SNOOZE CRUISERS

  The Atlanta hypersomnia support group began meeting around the same time as the Living with Hypersomnia conference. Sometimes the group assembled at a casual restaurant. The group later met on Saturdays at Emory Sleep Center, which moved to a modern building with a spacious lobby in 2015. “We realized that we had to do it every month, whether we knew people were going to show up or not,” Diana said. Atlanta support group members have exchanged tips about doctors, how to deal with insurance companies, and specific medications. Diana has also organized week-long support retreats on cruise ships—known as Snooze Cruises.

  Many in the Atlanta group have tried various dietary supplements and low-carbohydrate or ketogenic diets. They’ve talked about their struggles to have friends and relatives see what they have as real. One woman discussed whether she should disclose her diagnosis to her supervisor, since she was managing well enough with a remote work schedule. Another said she had trouble relating to those she met at a Narcolepsy Network conference because everyone seemed to have a job.

  Members have discussed “spoon theory,” a term used in several chronic illness communities when fatigue limits daily activities.28 Spoons refer to units of motivation and concentration that are used up as someone makes it through the day. At one meeting, a supporter of an IHer explained that he envisioned spoons like magical energy in a fantasy role-playing game, consumed upon casting spells.

  Notes of levity appear when members discuss alarm clocks that flash and shake or make someone perform math calculations before they shut off. At another meeting, members couldn’t stop laughing at a phone-based alarm with an opera singer proclaiming “You must get up! You must get up!”

  “Hah, I could sleep through that!” someone said.

  CHAPTER 9

  THE STORY OF FLUMAZENIL

  Although our prime intention was to create a powerful scientific tool for future studies of the function of the BZR [benzodiazepine receptor], we did quite clearly foresee a number of therapeutic applications for a BZ antagonist. The enthusiasm of the marketing department was modest at the time.

  —Walter Hunkeler and Willy Haefely, 1988

  David Rye was not the first to test flumazenil in the context of sleep disorders. Working decades before its repurposing for hypersomnia, the scientists who first synthesized flumazenil cited it as “a classic example of preclinical serendipitous drug discovery.”1 While searching for an improved version of Valium, they found its opposite along the way.

  In the form of a little yellow pill, Valium inspired Mick Jagger’s barbs in “Mother’s Little Helper” and made Roche the biggest drug company in the world in the 1960s and 1970s.2 Riding a wave of profit, the Swiss company invested in research facilities in Europe and the United States. But since Valium-related patents were anticipated to expire, Roche was headed for trouble, and competitors were already encroaching.3 Users’ difficulties with dependence and withdrawal made Valium a target for policy makers. Senator Edward Kennedy of Massachusetts held hearings in 1979 denouncing the drug. “Millions of dollars are spent each year to convince physicians to use tranquilizers for a wide variety of things—some legitimate, and some, like the stress of everyday life, not,” Kennedy said then.4 “Our message to the American people is clear. If you require a daily dose of Valium to get through each day, you are hooked and you should seek help.”

  The Roche of the twenty-first century became more conservative and more focused on cancer, its largest source of revenue. An official history criticized the company’s scattered approach in the 1970s: “The underlying hope that someone would somehow stumble on another money-spinner like the benzodiazepines appears to have distracted people’s attention from the key issues.”5

  GRIND AND BIND

  Willy Haefely, the leader of Hoffmann-La Roche’s central nervous system research unit, is not as celebrated as Leo Sternbach, the chemist who first synthesized Valium and other benzodiazepines. As a colleague wrote: “He had no desire to place himself in the limelight; he did not covet titles or seek power. For all his outward robustness he was a sensitive man, easily hurt.”6

  Haefely (figure 9.1) was known for his fierce intelligence, and he deserves some credit for correctly deducing that benzodiazepines exert their effects through GABA, years before the mechanism became clear.7 Today, we have amassed a huge amount of information about neurotransmitters, drugs, and receptors. Back then, these relationships were still being teased out. More pharmaceutical companies like Roche were conducting basic neuroscience research. Scientists didn’t know much about the complex variety of receptors they were dealing with, and GABA receptors’ diversity did not unfold until later.

  A standard “grind and bind” experiment from the period went like this. Put rat brains in a blender, breaking apart the cells, and centrifuge the mixture to obtain membranes. Add something “hot”: a radioactive drug. Then put the membranes on a filter and slurp off the liquid with a vacuum. A scientist can test how tightly the drug is sticking to its receptors by checking if a surplus of the “cold” drug or a similar compound can make the radioactivity wash away.

  In 1977, grind and bind was how Hanns Möhler, a pharmacologist working for Haefely in Basel, identified what were called “benzodiazepine receptors,” which are now called GABA-A receptors. In this case, the radioactive probe was Valium, whose generic name is diazepam. Möhler could see that diazepam and its relatives stuck to something on cell membranes all over the brain.8

  A year later, Möhler was using the grind-and-bind technique to screen a host of new compounds the Roche chemist Walter Hunkeler had been synthesizing. Their overall goal was to find new benzodiazepine drugs that would have antianxiety and anticonvulsant properties without being sedating, or vice versa. One of the compounds, a parent of flumazenil with the code number Ro 14-5974, caught the attention of Möhler, Haefely, and their colleagues. This compound was capable of competing with diazepam for binding, but by itself, it had none of diazepam’s classic effects in live animals, such as sedation or stopping seizures. “We checked if it entered the brain, and it clearly did,” Möhler said in a phone interview. “But it was inactive in all the assays we had at the time. So the idea came up: maybe it’s an antagonist, not an agonist, like the benzodiazepines.”

  FIGURE 9.1. Willy Haefely, circa 1980.

  Source: Roche Historical Archive.

  Remember what benzodiazepines do in molecular terms: they make it easier for GABA receptors to open their gates. Haefely and Möhler thought it was possible for a molecule to occupy the same slot that diazepam did but just sit there without facilitating anything: an antagonist. They had searched for benzodiazepine antagonists in the early 1970s without success. Ro 14-5974 still had some weak activity on animals’ spinal cords, but Haefely and Möhler knew they were close to having something “with as little agonistic (intrinsic) activity as possible.” Grind and bind made it possible to look for antagonists more systematically.

  In a “crash program,” Hunkeler synthesized many variations of molecules resembling the parent, and the group tested them all over again. One with the code number Ro 15-1788, later named flumazenil, came out on top. In an initial test, Möhler had a technician give enough diazepam to a mouse to make it fall asleep, then inject the proposed antagonist. The mouse promptly woke up and began running around, as if the diazepam in its body had been wiped away.

  In 1981 in Nature, the Roche team reported that in experiments with several types of animals, flumazenil had none of the benzodiazepines’ characteristic effects.9 At the same time, it did not sensitize animals to seizures or act as a stimulant. These results agreed with the Roche researchers’ emerging view of the drug as biochemically inert.10

  When it came time for human experimentation, a physician named Walter Ziegler insisted on testing flumazenil on himself first. On the morning of April 3, 1981, at Zürich’s university hospital, Ziegler was first put to sleep with the fast-acting benzodiazepine flunitrazepam (Rohypnol, now banned in the United States). Before the entire intravenous dose of flumazenil entered his body, he turned his head and opened his eyes. Ziegler was quoted in Roche’s in-house magazine as saying that he felt “as if I have been torn from sleep by an alarm clock, and want to get up.”11 He was still dizzy, but within another minute, he stood up, went to lunch as usual, and then got back to work.

  The first published reports on reversing benzodiazepine-induced sedation in humans with flumazenil came from a Roche-affiliated clinical research facility in Ireland.12 These studies were mainly aimed at demonstrating flumazenil was safe enough for one-time use. In less than a minute, intravenous flumazenil could rouse heavily drugged volunteers, all healthy young men, who otherwise would not respond to vigorous shaking, having their names called, or ten seconds of alarm bells. Separately, flumazenil—in large doses, from 200 up to 600 milligrams by mouth—did not affect volunteers’ performance on a series of benzodiazepine-sensitive tests, such as reaction time and the ability to quickly copy numbers on paper.13

  THE BRAIN’S OWN ANTIANXIETY SUBSTANCE

  In the laboratory, flumazenil by itself sometimes displayed weak effects. Since the drug was thought to be inert neurochemically, this was explained by the proposed existence of endogenous benzodiazepines, or “endozepines.” Speculation about endogenous ligands—something that binds to a receptor, presumably fulfilling a physiological role—had begun even before flumazenil was identified.14 Opiate receptors, the molecules that morphine and heroin act upon in the brain, were a hot topic at the time. When opiate receptors were identified in 1973, Newsweek quoted Solomon Snyder—one of the flashier neuroscientists of his time—as predicting prophetically: “We can assume that nature did not put opiate receptors in the brain solely to interact with narcotics.”15

  A similar set of assumptions lay behind research on benzodiazepine receptors. Scientists rationalized that something in the body must take advantage of the sites accessible to benzodiazepines because their receptors were present in all vertebrates, and evolutionary pressure kept them there.16 A commentary in Nature described the unknown entity as “the brain’s own anti-anxiety substance.” At Roche, Willy Haefely was skeptical of arguments based on divining nature’s intent but said that the possibility that an endogenous ligand bound at the same sites as benzodiazepines “cannot be dismissed and has to be examined seriously.”17

  In the 1970s, major pharmaceutical companies such as Eli Lilly and Sandoz were developing enkephalins, endogenous pain relievers that interact with opiate receptors, as potential drugs. The hope was that these molecules might be nonaddictive, or less harmful compared with opiates, because they were naturally found in the body. However, because they were peptides, they didn’t last long in the body and did not become commercial drugs.18

  NOT TOTALLY INACTIVE

  The prevailing view of flumazenil as an inert antagonist began to change as others began trying it out. While not sedating like diazepam, the drug had antiepilepsy properties, which Roche and allied investigators puzzled over.19 In healthy people, the drug could elicit “increased discontent, increased headache and sweating” and did have the potential to make people dizzy or uncomfortable in high doses.20

  Jean-Michel Gaillard, a psychiatrist in Geneva, was one of the first to test flumazenil’s effects on sleep. Beginning in 1981, Gaillard compared flumazenil’s effects on how long it takes someone to fall asleep with that of caffeine. With a small number of volunteers, he found its activity profile was similar to caffeine’s, although weaker. Gaillard wrote: “To our surprise, flumazenil alone was not totally inactive, but exhibited some alerting effect. When we presented these results to Haefely, he was somewhat skeptical, because there was not the slightest evidence for such an effect in animal experiments.… We continued to believe that our results were not due to chance and this little controversy became a joke between Haefely and us.”21

  Other investigators gave flumazenil to sleep-deprived people, reasoning that the stimulant effect Gaillard observed would then be easier to detect. Peretz Lavie, a sleep researcher from Israel, had volunteers stay up all night and take flumazenil in the morning.22 Large doses (60 or 120 milligrams by mouth, every four hours) helped them resist slumber. Here, the idea of endozepines resurfaced. Since flumazenil was supposed to be inactive, Lavie interpreted his results as revealing that the drug was displacing a benzodiazepine-like substance that had accumulated during his charges’ sleepless nights.

  Other studies from Germany and Switzerland support Lavie’s finding that flumazenil can dispel drowsiness resulting from lack of sleep, although ready comparisons between these studies are challenging.23 Alertness tests were conducted in the morning or the evening; sometimes the drug was given orally, sometimes intravenously, and doses vary. One study from the United Kingdom found that flumazenil could trigger panic attacks, but this effect was limited to people who had been taking benzodiazepines for panic attacks to begin with.24 Generally, in healthy study participants without a sleep disorder diagnosis, the effects of flumazenil were mild.

  Research on flumazenil’s effects in animals was more provocative. One of the better-publicized examples comes from Thaddeus Marczynski at the University of Illinois. Marczynski, funded by the Air Force, gave flumazenil to older rats for months. Even after the drug was removed from the rats’ drinking water, on memory tests, the older rats performed as well as control animals that were a year younger.25

  Marczynski was so enthusiastic about flumazenil that he had to promise Roche executives not to try it himself.26 He told Newsday: “I rejuvenated the rats. Exploratory behavior is extraordinarily enhanced. They want to know everything.” Science noted: “Though Marczynski’s results are provocative, that’s nothing new in a field that has seen many leads fail to pan out.”27

  By the time Rye, Parker, and Jenkins became interested in flumazenil, the stream of publications on the cognitive or sleep-related effects of the drug had dwindled. For his part, Rye relied mainly on a study from researchers from the United Kingdom,28 which concluded that “the beneficial effects of flumazenil on cognitive performance appear limited to the reversal of benzodiazepine-induced impairment (in humans at least).”

  “All those early experiments with flumazenil were interesting, but you have to look at the effect sizes,” Rye said. “They weren’t that big. Not like what we saw with Anna.”

  A ROMANTICIZED QUEST

  A full account of the search for endozepines, which some scientists have called a “romanticized quest,” would take up more space than available here.29 The quest was not resolved to the satisfaction of the person who had the largest role in driving it: Erminio Costa. A formidable neuroscientist, Costa ran a National Institute of Mental Health lab located at St. Elizabeth’s Hospital, across town from the intramural campus in Bethesda. He and Haefely were both friends and allies in debates over benzodiazepines’ mechanism of action, since they had both inferred the benzodiazepine-GABA connection before the receptors were fully identified.

  Costa, originally from Sardinia, was a lover of opera, outgoing and imaginative. He was known for asking tough questions in public, but he was devoted to supporting the careers of his trainees from all over the world.30 Colleagues described him as having a fiery temper and as “a truly brilliant man who simply did not tolerate fools easily.”31 “He was a very excitable man,” said Alessandro Guidotti, Costa’s partner in running his lab for many years. “He believed in what he was doing and would strongly defend his ideas.”

  Starting in the late 1970s, Guidotti and Costa were looking for proteins that would interfere with diazepam at its receptor sites in the brain. They identified a protein called “diazepam binding inhibitor,” or DBI, which fit criteria for an endogenous benzodiazepine receptor ligand. DBI’s purification began with the extraction of rat brains with steaming acetic acid—a pungent, old-school technique that left other proteins behind in a scrambled mass. But when DBI was injected into the brains of animals, it seemed to heighten anxiety rather than dampen it, a result that Costa rationalized. “It makes sense that the endogenous system in which the benzodiazepines operate is there to create anxiety, not to limit anxiety,” he told Science News.32

  Researchers looked for DBI in the cerebrospinal fluid of people with various mental illnesses, finding that its levels were elevated in depression. However, the more DBI was probed, the murkier its role became. Smaller fragments of DBI sometimes had opposite effects from the full peptide. And DBI appeared to have another identity; it was the same as another protein whose function was escorting fatty acids inside cells, instead of modulating GABA signals outside.

  Doubts accumulated. Over time, Costa changed his ideas on DBI’s function, postulating that its effects on GABA receptors might be indirect. In a 1994 review, he wrote that “the term endozepine should not be used to refer to DBI, because it is now reserved for endogenous ligands of benzodiazepine recognition sites which do not include peptidic bonds in their chemical structure.”33

  Costa’s competitors at the National Institutes of Health’s Bethesda campus identified several other candidates for “the body’s own Valium,” as the hypothetical benzodiazepine-like compounds were called. The major candidates to emerge were derivatives of progesterone and other steroid hormones, known as neuroactive steroids or neurosteroids.34

  In 1986, a team led by Steven Paul at NIH discovered that these hormones signal through GABA-A receptors. Progesterone was known to have anesthetic properties since the 1940s, and Paul’s research explained why. Paul went on to lead Eli Lilly’s research in the 1990s and later cofounded a company, Sage Therapeutics, focused on developing related compounds to treat disorders such as depression. In an interview, Paul said: “After all these years, except for GABA itself, the only truly physiological ligands for GABA-A receptors appear to be neuroactive steroids.”