^Australian Computing and Communications Institute, Clinical Neuroscience, St Vincent's Hospital, Victoria

Transcription

1 Brain (996), 9, 4 Temporal lobe epilepsy caused by mesial temporal sclerosis and temporal neocortical lesions A clinical and electroencephalographic study of 46 pathologically proven cases Terence J. O'Brien, Christine Kilpatrick, Vanessa Murrie, Simon Vogrin, Kevin Morris and Mark J. Cook Royal Melbourne Hospital, St Vincent's Hospital, and the Correspondence to: Dr Mark J. Cook, Department of ^Australian Computing and Communications Institute, Clinical Neuroscience, St Vincent's Hospital, Victoria Melbourne, Australia Parade, Fitzroy, 065 Victoria, Australia Summary This study aims to determine whether there are important clinicoelectrical differences between patients with temporal lobe epilepsy (TLE) secondary to mesial temporal sclerosis (MTS) and those with TLE secondary to a discrete temporal neocortical lesion (NL). The case histories, interictal EEG, seizure semiology, ictal EEG and postoperative outcome of 46 pathologically proven patients ( MTS and 5 NL) were compared. A history of febrile convulsions (FC) was more common in MTS patients (5% versus 6%, P < 0.05), as was a history of a significant cerebral event at <4 years of age (% versus 0%, P < 0.05). There were no statistically significant differences in the incidence or nature of auras. No statistically significant differences between the groups were found in the interictaleeg. With ictal semiology dystonic posturing occurred more frequently in MTS patients (mean 5% versus 6%, P < 0.05). Facial grimacing/ twitching occurred earlier in the seizures of NL patients (median 9 s versus 5 s, P < 0.05). There was an increased frequency offast rhythmic sharp waves (>4 Hz) in the ictal EEG of MTS patients (mean % versus 60%, P = 0.05). The patients with NL developed bilateral ictal EEG changes more often (mean 55% versus 6%, P < 0.05) and more rapidly (mean s versus 74 s, P < 0.005). The onset of ictal EEG seizure activity was bilateral more often in patients with NL (0% versus 4%, P < 0.005). There were no significant differences between the two groups for any of the videoeeg features, in terms of whether or not the feature occurred at least once in an individual patient. There was a tendency for MTS patients to have a higher seizurefree postsurgical outcome (7% versus 60%, P = 0.057). However, all the NL patients who were not free of seizures had had an incomplete lesion resection. We conclude that there are a number of clinicoelectrical differences between patients with mesial TLE (MTLE) and patients with neocortical TLE (NCTLE), but that none of these are sufficient to allow a distinction to be made in an individual patient. Keywords: temporal lobe epilepsy; mesial temporal sclerosis; neocortical foreign tissue lesions; mesial temporal lobe epilepsy; neocortical temporal lobe epilepsy Abbreviations: FC = febrile convulsions; FDG = [ F]fluorodeoxyglucose; MTLE = mesial temporal lobe epilepsy; MTS = mesial temporal sclerosis; NCTLE = neocortical temporal lobe epilepsy; NL = neocortical lesion; PPV = positive predictive value; TLE= temporal lobe epilepsy Downloaded from by guest on 07 September 0 Introduction A syndrome of MTLE has been proposed that can be readily distinguished from neocortical temporal lobe epilepsy (NCTLE) (Commission on Classification and Terminology of the ILAE, 99; Engel, 99; Wieser et al., 99). For such a subdivision to be clinically useful the two syndromes should differ in at least some of the following: historical Oxford University Press 996 features; interictal and ictal EEG; seizure semiology; neuroimaging; natural history and response to treatment; aetiology and underlying pathology. On pathological examination, MTLE is characterized by the specific finding of MTS with marked neuronal loss in the CA region and relative sparing of CA (Wieser et al., 99). If hippocampal

2 4 T. J. O'Brien et al. atrophy or a discrete NL is present, MTLE and NCTLE can be readily distinguished on high resolution volumetric MRI (Kuzniecky et al., 99). Scanning with [ F]fluorodeoxyglucose (FDG)PET shows different patterns of glucose metabolism in patients with EEGdefined mesial temporal onset seizures compared with those with lateral neocortical onset seizures (Hajek et al., 99). Whether the two subtypes are distinguishable on clinicoelectrical grounds is much less certain. Many studies of patients with presumed MTLE have reported that this syndrome is characterized by relatively homogeneous clinical features, typical EEG changes and a good prognosis following temporal lobectomy (see review by Wieser et al., 99). However, most of these studies have not directly compared MTLE patients and those with NCTLE. The few comparative studies published have found some relatively minor differences between the groups, but none have compared all the above features (Mihara et al., 99; Ebner, 994; Saygi et al., 994; Burgerman et al., 995). Furthermore, with the exception of the study by Saygi et al. (994) which assessed auras and the seizure semiology only, these studies have all divided the patients into subtypes on the basis of presurgical depth electrode studies and not on the presence of a localized pathological lesion. The NCTLE groups in these studies included patients whose seizures had begun simultaneously in the mesial and the temporal neocortex which therefore raises the possibility that some were falsely localized as a result of sampling error (Gloor, 97). Freedom from postoperative seizures is not by itself proof of correct intratemporal localization, as most patients in these studies underwent an anterior temporal lobectomy with removal of both mesial and neocortical structures. In contrast, if an isolated NL is found in a patient with TLE, it is highly likely that the seizure onset zone is closely related to this lesion (Gloor, 97; Walczak, 995). Conversely MTLE is almost invariably associated with MTS (Babb and Brown, 97). We have therefore compared the historical features, interictal EEG, seizure semiology, ictal EEG and surgical prognosis in a series of patients who had intractable TLE and pathologically proven MTS or an isolated NL. The aim of the study was to determine whether there is a useful clinicoelectrical difference between patients with MTLE and those with NCTLE. No previous study has compared all these features in a single series. Methods Patients Fortysix patients with TLE who had undergone epilepsy surgery for intractable seizures were studied. Preoperative assessment was done using an exclusively noninvasive protocol (volumetric MRI, interictal EEG, videoeeg, ictal and interictal single photon emission computed tomography and FDGPET) (Kilpatrick et al. 995). Selection of the site and side of the surgery was primarily based on the detection of a focal temporal abnormality on volumetric MRI (visual inspection and hippocampal volume measurements) provided the videoeeg findings confirmed partial seizures with an EEG onset that was not consistently contralateral (a nonlateralized ictal EEG was not a contraindication). Thirtyone patients had pathologically proven MTS, and all had undergone a standard anterior temporal lobectomy ( cm lateral resection) with en block removal of the mesial temporal structures. None had evidence of hippocampal sclerosis in the contralateral temporal lobe as judged by volumetric MRI hippocampal measurements and inspection for T signal changes. Fifteen patients had an isolated NL (lateral and anterior temporal neocortex not involving the mesial structures) and all had undergone lesionectomies only. The pathological diagnoses for NL patients were; dysembryoblastic neuroepithelial tumour (n = 6), cortical dysplasia (n = 4), low grade astrocytoma (n = ), high grade astrocytoma (n = ), cavernous haemangioma (n = ). All the NL patients had MRI hippocampal volume measurements that were within the range established on normal controls (i.e. hippocampal ratio >0.95) (Cook et al. 99), and coronal T weighted scans showed no signal abnormalities in the mesial temporal structures. Two patients with temporal NL who had epilepsy surgery during the period of this study were excluded from analysis (one had a tumour involving the mesial temporal structures and the other had temporal encephalomalacia with coexistent hippocampal atrophy). Historical evaluation The patients' medical records were retrospectively reviewed for the features listed in Tables and. If the records were incomplete a direct inquiry of the patient, relatives and/or referring medical practitioner was undertaken. Fisher's exact test (twotailed) was used to test for differences in the nominal variables, e.g. a history of FC. MannWhitney U test (twotailed) was used to compare the groups with regard to quantifiable variables (e.g. age of onset of FC). Interictal EEG A single premonitoring, h, 6channel interictal EEG recording was reviewed in all patients when available. The EEG was analysed according to the features listed in Table. When an abnormality was present it was subclassified by its side and site. Differences between the groups for the above variables were tested using the Fisher's exact test (twotailed). VideoEEG All patients were monitored with continuous. 4 h video EEG using 6 scalp electrodes. Monitoring was continued until at least three seizures had been recorded. Medications were reduced as necessary to induce seizures. A seizure was Downloaded from by guest on 07 September 0

3 Table Demographic and historical data from 46 patients with TLE Mesial versus neocortical temporal lobe epilepsy 5 Age in years: median (range) Sex (M/F) Side of surgery (R/L) Febrile convulsions (FC) Complicated or prolonged FC Time of FC in months: median (range) Cerebral event before the age of 4 years Onset of AS (years): median (range) Duration of AS (years): median (range) Seizure frequency per yean median (range) GS (% of all seizures): median (range) No. of medications: median (range) Relationship to menstrual cycle Psychiatric history Family history of epilepsy MTS (n = ) 6(757) 5/6 6/5 (5.%) 4 (.9%) 4 (0.56) 7 (.6%) (0.54) 4 (^) (6095).5 (05) (4) 7 (4.7%) (9.7%) 5 (6.%) NL (n = 5) (65) 9/6 7/ 4 (6.7%) 0 (0.0%) 0 (60) 0 (0.0%) 6 d9) 5 (4) 7 (5095),.5 (00). 5 (5) (.%) (0.0%) 4 (6.7%) TLE = temporal lobe epilepsy; MTS = mesial temporal sclerosis; NL = neocortical lesion; FC = febrile convulsions; AS = afebrile seizures; GS = generalized seizures. Table Auras: localizing and materializing value Patients with auras (% ) Psychic auras (% of auras) Autonomic auras (% of auras) Hallucination (% of auras) Olfactory/gustatory Auditory Visual Motor auras (% of auras) Sensory auras (% of auras) Other auras (% of auras) Right MTS (n = 6) NL(n = 7) 4 (%) 6 (4%) 7 (50%) (4%) (%) 4 (9%) (%) 4 (57%) (60%) (0%) (0%) (0%) Left MTS («= 5) NL (n = ) (0%) 5 (%) (6%) (5%) (5%) (%) 4(%) Table Interictal EEG results of 46 patients with TLE Abnormal EEG Contralateral Epileptiform changes Contralateral Anterior temporal Mid temporal Posterior temporal Focal slow waves Contralateral 7 (4%) (9%) (4%) I (4%) (4%) (4%) Total P value <0.05 <0.05 MTS (n = ) NL (n = 5) 6 (4%) (4%) 5 (4%) 4 (5%) (9%) (9%) (7%) MTS (n = 7) NL(n = 5) Total (n = 4) 5 (57%) 0 0(7%) 0 6 (4%) 5 (7%) 9 (60%) 6 6 0(66%) 6 (6%) 5 9 (45%) (50%) 4 4 (7%) 6 (55%) (7%) (%) (9%) (9%) 4 (6%) Downloaded from by guest on 07 September 0 /contralateral refers to side of pathology; anterior temporal = phase reversal at F7/, midtemporal = phase reversal at T/4, posterior temporal = phase reversal at T5/6. selected for analysis only if the patient was visible throughout the event. The semiology and ictal EEG's were repeatedly reviewed for the occurrence of the features listed in Table 4 by one author (T.J.O'B.) who was blinded to the type and location of the lesion. The side, site and time of onset, maximum spread (including time to become bilateral) and

4 6 T. J. O'Brien et al. Table 4 VideoEEG results for 59 recorded seizures in 46 patients with intractable TLE Mesial temporal sclerosis (n = ) Neocortical lesions (n = 46) No of patients* (n = ) % Seizures per patienl: mean Onset time: median (range; in s) Duration: median (range; in s) No of patients* (n = 5) % Seizures per patient mean Onset time median (range; in s) Duration median (range, in s) Semiology Seizure duration Motionless stare Restlessness Oral automatisms Motor automatisms Dystomc posturing Unilateral arm paresis Facial clonus/gnmace Limb clonus Nonforced HT Ictal speech Generalization HT on generalization Postictal dysphasia Ictal EEG Seizure activity* Sharp waves (>4 Hz) seizure activity* Postictal slowing 9 (94%) 4 (45%) 5 (0%) 6 (%) (74%) (74%) 5 (6%) 6 (5%) 0 (65%) 7 (%) (4%) 9 (9%) 4(%) 9 (94%) (90%) 4 (45%) 7 (54%) " * 6" 44 5 (040) 9 (05) (06) 4(7) (56) 7 (79) 5 (60)" 4 (709) 6 (%) (6) 0 (4690) 74 (475) (559) 5 (40) 7(77)*" 94(567) (5) (404) 7 (54) 4(94) (9) (64) 6 (56) 4 (^49) (476) (444) 5 (776) 4 (976) 05 (450) 79(965) 7 (5) 65 (55) 65 (056) _ (7%) 7 (47%) (7%) (0%) 9(60%) 0 (67%) 5 (%) 7 (47%) 9(60%) (0%) 5 (%) 4 (7%) (0%) 4 (9%) (7%) (7%) 7 (47%) " II 7 60* 55" 0(0) (0) (06) 4 (46) 0 (76) (4) 9()" (65) 4 (557) 5 (0) 5 (577) 6(54) 4 (46) (54) (6)*" 7(6550) 9 (470) 0(597) (59) 5 () 9 (4_6) 9 (7) 5 (06) 47 ( 49) (54) (64) 6(5) 5 (0) 00(60600) 6 (4546) 7 (650) 56(4) 5 (600) HT = head turning. + Number in whom the feature was observed at least once; ^rhythmic sharp waves or spike wave complexes; 'P = 0.05, "P < 0.05: '"P < duration of each feature on semiology and ictal EEG was determined. These findings were then reviewed by a second author (C.K. or M.C.) and if there was disagreement the recording was reviewed by two authors simultaneously and a consensus obtained. Clinical ictal onset was defined as the time when the patient indicated a warning (e.g. pushing a button), or from the first evidence of abnormal movement or altered responsiveness (if no warning). The end of the seizure was defined as the cessation of abnormal movements or the resumption of normal responsiveness (if no abnormal movements occurred). The EEG seizure activity was defined as rhythmic sharp waves or spike wave complexes. Differences between the two groups, in the percentage of seizures in a patient in which each videoeeg feature occurred, was tested using the MannWhitney U test (twotailed) (Table 4). In addition the groups were compared for differences in the occurrence of each feature on at least one occasion in an individual using Fisher's exact test (twotailed) (Table 4). The time of onset and duration of each feature was averaged for the seizures in which it occurred in an individual, and the groups were then compared using the MannWhitney U test (twotailed) (Table 4). The nature of the first feature to appear on semiology and the ictal EEG, and the site of onset of the EEG change, were compared using the x test (Table 5). The side of each videoeeg feature that has been proposed to have value in lateralizing the seizure onset was checked using Fisher's exact test (twotailed), and the positive predictive value (PPV) for correct lateralization of each feature was calculated (Table 6). Table 5 Initial clinical and EEG features on video for 59 seizures in 46 patients with TLE Initial clinical feature Motionless staring Automatisms Restlessness Head turning Dystonic posturing Vocalization Initial EEG change Sharp waves (>4 Hz) Sharp waves (<4 Hz) Slow waves Obscured by Artefact No change Site of initial EEG change Anterior temporal Mid temporal Posterior temporal Diffuse temporal Diffuse hemisphere Bitemporal MTS {n = ) 59 (5%) 6 (%) 0(%) 7 (6%) (%) (n = ) 4 (4%) 40 (5%) 9(%) (%) (%) (7! = 97) 9(0%) (%) (%) 7 (%) 5 (6%) (%) NL (n = 46) 7 (59%) 6(%) 7 (5%) 4(9%) (4%) (7i = 46) (9%) 7 (7%) 4(9%) (4.%) 5 (%) (n = ) 5 (%) 5 (%) 4(%) 5 (40%) 6(6%) (%) MTS = patients with mesial temporal sclerosis; NL = patients with neocortical lesions. Surgical outcome Postoperatively, patients were reviewed at least every months for the duration of the followup period. They, and their families, were questioned closely about the occurrence of auras or seizures. Outcome was classified according to a Downloaded from by guest on 07 September 0

5 Mesial versus neocortical temporal lobe epilepsy 7 Table 6 Materializing and predictive value of features on videoeeg from 59 seizures in 46 patients with TLE MTS (n = ) NL (n = 46) Total PPV(%) Contralat. Contralat Contralat. Semiology: Dystonic posturing Unilateral arm paresis Early HT HT on generalization Facial clonus/grimace Clonic limb movement Ictal speech Postictal dysphasia Ictal EEG: Seizure activity Sharp waves (>4 Hz) Postictal slowing 4(7%) 0(%) 50 (5%) 0 (0%) 4(5%) (%) (6%) 9 (00%) 9 (95%)" 4 (97%)" 45 (%) 47 (7%) 45 (%) 9(5%) (00%) 9 (70%) (6%) 6 (94%) d%) 0 (0%) 9 (5%) 4(5%) 0 (9%) 4 (4%)" (%)* 6(%) (7%) 4(9%) (%) (5%) (75%) 5 (00%) (7%)" 5 (%)" (0%) (7%) 7 (%) (%) (75%) 4 (57%) 5 (50%) (5%) (%) (4%) 5 (50%) (0%)" 5 (7%)' (0%) 7 (0%) 4(%) (7%) (6%) 4(%) (%) 4 (9%) 4(00%) (90%) 09 (9%) 5 (00%) HT = head turning; PPV = positive predictive value for the feature in predicting the side of seizure onset,. *P < "P < four point scale, i.e. seizure free (with or without auras), >90% improvement in seizure frequency, <90%, but still worthwhile, improvement in seizure frequency and no worthwhile improvement in seizure frequency. The seizure free outcome in the two groups was compared using Fisher's exact test (twotailed). Results Historical details The historical details are summarized in Table. The patients were well matched for age, sex and side of lesion. Patients with MTS more commonly had a history of FC in infancy (P < 0.05) and also more commonly had a history of a major cerebral event earlier before the age of 4 years; four had prolonged febrile convulsions, one had a major head injury and two had encephalitis (P < 0.05). In these latter two instances, encephalitis had been diagnosed in childhood (when 5 and 7 years old, respectively), when presenting with fever, convulsions, raised CSF white cells, and prolonged alteration in conscious state. Neither had imaging performed in the acute phase of the illness, and no viral agent was identified. Other than seizures, there were no neurological sequelae. No statistical significant differences in the incidence or nature of the auras were found and they were not found to be of significant lateralizing value (Table ). Interictal EEG Interictal EEG recordings were available for analysis in 4 patients (7 MTS and 5 NL) and the results are summarized in Table. In none of the categories did the difference between the groups achieve statistical significance, but there was a strong tendency for the patients with NL to have a higher incidence of epileptiform changes (P = 0.5), and lateralized slow waves (P = 0.). 55 (77%) 6 (%) (%) 6 (94%) (6%) (64%) 7 (%) (%) 9 (%) 7(%) 5 (4%) (9%) (7%) (%) VideoEEG A total of 59 seizures were analysed ( in the MTS group and 46 in the NL group) and the results are summarized in Tables 46. There were no significant differences between the groups in the number of seizures analysed per patient. On semiology, dystonic posturing occurred in a significantly higher percentage of seizures in the MTS patients (P < 0.05) and facial grimacing/ clonus occurred earlier in the NL patients (P < 0.05) (Table 4). On the ictal EEG, bilateral seizure activity occurred in a higher percentage of seizures (P < 0.05) and earlier (P < 0.005) in the NL group. There was also a strong tendency for fast rhythmic sharp waves (>4 Hz) to occur more often in the MTS group (P = 0.05) (Table 4). There were no statistically significant differences between the two groups in the incidence of a videoeeg feature occurring in at least one seizure in the patient (Table 4). No significant differences were found between the two groups in the initial clinical features or ictaleeg change (Table 5). The groups also did not differ significantly in the lateralization of the videoeeg features (Table 6). The overall PPV of these features for lateralization of the seizure focus is given in Table 6. Surgical outcome Mean followup was 0 months (range months) and did not differ significantly between the groups. All patients had at least a worthwhile improvement in seizure frequency. At the time of the last followup in the MTS group 7 patients (7%) were seizure free (two suffered rare isolated auras) and four patients had a >90% improvement in seizure frequency. In comparison, only nine (60%) of the NL patients have remained seizure free postoperatively, but all are free of auras (P = 0.057). Five have a >90% improvement, and one has a <90%, but still worthwhile, improvement in seizure frequency. All of the NL patients with continuing seizures Downloaded from by guest on 07 September 0

6 T. J. O'Brien et al. had incomplete resections of their lesions, as judged by pathological examination and postoperative MRI. Discussion Past history There are only two other comparative studies of historical features in MTLE and NCTLE. Saygi et al. (994) compared MTS patients with NL patients and found that a history of both FC and intracranial infections were more common with MTS, and that head trauma was more common with NL. Burgerman et al. (995) found that a history of FC was significantly more common in patients with depth EEGdefined mesial temporal seizure onset, and there was a strong tendency for an increased incidence of a major cerebral insult before the age of years, compared with patients with neocortical onset seizures. The aetiology of MTS is still uncertain but the most popular theory is that it results from a cerebral insult in early childhood, especially prolonged FC (Engel, 99). Our finding, that a history of both FC and a major early cerebral event before the age of 4 years was significantly more common in MTS patients (Table ), is consistent with this hypothesis. However, 0 MTS patients (.9%) had no obvious causal event, and four NL patients (6.7%) had a history of FC. An alternate explanation is that the presence of a preexisting temporal lobe lesion, either NL or MTS, lowers the threshold for febrile convulsions. We found no significant differences between the groups with regard to seizure frequency, proportion of generalized convulsions, number of medications, precipitating factors, a family history of epilepsy or psychiatric history. Burgerman et al. (995) found a nonsignificant tendency for a higher frequency of seizures in MTLE (P = 0.4), but found no difference in the duration of afebrile seizures or in the presence of a family history of epilepsy. Auras Auras consisting of psychic symptoms or auditory and visual hallucinations have been proposed as reliable indicators of lateral temporal seizure onset (Penfield, 96; Williamson et al., 97). Conversely, autonomic auras (especially epigastric) have been considered typical of MTLE and to be rare with NCTLE (Williamson et al, 97; Mihara, 99; Ebner, 994). While we found no statistically significant differences in the incidence or nature of the auras between the groups, some interesting trends were present (Table ). Olfactory and gustatory hallucinations only occurred in the MTS group, and autonomic auras were more common in these MTS patients (P = 0.09). Visual and auditory hallucinations occurred only in the NL patients. Ebner (994) found no significant difference in the incidence of auras overall between NL and MTLE, although epigastric symptoms were more common in the MTLE group, while occurring in 50% of the NL group, and visual/auditory hallucinations were more common in the NL patients. Burgerman etal.(\ 995) found a tendency for a higher incidence of auras in MTLE (P = 0.07), and Saygi etal. (994) found no significant differences between the auras of patients with MTS and NL. Mihara et al. (99) correlated the type of aura with the location of seizure discharges on the depth EEG in 5 patients with amygdalohippocampal onset seizures and 5 patients with lateral temporal onset seizures. They noted that the autonomic symptoms were most frequently reported when the seizure discharges involved the amygdala/hippocampal region (either primarily or from secondary spread). Conversely, psychic symptoms were usually reported when the discharge involved lateral temporal neocortex. However, when the types of auras were compared between the groups of patients there was only a modest tendency for autonomic auras to be more common in mesial onset seizures, and for psychic auras to be more common in lateral onset seizures. It seems likely that, with the possible exception of auditory or visual symptoms, the nature of the aura is not reliable in differentiating between MTLE and NCTLE. Some recent cortical stimulation studies have suggested that much of the temporal neocortex is symptomatically silent {see review by Ebner, 994), and therefore seizures that begin in these areas would not produce an aura until they spread to a symptomatic region (So, 99). There is evidence that reciprocal connections between the mesial and lateral temporal cortex have to be activated before the patient reports an aura (Gloor, 990; Blume et al., 99). Therefore the type of aura in TLE is probably more indicative of the pattern of seizure spread than of the site of seizure onset. Interictal EEG Although no significant differences were found between the two groups in the interictal EEG, there was a strong tendency for NL patients to have an increased incidence of epileptiform changes and lateralized slow waves (which were more likely to be ipsilateral) (Table ). This is likely to reflect the fact that epileptiform activity confined to the mesial temporal region is often silent to standard scalpeeg electrodes, whereas activity in the lateral temporal neocortex is transmitted well to the scalp (Pacia and Ebersole, 99; Ebner and Hoppe, 995). There are no previous studies that directly compare routine h surfaceeegs of awake patients with MTLE with those from patients with NCTLE, but Burgerman et al. (995) analysed the interictal sphenoidal/ scalp EEG recordings during continuous 4 h videoeeg monitoring and found no significant differences in the frequency, laterality or location of epileptiform discharges or focal slowing. Engel (99) stated that in patients with MTLE, unilateral temporal interictal spikes lateralize the epileptogenic side correctly in ~5% of cases. In support of this statement we found that in MTS patients with lateralized epileptiform discharges, 0% of them were ipsilateral to epileptogenic lesion. A similar accuracy, however, was seen in NL and this Downloaded from by guest on 07 September 0

7 therefore seems to be a characteristic of TLE in general, rather than one of MTLE specifically. Anterior temporal cortex is the most common site for the phase reversal of the epileptiform activity in studies of interictal EEG in patients with MTS (Williamson et al., 99; Ebner and Hoppe, 995). We found, however, that the site of the interictal epileptiform abnormalities was not useful in distinguishing between MTS and NL (Table ). Semiology Dystonic posturing occurred significantly more frequently in the seizures of patients with MTS, and facial clonus or grimacing occurred earlier in patients with NL (Table 4). These findings may indicate that the mesial temporal seizures preferentially propagate to the basal ganglia, while neocortical onset seizures spread more rapidly to cortical motor areas. However, there were no significant differences for either of these features when the groups were compared for feature occurrence in * of the patients seizures (Table 4). This indicates that in an individual patient they are not reliable distinguishing features. Mihara et al. (99) found that dystonic posturing and oroalimentary automatisms occurred more frequently in mesial temporal onset seizures. Saygi et al. (994) found no significant differences between patients with MTS and NL in the frequency that any particular feature occurred. They did, however, find differences in the timing of appearance of a number of features; contralateral head turning (within the first 5 s only in the NL group), ipsilateral hand automatisms (in the first 60 s more often with MTS), leg automatisms and oral automatisms (more common between 5 and 60 s in the NL and MTS patients, respectively). However, with the exception of early contralateral head turning, which occurred in only four of the MTS patients, none of these features occurred exclusively in one group. Furthermore the division of the seizures into three phases for the analysis of the timing in this study was arbitrary, and its clinical utility is uncertain. Three other reports comparing the semiology in MTLE and NCTLE have been published in abstract form: Foldvary et al. (994) found dystonic posturing, automatisms and large body movements more common in MTLE; GilNagel and Risinger (99) found early oral automatisms and early motor involvement were more common is MTLE and NCTLE, respectively; and Ho et al. (994) found no significant differences. The first manifestation of a seizure may be an important clue to the site of seizure onset. However, we found no significant differences between the groups with respect to motionless staring as the initial clinical feature, which occurred in >50% of patients in both groups (see Table 5). Our results are consistent with the findings of Mihara et al. (99) which suggested that motionless staring was related to the presence of bilateral mesial temporal lobe involvement and not to the site of seizure origin. In all the potential lateralizing features on semiology that were studied we found a PPV of >0% (Table 6). Postictal Mesial versus neocortical temporal lobe epilepsy 9 dysphasia (dominant hemisphere), clonic limb movements (contralateral) and tonic head turning prior to generalization (contralateral) were the most reliable features, but they occurred in a minority of seizures. Contralateral dystonic posturing and ipsilateral early head turning had a relatively high PPV and occurred in the majority of seizures and therefore may be of more clinical utility. Ictal speech (nondominant) had the lowest PPV. Mihara et al. (99) also found that contralateral dystonic posturing and contralateral somatomotor manifestations had a high specificity for lateralizing the seizure onset, and that ictal speech had relatively poor specificity. Seizure semiology is reflective of both the site of seizure onset and the areas of spread of the discharges. Seizure propagation in patients with MTS has been shown to be very variable, both between patients and within the same patient (Gates and Gumnit, 97; Spencer, 9). Furthermore, extensive connections between the temporal neocortex and the mesial temporal structures mean that seizures can rapidly spread between these sites (Gates and Gumnit, 97; Walczak, 995). Rapid propagation of seizure activity from extratemporal neocortex lesions to the mesial temporal structures may occur producing indistinguishable seizure semiology and EEG changes (Engel, 99; Fish etal., 99; Williamson, 99a, b). It is therefore not surprising that, whilst there are some differences between the semiology of MTLE and NCTLE when comparing groups, it is not reliable in differentiating individual patients. This is consistent with the conclusion of Kotagal, et al. (9) who found no evidence for discrete subtypes using cluster analysis of the auras and seizure semiology (including sequences) of patients with TLE. Our data strongly suggest that semiology is much more useful in lateralizing the side of origin of the seizures than localising its site of onset within the temporal lobes. Scalp ictal EEG Fast rhythmic sharp waves and postictal slowing occurred more often in patients with MTS than those with NL, while bilateral seizure activity occurred more often and earlier in patients with NL (Table 4). There are two previous preliminary reports comparing scalp ictaleeg in patients with MTLE and in those with NCTLE (Pacia and Ebersole, 99; Walczak et al., 994). As with our study, both found fast frequency activity in the scalp EEG more commonly with MTLE, and Pacia and Ebersole et al. (99) also noted that its appearance correlated with the spread of seizure activity into the lateral neocortex on simultaneous depth electrode recording. Other depth electrode studies of patients with MTS have found that this spread is associated with an increase in the frequency of the activity (Ebner and Hoppe, 995). The finding that the scalpeeg changes become bitemporal earlier in patients with NL is likely to reflect earlier spread in these patients to the contralateral temporal neocortex. The presence of lateralized onset seizure activity, fast rhythmic sharp waves and lateralized postictal slowing were Downloaded from by guest on 07 September 0

8 40 T. J. O'Brien et al. all highly predictive for lateralization of the side of the lesion (Table 6). Previous studies assessing the lateralizing value of ictal scalpeeg in MTLE have found a somewhat lower degree of accuracy (5%) (Risinger et al., 99; Engel, 99). The high lateralization accuracy of scalp ictaleeg in our study may partially represent selection bias, but it is consistent with a recent study of 7 seizures in 4 patients with MTS which found that only % of seizures were lateralized incorrectly (Ebner and Hoppe, 995). The initial EEG change was localized to the temporal lobe in a high proportion of seizures in both MTS and NL patients (Table 5). A significantly higher proportion of seizures in NL patients had bilateral onset of both seizure activity and fast rhythmic sharp waves (Table 6), presumably reflecting rapid spread to the contralateral temporal neocortex. Surgical results After surgery, more MTS patients were seizure free than NL patients. However, all the NL patients with persistent seizures had an incomplete lesion excision. In some cases of incomplete resection the surgeon had judged intraoperatively that complete resection was hazardous or, more frequently, the excision was felt complete on macroscopic criteria, but evidence of residual lesion was noted at postoperative imaging, or pathologically. This is probably a more common occurrence than is generally appreciated. Such residual lesions are known to be associated with worse outcome, but they are not always avoidable. Spencer et al. (990) reported a better outcome in patients with mesiotemporal onset seizures than those with lateral temporal onset (9% versus 0%), but the lateral temporal resections in this study were relatively small (.0.5 cm) and therefore the epileptogenic region may have been missed or incompletely excised. Hajek et al. (99) also reported a higher proportion of seizure free patients in the mesial temporal onset group compared with lateral temporal onset group (4% versus 0%), but none of these patients had resection of the lateral temporal neocortex (all had only amygdalohippocampectomies). In contrast, Burgerman et al. (995) employing larger lateral temporal resections (44.5 cm left or 55.5 cm right) found no significant difference in the seizure free outcomes (0% for MTLE versus 7% for NCTLE). Conclusion Overall we identified some differences between patients with MTS and patients with NL with regard to historical features, interictal scalpeeg, seizure semiology and ictal scalpeeg. This is not unexpected given the differences in the nature of the underlying pathological lesion and the site of seizure onset. However, none of these differences were sufficient to enable a distinction between the groups to be used on an individual basis. Seizure semiology, and interictal and ictal scalpeeg were much more predictive of seizure lateralization than of intratemporal localization. The differing postoperative seizure free outcome was dependant on the success in completely excising the epileptogenic MRI lesion. In view of the findings of this study we would question the practical usefulness in an individual patient of attempting to make the distinction between MTLE and NCTLE on clinicoelectrical grounds alone. It is acknowledged, however, that neocortical seizure onset can occur in the absence of a visible structural lesion on MRI (Burgerman et al., 995) and it is possible that the findings of this study may not extend to these, apparently lesionfree, cases of NCTLE. Acknowledgements The authors are grateful to Dr J. Hopper, Department of Public Heath and Community Medicine, The University of Melbourne, for his advice regarding the statistical methods. T.J.O'B. was supported by a Reckitt and Colman Neurology Research Fellowship in Epilepsy. M.J.C. was partially supported by a grant from the Epilepsy Foundation of Victoria. References Babb TL, Brown WJ. Pathological findings in epilepsy. In: Engel J Jr. editor. Surgical treatment of the epilepsies. New York: Raven Press, 97: 540. Blume WT, Girvin JP, Stenerson P. Temporal neocortical role in ictal experiential phenomena. Ann Neurol 99; : 057. Burgerman RS, Sperling MR, French JA, Saykin AJ, O'Connor MJ. Comparison of mesial versus neocortical onset temporal lobe seizures: neurodiagnostic findings and surgical outcome. Epilepsia 995; 6: Commission on Classification and Terminology of the International League Against Epilepsy. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 99; 0: 999. Cook MJ, Fish DR, Shorvon SD, Straughan K, Stevens JM. Hippocampal volumetric and morphometric studies in frontal and temporal lobe epilepsy. Brain 99; 5: 005. Ebner A. Lateral (neocortical) temporal lobe epilepsy. In: Wolf P, editor. Epileptic seizures and syndromes. London: John Libbey, 994: 75. Ebner A, Hoppe M. Noninvasive electroencephalography and mesial temporal sclerosis. J Clin Neurophysiol 995; :. Engel J Jr. Recent advances in surgical treatment of temporal lobe epilepsy. [Review]. Acta Neurol Scand 99; 6 Suppl 40: 70. Falconer MA. Genetic and related aetiological factors in temporal lobe epilepsy: a review. Epilepsia 97; :. Fish D, Andermann F, Olivier A. Complex partial seizures and small posterior temporal or extratemporal structural lesions: surgical management. Neurology 99; 4: 7 4. Foldvary NR, Lee N, Mayes BN, Friedman AH, Van Landingham KE, Lewis DV. et al. Extrahippocampal temporal lobe epilepsy: clinical manifestations [abstract]. Epilepsia 994; 5 Suppl : 09. Gates JR, Gumnit RJ. Partial seizures of temporal lobe origin. In: Downloaded from by guest on 07 September 0

9 Engel J Jr, editors. Surgical treatment of the epilepsies. New York: Raven Press, 97: 795. GilNagel A, Risinger MW. lctal semiology in hippocampal vs. extrahippocampal temporal lobe epilepsy [abstract]. Neurology 99; 4 ( Suppl ): A7. Gloor P. Commentary': approaches to localization of the epileptogenic lesion. In: Engel J Jr, editor. Surgical treatment of the epilepsies. New York: Raven Press, 97: Gloor P. Experiential phenomena of temporal lobe epilepsy. [Review]. Brain 990: : Hajek M, Antonini A, Leenders KL, Wieser HG. Mesiobasal versus lateral temporal epilepsy: metabolic differences in the temporal lobe shown by interictal FFDG positron emission tomography [see comments]. Neurology 99; 4: 796. Comment in: Neurology 99; 4: 56. Ho SS, Berkovic SF, McKay WJ, Bladin PF. Subtypes of temporal lobe epilepsy: clinical and ictal "TcHMPAO SPECT analysis [abstract]. Aust NZ J Med 994; 4: 6. Kilpatrick C, Cook M, Kaye A, Murphy M. Noninvasive investigations successfully select patients for temporal lobectomy [abstract]. Epilepsia 995: 6 Suppl : S97. Kotagal P, LUders H, Williams G, Wyllie E, Nichols T, McPherson J. Temporal lobe complex partial seizures: analysis of symptom clusters and sequences [abstract], Epilepsia 9; 9: 66. Kuzniecky RI, Cascino GD, Palmini A, Jack CR Jr, Berkovic SF, Jackson GD, et al. Structural neuroimaging. In: Engel J Jr, editor. Surgical treatment of the epilepsies. nd ed. New York: Raven Press, 99: Mihara T, Inoue Y, Hiyoshi T, Watanabe Y, Kubota Y, Tottori T, et al. Localizing value of seizure manifestations of temporal lobe epilepsies and the consequence of analyzing their sequential appearance. Jpn J Psychiatry Neurol 99; 47: 75. Pacia SV, Ebersole JS. Intracranial determinants of scalp ictal EEG patterns in temporal lobe seizures [abstract]. Electroencephalogr Clin Neurophysiol 994: 9: 4P. Penfield W. Perot P. The brain's record of auditory and visual experience. A final summary and discussion. Brain 96; 6: Risinger MW. Engel J Jr, Van Ness PC, Henry TR. Crandall PH Mesial versus neocortical temporal lobe epilepsy 4 Ictal localization of temporal lobe seizures with scalp/sphenoidal recordings. Neurology 99; 9: 9. Saygi S, Spencer SS, Scheyer R, Katz A, Mattson R, Spencer DD. Differentiation of temporal lobe ictal behavior associated with hippocampal sclerosis and tumors of temporal lobe. Epilepsia 994; 5: 774. So NK. Epileptic auras. In: Wyllie E, editor. The treatment of epilepsy: principles and practice. Philadelphia: Lea and Febiger 99: Spencer SS. Cortical and intercortical seizure spread. In: Meldrum BS, Ferendelli JA, Wieser HG, editors. Anatomy of epileptogenesis. London: John Libbey, 9: 954. Spencer SS, Spencer DD, Williamson PD, Mattson R. Combined depth and subdural electrode investigation in uncontrolled epilepsy. Neurology 990; 40: 749. Walczak TS. Neocortical temporal lobe epilepsy: characterizing the syndrome. Epilepsia 995; 6: 65. Walczak T, Bazil C, Lee N, Radtke R, Goodman R, Friedman A. Scalp ictal EEG differs in temporal neocortical and hippocampal seizures [abstract]. Epilpesia 994; 5 Suppl : 4. Wieser HG, Engel J, Williamson PD, Babb TL, Gloor P. Surgically remedial temporal lobe syndromes. In: Engel J Jr, editor. Surgical treatment of the epilepsies, nd ed. New York: Raven Press, 99: 496. Williamson PD. Wieser HG, DelgadoEscueta AV. Clinical characteristics of partial seizures. In: Engel J Jr, editor. Surgical treatment of the epilepsies. New York: Raven Press, 97: 00. Williamson PD, Thadani VM, Darcey TM, Spencer DD, Spencer SS, Mattson RH. Occipital lobe epilepsy: clinical characteristics, seizure spread patterns, and results of surgery. [Review]. Ann Neurol 99a; :. Williamson PD, Boon PA, Thadani VM, Darcey TM, Spencer DD, Spencer SS, et al. Parietal lobe epilepsy: diagnostic considerations and results of surgery. Ann Neurol 99b; 90. 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 99; 4: 77. Received March, 996. Revised June 9, 996. Accepted June 4, 996 Downloaded from by guest on 07 September 0