Electrocorticographic factors associated with temporal lobe epileptogenicity

Transcription

1 Pathophysiology 7 (2000) Electrocorticographic factors associated with temporal lobe epileptogenicity M.E. Weinand a, *, M. Deogaonkar a, M. Kester b, G.L. Ahern c, D.M. Labiner c a Di ision of Neurosurgery, Department of Surgery, P.O. Box , 1501 N. Campbell A enue, Uni ersity of Arizona College of Medicine, Tucson, AZ , USA b Arizona Pre ention Center, Uni ersity of Arizona College of Medicine, Tucson, AZ 85721, USA c Department of Neurology, Uni ersity of Arizona College of Medicine, Tucson, AZ 85724, USA Received 19 July 1999; received in revised form 1 October 1999; accepted 22 November 1999 Abstract Continuous subdural electrocorticographic (ECoG) monitoring was performed to test the hypothesis that human temporal lobe epileptogenicity, during long-term monitoring following antiepileptic drug (AED) withdrawal, regardless of the specific AED regimen, is dependent upon ECoG ictal onset and interhemispheric spread of epileptic activity. In 121 patients, ECoG parameters were analyzed for association with seizure frequency, a clinical measure of epileptogenicity. Significantly associated with increased seizure frequency were: ictal medial temporal lobe onset, absence of ictal frontal lobe desynchronization and short interhemispheric propagation time (IHPT). Seizure frequency during long-term ECoG monitoring was not predictive of post-operative seizure outcome. It is concluded that, following AED withdrawal, regardless of the specific AED regimen, increased seizure frequency is associated with medial temporal lobe ictal onset, short IHPT and absence of frontal lobe desynchronization. The results confirm the hypothesis that human temporal lobe epileptogenicity, after withdrawal, is dependent upon ECoG ictal onset and interhemispheric spread of epileptic activity. Future development of procedures which promote ECoG factors associated with increased seizure frequency following AED withdrawal might decrease duration of invasive long-term monitoring and improve efficiency for the pre-surgical selection of temporal lobectomy candidates. Intervention producing ictal frontal lobe desynchronization and increased IHPT might inhibit temporal lobe epileptogenicity and should be evaluated for therapeutic efficacy outside of the long-term monitoring context Elsevier Science Ireland Ltd. All rights reserved. Keywords: Electrocorticography; Epilepsy; Epilepsy surgery; Epileptogenicity; Seizure frequency; Temporal lobe seizure 1. Introduction Long-term seizure monitoring is commonly preceded by withdrawal of antiepileptic drugs (AEDs) [1 3]. Abrupt AED withdrawal precipitates increased frequency of seizures, thus reducing the time for continuous electrophysiologic recording [2]. The electrocorticographic (ECoG) seizure discharge during long-term monitoring of AEDs is generally similar in morphology and topography to epileptic activity prior to AED withdrawal. For instance, seizures which occur Presented in abstract form at The 22nd International Epilepsy Congress, Dublin, Ireland, June 29 July 6, * Corresponding author. Tel.: ; fax: address: ndw@u.arizona.edu (M.E. Weinand) during AED withdrawal can ordinarily be assumed to provide accurate ictal localization for single seizure foci as long as the clinical seizure phenomenology is consistent with the patient s habitual attacks [2,4,5]. However, seizure frequency is increased and seizures are more likely to spread and become secondarily generalized after AED withdrawal [6]. Although the duration of the seizure discharge is comparable to the usual clinical phenomenology [6], the increase in seizure frequency is long-lasting [1]. The effect of AED withdrawal in producing increased seizure frequency has been described as an early rebound phenomenon in association with removal of the suppressive effect of AEDs [6]. AED withdrawal does not ordinarily affect ictal ECoG pattern, interhemispheric propagation time (IHPT), or coherence between homologous regions of temporal lobes [6] /00/$ - see front matter 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S (99)

2 34 M.E. Weinand et al. / Pathophysiology 7 (2000) The ECoG factors associated with increased seizure frequency following AED withdrawal have not been delineated. This study was performed to test the hypothesis that human temporal lobe epileptogenicity, following AED withdrawal, regardless of the specific AED regimen, is dependent upon ECoG factors associated with ictal onset and interhemispheric spread of epileptic activity. The results should improve understanding of pathophysiology of increased temporal lobe epileptogenicity following AED withdrawal. Because this study was designed to delineate general electrophysiologic associations between ECoG parameters and seizure frequency, during long-term ECoG recording, no specific relationships between individual AEDs and ECoG factors were sought. This report defines the general relationships between ECoG parameters and seizure frequency during long-term ECoG recording. The results have implications for predicting anticipated duration of hospital stay during long-term monitoring and, possibly, for the future development of procedures promoting ECoG factors associated with increased seizure frequency during long-term ECoG monitoring. This study may also have relevance for future development of therapy designed to promote ECoG factors associated with inhibition of temporal lobe epileptogenicity. 2. Subjects and methods 2.1. Patient population A consecutive series of 121 patients with medically intractable seizures were studied. In each patient, an MRI brain scan was performed to confirm absence of a structural lesion. All patients underwent video/scalp- EEG monitoring which proved insufficient for seizure focus localization due to at least one of the following: (a) EEG onset too diffuse to be localized to a single lobe of brain; (b) bilateral EEG onset; (c) clinical seizure onset followed by EEG onset; or (d) EEG onset obscured by electromyographic or other artifact. Therefore, each patient underwent implantation of bilateral subdural strip electrodes (AD-TECH, Racine, WI) for long-term ECoG monitoring [7,8]. Immediately prior to implantation of subdural electrodes, anticonvulsant medication was discontinued Surgical technique: subdural strip electrodes In each patient, four to eight subdural strips were placed using standard surgical techniques [7,8]. In all patients, temporal burr holes were placed for implantation of bilateral 4-contact medial and 4-contact lateral temporal strips. In some patients, bilateral frontal paramedian burr holes were made, through which additional bilateral 4-contact medial interhemispheric and 8-contact orbital-frontal subdural strips were placed. In some patients, frontal burr holes were not made and the temporal burr holes were also used for subdural implantation of bilateral 8-contact lateral frontal strips Interictal and ictal ECoG data acquisition In all patients, continuous 24-h subdural ECoG recording was performed. ECoG data were digitized in time-sequence with clinical phenomenology and stored on videotape. Seizure onset was defined as the first sustained change from background ECoG activity [7]. In all ECoG records, the following were analyzed for association with seizure frequency: (a) interictal seizure focus location [lobe of predominate interictal epileptic discharges ] [7]; (b) ictal seizure focus location; (c) presence or absence of frontal lobe desynchronization at ictal ECoG onset; (d) time from ECoG seizure onset to beginning of clinical seizure phenomenology ( electrical-clinical onset time ); (e) ictal ECoG seizure onset pattern (fast spike trains or low voltage fast, electrodecremental activity); (f) first location of ECoG epileptic activity spread following ictal ECoG onset; (g) time from ECoG seizure onset to contralateral ECoG epileptic activity spread (IHPT), and (h) presence or absence of ECoG secondary generalization. In every case, to eliminate potential inaccuracies associated with patient self-reporting, seizure frequency was calculated using long-term ictal ECoG data digitized in time-sequence with clinical seizure phenomenology Surgical technique: temporal lobectomy Based on ictal ECoG data, all patients underwent anterior temporal lobectomy with amygdalohippocampectomy. In all patients, a standard 4.5-cm lateral temporal corticectomy was performed. Amygdalectomy was performed and the hippocampus was resected to at least the level of the cerebral peduncle. In patients with a lateral temporal lobe focus, a more extensive ECoGguided lateral temporal cortical resection was performed. For regional (medial and lateral) temporal lobe onset, a more extensive removal of parahipppocampal gyrus, hippocampus and lateral temporal cortex was done [7] Seizure outcome Following temporal lobectomy, all patients continued on anti-convulsant medication. At minimum 12 months post-temporal lobectomy, seizure outcome was classified as: seizure-free (no seizures, including no auras), significantly improved ( 90% seizure frequency reduction), improved (50 90% seizure frequency reduction), or unchanged ( 50% seizure

3 M.E. Weinand et al. / Pathophysiology 7 (2000) frequency reduction). Patients who seized due to anticonvulsant noncompliance were assigned to the appropriate outcome category based on seizure frequency Statistical analysis In each patient, all ECoG and post-operative seizure outcome data were analyzed for association with seizure frequency. In all patients, bivariate statistical analysis was performed on all ECoG data. Median seizure frequency (1.00) was used as a cut-off for defining a binary variable indicating high or low seizure frequency. Chi-square tests were performed to determine significant bivariate associations between this new binary outcome variable and ECoG variables. The P- values of these tests are reported. The odds ratio is presented as a measure of the strength of association. Logarithmic regression was used to explore multivariate relationships of ECoG parameters with seizure frequency. A significance level of 0.05 was used for all statistical tests. 3. Results 3.1. Patient population In all 121 patients, analysis of clinical phenomenology confirmed the diagnosis of complex partial seizures. Among 121 patients, eight subdural strips were placed in 58 patients (47.9%), six strips in 54 patients (44.6%), and four strips in seven patients (5.8%). Seven strips were placed in one patient (0.8%) and five in another (0.8%); one medial and one lateral frontal strip, respectively, omitted because of bleeding from the superior Table 1 Electrocorticograhic (ECoG) factors associated with temporal lobe epileptogenicity (seizure frequency) during long-term ECoG monitoring following antiepileptic drug withdrawal a Bivariate analysis b Odds ratio n P-value Medial temporal lobe ictal focus Ictal frontal lobe desynchronization Multivariate analysis Parameter P-value estimate Linear regression model Dependent variable: log 10 seizure frequency n=44 Frontal lobe desynchronization Interhemispheric propagation time Constant a n=number of patients. b Outcome variable: median seizure frequency (1.00 seizure/day). sagittal sinus. For 89 patients, some clinical and demographic data have been previously reported [7]. There were 64 males and 57 females. In every patient, MRI brain scan confirmed the absence of a structural lesion. The mean age was 30 years (range: 4 54 years). For 117 patients in whom ECoG seizure onset was localized, the ictal epileptic focus was identified as follows: medial (n=74), lateral (n=12), regional (medial and lateral) (n=29) and bilateral temporal lobe (n=2). During long-term ECoG recording, median and mean seizure frequency were 1.00 and seizures/day, respectively (mean standard error, S.E.; range: seizures/day) Bi ariate statistical analysis Factors associated with seizure frequency During long-term ECoG monitoring of AED medications, the odds of averaging one or more seizures per day for patients with medial temporal lobe ictal ECoG onset (n=74) were 2.3 times greater than for patients with lateral, regional or bilateral temporal lobe onset (n=43) (P=0.037) (Table 1, Fig. 1). The odds of averaging less than one seizure per day for patients with ictal frontal lobe ECoG desynchronization (n=13) were 3.91 times that of patients without frontal lobe ECoG desynchronization (n=93) (P=0.025) (Fig. 2) Factors not associated with seizure frequency The odds of averaging one or more seizures per day for patients with focal interictal frontal or temporal lobe epileptic discharges (n=90) were 0.4 times that of patients with diffuse frontal-temporal lobe epileptic spikes (n=20) (P 0.05). The odds of averaging one or fewer seizures per day for patients with unilateral temporal lobe onset (n= 103) were 0.43 times that of patients in whom ECoG ictal onset began in both hemispheres followed by focal temporal lobe epileptic discharge (n=11) (P 0.05). The odds of averaging one or fewer seizures per day for patients with ECoG ictal onset with low voltage fast (electrodecremental) pattern (n=61) were 0.51 times that of patients with ictal ECoG onset beginning in the form of high voltage, fast spike trains (n=52) (P 0.05). The odds of averaging one or fewer seizures per day for patients with post-ictal ECoG secondary generalization (n=24) were 1.86 times that of patients without secondary generalization (n=90) (P 0.05). The odds of averaging one or fewer seizures per day for patients in whom ECoG preceded clinical seizure onset (n=42) were 1.80 times that of patients in whom ECoG followed clinical seizure onset (n=6) (P 0.05). The odds of averaging one or fewer seizures per day for patients in whom IHPT was 8 s were 1.05 times that of patients in whom IHPT was 8 s(p 0.05).

4 36 M.E. Weinand et al. / Pathophysiology 7 (2000) Fig. 1. Subdural electrocorticographic (ECoG) record of medial temporal lobe ictal onset (arrow). Compared to lateral or regional (medial and lateral) temporal lobe seizure onset, medial temporal lobe ECoG ictal onset is associated with greatest seizure frequency. LMT, left medial temporal; RMT, right medial temporal lobe subdural electrodes, respectively, (1, most distal electrode; 4, most proximal electrode). Fig. 2. Subdural electrocorticographic (ECoG) record of frontal lobe desynchronization (ROF 1 8) in association with temporal (RMT 1 4) lobe ictal ECoG onset. Presence of ictal frontal lobe desynchronization is associated with low seizure frequency. ROF, right orbital-frontal; RMF, right medial frontal; RMT, right medial temporal subdural strips. (1, most distal electrode; 4, most proximal electrode). In 106 patients, the first subdural ECoG epileptic activity spread after ictal onset was localized ipsilateral (n=60), contralateral (n=11) and bilateral (n=35) relative to ictal ECoG temporal lobe onset. The odds of averaging one or fewer seizures per day for patients in whom ictal ECoG epileptic activity first spread within

5 M.E. Weinand et al. / Pathophysiology 7 (2000) the ipsilateral hemisphere (n=60) were 1.05 times that of patients in whom initial spread was simultaneous in both hemispheres (n=35) (P 0.05) Multi ariate analysis Logarithmic regression analysis of all ECoG data showed that the strongest association with seizure frequency was a model consisting of ictal frontal lobe desynchronization and IHPT. Seizure frequency decreased with (a) ictal frontal lobe desynchronization and (b) long IHPT (Table 1, Fig. 3). Patients with ictal frontal lobe desynchronization had, on average, 2.45 fewer seizures per day than those without ictal frontal lobe desynchronization. For every 10 s of increase in IHPT, the number of seizures per day decreased, on average, by Several additional multivariate logistic regression models were estimated based on the bivariate associations that were discovered, but none of these models fit the data well Epileptogenicity and seizure outcome Following temporal lobectomy, seizure outcome data in 114 patients with minimum 1 year (mean: 23 months; range: months) follow-up was: 73 patients (64.0%) seizure-free; 32 patients (28.1%) significantly improved; four patients (3.5%) improved; and five patients (4.4%) unchanged. In seven patients, post-operative follow-up was less than 1 year. There was no association between pre-operative seizure frequency, during long-term ECoG monitoring, and post-temporal lobectomy seizure outcome (1 unit seizure frequency increase odds ratio=0.94, P 0.05). 4. Discussion Among epilepsy centers, there is considerable diversity regarding electrophysiologic selection criteria for temporal lobectomy. Some centers employ prolonged interictal EEG and/or intraoperative interictal ECoG for surgical decision making [9,10]. Others favor longterm extraoperative ictal EEG and/or ECoG data for seizure focus localization [7,8]. Because the neurons and physiologic pertubations responsible for interictal and ictal epileptogenic activity are probably different [11], this study was performed to analyze individual interictal, ictal and post-ictal ECoG parameters for association with human temporal lobe epileptogenicity during long-term subdural ECoG monitoring following antiepileptic drug withdrawal Factors associated with epileptogenicity Ictal focus There has been considerable documentation of the epileptogenic potential of medial temporal-limbic structures, with emphasis on the intrinsic epileptogenicity of the hippocampus [10,12,13]. The particular vulnerability of the medial temporal-limbic region to produce seizures appears to be multi-factorial, comprising numerous anatomic, pharmacological and physiologic abnormalities [14]. These include susceptibility to anoxia Fig. 3. Subdural electrocorticographic (ECoG) record of interhemispheric propgation time (IHPT). The best statistical model finds that seizure frequency decreases logarithmically as IHPT increases. In this illustration, IHPT=12.5 s. LMT, left medial temporal; RMT, right medial temporal lobe subdural electrodes. (1, most distal electrode; 4, most proximal electrode).

6 38 M.E. Weinand et al. / Pathophysiology 7 (2000) and herniation [15]; low threshold for seizure activity and rapid spread of epileptic discharges [15]; ease of kindling [15]; distinctive pharmacologic and electrophysiologic charcteristics including rapid rates of protein synthesis [15], abnormal glutamate regulation [16] and enhanced inhibition and excitation [13]; and hippocampal structural and biochemical genetic abnormalities [15]. The data suggest that the medial temporal lobe ictal focus is associated with significantly greater epileptogenicity, following AED withdrawal, than that associated with lateral or regional temporal lobe seizure onset. Because the topography and morphology of ictal ECoG data are not commonly altered by AED withdrawal, long-term ECoG monitoring in patients with medial temporal lobe ictal foci should be expected to be more efficient (i.e. with shorter duration of hospital stay due to more frequent seizures) than patients with lateral or regional temporal lobe ictal foci Frontal lobe desynchronization in temporal lobe epilepsy Epileptogenesis and transition from the interictal to ictal state are influenced by neuronal excitability and synchronization [12,14]. Overt epileptic activity results from conversion of paroxysmal firing of a few neurons to more extensive neuronal synchronization [14]. The physiologic mechanisms underlying neuronal synchronization are complex and multifactorial, including excitatory (e.g. NMDA receptors and channels) and inhibitory (e.g. depolarizing GABA potentials) interactions, intrinsic low-threshold calcium currents and extracellular ephaptic electrical interactions [10]. Fully developed epileptic activity probably requires recruitment of a critical mass of neurons (Group 2 neurons) [17,18] which are thought to modulate synchronization and desynchronization of epileptic activity in response to a complex array of internal and external factors. For example, during operant conditioning of monkeys with epileptogenic cortex, the electroencephalogram shows decreased voltage and increased frequency [17]. As opposed to Group 1 neurons which have more autonomous hyperexcitable firing patterns, Group 2 neurons are very labile in their propensity to burst and are also most affected by operant conditioning. Group 2 neurons may provide the critical mass for enlargement of the epileptic focus because of their proclivity to be synchronized by massive synaptic influences [17]. Desynchronization or disruption of synchrony of these neurons may produce elevation of the seizure threshold and reduction of seizure frequency. Human temporal lobe neuronal synchronization is limited by almost exclusive ipsilateral connectivity [10]. Once seizure activity spreads outside the temporal lobe, further synchronization may be responsible for augmenting and sustaining epileptic activity and clinical seizure phenomenology. It has been suggested that ECoG desynchronization should decrease neuronal recruitment [19] and facilitate seizure reducton [20]. The results of the current study suggest that ECoG desynchronization, when located in frontal cortex, is associated with decreased temporal lobe epileptogenicity following AED withdrawal. Conversely, absence of frontal lobe desynchronization is associated with increased temporal lobe seizure frequency. Consistent with experimental data [17,21] and human clinical studies [19], ictal frontal lobe desynchronization may be an inhibitory neurophysiologic process in human temporal lobe epilepsy. Desynchronization may prevent recruitment of adjacent cortex into the critical mass required for propagation and full development of the ictal event [19] IHPT IHPT is the duration from ictal ECoG onset to the first appearance of ECoG epileptic activity in contralateral cortex. A quantitative relationship between human IHPT and seizure frequency has, heretofore, not been reported. The logarithmic regression model shows that increased IHPT is significantly associated with decreased seizure frequency during long-term monitoring. Conversely, decreased IHPT is associated with increased seizure frequency. Because the best statistical model associated with seizure frequency is logarithmic, new behavioral, pharmacologic or other intervention designed to prevent ictal frontal lobe desynchronization and to produce even small decreases in IHPT could have great potential for increasing seizure frequency and reducing duration of hospital stay during long-term invasive monitoring. Alternatively, intervention producing ictal frontal lobe desynchronization and increased IHPT might inhibit temporal lobe epileptogenicity and should be studied for possible therapeutic efficacy outside of the long-term monitoring context Epileptogenicity and seizure outcome The results show that seizure frequency, following AED withdrawal during long-term ECoG monitoring, is not prognostic for seizure outcome following temporal lobectomy. Previously reported data have shown that pre-operative seizure frequency does not influence post-operative seizure outcome [22 24]. Therefore, preoperative seizure frequency, including data from longterm electrophysiologic monitoring following AED withdrawal, should not influence seizure outcome prognostic counseling in temporal lobectomy candidates.

7 M.E. Weinand et al. / Pathophysiology 7 (2000) Conclusion The results support the conclusion that temporal lobe epileptogenicity, during long-term ECoG monitoring following AED withdrawal, depends on ECoG factors invlolved in ictal onset and interhemispheric spread of epileptic activity. The following ECoG factors are associated with significantly increased temporal lobe seizure frequency during long-term ECoG recording of AEDs: medial temporal lobe seizure onset and lack of ictal frontal lobe desynchronization. The best logarithmic regression model associated with seizure frequency includes ictal frontal lobe desynchronization and IHPT. Temporal lobe seizure frequency increases, during longterm monitoring off AED medications, with (a) absence of ictal frontal lobe desynchronization and (b) short IHPT. Seizure frequency during long-term ECoG monitoring following AED withdrawal has no prognostic value for seizure outcome following temporal lobectomy. Therefore, during evaluation of temporal lobectomy candidates, seizure frequency during long-term monitoring should not influence seizure outcome prognostic counseling. This study should improve understanding of ECoG data which are associated with temporal lobe epileptogenicity during long-term electrophysiologic monitoring. Further research is needed to elucidate the fundamental pertubation responsible for ECoG changes associated with epileptogenicity following AED withdrawal. References [1] M. Marciani, J. Gotman, F. Andermann, A. Olivier, Patterns of seizure activation following withdrawal of antiepileptic medication, Neurology 35 (1985) [2] S. Spencer, D. Spencer, P. Williamson, R. Mattson, Ictal effects of anticonvulsant medication withdrawal in epileptic patients, Epilepsia 22 (1981) [3] S.S. Spencer, D.D. Spencer, P.D. Williamson, R.H. Mattson, The localizing value of depth electroencephalography in 32 patients with refractory epilepsy, Ann. Neurol. 12 (1982) [4] W. Theodore, Abrupt withdrawal of antiepileptic drugs for intensive video EEG monitoring, in: E. Wyllie (Ed.), The Treatment of Epilepsy: Principles and Practice, Lea & Febiger, Philadelphia, [5] J. Duncan, S. Shorvon, M. Trimble, Discontinuation of phenytoin, carbamazepine, and valproate in patients with active epilepsy, Epilepsia 31 (1990) [6] N. So, J. Gotman, Changes in seizure activity following anticonvulsant drug withdrawal, Neurology 40 (1990) [7] M. Weinand, A. Wyler, E. Richey, B. Phillips, G. Somes, Long-term ictal monitoring with subdural strip electrodes: prognostic factors for selecting temporal lobectomy candidates, J. Neurosurg. 77 (1992) [8] A.R. Wyler, E.T. Richey, B. Hermann, Comparison of scalp to subdural recordings for localizing epileptogenic foci, J. Epilepsy 2 (1989) [9] G.A. Ojemann Jr, Intraoperative tailoring of temporal lobe resections, in: J. Engel Jr (Ed.), Surgical Treatment of the Epilepsies, Raven Press, New York, 1993, pp [10] P. Schwartzkroin Jr, Basic research in the setting of an epilepsy surgery center, in: J. Engel Jr (Ed.), Surgical Treatment of the Epilepsies, Raven Press, New York, 1993, pp [11] W. Wilson, H. Swartzwelder, W. Anderson, D.V. Lewis, Seizure activity in vitro: a dual focus model, Epilepsy Res. 2 (1988) [12] J. Engel Jr, Seizures and Epilepsy, F.A. Davis, Philadelphia, [13] H. Wieser, J. Engel Jr, P.D. Williamson, T.L. Babb, P. Gloor Jr, Surgically remediable temporal lobe syndromes, in: J. Engel Jr (Ed.), Surgical Treatment of the Epilepsies, 2nd edition, Raven Press, New York, 1993, pp [14] E. Niedermeyer, The Epilepsies. Diagnosis and Management, Urban & Schwarzenberg, Baltimore, [15] G. Glaser Jr, Natural history of temporal lobe-limbic epilepsy, in: J. Engel Jr (Ed.), Surgical Treatment of the Epilepsies, Raven Press, New York, 1987, pp [16] P. Lleu, D. Labiner, M. Weinand, R. Huxtable, Neuroactive amino acids in synaptosomes from focal and nonfocal temporal lobe tissue of patients with intractable complex partial seizures, in: R. Huxtable, D. Michalk Jr (Eds.), Taurine in Health and Disease, Plenum Press, New York, 1994, pp [17] A.R. Wyler, Operant conditioning of epileptic neurons and its theoretical application to EEG operant conditioning in humans, Pav. J. Biol. Sci. 12 (1977) [18] A. Wyler, K. Burchiel, A. Ward Jr, Chronic epileptic foci in monkeys: correlation between seizure frequency and proportion of pacemaker neurons, Epilepsia 19 (1978) [19] A.R. Wyler, J.S. Lockard, A.A. Ward Jr, C.A. Finch, Conditioned EEG desynchronization and seizure occurrence in patients, Electroencephalogr. Clin. Neurophysiol. 41 (1976) [20] J.S. Lockard, A.R. Wyler, C. Finch, K.E. Hurlburt, EEG operant conditioning in a monkey model: I. Seizure data, Epilepsia 18 (1977) [21] J.S. Lockard, A.R. Wyler, The influence of attending on seizure activity in epileptic monkeys, Epilepsia 20 (1979) [22] J. Engel Jr, L. Cahan, Potential relevance of kindling to human partial epilepsy, in: J. Wada (Ed.), Kindling 3, Raven Press, New York, 1986, pp [23] R. Goldstein, A.S. Harvey, M. Duchowny, P. Jayakar, N. Altman, T. Resnick, B. Levin, P. Dean, L. Alvarez, Preoperative clinical, EEG and imaging findings do not predict seizure outcome following temporal lobectomy in childhood, J. Child Neurol. 11 (1996) [24] C. Kilpatrick, M. Cook, Z. Matkovic, T. O Brien, A. Kaye, M. Murphy, Seizure frequency and duration of epilepsy are not risk factors for postoperative seizure outcome in patients with hippocampal sclerosis, Epilepsia 40 (1999)