Deep Brain Stimulation of the Subthalamic Nucleus as Adjunct Treatment for Refractory Epilepsy


Address correspondence and reprint requests to Dr. A. Handforth at VA Greater Los Angeles Healthcare System, Neurology Service (W127), 11301 Wilshire Blvd., Los Angeles, CA 90073, U.S.A. E-mail:


Summary: Purpose: We studied the efficacy and safety of bilateral subthalamic deep brain stimulation (DBS) for refractory partial-onset epilepsy in two cases.

Methods: This was an open treatment pilot study for subjects who had failed numerous medications and had seizure injuries. Seizure counts and adverse events were collected during a 3-4 month baseline, and for 26-32 months after DBS surgery, with AEDs held constant.

Results: Case 1, age 45, with bitemporal seizures, had about half the seizure frequency but still fell with injuries. Case 2, age 46, with left frontal encephalomalacia, had a frequency reduction of about one-third, but a more meaningful reduction of seizure severity and injuries.

Conclusions: Subthalamic DBS partly reduced partial-onset seizures, but the quality of life was more affected by seizure-related injuries.

Numerous studies in animal models over two decades have demonstrated that the basal ganglia modulate seizure susceptibility through the substantia nigra reticulata (SNr) and globus pallidus interna (GPi) (1–4). Evidence indicates that SNr/GPi output neurons exert a tonic proseizure effect. During seizures, these neurons increase firing, further promoting seizure propagation (5). The dominant excitatory control of SNr/GPi is by subthalamic nucleus (STN) afferents. Suppression of the STN is anticonvulsant (6–8). These observations suggest that STN deep brain stimulation (DBS) should suppress seizures in refractory epilepsy. A report of STN-DBS efficacy has been published by the Grenoble group (9). We compare our experience with two cases.

Subjects were eligible if they had six or more refractory partial-onset seizures per 30 days involving altered consciousness, were aged 18–65 years, were taking stable doses of one to three antiepileptic drugs (AEDs), and experienced seizure injuries. This was an Internal Review Board (IRB)- and Food and Drug Administration (FDA)-approved open-treatment pilot study of bilateral STN-DBS as adjunct treatment. Information was collected at each visit about adverse events, seizure counts, medications, and quality of life. Seizure counts did not include simple partial seizures. Subjects were seen in a 3- to 4-month baseline, and then at intervals of ≤2 months during the first postoperative year, and then at intervals of ≤6 months thereafter.

At implantation, MR images generated while a Leksell cranial frame (Elekta, Stockholm, Sweden) was attached were used to calculate STN coordinates with BrainLAB @Target software (Brain LAB, Heimstetten, Germany). Cell recordings were performed, and passes at 2-mm adjustments performed as necessary. To guide the final placement, macrostimulation using the inserted lead (model 3387-40; Medtronic, Minneapolis MN, U.S.A.) connected to a test stimulator (model 3625; Medtronic) assessed for stimulation-related adverse symptoms in the awake subject. Both STNs were implanted in the same procedure, and each connected to a subcutaneous pulse generator (model 7424; Soletra; Medtronic) via an extension (model 7494-51; Medtronic).

Initially, 60-μs pulse duration and 130-Hz pulse-frequency settings were chosen, but changed after 3 months to 90 μs and 185 Hz, respectively. Voltage did not exceed 3.5 V.


Case 1 was a 45-year-old right-handed homemaker with seizures since age 1 year after head trauma. Trials of phenytoin (PHT), phenobarbital (PB), diazepam (DZP), valproate (VPA), felbamate (FBM), topiramate (TPM), and primidone (PRM) failed. Videotelemetry evaluations showed bitemporal independent ictal origination. On enrollment, she had psychomotor complex partial seizures, numerous injuries, and was taking carbamazepine (CBZ; 1,200 mg/day) and gabapentin (GBP; 3,600 mg/day).

The baseline seizure frequency averaged 0.31/day. At implantation in October 2000, cell recordings compatible with STN activity were obtained bilaterally. A postoperative MRI confirmed accurate STN placements. Initially, two negative contacts were chosen on each lead; this induced adverse symptoms. Next, contact 0 was set positive, and the negative site varied on both leads at the same time. Finally, the left lead setting was unchanged while the right was adjusted to optimize seizure control.

Figure 1A displays seizure frequency up to February 2003, when AEDs were then changed. The final long-term setting achieved about a 50% reduction, but seizure-related injuries continued to occur (Fig. 1B). Stimulation-associated disequilibrium and leg dyskinesias resolved within 2 hours after reprogramming. A 10-day period of disinhibited talking and imbalance promptly resolved with reduction of right-lead voltage.

Figure 1.

A: Seizure frequency compared with baseline in successive 3-month STN DBS treatment epochs for case 1 (left bar of each pair) and case 2 (right bar). Case 2 had a second baseline with continuing right-side DBS, but the left side was turned off. The mean seizure-frequency percentage reduction after baseline in case 1 was 44%; for case 2, the mean seizure-frequency reduction after the second baseline was 29%. B: Seizure-related injuries incurred by cases 1 and 2, expressed as frequency per 100 days. Injuries decreased for case 2, but not for case 1.

Case 2 was a 46-year-old left-handed homemaker with seizures since encephalitis at age 6 months. Trials of PHT, CBZ, VPA, TPM, FBM, clonazepam (CZP), TGB, GBP, partial left frontal cortical resection, and vagus nerve stimulation failed to control seizures. She was taking LTG (500 mg/day) and PRM (750 mg/day), with rectal DZP rescue, and had numerous injuries. The vagus nerve stimulator was removed 1 month before the study. Her most common seizure was complex partial with head drop, bilateral upper extremity abduction, a fall, and incontinence. Secondarily generalized seizures were less frequent. An MRI showed left cerebral hemiatrophy, with opercular-frontal encephalomalacia. The baseline seizure frequency averaged 3.05/day, including 15 secondarily generalized seizures.

At surgery, distortion by atrophy and prior craniotomy led to a left-lead assignment more dorsal than that of the right. Initially, the left-side settings were unchanged while the right side was adjusted, resulting in reduced seizure severity and fewer injuries (Fig. 1). No more generalized convulsions occurred. An MRI showed intra-STN placement of the right but not the left lead. The left generator was then programmed OFF, and a second baseline elapsed, with the right-generator settings unchanged. After a second surgery, an MRI indicated the left lead was now intra-STN.

The right-side settings were then held constant while adjustments were made on the left. At the final setting before AEDs were changed in July 2003, the seizure frequency was reduced by one third (Fig. 1A). Seizures were milder, with the emergence of simple partial seizures. Because seizure injuries were fewer (Fig. 1B), case 2 considered her quality of life improved, in contrast to case 1. Adverse events resolved with reprogramming and included left arm and leg dyskinesia, diplopia, and disequilibrium.

In conclusion, initial or revised STN-DBS lead placement was successful based on MRI and characteristic stimulation-related symptoms in our cases. Our observation of dyskinesias demonstrates that STN-DBS of non-Parkinson subjects can produce this reaction. Of interest was DBS-induced behavioral disinhibition in case 1, well known to occur with STN-DBS in Parkinson disease (10). Seizure-frequency reductions were partial at one third to one half, and improved quality of life corresponded with reduced injuries.

Our results may be compared with observations on five cases by Chabardes et al. (9). A 17-year-old with insulofrontal epilepsy did not respond to STN-DBS. Their other four cases had epilepsy from central cortical dysplasia or porencephaly, or anterior seizure onset in a case of Dravet syndrome, and responded with seizure-frequency reductions averaging 64%. Differences in subject selection or therapeutic technique may have resulted in these more favorable outcomes. One case had three separate leads implanted in the STN on one side and two in the other. Another had two leads in one STN and one in the other. The other two cases had single-lead DBS, but received 5.2- and 4.0-V stimulation.

In contrast, our STN-DBS therapy followed that of community practice for Parkinson's disease, with a single lead in each STN and amplitudes <3.6 V to avoid excessive battery drainage. Our observations suggest that STN-DBS is capable of reducing seizure severity and frequency, but the response in these two cases did not exceed that often seen with a new antiepileptic therapy.


Acknowledgment:  This study was supported by Medtronic Inc., Minneapolis, MN, and by the U.S. Department of Veterans Affairs.