Neurons in human epileptic cortex. Department of Neurological Surgery, University of Washington, Seattle, Washington

Transcription

1 J Neurosurg 55: , 1981 Neurons in human epileptic cortex Response to direct cortical stimulation ALLEN R. WYLER, M.D., AND ARTHUR A. WARD, JR., M.D. Department of Neurological Surgery, University of Washington, Seattle, Washington ~/ During craniotomy for surgical treatment of medically intractable epilepsy, single neurons were recorded from the lateral temporal cortex of 11 awake patients. A total of 83 neurons were recorded, and their response to repetitive direct cortical stimulation with ascending and descending frequency ramps between 1 to 10 Hz was evaluated. More normal units from the suspected epileptogenic cortex responded to repetitive stimulation at frequencies between 5 to 10 Hz with augmented action potential bursts than did units from cortex thought not to be primarily epileptogenic. This burst response might persist for up to 30 seconds after the frequency ramp had descended from 10 to 1 Hz. Except in' two cases, this augmentation of burst response was not accompanied by afterdischarge on electroencephalography. These data would indicate that neurons within the region of suspected epileptogenic cortex demonstrate a greater propensity for afterdischarge to repetitive stimuli than do neurons in more normal cortex. KEY WORDS 9 single neuron 9 direct cortical response 9 epilepsy 9 temporal lobe 9 seizure T HE direct cortical response (DCR) was first described by Adrian 1 in It has since been studied in a variety of laboratory animals 4-6,s'1L~2 and measured from the cortex of humans during the course of craniotomies2,9 Although the neural mechanism involved in the generation of the DCR waveform is not completely understood, it is clear that direct cortical stimulation can evoke responses in nearby neurons synaptically. Since little is known about neuronal excitability in focal epilepsy of humans, we have used direct cortical stimulation to study evoked responses of single neurons in lateral temporal cortex of patients undergoing resection of epileptogenic foci. The data suggest that neurons within epileptogenic cortex (as defined by direct electrocorticography, ECoG) are more easily driven to generate single-unit afterdischarge than similar units recorded from less epileptogenic cortex. Selection of Patients Clinical Material and Methods All patients had medically intractable complex partial seizures originating from one temporal lobe and had elected cortical resection as the next therapeutic attempt to control their seizure disorder. All patients were informed that single-unit recordings during the course of their operations were of an experimental nature and signed a separate consent form which had been approved by the Human Experimentation Committee of the University of Washington. All data were obtained during the course of operations performed under local anesthesia with, on some occasions, supplemental narcotic analgesia. However, all patients were awake at the time of recording and no data were collected from patients under general anesthesia. Recording Sites All data were obtained from lateral temporal cortex and included superior, middle, and inferior temporal gyri not more than 7 cm posterior to the temporal tip. Prior to single-unit recording, extensive ECoG monitoring had been undertaken to determine areas of maximum epileptiform activity as well as minimum afterdischarge thresholds. In some cases, strip electrodes were placed along the inferior temporal surface to better defme the existence of mesial temporal foci when suspected by preoperative sphenoidal recordings. Small, numbered paper tags were placed at 904 J. Neurosurg. / Volume 55 / December, 1981

2 Epileptic neurons ECoG sites of maximum spontaneous epileptiform spiking for later correlation with microelectrode recording sites. The exposed cortex was then photographed for permanent documentation. No electrode penetrations were made in cortex outside of the proposed resection site. Since many patients had presumed mesial epileptogenic foci, this allowed recording to be made from relatively normal lateral cortex in such cases. Stimulation A Trent Wells micromanipulator* (attached to the craniotomy margin) had a Lucite pressor foot with a 2-mm hole in its center for the microelectrode. At 1 mm to either side of this hole were two silver ball electrodes which were partially embedded into the Lucite so as to provide electrical contact with the cortical surface, without undue pressure. A third silver ball electrode was placed equidistant between these two stimulation electrodes (also 1 mm away from the central hole) and was used for recording the focal ECoG (similar to methods of Phillis and Ochs8). Stimuli were generated by a Grass $44 unit in series with a Grass SIU5 stimulus isolation unit and a 7.5 KOhm resistor to approximate constant current.t Pulse durations were most frequently between 0.2 and 0.5 msec. Stimulation current could be varied between 89 to 10 mamps. Bipolar stimuli were delivered between the two silver ball electrodes. After a unit was isolated, l/sec stimulation was given and the current increased until a maximum of 10 mamps at 0.5 msec was reached. If the unit's response could not be evoked at this level of maximum stimulation, it was considered unresponsive. If unit activity was evoked, then the pulse width was narrowed to 0.2 msec and the current decreased until threshold was found. In most cases, the polarity of the stimulation was reversed to insure minimum current threshold determination. After the threshold for each unit's response was determined, the frequency of stimulation was increased by 1 Hz/sec until 10-Hz stimulation was reached and held for 10 seconds. The stimuli were then decreased to l/sec in a similar fashion. This frequency ramp was repeated several times for units with well maintained action potential (AP) isolation. Unit Recordings Single-unit AP's were recorded through electrolytically etched tungsten microelectrodes coated with Epoxylite, with tip impedances between 1 to 1.5 megohms. Microelectrode output was amplified by a * Micromanipulator manufactured by Trent Wells Instrument Co., 8120 Otis Street, South Gate, California. t Grass $44 stimulus isolation unit manufactured by Grass Medical Instruments, Quincy, Massachusetts. PFC-2 Biolectric probe~: in series with an AC-coupled Grass P511 amplifier (bandpass of 300 Hz to 10 khz). The DCR and ECoG (recorded through the silver ball electrode embedded in the presser foot) were amplified in a similar fashion but using an AC coupled P511 amplifier with a bandpass of I to 100 Hz. Classification of Units Single neurons were classified as "epileptic" if they spontaneously fired high-frequency bursts of AP's. However, quantification of burst activity based on the percentage of interspike intervals (ISI's) less than 5 msec was not performed, as has been done with precentral units in monkey cortex. TM This is because there has been no systematic quantification of the ISI distribution of normal neurons from non-epileptic human cortex. Although "normal" neurons recorded from the amygdala and hippocampus of humans have been shown to produce burst firing, z,a~ this has not been recorded for neurons in non-epileptogenic lateral temporal cortex. Therefore, since units that fire highfrequency AP bursts are considered epileptic in animal models of epilepsy, a4 we have arbitrarily defined neurons with similar burst patterns as epileptic. Definition of the Focus Clear parameters for defining an "epileptogenic focus" do not exist, and are not easily defined. That epileptic foci exist at all is an empirical observation; that is, if the focus is removed, the seizures cease. By classical criteria, 7,~3 epileptogenic foci are regions of brain that demonstrate the most active epileptiform spiking on ECoG and sustain the longest periods of electrically evoked afterdischarge. However, neither criterion is absolute, since various regions of the temporal lobe have different inherent afterdischarge thresholds and since there is no consensus as to what neuronal mechanisms generate epileptiform spikes. For example, when recording from the cortical surface, one can never be sure if a spike is generated locally or has been transmitted from other neurons (projected spikes). Hence, we have arbitrarily divided cortical regions into epileptic and non-epileptic areas, based upon the above criteria while realizing that such distinctions have an inherent uncertainty. Results A total of 83 units were recorded from the lateral temporal cortex of 11 patients who were not receiving general anesthesia. All patients had been receiving anticonvulsant medication including phenytoin, but, in addition, seven had also received Tegretol (carbamazepine) or Depakene (valproic acid) preoperatively. The number of normal and epileptic units PFC Biolectric probe manufactured by Biolectric Instruments, Hastings-on-Hudson, New York. J. Neurosurg. / Volume 55 / December,

3 A. R. Wyler and A. A. Ward, Jr. recorded from relatively normal and primarily epileptogenic cortex are summarized in Table 1. Fro. 1. Unit activity from a weakly epileptic unit recorded from the middle temporal gyrus. A and B: Spontaneous unit activity. C: Spontaneous burst at faster sweep-speed than above. D: Spontaneous unit and correlative electrocorticogram activity. E: Evoked response at 1 Hz. F: Evoked response at 5 Hz. G: Evoked response at 10 Hz. H: Evoked response 10 seconds after returning stimulation to 1 Hz. I: Expanded sweep of direct cortical response recorded through AC-coupled amplifier; peak-topeak amplitude: 1 mv. Peak-to-peak unit activity varied somewhat with respiration but was 200 ttv in the sweep shown in D. Unit Responses For both normal and epileptic neurons, there were three possible outcomes to the stimulation frequency ramps as follows: 1) unit activity was evoked by stimuli; 2) the units were unaffected and would fire independently of cortical stimuli (that is, they could not be driven even at maximum frequencies, current, and pulse width); and 3) firing of the units was inhibited at stimulation frequencies greater than 3/ sec. For both normal and epileptic units, evoked during frequency ramps, the latency of evoked AP's shortened and became less variable as the frequency increased toward 10/sec (Fig. 1). The number of units in each of these categories are listed in Table 2. The three "epileptic" units that became inhibited during frequency ramps resumed spontaneous burst firing shortly after the stimuli were returned to l/sec. For all the epileptic units which were stimulated, the burst became more intense as the frequency increased and often would continue to generate augmented burst firing for several seconds (10 to 30 seconds) after the cortical stimulation had returned to 1/sec (Fig. 2). In two patients, the individual's typical aura was reliably precipitated by cortical stimulation at 10/sec for 10 seconds and, in one case, a secondary generalized complex partial seizure followed the aura. In both cases, the aura was correlated with augmented spontaneous unit burst firing, and the aura ceased as the unit activity returned to normal. The striking difference between units recorded in normal and epileptogenic cortex was that eight of 20 normal units from the epileptic cortex could be driven TABLE 1 Normal and epileptic neurons from lateral temporal cortex thought to be primarily normal or epileptogenic (focus) Type of Neuron Focus Normal Cortex normal epileptic 12 4 total TABLE 2 and normal units to repetitive cortical stimulation from 1 to 10 Hz FIG. 2. Evoked activity of a mildly epileptic unit. Sweeps A, B, and C: Consecutive evoked responses to l/sec stimuli at threshold. D and E: Evoked response at 5 Hz (D) and at 10 Hz (E). F: Three consecutive superimposed evoked bursts of 1 Hz, 10 seconds after stimulation had returned to 1 Hz. Peak-to-peak amplitude: 250/~V. Epileptic Normal Response No. Percent No. Percent evoked unaffected inhibited total J. Neurosurg. / Volume 55 / December, 1981

4 Epileptic neurons to burst firing during 10/sec stimulation whereas only five of 47 units from normal cortex could be driven to burst in a like manner (Fig. 3). Figure 3A and B shows a long latency response to 1 Hz stimuli, Fig. 3C is an evoked burst of the same unit at 10 Hz (note the markedly decreased latency), and Fig. 3D is the same type of response 5 seconds after the stimulation frequency had been returned to 1 Hz. After 30 seconds, the unit's evoked activity had returned to its original response (Fig. 3A). In these cases, the burst firing continued for several seconds after the stimulation returned to l/see, and the ECoG was not accompanied by typical afterdischarge. Discussion In the present study, only 48% of the total unit responses were evoked by surface stimulation. From animal experiments, Li and Chou 4 found only 30% of units responsive to cortical stimulation. Phillis and Ochs ~ reported that 78% of units in cat postcruciate cortex could be stimulated in this fashion. However, in both studies, the animals were anesthetized with various agents, including pentobarbital sodium, thiopentone sodium, and chloralose. In Phillis' study, many of the units were responsive only after small amounts of L-glutamate were applied in the close vicinity of the cell by iontophoresis. In addition, the cell type appeared important, since 70% of all antidromically identified Betz cells could be driven by surface stimulation. The units in the present study are therefore different from those in animal studies in that: 1) they were not under the influence of general anesthetics; 2) they were recorded from lateral temporal cortex rather than sensorimotor cortex; and 3) they were influenced by the presence of phenytoin in all cases, and a second anticonvulsant in many other cases. Phillis and Ochs ~ also reported that the response of some normal cells, especially those in more superficial layers, could be a short burst with a latency of 1.5 to FIG. 3. Evoked activity of a "normal" unit recorded within epileptogenic cortex. A and B: Consecutive responses to 1 Hz stimulation. C: Unit burst response at 10 Hz. Note that at this frequency the unit's latency decreased. D: Response 10 seconds after return to 1 Hz. Peak-to-peak unit amplitude: 200/tV. 24 msec. However, inspection of the illustrations revealed that this was infrequent and more often than not was a doublet response. Moreover, it is unclear whether the doublet response occurred with the stimulation frequencies of l/sec or 20/sec used in their experiments. Sugaya, et al., 12 reported that repetitive stimuli could "produce a brief burst of cell discharge." In their experiments, 1) stimuli were given at higher rates than 10/sec; 2) stimuli were at maximum intensity; 3) the illustrated bursts were only doublets; and 4) the animals had been given varying amounts of Metrazol (pentylenetrazol). They reported that repetitive stimuli (at 20 Hz) produced a sustained membrane depolarization, but the only cell discharge that occurred was that evoked by each shock. As the membrane depolarization increased, the cell discharge changed from one of "relative quiescence to a buildup of high frequency firing." And, "If the stimulus terminates at this point, there is an intensification of firing in the immediate poststimulation period when the membrane depolarization reaches a plateau." "With weak repetitive stimulus insufficient to evoke afterdischarge, sustained membrane depolarization does not develop, although the surface record was a negative SP shift. The intracellular records then show only hyperpolarization." They also noted that the activity of some units was inhibited during repetitive stimuli, and this was thought to result from hyperpolarization of some cells. A similar finding was also reported by Li and Salmoiraghi) It appears, therefore, that the response of single neurons in lateral temporal cortex of humans is, in many ways, similar to what has been documented from sensorimotor cortex of cats and monkeys. However, a major difference between cells recorded in what was presumed to be most epileptogenic cortex in comparison to cells from less epileptogenic cortex is that cells from the former more readily demonstrate single-unit afterdischarge when compared to cells of the latter. Although a similar type of afterdischarge has been recorded from laboratory animals, it appears not to be as intense as that recorded from humans. In many of the animal experiments, the preparations were studied under conditions that would augment repetitive firing (such as under chloralose anesthesia), whereas in the human subjects the unit response to direct cortical stimulation was recorded while the patient had been receiving anticonvulsant medications which should decrease repetitive firing. In 1949, Walker 1~ suggested that afterdischarge threshold was an adjunct to determining the epileptogenicity of the cortex. Penfield and Jasper 7 confirmed this observation; however, they placed some limitations on the accuracy of this technique because they felt it was a reliable indication of the epileptogenic focus in only 75% of patients. The data from the present study would suggest that single units may demonstrate afterdischarge without concomitant surface epileptiform activity. It is suggested that this represents a single- J. Neurosurg. / Volume 55 / December,

5 A. R. Wyler and A. A. Ward, Jr. unit analogy to ECoG afterdischarge, but that it recruits a more focal neuronal aggregate. Some ambiguity remains between the ECoG-defined epileptogenic focus and the presence of either spontaneous or evoked abnormal unit activity. Although the majority of units that spontaneously fired epileptic burst patterns were within cortical regions thought to be primarily epileptic, similar units were recorded from relatively normal areas. In addition, neurons that could be evoked to sustained afterdischarge were not completely confined to ECoG-defined foci. The fact that abnormal unit activity is not consistently associated with the presumed focus can indicate several possibilities, including: 1) the ECoG (as presently interpreted) is not entirely accurate for defining primary epileptogenic cortex, since it cannot discriminate whether an epileptiform spike is generated locally or has been projected from another region; and 2) our criteria for defining epileptic firing patterns for single units recorded from regions other than precentral cortex of awake primates is not complete. Regardless of these ambiguities, the finding that the majority of spontaneous and evoked unit abnormalities are associated with those regions of presumed cortical epileptogenicity supports the concept that epilepsy results from "hyperphysiological" neurons. References 1. Adrian ED: The spread of activity in the cerebral cortex. J Physiol (Lond) 88: , Babb TL, Crandall PH: Epileptogenesis of human limbic neurons in psychomotor epileptics. Eiectroencephalogr Clin Neurophysioi 40: , Goldring S, Jerva M J, Holmes TG, et al: Direct response of human cerebral cortex. Arch Neurol 4: , Li CL, Chou SN: Cortical intracellular synaptic potentials and direct cortical stimulation. J Cell Comp Physlol 60:1-16, Li CL, Salmoiraghi GC: Cortical steady potential changes: extracellular microelectrode investigations. Nature 198: , Ochs S, Clark F J: Interaction of direct cortical responses-a possible dentritic site of inhibition. Eiectroencephalogr Clin Neurophysiol 24: , Penfield W, Jasper HH: Epilepsy and the Functional Anatomy of the Human Brain. Boston: Little, Brown and Co, 1954, p Phillis JW, Ochs S: Occlusive behavior of negativewave direct cortical response (DCR) and single cells in the cortex. J Neurophysioi 34: , Purpura DP, Pool JL, Ransohoff J, et al: Observations on evoked dendritic potentials of human cortex. Electroencephalogr Clin Neurophysiol 9: , Rayport M: Single neurone studies in human epilepsy, in Somjen GG (ed): Neurnphysiology Studied in Man. Amsterdam: Excerpta Medica, 1972, pp Stohr PE, Goldring S, O'Leary JL: Patterns of unit discharge associated with direct cortical response in monkey and cat. Electroencephalogr Ciin Neurophysiol 15: , Sugaya E, Goldring S, O'Leary JL: Intracellular potentials associated with direct cortical response and seizure discharge in cat. Electroencephalogr Clin Neurophysiol 17: , Walker AE: Electrocorticography in epilepsy. Electroencephalogr Clin Neurophysiol Suppl 2:30-37, Wyler AR, Ward AA Jr: Epileptic neurons, in Lockard JS, Ward AA Jr." Epilepsy: A Window to Brain Mechanisms. New York: Raven Press, 1980, pp Manuscript received December 30, Accepted in fmal form July 6, This work was supported by NIH Grant NS and Teacher Investigator Award NS (Dr. Wyler) awarded by the National Institute of Neurological and Communicative Disorders and Stroke, PHS/DHHS. Drs. Wyler and Ward are affiliates of the Child Development and Mental Retardation Center, University of Washington. Address reprint requests to: Allen R. Wyler, M.D., Department of Neurological Surgery RI-20, University of Washington, Seattle, Washington J. Neurosurg. / Volume 55 / December, 1981