Journal of Neuro-Ophthalmology, March 2004, Volume 24, Issue 1

Occipital Hypometabolism Demonstrated by Positron Emission Tomography After Temporal Lobectomy for Refractory Epilepsy

BACKGROUND: Epilepsy surgery involves well-planned discrete injury to the brain and may create visual deficits. This study seeks to evaluate the indirect effects of temporal lobectomy on brain metabolism by correlating visual field defects and glucose metabolism in the visual cortex of patients before and after undergoing epilepsy surgery. METHODS: A retrospective survey of 11 patients who had undergone temporal lobectomy for refractory epilepsy in a single institution from 1986 to 1989, and who had pre-lobectomy and post-lobectomy visual field examinations and F-18 2-fluorodeoxyglucose positron emission tomography (FDG-PET) as part of a standard comprehensive epilepsy surgery evaluation. The PET images were analyzed to provide a correlation with the visual field defects that developed after the temporal lobectomy. RESULTS: Occipital hypometabolism in the absence of structural lesions of the occipital lobe was noted in seven patients with contralateral visual field defects and in one of four patients without a visual field defect. FDG-PET studies in three patients repeated for as long as 20 months after lobectomy showed no significant change in the occipital hypometabolism pattern. CONCLUSIONS: Although the occipital cortex was not directly injured during temporal lobectomy, the resulting hypometabolism correlates with the clinical findings of visual field defects. The hypometabolism may be due to deafferentation after interruption of the optic pathways and appears to be persistent.

ORIGINAL CONTRIBUTION Occipital Hypometabolism Demonstrated by Positron Emission Tomography After Temporal Lobectomy for Refractory Epilepsy Franklin C. L. Wong, MD, PhD, JD, Barbara E. Swartz, MD, PhD, Manyee Gee, PhD, and Mark Mandelkern, MD, PhD Background: Epilepsy surgery involves well- planned dis­crete injury to the brain and may create visual deficits. This study seeks to evaluate the indirect effects of temporal lobectomy on brain metabolism by correlating visual field defects and glucose metabolism in the visual cortex of pa­tients before and after undergoing epilepsy surgery. Methods: A retrospective survey of 11 patients who had undergone temporal lobectomy for refractory epilepsy in a single institution from 1986 to 1989, and who had pre-lobectomy and post- lobectomy visual field examinations and F- 18 2- fluorodeoxyglucose positron emission tomog­raphy ( FDG- PET) as part of a standard comprehensive epi­lepsy surgery evaluation. The PET images were analyzed to provide a correlation with the visual field defects that de­veloped after the temporal lobectomy. Results: Occipital hypometabolism in the absence of structural lesions of the occipital lobe was noted in seven patients with contralateral visual field defects and in one of four patients without a visual field defect. FDG- PET studies in three patients repeated for as long as 20 months after lobectomy showed no significant change in the occipital hypometabolism pattern. Conclusions: Although the occipital cortex was not di­rectly injured during temporal lobectomy, the resulting hy­pometabolism correlates with the clinical findings of visual field defects. The hypometabolism may be due to deaffer-entation after interruption of the optic pathways and ap­pears to be persistent. ( JNeuro- Ophthalmol 2004; 24: 19- 23) From the Neurology Service ( FCLW, BES, MG) and PET Center ( MM), West Los Angeles VAMC, Los Angeles, California; the University of Texas, M. D. Anderson Cancer Center, Houston, Texas ( FCLW); Uni­versity Hospitals of Cleveland, Case Western University, Cleveland, Ohio ( BES); and Deptartment of Physics, University of California at Irvine, California ( MM). Address correspondence to Dr. Franklin C. Wong, University of Texas, M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Box 59, Houston, TX 77030. E- mail: Fwong@ di. mdacc. tmc. edu. Cerebral glucose metabolism, which correlates with synaptic activity in the brain, may be evaluated by pos­itron emission tomography ( PET) using F- 18 fluorodeoxy-glucose ( FDG) ( 1,2). Altered patterns of regional cerebral metabolism of glucose after cerebral insults are most prominent in the primary site of focal injury ( 3,4), but distal effects such as crossed cerebellar diaschisis are commonly seen ( 5,6,7). Remote metabolic effects have been noted in themesial temporal lobe ipsilateral to an ictal focus ( 8) and contralateral to a temporal lobe resection ( 8,9). The effects of most cerebral injuries on the cerebral metabolic pattern are often not clear because of the lack of a pre- injury baseline metabolic study and uncertainty about the extent of the injury or preexisting disease. However, patients who are evaluated for epilepsy surgery often have baseline FDG- PET studies. Occasionally, they also have post- lobectomy FDG- PET scans to confirm the removal of the epileptic focus. Such patients provide an opportunity to study the regional metabolic effects of a discrete, planned cerebral injury by comparing the FDG- PET scans before and after lobectomy. After lobectomy, visual field defects are often identified ( 10- 13) because of damage to the Mey­er's loop portion of the optic pathway traversing the tem­poral lobe. Therefore, the correlation between FDG- PET changes and visual field defects after lobectomy is a good model for the effect of a cerebral lesion on cerebral metabo­lism in relation to specific neural deficits. Hemifield visual stimulation has demonstrated left-right asymmetry in the local cerebral metabolic rate of glu­cose ( 13). The local cerebral metabolic rate of glucose in­creases by 10- 15% in response to opening a patched eye to white light, by 20% to the viewing of a checkerboard, and by 45- 60% to the viewing of a park scene ( 14). Studies of blind patients have demonstrated that metabolic changes in the striate cortex depend on the age of onset of blindness. Late- onset blindness ( in children older than 3 years) is as­sociated with a 10- 15% decrease in the occipital local ce­rebral metabolic rate of glucose ( 15), while early- onset blindness ( in children less than 2 years of age) is associated J Neuro- Ophthalmol, Vol. 24, No. 1, 2004 19 JNeuro- Ophthalmol, Vol. 24, No. 1, 2004 Wong et al. with a 10- 20% increase ( 15), a difference that is explained by neuronal reorganization. Occipital metabolic dysfunction has been correlated with specific visual field defects in patients with ischemic lesions of the optic radiations and/ or occipital cortex ( 16). Strict quantitative measurement of cerebral glucose re­quires invasive procedures, including arterial catheteriza­tion. Such rigorous procedures are mostly limited to studies of normal volunteers in cerebral PET research programs. For patient comfort and compliance, a less invasive semi­quantitative measure, the laterality index ( LI), has been used to measure the asymmetry of metabolic activity ( 16). It is semi- quantitative because it is a ratio that carries no unit and does not measure the absolute glucose metabolic rate. Yet LI is sufficient to compare the metabolic activities of different parts of the brain with PET scans at one point in time. It also allows for the comparison of PET scans from one point in time to another. After initial cerebral infarction, patients with a relatively small LI are likely to have suffered a primary lesion outside of the visual cortex and have a good chance of subsequent improvement in visual field de­fects ( 17). Patients with a relatively large LI probably have had primary damage to the visual cortex itself, so that the visual field defects are less likely to recover ( 17). Patients who undergo temporal lobectomy for epi­lepsy are generally young and unlikely to have other cere­bral lesions. Furthermore, the surgical lesion is limited and well- documented. Therefore, they constitute an ideal group for correlating metabolic changes in the striate cortex and visual field defects in response to a remote cerebral injury. METHODS Records from 16 consecutive post- temporal lobec­tomy patients treated by the California Comprehensive Epi­lepsy Program were collected over 2 years and analyzed. Records from 11 of these patients were selected because their preoperative visual fields were normal and their post­operative visual fields and FDG- PET studies were avail­able. The absence of occipital lesions was confirmed by magnetic resonance imaging of Tl and T2 sequences. There were seven men and four women, aged 19- 37 years ( mean, 29.8 years). The formal visual field studies were done with the Octopus ( Interzeag, Berne, Switzerland) or Humphrey devices ( Allergan- Humphrey Instruments, Palo Alto, CA). The visual field studies and PET studies were performed with informed consent. The preoperative and postoperative static FDG- PET studies were performed using an ECAT- III ( CTI Inc., Knoxville, TN) within 6 months of the visual field studies. The patients were awake and lying in a quiet room with low ambient light. PET scanning was initiated 40 minutes after injection of approximately 7- 10 mCi of F- 18 FDG. The PET scanner had a planar resolution of 6 mm full- width, half- maximum. The slice thickness was set to 22 mm, and successive slices were separated by 11 mm. The patients were positioned in the gantry so that the scan plane was parallel to the canthomeatal line. Starting at the canthomeatal line, 10 to 12 sections were obtained, each section representing a counting period of several minutes, leading to 1- 2 million true coincidences. To correct for the different times at which the data acquisition was made for the different sections, an average input function was applied to relate different sections. Initial visual inspection of the printed images was made by three physician evaluators masked to the visual field examinations and experienced in interpreting PET scans. The result was based on agreement of at least two of the evaluators. More detailed, semi- quantitative evaluation was performed on six sets of preoperative and postoperative PET scans. These scans were chosen because they provided sufficient technical quality as defined by their coverage of the entire brain, allowing summation of whole- brain radio­activities. According to published methods ( 18- 20), radio­activity measurements involved placing 87 different re­gions of interest ( ROIs) over the entire brain. The ROIs were placed without knowledge of the visual field findings and according to templates from the atlas of Matsui and Hirano ( 18). In the occipital lobe, the regions were VI and V2 ( further subdivided into surface and depth regions) ac­cording to the neuroanatomic divisions of Brodmann. This methodology has been found to have interobserver and in-traobserver correlation coefficients of up to 0.9, depending on the size of the ROI ( 20). The mean radioactivity uptake ( counts/ min/ U volume averaged over the ROI and cor­rected for decay) was obtained for each ROI. To enable semi- quantitative evaluation of the regional metabolic ab­normalities, Lis were calculated as the ratio of the differ­ence ( right - left) to the sum ( right + left) of the radioactivity for each ROI and its contralateral counterpart, according to the method of Bosley ( 16,17). Four patients with homonymous left superior quad­rant visual defects, three with homonymous right superior quadrant visual field defects, and four without any visual defects were studied. None of these patients had significant occipital lesions as demonstrated by magnetic resonance imaging ( Table 1). RESULTS All four patients with left superior quadrant cuts were found to have hypometabolism of the right striate cortex in their postoperative FDG- PET scans ( Fig. 1). All had under­gone a right temporal lobectomy. All three patients with right superior quadrant visual field defects had postopera­tive hypometabolism in the left striate cortex following left temporal lobectomy ( Fig. 2). Of the four patients with no 20 © 2004 Lippincott Williams & Wilkins Occipital Hypometabolism After Temporal Lobectomy JNeuro- Ophthalmol, Vol. 24, No. 1, 2004 TABLE 1. Qualitative evaluation of PET scans in temporal lobectomy patients who had postoperative visual field defects Patient Surgical date Tissue resected Postoperative visual field defect Postoperative PET finding 1 2 3 4 5 6 7 8 9 0 1 5/ 88 10/ 87 11/ 88 11/ 87 12/ 88 9/ 88 3/ 87 8/ 87 9/ 87 9/ 86 2/ 88 R ATL 6 cm R Temp cortectomy R ATL + AH R ATL + AH L ATL + AH L ATL + AH L ATL + AH RATL R ATL + AH LAH R ATL + AH LSQ LSQ LSQ LSQ RSQ RSQ RSQ Normal Normal Normal Normal 03/ 89 R SC Hypometabolism 07/ 88 R SC Hypometabolism 05/ 89 R SC Hypometabolism 08/ 88 R SC Hypometabolism 06/ 89 L SC Hypometabolism 10/ 88 L SC Hypometabolism 08/ 88 L SC Hypometabolism 02/ 88 no SC Asymmetry 04/ 88 no SC Asymmetry 11/ 87 no SC Asymmetry 02/ 88 R SC Hypometabolism ATL, anterior temp, lobectomy; AH, amygdal. hippocampectomy; SC, striate cortex; LSQ, left superior quadrant homonymous hemianopia; RSQ, right superior quadrant homonymous hemianopia. visual field defects, three did not have significant asymme­try in the FDG uptake in the striate cortex, while one was found to have hypometabolism in the right striate cortex. The Fisher's exact probability of the above observation ( contralateral visual field defect versus no visual field de­fects occurring by random chance) was 0.0042 ( Table 1). Patients 2, 4, and 6 underwent a repeat FDG- PET that did not show significant changes in striate cortical hypometab­olism 7 to 20 months after the first postoperative FDG- PET. Repeated visual field examination on return clinic visits showed no change in deficits. These results suggest that the relationship between striate cortex hypometabolism, sur­gery, and visual field deficits is long- lasting. The Lis showed changes similar to those seen in the visual inspection of the scan and corresponded well to lat­erality of the surgery ( Table 2). Only 6 of the 11 patient records had sets of preoperative and postoperative PET scans of sufficient quality ( coverage of the entire brain in both sets of images) to allow semi- quantification. We noted that for patients with postoperative hypometabolism by visual inspection and a visual field defect, the most com­mon abnormal region was the lateral portion of the second­ary visual cortex. The changes in LI after lobectomy were in a similar direction for VI, V2 ( surface), and V2 ( deep), al­though the small sample sizes precluded further analysis ( Table 2). Patient 11 had right striate cortex post- lobectomy hypometabolism ( as seen in Fig. 3 and in the V2 depth re­gion in Table 2) without a visual field cut. We regarded this case as a false positive for the correlation between PET findings and field loss. No significant asymmetry in thalamic metabolism appeared after lobectomy. Patient 1 had right thalamic hy­pometabolism before and after surgery. FIGURE 1. F- 18 2- fluorodeoxyglucose positron emission tomography ( FDG- PET) before and after right temporal lobectomy in Patient 2. Preoperative scan ( left) shows nor­mal metabolic pattern. Postoperatively, the patient had a left superior quadrant homonymous hemianopia. Postop­erative scan ( right) shows right striate cortex hypometab­olism ( blue). FIGURE 2. FDG- PET before and after left temporal lobec­tomy with amygdal- hippocampectomy in Patient 6. Preop­erative scan ( left) is normal. Postoperatively, the patient had a right superior quadrant homonymous hemianopia. Post­operative scan ( right) shows left striate cortex hypometab­olism ( blue). 21 JNeuro- Ophthalmol, Vol. 24, No. 1, 2004 Wong et al. TABLE 2. Semi- quantitative evaluation of glucose metabolism in the visual cortex and thalamus in post- lobectomy patients Patient 1 2 5 6 7 11 Tissue resected RATL RTL LATL LATL LATL RATL Postoperative visual field defect LSQ LSQ RSQ RSQ RSQ Normal VI Pre - 3.7 2.4 3.7 0.0 - 0.6 - 2.0 ( SC) Post - 6.3 50.7 17.7 - 0.3 17.0 - 10.5 V2 Pre - 6.6 - 3.3 3.6 2.2 - 4.1 - 2.4 Laterality ( surface) Post - 18.5 - 82.2 25.2 12.0 0.3 - 4.5 index ( LI)* V2 ( depth) Pre - 3.4 - 6.2 4.1 - 6.0 - 2.9 - 4.1 Post - 10.4 - 29.8 7.0 - 4.9 17.6 - 11.3 Thalamus Pre - 15.4 - 5.6 2.3 0.5 3.9 - Post - 26.9 - 7.3 4.8 3.4 9.0 - 8.7 Normal ranges ( ref 16, 17) (- 4.8- 14.4) (- 9.4- 10.6) (- 9.4- 10.6) (- 10.8- 12.0) * Laterality Index = [ ICMRglc( R) - ICMRglc( L)]/[ LCMRglc( R) + LCMRglc( L)] x 200 SC, striate cortex; ATL, anterior temporal lobectomy; RSQ, right superior quadrant homonymous hemianopia; LSQ, left superior quadrant homony­mous hemianopia. DISCUSSION We found that postoperative visual field defects oc­curred in 7 of 11 patients, a frequency similar to that previ­ously reported ( 11- 13). Our study did not differentiate the effects on visual field defects from anterior temporal lobec­tomy ( ATL) and amygdalohippocampectomy ( AH), partly because of the large degree of overlap ( 7 of 11 patients had both ATL and AH). In light of recent findings of similar effects on visual field deficits from ATL and AH ( 21), such a distinction may not be important. Our study of patients with both ATL and AH demonstrates a frequency of visual field deficits similar to that found by Egan et al. ( 21) in patients who had undergone either ATL or AH. The correlation between visual field defects and oc­cipital hypometabolism is in excellent agreement with the findings of Phelps et al. ( 14) and Bosley et al. ( 16,17), who studied cerebral infarction. In general, the surgical lesions in our patient group are smaller, better defined, and better documented than the infarctions suffered by the patients studied by these authors. Our patients were also less likely to be affected by additional lesions, which could confound the results of the clinical examinations and PET studies. Our data did not show thalamic hypometabolism after lobectomy in comparison with Bosley et al. ( 16,17), who found significant thalamic asymmetries associated with in­farcts remote from the thalamus. There are three possible explanations for the discrepancy. First, in the series of Bos­ley et al. ( 16,17), there may have been infarction or ische­mia of part or all of the affected thalamus, or abnormalities of cortical structures causing secondary hypometabolism of the thalamus. Second, there are differences in the tech­niques we used to evaluate the metabolism of the thalamus and other structures. We measured the average metabolism of the thalamus using multiple sections. The comparison ( 14,16,17) studies sampled the thalamus using a single sec­tion. It is possible that a small region of hypometabolism in, for example, the lateral geniculate body may manifest itself in a single sample but not show up in the measurement of total thalamic metabolism. Third, infarction causes damage in multiple brain locations, which could have diverse re­mote sequelae on metabolism. We believe that the current spatial resolution of PET and the ROI technique precludes reliable measurement of the metabolism of an individual thalamic nucleus and that limited sampling of the thalamus gives unreliable measurements. We have also observed that thalamic hypometabolism is common in subjects with epi­lepsy ( 8) but generally does not involve the geniculate bod- FIGURE 3. FDG- PET before and after left temporal lobec­tomy in Patient 11. Postoperatively, the patient had normal visual fields, but FDG- PET demonstrates right striate cortex hypometabolism ( yellow). There was a conversion of the laterality index ( LI) from normal to abnormal in V2 depth ( see Table 2). The FDG- PET findings are therefore a false-positive correlation with the visual field loss. 22 © 2004 Lippincott Williams & Wilkins Occipital Hypometabolism After Temporal Lobectomy JNeuro- Ophthalmol, Vol. 24, No. 1, 2004 ies. In this study, we observed thalamic hypometabolism both preoperatively and postoperatively in one subject. All post- surgical FDG- PET studies were done at least 3 months after surgery. Thus, the earlier course of occipital hypometabolism secondary to lobectomy was not studied. In the three patients on whom a second postoperative scan was performed, no further changes were seen for up to 20 months after the first scan, suggesting that the postoperative changes are likely to be permanent. Because these changes occur in the absence of further structural damage, deaffer-entation is a reasonable explanation for occipital hypome­tabolism in this setting. Remote metabolic effects and di-aschisis from cerebral injuries have been described by Von Monakow ( 22). These effects may be related to neuronal loss and/ or depressed perfusion. They have recently gained attention, partly because of advances in in vivo imaging techniques such as PET and single photon emission com­puted tomography. These methods allow metabolic studies of primary and remote effects from neurologic lesions, in­cluding the measurement of glucose metabolism, oxygen metabolism, cerebral perfusion, and neuroreceptor density ( 23- 26). An enhanced vascular response, as measured by single photon emission computed tomography, has been demonstrated in crossed cerebellar diaschisis using acet-azolamide, suggesting increased perfusion reserve and pre­served vascular reactivity to a metabolic stimulus in brain tissues affected by the remote stimulus ( 27). Pharmacologi­cal and functional stimulation combined with imaging is likely to provide further opportunities to evaluate remote effects such as those investigated here ( 28). In this study, we have demonstrated a relationship be­tween lesions of the thalamostriate pathway, visual field de­fects, and depression of glucose metabolism in the corre­sponding striate cortex. However, further studies are needed to illuminate the mechanisms of remote effects. De­termination of the vascular reserve of these regions and the evaluation of their neuronal constituency may allow better understanding of these phenomena. REFERENCES 1. Sokoloff L. Relationship among local functional activity, energy metabolism and blood flow in the CNS. Federation Proc 1981 ; 40: 2311- 2316. 2. Lebrun- Grandie P, Baron JC, Sourssaline F, et al. Coupling be­tween regional blood flow and oxygen utilization in the normal brain: a study with positron tomography and oxygen 15. Arch Neu­rol 1983 ; 40: 230- 236. 3. Broich K, Alavi A, Kushner M. Positron emission tomography in cerebrovascular disorders. Sent NuclMed 1992; 22: 224- 232. 4. Ruff RM, Crouch JA, Troster AL, et al. Selected cases of poor out­come following a minor brain trauma: comparing neuropsycholog­ical and positron emission tomography assessment. Brain Inj 1994; 8: 297- 308. 5. Tien RD, Ashdown BC. Crossed cerebellar diaschisis and crossed cerebellar atrophy: correlation of MR findings, clinical symptoms, and supratentorial diseases in 26 patients. Am J Roentgenol 1992; 158: 1155- 1159. 6. Tanaka M, Kondo S, Hirai S, et al. Crossed cerebellar diaschisis accompanied by hemiataxia: a PET study. J Neurol Neurosurg Psy­chiatry 1992; 55: 121- 125. 7. Infeld B, Davis SM, Lichtenstein M, et al. Crossed cerebellar di­aschisis and brain recovery after stroke. Stroke 1995; 26: 90- 95. 8. Sperling MR. Neuroimaging in epilepsy: contribution of MRI, PET and SPECT. Semin Neurol 1990; 10: 349- 356. 9. Hajek M, Wieser HG, Khan N, et al. Preoperative and postoperative glucose consumption in mesiobasal and lateral temporal lobe epi­lepsy. Neurology 1994; 44: 2125- 2132. 10. Jensen I, Seedorff HH. Temporal lobe epilepsy and neuro-ophthalmology. Ophthalmological findings in 74 temporal lobe re­sected patients. Acta Ophthalmol 1976; 54: 827- 881. 11. Katz A, Awad IA, Kong AK, et al. Extent of resection in temporal lobectomy for epilepsy. II. Memory changes and neurological com­plications. Epilepsia 1989; 30: 763- 771. 12. Tecoma ES, Laxer K D, Barbara NM, et al. Frequency and charac­teristics of visual field deficits after surgery for mesial temporal sclerosis. Neurology 1993; 43: 1235- 1238. 13. Reivich M, Greenberg J, Alavi A, et al. The use of 18- F fluorode-oxyglucose technique for mapping functional neural pathways in man. Acta Neurol Scand 1979; 72( suppl): 198- 199. 14. Phelps ME, Mazziotta JC, Kulh DE. Tomographic mapping of hu­man cerebral metabolism: visual stimulation and deprivation. Neu­rology 1981; 31: 517- 529. 15. Veraart C, De Voider AG, Wanet- Defalque MC, et al. Glucose uti­lization in human visual cortex is abnormally elevated in blindness of early onset but decreased in blindness of late onset. Brain Res 1990; 510: 115- 121. 16. Bosley TM, Rosenquist AC, Kushner M, et al. Ischemic lesions of the occipital cortex and optic radiations: positron emission tomog­raphy. Neurology 1985; 35: 470- 484. 17. Bosley TM, DannR, Silver FL, etal. Recovery of vision after ische­mic lesions: Positron Emission Tomography. Ann Neurol 1987; 21: 444^ 150. 18. Matsui T, Hirano A. An atlas of the human brain for computed to­mography. Tokyo: Igaku- shin, 1978. 19. Swartz BE, Halgren E, Fuster JM, et al. Cortical metabolic activa­tion in humans during a visual memory task. Cereb Cortex 1995; 5: 205- 214. 20. Small GW, Stern CE, Mandelkern MA, et al. Reliability of drawing regions of interest for positron emission tomographic data. Psychi-atrRes 1992; 45: 177- 185. 21. Egan RA, Shults WT, So N, et al. Visual field deficits in conven­tional anterior temporal lobectomy versus amygdalohippocampec-tomy. Neurology 2000; 55: 1818- 1822. 22. Pearce JM. Von Monakow and diaschisis. JNeurol Neurosurg Psy­chiatry 1994; 57: 197'. 23. Baron JC, Levasseur M, Mazoyer B, et al. Thalamocortical diaschi­sis: positron emission tomography in humans. J Neurol Neurosurg Psychiatry 1992; 55: 935- 942. 24. Yamauchi H, Fukuyama H, Kimura J. Hemodynamic and metabolic changes in crossed cerebellar hypoperfusion. Stroke 1992; 23: 855- 860. 25. Theodore WH, Sato S, Kufta C, et al. Temporal lobectomy for un­controlled seizures: the role of positron emission tomography. Ann Neurol 1992; 32: 789- 694. 26. Henry TR, Babb TL, Engel J Jr, et al. Hippocampal neuronal loss and regional hypometabolism in temporal lobe epilepsy. Ann Neu­rol 1994; 36: 925- 927. 27. Terada H, Gomi T, Sasaki K, et al. Acetazolamide effect on vascular response in crossed cerebellar diaschisis as measured by 99mTc- HMPAO SPECT [ in Japanese]. Kaku Igaku 1993; 30: 1075- 1080. 28. Swartz BE, Simpkins F, Halgren E, et al. Visual working memory in primary generalized epilepsy: an 18 FDG- PET study. Neurology 1996; 47: 1203- 1212. 23