- Email: [email protected]
Comparison of localizing values of various diagnostic tests in non-lesional medial temporal lobe epilepsy
Seizure 1999; 8: 465–470
Article No. seiz.1999.0344, available online at http://www.idealibrary.com on
Comparison of localizing values of various diagnostic tests in non-lesional medial temporal lobe epilepsy ∗
∗
†
‡
§
YOUNG-JE SON , CHUN-KEE CHUNG , SANG-KUN LEE , KEE HYUN CHANG , DONG SOO LEE , ¶ ∗ YUNG NAHN YI & HYUN JIB KIM ∗
Department of Neurosurgery, ; † Department of Neurology, ; ‡ Department of Diagnostic Radiology, ; of Nuclear Medicine, Seoul National University College of Medicine, Seoul, Korea; ¶ Neural Systems Research, Seoul, Korea § Department
Correspondence to: Chun-Kee Chung, Department of Neurosurgery, Seoul National University College of Medicine, Seoul, Korea, 28 Yongon-dong, Chongno-gu, Seoul 110-744, South Korea, E-mail: [email protected]
Though the surgical treatment for medial temporal lobe epilepsy yields a high success rate, more studies are needed in order to determine the most efficacious pre-operative algorithm. The authors studied the relationship between surgical outcome and the localization results of various pre-operative diagnostic tests to assess the predictive value. Seventy-one consecutive patients who had undergone anterior temporal lobectomy with amygdalohippocampectomy with the diagnosis of non-lesional medial temporal lobe epilepsy, who had been followed up more than 24 months, were analyzed retrospectively. Electroencephalogy (EEG), magnetic resonance imaging (MRI), proton emission tomography (PET), single photon emission computed tomography (SPECT), the Wada test, and neuropsychological testing were analyzed. There was no diagnostic test that was found to have a statistically significant relationship between Engel Class I outcome and localization results (P > 0.05). SPECT, neuropsychological testing, and the Wada test all had less predictive values (P < 0.01). EEG and PET had comparable predictive values for Engel Class I with MRI (P > 0.05). No single diagnostic test alone is sufficient to make a diagnosis of non-lesional medial temporal lobe epilepsy. MRI, EEG and PET had comparable predictive values for Engel Class I. SPECT, neuropsychological testing, and the Wada test had less predictive values. c 1999 BEA Trading Ltd
Key words: diagnosis; localization; outcome; surgery; non-lesional temporal lobe epilepsy.
INTRODUCTION The paradigm for assessing patients prior to epilepsy surgery varies widely depending on each individual institution. In order to decrease medical costs and to decrease the pre-operative evaluation period, we need to determine the most efficacious pre-operative algorithm and eliminate redundant tests1 . Anterior temporal lobectomy with amygdalohippocampectomy is a safe and effective surgical treatment widely performed for medically refractory medial temporal lobe epilepsy. Though the surgical treatment for medial temporal lobe epilepsy results in a high success rate, more studies are needed to determine the most efficacious pre-operative algorithm. A literature review suggests that proton emission tomography (PET) and magnetic resonance imaging (MRI) are much more sensitive to mesial temporal scle1059–1311/99/080465 + 06
$12.00/0
rosis (MTS) than single photon emission computed tomography (SPECT)2–10 . MRI-identified unilateral hippocampal atrophy11–13 , restrictive temporal lobe hypometabolism by PET analysis14 , febrile convulsions11 , Wada memory performance15 , and electroencephalography (EEG) localization to temporal lobe only11, 16 were found to be significant predictors of good post-operative outcome, using bivariate analysis. By using multivariate analysis, the percentage of epileptic EEG activity arising from the site of resection, and either imaging localization or lack of the use of invasive monitoring, were the only statistically significant pre-operative predictors for good outcome at 2 years17 . Therefore, diagnostic tests with statistically significant prognostic importance are thought to be MRI, PET, the Wada test, and EEG. The diagnostic tests, however, were not always redundant. c 1999 BEA Trading Ltd
466
Therefore, the fact that there has been no definitive single diagnostic indicator for medial temporal lobe epilepsy until now has led the authors to study the relationships between surgical outcome and the localization results of various diagnostic tests, in order to assess the predictive value of each diagnostic test. We hypothesized that MRI is the most important diagnostic test and that EEG, PET, SPECT, the Wada test, neuropsychological testing, and even ictal EEG contribute no additional diagnostic efficacy.
MATERIALS AND METHODS Patients Seventy-one consecutive patients who underwent standard anterior temporal lobectomy along with amygdalohippocampectomy with the diagnosis of medial temporal lobe epilepsy were studied. Patients in whom extrahippocampal focal epileptogenic lesions, such as tumors, vascular malformations, developmental anomalies, and malacia were seen on MR images were excluded. The group of patients included 45 males and 26 females, ranging in age from 11 to 51 years (mean age, 28.9 years). All the patients were followed for more than 24 months post-operatively at Seoul National University Hospital Comprehensive Epilepsy Program from September 1994 to March 1999. The mean duration of follow-up was 38.4 months, ranging from 24 to 60 months.
Pre-operative evaluation protocol Evaluation protocol included a routine history, physical and neurological examination, prolonged interictal EEG, ictal video-EEG monitoring (of at least three usual seizures), MRI for specific hippocampal sequence, PET, interictal and ictal SPECT, the Wada test, and neuropsychological testing. MRI protocol is described in detail by Cheon et al.18 . Briefly, all MRI studies were performed on either 1.5-T or 1.0-T units and T1-weighted sagittal sequences and proton density- and T2-weighted axial fast spin-echo sequences were obtained with 5 mm section thicknesses. Oblique coronal images were obtained perpendicular to the long axis of the hippocampus with 2-mm-thick T1-weighted 3D magnetization-prepared rapid acquisition gradientecho sequences and 3-mm-thick T2-weighted fast spin-echo sequences. The MR diagnosis of hippocampal sclerosis was based on the presence of either unilateral atrophy, high T2 signal intensity of the hippocampus, or both, and on the absence of any other abnormality.
Y.-J. Son et al.
Lee et al.19 have described PET protocol in detail. Briefly, after preliminary transmission scans with use of germanium-68 (Ge-68), patients underwent injection of 10 mCi (370 MBq) F-18fluorodeoxyglucose (FDG). PET was begun 40 minutes after F-18 FDG had been administered. Lee et al.19 have described SPECT protocol in detail. Briefly, interictal SPECT was done at 10 minutes after intravenous administration of 1110 MBq of Technetium-99m hexamethylepropyleneamine oxime (HMPAO). Patients who were to undergo ictal SPECT were monitored with synchronized video and EEG in the Epilepsy Monitoring Unit (EMU). Soon after seizure onset, Tc-99m HMPAO was administered intravenously and ictal SPECT were acquired from 30 minutes to 2 hours after the seizure calmed down. The Wada test was done for evaluation of language and memory with concurrent scalp EEG recording. After puncture of the femoral artery, a catheter tip was positioned 5 cm 43 distal to the carotid bifurcation, at first ipsilaterally to the presumed epileptic focus. Digital subtraction angiography was performed with pressurized injection of contrast medium. Then, while the patient raised both arms and loudly counted backward from 100, a bolus of 80–120 mg sodium amobarbital was injected through the internal carotid artery over a period of 3 to 5 seconds, until the contralateral hemiplegia was obtained. Ten stimulus cards were used to evaluate memory function: five common objects, one word, one phrase, one colored object, one geographic object, and one mathematical expression. Each card was presented a sufficient number of times on the side contralateral to hemiparesis. While the memory stimuli were being presented, evaluation of spontaneous speech and comprehension was also performed. The language function was lateralized by the presence of spontaneous speech and comprehension. The criteria of memory failure were below five items recalled in dominant side injection and below six items recalled in non-dominant side injection. The criterion of significant interside difference was that the asymmetry score be above four items recalled. After confirming normalization of language, speech, hemiparesis, and EEG abnormality, memory function was assessed by free recall and multiple choice recognition. Injection on the other side was performed after the EEG was normalized and at least 30 minutes after the first injection. Neuropsychological testing included general intellectual assessment, tests of attention, memory, language and reasoning, visuospatial or perceptual testing, affective testing and sensory-motor examination through clinical interviews. Seven of the 71 patients required invasive monitoring because of indeterminate or discordant results on the pre-surgical tests.
Diagnostic tests in non-lesional medial TLE
Terms used to describe determination of localization
467
RESULTS Pathology
If each diagnostic test resulted in localization or lateralization of the surgically resected medial temporal lobe, the term true is used; localization or lateralization to the opposite site, false; neither localization nor lateralization, inconclusive; and unchecked.
Statistical analysis Surgical outcome was classified according to the classification of J. Engel20 . Class I (‘seizure free’) includes those who have been seizure free since surgery, as well as those who experienced auras only. It also includes atypical generalized seizures with withdrawal of antiepileptic medication, or at least 2 years of seizure-free remission with no further seizures up to the most recent follow-up. Class II indicates rare seizures. Class III indicates a worthwhile improvement, i.e. at least a 90% improvement in seizure frequency compared with pre-operative status. Class IV denotes no worthwhile improvement. The sensitivity estimations were acquired by dividing the numbers of the true localized or lateralized Engel Class I group by the numbers of the Engel Class I group including the inconclusive group, and the true and false localized or lateralized groups. The specificity estimations were acquired by dividing the numbers of the false localized or lateralized Engel Class II–IV group by the numbers of the Engel Class II– IV group. Logistic regression techniques were applied to assess the relationships between pre-operative predictors and surgical outcome. A chi-squared test was used to evaluate each diagnostic test vs. seizure outcome. The logit log linear model was used for comparing MRI with the other diagnostic tests. This logistic regression method can be used to describe the relationship between a dichotomous response variable and a set of explanatory variables. Given a sample of n subjects, let Y j denote the binary response variate of interest and χ j = (χ1 j , χ2 j , . . . , χk j ) be the vector for explanatory variables corresponding to the jth individual selected in the sample. βi is the regression coefficient of the jth individual. Let the conditional probability that the response outcome is present, Pr = {Y j = 1|X j = χi }, be denoted by Pr (χ j ), then the logistic regression model can be written as below21, 22 . Pk exp β0 + i=1 βi χi j . Pr (χi j ) = Pk 1 + exp β0 + i=1 βiχi j In our use of the logit log–linear model, P < 0.05 was considered statistically significant.
The surgically excised temporal lobes showed varying degrees of hippocampal neuronal loss with gliosis and contained no pathologically verified mass lesions. Sixty-one patients had hippocampal sclerosis. Fifty patients showed cortical dysplasia in the temporal lobe. One patient had no pathologic report. One patient had gliosis only, and one patient showed an old contusion. Forty-seven patients had hippocampal sclerosis with mild cortical dysplasia. Three patients with hippocampal sclerosis were found to have normal findings on MRI.
Surgical outcome Sixty-six patients (92.9%) who have been seizure free since surgery had Engel Class I outcomes. Two patients (2.8%) who had rare seizures had Engel Class II outcomes. Three patients (4.3%) who had worthwhile improvement in seizure frequency, compared with their pre-operative status, had Engel Class III outcomes (Table 1). Two patients had post-operative hemiparesis with intracranial hemorrhage and two patients suffered from surgical wound infections. Thus the mortality was 0% and the permanent morbidity was 2.8% (Table 2). The sensitivities for various diagnostic tests were as follows: MRI, 93%; PET, 83%; interictal EEG, 87%; video-EEG, 89%; ictal SPECT, 76%; Wada test, 60%; neuropsychological test, 55%; interictal SPECT, 37%. The specificities were as follows: MRI, 13%; PET, 25%; interictal EEG, 13%; video-EEG, 40%; ictal SPECT, 25%; Wada test, 33%; neuropsychological test, 50%; interictal SPECT, 75% (Table 3). Also, the positive predictive values for these tests were as follows: MRI, 89%; PET, 89%; interictal EEG, 89%; video-EEG, 95%; ictal SPECT, 91%; Wada test, 90%; neuropsychological test, 90%; interictal SPECT, 92%. The negative predictive values for these tests were as follows: MRI, 20%; PET, 17%; interictal EEG, 11%; video-EEG, 22%; ictal SPECT, 9%; Wada test, 7%; neuropsychological test, 12%; interictal SPECT, 13% (Table 4). No statistically significant relationships were found between the surgical outcomes and the localization results from the various diagnostic tests (R-by-C table, P > 0.05) (Table 5). However, in the logit log–linear model comparing MRI with the other tests, SPECT, neuropsychological testing, and the Wada test had less predictive values (P < 0.01). Interictal EEG, video-EEG,
468
Y.-J. Son et al.
and PET had comparable predictive values for Engel Class I with MRI (P > 0.05) (Table 6). Like the cases in Casino et al.13 , most cases in our series could be localized using only MRI and EEG. However, even the false localized group (N-8, Engel Class I: 87.5%) had comparable prognoses to the true localized group (N = 63, Engel Class I: 93%) (Tables 7 and 8). Therefore, we conclude that cases with good prognoses could be lost if only EEG and MRI are used pre-operatively.
Diagnostic test
P-value
Interictal EEG Video-EEG MRI PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
0.987 0.079 0.723 0.561 0.978 0.502 0.863 0.738
Table 6: The logit log–linear model for comparison with MRI.
Table 1: Post-operative outcome. Engel Class
No.
Percentage
I II III IV
66 2 3 0
92.9 2.8 4.3 0
Table 2: Post-operative complications.
Mortality Morbidity Post-operative hemiparesis Wound infection
Table 5: The result of chi-squared testing (each diagnostic test and Engel Class I vs. Engel Class II–IV).
Percentage
No. (n = 71)
0.0 5.6 2.8 2.8
0 4 2 2
Permanent complications.
Diagnostic test
P-value
Interictal EEG Video-EEG PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
0.4924 0.2030 0.2678 0.0236 0.0001 0.0001 0.0001
Table 7: Localization with interictal EEG or video-EEG monitoring, and MRI (n = 63). Engel Class
No.
Percentage
I II III IV
59 1 3 0
93.7 1.6 4.7 0
Table 8: No localization with interictal EEG or video-EEG monitoring, and MRI (n = 8). Table 3: Sensitivities and specificities.
Engel Class
Diagnostic test
Sensitivities
Specificities
Interictal EEG Video-EEG MRI PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
87% (55/63) 89% (56/63) 93% (59/63) 83% (50/60) 76% (31/41) 37% (23/62) 55% (18/33) 60% (37/62)
13% (1/8) 40% (2/5) 13% (1/8) 25% (2/8) 25% (1/4) 75% (6/8) 50% (2/4) 33% (2/6)
Table 4: Positive and negative predictive values. Diagnostic test Interictal EEG Video-EEG MRI PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
Positive predictive value n 89% (55/62) 95% (56/59) 89% (59/66) 89% (50/56) 91% (31/34) 92% (23/25) 90% (18/20) 90% (37/41)
I II III IV
No.
Percentage
7 1 0 0
87.5 12.5 0 0
DISCUSSION Precise localization and removal of the epileptogenic focus is the ultimate treatment goal in medically intractable epilepsy patients. Until now, no single diagnostic test has been shown to be solely sufficient in localizing the site to be surgically resected. During the 40 years that EEG was used to localize epileptogenic foci, false localization commonly occurred. The irritative zone may correspond to the site of seizure onset, but it does not always approximate the critical epileptogenic brain region23–25 . After the use of depth electrodes, video monitoring, and subdural electrodes began, diagnostic accuracy increased. However, other problems concerning time and cost arose. In the
Diagnostic tests in non-lesional medial TLE
studies of interictal techniques by Spencer et al.9, 10 , PET had the highest diagnostic sensitivity in temporal lobe epilepsy (TLE) (84% vs. 66% for SPECT, 55% for qualitative MRI, 71% for quantitative MRI) while SPECT has the highest sensitivity in extratemporal epilepsy (ETE) (60% vs. 43% for MRI and 33% for PET). The highest diagnostic sensitivity and specificity were achieved by ictal imaging with SPECT (90% in TLE, 81% in ETE). The techniques, however, were not always redundant. One reason for the wide discrepancy of results between TLE and ETE might be the differing pathologic substrates. A literature review suggests that PET and MRI are much more sensitive to MTS than SPECT (100%, 95% vs. 70%)2–8 . The use of these modalities can save the expense of pre-operative EEG evaluation, and invasive and risky diagnostic tests. For example, the use of depth electrodes and subdural electrodes can be omitted when using PET and MRI. The principle of PET is based on the functional state of hypometabolic interictal epileptogenic foci, and its sensitivity is high, although there is no definite anatomical abnormality. Because spatial discrimination has improved recently, PET’s sensitivity is approximately 85%, and its specificity approaches 90% to 98%14, 26 . On the other hand, MRI is sensitive when there are structural changes27 . In medial TLE, it has been found that the sensitivity of MRI ranges from 75% to 86%, and its specificity ranges from 64% to 100%9, 10, 12, 13 . In our study, the specificity of MRI was 13% and the specificity of PET was 25%. These numbers are lower than those in other studies. There could be several explanations for this, but case selection bias may be the biggest influence. Another possibility may be due to our protocol. Even when EEG, MRI or PET failed to localize concordantly, we did perform the surgery if two of the three tests succeeded to localize and the other test was inconclusive. We also performed surgery if invasive monitoring resulted in conclusive data, even when two of the three tests succeeded in localizing and the other test had a discordant result. Another possible reason for our lower numbers may be the presence of hippocampal sclerosis without MRI-detectable abnormality. The last possible reason is that the false localized group included an inconclusive group as well as the false localized group itself, unlike other studies. Low negative predictive values may also have been influenced by the same reasons. Our study differs from others in the criteria for diagnosing medical temporal lobe epilepsy. In other studies, the correct localization was based on EEG results only10, 13, 28 . Theoretically, however, to diagnose medial temporal lobe epilepsy, the patient should be seizure free after removal of the medial temporal lobe. This study is based on the premise that only patients who are seizure free after removal of
469
the medial temporal lobe had medial temporal lobe epilepsy pre-operatively. Patients who are not seizurefree post-operatively did not have medial temporal lobe epilepsy. Our data were analyzed with surgical outcome as a gold standard. In this study, we evaluated the localizing power of each diagnostic test retrospectively. In medial temporal lobe epilepsy, we concluded that interictal EEG, video-EEG, MRI, and PET are basically necessary to localize epileptogenic foci pre-operatively, because any one of them alone does not completely predict the measure of seizure relief after surgery. Hence, these tests supplement one another. However, ictal SPECT had less diagnostic power than the above tests. A review of the published literature suggests that interictal SPECT is less reliable than PET for detection of medial temporal lobe epilepsy. However, it is important to note that further studies are needed because ictal SPECT has diagnostic validity, especially in extra-temporal epilepsy, and may be more accurate in diagnosing epilepsy associated with developmental lesions and tumors10 . In addition, we think that while pre-operative Wada tests or neuropsychological tests have less diagnostic power, they are necessary in predicting the post-operative changes of cognitive and memory functions. Although the Wada test identifies the language-dominant hemisphere and can evaluate memory in the contralateral temporal lobe to assure the safety of performing anterior temporal lobectomy, it is not always used to lateralize the seizure focus29–32 . Neuropsychological testing is employed during pre-operative evaluation for epilepsy surgery to measure functional deficits associated with unilateral seizure onset15 . Also, the outcomes of neuropsychological testing may be related to post-operative seizure outcome, although the results have been mixed33 . At present, various diagnostic tests should be used to localize the epileptic focus correctly. In medial temporal lobe epilepsy, the reason is that hippocampal sclerosis may not be the only pathologic finding, and the cause of hippocampal sclerosis is obscure. In addition, the standard anterior temporal lobectomy with amygdalohippocampectomy as our protocol of surgical treatment can include the temporal neocortex. Therefore, although patients who are diagnosed with medial temporal lobe epilepsy are seizure free after the surgical treatment, lateral TLE cannot be ruled out. In addition, if many variables are examined in a large number of patients, and the indications and limits of each diagnostic test are clear, costs can be reduced.
CONCLUSION We found that no diagnostic test was sufficient to diagnose medial temporal lobe epilepsy by itself. MRI had
470
the highest diagnostic sensitivity in non-lesional medial temporal lobe epilepsy. However, it is suggested that interictal EEG, video-EEG and PET be performed with MRI. SPECT, neuropsychological tests and the Wada test were found to have less predictive value and limited diagnostic value.
ACKNOWLEDGEMENT This work was supported by a grant from the Ministry of Health and Welfare, Korea (HMP-99-N-02-0003).
REFERENCES 1. Bronen, R. MR imaging in epilepsy. AES News Summer 1998; 3–5. 2. Hajek, M., Siegel, A., Haldemann, R., von Schulthess, G. K. and Wieser, H. Value of HMPAO-SPECT in selective temporal lobe surgery for epilepsy. Journal of Epilepsy 1991; 4: 43–51. 3. Markand, O., Shen, W., Park, H. M., Siddiqui, A. R., Wellman, H. H. and Worth, R. H. Single photon imaging computed tomography (SPECT) for localization of epileptogenic focus in patients with intractable complex partial seizures. Epilepsy Research 1992; 5: 121–126. 4. Newton, M. R., Berkovic, S. F., Austin, M. C., Reutens, D. C., Mckay, W. J. and Bladin, P. F. A postictal switch in blood flow distribution characterizes human temporal lobe seizures. Journal of Neurology, Neurosurgery and Psychiatry 1992; 55: 884–891. 5. Rowe, C. C., Berkovic, S. F., Sia, S. T. B. et al. Localization of epileptic foci with postictal single photon emission computed tomography. Annals of Neurology 1989; 26: 660–668. 6. Shen, W., Lee, B. I., Park, H. M. et al. HIPDM-SPECT brain imaging in the presurgical evaluation of patients with intractable seizures. Journal of Nuclear Medicine 1990; 31: 1280–1284. 7. Stefan, H., Bauer, J., Feistel, H. et al. Regional cerebral blood flow during focal seizures of temporal and frontocentral onset. Annals of Neurology 1990; 27: 162–166. 8. Grunwald, F., Durwen, H. F., Bockisch, A. et al. Technetium99-m-HMPAO brain SPECT in medically intractable temporal lobe epilepsy: a post-operative evaluation. Journal of Nuclear Medicine 1991; 32: 388–394. 9. Spencer, S. S. The relative contributions of MRI, SPECT, and PET imaging in epilepsy. Epilepsia 1994; 35 (Suppl. 6): 72– 89. 10. Spencer, S. S., Theodore, W. H. and Berkovic, S. F. Clinical applications: MRI, SPECT, and PET. Magnetic Resonance Imaging 1995; 13 (Suppl. 8): 1119–1124. 11. Hufnagel, A., Elger, C. E., Pels, H. et al. Prognostic significance of ictal and interictal epileptiform activity in temporal lobe epilepsy. Epilepsia 1994; 35 (Suppl. 6): 1146–1153. 12. Cascino, G. D., Trenerry, M. R., Jack, C. R. Jr. et al. Electrocorticography and temporal lobe epilepsy: relationship to quantitative MRI and operative outcome. Epilepsia 1995; 36 (Suppl. 7): 692–696. 13. Cascino, G. D., Trenerry, M. R., So, E. L. et al. Routine EEG and temporal lobe epilepsy: relation to long-term EEG monitoring, quantitative MRI, and operative outcome. Epilepsia 1996; 37 (Suppl. 7): 651–656.
Y.-J. Son et al. 14. Manno, E. M., Sperling, M. R., Ding, X. et al. Predictors of outcome after anterior temporal lobectomy: positron emission tomography. Neurology 1994; 44 (Suppl. 12): 2331–2336. 15. Loring, D. W., Meador, K. J., Lee, G. P. et al. Wada memory performance predicts seizure outcome following anterior temporal lobectomy. Neurology 1994; 44 (Suppl. 12): 2322–2324. 16. Chung, M. Y., Walczak, T. S., Lewis, D. V., Dawson, D. V. and Radtke, R. Temporal lobectomy and independent bitemporal interictal activity: what degree of lateralization is sufficient? Epilepsia 1991; 32 (Suppl. 2): 195–201. 17. Armon, C., Radtke, R. A., Friedman, A. H. and Dawson, D. V. Predictors of outcome of epilepsy surgery: multivariate analysis with validation. Epilepsia 1996; 37 (Suppl. 9): 814–821. 18. Cheon, J. E., Chang, K. H., Kim, H. D. et al. MR of hippocampal sclerosis: comparison of qualitative and quantitative assessments. American Journal of Neuroradiology 1988; 19: 465–468. 19. Lee, D. S., Seo, J. M., Lee, J. S. et al. Diagnosis of ictal hyperperfusion using subtraction image of ictal and interictal brain perfusion SPECT. Korean Journal of Nuclear Medicine 1998; 32: 20–31. 20. Engel, J. Jr., Van Ness, P. C., Rasmussen, T. B. and Ojemann, I. M. Outcome with respect to epileptic seizures. In: Surgical Treatment of the Epilepsies. 2nd Edition (Ed. J. Jr. Engel). New York, Raven Press, 1993: pp. 609–621. 21. Hardeo, S. Statistics in Epidemiology. Boca Raton, FL, CRC Press, 1996: pp. 215–218; 265–268. 22. Thomas, H. Understanding Biostatistics. Mosby-Year Book, pp. 100–102. 23. Luders, H. O., Engel, J. Jr. and Munari, C. General principles. In: Surgical Treatment of the Epilepsies. 2nd Edition (Ed. J. Jr. Engel). New York, Raven Press, 1993: pp. 137–153. 24. Quesney, L. F. Clinical and EEG features of complex partial seizure of temporal lobe origin. Epilepsia 1986; 26: 27–46. 25. Sperling, J. R. and Engel, J. Jr. Electroencephalographic recording from the temporal lobes: a comparison of ear, anterior temporal and nasopharyngeal electrodes. Annals of Neurology 1985; 17: 510–513. 26. Suzanne, T. M., Walter, E. D., Robin, L. G. et al. Prospective localization of epileptogenic foci: comparison of PET and SPECT with site of surgery and clinical outcome. Radiology 1996; 199: 375–380. 27. Berkovic, S. F., McIntosh, A. M., Kalnins, R. M. et al. Preoperative MRI predicts outcome of temporal lobectomy: an actuarial analysis. Neurology 1995; 45 (Suppl. 7): 1358–1363. 28. Radhakrishnan, K., So, E. L., Silbert, P. L. et al. Predictors of outcome of anterior temporal lobectomy for intractable epilepsy. Neurology 1998; 51: 465–471. 29. Wada, J. and Rasmussen, T. Intracarotid injection of sodium amytal for the lateralization of cerebral speech dominance: experimental and clinical observation. Journal of Neurosurgery 1960; 17: 266–282. 30. Milner, B., Branch, C. and Rasmussen, T. Study of short-term memory after intracarotid injection of sodium amytal. Transactions of the American Neurology Association 1962; 87: 224– 226. 31. Perrine, K., Westerveld, M. , Sass, K. J. et al. Wada memory disparities predict seizure laterality and postoperative seizure control. Epilepsia 1995; 36 (Suppl. 9): 851–856. 32. Sperling, M. R., Saykin, A. J., Glosser, G. et al. Predictors of outcome after anterior temporal lobectomy. The intracarotid amobarbital test. Neurology 1994; 44 (Suppl. 12): 2325–2330. 33. Dodrill, C. B., Hermann, B. P., Rausch, R. et al. Neuropsychological testing for assessing prognosis following surgery for epilepsy. In: Surgical Treatment of the Epilepsies. 2nd Edition (Ed. J. Jr. Engel). New York, Raven Press, 1993: pp. 263– 271.
Article No. seiz.1999.0344, available online at http://www.idealibrary.com on
Comparison of localizing values of various diagnostic tests in non-lesional medial temporal lobe epilepsy ∗
∗
†
‡
§
YOUNG-JE SON , CHUN-KEE CHUNG , SANG-KUN LEE , KEE HYUN CHANG , DONG SOO LEE , ¶ ∗ YUNG NAHN YI & HYUN JIB KIM ∗
Department of Neurosurgery, ; † Department of Neurology, ; ‡ Department of Diagnostic Radiology, ; of Nuclear Medicine, Seoul National University College of Medicine, Seoul, Korea; ¶ Neural Systems Research, Seoul, Korea § Department
Correspondence to: Chun-Kee Chung, Department of Neurosurgery, Seoul National University College of Medicine, Seoul, Korea, 28 Yongon-dong, Chongno-gu, Seoul 110-744, South Korea, E-mail: [email protected]
Though the surgical treatment for medial temporal lobe epilepsy yields a high success rate, more studies are needed in order to determine the most efficacious pre-operative algorithm. The authors studied the relationship between surgical outcome and the localization results of various pre-operative diagnostic tests to assess the predictive value. Seventy-one consecutive patients who had undergone anterior temporal lobectomy with amygdalohippocampectomy with the diagnosis of non-lesional medial temporal lobe epilepsy, who had been followed up more than 24 months, were analyzed retrospectively. Electroencephalogy (EEG), magnetic resonance imaging (MRI), proton emission tomography (PET), single photon emission computed tomography (SPECT), the Wada test, and neuropsychological testing were analyzed. There was no diagnostic test that was found to have a statistically significant relationship between Engel Class I outcome and localization results (P > 0.05). SPECT, neuropsychological testing, and the Wada test all had less predictive values (P < 0.01). EEG and PET had comparable predictive values for Engel Class I with MRI (P > 0.05). No single diagnostic test alone is sufficient to make a diagnosis of non-lesional medial temporal lobe epilepsy. MRI, EEG and PET had comparable predictive values for Engel Class I. SPECT, neuropsychological testing, and the Wada test had less predictive values. c 1999 BEA Trading Ltd
Key words: diagnosis; localization; outcome; surgery; non-lesional temporal lobe epilepsy.
INTRODUCTION The paradigm for assessing patients prior to epilepsy surgery varies widely depending on each individual institution. In order to decrease medical costs and to decrease the pre-operative evaluation period, we need to determine the most efficacious pre-operative algorithm and eliminate redundant tests1 . Anterior temporal lobectomy with amygdalohippocampectomy is a safe and effective surgical treatment widely performed for medically refractory medial temporal lobe epilepsy. Though the surgical treatment for medial temporal lobe epilepsy results in a high success rate, more studies are needed to determine the most efficacious pre-operative algorithm. A literature review suggests that proton emission tomography (PET) and magnetic resonance imaging (MRI) are much more sensitive to mesial temporal scle1059–1311/99/080465 + 06
$12.00/0
rosis (MTS) than single photon emission computed tomography (SPECT)2–10 . MRI-identified unilateral hippocampal atrophy11–13 , restrictive temporal lobe hypometabolism by PET analysis14 , febrile convulsions11 , Wada memory performance15 , and electroencephalography (EEG) localization to temporal lobe only11, 16 were found to be significant predictors of good post-operative outcome, using bivariate analysis. By using multivariate analysis, the percentage of epileptic EEG activity arising from the site of resection, and either imaging localization or lack of the use of invasive monitoring, were the only statistically significant pre-operative predictors for good outcome at 2 years17 . Therefore, diagnostic tests with statistically significant prognostic importance are thought to be MRI, PET, the Wada test, and EEG. The diagnostic tests, however, were not always redundant. c 1999 BEA Trading Ltd
466
Therefore, the fact that there has been no definitive single diagnostic indicator for medial temporal lobe epilepsy until now has led the authors to study the relationships between surgical outcome and the localization results of various diagnostic tests, in order to assess the predictive value of each diagnostic test. We hypothesized that MRI is the most important diagnostic test and that EEG, PET, SPECT, the Wada test, neuropsychological testing, and even ictal EEG contribute no additional diagnostic efficacy.
MATERIALS AND METHODS Patients Seventy-one consecutive patients who underwent standard anterior temporal lobectomy along with amygdalohippocampectomy with the diagnosis of medial temporal lobe epilepsy were studied. Patients in whom extrahippocampal focal epileptogenic lesions, such as tumors, vascular malformations, developmental anomalies, and malacia were seen on MR images were excluded. The group of patients included 45 males and 26 females, ranging in age from 11 to 51 years (mean age, 28.9 years). All the patients were followed for more than 24 months post-operatively at Seoul National University Hospital Comprehensive Epilepsy Program from September 1994 to March 1999. The mean duration of follow-up was 38.4 months, ranging from 24 to 60 months.
Pre-operative evaluation protocol Evaluation protocol included a routine history, physical and neurological examination, prolonged interictal EEG, ictal video-EEG monitoring (of at least three usual seizures), MRI for specific hippocampal sequence, PET, interictal and ictal SPECT, the Wada test, and neuropsychological testing. MRI protocol is described in detail by Cheon et al.18 . Briefly, all MRI studies were performed on either 1.5-T or 1.0-T units and T1-weighted sagittal sequences and proton density- and T2-weighted axial fast spin-echo sequences were obtained with 5 mm section thicknesses. Oblique coronal images were obtained perpendicular to the long axis of the hippocampus with 2-mm-thick T1-weighted 3D magnetization-prepared rapid acquisition gradientecho sequences and 3-mm-thick T2-weighted fast spin-echo sequences. The MR diagnosis of hippocampal sclerosis was based on the presence of either unilateral atrophy, high T2 signal intensity of the hippocampus, or both, and on the absence of any other abnormality.
Y.-J. Son et al.
Lee et al.19 have described PET protocol in detail. Briefly, after preliminary transmission scans with use of germanium-68 (Ge-68), patients underwent injection of 10 mCi (370 MBq) F-18fluorodeoxyglucose (FDG). PET was begun 40 minutes after F-18 FDG had been administered. Lee et al.19 have described SPECT protocol in detail. Briefly, interictal SPECT was done at 10 minutes after intravenous administration of 1110 MBq of Technetium-99m hexamethylepropyleneamine oxime (HMPAO). Patients who were to undergo ictal SPECT were monitored with synchronized video and EEG in the Epilepsy Monitoring Unit (EMU). Soon after seizure onset, Tc-99m HMPAO was administered intravenously and ictal SPECT were acquired from 30 minutes to 2 hours after the seizure calmed down. The Wada test was done for evaluation of language and memory with concurrent scalp EEG recording. After puncture of the femoral artery, a catheter tip was positioned 5 cm 43 distal to the carotid bifurcation, at first ipsilaterally to the presumed epileptic focus. Digital subtraction angiography was performed with pressurized injection of contrast medium. Then, while the patient raised both arms and loudly counted backward from 100, a bolus of 80–120 mg sodium amobarbital was injected through the internal carotid artery over a period of 3 to 5 seconds, until the contralateral hemiplegia was obtained. Ten stimulus cards were used to evaluate memory function: five common objects, one word, one phrase, one colored object, one geographic object, and one mathematical expression. Each card was presented a sufficient number of times on the side contralateral to hemiparesis. While the memory stimuli were being presented, evaluation of spontaneous speech and comprehension was also performed. The language function was lateralized by the presence of spontaneous speech and comprehension. The criteria of memory failure were below five items recalled in dominant side injection and below six items recalled in non-dominant side injection. The criterion of significant interside difference was that the asymmetry score be above four items recalled. After confirming normalization of language, speech, hemiparesis, and EEG abnormality, memory function was assessed by free recall and multiple choice recognition. Injection on the other side was performed after the EEG was normalized and at least 30 minutes after the first injection. Neuropsychological testing included general intellectual assessment, tests of attention, memory, language and reasoning, visuospatial or perceptual testing, affective testing and sensory-motor examination through clinical interviews. Seven of the 71 patients required invasive monitoring because of indeterminate or discordant results on the pre-surgical tests.
Diagnostic tests in non-lesional medial TLE
Terms used to describe determination of localization
467
RESULTS Pathology
If each diagnostic test resulted in localization or lateralization of the surgically resected medial temporal lobe, the term true is used; localization or lateralization to the opposite site, false; neither localization nor lateralization, inconclusive; and unchecked.
Statistical analysis Surgical outcome was classified according to the classification of J. Engel20 . Class I (‘seizure free’) includes those who have been seizure free since surgery, as well as those who experienced auras only. It also includes atypical generalized seizures with withdrawal of antiepileptic medication, or at least 2 years of seizure-free remission with no further seizures up to the most recent follow-up. Class II indicates rare seizures. Class III indicates a worthwhile improvement, i.e. at least a 90% improvement in seizure frequency compared with pre-operative status. Class IV denotes no worthwhile improvement. The sensitivity estimations were acquired by dividing the numbers of the true localized or lateralized Engel Class I group by the numbers of the Engel Class I group including the inconclusive group, and the true and false localized or lateralized groups. The specificity estimations were acquired by dividing the numbers of the false localized or lateralized Engel Class II–IV group by the numbers of the Engel Class II– IV group. Logistic regression techniques were applied to assess the relationships between pre-operative predictors and surgical outcome. A chi-squared test was used to evaluate each diagnostic test vs. seizure outcome. The logit log linear model was used for comparing MRI with the other diagnostic tests. This logistic regression method can be used to describe the relationship between a dichotomous response variable and a set of explanatory variables. Given a sample of n subjects, let Y j denote the binary response variate of interest and χ j = (χ1 j , χ2 j , . . . , χk j ) be the vector for explanatory variables corresponding to the jth individual selected in the sample. βi is the regression coefficient of the jth individual. Let the conditional probability that the response outcome is present, Pr = {Y j = 1|X j = χi }, be denoted by Pr (χ j ), then the logistic regression model can be written as below21, 22 . Pk exp β0 + i=1 βi χi j . Pr (χi j ) = Pk 1 + exp β0 + i=1 βiχi j In our use of the logit log–linear model, P < 0.05 was considered statistically significant.
The surgically excised temporal lobes showed varying degrees of hippocampal neuronal loss with gliosis and contained no pathologically verified mass lesions. Sixty-one patients had hippocampal sclerosis. Fifty patients showed cortical dysplasia in the temporal lobe. One patient had no pathologic report. One patient had gliosis only, and one patient showed an old contusion. Forty-seven patients had hippocampal sclerosis with mild cortical dysplasia. Three patients with hippocampal sclerosis were found to have normal findings on MRI.
Surgical outcome Sixty-six patients (92.9%) who have been seizure free since surgery had Engel Class I outcomes. Two patients (2.8%) who had rare seizures had Engel Class II outcomes. Three patients (4.3%) who had worthwhile improvement in seizure frequency, compared with their pre-operative status, had Engel Class III outcomes (Table 1). Two patients had post-operative hemiparesis with intracranial hemorrhage and two patients suffered from surgical wound infections. Thus the mortality was 0% and the permanent morbidity was 2.8% (Table 2). The sensitivities for various diagnostic tests were as follows: MRI, 93%; PET, 83%; interictal EEG, 87%; video-EEG, 89%; ictal SPECT, 76%; Wada test, 60%; neuropsychological test, 55%; interictal SPECT, 37%. The specificities were as follows: MRI, 13%; PET, 25%; interictal EEG, 13%; video-EEG, 40%; ictal SPECT, 25%; Wada test, 33%; neuropsychological test, 50%; interictal SPECT, 75% (Table 3). Also, the positive predictive values for these tests were as follows: MRI, 89%; PET, 89%; interictal EEG, 89%; video-EEG, 95%; ictal SPECT, 91%; Wada test, 90%; neuropsychological test, 90%; interictal SPECT, 92%. The negative predictive values for these tests were as follows: MRI, 20%; PET, 17%; interictal EEG, 11%; video-EEG, 22%; ictal SPECT, 9%; Wada test, 7%; neuropsychological test, 12%; interictal SPECT, 13% (Table 4). No statistically significant relationships were found between the surgical outcomes and the localization results from the various diagnostic tests (R-by-C table, P > 0.05) (Table 5). However, in the logit log–linear model comparing MRI with the other tests, SPECT, neuropsychological testing, and the Wada test had less predictive values (P < 0.01). Interictal EEG, video-EEG,
468
Y.-J. Son et al.
and PET had comparable predictive values for Engel Class I with MRI (P > 0.05) (Table 6). Like the cases in Casino et al.13 , most cases in our series could be localized using only MRI and EEG. However, even the false localized group (N-8, Engel Class I: 87.5%) had comparable prognoses to the true localized group (N = 63, Engel Class I: 93%) (Tables 7 and 8). Therefore, we conclude that cases with good prognoses could be lost if only EEG and MRI are used pre-operatively.
Diagnostic test
P-value
Interictal EEG Video-EEG MRI PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
0.987 0.079 0.723 0.561 0.978 0.502 0.863 0.738
Table 6: The logit log–linear model for comparison with MRI.
Table 1: Post-operative outcome. Engel Class
No.
Percentage
I II III IV
66 2 3 0
92.9 2.8 4.3 0
Table 2: Post-operative complications.
Mortality Morbidity Post-operative hemiparesis Wound infection
Table 5: The result of chi-squared testing (each diagnostic test and Engel Class I vs. Engel Class II–IV).
Percentage
No. (n = 71)
0.0 5.6 2.8 2.8
0 4 2 2
Permanent complications.
Diagnostic test
P-value
Interictal EEG Video-EEG PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
0.4924 0.2030 0.2678 0.0236 0.0001 0.0001 0.0001
Table 7: Localization with interictal EEG or video-EEG monitoring, and MRI (n = 63). Engel Class
No.
Percentage
I II III IV
59 1 3 0
93.7 1.6 4.7 0
Table 8: No localization with interictal EEG or video-EEG monitoring, and MRI (n = 8). Table 3: Sensitivities and specificities.
Engel Class
Diagnostic test
Sensitivities
Specificities
Interictal EEG Video-EEG MRI PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
87% (55/63) 89% (56/63) 93% (59/63) 83% (50/60) 76% (31/41) 37% (23/62) 55% (18/33) 60% (37/62)
13% (1/8) 40% (2/5) 13% (1/8) 25% (2/8) 25% (1/4) 75% (6/8) 50% (2/4) 33% (2/6)
Table 4: Positive and negative predictive values. Diagnostic test Interictal EEG Video-EEG MRI PET Ictal SPECT Interictal SPECT Neuropsychological test Wada test
Positive predictive value n 89% (55/62) 95% (56/59) 89% (59/66) 89% (50/56) 91% (31/34) 92% (23/25) 90% (18/20) 90% (37/41)
I II III IV
No.
Percentage
7 1 0 0
87.5 12.5 0 0
DISCUSSION Precise localization and removal of the epileptogenic focus is the ultimate treatment goal in medically intractable epilepsy patients. Until now, no single diagnostic test has been shown to be solely sufficient in localizing the site to be surgically resected. During the 40 years that EEG was used to localize epileptogenic foci, false localization commonly occurred. The irritative zone may correspond to the site of seizure onset, but it does not always approximate the critical epileptogenic brain region23–25 . After the use of depth electrodes, video monitoring, and subdural electrodes began, diagnostic accuracy increased. However, other problems concerning time and cost arose. In the
Diagnostic tests in non-lesional medial TLE
studies of interictal techniques by Spencer et al.9, 10 , PET had the highest diagnostic sensitivity in temporal lobe epilepsy (TLE) (84% vs. 66% for SPECT, 55% for qualitative MRI, 71% for quantitative MRI) while SPECT has the highest sensitivity in extratemporal epilepsy (ETE) (60% vs. 43% for MRI and 33% for PET). The highest diagnostic sensitivity and specificity were achieved by ictal imaging with SPECT (90% in TLE, 81% in ETE). The techniques, however, were not always redundant. One reason for the wide discrepancy of results between TLE and ETE might be the differing pathologic substrates. A literature review suggests that PET and MRI are much more sensitive to MTS than SPECT (100%, 95% vs. 70%)2–8 . The use of these modalities can save the expense of pre-operative EEG evaluation, and invasive and risky diagnostic tests. For example, the use of depth electrodes and subdural electrodes can be omitted when using PET and MRI. The principle of PET is based on the functional state of hypometabolic interictal epileptogenic foci, and its sensitivity is high, although there is no definite anatomical abnormality. Because spatial discrimination has improved recently, PET’s sensitivity is approximately 85%, and its specificity approaches 90% to 98%14, 26 . On the other hand, MRI is sensitive when there are structural changes27 . In medial TLE, it has been found that the sensitivity of MRI ranges from 75% to 86%, and its specificity ranges from 64% to 100%9, 10, 12, 13 . In our study, the specificity of MRI was 13% and the specificity of PET was 25%. These numbers are lower than those in other studies. There could be several explanations for this, but case selection bias may be the biggest influence. Another possibility may be due to our protocol. Even when EEG, MRI or PET failed to localize concordantly, we did perform the surgery if two of the three tests succeeded to localize and the other test was inconclusive. We also performed surgery if invasive monitoring resulted in conclusive data, even when two of the three tests succeeded in localizing and the other test had a discordant result. Another possible reason for our lower numbers may be the presence of hippocampal sclerosis without MRI-detectable abnormality. The last possible reason is that the false localized group included an inconclusive group as well as the false localized group itself, unlike other studies. Low negative predictive values may also have been influenced by the same reasons. Our study differs from others in the criteria for diagnosing medical temporal lobe epilepsy. In other studies, the correct localization was based on EEG results only10, 13, 28 . Theoretically, however, to diagnose medial temporal lobe epilepsy, the patient should be seizure free after removal of the medial temporal lobe. This study is based on the premise that only patients who are seizure free after removal of
469
the medial temporal lobe had medial temporal lobe epilepsy pre-operatively. Patients who are not seizurefree post-operatively did not have medial temporal lobe epilepsy. Our data were analyzed with surgical outcome as a gold standard. In this study, we evaluated the localizing power of each diagnostic test retrospectively. In medial temporal lobe epilepsy, we concluded that interictal EEG, video-EEG, MRI, and PET are basically necessary to localize epileptogenic foci pre-operatively, because any one of them alone does not completely predict the measure of seizure relief after surgery. Hence, these tests supplement one another. However, ictal SPECT had less diagnostic power than the above tests. A review of the published literature suggests that interictal SPECT is less reliable than PET for detection of medial temporal lobe epilepsy. However, it is important to note that further studies are needed because ictal SPECT has diagnostic validity, especially in extra-temporal epilepsy, and may be more accurate in diagnosing epilepsy associated with developmental lesions and tumors10 . In addition, we think that while pre-operative Wada tests or neuropsychological tests have less diagnostic power, they are necessary in predicting the post-operative changes of cognitive and memory functions. Although the Wada test identifies the language-dominant hemisphere and can evaluate memory in the contralateral temporal lobe to assure the safety of performing anterior temporal lobectomy, it is not always used to lateralize the seizure focus29–32 . Neuropsychological testing is employed during pre-operative evaluation for epilepsy surgery to measure functional deficits associated with unilateral seizure onset15 . Also, the outcomes of neuropsychological testing may be related to post-operative seizure outcome, although the results have been mixed33 . At present, various diagnostic tests should be used to localize the epileptic focus correctly. In medial temporal lobe epilepsy, the reason is that hippocampal sclerosis may not be the only pathologic finding, and the cause of hippocampal sclerosis is obscure. In addition, the standard anterior temporal lobectomy with amygdalohippocampectomy as our protocol of surgical treatment can include the temporal neocortex. Therefore, although patients who are diagnosed with medial temporal lobe epilepsy are seizure free after the surgical treatment, lateral TLE cannot be ruled out. In addition, if many variables are examined in a large number of patients, and the indications and limits of each diagnostic test are clear, costs can be reduced.
CONCLUSION We found that no diagnostic test was sufficient to diagnose medial temporal lobe epilepsy by itself. MRI had
470
the highest diagnostic sensitivity in non-lesional medial temporal lobe epilepsy. However, it is suggested that interictal EEG, video-EEG and PET be performed with MRI. SPECT, neuropsychological tests and the Wada test were found to have less predictive value and limited diagnostic value.
ACKNOWLEDGEMENT This work was supported by a grant from the Ministry of Health and Welfare, Korea (HMP-99-N-02-0003).
REFERENCES 1. Bronen, R. MR imaging in epilepsy. AES News Summer 1998; 3–5. 2. Hajek, M., Siegel, A., Haldemann, R., von Schulthess, G. K. and Wieser, H. Value of HMPAO-SPECT in selective temporal lobe surgery for epilepsy. Journal of Epilepsy 1991; 4: 43–51. 3. Markand, O., Shen, W., Park, H. M., Siddiqui, A. R., Wellman, H. H. and Worth, R. H. Single photon imaging computed tomography (SPECT) for localization of epileptogenic focus in patients with intractable complex partial seizures. Epilepsy Research 1992; 5: 121–126. 4. Newton, M. R., Berkovic, S. F., Austin, M. C., Reutens, D. C., Mckay, W. J. and Bladin, P. F. A postictal switch in blood flow distribution characterizes human temporal lobe seizures. Journal of Neurology, Neurosurgery and Psychiatry 1992; 55: 884–891. 5. Rowe, C. C., Berkovic, S. F., Sia, S. T. B. et al. Localization of epileptic foci with postictal single photon emission computed tomography. Annals of Neurology 1989; 26: 660–668. 6. Shen, W., Lee, B. I., Park, H. M. et al. HIPDM-SPECT brain imaging in the presurgical evaluation of patients with intractable seizures. Journal of Nuclear Medicine 1990; 31: 1280–1284. 7. Stefan, H., Bauer, J., Feistel, H. et al. Regional cerebral blood flow during focal seizures of temporal and frontocentral onset. Annals of Neurology 1990; 27: 162–166. 8. Grunwald, F., Durwen, H. F., Bockisch, A. et al. Technetium99-m-HMPAO brain SPECT in medically intractable temporal lobe epilepsy: a post-operative evaluation. Journal of Nuclear Medicine 1991; 32: 388–394. 9. Spencer, S. S. The relative contributions of MRI, SPECT, and PET imaging in epilepsy. Epilepsia 1994; 35 (Suppl. 6): 72– 89. 10. Spencer, S. S., Theodore, W. H. and Berkovic, S. F. Clinical applications: MRI, SPECT, and PET. Magnetic Resonance Imaging 1995; 13 (Suppl. 8): 1119–1124. 11. Hufnagel, A., Elger, C. E., Pels, H. et al. Prognostic significance of ictal and interictal epileptiform activity in temporal lobe epilepsy. Epilepsia 1994; 35 (Suppl. 6): 1146–1153. 12. Cascino, G. D., Trenerry, M. R., Jack, C. R. Jr. et al. Electrocorticography and temporal lobe epilepsy: relationship to quantitative MRI and operative outcome. Epilepsia 1995; 36 (Suppl. 7): 692–696. 13. Cascino, G. D., Trenerry, M. R., So, E. L. et al. Routine EEG and temporal lobe epilepsy: relation to long-term EEG monitoring, quantitative MRI, and operative outcome. Epilepsia 1996; 37 (Suppl. 7): 651–656.
Y.-J. Son et al. 14. Manno, E. M., Sperling, M. R., Ding, X. et al. Predictors of outcome after anterior temporal lobectomy: positron emission tomography. Neurology 1994; 44 (Suppl. 12): 2331–2336. 15. Loring, D. W., Meador, K. J., Lee, G. P. et al. Wada memory performance predicts seizure outcome following anterior temporal lobectomy. Neurology 1994; 44 (Suppl. 12): 2322–2324. 16. Chung, M. Y., Walczak, T. S., Lewis, D. V., Dawson, D. V. and Radtke, R. Temporal lobectomy and independent bitemporal interictal activity: what degree of lateralization is sufficient? Epilepsia 1991; 32 (Suppl. 2): 195–201. 17. Armon, C., Radtke, R. A., Friedman, A. H. and Dawson, D. V. Predictors of outcome of epilepsy surgery: multivariate analysis with validation. Epilepsia 1996; 37 (Suppl. 9): 814–821. 18. Cheon, J. E., Chang, K. H., Kim, H. D. et al. MR of hippocampal sclerosis: comparison of qualitative and quantitative assessments. American Journal of Neuroradiology 1988; 19: 465–468. 19. Lee, D. S., Seo, J. M., Lee, J. S. et al. Diagnosis of ictal hyperperfusion using subtraction image of ictal and interictal brain perfusion SPECT. Korean Journal of Nuclear Medicine 1998; 32: 20–31. 20. Engel, J. Jr., Van Ness, P. C., Rasmussen, T. B. and Ojemann, I. M. Outcome with respect to epileptic seizures. In: Surgical Treatment of the Epilepsies. 2nd Edition (Ed. J. Jr. Engel). New York, Raven Press, 1993: pp. 609–621. 21. Hardeo, S. Statistics in Epidemiology. Boca Raton, FL, CRC Press, 1996: pp. 215–218; 265–268. 22. Thomas, H. Understanding Biostatistics. Mosby-Year Book, pp. 100–102. 23. Luders, H. O., Engel, J. Jr. and Munari, C. General principles. In: Surgical Treatment of the Epilepsies. 2nd Edition (Ed. J. Jr. Engel). New York, Raven Press, 1993: pp. 137–153. 24. Quesney, L. F. Clinical and EEG features of complex partial seizure of temporal lobe origin. Epilepsia 1986; 26: 27–46. 25. Sperling, J. R. and Engel, J. Jr. Electroencephalographic recording from the temporal lobes: a comparison of ear, anterior temporal and nasopharyngeal electrodes. Annals of Neurology 1985; 17: 510–513. 26. Suzanne, T. M., Walter, E. D., Robin, L. G. et al. Prospective localization of epileptogenic foci: comparison of PET and SPECT with site of surgery and clinical outcome. Radiology 1996; 199: 375–380. 27. Berkovic, S. F., McIntosh, A. M., Kalnins, R. M. et al. Preoperative MRI predicts outcome of temporal lobectomy: an actuarial analysis. Neurology 1995; 45 (Suppl. 7): 1358–1363. 28. Radhakrishnan, K., So, E. L., Silbert, P. L. et al. Predictors of outcome of anterior temporal lobectomy for intractable epilepsy. Neurology 1998; 51: 465–471. 29. Wada, J. and Rasmussen, T. Intracarotid injection of sodium amytal for the lateralization of cerebral speech dominance: experimental and clinical observation. Journal of Neurosurgery 1960; 17: 266–282. 30. Milner, B., Branch, C. and Rasmussen, T. Study of short-term memory after intracarotid injection of sodium amytal. Transactions of the American Neurology Association 1962; 87: 224– 226. 31. Perrine, K., Westerveld, M. , Sass, K. J. et al. Wada memory disparities predict seizure laterality and postoperative seizure control. Epilepsia 1995; 36 (Suppl. 9): 851–856. 32. Sperling, M. R., Saykin, A. J., Glosser, G. et al. Predictors of outcome after anterior temporal lobectomy. The intracarotid amobarbital test. Neurology 1994; 44 (Suppl. 12): 2325–2330. 33. Dodrill, C. B., Hermann, B. P., Rausch, R. et al. Neuropsychological testing for assessing prognosis following surgery for epilepsy. In: Surgical Treatment of the Epilepsies. 2nd Edition (Ed. J. Jr. Engel). New York, Raven Press, 1993: pp. 263– 271.