As the dispute proceeded, genetics was beginning to show the differences between the two types of narcolepsy. When Stanford investigators surveyed people with narcolepsy, those with severe cataplexy were almost always DQB1*0602 positive.10 The proportion of people with narcolepsy without cataplexy who had the same HLA risk factor was 40 to 60 percent: higher than the general population but lower than those with cataplexy. These findings bolstered the view of narcolepsy without cataplexy as a separate, possibly nonuniform category.

  A LONG MARCH, AIDED BY DOGS

  Although many sleep researchers had a part, one who was central to defining hypocretin’s relationship to narcolepsy was Emmanuel Mignot, director of the Center for Narcolepsy at Stanford (figure 13.1). Mignot served as chair of Narcolepsy Network’s medical advisory board for many years. David Rye viewed him as a friendly rival; they’ve exchanged patient samples and reviewed each others’ papers. Mignot’s ambition to solve the puzzle of narcolepsy was tempered by empathy for his patients and a willingness to tell self-deprecating jokes.

  FIGURE 13.1. Emmanuel Mignot and his dog Watson, who has narcolepsy.

  Source: Lenny Gonzalez.

  One of Mignot’s signature achievements was identifying the genetic basis for narcolepsy in dogs, after Dement had established a colony of the occasionally floppy animals at Stanford in the 1970s. When he first arrived from France, Mignot spent a great deal of time in close contact with the dogs, playing with them and bottle-feeding puppies if they were rejected by their mothers. “Even after I took a shower, you could still smell the dogs,” he told the magazine The Scientist. “It was a strange part of my life.”11

  In dogs, the inheritance pattern for narcolepsy was simpler than in humans: there appeared to be a mutation in one gene. However, this project played out before the human or dog genomes had been sequenced. There was no link to the canine version of HLA. It took a decade of painstaking work and exploring dead ends to find the mutation.

  The long march through canine chromosomes was nearly finished in the 1990s when the work of Mignot’s team converged with independent research on hypocretin in mice. Masashi Yanagisawa’s laboratory, originally focused on appetite rather than sleep, had generated mice missing the hypocretin gene. For months, it looked like the mice didn’t have altered behavior, because the Texas researchers were monitoring them during the day. At night, when mice are more active, their behavior was different. The mice would be running around burrowing and grooming themselves but would suddenly become immobile.

  Yanagisawa’s group first thought the mice were having seizures, but EEG and muscle activity measurements revealed that what they were observing was a mouse version of cataplexy. The two labs sniffed each other out at a sleep research meeting and published papers in 1999 two weeks apart.12 It turned out that canine narcolepsy came from a mutation in a hypocretin receptor gene.

  While this did not explain human narcolepsy, Mignot and his colleague Seiji Nishino then guessed that it would be possible to see changes in hypocretin levels in cerebrospinal fluid.13 This was a confident leap, given the disappointing track record of using CSF to reveal neurochemical alterations in narcolepsy. However, using an antibody-based laboratory test for the hypocretin peptide, it worked. A bank of postmortem brain tissue established at the University of Michigan enabled other investigators to see directly that a large number of hypocretin-producing neurons (more than 90 percent) were missing in people with narcolepsy and cataplexy.14 While someone was alive, hypocretin CSF measurement was a way to see into the hypothalamus.

  Since then, surveys of hundreds of people with narcolepsy and other sleep disorders found that CSF hypocretin levels are generally low or undetectable in the classic form of narcolepsy with cataplexy.15 Low was defined as less than 30 percent of the average in healthy controls, and some people with traumatic brain injury or encephalitis had in-between levels. Most people with narcolepsy without cataplexy have normal hypocretin CSF levels, an indication that something else may be going on in their brains that does not involve elimination of hypocretin neurons. There are a few exceptions to this general rule. Low CSF hypocretin can precede the appearance of cataplexy by years, and a drop in an already low hypocretin level can follow development of cataplexy.16

  Overall, hypocretin’s discovery resulted in an incomplete redefinition of narcolepsy. Previously, the disorder was defined by disturbances in REM sleep, but with hypocretin, there was something more tangible: a molecule. Experts such as Mignot expressed confidence that their discoveries would lead quickly to new treatments. They also said they thought an autoimmune mechanism for narcolepsy was likely. What remained was to provide evidence of the immune system’s treachery.

  XENOPHOBIC POLICE

  In a sense, autoimmune diseases come in two groups. In the first, the immune system attacks the body but never succeeds in eliminating the source of its irritation. Symptoms come from inflammation and the debris of battle left behind. Examples include systemic lupus erythematosus and rheumatoid arthritis. In the second, such as type 1 diabetes, the immune system “wins” by removing the cells it objects to. Narcolepsy type 1 appeared to be a stealthy example of the second type because of the lack of obvious inflammation in the brain. Hypocretin-producing cells were apparently eliminated while their neighbors in the hypothalamus were left intact.

  With the HLA genetic link in mind, several narcolepsy research groups went looking for autoantibodies, a common feature of both types of autoimmune disorders. Antibodies are blood proteins whose usual function is to grab onto invaders, and autoantibodies are reactive toward components of our own cells. They specify what the immune system is going to mistakenly attack. Despite several high-profile papers on the topic, it is still debated whether autoantibodies play a role in narcolepsy.17

  A 2009 Nature Genetics paper from Mignot’s group pointed in a different direction than antibodies. The study of more than eight hundred people with narcolepsy, all with DQB1*0602 and cataplexy, showed that a variant of one T cell receptor gene was a risk factor. This was a substantial clue but one whose significance did not become clear until later.

  T cells represent the other major arm of the adaptive immune system, complementary to antibody-producing B cells. When both B and T cells develop, each goes through a set of mix-and-match genetic changes called V(D)J recombination. In T cells, the process starts with a large array of T cell receptor genes, rearranging and gluing together just two of them and leaving the rest intact. As a result, each mature T cell is capable of recognizing a different set of protein fragments, displayed within a specific frame provided by HLA proteins. The HLA proteins appear on a partner cell called an antigen-presenting cell, which holds recycled protein fragments on the outer membrane as signals that may excite T cells.

  If pathogens are present, unusual fragments of digested protein appear on the antigen-presenting cell’s surface. This is how “killer” T cells recognize cells that are infected by a virus. It’s like having security police sort through food leftovers in a garbage can outside a house, to figure out whether there were foreign travelers inside. Another group of “helper” T cells looks through extracellular waste. T cells that recognize protein fragments from the body itself are supposed to be edited out during development, but this process is incomplete, and T cells can usually recognize more than one sequence. The capability of launching an autoimmune attack lies within many people, but the checks and balances of the immune system usually prevent such an attack from gaining enough momentum to do any real damage.

  Mignot’s group was also able to show that people whose narcolepsy symptoms had appeared in the last three years were more likely to have antibodies against streptococcal bacteria.18 This made narcolepsy appear like a neurological version of rheumatic fever, a delayed inflammatory complication of strep throat infection that can damage heart valves and joints. An environmental trigger of some kind of infection seemed likely; immune cells can migrate into the nervous system via the upper respiratory tract and olfactory nerves.

  FORCED BY THE FLU

  What removed uncertainty about an autoimmune mechanism was the H1N1 flu outbreak, which began in mid-2009. Public health officials feared that the impact could be similar to the 1918 Spanish flu, killing millions of people around the world. Because of the anticipated impact of H1N1, many countries organized vaccination campaigns, and vaccines were rushed into service. Supply was limited because vaccine production was lower than expected.19 To extend that supply, some vaccines contained an adjuvant: a mixture of ingredients that boost the immune response. Adjuvants, previously used with other vaccines but not with flu vaccines, would allow a reduced dose and faster deployment.

  In Helsinki, the neurologist Markku Partinen was the first to detect a possible association between one particular flu vaccine called Pandemrix and new cases of narcolepsy in children.20 Partinen saw several children in the spring of 2010 in whom symptoms had developed rapidly, some with severe psychiatric symptoms. Almost all had cataplexy, and all those tested were DQB1*0602 positive. Fourteen were diagnosed before August, when public health officials in Finland suspended the flu vaccination program and launched an investigation. Similar findings were reported from Sweden and later from France and other European countries.21

  It remains in doubt whether Pandemrix’s adjuvant contributed to the problem, since a similar vaccine, Arepanrix, did not lead to a significant increase in narcolepsy cases in Canada.22 The differences between Arepanrix and Pandemrix, made by GlaxoSmithKline in different locations, may point to manufacturing or processing issues.23 GlaxoSmithKline sponsored research to examine the connection, bringing in narcolepsy experts such as Mignot. He worked with narcolepsy patient groups such as SOUND (Sufferers of Unique Narcolepsy Disorder) in Ireland to find those who had recently developed narcolepsy after being vaccinated.

  Of the roughly seven hundred cases of Pandemrix-related narcolepsy reported in Europe at that time, more than half were found in Sweden—but more than five million Swedes were vaccinated with Pandemrix.24 Given that risk ratio, it would have been difficult to catch the side effect in small-scale vaccine safety studies beforehand. Eventually, studies from several European countries showed that the apparent risk of developing narcolepsy for children immunized with Pandemrix was between five and fourteen times higher than the baseline in previous years.

  After the H1N1 pandemic swept through China, a spike in narcolepsy was reported from a large pediatric clinic in Beijing. The effect was independent of vaccination, but it strengthened the argument that an immune response to flu was involved.25 A network of pediatricians in the United States has observed a similar but weaker seasonal pattern of new narcolepsy cases.26 Neither Arepanrix or Pandemrix was used in the United States.

  VACCINE SKEPTICISM

  Partinen said that after making his concerns about Pandemrix known, other scientists ridiculed or avoided him and even raised doubts about his mental stability.27 He and his colleagues encountered resistance when trying to publish their findings in medical journals. The caution grew out of concern that antivaccine activists might exploit the information to erode public confidence in vaccines.28 Mignot told Reuters that “No one wants to be the next Wakefield.” He was referring to Andrew Wakefield, who was discredited for claiming a connection between the MMR vaccine and autism.

  A decade later, when I conducted an informal survey of immunologists outside the narcolepsy field, several said they were skeptical of the proposed mechanism. Past precedent, such as a link between the 1976 swine flu vaccine and Guillain-Barre syndrome, did exist. Public health officials have said that media attention in Europe made it more difficult to sort out the epidemiology, because greater awareness accelerated narcolepsy diagnoses that might have occurred later.29 An international study supported by the CDC found evidence for increased narcolepsy risk related to Pandemrix in Sweden but not in several other countries.30

  While 2009 H1N1 flu vaccination programs are estimated to have prevented hospitalizations and saved lives in the United States,31 several confounding issues make it difficult to tally risks versus benefits. These include increased awareness of narcolepsy coming from media attention, viral infections occurring in the same timeframe, and apparently greater risk for vaccine-induced narcolepsy among children compared with adults. The H1N1 flu pandemic’s relative mildness was welcome, but it also complicates retrospective analysis.

  THE IMPACT OF DISCUSSION

  Limited research exists on the impact that public discussion of the flu vaccine–narcolepsy link has had on vaccine hesitancy or antivaccine activism. In France, a country where vaccine skepticism is relatively high, a survey of general practitioners found that flu/narcolepsy represented one of several vaccine-related controversies and not the most prominent one.32 In the United Kingdom and Ireland, a steady stream of news articles has appeared about lawsuits and compensation claims against government agencies because of the vaccine-narcolepsy link. However, those driving them have said they’re not against vaccines but are only advocating for people with narcolepsy. The Narcolepsy UK website simply says: “We believe that the causal link between the Pandemrix vaccine and a rise in narcolepsy cases has been established, and that our society has an obligation to look after individuals who develop narcolepsy as a result.” Similarly, an in-depth study of the group Narkolepsiföreningen in Sweden, which was established specifically for people who developed narcolepsy after H1N1 vaccination, concluded: “None of them opposed vaccination in general, and none would place themselves in an anti-vaccination movement.”33

  Before the uptake of COVID-19 vaccines became politically charged in the United States, a few people at Atlanta narcolepsy support group meetings expressed worry about flu vaccines or others such as the human papillomavirus vaccine. Non-flu vaccines would be unlikely to have the same effect as Pandemrix. Still, given the known genetic risk, we can understand individuals’ concerns regarding possible exacerbation of existing narcolepsy or triggering of narcolepsy in their relatives.

  RETRACTION AND VINDICATION

  An initial fumble by Mignot’s group probably delayed wider scientific acceptance of the flu-narcolepsy connection. Researchers in his lab and another at Stanford had been looking at whether fragments of hypocretin could fit into the DQB1*0602 protein and activate T cells. In blood samples from Pandemrix-vaccinated children from Ireland, they could detect T cells that became activated in response to hypocretin. When comparing those children with their siblings, they could only find hypocretin-reactive T cells in those with narcolepsy. In addition, they noticed a sequence similarity between a fragment of hypocretin and part of the H1N1 flu virus. The idea was that through “molecular mimicry,” flu virus fragments that resembled hypocretin were arousing the immune system to begin targeting hypocretin-producing neurons. The paper didn’t specify whether vaccination or infection was the trigger—both could accelerate the immune confusion.

  The Stanford team published its results in 2013, receiving acclaim from other narcolepsy researchers. Unfortunately, after a scientist who performed several experiments in the paper left his lab, Mignot’s group was unable to reproduce their results. He and his colleagues decided to retract the paper in 2014. The NIH’s grant reviewers took note of the retraction, and one of Mignot’s major narcolepsy grants was not renewed. Some of his research staff had to leave to work elsewhere. “It was really painful and the worst time in my career,” Mignot said.34

  Four years later, researchers in Switzerland published more solid evidence for the T cell–mediated mechanism. The Swiss team, led by the neurologist Claudio Bassetti and immunologist Federica Sallusto, used a sensitive “T cell library” approach to detect rare hypocretin-sensitive cells in narcolepsy patients’ blood that others had trouble spotting before.35 In the intervening time, Mignot’s lab had redone its work. By that point, they could identify troublesome T cells directly, pulling out the T cell receptors’ genetic sequences. They published their follow-up a few months after the Bassetti/Sallusto paper.36

  The Stanford and Swiss publications diverged in a few ways. Unlike some of the narcolepsy cases in the Stanford paper, none from the Swiss paper were connected with Pandemrix. In addition, the Stanford paper showed T cells recognizing only a couple sections of hypocretin. In contrast, the Swiss paper had T cells recognizing fragments spread out along the entire protein.

  Despite differences between the papers, when both groups presented their work at the 2018 International Symposium on Narcolepsy in Massachusetts, there was a feeling that the field had turned a corner. The findings pointed toward a possible blood test for T cells recognizing hypocretin, which could facilitate narcolepsy type 1 diagnosis.

  A hot topic of discussion at the symposium was early intervention. Part of the excitement about narcolepsy type 1’s emergence as an autoimmune disorder is that there are many “off-the-shelf” options to try. Over the previous fifteen years, physicians have tried several times to arrest new-onset narcolepsy with immune-calming medications, such as a blend of antibodies (intravenous immunoglobulin) or corticosteroids.37 The thinking behind this was that in patients who had started to experience narcolepsy symptoms, it might be possible to quench the immune destruction. However, past attempts at early intervention had occurred on a small scale and were inconclusive.

  At the symposium, Michel Lecendreux, a child psychiatrist and sleep researcher from Paris, reviewed his and others’ past efforts at early intervention and suggested “the clock is ticking” for recently diagnosed narcolepsy patients. Still, there was a lack of consensus over whether a randomized clinical trial or even open-label studies would be appropriate.