204

EPIRES 00399

&i&y Res., 8 (1991) 204-212 Elsevier

Effects of seizures and carbamazepine on interictal spiking in amygdala kindled cats

Gian Luigi Gigli* and Jean Gotman

Montreal Neurological Institute and Department of Neurology and Neurosurgery, McGill University, Montreal, Quebec, Canada

(Received 29 July 1990; revision received 14 December 1990; accepted 18 December 1990)

Key wora!s: Interictal spiking; Kindling; Carbamazepine; REM sleep

We examined the influence of seizures and carbamazepine (CBZ) on spiking rates in kindled cats. In the first experiment, spiking rates were measured before and after seizures, with and without CBZ. CBZ was administered immediately after seizures in order not to affect them. Spiking rates were measured over 9 h during the different sleep stages. In a second experiment, CBZ was administered before and after seizures so as to affect seizure strength and thus measure its effect on spiking. Results confirmed earlier findings of a large increase in spiking following a stage 6 seizure in fully kindled animals. We also established that: (1) repeated daily seizures caused a further increase in spiking until a ceiling was reached; (2) increased spiking was not a direct effect of postictal alterations in sleep stages; (3) CBZ, despite its effectiveness as an anticonvulsant, did not reduce spiking but rather increased it; (4) postictal in- creases in spiking were related to seizure ‘strength’. These findings support the hypothesis that spiking rates are primarily influenced by seizure occurrence, as was found in patients with temporal lobe seizures, and that anticonvulsants act differently on seizures and spikes. This emphasizes the possibility of distinct pathophysiological mechanisms for interictal spikes and seizures.

INTRODUCTION

In chronic models of focal epilepsy, as in human epilepsy, the rate of interictal spiking changes greatly over time. The factors governing these changes are only partially known. In human focal epilepsy, sleep changes frequently affect the rate of spiking, which is much higher in non-REM sleep

* Present address: Clinica Neurologica, II Universita di Roma (Tor Vergata), Rome, Italy.

Correspondence roe: Jean Gotman, Ph. D., Montreal Neuro- logical Institute, 3801 University Street, Montreal, Quebec, Canada. H3A 2B4.

than in REM sleep and wakefulness’7~‘8,20,21. A major factor influencing the occurrence of spikes is also the occurrence of seizures: spikes increase in frequency after many seizures, particularly sec- ondarily generalized seizuresg,10F12. One factor that, surprisingly, does not appear to influence spiking, is the plasma level of antiepileptic medica- tiongy’O.

In amygdala kindling, the most studied model of human temporal lobe epilepsy, a similar situation exists, with respect to sleep and to the effect of sei- zures. In fully kindled animals, spiking rate is higher during slow wave sleep than during REM sleep or wakefulness’3~‘5Jg~28. Spiking also in- creases in the hours that follow each seizure7,16. As

0920-1211/91/$03.50 0 1991 Elsevier Science Publishers B.V.

205

in humans, the probability of occurrence of a spon- taneous seizure appears independent of the spik- ing rate (high spiking rates do not increase the likelihood of seizure occurrence).

In the experiments presented here, we address some questions that still remain from animal data and their comparison to human data. Secondarily generalized convulsive seizures disrupt sleep, es- pecially by reducing REM sleep, in humans3T4 and in animals5,‘9~23,24,25’28. The following questions therefore arise: is the postictal increase in spiking related to the postictal changes in sleep patterns? Could changes in sleep patterns fully explain changes in spike frequency? In addition to at- tempting to answer this question of sleep and spik- ing, we investigated how spiking evolves from mid-kindling to full kindling and with continuing seizures in fully kindled animals. Different results have been found with respect to the appearance of spikes and the timing of maximal spiking in the course of kindling’9,28,30.

We also wanted to investigate the effect of a known antiepileptic drug on spiking activity in the kindling model: carbamazepine (CBZ) is effective in many patients with partial seizures and was shown to increase the seizure threshold and re- duce seizure activity in kindled animals’,2’26. In particular we wanted to know its effect on spiking, having already studied its effect on sleep5 and hav- ing observed its lack of effect on spiking in pa- tients’.

By these studies, we hope to gain a better under- standing of the influence of seizures, sleep stages and antiepileptic medication on spiking activity in kindled animals.

METHODS

Two experiments were carried out. Common as- pects will be described while explaining the first. For the second, differences will be mentioned.

In the first experiment, after 5 weeks of observa- tion and adaptation, nine adult cats (3.5-4.5 kg) were implanted with chronic electrodes under gen- eral anaesthesia (sodium pentobarbital i.v.). Epi- dural screws were inserted in frontal and occipital bones to record epicortical EEG and supraorbital- ly for the electro-oculogram (EOG). Bipolar

depth electrodes were positioned stereotaxically in basal amygdalae and dorsal hippocampi bilater- ally. Wire electrodes were inserted in neck mus- cles for EMG. One of the electrodes implanted in basal amygdalae (left or right) served also as a stimulating electrode for kindling. After implanta- tion animals were allowed ten days to recover.

In the second experiment, in which effects of sei- zures and CBZ were more thoroughly investi- gated, a single animal with active spikes was se- lected for recordings.

EEG recordings During polygraphic recordings, animals were

kept in a large, partially sound-proof cage (1 x 1 x 1 m), connected to a cable that allowed free move- ment by using counterweights and a rotating con- nector. Signals used to score sleep stages were am- plified, filtered and recorded on paper using an 8- channel Grass electroencephalograph. Paper speed was set at 6 mm/set. Sensitivities were be- tween 20 and 30 ,uV/mm, time constants at 0.3 and low pass filters at 15 Hz (except for the EMG channel at 70 Hz). The montage used for paper tracing included left and right fronto-occipital EEG leads, left and right hippocampal EEG leads, the EEG from one of the amygdalae, 1 EOG and 1 EMG channel. One channel was used as a time marker. All recordings took place be- tween 6:00 p.m. and 3:00 a.m., corresponding to a total recording time (TRT) of 540 min.

Since they were not used for sleep scoring, EEGs of Expt. 2 were only recorded on disk by an automatic spike detection program (see section below). Recordings of experiment 2 took place be- tween 690 p.m. and 8 a.m., corresponding to a to- tal recording time of 840 min.

Kindling Stimuli consisted of 1-set trains of 1-msec square

waves at 60 Hz. Behavioral modifications asso- ciated with stimulation were observed and noted. Thresholds for kindling stimulation were estab- lished. Thresholds varied in different animals and were defined as the lowest intensity of current able to provoke a clear afterdischarge in a stepwise stimulation procedure. Animals were stimulated once a day at 390 p.m. with an intensity slightly

206

below the established threshold. Stimulation con- tinued until full kindling occurred and for some time after, as described below in the ‘Schedule of experiments’. Stimulation was provided only when EEG and behaviour indicated that the animal was alert, since it has been shown that sleep disorders depend on the stage at which seizures are elicited in the sleep waking cycle24. The behavioral and EEG progression of kindling was classified in six stages, with stage 6 representing the condition of generalized tonic-clonic convulsive seizures3’.

Sleep scoring Details on sleep scoring procedures have been

previously published6. Each 1-min epoch was clas- sified as either wakefulness (W), NREM sleep, or REM sleep.

Automatic spike detection The amplified and filtered EEG signals (2 epi-

cortical, 2 from amygdala, and 2 from hippocam- pal leads) were sent to a computer for automatic spike recognition. The system, routinely used for clinical purposes’ was instructed to detect spikes only on amygdala and hippocampal channels and to store each detected event together with the 2.5 set of EEG that preceded and followed it (Fig. 1). The computer also stored 30-set samples of EEG every 30 min. In the morning following the record- ing session, the data stored on disc were played back on paper for visual editing. Artifacts were eliminated and each spike was assigned to the cor- responding sleep epoch, as scored on the continu- ous recording on paper.

In experiment 1, spiking rates (number of spikes per min) were calculated for each animal during

total recording time, during wakefulness, NREM sleep and REM sleep. Because of the great inter- individual variability in spiking rates, results of ex- periment 1 were normalized, assuming as 100% each animal’s spiking rate during total recording time when the animal became fully kindled (SZ-A condition described below). In Expt. 2, spiking was measured only during total recording time, over a 14-h period of continuous EEG recording, starting 3 h after stimulation. There was no need to normalize these data since all results were from one animal.

In order to differentiate the production of spikes on the hemisphere homolateral to stimulation from that in the contralateral hemisphere, we sep- arated spikes detected from left amygdala and hip- pocampus from those detected on the right.

Experiment 1. The sequence of recordings in- Schedule of experiments

cluded recordings at mid-kindling (MID), at the end of the kindling process (SZ-A), after approxi- mately 2 weeks of further daily stage 6 seizures (SZ-B), 1 week and 1 month after the last seizure (Post-7 and Post-30). In SZ-A and SZ-B condi- tions, we made recordings with and without CBZ.

A synopsis of the schedule is given in Fig. 2. CBZ was administered (100 mg p.o.) between

3130 and 4:00 p.m. We used a dosage of 100 mg, an amount that has been shown to be effective in sup- pressing kindled seizures or reducing th$r intensi- ty, and a dosage not very different from clinical value&. The recording took place between 6 p.m. and 390 a.m. Our pilot pharmacokinetic study es- tablished that this dose and schedule provided high plasma levels during the period of recording6.

Stage I -- Wakefulness

t Hippocamws v

R Fronto-Occipital fl ‘W

R Hippocampus S-.+.X+ m

2r

Fig. 1. Examples of automatic detection of epileptic spikes during wakefulness and stage I sleep. The stimulated amygdala is indicated by an asterisk. The channels in which spikes were detected are indicated by solid circles.

207

A Begin stimulation

1 Stage 6 seizures

Observation t

Mid-kindling

I I I I,

W Sz-A

1 Stop stimulation

~r,~,,“I”I,~i”~“““~“““““‘~

Sz-A + CBZ Sz-B Sz-B + CBZ post-7 Post-30

Fig. 2. Schedule of Expt. 1. The second line is a direct continuation of the first. Small vertical marks represent one day. The thinnest horizontal line represents absence of stimulation. The thickest horizontal line represents the fully kindled state with daily seizures. Jn- termediate thickness represents the evolution of kindling. Each condition is the average of 2 nights (consecutive without CBZ and sep-

arated by 3 days with CBZ).

The study also established that total clearance of CBZ was obtained 72 hours after administration. Two administrations of CBZ were therefore al- ways separated by a minimum of 3 days. CBZ was administered on the same day as electrical stimula- tion, but it was always given after the stimulation in order not to modify the seizure triggered by stimulation. Thus, we could study the effect of CBZ on spiking independently from its effect on seizures.

For each condition, data from 2 nights were av- eraged; in the absence of CBZ, these 2 nights were consecutive. For CBZ nights, we averaged 2 nights separated by 3 days in order to allow for CBZ clearance. As an example, the SZ-A condi- tion was the average of 2 consecutive nights at the end of kindling. SZ-A + CBZ was the average of 2 nights separated by 3 days, while daily stimulation continued.

The entire experimental protocol included 14 re- cording sessions for each of the 9 cats and lasted around 3% months for each animal. Two of 9 ani- mals had to be discarded from analysis, because the electrodes were too sensitive to movement ar- tifacts which could not be separated from genuine spikes with confidence. Due to overlapping in re- cording sessions, 1 animal was not recorded during the SZ-B and SZ-B + CBZ conditions, and anoth- er was not recorded during the Post-7 condition.

In order to assess the effect of seizures on inter- ictal spikes independently from the effect of CBZ, we compared the mid-kindling, SZ-A, SZ-B, Post-

7 and Post-30 conditions. To measure the effect of CBZ, we made paired comparisons: SZ-A vs. SZ- A + CBZ, SZ-B vs. SZ-B + CBZ. Because of the great interindividual variability, statistical analysis was done only for correlation between spiking rates during TRT and sleep variables (percentage of wakefulness, NREM sleep and REM sleep). Experiment 2. In order to investigate more thor-

oughly the effect of seizures and of CBZ on spik- ing, we selected one cat with active spiking and performed multiple studies in which CBZ dosage and timing of administration were changed. The spiking rate was measured during TRT (not during the different wakefulness and sleep stages) in the following conditions: (1) Baseline. Spiking activity was measured 3 days after the last seizure (total number of recordings in this condition: n = 4). (2) After administration of CBZ (no stimulation and no seizure). This allowed observation of changes in baseline spiking caused by CBZ, independently of seizures. Dosages were 50 mg (n = 5) and 100 mg (n = 5). (3) After a stage 6 seizure (n = 6). The stimulation was administered 3 days after the last stimulation in order to ensure that the postictal in- crease in spiking had subsided. This condition al- lowed measurement of the increase in spiking due to the seizure alone. Stimulation in all subsequent conditions was carried out with a similar 3-day in- terval. (4) After a stage-6 seizure, itself followed by CBZ administration, as in Expt. 1. CBZ doses of 50 mg (n = 6) and 100 mg (n = 7) were adminis- tered immediately after the seizure to compare the

208

CBZ to the non CBZ condition, without affecting seizure intensity. Thus, we measured the effect of CBZ when spiking rate was high because of a pre- ceding seizure. (5) After a seizure preceded by CBZ administration (CBZ was administered 1 h before stimulation, thus allowing sufficient ab- sorption for a clear effect on seizures). Comparing this condition to that with CBZ administered just after the seizure allowed measuring the effect of seizure strength: CBZ administered before the seizure renders the seizure less violent and usually removes the tonic-clonic component (it does not necessarily alter seizure duration26). Given that CBZ is administered 1 h before the seizure in this condition and just after the seizure in the previous condition, effects of CBZ itself on the spiking ac- tivity are identical; only seizure strength is changed. Dosages were 50 mg (n = 6) and 100 mg (n = 7).

At the end of each experiment, each animal was killed with an overdose of barbiturate. The brain was removed and immersion fixed in formaline for verification of electrode placement.

RESULTS

Experiment 1 The effect of seizures on interictal spikes is

shown for each animal in Fig. 3. The results show a great interindividual variability in spiking levels. There is, however, a fairly consistent profile, showing a higher spiking rate during total record- ing time (TRT) when the animal was fully kindled (SZ-A) than during the development of kindling (MID) in 6 out of 7 animals (in one there was no difference). Continuing to stimulate animals (SZ- B), resulted in a further increase in the spiking rate in 4 out of 6 animals (one did not enter SZ-B re-

,O- o 9 Alabaster II8 - 020 -

'y O?-

g 06-

F 05

104- z 03 -

02 -

01 -

OO-

OS- Cedrick

1 75 - 04 -

5 03- z

02 -

s

01 -

0 Mid

n Sz-A

q Sz-B

gJ Post 7

q Post-30

q Sz-A + CBZ

q Sz.B + CBZ

Fig. 3. Expt. 1. Spiking rates during total recording time in the 7 cats. Each bar represents the mean of 2 recording sessions in the same experimental condition.

cordings). The end of stimulation was followed by a clear decrease in spiking rates, but spiking defi- nitely persisted.

The first CBZ administration consistently in- creased spiking rates (SZ-A vs. SZ-A + CBZ). The second administration of CBZ did not repli- cate this increase in spiking.

Data were normalized because of high vari- ability between animals. The normalized data show an averaged profile of spiking rates during TRT that confirms the increase in spiking with the progression of seizures and with the first adminis- tration of CBZ (Fig. 4). A similar profile was ob- served during wakefulness and NREM sleep. The

Total Necordkbg The 200 r 0 Mid

n 9-A

H Sz-a H Post.7

q post.30

q Sz-A + CBZ

q Sz-B + CBZ

Wakotoleets NREM Steep Experiment 2

200 r REM Skcp

Fig. 4. Expt. 1. Average profiles (7 cats) of spiking rates across experimental conditions in different states of vigilance. Be- cause of great interindividual variability, spiking rates are normalized, taking as a reference (100%) the SZ-A condition during total recording time (fully kindled animals). The left half of each graph illustrates the effect of seizures on spiking (conditions: Mid, SZ-A, SZ-B, Post-7, Post-30). The right half illustrates the effect of CBZ (SZ-A, SZA + CBZ, SZB, SZ-B

+ CBZ).

209

pattern was similar in REM sleep, with the excep- tion of the SZ-B + CBZ condition, where spiking was lower than in SZ-B. Absolute spiking rates were generally very low during REM sleep and re- sults in that state may be less reliable.

The separation of spikes according to the side in which they originate (stimulated or contralateral) shows that bilateral spiking is already present at midkindling and that the increase with progression of seizures occurs first on the homolateral hemi- sphere and later on the contralateral one (Fig. 5, left). Following CBZ administration (Fig. 5, right), spiking rates are high in both hemispheres.

The spiking rates during TRT in the different conditions were correlated with the percentage of wakefulness, NREM and REM sleep that animals were presenting at the corresponding recording session. Spiking rate and percentage of NREM sleep were not significantly correlated (r = -0.250); percentage of W presented a positive sig- nificant correlation (r = 0.471, P < 0.01) and per- centage of REM sleep a negative significant corre- lation (r = -0.421, P < 0.01) with spiking rates.

Results are shown in Fig. 6. Spiking rate in the

SEA SZ-0 post7 Post30 SZ-A SZ-A SZ-S SI-S

22 CL Fig. 5. Expt. 1. Comparison of average spiking rates during to- tal recording time across conditions for the stimulated and con- tralateral hemispheres. The data are normalized taking as ref- erence (100%) the total spiking rate (stimulated + contralater-

al side) at SZ-A condition during total recording time.

210

11 r

10

0.9

0.6

07

aI

g 0.6

F

:s

w

05

0.4

03'

02'

0.1

0.0 A- Baseline

1 50 100

CBZ 0x0 SZ CBZ (mg) +

CBZ (w ST Fig. 6. Expt. 2. Spiking rates during different experimental conditions. Each result is the average of 4-7 recording sessions

in the same animal.

baseline condition (3 days after the last seizure) was very low (0.08 spikes per minute). CBZ (with- out seizure) caused a slight increase in spiking rate at both doses (0.12 with 50 mg; 0.19 with 100 mg). Seizures alone (without drug administration) pro- voked a very large increase of spiking (0.41). However, the most dramatic increases in spiking rate were observed when CBZ was given after sei- zures (thus not affecting seizure intensity). In this condition, spiking rates were more than ten times the baseline values (0.94 after 50 mg and 1.04 after 100 mg). Finally, when CBZ was administered 1 h before seizures, thus reducing seizure intensity, the increase in spiking rate was much less promi- nent (0.51 with 50 mg and 0.52 with 100 mg CBZ).

DISCUSSION

Our findings confirm previous reports on the spike-inducing role of seizures7,10,16,27. In accord- ance with other studies, the production of spikes began long before the stage of generalized convul-

sions appeared’6~‘9~22’2R’30. The appearance of the state of generalized tonic-clonic seizures caused an activation of spikes, differently from findings of other studies’4,2s,30, but similarly to findings of Rondouin et a1.19 in hippocampal kindling. A pos- sible explanation for this contradiction is that the

time elapsed from stimulation and duration of re- cording can influence the results. In particular, re- cordings made immediately or 24 h after a seizure are likely to miss the spike-activating effect of gen- eralized convulsions7,16.

We demonstrated in Expt. 2 that the spike-acti- vating ability of seizures is a function of their strength: reducing their strength with CBZ admin- istration before stimulation resulted in a lesser in- crease in spiking. The possibility that this reduc- tion was caused by CBZ itself, and not by the re- duction in seizure intensity, was excluded by com- paring that condition with CBZ administration im- mediately after the seizure. In these 2 conditions, levels of CBZ were identical; only seizure intensi- ty was different.

Once generalized convulsions occur, further sei- zures, evoked with a stimulation interval of 24 h, continue to increase spiking (SZ-B condition). This is probably due to the occurrence of a new sei- zure before the activating effects of the previous one have subsided and ‘baseline’ spiking levels have been reached. However, this cumulating ef- fect of daily stimulations cannot progress beyond a ‘ceiling’, specific to each animal. In contrast to a previous reportsc’, spikes were recorded bilaterally at midkindling. The hemisphere contralateral to the stimulated amygdala became active more slow- ly than the homolateral one, reaching its ceiling later.

Since spiking rates are usually higher in NREM sleep 15,19,22,28, the possibility could be considered that the postictal increase in spiking was simply a result of an increase in NREM sleep. This possibil- ity can now be excluded for 2 reasons. First, sei- zures disturb sleep, but do not significantly change the amount of NREM sleep, at least when there is a sufficient interval between stimulation and re- cording5~“,‘9~*9. Second, in this study, spiking rates were not correlated with the percentage of NREM sleep; rather, they were correlated with the per- centage of wakefulness. It is notable that spikes

211

were observed during REM sleep, in agreement with Rondouin et al. , l9 but in contrast to other studies13,22,28. This is similar to human focal epilep- sy in which spikes are rare but present during REM sleep. Our studies show that seizures in- crease wakefulness and spiking at the same time. Whether this is simply a correlation, or whether disturbed sleep (more wakefulness) increases spiking needs further elucidation. In agreement with Leung16, spikes continue to be present long after the last seizure, despite the end of stimula- tion, but spikes become much less frequent.

Contrary to expectation, CBZ, an effective anti- convulsant, not only failed to reduce spikes, but rather increased their frequency. This finding con- firms indirectly that spikes and seizures are dis- tinct phenomena probably generated by different pathophysiologic mechanisms, seizures being able to increase spiking, but spiking not being predic- tive of seizures7,i0. The effect of CBZ on spiking is dose related and is not due to an increase of NREM sleep: CBZ does not increase NREM sleep in kindled animals’, and spiking rates and

REFERENCES

1 Albertson, T.E., Joy, R.M. and Stark, L.G., Carbamaze- pine. A pharmacological study in the kindling model of epi- lepsy, Neuropharmacology, 23 (1984) 1117-1123.

2 Albright, P.S., Effects of carbamazepine, clonazepam and phenytoin on seizure threshold in amygdala and cortex, Exp. Neural., 79 (1983) 11-17.

3 Baldy-Moulinier, M., Temporal lobe epilepsy and sleep or- ganization. In: M.B. Sterman, M.N. Shouse and P. Pas- souant (Eds.), Sleep and Epilepsy, Academic Press, New York, 1982, pp. 347-360.

4 Baldy-Moulinier, M., Touchon, J., Besset, A., Billiard, M., Cadilhac, J. and Passouant, P., Sleep architecture and epileptic seizures. In: R. Degen and E. Niedermeyer (Eds.), Epilepsy, Sleep and Sleep Deprivation, Elsevier, Amsterdam, 1984, pp. 109-118.

5 Gigli, G.L. and Gotman, J., Effects of seizures, kindling and carbamazepine on sleep organization in cats, Epilepsia,

in press. 6 Gigli, G.L., Gotman, J. and Thomas, S.T., Sleep altera-

tions after acute administration of carbamazepine in cats, Epilepsia, 29 (1988) 748-752.

7 Gotman, J., Relationships between triggered seizures, spontaneous seizures, and interictal spiking in the kindling model of epilepsy, Exp. Neurol., 84 (1984) 259-273.

amount of NREM sleep are not correlated. The second administration of CBZ in Expt. 1 (SZ-B + CBZ) did not reproduce the activating effect of the drug on spiking observed in the SZ-A condi- tion. This could be due to a ceiling effect, the spik- ing rate being already very high at the SZ-B condi- tion.

A critical factor appears to be REM sleep. Both seizures and CBZ cause a decrease in REM sleep5 and REM sleep is negatively correlated to spiking rate. The decrease of REM sleep could be one of the possible mechanisms by which seizures and CBZ increase spiking.

ACKNOWLEDGEMENTS

We are grateful for technical assistance from S. Schiller, S. Thomas and E. Madevu, and to J. Thi- baudeau for typing this manuscript.

G.L.G. is the recipient of the W.G. Lennox Fel- lowship of the American Epilepsy Society. J.G. is the recipient of an I.W. Killam Scholarship from the Montreal Neurological Institute.

8 Gotman, J., Ives, J.R., Gloor, P., Quesney, L.F. and Bergsma, P., Long-term monitoring at the Montreal Neu- rological Institute. In: J. Gotman, J.R. Ives and P. Gloor (Eds.), Long-Term Monitoring in Epilepsy, Electroenceph.

Clin. Neurophysiol., Suppl. 37, Elsevier, Amsterdam, 1985, pp. 327-340.

9 Gotman, J. and Koffler, D.J., Interictal spiking increases after seizures but does not after decrease in medication, Electroenceph. Clin. Neurophysiol., 72 (1989) 7-15.

10 Gotman, J. and Marciani, M., Electroencephalographic spike activity, drug levels and seizure occurrence in epilep- tic patients, Ann. Neurol., 17 (1985) 597-603.

11 Hiyoshi, T., Mou, M. and Wada, J.A., Feline amygdaloid kindling and sleep, Electroenceph. Clin. Neurophysiol., 73

(1989) 254-259. 12 Katz, A. and Spencer, S., Spatial and temporal relations of

interictal spikes and seizures, Epilepsia, 5 (1989) 664.

13 Kawahara, R. and Wada, J.A., Effect of REM sleep depri- vation on amygdaloid kindling in cats, Jpn. J. EEG EMG, ll(l983) 176-184.

14 Lange, H., Tanaka, T. and Naquet, R., Temporal-spatial pattern of subcortical spike activity in kindling epilepsy. A statistical approach, Electroenceph. Clin. Neurophysiol.,

42 (1976) 554-574.

15 Leung, L.S., Hippocampal interictal spikes induced by kindling: relations to behavior and EEG, Behav. Bruin

212

Res., 31 (1988) 75-84. 16 Leung, L.S., Spontaneous hippocampal interictal spikes

following local kindling: time-course of change and relation to behavioral seizures, Brain Res., 513 (1990) 308-314.

17 Lieb, J.P., Joseph. J.P., Engel, J., Walker, J. and Crandall, P.H., Sleep state and seizure foci related to depth spike ac- tivity_ in patients with temporal lobe epilepsy, Electroen- mph. Clin. Neurophysiol., 49 (1980) 538-557.

18 Montplaisir, J., Laverditre, M., Saint-Hilaire, J.M. and Rouleau, I., Nocturnal sleep recording in partial epilepsy: a study with depth electrodes, .I. Clin. Neurophysiol., 4 (1987) 383-388.

19 Rondouin, G., Baldy-Moulinier, M. and Passouant, P., The influence of hippocampal kindling on sleep organiza- tion in cats. Effects of alpha-methyl-paratyrosine, Bruin

Res., 181 (1980) 413-424. 20 Rossi, G.F., Colicchio, G. and Pola, P., Interictal epileptic

activity during sleep: a stereo-EEG study in patients with partial epilepsy, Electroenceph. Clin. Neurophysiol., 58 (1984) 97-106.

21 Sammaritano, M., Gigli, G.L. and Gotman, J., Interictal spiking during wakefulness and sleep and the localization of foci in temporal lobe epilepsy, Neurology, 41 (1991) 290-297.

22 Sato, M. and Nakashima, T., Kindling: secondary epilepto- genesis, sleep and catecholamines, Can. J. Neural. Sci., 2 (1975) 439-446.

23 Shouse, M.N., State disorders and state-dependent sei- zures in amygdala-kindled cats, Exp. Neural., 92 (1986) 601-608.

24 Shouse, M.N., Sleep disorders in kindled cats differ accord- ing to the timing of polygraphic recordings and of seizures in the sleep-wake cycle, Exp. Neural., 96 (1987) 158-162.

2.5 Shouse, M.N. and Sterman, M.B., Sleep and kindling: II. Effects of generalized seizure induction, Exp. Neurol., 71 (1981) 563-580.

26 So, N. and Gotman, J., Quantitative EEG analysis of car- bamazepine effects on amygdaloid kindled seizures in cats, Elecrroenceph. Clin. Neurophysiol., 76 (1990) 63-72.

27 Sundaram, M., Hogan, T., Hiscock, M. and Pillay, N., Fac- tors affecting interictal spike discharges in adults with epi- lepsy, Electroenceph. Clin. Neurophysiol., 75 (1990)

358-360.

28 Tanaka, T., Lange, H. and Naquet, R., Sleep, subcortical stimulation and kindling in the cat, Can. J. Neural. Sci., 2

(1975) 447-455.

29 Tanaka, T. and Naquet, R., Kindling effect and sleep orga- nization in cats, Electroenceph. Clin. Neurophysiol., 39

(1975) 449-454.

30 Wada, J.A. and Sato, M., Generalized convulsive seizures induced by daily electrical stimulation of the amygdala in cats. Correlative electrographic and behavioral features, Neurology, 24 (1974) 565-574.