Antiepileptic drugs have provided tremendous relief for many people with epilepsy. But around one-third of patients with epilepsy continue to have seizures despite medication, according to Brian Litt, M.D., an associate professor of neurology and bioengineering at the University of Pennsylvania, and a member of the Epilepsy Foundation’s professional advisory board.

Epilepsy surgery can cure another 7 to 8 percent of patients, but many patients cannot undergo the procedure because their seizures originate at multiple seizure focuses, or because a focus is in or close to a critical region of the brain where surgical removal carries a high risk of impairing normal function.

Currently, such patients have little recourse. One implantable treatment device is FDA approved: the vagus nerve stimulator – which is inserted under the skin near the collarbone, where it periodically applies mild electrical stimulation to the vagus nerve that connects parts of the upper body to the brain, which can lead to a decrease in the number and severity of seizures. Although the device is safe, it isn’t universally successful. About one-third of patients experience a major reduction in the number of seizures, another one-third experience moderate improvement, and the final third experience no change or even an increase in seizures.

Today, however, advances in computer software and engineering have produced a new generation of implantable devices that may hold greater promise, though proponents insist it’s too early to be sure they are a significant improvement over the VNS. The new devices take a more direct approach. They provide electrical stimulation to the seizure focus. ‘Open loop’ devices provide constant or intermittent stimulation, similar to the VNS.

Others are taking this concept one step further, taking advantage of advances in computer software that can analyze an EEG signal in real time and recognize the beginnings of a seizure. An implantable device with a microchip running such software, or a ‘closed-loop’ system, only delivers a therapeutic intervention when it is needed. Part of the reason for a more controlled approach is that animal studies have shown electrical stimulation – the most common therapeutic option – can have a variety of effects, including causing local changes in function and even seizures, in rare cases.

“Nobody is suggesting that an open-loop device can cause injury or epilepsy, but the brain is a very dynamic organ, so, theoretically, less stimulation is better than more. But this has not been clearly proven,” Litt said.

In fact, efforts to treat epilepsy using electrical stimulation of the brain date to the 1970s, with a number of small human studies showing occasional glimmers of improvement. But there have also been conflicting results, in part because researchers used a variety of devices and parameters on patients with different types of epilepsy. More controlled studies using closed loop devices could improve that track record. There is a clear precedent in the treatment of arrhythmias, which are abnormal heart rhythms that are also related to electrical disturbances.

“Closed-loop seizure devices were inspired by implantable cardiac defibrillators (pacemakers), which have been wildly successful… so people are trying to port that same type of technology over to seizure control,” Litt said.

The ‘Vanguard’

NeuroPace Inc., based in Mountain View, Calif., is currently developing an implantable device called the Response Neurostimulator. The RNS is a closed-loop device with programmable software that continuously records an EEG. It then sends an electrical pulse into the surrounding areas of the brain when it detects signals characteristic of an oncoming seizure in hopes of stopping it before it spreads and gathers strength.

“The analogy is, you detect a little spark and try to put it out before it causes a fire,” said Martha Morrell, M.D., who is a clinical professor of neurology at Stanford University, and the chief medical officer of NeuroPace Inc.

In an earlier NeuroPace-sponsored feasibility study of 65 patients with medically-uncontrolled seizures, patients experienced no serious side effects related to the device, and the majority responded with a decrease in both the frequency and severity of seizures. Based on that trial, the FDA granted the company approval to conduct a pivotal investigational trial, the results of which the agency will use in determining whether to grant approval. That trial, launched in December 2005, will enroll 220 people with medically-uncontrolled partial-onset seizures at 28 sites across the United States. The company hopes to complete enrollment in 2007.

NeuroPace researchers hope to learn more from the trial than just the efficacy of the device. The RNS will record EEG data from the patient’s brain, which can then be uploaded through a hand-held device to a computer for further study. Morrell anticipates that the devices will detect seizures so mild that the patient does not even notice them.

“We don't necessarily know whether these electrical seizures are significant or not,” Morrell said, but she continued by saying they should inform future research.

Steven Rothman, M.D., a professor of neurology at the Washington University School of Medicine who is working on his own innovative device, said he is impressed with NeuroPace’s device. “The results I’ve seen suggest that the RNS is very promising."

The ‘Smooth Operator’

Most electrical stimulation devices use high-frequency, and higher energy, electrical stimulation. This, in effect, overwhelms the neural network so that it becomes temporarily quiet. The approach shows some promise, but some researchers suspect that this brute force approach is a bit extreme.

“It’s often the same effect as surgically removing that part of the brain. We refer to it as a reversible lesion,” said Steven Schiff, M.D., Ph.D., a professor of neurobiology and the director of the Center for Neural Dynamics at George Mason University.

Along with his collaborator, Bruce Gluckman, Ph.D., a professor of physics and astronomy at George Mason, Schiff is working on a second generation electrical stimulation approach. The method would use more frequent but milder stimulation to gently guide neural systems and modify the excitability of the individual cells.

The idea is that a device could continuously read EEG signals and react to an unusual pattern by giving it a gentle electrical ‘nudge.’ The device would then note the effect in the EEG and make more adjustments as necessary, gently guiding the brain back to a normal electrical pattern.

The idea is analogous to “the way you keep your eyes open as you drive, getting continual feedback from the road so that you make small adjustments to the left or right to keep the car centered in the lane,” Schiff said. Gentle corrections work better than waiting until the car is significantly off course and then jerking the wheel back to center. It works, but the ride is much less smooth.

Schiff became interested in feedback mechanisms because he felt that that high-frequency stimulation is too unpredictable.

“There was no knob that we could turn left and right to balance the system without some operating point. We thought we probably had to use electric fields and currents in a continual mode – you couldn’t just bang away at it with occasional pulses,” he said.

Schiff has begun testing the system in rats. If successful, a feedback system could be incorporated into the next generation of implantable devices. Schiff estimates that his approach would use between one-tenth and one-hundredth of the electrical power used by current devices. By minimizing the amount of stimulation required, he also hopes to make such stimulation safer than the current higher powered approaches. He believes that a feedback mechanism could potentially control the wayward patterns of seizures without affecting normal brain activity, and do it using the least possible electrical stimulation.

The ‘Ice Man’

Electrical stimulation is not the only therapeutic intervention being investigated for implantable devices. Another approach, headed by the aforementioned Rothman, is still in the animal experimentation stage. It seeks to chill neurons, which dampens their activity and interrupts a seizure. After the treatment, cells warm up again and return to normal functioning.

The device is based on work dating back to the late 19th century, when German scientists showed localized cooling could diminish neurological function. Throughout the 20th century, researchers used cooling in animal studies to identify the location of specific brain functions. Rothman’s device consists of an electrical circuit that, when activated, cools down at one end and heats up at the other. Implanted near the focal point of a patient’s seizures, the cool end could chill nearby cells, dampening their activity. His team demonstrated it can interrupt seizures in rats.

Rothman says technical hurdles must be overcome to build a device small enough to implant in human subjects, though his team is working on the problem. In the meantime, he plans to use a water-cooled device to test on patients undergoing brain-mapping to find the seizure onset zones in preparation for surgery. He hopes the study will demonstrate the amount of cooling necessary to interrupt a seizure in humans. The results will help in the design of a device for human use.

“An educated guess would be that in situations where you can localize a patient’s seizures to an area on the surface of the brain, cooling might have the most specific and localized effect. [Electrical] stimulation, which can operate at larger distances, may not require such precise localization of the seizure focus,” Rothman said.

That could be an advantage because some patients’ seizures are more difficult to pinpoint.

The ‘Oracle’

Electrical stimulation and chilling show promise for stopping a seizure once it starts, but other researchers have in mind a more ambitious challenge. They seek to predict seizures well before they begin.

Software advances make it conceivable that a device could detect an unusual EEG pattern that starts minutes or hours before a seizure actually begins. It could allow intervention far earlier in the process, perhaps when a nascent seizure is more easily corrected.

“Many of us believe that if you wait until the seizure has taken hold, it has already spread to a larger portion of the neuronal network, and it could be more difficult to stop,” Litt said. “But no trials have clearly demonstrated that.”

Klaus Lehnertz, Ph.D., who heads the neurophysics group in the department of epileptology at Germany’s University of Bonn Medical Center, has been in the forefront of developing predictive software. He doesn’t believe that standard statistical methods will be capable of it, because EEG signals, like the brain that generates them, are constantly changing and adapting. To tackle the problem, he is turning to cellular neural networks, which are a little like miniature computers linked to one another, much like nerve cells are linked. Like nerve cells, the miniature computers ‘talk’ to each other, and this structural similarity might allow it to better mimic brain activity than a standard computer.

Lehnertz reports success in predicting seizures retrospectively - that is, applying the cellular neural network’s prediction software to a range of already collected EEG data and successfully identifying a pre-seizure state. Doing it prospectively – using EEG data to predict a seizure that hasn’t happened yet, as would be required if it were used in a device – is a more difficult challenge.

“I think it’s a little too early for that,” he said, but he continued by saying the day will come when it is possible.

Will prediction and early intervention in an oncoming seizure be more effective than detecting a seizure and reacting to it? NeuroPace’s Morrell isn’t sure.

“I think the ability to predict seizures would be really helpful in the sense that it would warn individuals, and hopefully people would know not to drive a car, or to go to a safe place, or even take medicine prophylacticly,” Morrell said. “I have no idea whether it would be a benefit to provide electrical stimulation to the brain well in advance, or if it is better to save the stimulation for the moment that the event starts.”

Whether or not researchers ever succeed in predicting seizures, it seems likely that devices will play an important role in epilepsy therapy in the coming years. What lies ahead? Safety is a key concern, says Ivan Osorio, M.D., who is a professor of neurology at the University of Kansas Medical Center, and is also working on a closed-loop electrical stimulation device.

“I think the primary challenge we face is to demonstrate that implantable devices do not trigger more seizures... or change the brain in other ways. They may have cognitive or behavioral effects,” Osorio said.

Although they are currently being tested only in patients that have failed drug therapy, implantable devices may eventually find use as the frontline therapy for epilepsy. Litt points to the example of pacemakers. Early clinical trials of the devices were actually halted because they were so effective that it was deemed unethical to continue to withhold them from people in the control group.

“Now the implants are the first-line therapy for this medical condition, and the market for the drugs has diminished considerably,” Litt said.