Epilepsy Research (2014) 108, 1790—1796
journal homepage: www.elsevier.com/locate/epilepsyres
Subclinical seizures during intracranial EEG recording: Are they clinically significant? Pue Farooque ∗, Robert Duckrow Yale University Comprehensive Epilepsy Center, Department of Neurology, New Haven, CT, United States Received 6 February 2014; received in revised form 9 September 2014; accepted 20 September 2014 Available online 28 September 2014
KEYWORDS Seizures; Subclinical; Epilepsy surgery
Summary Purpose: To evaluate the clinical significance, characteristics and prognostic value of subclinical seizures of temporal and extra-temporal origin. Methods: Our epilepsy database from 2003 to 2011 was reviewed to identify patients with subclinical seizures during intracranial EEG recording who underwent surgical resection. Two groups were formed: Group 1 where both the clinical and subclinical seizures had the same seizure onset region, and Group 2 where some or all of the clinical and subclinical seizures originated from different regions. Results: A total of 27 patients were identified with 791 seizures, of which 310 were subclinical. In Group 1 (n = 14), 64% had good surgical outcome, and 36% had poor surgical outcome. In Group 2 (n = 12), 83% had poor outcome while 17% had good outcome. One patient had only subclinical seizures. Eleven patients had subclinical seizures that propagated to a region beyond their onset zone. Of those 11 patients, 6 patients had subclinical seizures propagate to a different anatomical region than their clinical seizures. These six patients had poor surgical outcome. Discussion: Our study, like others, found that subclinical seizures are clinically significant and including their onset region in the volume of surgical resection correlates with good surgical outcome for both temporal and extra-temporal lobe epilepsy (Fisher’s exact test p = 0.017). Propagation of subclinical seizures to a different region than clinical seizures can affect surgical outcome (Fisher’s exact test p = 0.06). Subclinical seizures may represent a distinct epileptic network. © 2014 Elsevier B.V. All rights reserved.
Introduction ∗ Corresponding author at: Yale University School of Medicine, Department of Neurology, P.O. Box 208018, New Haven, CT 06520, United States. Tel.: +1 203 785 3865; fax: +1 203 737 2799. E-mail address: [email protected] (P. Farooque).
http://dx.doi.org/10.1016/j.eplepsyres.2014.09.020 0920-1211/© 2014 Elsevier B.V. All rights reserved.
Localizing the onset zone of habitual clinical seizures with intracranial electroencephalographic (icEEG) monitoring is a frequent component of the evaluation of patients with medically intractable epilepsy being considered for
Subclinical seizures therapeutic resective surgery. During intracranial monitoring other epileptiform patterns such as sharp transients, high frequency bursts, or subclinical electrographic seizures are often recorded. Considerable effort has been directed to understand their relationship to the seizure onset zone, so that these patterns can be used as supportive evidence during the process of onset localization. Study of interictal spike wave patterns (Walczak et al., 1990) and high frequency oscillations at ictal onset (Fujiwara et al., 2012; Guggisberg et al., 2008) suggests the localizing value of these patterns as a predictor of surgical success. However, there are few studies evaluating the clinical significance and localizing value of subclinical seizures when assessing patients for epilepsy surgery (Sperling and O’Connor, 1990; Velkey et al., 2011; Zangaladze et al., 2008). Subclinical seizures are defined as electrographic seizures with rhythmic ictal discharges that evolve in frequency and space that lack any objective or subjective alteration in behavior or consciousness (Sperling and O’Connor, 1990). They are observed during long-term monitoring with either scalp or intracranial recordings; however their exact prevalence is unknown. Velkey et al. found that 18% of children had subclinical seizures during long-term scalp EEG monitoring and the presence of subclinical seizures influenced the diagnostic process in 85% of cases. Sperling and O’Connor evaluated the clinical characteristics and prognostic value of subclinical seizures of predominantly temporal lobe origin during icEEG monitoring. They found the presence of subclinical seizures to be a favorable prognostic sign for patients undergoing temporal lobectomy (Sperling and O’Connor, 1990) and that subclinical seizures had the same localizing value as clinical seizures (Zangaladze et al., 2008). However, they were unable to draw the same conclusion for subclinical seizures of extra-temporal onset due to lack of sample size. The purpose of this study was to compare the electrographic characteristics of subclinical and clinical seizures of both temporal and extra-temporal origin and to extend our understanding of the localizing and prognostic value of subclinical seizures to patients with extra-temporal lobe epilepsy.
Methods We reviewed the Yale Epilepsy Surgery Program database from 2003 to 2011 and identified 216 patients, who underwent intracranial EEG recording and subsequent resective surgery. Of these patients, 27 had subclinical seizures. At our institution, all patients regardless of epilepsy classification undergo similar intracranial studies consisting of a combination of depth electrodes, subdural strips and a 8 × 8 grid. The maintenance of the database and the methods of this study were approved by the Yale Human Research Protection Program. The clinical behavior and electrographic accompaniment of all seizures were visually analyzed. Subclinical seizures were defined as rhythmic ictal discharges that evolved in frequency and space that lacked any objective or subjective alteration in behavior or consciousness (Sperling and O’Connor, 1990). Seizures were excluded if they caused an arousal during sleep.
1791 Clinical and subclinical seizures were characterized by their onset region, the electrographic features at onset, and the pattern of propagation. Clinical seizures were further classified as typical or atypical based on their habitual semiology. Subclinical seizures were further classified as occurring during the awake or sleep state. The seizure onset region was defined by the electrode contact or contacts where the initial ictal rhythm was seen. Seizure onsets that involved the grid appeared within the grid and did not include contacts at the edge of the grid. Lobar classification of the seizure onset region was temporal, either medial or lateral, or extra-temporal, either frontal or parietal/occipital. Seizure propagation was defined as a definite ictal pattern at electrode contacts occurring not simultaneously at least 2 cm from the seizure onset region (Götz-Trabert et al., 2008). The presence or absence of an MRI lesion including hippocampal sclerosis was noted. The location and extent of the surgical resection was based on all available information. Treatment outcome was staged at least 2 years post surgery using the Engel classification with good outcome denoted by Engel Class I and poor outcome by Engel Class II—IV (Engel et al., 1993). Based on previous studies (Zangaladze et al., 2008), patients were also grouped based on the concordance of clinical and subclinical seizure onset regions. All statistical analysis was done with Fischer Exact test.
Results A total of 27 patients had subclinical seizures during their intracranial EEG recording and underwent focal resective epilepsy surgery of the presumed seizure onset zone. Three of these patients had additional multiple subpial transections (MSTs) of overlapping functional regions. There were 14 patients with temporal lobe epilepsy (12 mesial and 2 lateral) and 13 with extra-temporal epilepsy (5 frontal, 8 parietal/occipital). Of the 791 seizures, 310 were subclinical with 210 of the subclinical seizures being extra-temporal in origin. Fig. 1 illustrates the number of subclinical and clinical seizures arising from each lobar region. Post-operative follow-up ranged from 2 to 9 years with a median of 5.6 years. The surgical outcome groups did not change from two years post surgery to the last follow up as depicted in Table 3. Seventeen patients had lesions on MRI including mesial temporal sclerosis, cavernoma, cortical dysplasia/dysgenesis and nodular heterotopia. Of patients with lesional epilepsy, eight (47%) had good surgical outcome. Of all 27 patients, twelve (44%) had good surgical outcome; of which six had extra-temporal lobe epilepsy. Fifteen patients (56%) had poor surgical outcome with seven of these having extra-temporal lobe epilepsy. The proportion of subclinical seizures to habitual clinical seizures did not correlate to outcome (Data not shown). Table 1 depicts patient demographic and clinical information.
Localization of seizure onset for subclinical seizures In 14 of 27 patients clinical and subclinical seizures arose from the same seizure onset region (Group 1). Of these patients, nine had extra-temporal lobe epilepsy (4 frontal, 5
1792 Table 1 Patient number
P. Farooque, R. Duckrow Patient demographic information. Age
Gender MRI
Surgical resection
Engel classification
Epilepsy classification
Post-op follow-up
1
22
F
RAMTL
4
Temp
5
2
33
F
R medial frontal
1
ET-frontal
3
3 4
38 35
F F
3 1
Temp ET-parietal/occipital
7 4
5
31
F
3
ET-TPO
6
6 7 8 9
16 26 31 59
M F M F
4 3 1 3
Temp Temp ET-frontal ET-occipital
8 9 8 5
10 11
35 45
F M
LAMTL R posterior temporoparietal occipital L mid-temporal to occipital region LAMTL RAMTL R frontal R occipital posterior temporal RAMTL LAMTL
4 1
Temp Temp
8 7
12
29
F
1
ET-parietal
5
13
44
F
1
ET-frontal
7
14
51
M
R parietal occipital junction R frontal anteroinferior parietal RAMTL
1
Temp
6
15
31
M
3
Lateral temp
8
16 17
45 33
F M
R MTS R Nodular heterotopia of lateral ventricle
1 3
Temp ET-parietal
4 3
18
14
F
Gliosis L parietal-occipital lobe
4
ET-TPO
4
19
13
M
3
Lateral temp
7
20 21 22
35 53 47
F F F
Cavernoma L lateral temp R MTS Normal R MTS
RAMTL with 7 cm lateral resection RAMTL MSTs of parietotemporal region, Resection of dysplasia L temporoparietooccipital and occipital lobectomy L lateral temporal
3 1 1
Temp Temp ET-parietal
7 4 8
23
23
M
Normal
4
ET-frontal
8
24 25
9 31
M F
R MTS Normal
1 3
Temp ET-frontal
2 2
26
26
F
Normal
4
ET-occipital
2
27
58
M
L hippocampal atrophy
1
Temp
6
Heterotopia R temp Cortical dysplasia R superior frontal gyrus L MTS R TPO encephalomalacia Cortical dysplasia L occipital lobe L MTS Normal Normal R hippocampal atrophy Normal L temp encephalomalacia S/P R temp lobectomy Normal
Bilat HA, worse on R Normal
RAMTL RAMTL R posterior temporal with MSTs of parietal region Right frontalpolar and orbital RAMTL R frontal lobectomy with MSTs L parietal occipital lobectomy L amygdalohippocampectomy
M, Male; F, Female; R, Right; L, Left; Temp, temporal; ET, Extra temporal; TPO, temporal/parietal/occipital; AMTL, anterior medial temporal lobectomy; MST, Mesial Temporal Sclerosis.
Subclinical seizures
1793
27 paents 791 seizures
Temporal 14 paents 167 seizures
YES
*Co-localized NO
5 paents 70 seizures
42 clinical seizures
Extra-temporal 13 paents 624 seizures
28 subclinical seizures
YES
8 paents 94 seizures
25 clinical seizures
69 subclinical seizures
Co-localized NO
9 paents 547 seizures
358 clinical seizures
189 subclinical seizures
4 paents 77 seizures
56 clinical seizures
21 subclinical seizures
Figure 1 Subgroup of clinical seizures, subclinical seizures, and co-localization. * One patient in the temporal lobe epilepsy group only had three subclinical seizures and no clinical seizures, co-localization was not determined for this patient.
parietal/occipital) and four had lesions. In 12 of 27 patients clinical and subclinical seizures arose from different regions (Group 2). Within this discordant group, four patients had at least one subclinical seizure arise from a different location than their clinical seizures (Group 2a) and eight patients had all their subclinical seizures arise from a different location (Group 2b). For Group 2a (some discordant), there were a total of 105 seizures in four patients where 43 were clinical and 62 were subclinical. Four of the clinical seizures were atypical and only 15 subclinical seizures had the same seizure onset region as their respective clinical seizures (14%). The majority of subclinical seizures within this subgroup did not co-localize to the same region as their respective clinical seizures. Fig. 1 depicts the total number of clinical and subclinical seizures for each onset region where the seizures either co-localized or did not co-localize. Of note, one patient had only subclinical seizures during her intracranial monitoring as her study abruptly ended due to surgical complication prior to clinical seizures being recorded. Her data was excluded from analysis.
Seizure spread Subclinical seizures remained within their seizure onset region in 15 patients (224 subclinical seizures) and propagated beyond their onset region in 11 patients (86 subclinical seizures). Of those 11 patients, six patients (3 temporal, 3 extra-temporal) had subclinical seizures (17 subclinical seizures) that propagated to a different region than their respective clinical seizures. Of the six patients that had subclinical seizures propagate to a different region than their clinical seizures, two had their clinical and subclinical seizures arise from the same area at onset. Of note, both of these patients had the same electrographic
pattern at seizure onset and both patients had abnormalities seen on MRI. Of the five patients who had their subclinical patients spread to the same region as their respective clinical seizures, two of them had their clinical and subclinical seizures arise from the same region.
Surgical outcome Of the 14 patients with co-localized clinical and subclinical seizures (Group 1), 9 (64%) had good surgical outcome, and 5 (36%) had poor surgical outcome. Of the five patients with poor surgical outcome, two underwent focal resection and multiple subpial transections through a functional area that overlapped the seizure onset region, one patient underwent temporal lobectomy although the seizure semiology suggested insular onset, and two patients had subclinical seizures that spread to a different region than their clinical seizures. Of the 12 patients with partially co-localized or not co-localized subclinical and clinical seizures (Group 2) 10 (83%) had poor outcome and 2 (17%) had good outcome. The two patients with good outcome had mesial temporal lobe sclerosis. The side of the sclerosis was resected and either all or some of the subclinical seizures came from the contralateral side. Fig. 2 shows surgical outcome based on the concordance of subclinical and clinical seizure onset, either co-localized or not co-localized. Good surgical outcome assessed at 2 years post surgery is more likely when the subclinical seizure onset region is included in the volume of surgical resection (Fisher’s exact test p = 0.017). Good surgical outcome assessed at last follow up (range 2—9 years) was also more likely when the subclinical seizure onset region is included in the volume of surgical resection (Fisher’s exact test p = 0.017).
1794
P. Farooque, R. Duckrow 16
Surgical Outcome
Number of Paents
14 12 10 Not Co-localized
8
Co-localized
6 4 2 0
Good Outcome
Poor Outcome
Figure 2 Relationship between subclinical and clinical seizures arising from either the same or different region of onset and its affect on surgical outcome. Treatment outcome was staged at least 2 years post surgery using the Engel classification with good outcome denoted by Engel Class I and poor outcome by Engel Class II—IV. Good surgical outcome is more likely when the subclinical seizure onset region is included in the volume of surgical resection (Fisher’s exact test p = 0.017). This was true for 2 years post surgery as well as at last follow-up.
Eleven patients had subclinical seizures spread beyond their onset region. Of these eleven patients, six had subclinical seizures that spread to a different region than their respective clinical seizures and all six had poor surgical outcome. Three of these six had temporal onset and three had extra-temporal onset. Of the six patients with disparate spread patterns and poor outcome, two patients had lesions with clinical and subclinical seizures arise from the same onset region, elements usually associated with good outcome. Among the five patients that had subclinical seizures spread to the same region as their clinical seizures, two had poor surgical outcome. Their subclinical and clinical seizures had different onset regions. Of the three patients whose seizures spread to the same region and had good surgical outcome only one had their subclinical seizures partially co-localize with their clinical seizures at onset (Group 2a). This patient had mesial temporal sclerosis with five clinical seizures and sixteen subclinical seizures. One clinical seizure was atypical in character and had onset in the temporal lobe opposite to the side of the sclerosis. Eleven subclinical seizures had the same seizure onset zone as the typical clinical seizures on the side of the sclerosis. The other five subclinical seizures arose from the contralateral temporal lobe, again opposite to the side of the sclerosis. Table 2 correlates the pattern of subclinical and subclinical seizure spread with seizure onset to surgical outcome. The propagation of subclinical seizures to a different region than clinical seizures correlates with poor surgical outcome (Fisher’s exact test p = 0.06).
Discussion We studied patients with subclinical seizures during intracranial EEG recording who subsequently underwent epilepsy surgery to investigate the localizing value and prognostic significance of subclinical seizures in relation to surgical outcome. We specifically wanted to evaluate if conclusions based on subclinical seizures of extra-temporal onset would be the same as those made for temporal lobe epilepsy. Our cohort of patients was equally split among temporal and extra-temporal epilepsy patients, which aided the comparison. We found that our extra-temporal lobe patients had twice the number of subclinical seizures as our temporal lobe patients while previous studies found the majority of subclinical seizures were temporal in origin (Zangaladze et al., 2008). This difference may be secondary to increased spatial sampling during the intracranial implant, where our average number of electrodes implanted for an extra-temporal case was 205. Another contrast is that more than half of our subclinical seizures of temporal onset did not arise from the same region as the coresponding clinical seizures while previous studies found the majority of temporal lobe subclinical seizures co-localized with clinical seizures (Zangaladze et al., 2008). It is possible that our temporal lobe epilepsy patients had more diffuse disease that was refractory to treatment, as evidenced by 57% having poor surgical outcome. We found that when subclinical seizures and clinical seizures arise from the same region and that region is included in surgical resection, 64% of patients had good
Table 2 This table shows subclinical seizures (SCS) either co-localizing or not co-localizing at onset with their clinical seizures (CS) in relation to the spread of SCS to either the same or different region than its respective CS and the effect on surgical outcome. Spread pattern
Good outcome
Poor outcome
Patients with SAME SCS as CS onset and SAME SCS as CS spread Patients with SAME SCS as CS onset but DIFFERENT spread All patients with spread that DIFFERED between CS and SCS, regardless of onset.
2 0 0
0 2 6
Pt #
Number of clinical seizures
Number of subclinical seizures
Spread
Spread different than clinical
MRI lesion
Engel classification at Last follow-up
Co-localization of subclinical and clinical seizures
Seizure classification
N/A N/A Yes, different-parietal N/A Yes, differentorbitofrontal N/A N/A No, same Yes, different-parietal N/A N/A N/A N/A No, same Yes, different-frontal N/A N/A No, same No, same Yes, different-parietal N/A No, same Yes, different-lateral temporal N/A N/A N/A N/A
Yes Yes Yes
4 1 3
2b 1 1
T ET-F T
Yes Yes
1 3
1 1
ET-P/O ET-TPO
Yes No No Yes
4 3 1 3
2b 2a 1 2b
T T ET-F ET-O
No Yes No No Yes No
4 1 1 1 1 3
1 2b 1 1 2a 2a
T T ET-P ET-F T LT
Yes Yes Yes Yes Yes
1 3 4 3 3
N/A 1 2a 2b 2b
T ET-P ET-P/O LT T
No Yes No
1 1 4
1 1 2b
T ET-P ET-F
Yes No No Yes
1 3 4 1
1 1 2b 1
T ET-F ET-O T
1 2 3
4 11 5
3 90 2
No No Yes
4 5
3 107
11 3
No Yes
6 7 8 9
8 8 12 9
1 41 10 5
No No Yes Yes
10 11 12 13 14 15
11 2 141 12 5 2
7 2 44 2 16 2
No No No No Yes Yes
16 17 18 19 20
0 4 28 1 4
3 1 3 16 1
No No Yes Yes Yes
21 22 23
6 5 37
1 24 4
No Yes Yes
24 25 26 27
8 35 10 3
2 2 11 3
No No No No
1795
T, temporal; LT, lateral temporal; P, parietal; O, occipital; F, frontal; ET, extra temporal; OF, orbitofrontal; 1, subclinical and clinical seizures onset match; 2a, partial match of onset between subclinical and clinical seizures; 2b, no match between onset of clinical and subclinical seizures; N/A, not applicable.
Subclinical seizures
Table 3 Number of clinical seizures, subclinical seizures, presence of spread for subclinical seizures, region of spread for subclinical seizures, Presence of MRI abnormality, surgical outcome at 2 years post surgery and last follow-up and degree of co-localization for all epilepsy subgroups.
1796 surgical outcome. This confirms the earlier observations of Sperling and O’Connor (1990) and Zangaladze et al. (2008) and extends them to extra-temporal epilepsy. Conversely, if subclinical seizures arise from a different region than clinical seizures and that region was not included in the area of resection, 83% of patients had poor surgical outcome. Two patients had good surgical outcome despite their subclinical and clinical seizures coming from different regions. Both of these patients had unilateral mesial temporal sclerosis, where some or all of their subclinical seizures arose from the temporal lobe contralateral to the scelerosis and region of clinical seizure onset. These subclinical seizures either did not spread beyond their seizure onset zone or spread to the same region as their respective clinical seizures. When bilateral independent seizure onset is seen in patients with unilateral medial temporal sclerosis, resecting the side with the sclerosis provides a high probability of becoming seizure free, as reported in previous studies (Hirsch et al., 1991a, 1991b; King et al., 1997a, 1997b; Mintzer et al., 2004; Spencer, 2002). This suggests that the presence of subclinical seizures may not influence surgical outcome as much as the presence of mesial sclerosis, at least in the case of medial temporal lobe epilepsy. A novel finding in our study involved the relationship of subclinical seizure spread to surgical outcome. Although the majority of subclinical seizures did not propagate, eleven patients had subclinical seizures that spread beyond their onset zone. In these eleven patients, the subclinical seizures of six spread to a different region than their respective clinical seizures. All these patients had poor surgical outcome. Of note, one of these patients with poor outcome had mesial temporal sclerosis with clinical and subclinical seizures arising from the same area; however, their subclinical seizures propagated to a region different from the clinical seizures. The clinical seizures initially propagated medially then laterally throughout the temporal lobe and the subclinical seizures spread to the parietal lobe. In cases of mesial temporal sclerosis, the region of propagation of subclinical seizures may correlate with surgical outcome more than the prescence of sclerosis or the character of the clinical seizures; however, we did not have enough patients with this particular pattern to allow a general statement with statistical confidence. When subclinical seizures came from a different region than the clinical seizures but spread to the same region as the clinical seizures, surgical outcome was poor. This was seen in all patients except for those with mesial temporal lobe sclerosis, as previously discussed. This suggests that correlation with surgical outcome is not only influenced by concordance of the onset zones of subclinical and clinical seizures but also by the ability of subclinical seizures to spread beyond their onset zone and the nature of that spread. Specfically, if the spread pattern differs from the clinical seizures. These three elements suggest that some subclinical seizures may signify an additional underlying epileptic network that is distinct but connected to the clinical epileptic network. Based on the findings of this study and others we conclude and agree that the onset zone of subclinical seizures has similar clinical significance to their clinical counterpart and if included in the region of resection can yield greater surgical success. This appears to be the case for both temporal
P. Farooque, R. Duckrow and extra-temporal lobe epilepsy but, based on our findings, is particularly true for extra-temporal lobe epilepsy. It is known that outcome for surgically treated extra-temporal lobe epilepsy is worse than temporal lobe epilepsy (Haglund and Ojemann, 1993), and it is possible that incorporating subclnical seizures can be used to estimate surgical success. Recognizing that subclinical seizures may depict an additional distinct epileptic network with the potential to influence surgical outcome, they should be included in the surgical decision making process and when estimating the potential for seizure freedom.
Conflicts of interest None.
Acknowledgments Jennifer Bonito, Yale Research Co-ordinator for the maintenance of the Yale Epilepsy Surgery Database. Mark Youngblood, 2nd year Yale Medical Student for assistance with the statistical analysis.
References Engel Jr., J., Van Ness, P., Rasmussen, T., et al., 1993. Outcome with respect to epileptic seizures. In: Engel Jr., J. (Ed.), Surgical Treatment of the Epilepsies. Raven Press, New York, pp. 609—621. Fujiwara, H., et al., 2012. Resection of ictal high-frequency oscillations leads to favorable surgical outcome in pediatric epilepsy. Epilepsia 53, 1607—1617. Götz-Trabert, K., et al., 2008. Spread of ictal activity in focal epilepsy. Epilepsia 49, 1594—1601. Guggisberg, A.G., et al., 2008. Fast oscillations associated with interictal spikes localize the epileptogenic zone in patients with partial epilepsy. Neuroimage 39, 661—668. Haglund, M.M., Ojemann, G.A., 1993. Extratemporal resective surgery for epilepsy. Neurosurg. Clin. N. Am. 4, 283—292. Hirsch, L.J., et al., 1991a. Temporal lobectomy in patients with bitemporal epilepsy defined by depth electroencephalography. Ann. Neurol. 30, 347—356. Hirsch, L.J., et al., 1991b. Comparison of bitemporal and unitemporal epilepsy defined by depth electroencephalography. Ann. Neurol. 30, 340—346. King, D., et al., 1997a. Topographic distribution of seizure onset and hippocampal atrophy: relationship between MRI and depth EEG. Electroencephalogr. Clin. Neurophysiol. 103, 692—697. King, D., et al., 1997b. Surface and depth EEG findings in patients with hippocampal atrophy. Neurology 48, 1363—1367. Mintzer, S., et al., 2004. Unilateral hippocampal sclerosis with contralateral temporal scalp ictal onset. Epilepsia 45, 792—802. Spencer, S.S., 2002. Neural networks in human epilepsy: evidence of and implications for treatment. Epilepsia 43, 219—227. Sperling, M.R., O’Connor, M.J., 1990. Auras and subclinical seizures: characteristics and prognostic significance. Ann. Neurol. 28, 320—328. Walczak, T.S., et al., 1990. Anterior temporal lobectomy for complex partial seizures: evaluation, results, and long-term follow-up in 100 cases. Neurology 40, 413—418. Velkey, A., et al., 2011. Clinical value of subclinical seizures in children with focal epilepsy. Epilepsy Res. 95, 82—85. Zangaladze, A., et al., 2008. Characteristics and clinical significance of subclinical seizures. Epilepsia 49, 2016—2021.
journal homepage: www.elsevier.com/locate/epilepsyres
Subclinical seizures during intracranial EEG recording: Are they clinically significant? Pue Farooque ∗, Robert Duckrow Yale University Comprehensive Epilepsy Center, Department of Neurology, New Haven, CT, United States Received 6 February 2014; received in revised form 9 September 2014; accepted 20 September 2014 Available online 28 September 2014
KEYWORDS Seizures; Subclinical; Epilepsy surgery
Summary Purpose: To evaluate the clinical significance, characteristics and prognostic value of subclinical seizures of temporal and extra-temporal origin. Methods: Our epilepsy database from 2003 to 2011 was reviewed to identify patients with subclinical seizures during intracranial EEG recording who underwent surgical resection. Two groups were formed: Group 1 where both the clinical and subclinical seizures had the same seizure onset region, and Group 2 where some or all of the clinical and subclinical seizures originated from different regions. Results: A total of 27 patients were identified with 791 seizures, of which 310 were subclinical. In Group 1 (n = 14), 64% had good surgical outcome, and 36% had poor surgical outcome. In Group 2 (n = 12), 83% had poor outcome while 17% had good outcome. One patient had only subclinical seizures. Eleven patients had subclinical seizures that propagated to a region beyond their onset zone. Of those 11 patients, 6 patients had subclinical seizures propagate to a different anatomical region than their clinical seizures. These six patients had poor surgical outcome. Discussion: Our study, like others, found that subclinical seizures are clinically significant and including their onset region in the volume of surgical resection correlates with good surgical outcome for both temporal and extra-temporal lobe epilepsy (Fisher’s exact test p = 0.017). Propagation of subclinical seizures to a different region than clinical seizures can affect surgical outcome (Fisher’s exact test p = 0.06). Subclinical seizures may represent a distinct epileptic network. © 2014 Elsevier B.V. All rights reserved.
Introduction ∗ Corresponding author at: Yale University School of Medicine, Department of Neurology, P.O. Box 208018, New Haven, CT 06520, United States. Tel.: +1 203 785 3865; fax: +1 203 737 2799. E-mail address: [email protected] (P. Farooque).
http://dx.doi.org/10.1016/j.eplepsyres.2014.09.020 0920-1211/© 2014 Elsevier B.V. All rights reserved.
Localizing the onset zone of habitual clinical seizures with intracranial electroencephalographic (icEEG) monitoring is a frequent component of the evaluation of patients with medically intractable epilepsy being considered for
Subclinical seizures therapeutic resective surgery. During intracranial monitoring other epileptiform patterns such as sharp transients, high frequency bursts, or subclinical electrographic seizures are often recorded. Considerable effort has been directed to understand their relationship to the seizure onset zone, so that these patterns can be used as supportive evidence during the process of onset localization. Study of interictal spike wave patterns (Walczak et al., 1990) and high frequency oscillations at ictal onset (Fujiwara et al., 2012; Guggisberg et al., 2008) suggests the localizing value of these patterns as a predictor of surgical success. However, there are few studies evaluating the clinical significance and localizing value of subclinical seizures when assessing patients for epilepsy surgery (Sperling and O’Connor, 1990; Velkey et al., 2011; Zangaladze et al., 2008). Subclinical seizures are defined as electrographic seizures with rhythmic ictal discharges that evolve in frequency and space that lack any objective or subjective alteration in behavior or consciousness (Sperling and O’Connor, 1990). They are observed during long-term monitoring with either scalp or intracranial recordings; however their exact prevalence is unknown. Velkey et al. found that 18% of children had subclinical seizures during long-term scalp EEG monitoring and the presence of subclinical seizures influenced the diagnostic process in 85% of cases. Sperling and O’Connor evaluated the clinical characteristics and prognostic value of subclinical seizures of predominantly temporal lobe origin during icEEG monitoring. They found the presence of subclinical seizures to be a favorable prognostic sign for patients undergoing temporal lobectomy (Sperling and O’Connor, 1990) and that subclinical seizures had the same localizing value as clinical seizures (Zangaladze et al., 2008). However, they were unable to draw the same conclusion for subclinical seizures of extra-temporal onset due to lack of sample size. The purpose of this study was to compare the electrographic characteristics of subclinical and clinical seizures of both temporal and extra-temporal origin and to extend our understanding of the localizing and prognostic value of subclinical seizures to patients with extra-temporal lobe epilepsy.
Methods We reviewed the Yale Epilepsy Surgery Program database from 2003 to 2011 and identified 216 patients, who underwent intracranial EEG recording and subsequent resective surgery. Of these patients, 27 had subclinical seizures. At our institution, all patients regardless of epilepsy classification undergo similar intracranial studies consisting of a combination of depth electrodes, subdural strips and a 8 × 8 grid. The maintenance of the database and the methods of this study were approved by the Yale Human Research Protection Program. The clinical behavior and electrographic accompaniment of all seizures were visually analyzed. Subclinical seizures were defined as rhythmic ictal discharges that evolved in frequency and space that lacked any objective or subjective alteration in behavior or consciousness (Sperling and O’Connor, 1990). Seizures were excluded if they caused an arousal during sleep.
1791 Clinical and subclinical seizures were characterized by their onset region, the electrographic features at onset, and the pattern of propagation. Clinical seizures were further classified as typical or atypical based on their habitual semiology. Subclinical seizures were further classified as occurring during the awake or sleep state. The seizure onset region was defined by the electrode contact or contacts where the initial ictal rhythm was seen. Seizure onsets that involved the grid appeared within the grid and did not include contacts at the edge of the grid. Lobar classification of the seizure onset region was temporal, either medial or lateral, or extra-temporal, either frontal or parietal/occipital. Seizure propagation was defined as a definite ictal pattern at electrode contacts occurring not simultaneously at least 2 cm from the seizure onset region (Götz-Trabert et al., 2008). The presence or absence of an MRI lesion including hippocampal sclerosis was noted. The location and extent of the surgical resection was based on all available information. Treatment outcome was staged at least 2 years post surgery using the Engel classification with good outcome denoted by Engel Class I and poor outcome by Engel Class II—IV (Engel et al., 1993). Based on previous studies (Zangaladze et al., 2008), patients were also grouped based on the concordance of clinical and subclinical seizure onset regions. All statistical analysis was done with Fischer Exact test.
Results A total of 27 patients had subclinical seizures during their intracranial EEG recording and underwent focal resective epilepsy surgery of the presumed seizure onset zone. Three of these patients had additional multiple subpial transections (MSTs) of overlapping functional regions. There were 14 patients with temporal lobe epilepsy (12 mesial and 2 lateral) and 13 with extra-temporal epilepsy (5 frontal, 8 parietal/occipital). Of the 791 seizures, 310 were subclinical with 210 of the subclinical seizures being extra-temporal in origin. Fig. 1 illustrates the number of subclinical and clinical seizures arising from each lobar region. Post-operative follow-up ranged from 2 to 9 years with a median of 5.6 years. The surgical outcome groups did not change from two years post surgery to the last follow up as depicted in Table 3. Seventeen patients had lesions on MRI including mesial temporal sclerosis, cavernoma, cortical dysplasia/dysgenesis and nodular heterotopia. Of patients with lesional epilepsy, eight (47%) had good surgical outcome. Of all 27 patients, twelve (44%) had good surgical outcome; of which six had extra-temporal lobe epilepsy. Fifteen patients (56%) had poor surgical outcome with seven of these having extra-temporal lobe epilepsy. The proportion of subclinical seizures to habitual clinical seizures did not correlate to outcome (Data not shown). Table 1 depicts patient demographic and clinical information.
Localization of seizure onset for subclinical seizures In 14 of 27 patients clinical and subclinical seizures arose from the same seizure onset region (Group 1). Of these patients, nine had extra-temporal lobe epilepsy (4 frontal, 5
1792 Table 1 Patient number
P. Farooque, R. Duckrow Patient demographic information. Age
Gender MRI
Surgical resection
Engel classification
Epilepsy classification
Post-op follow-up
1
22
F
RAMTL
4
Temp
5
2
33
F
R medial frontal
1
ET-frontal
3
3 4
38 35
F F
3 1
Temp ET-parietal/occipital
7 4
5
31
F
3
ET-TPO
6
6 7 8 9
16 26 31 59
M F M F
4 3 1 3
Temp Temp ET-frontal ET-occipital
8 9 8 5
10 11
35 45
F M
LAMTL R posterior temporoparietal occipital L mid-temporal to occipital region LAMTL RAMTL R frontal R occipital posterior temporal RAMTL LAMTL
4 1
Temp Temp
8 7
12
29
F
1
ET-parietal
5
13
44
F
1
ET-frontal
7
14
51
M
R parietal occipital junction R frontal anteroinferior parietal RAMTL
1
Temp
6
15
31
M
3
Lateral temp
8
16 17
45 33
F M
R MTS R Nodular heterotopia of lateral ventricle
1 3
Temp ET-parietal
4 3
18
14
F
Gliosis L parietal-occipital lobe
4
ET-TPO
4
19
13
M
3
Lateral temp
7
20 21 22
35 53 47
F F F
Cavernoma L lateral temp R MTS Normal R MTS
RAMTL with 7 cm lateral resection RAMTL MSTs of parietotemporal region, Resection of dysplasia L temporoparietooccipital and occipital lobectomy L lateral temporal
3 1 1
Temp Temp ET-parietal
7 4 8
23
23
M
Normal
4
ET-frontal
8
24 25
9 31
M F
R MTS Normal
1 3
Temp ET-frontal
2 2
26
26
F
Normal
4
ET-occipital
2
27
58
M
L hippocampal atrophy
1
Temp
6
Heterotopia R temp Cortical dysplasia R superior frontal gyrus L MTS R TPO encephalomalacia Cortical dysplasia L occipital lobe L MTS Normal Normal R hippocampal atrophy Normal L temp encephalomalacia S/P R temp lobectomy Normal
Bilat HA, worse on R Normal
RAMTL RAMTL R posterior temporal with MSTs of parietal region Right frontalpolar and orbital RAMTL R frontal lobectomy with MSTs L parietal occipital lobectomy L amygdalohippocampectomy
M, Male; F, Female; R, Right; L, Left; Temp, temporal; ET, Extra temporal; TPO, temporal/parietal/occipital; AMTL, anterior medial temporal lobectomy; MST, Mesial Temporal Sclerosis.
Subclinical seizures
1793
27 paents 791 seizures
Temporal 14 paents 167 seizures
YES
*Co-localized NO
5 paents 70 seizures
42 clinical seizures
Extra-temporal 13 paents 624 seizures
28 subclinical seizures
YES
8 paents 94 seizures
25 clinical seizures
69 subclinical seizures
Co-localized NO
9 paents 547 seizures
358 clinical seizures
189 subclinical seizures
4 paents 77 seizures
56 clinical seizures
21 subclinical seizures
Figure 1 Subgroup of clinical seizures, subclinical seizures, and co-localization. * One patient in the temporal lobe epilepsy group only had three subclinical seizures and no clinical seizures, co-localization was not determined for this patient.
parietal/occipital) and four had lesions. In 12 of 27 patients clinical and subclinical seizures arose from different regions (Group 2). Within this discordant group, four patients had at least one subclinical seizure arise from a different location than their clinical seizures (Group 2a) and eight patients had all their subclinical seizures arise from a different location (Group 2b). For Group 2a (some discordant), there were a total of 105 seizures in four patients where 43 were clinical and 62 were subclinical. Four of the clinical seizures were atypical and only 15 subclinical seizures had the same seizure onset region as their respective clinical seizures (14%). The majority of subclinical seizures within this subgroup did not co-localize to the same region as their respective clinical seizures. Fig. 1 depicts the total number of clinical and subclinical seizures for each onset region where the seizures either co-localized or did not co-localize. Of note, one patient had only subclinical seizures during her intracranial monitoring as her study abruptly ended due to surgical complication prior to clinical seizures being recorded. Her data was excluded from analysis.
Seizure spread Subclinical seizures remained within their seizure onset region in 15 patients (224 subclinical seizures) and propagated beyond their onset region in 11 patients (86 subclinical seizures). Of those 11 patients, six patients (3 temporal, 3 extra-temporal) had subclinical seizures (17 subclinical seizures) that propagated to a different region than their respective clinical seizures. Of the six patients that had subclinical seizures propagate to a different region than their clinical seizures, two had their clinical and subclinical seizures arise from the same area at onset. Of note, both of these patients had the same electrographic
pattern at seizure onset and both patients had abnormalities seen on MRI. Of the five patients who had their subclinical patients spread to the same region as their respective clinical seizures, two of them had their clinical and subclinical seizures arise from the same region.
Surgical outcome Of the 14 patients with co-localized clinical and subclinical seizures (Group 1), 9 (64%) had good surgical outcome, and 5 (36%) had poor surgical outcome. Of the five patients with poor surgical outcome, two underwent focal resection and multiple subpial transections through a functional area that overlapped the seizure onset region, one patient underwent temporal lobectomy although the seizure semiology suggested insular onset, and two patients had subclinical seizures that spread to a different region than their clinical seizures. Of the 12 patients with partially co-localized or not co-localized subclinical and clinical seizures (Group 2) 10 (83%) had poor outcome and 2 (17%) had good outcome. The two patients with good outcome had mesial temporal lobe sclerosis. The side of the sclerosis was resected and either all or some of the subclinical seizures came from the contralateral side. Fig. 2 shows surgical outcome based on the concordance of subclinical and clinical seizure onset, either co-localized or not co-localized. Good surgical outcome assessed at 2 years post surgery is more likely when the subclinical seizure onset region is included in the volume of surgical resection (Fisher’s exact test p = 0.017). Good surgical outcome assessed at last follow up (range 2—9 years) was also more likely when the subclinical seizure onset region is included in the volume of surgical resection (Fisher’s exact test p = 0.017).
1794
P. Farooque, R. Duckrow 16
Surgical Outcome
Number of Paents
14 12 10 Not Co-localized
8
Co-localized
6 4 2 0
Good Outcome
Poor Outcome
Figure 2 Relationship between subclinical and clinical seizures arising from either the same or different region of onset and its affect on surgical outcome. Treatment outcome was staged at least 2 years post surgery using the Engel classification with good outcome denoted by Engel Class I and poor outcome by Engel Class II—IV. Good surgical outcome is more likely when the subclinical seizure onset region is included in the volume of surgical resection (Fisher’s exact test p = 0.017). This was true for 2 years post surgery as well as at last follow-up.
Eleven patients had subclinical seizures spread beyond their onset region. Of these eleven patients, six had subclinical seizures that spread to a different region than their respective clinical seizures and all six had poor surgical outcome. Three of these six had temporal onset and three had extra-temporal onset. Of the six patients with disparate spread patterns and poor outcome, two patients had lesions with clinical and subclinical seizures arise from the same onset region, elements usually associated with good outcome. Among the five patients that had subclinical seizures spread to the same region as their clinical seizures, two had poor surgical outcome. Their subclinical and clinical seizures had different onset regions. Of the three patients whose seizures spread to the same region and had good surgical outcome only one had their subclinical seizures partially co-localize with their clinical seizures at onset (Group 2a). This patient had mesial temporal sclerosis with five clinical seizures and sixteen subclinical seizures. One clinical seizure was atypical in character and had onset in the temporal lobe opposite to the side of the sclerosis. Eleven subclinical seizures had the same seizure onset zone as the typical clinical seizures on the side of the sclerosis. The other five subclinical seizures arose from the contralateral temporal lobe, again opposite to the side of the sclerosis. Table 2 correlates the pattern of subclinical and subclinical seizure spread with seizure onset to surgical outcome. The propagation of subclinical seizures to a different region than clinical seizures correlates with poor surgical outcome (Fisher’s exact test p = 0.06).
Discussion We studied patients with subclinical seizures during intracranial EEG recording who subsequently underwent epilepsy surgery to investigate the localizing value and prognostic significance of subclinical seizures in relation to surgical outcome. We specifically wanted to evaluate if conclusions based on subclinical seizures of extra-temporal onset would be the same as those made for temporal lobe epilepsy. Our cohort of patients was equally split among temporal and extra-temporal epilepsy patients, which aided the comparison. We found that our extra-temporal lobe patients had twice the number of subclinical seizures as our temporal lobe patients while previous studies found the majority of subclinical seizures were temporal in origin (Zangaladze et al., 2008). This difference may be secondary to increased spatial sampling during the intracranial implant, where our average number of electrodes implanted for an extra-temporal case was 205. Another contrast is that more than half of our subclinical seizures of temporal onset did not arise from the same region as the coresponding clinical seizures while previous studies found the majority of temporal lobe subclinical seizures co-localized with clinical seizures (Zangaladze et al., 2008). It is possible that our temporal lobe epilepsy patients had more diffuse disease that was refractory to treatment, as evidenced by 57% having poor surgical outcome. We found that when subclinical seizures and clinical seizures arise from the same region and that region is included in surgical resection, 64% of patients had good
Table 2 This table shows subclinical seizures (SCS) either co-localizing or not co-localizing at onset with their clinical seizures (CS) in relation to the spread of SCS to either the same or different region than its respective CS and the effect on surgical outcome. Spread pattern
Good outcome
Poor outcome
Patients with SAME SCS as CS onset and SAME SCS as CS spread Patients with SAME SCS as CS onset but DIFFERENT spread All patients with spread that DIFFERED between CS and SCS, regardless of onset.
2 0 0
0 2 6
Pt #
Number of clinical seizures
Number of subclinical seizures
Spread
Spread different than clinical
MRI lesion
Engel classification at Last follow-up
Co-localization of subclinical and clinical seizures
Seizure classification
N/A N/A Yes, different-parietal N/A Yes, differentorbitofrontal N/A N/A No, same Yes, different-parietal N/A N/A N/A N/A No, same Yes, different-frontal N/A N/A No, same No, same Yes, different-parietal N/A No, same Yes, different-lateral temporal N/A N/A N/A N/A
Yes Yes Yes
4 1 3
2b 1 1
T ET-F T
Yes Yes
1 3
1 1
ET-P/O ET-TPO
Yes No No Yes
4 3 1 3
2b 2a 1 2b
T T ET-F ET-O
No Yes No No Yes No
4 1 1 1 1 3
1 2b 1 1 2a 2a
T T ET-P ET-F T LT
Yes Yes Yes Yes Yes
1 3 4 3 3
N/A 1 2a 2b 2b
T ET-P ET-P/O LT T
No Yes No
1 1 4
1 1 2b
T ET-P ET-F
Yes No No Yes
1 3 4 1
1 1 2b 1
T ET-F ET-O T
1 2 3
4 11 5
3 90 2
No No Yes
4 5
3 107
11 3
No Yes
6 7 8 9
8 8 12 9
1 41 10 5
No No Yes Yes
10 11 12 13 14 15
11 2 141 12 5 2
7 2 44 2 16 2
No No No No Yes Yes
16 17 18 19 20
0 4 28 1 4
3 1 3 16 1
No No Yes Yes Yes
21 22 23
6 5 37
1 24 4
No Yes Yes
24 25 26 27
8 35 10 3
2 2 11 3
No No No No
1795
T, temporal; LT, lateral temporal; P, parietal; O, occipital; F, frontal; ET, extra temporal; OF, orbitofrontal; 1, subclinical and clinical seizures onset match; 2a, partial match of onset between subclinical and clinical seizures; 2b, no match between onset of clinical and subclinical seizures; N/A, not applicable.
Subclinical seizures
Table 3 Number of clinical seizures, subclinical seizures, presence of spread for subclinical seizures, region of spread for subclinical seizures, Presence of MRI abnormality, surgical outcome at 2 years post surgery and last follow-up and degree of co-localization for all epilepsy subgroups.
1796 surgical outcome. This confirms the earlier observations of Sperling and O’Connor (1990) and Zangaladze et al. (2008) and extends them to extra-temporal epilepsy. Conversely, if subclinical seizures arise from a different region than clinical seizures and that region was not included in the area of resection, 83% of patients had poor surgical outcome. Two patients had good surgical outcome despite their subclinical and clinical seizures coming from different regions. Both of these patients had unilateral mesial temporal sclerosis, where some or all of their subclinical seizures arose from the temporal lobe contralateral to the scelerosis and region of clinical seizure onset. These subclinical seizures either did not spread beyond their seizure onset zone or spread to the same region as their respective clinical seizures. When bilateral independent seizure onset is seen in patients with unilateral medial temporal sclerosis, resecting the side with the sclerosis provides a high probability of becoming seizure free, as reported in previous studies (Hirsch et al., 1991a, 1991b; King et al., 1997a, 1997b; Mintzer et al., 2004; Spencer, 2002). This suggests that the presence of subclinical seizures may not influence surgical outcome as much as the presence of mesial sclerosis, at least in the case of medial temporal lobe epilepsy. A novel finding in our study involved the relationship of subclinical seizure spread to surgical outcome. Although the majority of subclinical seizures did not propagate, eleven patients had subclinical seizures that spread beyond their onset zone. In these eleven patients, the subclinical seizures of six spread to a different region than their respective clinical seizures. All these patients had poor surgical outcome. Of note, one of these patients with poor outcome had mesial temporal sclerosis with clinical and subclinical seizures arising from the same area; however, their subclinical seizures propagated to a region different from the clinical seizures. The clinical seizures initially propagated medially then laterally throughout the temporal lobe and the subclinical seizures spread to the parietal lobe. In cases of mesial temporal sclerosis, the region of propagation of subclinical seizures may correlate with surgical outcome more than the prescence of sclerosis or the character of the clinical seizures; however, we did not have enough patients with this particular pattern to allow a general statement with statistical confidence. When subclinical seizures came from a different region than the clinical seizures but spread to the same region as the clinical seizures, surgical outcome was poor. This was seen in all patients except for those with mesial temporal lobe sclerosis, as previously discussed. This suggests that correlation with surgical outcome is not only influenced by concordance of the onset zones of subclinical and clinical seizures but also by the ability of subclinical seizures to spread beyond their onset zone and the nature of that spread. Specfically, if the spread pattern differs from the clinical seizures. These three elements suggest that some subclinical seizures may signify an additional underlying epileptic network that is distinct but connected to the clinical epileptic network. Based on the findings of this study and others we conclude and agree that the onset zone of subclinical seizures has similar clinical significance to their clinical counterpart and if included in the region of resection can yield greater surgical success. This appears to be the case for both temporal
P. Farooque, R. Duckrow and extra-temporal lobe epilepsy but, based on our findings, is particularly true for extra-temporal lobe epilepsy. It is known that outcome for surgically treated extra-temporal lobe epilepsy is worse than temporal lobe epilepsy (Haglund and Ojemann, 1993), and it is possible that incorporating subclnical seizures can be used to estimate surgical success. Recognizing that subclinical seizures may depict an additional distinct epileptic network with the potential to influence surgical outcome, they should be included in the surgical decision making process and when estimating the potential for seizure freedom.
Conflicts of interest None.
Acknowledgments Jennifer Bonito, Yale Research Co-ordinator for the maintenance of the Yale Epilepsy Surgery Database. Mark Youngblood, 2nd year Yale Medical Student for assistance with the statistical analysis.
References Engel Jr., J., Van Ness, P., Rasmussen, T., et al., 1993. Outcome with respect to epileptic seizures. In: Engel Jr., J. (Ed.), Surgical Treatment of the Epilepsies. Raven Press, New York, pp. 609—621. Fujiwara, H., et al., 2012. Resection of ictal high-frequency oscillations leads to favorable surgical outcome in pediatric epilepsy. Epilepsia 53, 1607—1617. Götz-Trabert, K., et al., 2008. Spread of ictal activity in focal epilepsy. Epilepsia 49, 1594—1601. Guggisberg, A.G., et al., 2008. Fast oscillations associated with interictal spikes localize the epileptogenic zone in patients with partial epilepsy. Neuroimage 39, 661—668. Haglund, M.M., Ojemann, G.A., 1993. Extratemporal resective surgery for epilepsy. Neurosurg. Clin. N. Am. 4, 283—292. Hirsch, L.J., et al., 1991a. Temporal lobectomy in patients with bitemporal epilepsy defined by depth electroencephalography. Ann. Neurol. 30, 347—356. Hirsch, L.J., et al., 1991b. Comparison of bitemporal and unitemporal epilepsy defined by depth electroencephalography. Ann. Neurol. 30, 340—346. King, D., et al., 1997a. Topographic distribution of seizure onset and hippocampal atrophy: relationship between MRI and depth EEG. Electroencephalogr. Clin. Neurophysiol. 103, 692—697. King, D., et al., 1997b. Surface and depth EEG findings in patients with hippocampal atrophy. Neurology 48, 1363—1367. Mintzer, S., et al., 2004. Unilateral hippocampal sclerosis with contralateral temporal scalp ictal onset. Epilepsia 45, 792—802. Spencer, S.S., 2002. Neural networks in human epilepsy: evidence of and implications for treatment. Epilepsia 43, 219—227. Sperling, M.R., O’Connor, M.J., 1990. Auras and subclinical seizures: characteristics and prognostic significance. Ann. Neurol. 28, 320—328. Walczak, T.S., et al., 1990. Anterior temporal lobectomy for complex partial seizures: evaluation, results, and long-term follow-up in 100 cases. Neurology 40, 413—418. Velkey, A., et al., 2011. Clinical value of subclinical seizures in children with focal epilepsy. Epilepsy Res. 95, 82—85. Zangaladze, A., et al., 2008. Characteristics and clinical significance of subclinical seizures. Epilepsia 49, 2016—2021.
