Patch clamping measures a very small electrical current. How small? A milliampere will produce a tingling sensation as it flows through a fingertip. A microampere is thousand times smaller and represents the current that flows while a finger touches a modern phone’s screen. With another thousand-fold squeeze, a nanoampere comes into view. Motion sensors and smoke detectors operate at this level of sensitivity. The currents measured with patch clamps are usually less than one nanoampere.

  HUGE POTENTIATION

  Prompted by her daughter’s conversation, Parker called Jenkins and explained the struggle with Anna’s sleepiness. She asked if he would be willing to help. Jenkins had just a few people in his lab and was doing hands-on experiments, exploring science that looked promising.

  At the time, Paul Garcia, an MD/PhD interested in neuroscience, was working off and on in Jenkins’s lab. Garcia, then beginning his anesthesiology residency, had a connection to Rye as well, having shadowed him as a medical student. Garcia said he had helped revamp Jenkins’s lab to handle clinical samples. “The system was all set up,” Garcia said.

  Parker already had a sample of Anna’s CSF when she called Jenkins. In a repeat of her diagnostic workup, Anna had fallen asleep quickly in a multiple nap test, even after sleeping for fourteen hours the night before. Anna’s lumbar puncture was performed in the hospital, since she tended to “get the vapors” and pass out, according to Parker. A headache often results after a lumbar puncture, with the intensity related to how much fluid is removed.

  Jenkins took charge of the first set of experiments on her CSF. His notebook from May 2007 read: “Huge potentiation.” If he squirted a bit of GABA onto a patch of kidney cell membrane, it let in a tiny amount of current, which his patch clamp apparatus was set up to detect. Anna’s cerebrospinal fluid contained a negligible amount of GABA, but whatever was in it made the effect of the GABA he provided more than twice as strong. Jenkins would write that her CSF contained a level of GABA-enhancing activity “equivalent to that caused by general anesthetics at loss-of-consciousness concentrations.”20 He was astounded at the neurochemical state Anna had been living in.

  There weren’t many “off the shelf” options available for diminishing GABA signals. Several relevant compounds, such as picrotoxin or cicutoxin, are known more as poisons than as beneficial drugs. They even have “-toxin” in their names, because taking away the calming influence of GABA can result in a seizure.

  The emergency room antidote flumazenil stood out as available and FDA approved. How the idea to try it with Anna solidified was remembered differently by the main players. Parker said she had already thought about flumazenil before contacting Jenkins: “Why go through all of that, if there wasn’t something we could do for her?” Jenkins said a specific remedy was not discussed initially but emerged as an option after his experiments suggested it.

  When Jenkins added flumazenil to a patch clamp experiment, it reversed the GABA-enhancing effects of Anna’s CSF. He used a high concentration of the drug, possibly surpassing what can feasibly be introduced into the nervous system of a living person.21 Even so, his laboratory findings raised the question: would flumazenil accomplish the same thing in Anna?

  ON MY RADAR

  Flumazenil, developed by Hoffmann-La Roche, was thought to be helpful in two situations. The first is when someone is sedated for an uncomfortable medical procedure. To make the patient relaxed and the process less memorable, a normal practice is for the doctor to give someone a sedative, such as diazepam (Valium) or midazolam (Versed). If the patient receives too much, the sedative can inhibit parts of the brain that control breathing, which is dangerous. Flumazenil can reverse the sedation.

  A second situation calling for flumazenil is when someone arrives at the hospital and appears to have overdosed on benzodiazepines. According to data from the American Association of Poison Control Centers, flumazenil is used in this way about two thousand times per year. It is analogous to naloxone, the fast-acting antidote against opiate overdose.

  At the time the Emory group was considering it, flumazenil had already accumulated a weird side story. Flumazenil could revive people with liver failure who had been in a coma minutes earlier. Some studies had explored whether it could be a cognitive enhancer—that is, whether it could make people smarter. A few addiction specialists latched onto it, claiming that flumazenil could help people with withdrawal symptoms.

  Flumazenil’s effects pointed to an unsolved puzzle. Neuroscientists still think that flumazenil doesn’t do much by itself. It elbows benzodiazepines out of their slots on GABA receptors, but its direct effects on those receptors are weak. That explains why endozepines, short for “endogenous benzodiazepines,” were proposed to exist. In concept, they resemble endorphins, messengers produced by the body that tickle the same pain-relief receptors that morphine does. If the body produces its own painkillers, it could make its own calming or sleep-inducing molecules too. In some situations, such as hepatic encephalopathy, a term for the effects of liver failure on the brain, the nervous system seems to accumulate benzodiazepine-like chemicals. These substances might be what flumazenil was kicking out of the way.

  Jenkins was familiar with previous research on endozepines and hepatic encephalopathy. After Imperial College, he had worked in the United States as a postdoc with Neil Harrison, another important player in defining the mechanisms of action for anesthetics. One of his former lab-mates had been testing whether fragments of heme, breakdown products from decaying red blood cells, might be partly responsible for symptoms of hepatic encephalopathy.22 This prepared him for his collaboration with Rye and Parker. “That was going on next to my bench for a few years,” he said. “It was the reason I was ready for Kathy’s call. Those ideas were already on my radar.”

  But a curious episode in Italy had cast a shadow over the endozepine idea. Neurologists from the University of Bologna had published several papers on a disorder they called “idiopathic recurring stupor.” People from Italy and other countries had been falling asleep suddenly, without adequate explanation. Flumazenil woke them up, so their sleepiness was attributed to endozepines. A cluster of cases was revealed to instead come from someone slipping sedatives to their neighbors.23 “Dave and I talked about those papers,” Parker said. “We agreed they were very quirky. Initially, we did wonder if Anna’s condition was self-inflicted. We did blood tests and urine tests. She was clean.”

  There was enough logic behind others’ use of flumazenil that Parker and Rye were willing to pursue it. Rye had briefly woken up a couple patients using flumazenil in the 1990s. One had elevated ammonia levels—a sign of liver trouble. Rye recalled these experiments as taking place “in a closet.” Also, because of his familiarity with Parkinson’s research, Rye knew that a colleague in Houston, Bill Ondo, had completed two small-scale studies on flumazenil’s effectiveness against Parkinson’s motor symptoms.24 While Rye was skeptical about endozepines, he was attracted to trying something first and figuring out the hows and whys later. “Kathy is a process person,” Rye said. “I’m a results person.”

  TREATMENT OR TOXIN?

  Rye and Parker also were aware this line of thinking was not risk-free, because flumazenil’s use in emergency departments had become a topic of debate. Along with naloxone, flumazenil was once proposed as part of a “coma cocktail” for reviving intoxicated people, but medical opinion had turned against it.25 A 2004 paper summarizes the problem: “Flumazenil—Treatment or Toxin.”26 The risks of flumazenil were often not worth the potential benefits, compared to introducing a tube to aid a patient who was having trouble breathing.

  Many papers showed that flumazenil can whisk away a benzodiazepine’s sedative effects and restore alertness. However, there were occasional reports of seizures and other adverse events. For most of these, researchers concluded that other drugs, such as tricyclic antidepressants, were responsible or that the antidote had unmasked underlying problems, such as epilepsy or panic disorder.

  In a few case reports, it’s difficult to say why things had taken a wrong turn. In a small town in Ireland, a doctor was performing an endoscopy on a thirty-year-old woman suspected of having an ulcer.27 She was given a large dose (15 milligrams) of diazepam. After the procedure, the woman displayed shallow breathing, an indicator that the dose may have been too high. Alarmed, the doctor provided oxygen and half a milligram of intravenous flumazenil. The woman promptly woke up, but a few minutes later she lost consciousness and began to have convulsions, which progressed to status epilepticus: a seizure lasting several minutes, carrying with it an increasing mortality risk. Once the seizure was quenched, she did survive and recover. The woman did not have other medical problems and had not taken other drugs. Conclusion: watch out—a message many doctors absorbed over the years.

  Benzodiazepines make brain cells less excitable, but the body and brain adjust to the drugs’ presence. It’s what makes them so addictive. When they wash out, brain circuits are then more excitable—producing withdrawal symptoms. Flumazenil makes the washout occur even faster than quitting cold turkey. It explains why flumazenil can provoke panic attacks and seizures in people who had been taking benzodiazepines chronically.

  In her 1979 book I’m Dancing as Fast as I Can, the television producer Barbara Gordon explained what happened when she tried to quit Valium. Gordon had started taking the drug for back pain and became dependent on it to stave off anxiety. When she stopped abruptly, she experienced withdrawal symptoms, which she described vividly: “My scalp started to burn as if I had hot coals under my hair. Then I began to experience funny little twitches, spasms, a jerk of a leg, a flying arm, tiny tremors that soon turned into convulsions. I held onto the bed, trying to relax. It was impossible.”28

  In 2007, Anna had been living in an abnormally sleepy state for years. If there was a benzodiazepine-like substance weighing down her brain, what would happen if it was suddenly stripped away?

  CHAPTER 3

  THE ANTIDOTE

  To justify the use of an off-label treatment, there is one and only one person to bear in mind: the patient. But disposing of the other interests in the delivery of medicine, for example the pharmaceutical company that makes the product, the doctor who prescribes it and the government or insurance company who pays for it, is not an easy task.

  —David Cavalla, Off-Label Prescribing, 2015

  Momentum around trying flumazenil with Anna was building. Prompted by Rye, Jenkins and Garcia looked into the medical literature on the danger it might pose. Garcia read a compilation of clinical data on seizures that had occurred in connection with flumazenil.1 He concluded the risk was minimal, since Anna wouldn’t have other drugs in her system. Jenkins wasn’t so sure. “A range of things could have happened,” he said later. “Anything from feeling crummy all the way to grand mal seizure.”

  Anna’s previous experiences had prepared her to take a risk, and she was fully aware of potential complications. As an attorney, she had dealt with cases when hospital procedures had gone wrong. One of the cases she had worked on as a junior associate in 2005, before her sleepiness became such an obstacle, involved defending an obstetrician from a lawsuit by a woman who had a gruesome miscarriage.2 She was still willing to go ahead, telling her doctors: “I am the most informed consent patient you are going to get.”

  The setting for the experiment—the epilepsy monitoring unit in Emory University Hospital—was on the third floor, away from busy Clifton Road. Each bland-looking room had a camera and an extra console for looking at EEG recordings. If a seizure did occur, Anna was already in the hospital, rather than at the sleep lab, which was more than a mile away. Parker was attentive to Anna, holding her hand at the start, but inside, she was “a nervous wreck.” Her maverick proposal was finally going to be carried out. “I think we didn’t fully grasp the gravity of what she was doing, until we saw all those specialists standing there,” Anna’s father, Jim, said.

  The flumazenil experiment attracted spectators, some of whom didn’t strictly need to be there for medical reasons. Jenkins, for example, wanted to see the human side of his experiment. Anna’s parents recalled him as charming. Photos her family took (figure 3.1) show Anna’s colorful pajamas, offset by EEG wires attached to her head, along with stark white walls and a box of crackers to snack on. Epilepsy patients and their families often hang around for days, waiting for a seizure so that doctors can determine what part of the brain it comes from.

  FIGURE 3.1. Anna at Emory, wearing her Spinal Tap shirt.

  Source: Courtesy of Anna Sumner Pieschel.

  Anna occupied herself with the psychomotor vigilance test, a way of tracking her alertness by timing how fast she could react to racing numbers on a screen. When researchers test the effects of sleep deprivation on reaction time and extrapolate to long-distance truckers or airline pilots, they use the psychomotor vigilance test.3

  It’s designed to be very simple, almost mind-numbing: one hundred challenges, randomly spaced, over ten minutes. It’s also difficult (but not impossible) to fake by consistently hitting a button a fraction of a second slower.4 Anna called the test “the world’s most boring video game,” but it was important for the team’s ability to validate what they were doing. Rye and Parker intended it as an objective measure of Anna’s alertness, in contrast to the questionnaires she filled out.

  Over two days in June 2007, flumazenil was introduced into Anna’s system intravenously. The doctors monitored her vital signs and EEGs. The effects were not immediately dramatic, because she started off with low doses. The anticipation might have contributed to a placebo effect, since Anna knew she was getting a real drug—although many other drugs previously had little effect.

  When the dose of flumazenil reached two milligrams, Anna exclaimed to her parents: “I feel alive!” She started talking rapidly and sent an email to friends and family. She said later: “The best way to describe it is that my eyes opened, after being half-closed for so long. It was as if a force grabbed my eyelids and pulled them upwards.” Subjectively, her eyes did seem a little brighter. And her reaction time, as measured by the racing numbers, was a fraction of a second faster. According to Parker, Jenkins broke out in tears, saying: “I sit all day behind a bench with rats and finally I get to see my work make a difference for someone.”5

  As exciting as it was, the effect wore off after a few hours. Several sleepy months would go by before Anna would be able to access the same relief.

  AN EXTENDED EXPERIMENT

  Parker calculated that Anna would need one or two milligrams of flumazenil every hour: several times the amount that reverses midazolam sedation in a healthy person.6 Flumazenil is usually given intravenously because enzymes in the stomach and liver break it down if swallowed. To deliver flumazenil as it comes from the manufacturer into her body, Anna would have to wear a pump infusing the drug directly into her body. That seemed cumbersome and would carry a risk of infection.

  Parker wanted to formulate the drug in a different way—as a nasal spray or delivered under the tongue. For access to a large quantity of flumazenil, she would need the cooperation of Hoffmann-La Roche. Parker began trying to reach executives there in July and August, to her initial frustration. “They would not put me through,” she said.

  Anna’s personal connections opened doors. Through a friend from school, she reached out to George Abercrombie, Roche’s chief executive officer, who had been a pharmacist in a small town in North Carolina before embarking on his management career.7 Abercrombie surprised Parker by calling her and referring her to an employee at Roche’s New Jersey headquarters named Bob Baker, whose official title was professional product information director. Baker’s job was to explain the fine print to health care professionals who needed information about Roche’s products. “I like odd questions, so I decided to take this one on myself,” Baker recalled.

  After talking with Parker, Baker wasn’t quite sure what to do. Using flumazenil to wake up people who weren’t taking benzodiazepines seemed a bit fishy. He also wondered how Anna’s case compared to other cases of sleep disorders. “What convinced me was talking with Parker and Rye about how thoroughly they had evaluated Anna and all the tests they had done,” Baker said. “I went back to my boss and said: ‘We should do something to help this young lady.’ ”

  Two main challenges were apparent. The first was physical. Flumazenil was scarce. The drug was not used that much and hadn’t been manufactured since around the time the patent expired in 2003.8 It was available as a generic, but not at the scale Parker was asking for. What she had requested was a substantial fraction of the existing world supply of flumazenil. Baker had to beat the bushes to locate some in Roche’s laboratories in Basel, Switzerland.

  The second challenge was regulatory. The initial June experiment, when Anna was given flumazenil intravenously, was considered a discretionary “off label” decision by her doctors. Although flumazenil was not approved for sleep disorders, as a physician, Rye had the professional capacity to decide to try it. But if Roche was going to get involved, the company would have to seek guidance from the Food and Drug Administration.

  As defined by the FDA, “compassionate use” is a pathway for a patient with a serious or life-threatening condition to try an experimental product outside of the context of a clinical trial, when there are no comparable or satisfactory therapies available. This process, sometimes called “expanded access,” was formalized in the 1980s in response to people with HIV infections pushing for access to experimental drugs. Through expanded access programs, Roche had provided HIV/AIDS medications to thousands of people in the 1980s and 1990s.9

  The request for flumazenil was different: there was no active clinical trial, and the company was opening its vaults just for Anna. The drug was already approved as a benzodiazepine antidote, but for Anna, the dose, the timeframe, and the indication—chronic treatment for a sleep disorder—were all different from flumazenil’s approved use. They were distinct enough that the FDA advised that the Emory team would need to make an “expanded access IND” (Investigational New Drug) application. This type of request is typically submitted by the patient’s physician and needs to be approved by the drug manufacturer as well as the FDA and an IRB (Institutional Review Board).10