Epilepsy affects some 2.7 million Americans—more than Parkinson’s disease, multiple sclerosis and amyotrophic lateral sclerosis (Lou Gehrig's disease) combined. More than half of patients can achieve seizure control with treatment, yet almost a third of people with epilepsy have a refractory form of the disease that does not respond well to existing antiepileptic drugs. Nor are these patients typically helped by the one implanted device—Cyberonics' Vagus Nerve Stimulator (VNS)—that has had U.S. Food and Drug Administration approval for treatment of epilepsy since 1997.

Because epilepsy causes repeated, sudden seizures, people with the condition would benefit greatly from a therapy that can detect seizures just as they are starting or, eventually, predict them before they begin and prevent them from happening. A new generation of implantable devices is looking to pick up where medications—and even the VNS—often leave off, at least for people whose seizures routinely begin in one part of the brain (the seizure focus). "Closed-loop" devices are designed to monitor the seizure focus, detect patterns of electrical activity that indicate a seizure is beginning, and quickly respond without external intervention. Such responses could include electrical stimulation, cooling or focused drug delivery—all meant to interrupt the activity and stop the seizure.

Closed-loop devices are considered a new frontier in epilepsy treatment because of their responsiveness. By comparison, the VNS is an open-loop device that stimulates the vagus nerve—a pair of nerves running from the brain stem to the abdomen—to deliver mild electrical pulses (which mitigate the electrical activity of seizures) to the brain on a consistent schedule rather than in response to detected seizure activity. The concept of a closed-loop device for epilepsy comes out of the cardiac world, jumping off from implanted defibrillators that monitor the heart and deliver stimulation in response to an event.

Responsive neurostimulation
So far, only one closed-loop device has reached human trials: NeuroPace's Responsive Neurostimulation System (RNS), an electrical-stimulation implant with two leads, each containing four electrodes, placed in the brain at the seizure focus. The RNS detects electrical activity that denotes the start of a seizure and delivers direct electrical stimulation to interrupt the activity and normalize the area. The device is surgically positioned in a section of the skull, can be accessed via outpatient surgery when the battery has to be changed, and is imperceptible to the patient and others—all strong design advantages for patients and doctors. The implant, which is now seeking FDA approval, also records information on electrical activity in the brain throughout the day for later review. The RNS has a laptop-based wand interface for remote patient monitoring.

Results of the RNS trials, which tested the implant in conjunction with medications, have been mixed: seizure frequency was reduced by about half in approximately 50 percent of patients. "For a patient to go though permanent implanting of the device on the skull, and electrodes over the brain, which is what is needed for RNS, you'd want it to eliminate most or all seizures, which isn't the result in most patients," says John Miller, director of the University of Washington School of Medicine's Regional Epilepsy Center at Harborview in Seattle. Possible ways to improve the device's effectiveness, Miller says, could include refining patient selection, improving electrode placement or honing the RNS's detection process so that it can pick up seizure activity earlier.

Work in closed-loop electrical stimulation is also happening at Boston’s Center for Integration of Medicine and Innovative Technology, where researchers are effectively attempting to turn the VNS into a closed-loop device by developing a nonimplanted add-on system to detect early seizure activity and automatically fire the VNS in response. The VNS comes with a therapy magnet wristband that allows wearers to stimulate the device if they feel a seizure coming on (a sensation called an aura), but not everyone is physically able to do so once the aura begins. The CIMIT system automates the process, activating the VNS once the start of a seizure is detected through electroencephalogram and electrocardiogram readings.

Cool it
Another key area of closed-loop research is focal cooling. Here, an implant—after detecting the onset of a seizure by sensing a rise in brain temperature at the seizure focus, which may slightly precede the start of abnormal electrical activity—rapidly cools the involved region to halt the event. The warming associated with the seizure focus makes thermal detection and cooling a potentially promising technique. One center of focal cooling research is the University of Kansas Medical Center, where Ivan Osorio, professor of neurology, has collaborated with an international research partnership to design a prototype implant with funding from the U.S. Department of Energy. Work on cooling is also in progress at other sites, including Yale University and the University of Minnesota.

"I think cooling is the most promising approach," says Miller, who collaborates on cooling research led by a University of Washington colleague. "If a particular cooling temperature can be found that prevents seizures, but does not injure the brain or interfere with normal brain function, it would be possible to maintain the region of brain around the seizure focus at that temperature all the time, so that it would not be necessary to detect the seizures to apply the therapy."

Targeted drug delivery
The third possible mode of operation for closed-loop devices would use convection-enhanced drug delivery (CED). CED involves feeding seizure-halting medications directly to specific areas of brain tissue through an implanted catheter; the concept of CED is designed to avoid the systemic side effects of giving medications orally and having them suffuse through the bloodstream in order to reach the brain.

Yet CED may ultimately prove more useful on a set infusion schedule, rather than linked to a responsive, automated seizure-detection system. "Our current conception of how CED would be used in epilepsy is that patients would receive periodic infusions of a long-lasting antiseizure agent into the epileptic brain region," says Michael Rogawski, chair of neurology at the University of California, Davis, whose lab is working with British Columbia–based biopharmaceutical company MedGenesis Therapeutix to develop an implantable CED device for epilepsy. "Seizure control might be maintained for months," he says. "This approach greatly simplifies the technical challenges in comparison with a device that must sense and deliver a drug on a moment-to-moment basis."

Deep-brain stimulation
With electrical stimulation, too, some patients will find that an open-loop device that fires consistently works better—like the VNS, or Medtronic's Deep Brain Stimulation (DBS) implant for epilepsy, which the FDA is now reviewing. Similar to the company's widely-used DBS technology for Parkinson’s disease, the DBS for epilepsy is placed within the brain and consistently stimulates a region called the anterior nucleus of the thalamus, which helps control the electrical excitability of the cortex.

Unlike closed-loop devices, which typically require a distinct seizure focus, the DBS can be used to treat patients whose seizures appear to engulf the entire brain, or large portions of it, at once. "If you look at the population of patients who have these very unlocalizable, diffuse seizure disorders, folks who are having many, many seizures a day and are just devastated—if you can control some of those seizures even in some of those patients, you've done a great good for the families and the patients," says Dennis Spencer, chair of neurosurgery and director of the Epilepsy Surgery Program at Yale University School of Medicine. "We think that the DBS will open up a path for therapy."

Closing the loop
Closed-loop technologies for epilepsy face several hurdles. Skeptics note that brain surgery poses significant risks, and that the benefits of implanted devices will not always outweigh those dangers. There are also concerns about the possibility of false positives—detection of electrical activity that turns out not to be a seizure. "If the intervention did cause a transient interruption in brain function, it would be undesirable for the patient," Miller says. "For example, if the area that was being affected mediated language, the person might have a brief interruption in the ability to speak."

Researchers also acknowledge that in a condition as variable as epilepsy, there will never be a single solution, such as cooling, stimulation or drug delivery alone. "We may need to use more than one modality to fully control epilepsy," Osorio says. "But all of that hinges on the ability to detect in real time—and to quantify—seizures."

Although the design of first-generation closed-loop devices is just beginning, theoretical development of the second generation is already underway. Because people with epilepsy never know when and where a seizure will occur, the goal of second-generation closed-loop devices will be finding a way to predict seizures before they begin and intervene to prevent them. "You can detect seizures, but you're still detecting them too late to really have a major therapeutic possibility," Spencer says. "Prediction is where we're really looking to put our eggs—in that basket."