Downloaded from by guest on October 15, 2014
Transcription
Downloaded from by guest on October 15, 2014
Brain (1996), 119, 989-996 The running down phenomenon in temporal lobe epilepsy Vicenta Salanova,* Frederick Andermann, Theodore Rasmussen, Andre Olivier and Luis Quesney Department of Neurology and Neurosurgery, McGill University and the Montreal Neurological Institute and Hospital, 3801 University Street, Montreal, PQ, Canada H3A 2B4 Correspondence to: Dr Salanova, Department of Neurology, Indiana University Medical Center, Riley Hospital, Room 5999C, 702 Barnhill Drive, Indianapolis, IN 46202-5200, USA * Present address: Department of Neurology, Indiana University Medical Center, Riley Hospital, Indianapolis, USA Summary areas often involving the lateral temporal and posterior temporal cortex. Other factors predictive of good outcome were: a history of febrile seizures, predominantly unilateral interictal spiking, anterior temporal localization, extent of resection of the mesial temporal structures, surgery under the age of 30 years, and the absence of habitual seizures in the immediate postoperative period. Patients with history of head trauma, encephalitis, posterior temporal localization and bitemporal spiking had a worse outcome. The frequency and types of aurae, and laterality of resection did not correlate with outcome. Keywords: epilepsy; surgery; outcome Abbreviations: ECOG = electrocorticogram Introduction Those patients whose epileptogenic areas are surgically removed, because of intractable epilepsy, are shown to follow one of three well-defined clinical courses: (i) no further seizures; (ii) continue to have seizures indefinitely; (iii) seizures which ultimately remit after a period of months to years following surgery. Rasmussen (1970) has called this later phenomenon the 'running down phenomenon'. Rasmussen (1970) postulated that in those patients manifesting the running down phenomenon, the entire epileptogenic area had not been removed, but that the remaining epileptogenic area was 'not quite autonomous', suggesting that a 'hard core, lowest threshold epileptogenic area' had been removed at surgery, and the remaining epileptogenic area was not sufficient to continue to generate seizures indefinitely. This implies that epileptogenic areas are of two types: one type must be so constituted that it © Oxford University Press 1996 autonomously generates seizure activity, while another is not fully autonomous, yet can generate seizure activity for some time but not indefinitely. The manner in which areas of differing epileptogenic potential can be generated, as postulated by Rasmussen, was defined by Morrell (1959/60, 1985); he showed that areas of secondary epileptogenesis could be recruited into activity by primary seizure foci, and that these areas of secondary epileptogenesis often become unable to generate seizures after periods ranging from months to years after removal of the primary epileptogenic areas, thus accounting for the running down phenomenon. Theoretically, the reason the running down phenomenon occurs after the removal of the pathologically damaged primary epileptogenic areas is that in these cases secondary epileptogenesis was caused by reversible pathophysiological phenomena such as transsynaptic facilitation. Downloaded from by guest on October 15, 2014 We compared 100 patients with temporal lobe epilepsy, who exhibited the running down phenomenon following temporal resections, with two groups of patients: 100 patients who became seizure-free, and 100 patients who continued to have frequent seizures following temporal resection. We found a significant correlation between prognosis and the size of the epileptogenic area as defined; patients with smaller epileptogenic areas had the best prognosis (seizure-free group). Patients exhibiting the running down phenomenon had intermediate size epileptogenic areas, while those patients who continued to have seizures had the largest epileptogenic 990 V. Salanova et al. Material and methods Initially, we had planned to analyse only those patients with the running down phenomenon; however, Dr Pierre Gloor suggested we compare those patients with two other groups of patients with medically refractory temporal lobe epilepsy treated surgically at the Montreal Neurological Institute. None of these patients had foreign tissue lesions. We used Rasmussen's classification of seizure outcome (1974). The first group (Grade O, follow-up 2-39 years), comprised 100 patients operated on during the period 1950-85 who remained seizure-free since surgery. The second group (Grade I, followup 2 ^ 0 years) comprised 100 patients operated on during the period 1950-80; these patients had seizures in the early postoperative months or years, but subsequently became and remained seizure-free (running down group). The third group consisted of 100 patients operated on during the period 1950— 85 who continued to have seizures (Grade IV, follow-up 239 years). These three groups of patients were selected from a large number of patients undergoing surgery for medically refractory temporal lobe epilepsy at the Montreal Neurological Institute during the period 1950-85. We did not include those patients in the intermediate groups who were assessed at Rasmussen's Grades II and III, with rare seizures or significant reduction in seizure frequency, because we reasoned that the comparison between diametrically opposed groups was most likely to show significant differences, and help to clarify the nature of the running down phenomenon. Rasmussen collected these data for years with a view to elucidating the nature of the running down phenomenon. He had kept a detailed follow-up on these patients through clinic visits and personal correspondence. The epilepsy surgery files were reviewed, and patients with the most complete data who also fulfilled the above criteria were chosen for the study. Most of these patients were studied prior to the era of video EEG recordings, and head CT and MRI. The epileptogenic area is defined as 'that portion of the brain which initiates the clinical seizure, whose removal or disconnection is necessary for the abolition of seizures' (Liiders et al., 1993). At the Montreal Neurological Institute all data available, including clinical semiology, neurological findings, imaging studies, neuropsychological testing, epileptiform and non-epileptiform EEG and electrocorticogram (ECOG) abnormalities are used to determine the localization of the epileptogenic area (Gloor, 1975; Quesney and Gloor, 1985). Most patients had serial surface interictal EEGs using sphenoidal electrodes; some also had ictal recordings. The neurophysiologist conducting the presurgical evaluation determined whether unilateral or bitemporal interictal spiking was present. This was a qualitative decision based on reviewing multiple recordings for several days. Most records were interpreted by Drs H. Jasper, P. Gloor and L. P. Quesney. Most patients had pneumoencephalograms, and many also had cerebral angiograms. As previously reported (Gloor, 1975; Rasmussen, 1975; Stefan et al., 1991), intraoperative electrocorticography, and cortical stimulation was used to confirm, and to determine the extent of the epileptogenic area. The ECOG tracing was obtained from a wide exposure of the fronto-centro-temporo-parietal region. In most patients, recordings were also obtained from the mesial temporal structures using two depth electrodes inserted through the second temporal gyrus. The first one was inserted 2.5-3 cm behind the temporal tip and the second one 2 cm behind the first. The degree of resection of the mesial temporal structures was assessed by the neurosurgeon at the time of surgery. Post-resection residual ECOG spiking was classified by the neurophysiologist and given a prognostic rating from I (poor prognosis, significant spiking) to 4 (good prognosis, no spiking). Identification of the mesial temporal structures in the pathological specimen could not be used to assess the degree of resection as many patients had removal of these structures by suction. Post-operative MRIs to confirm the extent of resection were not available. In most patients the pathological diagnosis consisted of different degrees of neuronal loss and gliosis; however, no further detail was possible. Most patients were treated with phenytoin, phenobarbital and primidone; however, details about anticonvulsant levels and dosages are not available. We analysed the following factors: (i) the age of seizure onset and age at time of surgery; (ii) duration of epilepsy; (iii) the types of aurae and other clinical manifestations; (iv) neurological examination; (v) EEG findings; (vi) pre-resection and post-resection ECOG; (vii) laterality of resection; (viii) extent of resection of the mesial temporal structures; (ix) Downloaded from by guest on October 15, 2014 Rasmussen (1983) observed that the running down phenomenon occurred irrespective of the cortical zone generating the seizures. In an earlier study Rasmussen (1982), compared 101 patients with temporal lobe epilepsy who became seizurefree following temporal resection with 46 patients who manifested the running down phenomenon, and suggested that 'the epileptic mechanisms essential for the production of the clinical seizures were somewhat more extensive in the 46 patient series.' The exact pathophysiology of the running down phenomenon remains unknown. Gloor (1987), suggested that 'gradations in the seizure producing potential' of epileptogenic areas exist, which may explain why removal of the area of seizure onset did not lead to complete cessation of attacks. • In order to study this phenomenon further, we compared 100 patients with temporal resections, who manifested the running down phenomenon, with an equal number of patients, who immediately became seizure-free, and with another 100 patients, who continued to have frequent habitual seizures. Unfortunately, the very nature of this phenomenon precludes any but a retrospective analysis. Surgery^ of temporal lobe epilepsy presence of seizures in the immediate postoperative period. We also analysed the duration of seizures in the early postoperative months or years for the group of patients who subsequently became seizure-free (running down phenomenon). Results Grade 0 (seizure-free group, n = 100 patients) Aetiological factors Thirty-eight patients had a history of febrile seizures under the age of 5 years; in 10 patients focal features with postictal hemiparesis were reported. Fifteen patients had a history of head trauma, 10 of difficult birth, eight had a history of meningitis, two had a history of encephalitis, one had a hypoxic event, one an arachnoid cyst, and another patient had onset of seizures during pregnancy. Twenty-four patients had no aetiological factors. Surface EEG The surface EEG showed unilateral anterior temporal interictal epileptiform discharges in 69 patients; in one maximum spiking was in the posterior temporal region. Thirty patients had bitemporal independent epileptiform discharges, but with predominance on the side of eventual resection; one of these patients also had spiking in the posterior temporal region. One patient had no epileptiform discharges. Pre-resection ECOG Sixty-five patients had spiking involving the mesial and lateral temporal structures; 27 of these patients also had spiking in the suprasylvian region. Of the remaining patients, in 24 the most active spiking was recorded from the mesial temporal structures, and in six from the lateral temporal region. Two patients had no spiking and in three the data were not available. Types of resection Fifty-seven patients had left- and 43 right-sided resections. Forty-nine patients had resection of the temporal neocortex, amygdala, and most of the hippocampus. The hippocampal resection ranged from 1 to 3 cm; most patients had resection of the anterior 1.5 cm of the hippocampus. Twelve patients had resection of the temporal neocortex, pes hippocampus, and adjacent part of the body of the hippocampus. Twentyseven had resection of the temporal neocortex, amygdala and pes hippocampus, six of the lateral temporal lobe and amygdala, and the remaining six patients had only resection of the lateral temporal region. Three patients also had small extra-temporal corticectomies, including the orbital surface of the frontal lobe (n = 2), and a small parietal corticectomy (n = 1). Post-resection ECOG Post-resection ECOG was available in 95 patients; 70 out of 95 (73.6%) showed no residual or minimal spiking (prognostic rating 3, 4) and 25 out of 95 (26.3%) showed residual spiking (prognostic rating 1, 2). Postoperative seizures Seven patients had habitual seizures in the immediate postoperative period, 15 had neighbourhood and/or generalized seizures, and in four patients the seizures were not clearly defined. The seizures occurred in the first few days following the surgery. The 'neighborhood seizures' were not the typical patients' typical seizures, and were felt to be precipitated by surgical trauma, oedema or metabolic factors. Grade I (running down phenomenon, n = 100 patients) The mean age of onset of seizures was 14.9 years (9 months to 48 years), the mean age at surgery 27.68 years (12-53 years), and the mean duration of epilepsy 12.8 years ( 1 48 years). Age at onset of seizures In 11 patients the age at onset of seizures ranged from 9 months to 4 years; in 19 from 5 to 9 years; in 26 from 10 to 14 years; in 18 from 15 to 19 years; in 18 from 20 to 29 years and in eight from 30 years to 48 years. Age at operation In 20 patients the age at surgery ranged from 10 to 19 years; in 26 from 20 to 24 years; in 18 from 25 to 29 years; in 25 from 30 to 39 years, and in 11 from 40 to 53 years. Downloaded from by guest on October 15, 2014 The mean age at onset of seizures was 10.8 years (range 3 months to 42 years). The mean age at surgery was 26.3 years (range 13-51 years), and the mean duration of epilepsy was 15 years (range 3—41 years). The neurological examination was abnormal in only two patients; one had a mild left hemiparesis and another smaller right extremities. The clinical characteristics were typical of temporal lobe epilepsy; 72% experienced aurae. The most common aurae were epigastric, present in 40 patients, followed by experiential aurae exhibited by 22 patients (deja vu, fear, complex visual or auditory hallucinations). Five patients had dizziness, four olfactory aurae, two gustatory, one conscious confusion, five aphasic aurae, one forced thinking and three were unable to describe the aura. In 28 patients no aura was mentioned. Some patients had more than one aura. 991 992 V. Salanova et al. Duration of epilepsy prior to surgery In nine patients the duration of epilepsy prior to surgery ranged from 1 to 4 years; in 31 from 5 to 9 years; in 30 from 10 to 14 years; in eight from 15 years to 19 years; in 18 from 20 to 29 years and in four 30 to 48 years. The neurological examination was abnormal in seven patients; two had left homonymous hemianopsia, one a right quadrantic visual field defect, one a mild hemiparesis, one smaller left extremities, one left-sided hyper-reflexia, and the remaining one hypotonia of the lower extremities. Forty-five patients had epigastric, and 28 experiential aurae (fear, deja vu, complex visual or auditory hallucinations). Six had cephalic sensation or dizziness sensations, four had aphasic aurae, three olfactory, four gustatory, two pilomotor, one forced thinking, one conscious confusion, one numbness, and two were unable to describe the aura. Seventeen patients had no aura. Some patients had more than one aura. Aetiological factors No. of patients Seizures in the early postoperative months/years 2-12 months 33 2 years 20 3 years 12 10 4 years 11 5-6 years 14 7-8 years Seizure-free years after early seizures 2-5 years 16 6-10 years 27 11-15 years 18 16-20 years 17 21-30 years 15 31-40 years 7 Types of resection Fifty-two patients had left- and 48 right-sided resections. Forty-eight patients had extensive resections involving the temporal cortex, amygdala, and part of the hippocampus; 14 of these patients also had small extratemporal corticectomies of orbitofrontal cortex (n = 7); frontal operculum (n = 3); post-central face area (n = 1); parietal operculum (n = 2) and central operculum (n = 1). Twenty-two patients had resection of the temporal cortex, pes hippocampus and adjacent body of the hippocampus. Eighteen had resections of the temporal cortex, amygdala and pes hippocampus; one of these also had a small removal of the orbitofrontal cortex. Nine had resections of temporal cortex and amygdala, and three had only resections of the temporal neocortex. Thus, 70% had some resection of the hippocampus, and another 18 had resection of the pes hippocampus. Scalp EEG Scalp EEG showed unilateral anterior temporal interictal epileptiform discharges in 71 patients, bitemporal independent spiking in 28, and no spiking in one patient. In eight out of 99 patients spiking also involved the suprasylvian region, and in five out of 99 the posterior temporal region. Pre-resection ECOG spiking Sixty-three patients had widespread spiking involving the temporal neocortex, and the mesial temporal structures; 29 of these patients also had spiking in the suprasylvian region, and in seven of these patients spiking involved also the posterior temporal region. Of the remaining patients, 18 had maximum spiking in the mesial temporal structures, and in two of these patients involved the posterior temporal structures. Fourteen had maximum spiking in the lateral temporal region, and in three of these spiking involved the posterior temporal region. Four patients had no spiking and the information was not available in one patient. Thus, in 13% (12 out of 95) of patients, active spiking involved the posterior temporal structures. Post-resection ECOG Twenty-eight patients had no or minimal post-resection spiking. Forty-five had residual spiking, mainly in the posterior resection margin; seven of these also had spiking in the suprasylvian region, and another five in the insula. Fifteen patients had spiking in the insula only; one of these also had spiking in Broca's area, and another in the posterior temporal region. Seven showed spiking only in the suprasylvian region, and three had no pre-resection or postresection spiking. In two patients the data were not available. Seizures in the immediate postoperative period Sixty-five patients had no seizures in the immediate postoperative period. Thirteen had habitual seizures, 18 neighbourhood seizures and in four patients the seizures were unclassified. Years of postoperative seizures Table 1 shows the number of years of postoperative seizures. The frequency of seizures ranged between one or two Downloaded from by guest on October 15, 2014 Twenty-four patients had a history of febrile seizures, 18 had a history of head trauma, and 10 had a history of difficult birth. Three patients had a history of meningitis; three had a history of brain abscess; one of toxaemia of pregnancy; one had a history of cerebral malaria; one of electroconvulsive treatment; one of CNS infection; one had a small hamartoma; one an occult vascular malformation and one had radon seeds implanted into the temporalis muscle. Thirty-five patients had no identifiable aetiological factors. Table 1 The running down phenomenon—Grade I (n =100) Surgery of temporal lobe epilepsy per year to several per month. The median duration of postoperative seizures was 2 years. A few patients had seizures up to 7-8 years before becoming seizure-free. Total number of seizure-free years after early seizures Table 1 shows the number of seizure-free years after early seizures. Follow-up range was 2-40 years. 993 these patients also had spiking in the suprasylvian region, and in 15 patients the maximum spiking was in the posterior temporal region. Of the remaining patients, in 21 the maximum spiking was in the lateral temporal region, and 11 out of 21 of these patients had maximum spiking in the lateral posterior temporal region. Eight patients had maximum spiking in the mesial temporal structures, and in two of these patients it was maximum in the posterior mesial temporal structures. Six had no or minimum spiking, and in two the data were not available. Grade IV (failure group, n = 100) Aetiological factors Thirty-four patients had a history of head trauma; 16 of difficult birth; 13 of encephalitis; eight of febrile seizures; one of meningitis; one of an infarct; one had an hypoxic episode and one was proven to have micropolygyria. Twentyfive patients had no recognizable aetiological factors. Scalp EEG Fifty-three patients had bitemporal interictal epileptiform discharges; 14 of these patients were studied with intracranial electrodes. Forty-five patients had unitemporal interictal epileptiform discharges, and one had no epileptiform discharges. In one patient the data were not available. The epileptogenic area extended to the posterior temporal region in 20 out of 98 (20.4%) of the patients. Pre-resection ECOG spiking Sixty-three patients had widespread spiking involving the lateral temporal and the mesial temporal structures; 40 of Types of resection Fifty-nine patients had left- and 51 right-sided resections. Thirty-six patients had resection of the temporal neocortex, amygdala, and part of the hippocampus; 19 had resection of the temporal neocortex, amygdala, and pes hippocampus, 16 of the temporal cortex, pes hippocampus and part of the adjacent body of the hippocampus, 18 of the temporal cortex and amygdala, and 11 of the temporal cortex only. Ten patients also had small extratemporal corticectomies most often including the parietal or frontal operculum or the orbitofrontal cortex. Post-resection ECOG The post-resection ECOG was available in 91 patients; 22 out of 91 (24.1%) had no residual or significant reduction of spiking. The remaining 69 patients had residual spiking mainly in the posterior resection margin and in the suprasylvian region, and a few patients had spiking also in the insula. Seizures in the immediate postoperative period Forty patients had seizures in the immediate postoperative period: of these, 18 were described as habitual, 13 as neighborhood and/or generalized, and in nine patients the seizures were not described in detail. Comparison between the groups (Tables 2 and 3) When we compared the three groups we found that 38% of seizure-free patients had a history of febrile seizures compared with 8% of those who were not seizure-free; this difference was highly significant statistically (P < 0.001). Twenty-four percent of patients in the running down group had a history of febrile seizures compared with 8% of those who did not become seizure-free; this was also statistically significant (P = 0.002). More patients in the group who were not seizure-free had an abnormal neurological examination. More patients in the group who were not seizure-free (Grade IV) had a history of head trauma and encephalitis, compared with the seizure-free patients (P = 0.002, and P = Downloaded from by guest on October 15, 2014 The mean age of onset of seizures was 12.8 years (1 day to 45 years), the mean age at surgery 28.01 years (range, 6-51 years), and the mean duration of epilepsy 15.4 years (2-43 years). The neurological examination was abnormal in 10 patients: three had a mild hemiparesis; two had smaller extremities contralateral to the epileptogenic area; one mild hemiparesis and mild dysphasia; one a left hemiparesis and a left upper quadrantic defect; two homonymous hemianopsias and the remaining patient had impaired contralateral two-point discrimination. Sixty-nine patients had aurae (69 out of 94, 73%), 25 had no aurae, and in six patients it was not mentioned. Twenty-eight patients had epigastric, and 24 experiential aurae, consisting predominantly of fear and deja vu, three patients exhibited complex visual and one complex auditory hallucinations; one patient exhibited elementary visual hallucinations described as 'coloured lights', and another patient manifested 'visual hallucinations'; however, no further description was available. Three had aphasic aurae, one olfactory, one gustatory, one conscious confusion, one forced thinking, two numbness in face and limbs, two chillin-the-back like an electric shock, and one could not describe the aura. 994 V. Salanova et al. Table 2 Clinical characteristics in temporal lobe epilepsy History of febrile seizures History of head trauma History of encephalitis* Age at surgery under 30 years Scalp EEG localization Anterior temporal Interictal spiking Unilateral temporal Bilateral Almost complete hippocampal resection Post-resection ECOG spiking Seizures in post-operative period Seizure-free patients (n = 100) Not seizure-free patients (n = 100) Statistical significance (P) 38 15 2 75 8 34 13 57 <0.001 0.002 0.00545 0.007 97/99 (97.9%) 78/98 (79.5%) <0.001 69/9 (69.6%) 30/99 (30.3%) 61 25/95 (26.3%) 26 45/98 (45.9%) 53/98 (54%) 52 69/91 (75.8%) 40 0.001 0.001 0.199 <0.001 0.035 The P values were obtained via 2X2 contingency tables. *Denotes Fisher's exact test was used. Table 3 Clinical characteristics in temporal lobe epilepsy Not seizure free patients (n = 100) 24 18 00 64 8 34 13 57 0.002 0.010 0.00016 0.311 94/99 (94.9%) 78/98 (79.5%) 0.001 71/99 (71.7%) 28/99 (28.2%) 70 67/98 (68.3%) 35 45/98 (45.9%) 53/98 (54%) 52 69/91 (75.8%) 40 <0.001 <0.001 0.009 0.254 0.465 Statistical significance (P) The P values were obtained via 2X2 contingency tables. *Denotes Fisher's exact test was used. 0.00545, respectively), and the running down group (P = 0.010, and P = 0.00016). These differences were statistically highly significant. Surface EEG localization showed anterior temporal epileptiform discharges in 98%, and 94% of patients in Grade O and Grade I, respectively. However, 20.4% of patients in Grade IV had maximum spiking in the posterior temporal region on surface recordings. Bitemporal independent interictal epileptiform discharges were also more frequent in those patients with poor outcome (Grade IV). These differences also reached statistical significance (Tables 2 and 3). Preresection ECOG showed larger interictal ECOG epileptogenic areas in the patients who did not become seizure-free (Grade IV), and in 28 out of 92 (30.4%) this extended into the posterior temporal region. Those patients exhibiting the running down phenomenon (Grade I) also had larger interictal ECOG epileptogenic areas, when compared with the seizure-free patients, and in 13% of the patients the interictal epileptogenic area extended into the posterior temporal region. In addition more patients in the group who did not become seizure-free had maximum spiking in the lateral temporal region when compared with the seizurefree patients. When we reviewed the extent of resection we found that more of the seizure-free, and running down patients had complete or almost complete hippocampal resection when compared with patients in Grade IV. This difference was statistically significant when comparing the running down patients with the patients who were not seizure-free patients (P = 0.009), but did not reach statistical significance when comparing the seizure-free with the not seizure-free patients (P = 0.199). In addition, fewer patients in Grade O and Grade I had ECOG post-resection residual epileptiform discharges compared with those in Grade IV. However, this was statistically significant only when comparing the seizurefree with the not seizure-free patients (P < 0.001). Other factors correlating with better outcome were surgery under the age of 30 years, and no seizures in the immediate Downloaded from by guest on October 15, 2014 History of febrile seizures History of head trauma History of encephalitis* Age at surgery under 30 years Scalp EEG localization Anterior temporal Interictal spiking Unilateral temporal Bitemporal Almost complete hippocampal resection Post-resection ECOG spiking Seizures in postoperative period Running down phenomenon (n = 100) Surgery' of temporal lobe epilepsy postoperative period, but these were statistically significant only when comparing the seizure-free with the not seizurefree patients. We found that the age of onset, duration of epilepsy, presence or absence of aurae, and laterality of resection were not significantly different between the three groups. Discussion involve a wider and wider area of cortical surface' (Morrell, 1959/60). Following removal of the primary epileptogenic area, Morrell (1959/60) observed recovery in the area of secondary epileptogenesis, which frequently stopped generating paroxysmal activity; the animals with more chronic lesions had a worse prognosis. He also studied patients with small tumours for evidence of secondary epileptogenesis, and observed that this occurred more frequently in those whose seizure disorder was of longer duration. Some patients developed independent foci, which did not resolve after surgery; but there were others, like the ones we present, who exhibited a running down period, in whom secondary epileptogenesis (intermediate stage) resolved after a period of months to years following removal of the primary lesion. We found that the running down phenomenon continued to occur even years after initial removal of the primary lesion. We also found that those patients who would be expected to have small areas of epileptogenesis, such as those whose seizures began after febrile convulsions, did, in fact, have the best prognosis, while those whose epilepsy was caused by more extensive pathology, such as encephalitis or head trauma, had the worst prognosis. This difference was statistically highly significant. The concept of secondary epileptogenesis as shown in animals and in patients with temporal and frontal lobe tumours has generally been equated with development of mirror foci in contralateral homologous cortex, which become capable of generating independent seizures (Morrell, 1985, 1989; Morrell et al., 1993; Gilmore et al., 1994). However, as Morrell et al. (1987) suggested, secondary epileptogenesis 'can also refer to the progressive extension of the primary epileptogenic zone'. Our study suggests that secondary epileptogenesis leading to larger epileptogenic areas in the same hemisphere by recruiting adjacent cortex, contributes significantly to the epileptogenic process, explaining why seizures tend to become more frequent and difficult to treat (Reynolds et al., 1983), more elaborate with the passage of time (French et al., 1993; Williamson et al., 1993), and may be related to the latent period between injury and the onset of clinical epilepsy. In view of our findings, it appears likely that the more widespread structural abnormalities produced by encephalitis and head injury produce larger areas of primary epileptogenesis, thus accounting for the poor prognosis these patients have after surgery, for it is difficult to remove all of the extensive autonomous, primary epileptogenic areas in these patients. These findings suggest that the group who were not seizure-free represents a different category of patients with temporal lobe epilepsy. Morrell (1985) found that patients with mirror foci had longer duration of epilepsy, and more than one seizure type. However, Gilmore et al. (1994) found that their patients with mirror foci did not. This indicates that longer duration of epilepsy is not always a prerequisite for the development of secondary epileptogenesis, and other factors like the severity Downloaded from by guest on October 15, 2014 Our findings suggest that the clinical course followed by a patient after surgery for epilepsy depends on the nature and extent of the interictal epileptogenic area. The patients who became immediately seizure-free were those who had the smallest interictal epileptogenic area, or who had a history of febrile seizures, which are likely to produce damage largely limited to sharply circumscribed vulnerable areas, such as the hippocampus. These patients were more likely to have unilateral anterior temporal spiking on scalp EEG, preresection ECOG spiking recorded predominantly over the anterior temporal region and from mesial temporal structures, and the least postresection ECOG spiking. Those patients who manifested the running down phenomenon had intermediate size interictal epileptogenic areas at times extending into the lateral temporal and posterior temporal region, while those who continued to have seizures had the largest epileptogenic areas with widespread spiking on scalp EEG and pre-resection ECOG, often involving the lateral and posterior temporal regions. This last group more often exhibited post-resection ECOG spiking, and a larger number of these patients had neurologic deficits prior to surgery. These results support Rasmussen's (1970) original hypothesis: that a minimal area of residual epileptogenic cortex is required to generate seizures, and that some of those who continue to have seizures will became seizure-free, because the residual epileptogenic area is 'not quite autonomous'. In those patients who manifested the running down phenomenon, Rasmussen (1970) postulated that the 'hard core, lowest threshold epileptogenic area' was removed at surgery. It is possible that this 'hard core area' represented structurally and functionally abnormal cortex, and that in these patients, the residual spiking on ECOG was caused by secondary epileptogenesis. That such a phenomenon actually occurs was shown by Morrell's clinical and experimental data (1959/60, 1985). He produced epileptogenic lesions in cats and rabbits by the application of ethyl chloride spray, and studied the development of secondary epileptogenesis, and the response of the foci of secondary epileptogenesis to removal of the primary epileptogenic lesion, and to neuronal isolation. Morrell (1959) found that 'the exact topographic relationships depended principally upon the size of the original ethyl chloride lesion, and to some extent, on the degree of paroxysmal abnormality developed at the primary site'. 'Those with actively discharging lesions >2—4 mm had a more widespread mirror focus, and 'In the course of time the paroxysmal abnormality in both hemispheres tended to 995 996 V. Salanova et al. and extent of the initial injury may play a role. Similarly, we did not find multiple-seizure types, or longer duration of epilepsy in our patients with non-tumoural lesions who exhibited the running down phenomenon. In addition to a history of febrile seizures and smaller interictal epileptogenic areas, other factors predictive of a good outcome were: unilateral interictal spiking, anterior temporal localization, and surgery under the age of 30 years. Many of these observations have been reported previously (Falconer and Serafetinides, 1963; Falconer, 1974; Davidson and Falconer, 1975; Nayel et al., 1991; Olivier, 1992; Abou-Khalil et al., 1993; French et al., 1993; Williamson et al., 1993; Salanova et al., 1994). Bengzon et al. (1968) also reported that surgery under the age of 30 years, complete hippocampal removal, no epileptiform abnormality in the post-resection ECOG and absence of seizures in the immediate postoperative period correlated with a better outcome. Acknowledgements References Abou-Khalil B, Andermann E, Andermann F, Olivier A, Quesney LF. Temporal lobe epilepsy after prolonged febrile convulsions: excellent outcome after surgical treatment. Epilepsia 1993; 43: 878-83. Morrell F. Secondary epileptogenic lesions. Epilepsia 1959/60; 1: 538-60. Morrell F. Secondary epileptogenesis in man. Arch Neurol 1985; 42: 318-35. Morrell F. Varieties of human secondary epileptogenesis. [Review]. J Clin Neurophysiol 1989; 6: 227-75. Morrell F, Wada J, Engel J Jr. Appendix III: potential relevance of kindling and secondary epileptogenesis to the consideration of surgical treatment for epilepsy. In: Engel J Jr, editor. Surgical treatment of the epilepsies. New York: Raven Press, 1987: 701-7. Morrell F, Smith MC, de Toledo-Morrell L. Secondary epileptogenesis and kindling. In: Wyllie E, ed. The treatment of epilepsy: principles and practice. Philadelphia: Lea & Febiger, 1993: 126-44. Nayel MH, Awad IA, Luders H. Extent of mesiobasal resection determines outcome after temporal lobectomy for intractable complex partial seizures. Neurosurgery 1991; 29: 55-61. Olivier A. Temporal resections in the surgical treatment of epilepsy. [Review]. Epilepsy Res Suppl 1992; 5: 175-88. Quesney LF, Gloor P. Localization of epileptic foci. In: Gotman J, Ives JR, Gloor P, editors. Long-term monitoring in epilepsy. Electroencephalogr Clin Neurophysiol, Suppl. 37. Amsterdam: Elsevier, 1985: 165-200. Rasmussen T. The neurosurgical treatment of epilepsy. In: Niedermeyer E, editor. Epilepsy. Modern Problems of Pharmacopsychiatry, Vol. 4. Basel: Karger, 1970: 306-25. Bengzon ARA, Rasmussen T, Gloor P, Dussault J, Stephens M. Prognostic factors in the surgical treatment of temporal lobe epileptics. Neurology 1968; 18: 717-31. Rasmussen T. Cortical excision for medically refractory focal epilepsy. In: Harris P, Maudsley C, editors. The natural history and management of epilepsy. Edinburgh: Churchill Livingstone, 1974; 227-39. Davidson S, Falconer MA. Outcome of surgery in 40 children with temporal-lobe epilepsy. Lancet 1975; 1: 1260-3. Rasmussen T. Surgical treatment of patients with complex partial seizures. Adv Neurol 1975; 11: 415-49. Falconer MA. Mesial temporal (Ammon's horn) sclerosis as a common cause of epilepsy. Aetiology, treatment, and prevention. Lancet 1974; 2: 767-70. Rasmussen T. Localizational aspects of epileptic seizure phenomena. In: Thompson RA, Green JR, editors. New perspectives in cerebral localization. New York: Raven Press, 1982: 177-203. Falconer MA. Serafetinides EA. A follow-up study of surgery in temporal lobe epilepsy. J Neurol Neurosurg Psychiatry 1963; 26: 154-65. Rasmussen T. Surgical treatment of complex partial seizures: Results, lessons, and problems. Epilepsia 1983; 24 Suppl 1: S65-76. French JA, Williamson PD, Thadani VM, Darcey TM, Mattson RH, Spencer SS, et al. Characteristics of medial temporal lobe epilepsy: I. Results of history and physical examination. Ann Neurol 1993; 34: 774-80. Reynolds EH, Elwes RD, Shorvon SD. Why does epilepsy become intractable? Prevention of chronic epilepsy. Lancet 1983; 2: 952—4. Salanova V, Markand ON, Worth R. Clinical characteristics and predictive factors in 98 patients with complex partial seizures treated with temporal resection. Arch Neurol 1994; 51: 1008-13. Gilmore R. Morris H, Van Ness C. Gilmore-Pollak W, Estes M. Mirror focus: function of seizure frequency and influence on outcome after surgery. Epilepsia 1994; 35: 258-63. Stefan H, Quesney LF, Abou-Khalil B, Olivier A. Electrocorticography in temporal lobe epilepsy surgery. Acta Neurol Scand 1991; 83: 65-72. Gloor P. Contributions of electroencephalography and electrocorticography to the neurosurgical treatment of the epilepsies. [Review]. Adv Neurol 1975: 8: 59-105. Williamson PD, French JA, Thadani VM, Kim JH, Novelly RA, Spencer SS, et al. Characteristics of medial temporal lobe epilepsy: II. Interictal and ictal scalp electroencephalography, neuropsychological testing, neuroimaging, surgical results, and pathology. Ann Neurol 1993: 34: 781-7. Gloor P. Commentary: approaches to localization of the epileptogenic lesion. In: Engel J Jr. editor. Surgical treatment of the epilepsies. 2nd ed. New York: Raven Press, 1987: 97-100. Liiders HO, Engel J Jr. Munari C. General principles. In: Engel J Received June 1, 1995. Revised December 18. 1995. Accepted February 13, 1996 Downloaded from by guest on October 15, 2014 We wish to thank Dr Pierre Gloor who contributed to the ideas and suggested the methods by which the subject should be approached and Mrs Martha Stanley for secretarial assistance in preparation of the manuscript. Jr, editor. Surgical treatment of the epilepsies. 2nd ed. New York: Raven Press, 1993: 137-53.