Postictal Language Assessment and Laterahation of Complex Partial Seizures Michael D. Privitera, MD, George L. Morris, MD, and Frank Gilliam, MD
We performed a prospective study of ictal and postictal language function after 105 temporal lobe complex partial seizures in 26 patients. Seizure localization was verified by a greater than YO%, reduction in seizure frequency after temporal lobectomy. At the time of the seizure, the patient was asked to read a test phrase aloud until it was read correctly and clearly. I n all 62 seizures originating from the left temporal lobe, the patient took more than 68 seconds to read the test phrase correctly (mean, 321.9 seconds); in 42 of 4 3 seizures from the right temporal lobe, the patient read the test phrase in less than 54 seconds (mean, 19.7 seconds). Postictal paraphasias occurred in 46 of 6 2 seizures from the left temporal lobe ( 1 1 of 14 patients). I n this study, quantifying the time delay in reading a test phrase lateralized seizure onset in all 26 patients tested, proving significantly more accurate than any other single noninvasive presurgical test. Privitera MD, Morris GL, Gilliam F. Postictal language assessment and lateralization of complex partial seizures. Ann Neurol 199 1;30:391-396
Surgical treatment of epilepsy has become more widely accepted as a highly effective treatment for medically intractable seizures El]. Accurate localization of the onset of partial seizures is a critical factor in surgical success [2]. Patients with temporal lobe complex partial seizures (TLCPS) have the best outcome after surgery compared with other seizure types 11). TLCPS have few lateralizing ictal behaviors, however, and electroencephalography (EEG) often shows bilateral abnormalities [ 3 ) ; therefore, most centers rely on a combination of diagnostic tests to localize the epileptogenic region [2, 4). Complex partial seizures (CPS) may produce a variety of speech and language manifestations during the ictal and postictal periods, which provide information about seizure lateralization 15- 12). With the advent of simultaneous video/EEG recording, several investigators have confirmed that ictal speech is a manifestation of nondominant TLCPS, whereas postictal dysphasia occurs with TLCPS from the dominant hemisphere C11, 12). Despite review of videotapes, however, 52% of the patients of Koerner and Laxer [ll) and 3796 of the patients of Gabr and colleagues 112) had neither detectable spontaneous speech nor postictal dysphasia. We performed a prospective study in which patients were continuously asked to read aloud a brief test phrase during and after each seizure. We assessed the lateralizing significance of ictal speech and postictal paraphasias, and additionally measured the time from the end of the EEG ictal discharge until the patient read a test phrase correctly. From the Deparrmcnr of Neurology, University of Cincinnari Medical Center, Cincinnati, OH. Received Sep 20, 1990, and in revised form Jan 17 and Mar 13, 1991. Accepted for publication Mar 14, 1991.
Methods We assessed postictal language function o n all 126 patients evaluated in the epilepsy monitoring unit of University Hospital of the University of Cincinnati Medical Center (Cincinnati, OH) from May 1988 to March 1990. For this study, we analyzed postictal language on 105 seizures in 26 patients who had TLCPS that did not secondarily generalize. We excluded those who had bilateral language representation o n the intracarotid sodium amobarbital test (ISA) (1 patient), bilateral independent seizure onset (1 patient), only seizures that spread to generalized tonic clonic ( 1 patient), or inadequate language testing in all of their CPS (1 patient). We considered language testing inadequate if the test phrase was not presented within the first 60 seconds after the end of the ictal EEG discharge. We determined seizure localization by interictal EEG, ictal videoiEEG, magnetic resonance imaging (MRI), and neuropsychological testing, and a11 patients had at least 90% reduction in seizure frequency after anterior temporal lobectomy. Eight patients had videoiEEG with invasive electrodes (stereotaxic depth or subdural grids and strips) because noninvasive studies failed to localize the epilcpcogenic focus. We performed an ISA on most patients (17 of 26 patients); patients who did not have an ISA were right-handed without a family history of left-handedness. To test language, patients read a phrase from the Boston Diagnostic Aphasia Test [l?] (“they heard him speak on the radio last night”), which was printed in 1-in. block letters on an 8.5 x 11-in. card. All patients were able to read the test phrase without errors during the interictal period. In our epilepsy monitoring unit, technicians or nurses watch the patients on a monitor with split-screen EEGivideo to detect clinical or electrographic evidence of a seizure, 24 hours per day, 7 days per week. When a seizure was detected, the Address correspondence to Dr Privitera, Department of Neurology (525), University of- Cincinnati Medical Center, Cincinnati, OH 45267-0525.
Copyright 0 1991 by the American Neurological Association
391
technician or nurse went to the patient’s bedside and, after ensuring patient safety, continuously instructed the patient to read aloud the test phrase until it was read clearly and correctly. Two of the authors (M.D.P. and F.G.; F.G. did not participate in the presurgical evaluation of these patients and was blind to the hemisphere of seizure origin) independently reviewed the videotapes and hardcopy EEG, and analyzed the following two language parameters: the presence or absence of dysphasic language errors and the time from the end of the EEG ictal discharge until the patient read the test phrase correctly. We defined the ictal EEG discharge as rhythmic sinusoidal waves, repetitive epileptiform potentials, or both [ 14, 151, and postictal EEG patterns as polymorphic slowing, attenuation, “activation of spikes,” or return to interictal nature {16]. We then determined the point where the EEG changed from an ictal to a postictal pattern, and measured the time from this point until the patient read the test phrase correctly.
Results Twenty-six patients had a total of 142 seizures; 17 patients were male, 9 were female. Mean age was 32.5 years (range, 17-58 years). Fourteen patients had left temporal lobe seizure foci, 12 had right. We evaluated a mean of 5.5 seizures per patient (range, 1-12 seizures). Twenty patients were right-handed, 4 were left-handed, and 2 were ambidextrous. We excluded seizures from analysis for the following reasons: There was secondary generalization after an initial CPS (17 patients); the seizure was not detected until videotapes were reviewed (17 patients), or the test phrase was not presented within 60 seconds of the end of the seizure (3 patients). Of 105 seizures, 62 seizures in 14 patients originated in the left temporal lobe; 43 seizures in 12 patients originated in the right temporal lobe. Twenty-two seizures in 8 patients occurred during video/EEG monitoring with invasive electrodes (2 patients had depth and 6 had subdural electrodes). The two authors who independently reviewed the EEG tracings differed by a mean of 2.4 2 0.4 (SEM) seconds (range, 0-37 seconds) in determining the time of EEG ictal termination. In all but one seizure the difference was 10 seconds or less. The two reviewers agreed on lateralization in all seizures using as a cutoff 60 seconds from ictal termination until accurate reading. Ictal termination was readily identifiable in most seizures, however, some EEG tracings required strict attention to our criteria for ictal and postictal patterns (see Methods). Four seizures in 1 patient had no obvious EEG ictal discharge, but there was an abrupt cessation of interictal spiking at the clinical onset of the seizures and a return of interictal spiking at the end of the clinical seizures. Using the time of EEG returned to an interictal pattern (spiking) as the EEG ictal termination, the EEG reviewers differed by 10, 1, 3, and 2 seconds in determining the time of EEG ictal termina-
tion in these four seizures. In 1 other patient, the EEG reviewers differed by 37 seconds on the time of ictal termination in one seizure where there was a waxing and waning discharge that defied accurate categorization into an ictal or postictal pattern. In nine other seizures in this patient, there was less than 10 seconds’ difference between the EEG reviewers in identifying ictal termination. Time to read the test phrase accurately predicted the hemisphere of seizure origin in all 26 patients. In 100% (62 of 62) of the seizures originating from the left temporal lobe, the patient took more than 68 seconds to read the sign correctly (Fig). In 98% (42 of 43) of the seizures originating from the right temporal lobe, the patient read the sign correctly in less than 54 seconds. One right-hemisphere patient took 106 seconds to read the sign correctly in her first seizure, but read correctly in 48, 54, 52, and 44 seconds in four subsequent seizures. In 10 seizures (4 patients) from the right temporal lobe, the patients read the test phrase correctly during the ictal EEG discharge. The mean time to read correctly in the left temporal lobe group was 321.9 31.3 (SEM) seconds (range, 68-1,276 seconds), whereas the mean time in the right temporal lobe group was 19.7 3.5 (SEM) seconds (range, 0-106 seconds). Eight patients had ictal EEG recordings with noninvasive electrodes (Phase I) and later with invasive recordings (Phase 11). We were able to analyze seizures recorded during both Phase I and Phase I1 in 5 of these patients (seizures from left temporal lobe in 3 patients, right temporal lobe in 2 patients). The time to read the test phrase after 27 seizures recorded during Phase I monitoring was significantly shorter than the time to read after 17 seizures recorded during Phase I1 monitoring (mean Phase I, 133.0 k 20.0 [SEMI seconds; mean Phase 11, 308.2 ir 75.8 [SEMI seconds; p = 0.012). The one seizure from the right temporal lobe that produced a delay in reading greater than 60 seconds was recorded in a patient during Phase I1 monitoring. The difference in duration of language dysfunction was not explained by a difference in seizure duration (mean seizure duration Phase I, 147.2 seconds; Phase 11, 138.9). To assess possible errors in using this technique, we analyzed the data a second time assuming a 10-second error (we added 10 seconds if the measured delay was less than 60 seconds, and subtracted 10 seconds if the measured delay was more than 60 seconds). A total of four seizures would have been misclassified. Two patients would have had a single seizure misclassified, but the majority of seizures would have correctly classified the patient. A single patient with right TLCPS would have had three of five seizures misclassified. This patient was undergoing Phase I1 monitoring where reading delay was longer in both the left- and right-hemisphere groups.
392 Annals of Neurology Vol 30 No 3 September 1991
*
*
15
I
= LEFT HEMISPHERE
X
O = RIGHT HEMISPHERE
., X
A
X X X X
X
10
X
5
X X X X X X X X
X
x x xxxxxxxx
1
-+ i 0
60
120
180
1
240
i
i i i
i
-
x x xx i
300
360
>420
TIME(SEC0NDS) Time (rounded to neawst 10 seconds)from the end of the electroencephalographic ictal discharge t o the correct wading of the test phrase for all seizures as scored by one of the two independent reviewets (M.D.P.). The two miewers agreed on lateralization of all seizwa wing a 60-second cu~ofi Mean time for the l4t temporal lobe complex partial seizwe (TLCPS) was 321.9 seconds; mean time for right TLCPS was 19.7 second.
Phonemic paraphasias were the only dysphasic language errors we detected while patients read the test phrase during the postictal period. These occurred in 46 of 62 seizures (74%) from the left temporal lobe and 0 of 43 seizures from the right temporal lobe. This corresponded to 11 of 14 patients with left TLCPS and 12 of 12 patients with right TLCPS. Thus, 3 patients (16 seizures) would have been misclassified using only postictal paraphasic errors during reading. There was no significant dlfference in the occurrence of paraphasic errors during Phase I compared with Phase I1 monitoring. Although the time to read the test phrase correctly lateralized in all 26 patients, the following other noninvasive tests were less frequently lateralizing: MRI (7 of 26 patients), interictal EEG (14 of 26 patients), ictal EEG (14 of 26 patients), and neuropsychological test battery (18 of 26 patients) (Table 1). Using McNemar’s Test [17), the time to read the test phrase was significantly more accurate in identifying the hemisphere of seizure origin than MRI ( p < 0.001), interictal EEG ( p = 0.0005),ictal EEG ( p = 0.0005),and neuropsychological test battery ( p = 0.008). Two patients are noteworthy because we believe they help determine the anatomical site of the postictal language dysfunction we measured. The first patient is a 40-year-old woman who underwent left temporal resection in September 1987 with resection of 5 cm of temporal neocortex, amygdala, and less than 1 cm
of hippocampus. ISA demonstrated left-hemisphere language dominance. She was seizure free for 1 year after surgery, but then had recurrence of medically intractable CPS. Repeat Phase I ictal video/EEG monitoring suggested TLCPS, but laterality could not be determined. She had stereotaxic implantation of depth electrodes in mesial temporal structures bilaterally from a horizontal approach. We recorded 18 subclinical (patient reported no symptoms) ictal EEG discharges consisting of beta frequency rhythmic discharges involving only the left hippocampus and lasting 30 seconds. The patient had no postictal paraphasias, read the sign correctly within 60 seconds each time, and read correctly during the EEG discharge in three instances. She had two habitual complex partial seizures with alteration in consciousness and masticatory automatisms, and was unable to read the sign for 167 and 260 seconds after the EEG discharge ended. These seizures began in the left hippocampus. The second patient is a 31-year-old man with medically intractable TLCPS. During Phase I video/EEG monitoring, he had six simple partial seizures consisting of a feeling of anxiety and an abdominal rising sensation (similar to the warning prior to his typical complex partial seizure). Each of the seizures had a simultaneous 6 to 8-Hq rhythmic, sharp wave discharge maximal at the left sphenoidal electrode. H e had no postictal difficulty with reading and, in two seizures, read the test phrase correctly during the EEG ictal discharge. After a complex partial seizure with alteration in consciousness, staring, and masticatory automatisms, however, he was unable to read the sign correctly for 526 seconds and made multiple paraphasic errors while reading. Privitera et al: Postictal Language Assessment
393
T&e 1 . Presurgical Noninvasive Tests
Patient 1" 2 3 4 5 6" 7" 8
9 10 1I" 12"
13 14 15" 16 17 18 19" 20 21 22 23 24" 25 26
Hemisphere of Surgery
NPT
MRI
Interictal EEG
Ictal EEG
L
L R R R L L R R NL L NL L L L NL R L R NL R NL R NL NL R NL
NL R NL L NL NL NL NL R L NL NL NL NL NL R NL NL NL NL NL NL NL NL R R
L NL NL L NL NL NL NL NL L L L L L NL R L R R R L NL NL NL R NL
NL L R NL L NL NL NL L NL NL L L NL NL R NL R R R L L NL L NL R
R R L L L
R R L
L L L L L R R L R R
R L L R L R R
aPatients had Phase JI monitoring with invasive electrodes.
NF'T
= neutopsychoiogical testing; MRI nonlateralizing.
=
magnetic resonance imaging; EEG
Discussion Our findings show that ictal and postictal language disturbances provide accurate information about the hemisphere of seizure origin. In our series of patients with TLCPS, the time from the end of the seizure to accurately read a test phrase aloud correlated with the hemisphere of seizure onset in all 26 patients tested, proving more accurate in seizure lateralization than any other noninvasive presurgical test. We believe our test is accurate because it is brief and repeatable and thus can distinguish between patients whose postictal ability to read is normal in less than 60 seconds (42 of 43 right-hemisphere seizures) compared with those whose language takes more than 60 seconds to normalize (62 of 62 left-hemisphere seizures). A more comprehensive language assessment may have provided more information about the type of language dysfunction, but may not have been able to classify patients within the first 2 minutes after the seizure, the critical time for most of our patients. Interpretation of ictal EEG recordings may be difficult, especially when recorded from scalp and sphenoidal electrodes. By using specific criteria to differentiate
=
electroencephalogaphy; R
=
right; L
=
left; NL =
ictal from postictal EEG, however, two independent EEG reviewers differed by a mean of only 2.4 seconds in assessing the time of ictal termination. The two reviewers agreed on lateralization in all seizures. We predicted that postictal confusion might lead to some inaccuracy in our test, but this was not the case. In 42 of 43 seizures originating in the right temporal lobe, the patient was able to read correctly within 60 seconds; in 10 seizures, the patient read correctly during the ictal EEG discharge. These patients may have been impaired on tests of nondominant-hemisphere function (2 patients appeared to have transient disturbances of language prosody), but they were able to read without errors. We believe that much of what has been previously described as "postictal confusion" in patients with TLCPS may represent transient disruption of receptive or expressive language abilities. The pattern of postictal language dysfunction in our patients appeared to depend only on the hemisphere of seizure origin and was not related to the spread of the ictal discharge to the contralateral hemisphere. TLCPS frequently show ictal spread to the contralateral hemisphere [18-201. Despite ictal interhemispheric
394 Annals of Neurology VoI 30 No 3 September 1991
Table 2.Seizure-related Language StudieJ
Left-hemisphere Seizures Study
Language Test
Bingley (1958) “Ictal aphasia” Serafetinides (1963) Dysphasia King (1977) Dysphasia McKeever (1983) Expressive dysphasia or speech arreste Impaired speechr Koerner (1988) Dysphasia‘ Gabr (198% Reading delay‘ Present study
Right-hemisphere Seizures
(5%)
Yield“ (%)
Accuracyb (p) Language Test
Yieldc (F)
Accuracyd
16/33 (48) 16/31 (52) 20131 (65) 717 (100)
16/16 (100) 16/17 (94) 20121 (95) 719 (78)
10117 (59) 11/22 (50) ...
l0/16 (63) 11/15 ( 7 3 ) ...
18/22 (82) 21/32 (66) 12/13 (92) 12/16 (75) 14/14 (100) 14/14 (100)
Speech automatism Speech automatism
... ...
Formed ictal speech Formed ictal speech Reading delay
...
...
12/40 (30) 12/13 (92) 10119 (53) 10112 (83) 12/12 (100) 12/12 (100)
”Proportion of patients with left-hemisphere seizures who showed posticd language disturbance. ‘Proportion of patients with postictal language disturbance who had left-hemisphere seizures. ‘Proportion of patients with right-hemisphere seizures who showed ictal speech. dProportion of patients with ictal speech who had right-hemisphere seizures. eStudies used videoiEEG for analysis.
EEG = electroencephalography.
spread, the postictal language disturbance correlated with the hemisphere of seizure origin in all our 26 patients. Gotman [18] used computer techniques to analyze interhemispheric spread of complex partial seizures recorded with depth electrodes and found that the hemisphere of seizure origin retained an electrophysiological “influence” over the contralateral discharge throughout the seizure. Our findings indicate that this influence extends to behavioral aspects of the seizure and persists into the postictal period. Jackson first noted that language disturbances were occasionally associated with seizures from the dominant hemisphere {5}. Studies by Bingley 171, Serafetinides and Falconer 181, and King and Ajmone-Marsan {9] found significant correlations between ictal or postictal dysphasia and dominant-hemisphere seizure foci; Bingley [7] and Serafetinides and Falconer [Sl also noted an association between ictal speech automatisms and nondominant seizure foci (Table 2). Later studies using video/EEG analysis of seizures [lo-12) confirmed the results of earlier investigators and increased detection of language abnormalities by careful review of video tapes. In our study, 74% of patients with dominant TLCPS had postictal paraphasic errors, and every patient with paraphasic errors had dominant TLCPS. Similarly, 33% of patients with nondominant TLCPS read correctly during the ictal EEG discharge (equivalent to formed ictal speech in earlier studies), and every patient who read during the EEG discharge had nondominant TLCPS. Our technique of measuring the time to read added a substantial amount of information because we were able to test 26 of 28 patients with 100% accuracy. One of the 2 patients we could not test adequately had only secondarily generalized seizures from a temporal lobe focus, and the second patient had all four of his seizures during the early
morning hours and was not tested within our 60-second limit. We did not analyze time to read in patients with generalized tonic-clonic seizures because these were followed by prolonged alteration of consciousness, which was beyond the time limits of our staff to perform the testing. Furthermore, we did not analyze testing in patients with extratemporal complex partial seizures. In the study by Koerner and Laxer [l I}, patients with extratemporal seizures did not have postictal language impairment. The 2 patients presented in detail demonstrate that subclinical electrographic seizures and simple partial seizures involving only dominanthemisphere mesial temporal structures do not produce language disturbances that are measureable using our techniques. Thus, if a patient does not demonstrate postictal language disturbance, other data should be used to determine whether the patient has experienced a nondominant TLCPS, a simple partial seizure, or an extratemporal CPS. We speculate that the brain region responsible for postictal language disturbances in dominant TLCPS is in the ipsilateral posterior perisylvian region. The 2 patients described in detail showed ictal EEG discharges without altered consciousness involving dominant-hemisphere mesial temporal structures and no language disturbance. During a complex partial seizure with identical EEG onset, however, both patients showed substantial delay in reading the test phrase and 1 patient had prominent postictal paraphasic errors. Thus, isolated mesial temporal ictal activity alone did not cause the language dysfunction detected by our technique. Stimulation studies have demonstrated interference with reading when electrically stimulating posterior-inferior frontal, posterior perisylvian, or basal temporal regions 121-231. We do not believe the basal Privitera et al: Postictal Language Assessment
395
temporal language area is responsible for postictal language dysfunction because the first patient we described in detail had most, if not all, of the basal temporal language area removed, using measurements recently reported by Burnstine and associates E241. Despite this, the patient had no language disturbance after mesial temporal electrographic seizures, yet had typical dominant temporal lobe postictal language dysfunction after a clinical complex partial seizure with lateral temporal spread. It is unlikely that the inferior frontal language area produced the postictal language dysfunction we saw because dominant-hemisphere extratemporal seizures rarely produce postictal language dysfunction rl11. Although we believe the posterior temporal region is responsible for postictal language disturbances in patients with TLCPS, it is possible that multiple language areas are impaired simultaneously in the postictal period. Detailed study of more patients may provide more information on the localization of postictal language dysfunction. The language assessment we report has important practical application as an additional noninvasive test to use in evaluating patients for surgical treatment of epilepsy. At most surgical centers, if there is inadequate data from noninvasive studies, then ictal video/ EEG monitoring with invasive (depth or subdural) electrodes is used. Invasive recordings provide a high degree of accuracy, but increase the risk, cost, and duration of the presurgical evaluation {25]. Our findings show that postictal language testing provides accurate data about seizure lateralization when the epileptogenic focus is in the temporal lobe and, in our series, this was the single most accurate test in lateralizing a temporal lobe focus. Our simple testing method cam be taught to nurses or EEG technologists in a few minutes so that an epileptologist or personnel trained in comprehensive language evaluation need not be present. This is helpful when patients have medications withdrawn and are monitored throughout a 24-hour period. The brief, repeatable nature of the test makes it easier to analyze from videotapes because the examiner knows the words he is listening for and may be able to detect subtle language problems despite background noise in the patient room. We thank the EEG technologists and nurses of the Epilepsy Monitoring Unit of the University of Cincinnati Hospital (Cincinnati, O H ) whose dedication and attention to detail helped make this study possible.
References 1. Engel J. Outcome with respect to epileptic seizures. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987:5 53-5 72 2. Engel J. Approaches to localization of the epileptogenic lesion. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987:75-95
3. So N, Gloor P, Quesney LF, et al. Depth electrode investigations in patients with bitemporal epileptiform abnormalities. Ann Neurol 1989;25:423-431 4. Engel J. Appendix 11: presurgical evaluation protocols. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987:669-698 5. Jackson JH. Relations of different divisions of the central nervous system to one another and to parts of the body. Lancet 1898;1:79-87 6. Hecaen H, Piercy M. Paroxysmal dysphasia and the problem of cerebral dominance. J Neurol Neurosurg Psychiatry 1956; 19:194-201 7 Bingley T. Mental symptoms in temporal lobe epilepsy and temporal lobe gliomas. Acta Psychiatr Neurol (Suppl) 1958; 120:95-101 8 Serafetinides EA, Falconer MA. Speech disturbances in temporal lobe seizures: A study in 100 epileptic patients submitted to anterior temporal lobectomy. Brain 1963;86:333-346 9. King DW, Ajmone-Marsan C. Clinical features and ictal patterns in epileptic patients with EEG temporal lobe foci. Ann Neurol 1977;2:138-147 10. McKeever M, Holmes GL, Russman BS. Speech abnormalities in seizures: a comparison of absence and partial complex seizures. Brain Lang 1983;19:25-32 11. Koerner M, Laxer KD. Ictal speech, postictal language dysfunction, and seizure lateralization. Neurology 1988;38:634636 12. Gabr M, Lueders H, Dinner D, et al. Speech manifestations in laterahation of temporal lobe seizures. Ann Neurol 1989; 25:82-87 13. Goodglass H , Kaplan E. The Boston Diagnostic Aphasia Examination. Philadelphia: Lea and Febiger, 1972 14. mass DW. Electroencephalographic manifestations of complex partial seizures. In:Penry JK, Daly DD, eds. Advances in neurology, vol 11. New York: Raven Press, 1975:113-140 15. Blume WT, Young GB, LemieuxJF. EEG morphology of partial epileptic seizures. Electroencephalogr Clin Neurophysiol 1984; 57:295-302 16. Kaibara M, Blume WT. The postictal electroencephalogram. Electroencephalogr Clin Neurophysiol 1988;70:99-104 17. Rosen B. Fundamentals of biostatistics, second edition. Boston: Duxbury Press, 1986:333-338 18. Gotman J. Interhemispheric interactions in seizures of focal onset: data from human intracranial recordings. Electroencephalogr Clin Neurophysiol 1987;67:120-133 19. Lieb JP, Hoque K, Skomer CE, Song XW. Inter-hemispheric propagation of human mesial temporal lobe seizures: A coherenceiphase analysis. Electroencephalogr Clin Neurophysiol 1987;67:101-1 19 20. Spencer SS, Williamson PD, Spencer DD, Mattson RH. Human hippocampal seizure spread studied by depth and subdural recording: the hippocampal commissure. Epilepsia 1987;28: 479-489 21. Penfield W, Rasmussen T. Localization and arrest of speech. Arch Neurol Psychiatry 1949;61:21-27 22. Lueders H, Lesser RP, Hahn J, et al. Basal temporal language area demonstrated by electrical stimulation. Neurology 1986; 36:505-510 23. Lueders H, Lesser RP, Dinner DS, et al. Localization of cortical function: new information from extraoperative monitoring of patients with epilepsy. Epilepsia 1988;29(suppl 2):S56-S65 24. Burnstine TH, Lesser RP, Hart J, et al. Characterization of the basal temporal language area in patients with left temporal lobe epilepsy. Neurology 1990;40:966-970 25. Van Buren JM. Complications of surgical procedures in the diagnosis and treatment of epilepsy. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987: 465-476
396 Annals of Neurology VoI 30 No 3 September 1991
We performed a prospective study of ictal and postictal language function after 105 temporal lobe complex partial seizures in 26 patients. Seizure localization was verified by a greater than YO%, reduction in seizure frequency after temporal lobectomy. At the time of the seizure, the patient was asked to read a test phrase aloud until it was read correctly and clearly. I n all 62 seizures originating from the left temporal lobe, the patient took more than 68 seconds to read the test phrase correctly (mean, 321.9 seconds); in 42 of 4 3 seizures from the right temporal lobe, the patient read the test phrase in less than 54 seconds (mean, 19.7 seconds). Postictal paraphasias occurred in 46 of 6 2 seizures from the left temporal lobe ( 1 1 of 14 patients). I n this study, quantifying the time delay in reading a test phrase lateralized seizure onset in all 26 patients tested, proving significantly more accurate than any other single noninvasive presurgical test. Privitera MD, Morris GL, Gilliam F. Postictal language assessment and lateralization of complex partial seizures. Ann Neurol 199 1;30:391-396
Surgical treatment of epilepsy has become more widely accepted as a highly effective treatment for medically intractable seizures El]. Accurate localization of the onset of partial seizures is a critical factor in surgical success [2]. Patients with temporal lobe complex partial seizures (TLCPS) have the best outcome after surgery compared with other seizure types 11). TLCPS have few lateralizing ictal behaviors, however, and electroencephalography (EEG) often shows bilateral abnormalities [ 3 ) ; therefore, most centers rely on a combination of diagnostic tests to localize the epileptogenic region [2, 4). Complex partial seizures (CPS) may produce a variety of speech and language manifestations during the ictal and postictal periods, which provide information about seizure lateralization 15- 12). With the advent of simultaneous video/EEG recording, several investigators have confirmed that ictal speech is a manifestation of nondominant TLCPS, whereas postictal dysphasia occurs with TLCPS from the dominant hemisphere C11, 12). Despite review of videotapes, however, 52% of the patients of Koerner and Laxer [ll) and 3796 of the patients of Gabr and colleagues 112) had neither detectable spontaneous speech nor postictal dysphasia. We performed a prospective study in which patients were continuously asked to read aloud a brief test phrase during and after each seizure. We assessed the lateralizing significance of ictal speech and postictal paraphasias, and additionally measured the time from the end of the EEG ictal discharge until the patient read a test phrase correctly. From the Deparrmcnr of Neurology, University of Cincinnari Medical Center, Cincinnati, OH. Received Sep 20, 1990, and in revised form Jan 17 and Mar 13, 1991. Accepted for publication Mar 14, 1991.
Methods We assessed postictal language function o n all 126 patients evaluated in the epilepsy monitoring unit of University Hospital of the University of Cincinnati Medical Center (Cincinnati, OH) from May 1988 to March 1990. For this study, we analyzed postictal language on 105 seizures in 26 patients who had TLCPS that did not secondarily generalize. We excluded those who had bilateral language representation o n the intracarotid sodium amobarbital test (ISA) (1 patient), bilateral independent seizure onset (1 patient), only seizures that spread to generalized tonic clonic ( 1 patient), or inadequate language testing in all of their CPS (1 patient). We considered language testing inadequate if the test phrase was not presented within the first 60 seconds after the end of the ictal EEG discharge. We determined seizure localization by interictal EEG, ictal videoiEEG, magnetic resonance imaging (MRI), and neuropsychological testing, and a11 patients had at least 90% reduction in seizure frequency after anterior temporal lobectomy. Eight patients had videoiEEG with invasive electrodes (stereotaxic depth or subdural grids and strips) because noninvasive studies failed to localize the epilcpcogenic focus. We performed an ISA on most patients (17 of 26 patients); patients who did not have an ISA were right-handed without a family history of left-handedness. To test language, patients read a phrase from the Boston Diagnostic Aphasia Test [l?] (“they heard him speak on the radio last night”), which was printed in 1-in. block letters on an 8.5 x 11-in. card. All patients were able to read the test phrase without errors during the interictal period. In our epilepsy monitoring unit, technicians or nurses watch the patients on a monitor with split-screen EEGivideo to detect clinical or electrographic evidence of a seizure, 24 hours per day, 7 days per week. When a seizure was detected, the Address correspondence to Dr Privitera, Department of Neurology (525), University of- Cincinnati Medical Center, Cincinnati, OH 45267-0525.
Copyright 0 1991 by the American Neurological Association
391
technician or nurse went to the patient’s bedside and, after ensuring patient safety, continuously instructed the patient to read aloud the test phrase until it was read clearly and correctly. Two of the authors (M.D.P. and F.G.; F.G. did not participate in the presurgical evaluation of these patients and was blind to the hemisphere of seizure origin) independently reviewed the videotapes and hardcopy EEG, and analyzed the following two language parameters: the presence or absence of dysphasic language errors and the time from the end of the EEG ictal discharge until the patient read the test phrase correctly. We defined the ictal EEG discharge as rhythmic sinusoidal waves, repetitive epileptiform potentials, or both [ 14, 151, and postictal EEG patterns as polymorphic slowing, attenuation, “activation of spikes,” or return to interictal nature {16]. We then determined the point where the EEG changed from an ictal to a postictal pattern, and measured the time from this point until the patient read the test phrase correctly.
Results Twenty-six patients had a total of 142 seizures; 17 patients were male, 9 were female. Mean age was 32.5 years (range, 17-58 years). Fourteen patients had left temporal lobe seizure foci, 12 had right. We evaluated a mean of 5.5 seizures per patient (range, 1-12 seizures). Twenty patients were right-handed, 4 were left-handed, and 2 were ambidextrous. We excluded seizures from analysis for the following reasons: There was secondary generalization after an initial CPS (17 patients); the seizure was not detected until videotapes were reviewed (17 patients), or the test phrase was not presented within 60 seconds of the end of the seizure (3 patients). Of 105 seizures, 62 seizures in 14 patients originated in the left temporal lobe; 43 seizures in 12 patients originated in the right temporal lobe. Twenty-two seizures in 8 patients occurred during video/EEG monitoring with invasive electrodes (2 patients had depth and 6 had subdural electrodes). The two authors who independently reviewed the EEG tracings differed by a mean of 2.4 2 0.4 (SEM) seconds (range, 0-37 seconds) in determining the time of EEG ictal termination. In all but one seizure the difference was 10 seconds or less. The two reviewers agreed on lateralization in all seizures using as a cutoff 60 seconds from ictal termination until accurate reading. Ictal termination was readily identifiable in most seizures, however, some EEG tracings required strict attention to our criteria for ictal and postictal patterns (see Methods). Four seizures in 1 patient had no obvious EEG ictal discharge, but there was an abrupt cessation of interictal spiking at the clinical onset of the seizures and a return of interictal spiking at the end of the clinical seizures. Using the time of EEG returned to an interictal pattern (spiking) as the EEG ictal termination, the EEG reviewers differed by 10, 1, 3, and 2 seconds in determining the time of EEG ictal termina-
tion in these four seizures. In 1 other patient, the EEG reviewers differed by 37 seconds on the time of ictal termination in one seizure where there was a waxing and waning discharge that defied accurate categorization into an ictal or postictal pattern. In nine other seizures in this patient, there was less than 10 seconds’ difference between the EEG reviewers in identifying ictal termination. Time to read the test phrase accurately predicted the hemisphere of seizure origin in all 26 patients. In 100% (62 of 62) of the seizures originating from the left temporal lobe, the patient took more than 68 seconds to read the sign correctly (Fig). In 98% (42 of 43) of the seizures originating from the right temporal lobe, the patient read the sign correctly in less than 54 seconds. One right-hemisphere patient took 106 seconds to read the sign correctly in her first seizure, but read correctly in 48, 54, 52, and 44 seconds in four subsequent seizures. In 10 seizures (4 patients) from the right temporal lobe, the patients read the test phrase correctly during the ictal EEG discharge. The mean time to read correctly in the left temporal lobe group was 321.9 31.3 (SEM) seconds (range, 68-1,276 seconds), whereas the mean time in the right temporal lobe group was 19.7 3.5 (SEM) seconds (range, 0-106 seconds). Eight patients had ictal EEG recordings with noninvasive electrodes (Phase I) and later with invasive recordings (Phase 11). We were able to analyze seizures recorded during both Phase I and Phase I1 in 5 of these patients (seizures from left temporal lobe in 3 patients, right temporal lobe in 2 patients). The time to read the test phrase after 27 seizures recorded during Phase I monitoring was significantly shorter than the time to read after 17 seizures recorded during Phase I1 monitoring (mean Phase I, 133.0 k 20.0 [SEMI seconds; mean Phase 11, 308.2 ir 75.8 [SEMI seconds; p = 0.012). The one seizure from the right temporal lobe that produced a delay in reading greater than 60 seconds was recorded in a patient during Phase I1 monitoring. The difference in duration of language dysfunction was not explained by a difference in seizure duration (mean seizure duration Phase I, 147.2 seconds; Phase 11, 138.9). To assess possible errors in using this technique, we analyzed the data a second time assuming a 10-second error (we added 10 seconds if the measured delay was less than 60 seconds, and subtracted 10 seconds if the measured delay was more than 60 seconds). A total of four seizures would have been misclassified. Two patients would have had a single seizure misclassified, but the majority of seizures would have correctly classified the patient. A single patient with right TLCPS would have had three of five seizures misclassified. This patient was undergoing Phase I1 monitoring where reading delay was longer in both the left- and right-hemisphere groups.
392 Annals of Neurology Vol 30 No 3 September 1991
*
*
15
I
= LEFT HEMISPHERE
X
O = RIGHT HEMISPHERE
., X
A
X X X X
X
10
X
5
X X X X X X X X
X
x x xxxxxxxx
1
-+ i 0
60
120
180
1
240
i
i i i
i
-
x x xx i
300
360
>420
TIME(SEC0NDS) Time (rounded to neawst 10 seconds)from the end of the electroencephalographic ictal discharge t o the correct wading of the test phrase for all seizures as scored by one of the two independent reviewets (M.D.P.). The two miewers agreed on lateralization of all seizwa wing a 60-second cu~ofi Mean time for the l4t temporal lobe complex partial seizwe (TLCPS) was 321.9 seconds; mean time for right TLCPS was 19.7 second.
Phonemic paraphasias were the only dysphasic language errors we detected while patients read the test phrase during the postictal period. These occurred in 46 of 62 seizures (74%) from the left temporal lobe and 0 of 43 seizures from the right temporal lobe. This corresponded to 11 of 14 patients with left TLCPS and 12 of 12 patients with right TLCPS. Thus, 3 patients (16 seizures) would have been misclassified using only postictal paraphasic errors during reading. There was no significant dlfference in the occurrence of paraphasic errors during Phase I compared with Phase I1 monitoring. Although the time to read the test phrase correctly lateralized in all 26 patients, the following other noninvasive tests were less frequently lateralizing: MRI (7 of 26 patients), interictal EEG (14 of 26 patients), ictal EEG (14 of 26 patients), and neuropsychological test battery (18 of 26 patients) (Table 1). Using McNemar’s Test [17), the time to read the test phrase was significantly more accurate in identifying the hemisphere of seizure origin than MRI ( p < 0.001), interictal EEG ( p = 0.0005),ictal EEG ( p = 0.0005),and neuropsychological test battery ( p = 0.008). Two patients are noteworthy because we believe they help determine the anatomical site of the postictal language dysfunction we measured. The first patient is a 40-year-old woman who underwent left temporal resection in September 1987 with resection of 5 cm of temporal neocortex, amygdala, and less than 1 cm
of hippocampus. ISA demonstrated left-hemisphere language dominance. She was seizure free for 1 year after surgery, but then had recurrence of medically intractable CPS. Repeat Phase I ictal video/EEG monitoring suggested TLCPS, but laterality could not be determined. She had stereotaxic implantation of depth electrodes in mesial temporal structures bilaterally from a horizontal approach. We recorded 18 subclinical (patient reported no symptoms) ictal EEG discharges consisting of beta frequency rhythmic discharges involving only the left hippocampus and lasting 30 seconds. The patient had no postictal paraphasias, read the sign correctly within 60 seconds each time, and read correctly during the EEG discharge in three instances. She had two habitual complex partial seizures with alteration in consciousness and masticatory automatisms, and was unable to read the sign for 167 and 260 seconds after the EEG discharge ended. These seizures began in the left hippocampus. The second patient is a 31-year-old man with medically intractable TLCPS. During Phase I video/EEG monitoring, he had six simple partial seizures consisting of a feeling of anxiety and an abdominal rising sensation (similar to the warning prior to his typical complex partial seizure). Each of the seizures had a simultaneous 6 to 8-Hq rhythmic, sharp wave discharge maximal at the left sphenoidal electrode. H e had no postictal difficulty with reading and, in two seizures, read the test phrase correctly during the EEG ictal discharge. After a complex partial seizure with alteration in consciousness, staring, and masticatory automatisms, however, he was unable to read the sign correctly for 526 seconds and made multiple paraphasic errors while reading. Privitera et al: Postictal Language Assessment
393
T&e 1 . Presurgical Noninvasive Tests
Patient 1" 2 3 4 5 6" 7" 8
9 10 1I" 12"
13 14 15" 16 17 18 19" 20 21 22 23 24" 25 26
Hemisphere of Surgery
NPT
MRI
Interictal EEG
Ictal EEG
L
L R R R L L R R NL L NL L L L NL R L R NL R NL R NL NL R NL
NL R NL L NL NL NL NL R L NL NL NL NL NL R NL NL NL NL NL NL NL NL R R
L NL NL L NL NL NL NL NL L L L L L NL R L R R R L NL NL NL R NL
NL L R NL L NL NL NL L NL NL L L NL NL R NL R R R L L NL L NL R
R R L L L
R R L
L L L L L R R L R R
R L L R L R R
aPatients had Phase JI monitoring with invasive electrodes.
NF'T
= neutopsychoiogical testing; MRI nonlateralizing.
=
magnetic resonance imaging; EEG
Discussion Our findings show that ictal and postictal language disturbances provide accurate information about the hemisphere of seizure origin. In our series of patients with TLCPS, the time from the end of the seizure to accurately read a test phrase aloud correlated with the hemisphere of seizure onset in all 26 patients tested, proving more accurate in seizure lateralization than any other noninvasive presurgical test. We believe our test is accurate because it is brief and repeatable and thus can distinguish between patients whose postictal ability to read is normal in less than 60 seconds (42 of 43 right-hemisphere seizures) compared with those whose language takes more than 60 seconds to normalize (62 of 62 left-hemisphere seizures). A more comprehensive language assessment may have provided more information about the type of language dysfunction, but may not have been able to classify patients within the first 2 minutes after the seizure, the critical time for most of our patients. Interpretation of ictal EEG recordings may be difficult, especially when recorded from scalp and sphenoidal electrodes. By using specific criteria to differentiate
=
electroencephalogaphy; R
=
right; L
=
left; NL =
ictal from postictal EEG, however, two independent EEG reviewers differed by a mean of only 2.4 seconds in assessing the time of ictal termination. The two reviewers agreed on lateralization in all seizures. We predicted that postictal confusion might lead to some inaccuracy in our test, but this was not the case. In 42 of 43 seizures originating in the right temporal lobe, the patient was able to read correctly within 60 seconds; in 10 seizures, the patient read correctly during the ictal EEG discharge. These patients may have been impaired on tests of nondominant-hemisphere function (2 patients appeared to have transient disturbances of language prosody), but they were able to read without errors. We believe that much of what has been previously described as "postictal confusion" in patients with TLCPS may represent transient disruption of receptive or expressive language abilities. The pattern of postictal language dysfunction in our patients appeared to depend only on the hemisphere of seizure origin and was not related to the spread of the ictal discharge to the contralateral hemisphere. TLCPS frequently show ictal spread to the contralateral hemisphere [18-201. Despite ictal interhemispheric
394 Annals of Neurology VoI 30 No 3 September 1991
Table 2.Seizure-related Language StudieJ
Left-hemisphere Seizures Study
Language Test
Bingley (1958) “Ictal aphasia” Serafetinides (1963) Dysphasia King (1977) Dysphasia McKeever (1983) Expressive dysphasia or speech arreste Impaired speechr Koerner (1988) Dysphasia‘ Gabr (198% Reading delay‘ Present study
Right-hemisphere Seizures
(5%)
Yield“ (%)
Accuracyb (p) Language Test
Yieldc (F)
Accuracyd
16/33 (48) 16/31 (52) 20131 (65) 717 (100)
16/16 (100) 16/17 (94) 20121 (95) 719 (78)
10117 (59) 11/22 (50) ...
l0/16 (63) 11/15 ( 7 3 ) ...
18/22 (82) 21/32 (66) 12/13 (92) 12/16 (75) 14/14 (100) 14/14 (100)
Speech automatism Speech automatism
... ...
Formed ictal speech Formed ictal speech Reading delay
...
...
12/40 (30) 12/13 (92) 10119 (53) 10112 (83) 12/12 (100) 12/12 (100)
”Proportion of patients with left-hemisphere seizures who showed posticd language disturbance. ‘Proportion of patients with postictal language disturbance who had left-hemisphere seizures. ‘Proportion of patients with right-hemisphere seizures who showed ictal speech. dProportion of patients with ictal speech who had right-hemisphere seizures. eStudies used videoiEEG for analysis.
EEG = electroencephalography.
spread, the postictal language disturbance correlated with the hemisphere of seizure origin in all our 26 patients. Gotman [18] used computer techniques to analyze interhemispheric spread of complex partial seizures recorded with depth electrodes and found that the hemisphere of seizure origin retained an electrophysiological “influence” over the contralateral discharge throughout the seizure. Our findings indicate that this influence extends to behavioral aspects of the seizure and persists into the postictal period. Jackson first noted that language disturbances were occasionally associated with seizures from the dominant hemisphere {5}. Studies by Bingley 171, Serafetinides and Falconer 181, and King and Ajmone-Marsan {9] found significant correlations between ictal or postictal dysphasia and dominant-hemisphere seizure foci; Bingley [7] and Serafetinides and Falconer [Sl also noted an association between ictal speech automatisms and nondominant seizure foci (Table 2). Later studies using video/EEG analysis of seizures [lo-12) confirmed the results of earlier investigators and increased detection of language abnormalities by careful review of video tapes. In our study, 74% of patients with dominant TLCPS had postictal paraphasic errors, and every patient with paraphasic errors had dominant TLCPS. Similarly, 33% of patients with nondominant TLCPS read correctly during the ictal EEG discharge (equivalent to formed ictal speech in earlier studies), and every patient who read during the EEG discharge had nondominant TLCPS. Our technique of measuring the time to read added a substantial amount of information because we were able to test 26 of 28 patients with 100% accuracy. One of the 2 patients we could not test adequately had only secondarily generalized seizures from a temporal lobe focus, and the second patient had all four of his seizures during the early
morning hours and was not tested within our 60-second limit. We did not analyze time to read in patients with generalized tonic-clonic seizures because these were followed by prolonged alteration of consciousness, which was beyond the time limits of our staff to perform the testing. Furthermore, we did not analyze testing in patients with extratemporal complex partial seizures. In the study by Koerner and Laxer [l I}, patients with extratemporal seizures did not have postictal language impairment. The 2 patients presented in detail demonstrate that subclinical electrographic seizures and simple partial seizures involving only dominanthemisphere mesial temporal structures do not produce language disturbances that are measureable using our techniques. Thus, if a patient does not demonstrate postictal language disturbance, other data should be used to determine whether the patient has experienced a nondominant TLCPS, a simple partial seizure, or an extratemporal CPS. We speculate that the brain region responsible for postictal language disturbances in dominant TLCPS is in the ipsilateral posterior perisylvian region. The 2 patients described in detail showed ictal EEG discharges without altered consciousness involving dominant-hemisphere mesial temporal structures and no language disturbance. During a complex partial seizure with identical EEG onset, however, both patients showed substantial delay in reading the test phrase and 1 patient had prominent postictal paraphasic errors. Thus, isolated mesial temporal ictal activity alone did not cause the language dysfunction detected by our technique. Stimulation studies have demonstrated interference with reading when electrically stimulating posterior-inferior frontal, posterior perisylvian, or basal temporal regions 121-231. We do not believe the basal Privitera et al: Postictal Language Assessment
395
temporal language area is responsible for postictal language dysfunction because the first patient we described in detail had most, if not all, of the basal temporal language area removed, using measurements recently reported by Burnstine and associates E241. Despite this, the patient had no language disturbance after mesial temporal electrographic seizures, yet had typical dominant temporal lobe postictal language dysfunction after a clinical complex partial seizure with lateral temporal spread. It is unlikely that the inferior frontal language area produced the postictal language dysfunction we saw because dominant-hemisphere extratemporal seizures rarely produce postictal language dysfunction rl11. Although we believe the posterior temporal region is responsible for postictal language disturbances in patients with TLCPS, it is possible that multiple language areas are impaired simultaneously in the postictal period. Detailed study of more patients may provide more information on the localization of postictal language dysfunction. The language assessment we report has important practical application as an additional noninvasive test to use in evaluating patients for surgical treatment of epilepsy. At most surgical centers, if there is inadequate data from noninvasive studies, then ictal video/ EEG monitoring with invasive (depth or subdural) electrodes is used. Invasive recordings provide a high degree of accuracy, but increase the risk, cost, and duration of the presurgical evaluation {25]. Our findings show that postictal language testing provides accurate data about seizure lateralization when the epileptogenic focus is in the temporal lobe and, in our series, this was the single most accurate test in lateralizing a temporal lobe focus. Our simple testing method cam be taught to nurses or EEG technologists in a few minutes so that an epileptologist or personnel trained in comprehensive language evaluation need not be present. This is helpful when patients have medications withdrawn and are monitored throughout a 24-hour period. The brief, repeatable nature of the test makes it easier to analyze from videotapes because the examiner knows the words he is listening for and may be able to detect subtle language problems despite background noise in the patient room. We thank the EEG technologists and nurses of the Epilepsy Monitoring Unit of the University of Cincinnati Hospital (Cincinnati, O H ) whose dedication and attention to detail helped make this study possible.
References 1. Engel J. Outcome with respect to epileptic seizures. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987:5 53-5 72 2. Engel J. Approaches to localization of the epileptogenic lesion. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987:75-95
3. So N, Gloor P, Quesney LF, et al. Depth electrode investigations in patients with bitemporal epileptiform abnormalities. Ann Neurol 1989;25:423-431 4. Engel J. Appendix 11: presurgical evaluation protocols. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987:669-698 5. Jackson JH. Relations of different divisions of the central nervous system to one another and to parts of the body. Lancet 1898;1:79-87 6. Hecaen H, Piercy M. Paroxysmal dysphasia and the problem of cerebral dominance. J Neurol Neurosurg Psychiatry 1956; 19:194-201 7 Bingley T. Mental symptoms in temporal lobe epilepsy and temporal lobe gliomas. Acta Psychiatr Neurol (Suppl) 1958; 120:95-101 8 Serafetinides EA, Falconer MA. Speech disturbances in temporal lobe seizures: A study in 100 epileptic patients submitted to anterior temporal lobectomy. Brain 1963;86:333-346 9. King DW, Ajmone-Marsan C. Clinical features and ictal patterns in epileptic patients with EEG temporal lobe foci. Ann Neurol 1977;2:138-147 10. McKeever M, Holmes GL, Russman BS. Speech abnormalities in seizures: a comparison of absence and partial complex seizures. Brain Lang 1983;19:25-32 11. Koerner M, Laxer KD. Ictal speech, postictal language dysfunction, and seizure lateralization. Neurology 1988;38:634636 12. Gabr M, Lueders H, Dinner D, et al. Speech manifestations in laterahation of temporal lobe seizures. Ann Neurol 1989; 25:82-87 13. Goodglass H , Kaplan E. The Boston Diagnostic Aphasia Examination. Philadelphia: Lea and Febiger, 1972 14. mass DW. Electroencephalographic manifestations of complex partial seizures. In:Penry JK, Daly DD, eds. Advances in neurology, vol 11. New York: Raven Press, 1975:113-140 15. Blume WT, Young GB, LemieuxJF. EEG morphology of partial epileptic seizures. Electroencephalogr Clin Neurophysiol 1984; 57:295-302 16. Kaibara M, Blume WT. The postictal electroencephalogram. Electroencephalogr Clin Neurophysiol 1988;70:99-104 17. Rosen B. Fundamentals of biostatistics, second edition. Boston: Duxbury Press, 1986:333-338 18. Gotman J. Interhemispheric interactions in seizures of focal onset: data from human intracranial recordings. Electroencephalogr Clin Neurophysiol 1987;67:120-133 19. Lieb JP, Hoque K, Skomer CE, Song XW. Inter-hemispheric propagation of human mesial temporal lobe seizures: A coherenceiphase analysis. Electroencephalogr Clin Neurophysiol 1987;67:101-1 19 20. Spencer SS, Williamson PD, Spencer DD, Mattson RH. Human hippocampal seizure spread studied by depth and subdural recording: the hippocampal commissure. Epilepsia 1987;28: 479-489 21. Penfield W, Rasmussen T. Localization and arrest of speech. Arch Neurol Psychiatry 1949;61:21-27 22. Lueders H, Lesser RP, Hahn J, et al. Basal temporal language area demonstrated by electrical stimulation. Neurology 1986; 36:505-510 23. Lueders H, Lesser RP, Dinner DS, et al. Localization of cortical function: new information from extraoperative monitoring of patients with epilepsy. Epilepsia 1988;29(suppl 2):S56-S65 24. Burnstine TH, Lesser RP, Hart J, et al. Characterization of the basal temporal language area in patients with left temporal lobe epilepsy. Neurology 1990;40:966-970 25. Van Buren JM. Complications of surgical procedures in the diagnosis and treatment of epilepsy. In: Engel J, ed. Surgical treatment of the epilepsies. New York: Raven Press, 1987: 465-476
396 Annals of Neurology VoI 30 No 3 September 1991
