Neurosteroids could also be candidates for the hypersomnia somnogen. At first glance, they don’t fit the detectives’ clues from chapter 6. They are not peptides, so peptide-chewing enzymes should have kept their activity intact—and they don’t bind at the same sites on GABA-A receptors as benzodiazepines. In her thesis research, Jenkins’s graduate student Olivia Moody did find that hypersomnia CSF samples, like neurosteroids, can modulate the activity of GABA-A receptors that are benzodiazepine insensitive.69 Rye and Jenkins said they had tested hypersomnia CSF samples for neurosteroids, but that her findings made them want to take a second look.

  For a time, Rye was having hypersomnia patients perform a visual test called critical flicker fusion, which has been used to identify people with covert hepatic encephalopathy.70 Benzodiazepine-like compounds synthesized by intestinal bacteria might account for some of the wake-promoting effects of the antibiotic clarithromycin. In future research, potential contributions of both neurosteroids and the intestinal microbiome to idiopathic hypersomnia should be examined more thoroughly.

  CHAPTER 10

  WEIRD DRUGS

  Any consideration of the development of psychopharmacology makes it quite clear that good marketing of such ideas can capture a field, either before the evidence is in on an issue or even in the face of considerable contradictory evidence. Ideas, such as the dopamine hypothesis of schizophrenia or the amine hypotheses of depression, have functioned very much as brand names.

  —David Healy, The Antidepressant Era, 1997

  After the 2014 Living with Hypersomnia conference, identifying the somnogen became a rallying cry for fundraising. A development officer at Emory’s medical school cooperated with a donor from Colorado to create a hypersomniaresearch.org webpage. The page, promoted on Facebook groups such as Idiopathic Hypersomnia and Major Somnolence Disorder, included photos of Jenkins’s patch clamp rigs. Rye was quoted as saying: “I believe that discovering the mystery somnogen holds the key to curing hypersomnia once and for all.”1

  Online enthusiasm was strong. Someone posted a note on one of the Facebook groups saying, “I TESTED POSITIVE FOR SLEEPY JUICE!!! Never have I been so excited for a positive test result,” and received forty-six comments in response. The fundraising page set out the goal of gathering about $70,000 for hypersomnia research. In a few months, that goal was fulfilled, with more than sixty donations contributing.

  At that point, Rye’s lab was running low on money. He called in a favor from Garcia, who had established his own lab with funding from the Department of Veterans Affairs. Garcia temporarily paid the salary of one of Rye’s researchers. During this period, Rye was able to manage with additional support from a university neuroscience initiative and from private donors, such as Anna’s father. In mid-2015, Anna lent her support by asking her wedding guests to give money to Emory, directed to Rye’s lab. Her wedding to fellow Atlanta lawyer Nick Pieschel demonstrated her recovery and her confidence in the future.

  With a high-profile publication in hand, Rye had applied to the National Institute of Neurological Disorders and Stroke (NINDS) for a large grant in early 2014. The application was well received, and after revisions, the grant was to start in August 2015. It resolved the funding problem, providing almost $400,000 per year in direct costs.2 Jenkins estimated that up to that point, he and Rye together had submitted at least twenty unsuccessful applications since 2009.

  With all the new funding, the sleepy juice idea was supposed to be tested in a more comprehensive way at Emory. Jenkins or one of his trainees would compare GABA-enhancing activity in CSF from patients with “primary hypersomnia” versus people with sleep apnea as well as nonsleepy controls. The elusive somnogen might be identified, with the help of mass spectrometry and proteomics, which had advanced enough to be able to identify minute amounts of substances in CSF. The hypersomnia team began working with biochemists at an Emory facility meant to support Alzheimer’s disease research.

  The scarcity of CSF, the abundance of other proteins in CSF, and the slow, low-capacity nature of patch clamping were all still obstacles. But according to preliminary data in the NINDS grant application, patient CSF samples induced sleep when directly injected into rats’ brains. Potential somnogen hits were emerging from the proteomics collaboration. Another aspect of hypersomnia would be investigated by Trotti, who had obtained a smaller grant from NINDS for a brain imaging study.3 “If our hypotheses prove correct, they would call for a sea change in scientific thought and clinical practice that has considered the brain’s wake promoting regions as the principal arbiters of pathological hypersomnolence,” Rye wrote.

  RETROSPECTIVE ANALYSIS

  In an August 2014 thank-you letter to donors, Rye noted that close to one hundred sleep disorders patients had received flumazenil in its unconventional form. He and colleagues at Emory’s sleep center were evaluating several new patients every week. In addition, doctors outside Emory were beginning to prescribe it.

  At the beginning of 2015, Trotti and Rye reviewed records of 153 patients with treatment-refractory hypersomnolence, to whom they had prescribed flumazenil. People who met the MSLT requirements for IH or narcolepsy type 2 comprised less than half of the group—a larger percentage either had sleep apnea or took longer to fall asleep than eight minutes on an MSLT. On average, the entire group reported habitual sleep of seventy hours per week and had tried more than four other medications already.

  This was not a randomized controlled clinical trial. Rather, it was a retrospective analysis, with patients choosing the dose, mode of delivery (lozenges or skin cream or both), and whether to stop. According to this chart review, flumazenil helped about 60 percent of the group feel more awake.4 A smaller number (39 percent) stuck with the drug long term. A cost of hundreds of dollars per month, with few people getting reimbursed through insurance, was another reason why some people stopped. The most common side effects were dizziness and anxiety. There were a few more serious adverse events, such as a transient ischemic attack in a patient with a history of atrial fibrillation, but determining if flumazenil was the cause was not possible.

  Flumazenil responders reported a large average drop, from 15 to 10, on the subjective Epworth Sleepiness Scale. Whether that corresponded to fewer hours of sleep or other measures of alertness during the day, the Emory authors could not say. In this group, testing positive for CSF “sleepy stuff” was a prerequisite. Flumazenil responders didn’t have more GABA-enhancing activity than nonresponders, but one predictor of flumazenil response was the symptom of sleep inertia.

  The Hypersomnia Foundation’s website acknowledged where flumazenil could be obtained and what it was prescribed for. Rye said he occasionally asked pharmacists to curb efforts to signal its availability online, because of concern about “off-label promotion” (discussed in chapters 14 and 15). There was no shortage of people waiting to see Rye and Trotti. But the painstaking patch clamp test had become a bottleneck for the process of diagnosis. Some people waited months for their results or never received them at all. Jenkins said: “They must be horribly frustrated with us. But this isn’t a diagnostic lab, it’s a research lab.”

  When Trotti presented the results of the retrospective analysis at a 2016 sleep research conference, she noted that she and Rye had gained enough confidence in flumazenil to stop requiring lumbar puncture and GABA activity measurements before a prescription. “Prospective controlled studies of flumazenil for treatment of hypersomnolence are certainly needed, but in the absence of those data there’s at least a rationale for people who are severely affected and have nowhere else to turn to consider flumazenil,” she said at the conference.5

  The overall conclusion was that flumazenil was another option for people with hypersomnolence who had found other drugs unsatisfactory. It wasn’t clearly better than drugs that acted by other mechanisms. Many people reported contrasting results with lozenges versus skin cream. It was difficult to determine whether someone didn’t respond to flumazenil because it didn’t work or because their body wasn’t really getting enough of the drug.

  Flumazenil could perhaps be formulated into a skin patch, similarly to methylphenidate or the Parkinson’s medication rotigotine. But for something like that, a pharmaceutical company would need to invest money in developing a delivery system. As it was, only a few people with hypersomnia were able to try a continuous infusion, Australian style. Diana Kimmel eventually did. “It felt amazing—but it was awkward to move around,” she said.

  Despite frustrations with flumazenil, enough people in this niche community depended on it that supply problems caused disruption. In mid-2015, the Atlanta compounding pharmacy Rye had engaged was temporarily unable to fill new or renewing prescriptions. The pharmacy had been receiving flumazenil from a company in Canada that imported it from China, and changes in pharmaceutical import regulations, along with purity issues, caused a delay. Members of hypersomnia interest groups posted frequent anxiety-tinged updates on Facebook until the shortage was resolved.

  SUCCESS STORIES

  In the summer of 2015, Romy, a woman from the French-speaking part of Switzerland, came to see Rye. She was his first hypersomnia patient from Europe. It was an indicator of how far word had spread, since she had learned about his research through a Facebook group.

  Romy recalled that when she was in school, she slept for twelve hours per day. She would ask her teachers for permission to go to the bathroom—and used the opportunity to nap. She found a job in fashion design but had to stop working because sleepiness was interfering too much.

  She saw several physicians, who mainly assumed that she was depressed. She was told: “Everybody is tired—you are just lazy.” Blood tests for metabolic disorders were negative. In early 2015, she visited the university hospital in Lausanne, where she finally got a confirmation that her long sleep times were “neurological, not psychological.” At a sleep lab, she dozed overnight for fourteen hours straight. Standard medications were unsatisfactory or ineffective. She slept for twenty hours per day after becoming tolerant to methylphenidate. “It feels like this is the last chance,” she said when she came to Atlanta. “After seeing several sleep neurologists in Switzerland and trying all the different treatments available, I decided to cross the ocean to meet this professor who everyone was talking about on social media groups.”

  Contacted after she returned home, Romy said that the flumazenil skin cream made her feel much better. The first week was “magical,” although the effect gradually decreased, and she perceived little benefit from flumazenil lozenges. Together with flumazenil, Adderall gave Romy a window of time when she felt safe enough to drive, but in the evening, she would have a headache and stomach pain. Over the next few years, Romy continued to visit Atlanta to see Rye and update her flumazenil prescription, sometimes staying with a member of the Atlanta hypersomnia support group. “It’s still hard to do things, but it [flumazenil] did help,” she said in 2018. “But not enough to feel like everybody else.”

  That same year, Anna’s brother James credited a higher dose of flumazenil, combined with conventional medications, with making it possible for him to get his dream job: animator at Industrial Light and Magic in San Francisco. In competitive fields like video game design and animation, James needed to be able to work hard at crunch time. After years of feeling stagnant, he was more productive.

  There were other celebrated success stories, such as Sigurjon, a young man from Iceland who came to see Rye in 2018. Formerly active and athletic, Sigurjon dropped out of college and lost touch with friends because of his ravenous need for sleep. His hypersomnolence seemed to vary with the strong seasonal light variations in Iceland; he slept almost ten days straight in the winter of 2017, and his MSLT results varied with the seasons as well. Standard medications such as modafinil and methylphendiate could keep him physically awake, but he still felt worn out inside.

  Rye had Sigurjon try flumazenil for the first time on camera, as part of a Georgia Public Broadcasting television feature on hypersomnia, with both his family and Anna watching. Sigurjon’s response sounded like Anna’s years ago: “It’s like my eyes are being lifted up … it feels really good. I don’t even remember feeling like this.” With flumazenil skin cream, he was able to wake up spontaneously for the first time in years, without the aid of an alarm clock. Rye visited Sigurjon in Iceland afterward and reported that he was working at his father’s construction company and was still taking flumazenil.

  THE ZOMBIE APOCALYPSE

  At the same time, plenty of people with IH have tried flumazenil without finding it a satisfactory solution. One of them was Meghan Mallare, who exemplified several elements of the IH experience. Meghan was referred to Rye in 2014, after having difficulty managing her sleep schedule at a university in North Carolina. She had started feeling excessive sleepiness in high school, falling asleep in class. Diagnosed with inattentive ADHD, she started off with the conventional stimulant methylphenidate, which her mother would wake her up to deliver at four in the morning. “Stimulants got me through high school, and then by college, I could take a higher dose, and I could just sleep through it,” Meghan said. “Sometimes they kind of work, but not really.”

  The greediness of her sleep schedule followed her to university, where Meghan set up special alarm clocks that would shake her bed in the morning. Her roommates called it the “zombie apocalypse” and called campus security a few times because they were unable to wake her. She went to see a psychiatrist through student health services, but the dominant assumption was “you’re a teenager, you’re depressed,” she said. The psychiatrist was willing to consider narcolepsy after she described a possible hypnogogic hallucination; during class, she saw herself at a desk floating in the middle of the ocean. The psychiatrist contacted Stanford, and somebody there recommended Rye.

  In a thorough examination, Rye measured her levels of iron, vitamin B12, and thyroid hormones. On an MSLT, she fell asleep in an average of 2.8 minutes and only entered REM once—clearly meeting ICSD criteria for idiopathic hypersomnia. Afterward, she returned to university with a letter from Rye to show her instructors, who laughed at the letter. She did not receive any disability accommodations, and Meghan eventually withdrew from university after sophomore year, feeling defeated. “They did not believe it [IH] was a real thing,” she said. “They thought it was a code name for partying too much.”

  Through Rye, Meghan was able to try both of the GABA antagonist medications flumazenil and clarithromycin, with mixed results. “Flumazenil made me feel dizzy and almost drunk and didn’t help with the tiredness,” she said. In contrast, clarithromycin was “amazing” and provided a definite improvement in sleepiness, but the side effects were too strong.

  When Meghan came to Emory as part of a grand rounds presentation, she said she doesn’t drive long distances and only drives short distances in the afternoons. By necessity, she worked part-time at a doctor’s office, only in the afternoons, since she was rarely able to wake up before noon. Her dog was trained to wake her up, and friends had a key to get in, just in case.

  Meghan’s case was noteworthy because she exhibited all the aspects of the classic long-sleep form of IH, including severe sleep inertia. She recalled being clumsy in the mornings and “confused and agitated” when others try to wake her up. Her later MSLT results also demonstrated the test’s limitations. A year after her initial diagnosis, she retook the test and entered REM in all five naps, but falling asleep slower: 7.9 minutes. In 2016, her average sleep latency was 11.4 minutes, which would be considered normal—but during this period, the amount of time she spent sleeping increased from ninety up to more than one hundred hours per week.

  CHALLENGE FROM FRANCE

  In 2016, the French sleep neurologist Yves Dauvilliers and colleagues published an Annals of Neurology paper saying they had not been able to replicate the Emory results.6 They observed no differences in GABA-enhancing activity between CSF samples from people with IH, narcolepsy, and controls. They argued that the findings did not support prescribing GABA antagonists such as flumazenil and clarithromycin to IH patients.

  Dauvilliers was the respected leader of a sleep disorders center in Montpellier and lead author of several studies on narcolepsy medications. A few years before, his group had brought down another theory about hypersomnia, involving levels of the neurotransmitter histamine in CSF.7 He was a foe of what he called “circular reasoning,” or making assumptions that ensure the result one is supposedly testing.

  Several factors could have explained discrepancies between the French and American research. The Montpellier group was using eggs from Xenopus toads as vehicles for GABA-A receptors, rather than the human kidney cells used by Jenkins’s lab. They were delivering puffs of CSF onto the toad eggs rather than bathing cells by continuous flow. Also, the patient populations were different; the Montpellier authors emphasized recruiting long sleepers. The majority had slept for more than sixteen hours total in an extended overnight protocol. In comparison, Emory patients were less uniform in their diagnoses.

  When the French paper was published, Jenkins was annoyed, saying that the Montpellier group had made a mistake in calibrating their assays. A ceiling effect imposed by the wrong GABA concentration was obscuring a modulating effect of the CSF, he said. “Basing therapeutic decisions on results of an in vitro assay that is methodologically flawed is a disservice to IH patients,” Jenkins and his colleagues wrote in a rebuttal to Dauvilliers’s Annals of Neurology paper.8