Journal of Neurology, Neurosurgery, and Psychiatry 1992;55:891-894 891

Postictal switch in blood flow distribution and J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.55.10.891 on 1 October 1992. Downloaded from temporal lobe seizures

Mark R Newton, Samuel F Berkovic, Mark C Austin, Christopher C Rowe, W John McKay, Peter F Bladin

Abstract ictal studies with PET can be performed only The ictal increase of regional cerebral serendipitously when seizures happen to occur blood flow has yet to be fully utilised in the during the scan. SPECT radionuclides such investigation of focal seizures. Although as o9mTechnetium-hexamethyl propyleneami- single photon emission tomography neoxime (HMPAO) rapidly cross the blood- (SPECT) is being increasingly used in the brain barrier and reside in cerebral tissue for localisation of epileptic foci, the evolution several hours. Isotope can thus be injected and time courses of the peri-ictal perfu- during an observed seizure and the patient can sion changes have yet to be clarified. We then be scanned within the next few hours, performed serial SPECT studies in the with the scan reflecting cerebral blood flow interictal, ictal and immediate postictal (CBF) at the time ofinjection."4These proper- states in 12 patients with refractory tem- ties permit the prospective and serial evalu- poral lobe epilepsy to define the patterns ation of changes in CBF associated with focal and duration of peri-ictal cerebral blood seizures, which typically last less than two flow changes. Visual analysis showed a minutes. constant pattern of unilateral global We previously reported the early postictal increases in temporal lobe perfusion dur- perfusion changes with SPECT, which allowed ing seizures which suddenly switched to a accurate prediction of seizure focus in 69% of pattern of relative mesial temporal (hip- a series of patients with refractory temporal pocampal) hyperperfusion and lateral lobe epilepsy."-3 We have since sought to temporal hypoperfusion in the immediate determine the diagnostic value of ictal and postictal period. Quantitative analysis postictal SPECT scans by examining a series confirmed the visual assessment. Lateral of ictal and postictal paired studies in the same temporal cortex ictal/normal side to side patients with qualitative and quantitative ratios were increased by mean 35-1% (95% image analysis. We found a remarkable differ- confidence interval 21-8% to 48-4%) more ence in the ictal and postictal perfusion in the ictal studies than in the interictal patterns, representing a postictal switch in studies and mesial temporal cortex ratios blood flow distribution. This finding is impor- increased by mean 30-8% (22.4% to tant for clinical diagnosis of seizure localisation 39-2%). In the postictal state, however, and for understanding the pathophysiology of lateral temporal ratios were reduced by temporal lobe epilepsy. mean 7-7% (-15-'8% to 0.4%) compared http://jnnp.bmj.com/ with interictal values, whereas mesial temporal perfusion was man*tained com- Methods pared with the interictal studies. These Twelve patients (mean age 28, (range 16-46) observations provide critical information years; seven women) with intractable unilateral for interpreting scans which can be used temporal lobe epilepsy diagnosed on clinical in the localisation of epileptic foci. This and EEG criteria, and MRI" and neuro-

postictal switch in blood flow patterns psychological data were studied (table 1). Peri- on October 4, 2021 by guest. Protected copyright. may reflect the underlying metabolic pro- ictal SPECT is a routine investigation for cesses ofneuronal activation and recovery prospective therapeutic temporal lobectomy in Department of and have implications for understanding our centre,' 1-13 and the 12 patients analysed Neurology, Austin Hospital, Melbourne, the neurobiology of human epileptic were selected because they had undergone Victoria, Australia seizures. paired SPECT studies for both ictal and M R Newton postictal states. S F Berkovic (7 Neurol Neurosurg Psychiatry CC Rowe 1992;55:891-894) An interictal study was performed when the P F Bladin patient had been seizure free for at least 24 Department of hours, with full medication. The 99mTc- Nuclear Medicine, The abnormal synchronised discharge of a HMPAO was injected with the patient in the Austin Hospital neuronal aggregate that characterises focal resting state (supine, eyes closed, and silent M C Austin J epileptic seizures is accompanied by focal with a minimum of background noise). W McKay Patients were Correspondence to: hyperperfusion''and hypermetabolism.67 continuously monitored with Dr Berkovic, Department of Functional imaging with positron emission video recording and EEG for periods of up to Neurology, Austin Hospital, Heidelberg (Melbourne), tomography (PET) and single photon emission two weeks (with reduced medication) and Victoria 3084, Australia computed tomography (SPECT) has shown a underwent 99Tc-HMPAO injection as soon Received 18 June 1991 variety of peri-ictal blood flow patterns., 8-13 as possible after onset of the seizure. SPECT and in revised form 28 August 1991. presumably reflecting the rapid changes of images derived from injections during or up to Accepted 3 September 1991 blood flow accompanying focal seizures. Peri- 30 seconds after the end of a seizure were 892 Newton, Berkovic, Austin, Rowe, McKay, Bladin

Table I Localisation data in 12 patients studied with ictal and postictal SPECT* J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.55.10.891 on 1 October 1992. Downloaded from Interictal Ictal Postictal Patient Ictal EEG SPECT SPECT SPECT MRIt Pathlogy I Left temporal discharge$ Not lateralised Left Not lateralised Left HS HS 2 Right hippocampal dischargeS Right Right Right Right HS HS 3 Left hippocampal dischargeS Left Left Left Left HS HS 4 Left temporal dischargell Not lateralised Left Left Left tumour Ganglioglioma 5 Left hippocampal dischargeS Left Left Left Normal Normal 6 Left temporal dischargell Left Left Left Left HS HS 7 Unlocalised. Active Right interictal Not lateralised Right Right Right HS HS 8 Left temporal dischargell Left Left Left Left dysplasia Dysplasia 9 Left temporal dischargell Left Left Not lateralised Left HS HS 10 Right temporal dischargell Not lateralised Right Right Right HS Cortical gliosis¶ 11 Right hippocampal dischargeS Right Right Not lateralised Normal Normal 12 Left temporal dischargell Left Left Left Left HS HS HS = hippocampal sclerosis. *SPECT results derived from blinded visual assessment. Interictal SPECT localised by focal hypoperfusion, ictal SPECT by focal hyperperfusion, and postictal SPECT by mesial hyerperfusion with lateral hyoperfusion. tMRI diagnosis of hippocampal sclerosis was made as previously described and validated.5 $Surface EEG. §Depth EEG. llSphenoidal EEG. ¶Hippocampal tissue not found on pathology specimen.

called "ictal studies", as the appearances of showed a pronounced global increase in tem- scans in all cases were consistently similar. All poral lobe blood flow on the epileptogenic side 12 patients underwent an additional injection (figure). This pattern was seen when injections on another occasion more than 30 seconds were given during clinical and EEG ictal after seizure end (postictal study; mean (SD) activity or up to 30 seconds after its cessation delay 1-46 (0-4) min). Exact timing of the and was easily identified in all 12 cases by the injections was made possible by the timer three blinded observers independently (table linked to the video and EEG, which was read 1). Postictal studies showed a switch in dis- on playback of all seizures. The paired seizures tribution of blood flow with hypoperfusion of studied were of a stereotyped pattern and the epileptogenic lobe and hemisphere in all 12 similar length for each patient. Patients were studies and residual hyperperfusion of the scanned within two hours of injection, with a mesial temporal region in seven. The postictal Starcam 400AC gammacamera (General Elec- changes were not always striking. In nine tric, Milwaukee, USA). Data acquisition, patients they were obvious to all three obser- reconstruction, and display methods have been vers but in the three others they were not described previously.'2 sufficiently clear for blinded lateralisation inde- Scans were qualitatively interpreted by three pendent of other data. blinded observers. The blood flow changes Hypoperfusion of the frontal and parietal were quantitated by determining a ratio of cortex was seen ipsilateral to the discharging counts, dividing the number in the epilepto- focus in the ictal studies and persisted into the genic side by that in the normal side, in the postictal period. One patient progressed to following regions of interest; mesial, lateral, hemigeneralisation after his complex partial and posterior temporal regions; lateral frontal seizure. SPECT showed increased perfusion in http://jnnp.bmj.com/ and parietal cortices; basal ganglia; and cere- the contralateral motor cortex in this case, in bellar hemispheres. Three contiguous 3-2 mm addition to the temporal lobe hyperperfusion. transverse slices were summed and regions of Perfusion ratios for all regions of interest in interest were drawn using rectangular boxes. the patients' interictal state did not differ from The volume studied was 4 9 cm3 (5 x 10 x 3 those in the control group, except for the pixels) for all regions except for the cerebellum mesial temporal region where there was a x x (7 7 3 pixels, 4-8 cm3) and basal ganglia significant decrease on the side of the epileptic on October 4, 2021 by guest. Protected copyright. (5 x 5 x 3 pixels, 25 cm3).Normal data were focus (table 2). Quantitative analysis of the obtained from scans of 18 healthy male volun- ictal and postictal studies confirmed the qual- teers where IJR ratios were calculated for the itative interpretation of the images (table 3). same regions of interest. Mean ratios for each region of interest in the ictal, postictal, and interictal patient studies were compared with each other by using paired t tests and with Table 2 Side to side perfusion ratios in the interictal state, * control values by using multiple unpaired t Values are means (SD) tests. Conversion of ratios to was logarithms Region of unnecessary as ratios varied beween 0 7 and Interest Controls Patients p Value 1V7. Lateral temporal 98 (2 5) 100 (4 39) NS Mesial temporal 101 (1 9) 94 (6 24) < 0-05 Posterior temporal 101 (2 5) 100 (3 04) NS Lateral frontal 99 (2 3) 99 (3-31) NS Results Parietal 99 (1-7) 101 (3 92) NS All patients were studied with ictal and post- Cerebellum 100 (1 9) 99 (3 75) NS Basal ganglia 99 (2-1) 100 (2-0) NS ictal SPECT; the length of seizures was similar in each patient for the two studies (ictal, mean *Side to side ratios (multiplied by 100) calculated from left or right side counts for controls, and ictal or normal side for length 1 min 33 s, postictal, mean length patients, were compared in each region of interest by using 1 min 25 s). Ictal images in all 12 patients unpaired t tests. Postictal switch in blood flow distribution and temporal lobe seizures 893

firmed the suspected hippocampal sclerosis. J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.55.10.891 on 1 October 1992. Downloaded from All patients underwent temporal lobectomy on the side of the ictal SPECT focus (table 1). Pathology was found in 10 cases (seven showed hippocampal sclerosis, and there were individ- ual cases of ganglioglioma, neocortical gliosis and dysplasia). All but one were seizure free (class 1 of Engel"6) during a mean follow up of 15 (range 4-32) months. Patient 11 had a pronounced improvement in seizure frequency (class 3 of Engel"6), although no pathology was suspected on MRI or found on histological examination.

Discussion We analysed the relative blood flow changes in the peri-ictal period as seen in the side to side differences of brain perfusion, by qualitative Transverse images through the temporal lobes for interictal, ictal, and postictal studies of and quantitative comparisons of cerebral blood patient 6 (left temporal lobe seizures). Colour scale from white to red to blue represents a spectrum of high to low counts. In the interictal study there is a small region of flow patterns in the interictal, ictal, and post- hypoperfusion in the left mesial temporal lobe (arrow). The ictal study shows a ictal phases. This was done because the deter- pronounced global increase in temporaI perfusion (large arrow). The 1 5 minute postictal mination of absolute CBF by SPECT requires study shows lateral temporal hypoperfusion (horizontal arrow) with residual mesial hyperperfusion (vertical arrow) on the left. an accurate attenuation correction technique which is not routinely available. It also requires an accurate arterial input function for the tracer, which requires elaborate and time consuming preparations similar to those for The lateral temporal, mesial temporal, and PET. Such an approach to blood flow meas- basal ganglia ratios were significantly increased urement is not currently feasible in the inves- in the ictal studies in comparison to the tigation of unpredictable events such as interictal, postictal and control data. The spontaneous seizures. increase over interictal values was 35 1% (95% Ictal scans in all patients showed significant confidence interval 21-8% to 48&4%) (p < relative global increases in blood flow in the 0 0005) for the lateral temporal area and discharging temporal lobe. The SPECT 30-8% (22.4% to 39 2%) (p < 0-0001) for the images produced from 99mTc-HMPAO injec- mesial temporal region. Comparison of the tion during or within 30 seconds of the end of postictal to the interictal scans showed that the the seizure suggests that the ictal temporal postictal mesial temporal ratios were unchan- hyperperfusion persists beyond the termina- ged, whereas the posterior and lateral temporal tion of EEG discharge for 60-90 seconds, as ratios were reduced (although the lateral tem- most of the HMPAO that crosses the blood poral change just failed to reach significance). brain barrier does so in the first 60 seconds Frontal and parietal ratios were significantly after the intravenous injection. 14 This distribu- decreased in the postictal state, confirming the tion period may be shorter, given the tachycar- http://jnnp.bmj.com/ visual interpretation of the scans. dia that often accompanies seizures.'7 No discrepancies were found between Injection after this time showed a switch in SPECT and EEG with regard to side of ictal blood flow patterns, with hypoperfusion of the activity. In patient 7, ictal EEG data were not lateral temporal cortex frequently associated useful in localizing the seizure focus, but with preserved perfusion or hyperperfusion of because clinical, interictal EEG, SPECT, the mesial temporal structures. The sequence MRI, and neuropsychological data concurred of changes was thus a significant relative on October 4, 2021 by guest. Protected copyright. on the localisation, a temporal lobectomy was increase in mesial and lateral temporal lobe performed; pathological examination con- blood flow with ipsilateral depression offrontal cortex perfusion during the seizure, followed by a reduction in perfusion ofthe epileptogenic hemisphere with relative preservation of mesial Table 3 Relative regional cerebral blood flow changes in the ictal and postictal states temporal blood flow in the immediate postictal compared to the interictal condition * state. Our previous observationsl 13 suggest that postictal blood flow changes gradually Ictal studies Postictal studies return to normal within 10-30 minutes. Mean % change Mean % change Lee et al noted in two of their ictal studies Region of (95% confidence (95% confidence interest interval) p Value that cortex surrounding the hyperperfused area interval) p Value was underperfused and suggested that a Lateral temporal 35-1 (21-8 to 48-4) < 0 0005 - 7-7 (- 15-8 to 0-4) 0-058 Mesial temporal 30-8 (22-4 to 39-2) < 0 0001 2-0 (-6-3 to 10-3) NS "steal" phenomenon may be responsible.8 Posterior temporal 11 1 (-3-5 to 25-7) NS -8-5 (- 14-0 to -30) < 0 01 Alternatively, there may be a depression of Lateral frontal -5-7 (- 13-9 to 2 5) NS - 12-3 (- 17-6 to -7-0) < 0-0005 neuronal in the corical zones Parietal 1-2 (-7-5 to 9 9) NS -77 (-13-9 to - 1-5) <0 05 activity adjacent Basal ganglia 9-6 (2-7 to 16-5) < 0-02 -0-2 (-4-3 to 3 9) NS to the disharging temporal lobe and a con- Cerebellum - 2-3 (-5-2 to 06) NS 0-6 (-1-7 to 2 9) NS sequent reduction in the demand for blood *Percentage differences in mean ratio changes (epileptic side/normal side) from interictal to ictal flow to those relatively inactive areas. Other and from interictal to postictal studies in 12 patients ictal studies using SPECT have shown 894 Newton, Berkovic, Austin, Rowe, McKay, Bladin

increased temporal lobe perfusion, but these a pathophysiological hallmark oftemporal lobe J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.55.10.891 on 1 October 1992. Downloaded from did not include observations of frontal and epilepsy. SPECT has expanded the under- parietal cortex hypoperfusion or the motor standing of pathophysiological events that strip hyperperfusion observed during hemi- accompany focal seizures in humans. generalisation.5 9 10 The motor strip hyper- found The authors gratefully acknowledge the contributions and perfusion during hemigeneralisation is cooperation they received from all patients, ward staff, and consistent with the expected haemodynamic members of the Department of Nuclear Medicine, Austin to ofneuronal Hospital, involved in this study. MRN, SFB, and CCR were response spread discharge to the supported by the National Health and Medical Research central region. The agreement of peri-ictal Council of Australia. SPECT data with ictal EEG localisation shows the reliability of side and site localisation of 1 Horsley V. An address on the origin and seat of epileptic SPECT in lobe seizures. The disturbance. BMJ 1892;i:693-6. temporal ratio 2 Gibbs FA, Lennox WG, Gibbs EL. Cerebral blood flow method of quantitation we used confirmed the preceding and accompanying epileptic seizures in man. visual and allowed a relative Arch Neurol Psychiatry 1934;32:257-72. interpretations 3 Penfield W The circulation of the epileptic brain. Ass Res measure of the magnitude of changes in Nerv Dis Proc 1938;18:605-37. CBF. 4 Kuhl DE, Engel J Jr, Phelps ME, Selin C. Epileptic patterns of local cerebral metabolism and perfusion in humans The hippocampus is the site of seizure determined by emission computed tomography of '8FDG in most cases of lobe and l3NH3. Ann Neurol 1980;8:348-60. generation temporal 5 Uren RF, Magistretti PL, Royal HD, et al. Single photon epilepsy.'8 '9 In the present study ictal mesial emission computed tomography. A method of measunng cerebral blood flow in three dimensions (preliminary temporal (hippocampal) hyperperfusion per- results of studies in patients with epilepsy and stroke). sisted into the postictal period in some cases. Med JAust 1983;i:411-3. This is consistent with the 6 Engel J Jr, Kuhl DE, Phelps ME. Patterns of human local findings previously cerebral glucose metabolism during epileptic seizures. reported in a larger series of patients, in which Science 1982;218:64-6. electrode EEG studies showed 7 Theodore WH, Newmark ME, Sato SJ, et al. [ 8F] Fluoro- depth total deoxyglucose positron emission tomography in refractory cessation of ictal activity at the time of complex partial seizures. Ann Neurol 1983;14: HMPAO Two factors 429-37. injection.'21 may 8 Lee BI, Markand ON, Siddiqui AR et al. Single photon underlie the ictal and postictal mesial hyper- emission computed tomography (SPECT) brain imaging seen our using HIPDM: intractable complex partial seizures. perfusion in SPECT studies. Experi- Neurology 1986;36: 1471-7. mentally, the extracellular cations (H+ and K+) 9 Duncan R, Patterson J, Hadley DM, et al. CT, MR and for the in cere- SPECT imaging in temporal lobe epilepsy. J Neurol responsible peri-ictal increases Neurosurg Psychiatry 1990;53: 11-5. bral microflow take some minutes to decline 10 Stefan H, Bauer J, Feistel H, et al. Regional cerebral blood after the end of the flow during focal seizures of temporal and frontocentral seizure.20 Secondly, peri- onset. Ann Neurol 1990;27:162-6. ictal regional CBF may be coupled to 11 Rowe CC, Berkovic SF, Austin M, McKay WJ, Bladin PF. increased non-oxidative Postictal SPECT in epilepsy. Lancet 1989;i:389-90. glucose metabolism, 12 Rowe CC, Berkovic SF, Sia STB, et al. Localization of which reflects hippocampal metabolic activity epileptic foci with postictal single photon emission in neural activation in computed tomography. Ann Neuroll989;26:660-8. physiological humans 13 Rowe CC, Berkovic SF, Austin MC, McKayWJ, Bladin PF. and experimental hippocampal seizures.22 Pattems of postictal cerebral blood flow in temporal lobe We conclude that the blood epilepsy qualitative and quantitative analysis. Neurology flow patterns in 199 1;41: 1096-103. temporal lobe epilepsy as shown by SPECT 14 Sharp PF, Smith FW, Gemmell HG, et al. Technetium-99m with HMPAO within a short time HMPAO stereoisomers as potential agents for imaging change regional cerebral blood flow: human volunteer studies. J frame, with global temporal hyperperfusion NuclMed 1986;27:171-7. 90 seconds of 15 Jackson GD, Berkovic SF, Tress BM, Kalnins RM, Fabinyi switching rapidly (within seizure CGA, et al. Hippocampal sclerosis can be reliably end) to with relative preserva- detected by magnetic resonance imaging. Neurology 1990;

hypoperfusion http://jnnp.bmj.com/ tion of mesial blood 40:1869-75. temporal flow. Ictal 16 Engel J Jr. Outcome with respect to seizures. Surgical SPECT images with global temporal hyper- treatment of the epilepsies. New York: Raven, 1987: are and 553-71. perfusion striking readily interpretable, 17 Blumhardt LD, Smith PEM, Owen L. Electrocardiographic whereas the postictal scans show more subtle accompaniments of temporal lobe epileptic seizures. but characteristic that know- Lancet 1986;i:1051-5. changes require 18 Quesney LF. Clinical and EEG features of complex partial ledge of the postictal switch in blood flow. seizures of temporal lobe origin. Epilepsia 1986;27(suppl accurate 2):S27-45. Clinically, therefore, documentation 19 Spencer SS, Williamson PD, Spencer DD, Mattson RH. of the time of injection with respect to the Human hippocampal seizure spread by depth and sub- on October 4, 2021 by guest. Protected copyright. video-EEG dural recording the hippocampal commissure. Epilepsia recording is crucial for proper 1987;28:479-89. interpretation. The peri-ictal perfusion pat- 20 Leniger-Follert E. Mechanisms of regulation of cerebral terns further evidence that the microflow during bicuculline-induced seizures in anaes- provide hippo- thetized cats. J Cereb Blood Flow Metab 1984;4: 150-65. campus is the generator for most seizures 21 Fox PT, Raichle ME, Mintun MA, Dence C. Non-oxidative in that The association an glucose consumption during focal physiologic neural arising region. of activity. Science 1988;241:462-4. abnormal synchronised hippocampal electrical 22 Ackermann RF, Lear JL. Glycolysis-induced discordance with the characteristic between glucose metabolic rates measured with radio- discharge rapidly chang- labelled fluorodeoxyglucose andglucose. J Cereb Metab ing blood flow patterns described here may be Blood Flow 1989;9:774-85.