Address correspondence and reprint requests to Dr. J.H. Goodman at Center for Neural Recovery and Rehabilitation Research, Helen Hayes Hospital, West Haverstraw, NY 10993, U.S.A. E-mail: email@example.com
Summary: Purpose: The use of electrical stimulation as a therapy for epilepsy is currently being studied in experimental animals and in patients with epilepsy. This study examined the effect of preemptive, low-frequency, 1-Hz sine wave stimulation (LFS) on the incidence of amygdala-kindled seizures in the rat.
Methods: Electrodes were implanted into the basolateral amygdalae of adult male rats. All animals received a kindling stimulus of 60-Hz, 400-μA, sine wave for 1 s twice a day. Experimental animals received an additional LFS consisting of 1 Hz, 50 μA for 30 s immediately before the kindling stimulus. Afterdischarge (AD) duration, behavioral seizure score, the number of stimulations required to elicit the first stage five seizure and to become fully kindled were measured. After 20 stimulations, a crossover procedure was performed. Fully kindled rats from each group were switched, so that the original controls received LFS plus the kindling stimulus, and the original experimental rats received only the kindling stimulus.
Results: During kindling acquisition, LFS induced a significant decrease in AD duration. A significant increase in the number of times the kindling stimulus failed to elicit an AD was noted. Control rats exhibited an AD 99% of the time compared with 70% in experimental rats (p < 0.0001; Fisher's Exact test). In fully kindled animals, the incidence of stage five seizures in the original controls significantly decreased from 98% to 42% (p < 0.0001) when the LFS was added to the kindling paradigm.
Conclusions: The dramatic decrease in the incidence of stage 5 seizures in fully kindled animals after preemptive LFS strongly suggests that LFS may be an effective therapy for the prevention of seizures in patients with epilepsy.
The failure of current therapies to treat effectively a large number of patients with epilepsy highlights the need for new, effective, and safe treatments. The recent success obtained by using deep brain stimulation to treat movement disorders and vagus nerve stimulation to treat epilepsy has led to a renewed research effort to develop direct stimulation of the brain as a treatment for epilepsy.
To date, the results from animal and clinical studies that have examined the effectiveness of brain stimulation for epilepsy have varied. A decrease in clinical and experimentally induced seizure activity has been reported to occur after stimulation of the brainstem (1), caudate (2,3), cerebellum (4–8), anterior thalamus (9,10), centromedian thalamus (11), subthalamic nucleus (12–14), the thalamic reticular nucleus (15), mamillary bodies (16), locus ceruleus (17,18), substantia nigra (19,20), and hippocampus (21–23). By using the in vitro hippocampal slice preparation, several laboratories have demonstrated that direct stimulation (24–27) and the application of electric fields (28–32) can disrupt experimentally induced epileptiform activity. However, many of these positive reports were anecdotal or were poorly controlled, and a number of other studies found no effect of electrical stimulation on seizure activity (33–36).
The fundamental questions of where to deliver the stimulus and what stimulus parameters will be optimal for preventing or disrupting seizure activity remain unresolved. Stimulus frequency appears to be an important factor as to whether a given stimulation will be successful (9,14,16). Although many studies have focused on the effectiveness of high-frequency stimulation, experimental and clinical evidence now exists that low-frequency stimulation (LFS) can decrease or prevent seizure activity. Initial reports stating that LFS could interfere with the generation of kindled seizures were largely ignored (37,38). The term quenching was used to describe interference in the kindling process after stimulation with direct current (39). More recently, Velisek et al. (40) demonstrated that LFS during kindling acquisition delayed the kindling process in immature animals. Several in vitro studies demonstrated an effective decrease in experimentally induced seizure activity by using LFS (27,41–43). Clinically, LFS has been reported to decrease interictal spiking in patients with temporal lobe epilepsy (44,45).
In the present study, we examined the effect of preemptive low-frequency sine wave stimulation on kindling acquisition and on kindled seizure frequency in adult amygdala-kindled rats. Our results support the hypothesis that preemptive LFS decreases the frequency of kindled seizures and suggest that LFS could become a new therapy for pharmacoresistant epilepsy.
All animals were treated in accordance with the guidelines set by the New York State Department of Health and the National Institutes of Health for the humane treatment of animals.
Bipolar, Teflon-coated, stainless steel electrodes (127-μm diameter; Plastics 1, Roanoke, VA, U.S.A.) were implanted bilaterally into the basolateral amygdalae of adult male Sprague–Dawley rats (275–325 g) (46,47). The coordinates measured from bregma were as follows: A-P, 2.3 mm; M-L, ±4.5 mm from the midline; and depth, 8.5 mm from skull surface (48). In brief, each animal was anesthetized with a combination of ketamine and xylazine (1 ml/kg; 80 mg/ml ketamine, 12 mg/kg xylazine, i.p.; Sigma-Aldrich, St. Louis, MO, U.S.A.) and placed in a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA, U.S.A.). A midline incision was made in the scalp, the calvarium was exposed, and burr holes were drilled for the placement of anchor screws. Additional holes were drilled for the placement of the electrodes. Each electrode was lowered to the appropriate stereotaxic coordinate, cemented in place with acrylic cement (Plastics 1) and connected to a socket (Ginder Scientific, Ottawa, Canada). The wound was treated with disinfectant and closed with wound clips. The animal was placed under a radiant heat source until it recovered from the anesthetic. All animals were allowed to recover from the surgery a minimum of 1 week before initiation of afterdischarge (AD)-threshold testing and kindling stimulation. Once stimulation was initiated, electrographic activity recorded from each animal was amplified with differential amplifiers (BAK Electronics, Mount Airy, MD, U.S.A.) and stored and analyzed with a computer program provided by NeuroPace Inc. (Mountain View, CA, U.S.A.)
AD threshold testing
The AD threshold was determined for the left and right electrodes for each animal by using the following kindling stimulus: 60-Hz, 1-ms pulse, sine wave for 1 s. AD testing was started with a current intensity of 25 μA and was increased in 25-μA steps until an AD was observed. The electrode with the lower AD threshold or better recording quality or both was selected for control and experimental stimulation. Rats with an AD threshold >250 μA were excluded from the study.
Effect of LFS on kindling acquisition
The animals were randomly assigned to one of two groups: control (n = 6) and experimental (n = 6). Control animals were stimulated twice each day with the following kindling stimulus: 60-Hz, 1-ms, 400-μA sine wave for 1 s. Experimental animals received the same kindling stimulus; however, they also received a preemptive stimulus: 1-Hz, 50-μA sine wave for 30s immediatelybefore and yoked to the kindling stimulus. AD duration and the behavioral seizure score as defined by Racine (49) were measured for each animal after each stimulation. In addition, the number of stimulations required to reach the first stage 5 seizure and the number of stimulations required to become fully kindled were measured for both groups of animals. An animal was considered fully kindled after three consecutive stage 5 seizures. The frequency of how often stimulation failed to elicit an AD also was measured for each animal.
Data were collected for the first 20 stimulations. A series of 20 stimulations is greater than that normally required for an amygdala-kindled animal to become fully kindled (50). However, the number of stimulations delivered in this phase of the experiment was set at 20 to allow the possibility of a delay in kindling acquisition due to the addition of LFS to the experimental animals.
Effect of LFS in fully kindled animals
After the first 20 stimulations, a crossover procedure was performed on fully kindled rats from both groups. Rats from the original control group (n = 5) received the LFS plus the kindling stimulus, and rats from the original experimental group (n = 4) received only the kindling stimulus. Twenty stimulations were delivered to each rat in this phase of the study so that animals that completed both phases of the study received a total of 40 stimulations. The frequency of stage 5 seizures was measured in both groups of animals.
Electrode location was determined at the end of the stimulation paradigms. Each animal was anesthetized with urethane (1.25 g/kg, i.p.) and sacrificed by perfusion-fixation with 4% paraformaldehyde by gravity feed through the aorta for 10 min. The brains were left in situ and stored at 4°C overnight. The next day, the brains were removed and sectioned on a vibratome. Coronal sections were cut 50 μm thick and slide mounted for Nissl staining with cresyl violet. A successful implant was defined as an electrode tip within the basolateral amygdala or directly adjacent to the basolateral amygdala. With these criteria, electrodes were successfully implanted in 10 of 12 animals. We were unable to determine the electrode location in one animal because it died during the study, and in the other animal, the electrode tip was located in the perirhinal cortex. Both of these animals were included in the initial phase of the study because they kindled in a similar manner to the other rats. In addition, because perirhinal kindling usually requires fewer stimulations than the amygdala to kindle (51), any bias induced by the inclusion of this animal would make it more difficult to observe an anticonvulsant effect.
During kindling acquisition, values for AD duration and behavioral seizure score were averaged after each stimulation. The data used in this analysis were generated from only those stimulations that elicited an AD. Stimulations that failed to elicit an AD were not included in these data to prevent a bias toward a decrease in AD duration. The statistical comparison between the control and experimental groups was made by using a two-way analysis of variance (ANOVA) with one repeated measure followed by the Hotelling T2 test, which allowed a comparison of individual points within the 20-stimulation series (52). The effect of LFS on mean behavioral seizure score was analyzed by using the nonparametric Mann–Whitney U test. Comparisons of AD or seizure incidence between the two groups were made by using Fisher's exact test. One animal from each group failed to become fully kindled after 20 stimulations, and therefore they were dropped from the study.
Effect of LFS on kindling acquisition
No statistical difference was found between the control and experimental animals in the number of stimulations required for the first stage 5 seizure or for the animals to become fully kindled. No statistical difference was noted in the mean afterdischarge duration over the first 20 stimulations between the two groups of animals when the data were analyzed by using a two-way ANOVA with one repeated measure (Fig. 1A). However, a comparison of the mean AD duration for stimulation numbers 16–20 revealed a significant decrease in mean AD duration (p < 0.01; Hotelling T2) in the experimental animals compared with the controls.
Coincident with the decrease in AD duration, the LFS also induced a significant decrease in the mean behavioral seizure score (Fig. 1B; p < 0.001; Mann–Whitney U). Although control animals consistently exhibited stage 5 seizures, the experimental animals exhibited mean behavioral seizure scores ranging from 2 to 4. An example of the AD from a control and experimental animal is illustrated in Fig. 2. The AD illustrated in Fig. 2A, taken from a control animal after the sixth stimulation, had a longer duration than the AD illustrated in Fig. 2B, taken from an experimental animal after the twelfth stimulation, consistent with an LFS-induced decrease in AD duration.
Unexpectedly, a significant decrease was seen in the number of times the kindling stimulus elicited an AD in the experimental animals. During kindling acquisition, amygdala stimulation rarely fails to elicit an AD, and the control animals in this study exhibited an AD 99% of the time in response to the kindling stimulus. However, the experimental animals exhibited an AD only 70% of the time (Fig. 3; p < 0.0001; Fisher's exact test). This suggested that the preemptive delivery of the LFS decreased the incidence of kindled AD during kindling acquisition.
Because the decrease in AD duration and behavioral seizure score in the experimental animals was greatest in response to stimulations 16 to 20 (Fig. 1), the possibility that the LFS was more effective after the experimental animals had exhibited a stage 5 seizure also was examined. However, the incidence of AD in the experimental animals was essentially the same before and after the first stage 5 seizure (data not shown; Fisher's exact test; p > 0.05). This suggests that the LFS was equally effective in preventing kindling-induced ADs throughout the acquisition phase of kindling.
Effect of LFS in fully kindled animals
After both groups of animals had been stimulated 20 times, a crossover procedure was performed. Fully kindled control animals (n = 5) were now stimulated with the LFS plus the kindling stimulus, and fully kindled experimental animals (n = 4) were stimulated with just the kindling stimulus. Each animal received a total of 20 additional stimulations. Fig. 4 compares the incidence of stage 5 seizures in fully kindled control animals before the crossover (only the kindling stimulus) to fully kindled control animals receiving LFS plus the kindling stimulus. The delivery of the LFS to fully kindled control animals dramatically decreased the mean incidence of stage 5 seizures to 42% compared with 98% before the crossover procedure (Fig. 4; p < 0.0001; Fisher's Exact test). However, adding LFS to the kindling stimulus did not lead to a decrease in the AD duration in this group when a seizure did occur.
When the original experimental animals received the kindling stimulus alone, the incidence of stage 5 seizures increased to 83% from 67%. Although this increase in the incidence was not significant (p > 0.05; Fisher's Exact test), it does suggest that once the LFS was no longer delivered, the incidence of stage 5 seizures increased. Interestingly, the AD duration in the original experimental animals did not increase to control levels after the LFS was no longer delivered.
Deep brain stimulation has the potential to become a new treatment for those patients with epilepsy who are currently unresponsive to pharmacologic therapy or who are not candidates for surgical resection. However, before brain stimulation can become a new therapy for epilepsy, the fundamental questions of where to stimulate and what type of stimulation will be most effective must be resolved. This study demonstrates that preemptive LFS stimulation decreases the incidence of amygdala-kindled seizure activity both during kindling acquisition and in fully kindled animals. These results support the hypothesis that LFS could be an effective therapy in some patients with epilepsy.
It has been hypothesized that electrical stimulation of specific brain areas can regulate seizure susceptibility by raising seizure threshold or by interfering with seizure propagation by activating seizure-gating networks (53). Stimulus frequency seems to be a key parameter for stimulation-induced disinhibition of seizure-gating networks (14). Many of the studies that reported a decrease in seizure activity after stimulation of the basal ganglia and thalamic nuclei used a high-frequency stimulus. Mirski et al. (9) observed an elevation in pentylenetetrazole (PTZ)-induced clonic seizure threshold after 100-Hz stimulation of the anterior thalamus, whereas 8-Hz stimulation was proconvulsant. Why high-frequency stimulation was effective is unclear. It has been suggested that high-frequency stimulation induces a functional lesion-like effect, causing the stimulated nucleus to stop functioning because of depolarization block, neural jamming (disruption of the network caused by stimulation-induced neuronal impulses), or preferential activation of inhibitory neurons (54).
Potential therapeutic effects of LFS
Initial reports that LFS could interfere with the generation of kindled seizures were ignored because of an awkward experimental design and poor presentation of the data (37,38). However, a recent study by Velisek et al. (40) demonstrated that the delivery of LFS during kindling acquisition delayed the kindling process and therefore interfered with epileptogenesis in immature animals. Clinically, LFS has been shown to decrease interictal spiking in patients with temporal lobe epilepsy (44,45).
Several in vitro studies demonstrated an effective decrease in experimentally induced seizure activity by using LFS (41–43), and Khosravani et al. (27) used LFS to prevent the transition from interictal activity to seizure-like events in the hippocampal slice. Albensi et al. (43) found that both high-frequency and low-frequency stimulation blocked bicuculline-induced interictal spiking in the hippocampal slice. When the high-frequency stimulation was discontinued, the interictal spikes returned, but when the low-frequency stimulation was discontinued, the interictal spikes did not return.
Effect of LFS on kindling acquisition
In the study by Velisek et al. (40), 15 min of low-frequency square-wave stimulation in the amygdala delayed kindling acquisition in immature rats. AD duration was shortened, and the mean seizure score was reduced. In the present study, a low-frequency sine wave stimulus was delivered for 30 s to the amygdala of adult animals. A similar decrease in AD duration and behavioral seizure score was observed; however, an additional observation was the significant decrease in the incidence of AD in response to the kindling stimulus. The control animals in this study exhibited an AD 99% of the time. The failure of the kindling stimulus to elicit an AD can be attributed only to the preemptive presentation of the LFS.
An interesting aspect to the data was that the experimental animals became fully kindled in the same number of stimulations as did the controls. Given the observation that the experimental animals had fewer ADs than the controls, one interpretation of the data is that the LFS accelerated epileptogenesis. This is difficult to reconcile, given the decrease in other parameters that are often used to measure progression of the kindling process. Reports exist of kindling with long trains of low-frequency stimulation (55–57). However, in these studies, in addition to long stimulus trains, a high current intensity (1,000 μA) was used. The low-frequency stimulus used in this study had a short duration with a low current intensity and therefore was unlikely to be a kindling stimulus.
Effect of LFS in fully kindled animals
When fully kindled control animals were stimulated with the LFS plus the kindling stimulus, they began to fail to exhibit stage 5 seizures. A failure was defined as any seizure with a behavioral score less than stage 5. However, in these animals when a failure occurred, 90% of the time, it was a complete failure. No AD occurred in response to the kindling stimulus. Because these animals exhibited a stage 5 seizure 98% of the time when receiving only the kindling stimulus, the significant decrease in stage 5 seizures after the LFS suggests this decrease in seizure activity is due to the preemptive presentation of the LFS. Although not statistically significant, it is interesting that once the original experimental animals no longer received LFS, the incidence of stage 5 seizures increased.
We do not have an explanation for the observation that the difference in AD duration between the two groups during acquisition did not change once the crossover procedure was performed. A contextual component may exist to when in the kindling process the LFS was presented.
Why is low-frequency stimulation effective?
LFS appears to interfere with seizure activity induced by focal stimulation in the kindling model (37,38,58–60) and by chemical convulsants in the in vitro slice preparation (26,41–43). The decrease in kindled seizures observed in the present study after LFS can be explained by the findings of McIntyre et al. (60). They observed a 200% increase in AD threshold that lasted for 2 to 3 days after the stimulation with the same LFS that was used in this study. Obviously, an elevation in seizure threshold would make it less likely that a given kindling stimulus would elicit a seizure.
The LFS used in this study differs from stimulus paradigms that lead to long-term depression (LTD) or depotentiation (61–64). The classic LTD stimulus usually consists of a minimum of 15 min of 1-Hz square wave stimulation (62). In this study, seizures were successfully prevented with only 30 s of a 1-Hz sine wave stimulus. The contribution of the sine wave to the success of the LFS is unclear. Field stimulation with sine waves at 5 Hz and 50 Hz has been shown to influence hippocampal excitability; often lasting for a long time (min) after the stimulation was terminated (65,66). However, it is difficult to compare the effects observed after focal 1-Hz sine wave stimulation with what occurs after field sinusoidal stimulation. Because more charge is delivered during sine wave stimulation, it may be more potent than square wave stimulation. Nevertheless, if kindling occurs through a potentiation of excitatory pathways, then an LFS-induced increase in seizure threshold may occur through a depotentiation-like process.
In conclusion, we demonstrated that preemptive, 1-Hz sine wave stimulation can significantly decrease the incidence of amygdala-kindled seizures. Further studies will be required to determine whether low-frequency sine wave stimulation can be an effective therapy for some types of pharmacoresistant epilepsy.
Acknowledgment: This work was supported by NeuroPace Inc., the CURE Foundation, and the Helen Hayes Hospital Foundation. We thank Dr. B.K. Shah and Mary Wootten for assistance with statistical analysis and Sheeja Thomas for technical assistance.