Box-and-whisker diagram of the global seizure duration and the focal seizure duration for each group: MTLE, NCTLE and NCXTLE. The bar in each box represents the median duration for each group, and the box extends from the first quartile (25% of the cases are inferior to it) to the third quartile (75% of the cases are inferior to it). The box height is the interquartile range (IQR). Closed circles represent the outliers (more than 1.5 IQR higher then the 3rd quartile) and stars represent the extreme outliers (more than 3 IQR above the 3rd quartile). Open boxes represent the global seizure duration. Filled boxes represent the focal seizure duration. This figure is presented in the standard format for box-and-whisker diagrams. The thin bars are not error bars as one would see in parametric analyses but instead represent the range (minus the outliers). The median for each group represents the analysis of each selected group, and the statistics are derived from ranks of the pooled data from the median seizure durations for each patient. This should not be used for comparison with data not included in this study without first incorporating any new data into the ranked pooled dataset. Epilepsia C ILAE two seizures. These analyses were therefore weighted to patients with more frequent seizures. Jenssen et al. (2006) compared the electroclinical duration of seizures (including complex partial) in patients undergoing scalp video-eeg recordings. They calculated the median seizure duration per patient. For CPS this was 78 s (range: s). There was no difference in seizure duration and side of cerebral onset; no differentiation between locations of seizure onset was made. Another study looked at duration of first unprovoked seizure in children (Shinnar et al., 2001). This study used subjective reports of witnesses, often parents. Duration of seizures in this study was considerably longer than our study or other studies. The results presented here are a systematic study that examines intracranial electrographic recordings to assess the duration of CPS and compares seizures originating from neocortical and mesial temporal lobe foci. While in many patients seizure termination occurs nearly simultaneously in all contacts, in others there may be asynchronous termination (Spencer & Spencer, 1996). Preliminary data from our laboratory suggests that partial seizures from a given patient that terminate synchronously are more likely to be shorter than seizures with asynchronous termination, if indeed both patterns are present in the same patient (Afra et al., 2006). Intracranial recording arrays offer the advantages of improved detection of seizure onset, seizure termination, and better determination of the onset of the postictal period. The placement of intracranial arrays is however individualized for each patient with some patients requiring bilateral recordings (depths or strips), while others have lateralized seizures and only require unilateral arrays. The possibility therefore that sampling error may influence the assessment of seizure duration exists and cannot be eliminated. Many neocortical onset seizures that spread bilaterally produce secondarily GTCS. These seizures were excluded from these analyses. While mesial onset seizures more commonly initially spread to the ipsilateral temporal and frontal lobe regions (Lieb et al., 1991) and may secondarily generalize, some mesial temporal lobe seizures (including some in our series) may initially spread to the contralateral mesial temporal structures without secondarily GTCS. In patients studied here the MTLE group CPS were significantly longer in duration than seizures originating from temporal or extratemporal neocortical regions, however, this difference was only apparent if bilateral recordings were included in the analyses. Therefore spread to the contralateral mesial temporal structures could be a factor in this observed difference. Contralateral spread was seen in 7 of the 12 patients with MTLE and bilateral recording arrays. The number of neocortical onset (temporal n = 2; extratemporal n = 3) with bilateral recording arrays (subdural strips) was not sufficient to permit any conclusions. Extratemporal neocortical onset CPS, particularly those originating from the frontal lobe regions are thought to be of briefer duration than CPS originating from mesial temporal regions; although no quantitative comparative studies exist. This clinical impression is probably reinforced by the fact that frontal lobe partial seizures often

8 684 P. Afra et al. have comparatively brief postictal periods or may be brief simple partial seizures without alteration of consciousness. It is interesting that the extratemporal neocortical onset CPS studied here had median global seizure durations of 78 s, perhaps longer than might be expected for this type and origin of seizures. The clinical duration of epileptic seizures is an important feature for the patient and this is why the GD measure was made. It is interesting, however, that the focal and global seizure durations were remarkably similar for the neocortical onset seizures (NCTLE and NCXTLE) when sampled unilaterally. The GDs and FDs of mesial temporal onset seizures sampled with bilateral arrays were longer than MTLE with unilateral sampling. While the increased GD may well reflect the bilateral sampling, it is less obvious why the focal seizure durations should be longer in these patients. Seizure duration may be a reflection of the inherent dynamic properties of the underlying abnormal network. Seizures originating from a single focus in a given patient have stereotyped patterns of seizure onset and predominant frequencies as determined by time-frequency analyses (Jouny et al., 2003). It has been suggested that the nonlinearity of the human brain contributes to the finite length of seizures and indeed in mesial temporal onset seizures there is a progressive decline in the frequency of the predominant monotonic organized rhythmic activity (Franaszczuk et al., 1998). Partial seizures also typically demonstrate increasing signal complexity prior to seizure termination (Bergey & Franaszczuk, 2001; Jouny et al., 2003). This may reflect decreased ability to sustain synchronized activity. Secondarily GTCS were not studied here, in part because intracranial monitoring is often limited to a single hemisphere. While it is not surprising to find that secondarily GTCS are longer in duration than are partial seizures in the same patient (Jenssen et al., 2006), this increased duration may not merely represent more extensive propagation. Recent time-frequency analysis (Jouny et al., 2007) indicate that some secondarily generalized seizures influence the seizure focus more than others; this was more common in neocortical seizures. Jenssen et al. observed that in patients with more than one complex partial seizure, the last seizure tended to last longer than the first. We did not observe this in the patients studied here. In a subset of these patients analyzed by Adamolekun et al. (2004) AED withdrawal did not affect the duration of partial seizures. Interestingly more than half the patients in each group had relatively similar duration for all of their seizures. The NCTLE group had the highest number (14 of 19; 74%) of patients with seizures of similar duration (low variability). This is further evidence supporting intrinsic seizure dynamics for a given patient with unifocal epilepsy. The shorter seizure duration of neocortical onset CPS may represent either different intrinsic dynamics at the seizure focus or different patterns (or speed) of seizure propagation. Mesial temporal lobe onset seizures tend to remain more localized to a limited area early in seizure evolution whereas neocortical onset seizures rapidly propagate regionally. This quantitative study of seizure duration of CPS therefore supports the commonly held, but previously unsubstantiated belief that extratemporal neocortical onset seizures are of shorter duration than mesial temporal onset CPS. ACKNOWLEDGMENTS We would like to thank Dr. Piotr J. Franaszczuk for his helpful comments and review of this manuscript. This research was supported in part by NIH grant NS R 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. The authors have no conflicts of interest to report. REFERENCES Adamolekun B, Jouny CC, Franaszczuk PJ, Bergey GK. (2004) Does antiepileptic drug withdrawal affect partial seizure duration. Epilepsia 45(Suppl 7):58. Afra P, Jouny CC, Franaszczuk PJ, Bergey GK. (2006) Partial seizure duration may influence patterns of seizure termination. Epilepsia 47(Suppl 4):48. Bergey GK, Franaszczuk PJ. (2001) Epileptic seizures are characterized by changing signal complexity. Clin Neurophysiol 112: Delgado-Escueta AV, Bacsal FE, Treiman DM. (1982) Complex partial seizures on closed-circuit television and EEG: a study of 691 attacks in 79 patients. Ann Neurol 11: Fisher DL, vanbelle G. (1993) Biostatistics: a methodology for the health sciences, Wiley Science Paperback Series, New York, NY. Franaszczuk PJ, Bergey GK, Durka PJ, Eisenberg HM. (1998) Timefrequency analysis using the matching pursuit algorithm applied to seizures originating from the mesial temporal lobe. Electroencephalogr Clin Neurophysiol 106: Jenssen S, Gracely EJ, Sperling MR. (2006) How long do most seizures last? A systematic comparison of seizures recorded in the epilepsy monitoring unit. Epilepsia 47: Jouny CC, Franaszczuk PJ, Bergey GK. (2003) Characterization of epileptic seizure dynamics using Gabor atom density. Clin Neurophysiol 114: Jouny CC, Adamolekun B, Franaszczuk PJ, Bergey GK. (2007) Intrinsic ictal dynamics at the seizure focus: effects of secondary generalization revealed by complexity measures. Epilepsia 48: Lieb JP, Dasheiff RM, Engel J. (1991) Role of frontal lobes in the propagation of mesial temporal lobe seizures. Epilepsia 32: Shinnar S, Berg AT, Moshe SL, Shinnar R. (2001) How long do new-onset seizures in children last? Ann Neurol 49: Small CG. (1990) A Survey of Multidimensional Medians. Int Stat Rev/Rev Int Stat 58: Spencer SS, Spencer DD. (1996) Implications of seizure termination location in temporal lobe epilepsy. Epilepsia 37: Theodore WH, Porter RJ, Penry JK. (1983) Complex partial seizures: clinical characteristics and differential diagnosis. Neurology 33: Theodore WH, Porter RJ, Albert P, Kelly K, Bromfield E, Devinsky O, Sato S. (1994) The secondary generalized tonic-clonic seizure: a videotape analysis. Neurology 44: Tukey JW. (1977) Exploratory data analysis. Addison-Wesley, Reading, MA.