"It's OK, they know. I came home intoxicated the other night at 5 a.m."
Thus begins another skirmish in Devinsky's long-standing war on the fecklessness of youth, their natural tendency to forget their medication, stay up all night working on a term paper and propel themselves into a hypernormal state by swilling vodka or cough medicine. The human brain, for all its marvels, has one glaring fault, the tendency to discount future losses—such as the risk of a fatal seizure—relative to present pleasure, if that's the right word for how 15 Benadryls makes you feel. This psychological quirk has cost the lives of more than 150 of his patients, mostly in their teens and 20s. "I had a patient who hadn't had a seizure in two years," Devinsky tells her. "Last fall I got a call. He went off to college, stayed up late one night at a party and never got up the next morning.
"Do you want to be the mother of two children wondering when your next seizure is coming?"
Striking the right balance of medication is especially important in children as they pass through the critical years for learning. Devinsky raises this with the father of a fourth-grade girl who suffers from brief absence seizures, just five seconds at a time, a few times a day. She has already been on Topamax, Keppra and Zarontin, and is now being treated with Lamictal. Devinsky is hesitant to give her more drugs, but he worries that the absence seizures, which seem so benign, might be causing subtle damage to her learning and behavior. "It's like you're trying to read something," he says, "and I keep tapping you on the arm, like this. For an adult it might not be such a big deal. But if you have this constant distraction during a critical learning window, you miss out, and you don't ever get that back." After talking it over with her father, he decides to up the dosage of Lamictal, and asks the father to let him know of any changes.
Devinsky doesn't disdain technology, but he has a keen sense of its limits. He has all the latest scanners and imagers at his disposal, but knows that most of the time whatever is causing a seizure won't show up on them, at least not definitively. The most important tool in his lab is the EEG machine, which monitors brain activity through electrodes on the scalp, a technology that was invented in 1929.
There are, at this time, only a few ways to treat epilepsy, and applying them is still an art as much as it is a science. What works for one patient often has no benefit for another with identical symptoms. Researchers still don't understand, 80 years after it was discovered, why some children can control seizures with a ketogenic diet, high in fat but so low in carbohydrates that even the amount of sugar in toothpaste can be too much. Nor do they know why two thirds of patients can control their seizures with drugs, but not the rest. Since the 1960s about 30 different compounds have been approved to treat epilepsy, although most neurologists, says Devinsky, have a stable of around 10 that they generally rely on. New ones are being developed and put into use regularly, but that progress is deceptive, says Michael Rogawski, a neurologist at University of California, Davis, who studies epilepsy therapies. The new drugs may have fewer side effects or less toxicity than older ones, but by and large they work only for the same percentage of patients who were already being helped; the number of refractory cases hasn't changed much over the years. And it's still impossible to know in advance which patients will benefit, and from which drugs. "I can look at a person, do all the testing, even see their seizure, and I can't tell which drug they'll respond to," says Carl Bazil, who heads Columbia's Comprehensive Epilepsy Center. "There must be something about their action in the brain, but we don't know what it is." Many researchers believe that the next important breakthrough won't be a new drug at all, but the development of a subcutaneous pump that can deliver medication directly to the right spot in the brain, bypassing both the organs of the rest of the body, and a delivery mechanism that relies on a teenager late for the school bus to remember where he left the bottle.
Devinsky tries not to use more than two different drugs on a patient at a time, but even so, the number of ways to combine different dosages of two drugs from among 10 is infinite, especially compared to the time he has to get someone's seizures under control before they fry their synapses beyond repair. When he first treated Wheeless he was on Depakote, which was making him lethargic and fat; Devinsky substituted Felbatol in combination with Lamictal. When that didn't work well, Devinsky switched him to Felbatol with Keppra. The improvement was dramatic—so much so that he kept Wheeless on Felbatol even after reports appeared of deaths from liver failure and aplastic anemia. For the relief Wheeless received, the risks seemed small enough to Devinsky, and manageable as long as he closely monitored his blood chemistry. Up till now, he's been right, but if the Felbatol-Keppra combo is losing its effectiveness, he will have to consider increasing the dosage, or switching him to another medication. He wants to improve Wheeless's sleep schedule and control his mood. Depression works synergistically with epilepsy; it can promote seizures, and seizures, of course, can make you depressed. And if those don't work … well, he'll think of something.
The complexity of managing epilepsy this way has led patients and their parents to agitate for research on a cure. Although "cure" is in the name of the foundation Devinsky founded, he uses the term sparingly, especially around patients; just hearing it can encourage them to stop taking their medications, and after a while they show up back in his office with new seizures. Cures do happen, except they're most often spontaneous and random; about half of all children with epilepsy outgrow their seizures and can be taken off meds after several years. In theory, gene therapy could someday cure some cases of generalized epilepsy. The first step would be to identify the mutations that cause the condition, and thousands of patients are being recruited for a study aimed at doing that. And, also in theory, and sometimes in practice, partial epilepsy can be cured by removing the part of the brain where the seizure originates—although most patients continue to take anti-seizure drugs afterward as a precaution.
The principles of epilepsy surgery have been known since the first operations took place more than a century ago: identify the area where the seizure begins and cut it out, as precisely as you can. Obviously this excludes most patients with generalized epilepsy, since you can't take out an entire brain (although in rare cases, you can remove half a brain; the patients sometimes compensate surprisingly well). Werner Doyle is one of two neurosurgeons at the NYU center; he does around 260 procedures a year, there and at St. Barnabas. In contrast to Devinsky's holistic approach to patient quality of life, Doyle's discipline requires an obsessive focus on brain stuff, on the paper-thin margin between diseased and healthy tissue. Sometimes, he says, he can feel the difference with his dissecting instruments, which cut easily through normal gray matter but meet resistance where the tissue is scarred from repeated seizure.
Like many surgeons, he is matter-of-fact about the manual dexterity he deploys inside the skull. The intellectual challenge for him is to find the focus of the seizure and trace the networks along which it spreads, while also mapping the boundaries of the unaffected areas. The basic tool for that is the EEG, but the standard device, with around 20 external leads, is a clumsy instrument for mapping a three-dimensional brain. That has led to the adoption of intracranial electrodes, which are inserted into the brain and can sample as many as 200 points. Inserting these requires opening the skull and takes as much of Doyle's skill and time—six or seven hours—as the subsequent operation itself. Patients may stay in the hospital for a week or more, trailing a Medusa-head of wires, under continuous monitoring both by EEG and video, while the doctors wait for the iconic seizure. Then Doyle goes to work.
In contrast to many types of neurosurgery—for tumors, notably—epilepsy surgery is generally performed on healthy people, and the risks are not that high. But Doyle believes, and hopes, that the future of his field lies not in the relatively blunt procedure of ablation, or cutting tissue, but in augmenting brain function through electronic devices. The first of these, the vagal nerve stimulator, has been in use since 1997, and more than 50,000 have been implanted. Its purpose is to disrupt an incipient seizure by sending an electrical signal to the brain through the vagus nerve in the neck. Again, no one knows exactly why it works, but it often does. Deep-brain stimulation, which has been successful in treating Parkinson's disease, is meant to do the same thing, by delivering a pulse directly to the brain itself, but clinical trials for its use in epilepsy so far have been disappointing. The VNS, though, is a primitive tool, which delivers a small current for a fixed duration at regular intervals of several minutes. Researchers now are working to develop a responsive device that could be implanted in the brain and "sense when a seizure is beginning, to release a small amount of medication or electrical stimulation where needed," says Harvard neurologist Steven Schachter, president of the American Epilepsy Society.
The key to such a device will be the computer algorithms that can predict a seizure, ideally in the "aura" stage that sets in before it even begins. We are accustomed to the idea that computers can process limitless amounts of data, but the brain, with its 100 billion neurons, each linked to as many as 10,000 others, is pushing the limits of information theory. The effort represents an audacious assault on a devastating disease, just what Devinsky was dreaming of in medical school. To conquer epilepsy we will have to outwit our own brains.
