Q&A on narcolepsy/hypersomnia

Quinn Eastman
13 min readDec 6, 2020

In October, someone with idiopathic hypersomnia contacted me with a series of insightful questions. It reminded me why I started writing my book in the first place. People with IH and narcolepsy type 2 are hungry to understand the origins of their sleep disorders — and science doesn’t have much to tell them, in contrast to narcolepsy type 1. I resolved to take a stab at answering the questions, although I may still need to add explanations and break things down more.

Most people in narcolepsy and IH registries are women; perhaps a question for another time

Do we have any new information about the etiology of narcolepsy type 2? Is it thought to be a dysfunction of orexin/hypocretin even though the CSF levels are normal?

The short answers are: not much, and maybe for some people. People with narcolepsy type 2 (NT2) are a heterogeneous group. In the United States, narcolepsy without cataplexy diagnoses have become more prevalent than narcolepsy with cataplexy — something that does not appear to be the case in other countries, such as Spain. This is likely because of differences in diagnostic practices, rather than differences between populations.

Keep in mind that the Multiple Sleep Latency Test, the standard procedure used to diagnose narcolepsy and IH, can rope in people who have high sleep pressure because they have sleep apnea or who work at night. That’s because the tendency to enter REM sleep during a daytime nap is not exclusive to narcolepsy. Also, the MSLT tends to provide inconsistent results when used to diagnose people with NT2 or IH, and the MSLT is often not implemented according to strict guidelines.

Sleep researchers in Zürich recently published a paper arguing that most apparent cases of NT2 could be attributed to insufficient sleep. They proposed to refine the MSLT by tightening the requirements (REM entry within about 5 minutes or directly from light sleep, and/or a high proportion of a nap spent in REM). Investigators in the UK have made similar observations about REM entry.

I am not prepared to say that most people diagnosed with NT2 just need to get more sleep, even if that’s what the Swiss authors imply, and they say it ain’t that simple. Anecdotally, it is possible for someone with an autoimmune disease such as lupus, or an endocrine disorder such as hypothyroidism, to be misdiagnosed with narcolepsy.

If insufficient sleep time is excluded and we are looking for a neurological explanation for NT2, there are several possibilities. Sleep researchers have been attracted to the idea that NT2 results from an incomplete autoimmune attack on hypocretin neurons. Perhaps the threshold of neuronal loss needed for someone to display excessive daytime sleepiness is lower than for cataplexy. This fits with observations that sleepiness can appear years before cataplexy. In this scenario, enough of the cells might remain to produce hypocretin visible in the CSF.

Unfortunately, there aren’t ways to directly check on those neurons in someone who is alive. We can only infer neuronal loss from the absence of CSF hypocretin, or from someone’s symptoms and performance on diagnostic tests. A Parkinson’s disease patient can undergo a PET scan, showing that they’ve lost dopaminergic neurons. Something analogous for narcolepsy doesn’t currently exist. In the United States, CSF hypocretin measurement itself has not been commercially available until recently.

To know what was going on in someone’s hypothalamus, researchers would need to take post-mortem tissue slices and stain them with antibodies against hypocretin. This was possible for several people with NT1 represented in the brain bank established by Michael Aldrich at the University of Michigan. But so far, people with NT2 are scarce in brain banks, and direct evidence for partial hypocretin neuron loss in humans rests on one case.

More indirect evidence comes from:

· monitoring hypocretin CSF levels in people who display sleepiness before cataplexy

· genetics: people with NT2 tend to have the HLA risk factor DQB1*0602 more than the general population. This also suggests that NT2 is heterogeneous.

· animal models of narcolepsy in which hypocretin neuron loss is incomplete

· hypocretin neuron loss in people with neurodegenerative diseases

Put together, this looks like a nice theory, but whether partial hypocretin cell loss applies to a significant number of people with a NT2 diagnosis is unclear. Harvard neurologist Cliff Saper has suggested that NT2 could instead come from loss or dysfunction of another group of neurons in the hypothalamus. NT2 could be like diabetes type 2 — instead of the brain losing the ability to produce hypocretin, other neurons may become insensitive to it.

What are the current hypotheses for proximate and ultimate causes of idiopathic hypersomnia?

One major idea is that IH is a variant of a circadian rhythm disorder. We can call this the “long biological night” theory: in IH, the nighttime phase is stretched, rather than shifted.

Several research groups have obtained supporting evidence. An influential study from Prague pointed to people with IH having a delayed and weaker onset of melatonin, the circadian marker that is usually high at night. Robert Thomas at Beth Israel has made related observations. In addition, in skin cells from a small number of IH patients, German sleep researchers have shown a reduced amplitude and a longer period for circadian rhythm gene oscillations. The German findings were intriguing and puzzling, because they suggested that the altered rhythms could be observed outside the brain.

Some details should be worked out. In IH, someone needs to look at melatonin levels and circadian rhythm gene patterns together. Altered circadian rhythm gene patterns might not be specific to IH; they’ve been observed in neurodegenerative diseases too. A hot topic in sleep research right now is the development of convenient circadian biomarkers. Why not take samples from people with IH? In those with very long sleep times, do their circadian markers cycle more weakly, or in a disorganized way?

Another theory, promoted by David Rye at Emory, is based on the neurotransmitter GABA. Emory investigators found that CSF samples from patients with IH and NT2 contained substances that neurochemically resemble benzodiazepines, enhancing responses to GABA. In addition, GABA-A receptor antagonists such as flumazenil and clarithromycin can increase wakefulness in some of these patients. The idea is that an endogenous GABA-A receptor modulator — possibly a peptide — is driving excess sleepiness.

This theory generated a wave of excitement several years ago, but it has been controversial since then. Peptide somnogens have a venerable history, and they keep popping up in studies of sleep regulation in organisms such as fruit flies. However, it is possible to explain how GABA-A receptor antagonists might work without invoking a peptide somnogen.

Neither the circadian or GABA theories explain the origin of IH’s neurochemical disturbances. Here’s where it gets murky. People with IH often report having sleepy relatives, and some say that they displayed excessive sleepiness since childhood. Do they represent extreme “long sleepers”, who are people who simply need more sleep? They might be the opposite of the short sleep families studied by Ying-hui Fu and Louis Ptacek at UCSF. IH could be an opportunity for geneticists.

Other people with IH recall a distinct onset, often in adolescence. Genetic susceptibility could play a role, analogously to how HLA risk factors set the stage for narcolepsy type 1. Perhaps an infection or other injury damages circuits within the brain, such as the suprachiasmatic nucleus, a region of the hypothalamus that governs circadian rhythms. In a study from the 1990s, squirrel monkeys with lesions of their suprachiasmatic nuclei displayed significantly longer sleep periods; the same effect is not observed in rodents.

Right now, we don’t know if the brains of people with IH have changes that can be consistently localized to a specific group of neurons. The hypothalamus could be a good place to look. People with IH often have disruptions of autonomic nervous functions (such as heart rate and blood pressure regulation), and the hypothalamus is important for those too.

Is it looking like Type 2 Narcolepsy and Idiopathic Hypersomnia will be merged into a single diagnosis?

Sleep specialists have recognized that the distinction between IH without long sleep and NT2 is flimsy. Two recent papers call for merging those two diagnostic categories, while carving off IH with long sleep. A difference between the two papers is in what the merged category would be called. One, from a group of American and European sleep specialists, proposes to rename NT2 + IH w/out long sleep “narcolepsy spectrum disorder.” The other paper, from European experts, proposes to name the merged category “idiopathic excessive sleepiness,” with a variety of sub-categories.

The European proposal raises concerns about what happens to medication access and insurance coverage if some diagnoses are redefined as not narcolepsy. But it is more rigorous about applying what has been learned in the last 20 years about narcolepsy. Fundamentally, is narcolepsy defined by a collection of symptoms, or by hypocretin deficiency? And how frequent or rare is genuine long sleep IH, given that its intensity waxes and wanes?

Is there in existence or planned a “fishing expedition” where lots of vital signs are taken of those with idiopathic hypersomnia/Type 2 Narcolepsy then big data is used to find similarities?

Yes (although “fishing expedition” can irritate scientists who want to be seen as hypothesis-driven.) A good place to look for current clinical studies on IH and related sleep disorders is the Hypersomnia Foundation’s web site.

One example is the study from David Plante and colleagues in Wisconsin, who used high-density EEG to map localized deficiencies with slow-wave activity in people with hypersomnia. These findings are suggestive because slow wave activity is associated with restorative sleep. Plante recently received a grant from the National Institute of Nursing Research to continue this work and validate additional measures of excessive daytime sleepiness. Thien Thanh Dang Vu’s group in Montreal is conducting a large brain imaging study of narcolepsy and IH — a continuation of their ground-breaking work in this area.

Currently, sleep researchers in Switzerland are recruiting people with NT1 and IH, taking CSF samples for proteomics and also looking at the gut microbiome. Their counterparts in Montpellier have a similar cohort study. In the United States, Lynn Marie Trotti and her team at Emory are conducting a clinical study of clarithromycin, collecting CSF and microbiome samples on the way.

Difficulties in designing studies on IH do come up. If we think that current distinctions between NT2 and IH are not meaningful, then maybe it makes sense to divide people up by different criteria, such as whether they display consistently long sleep periods. It’s hard to say that someone is getting insufficient sleep when they sleep 11 hours a day. Then the questions become: how do we establish whether someone habitually does sleep that much (wearable devices aren’t that great), and where do we draw lines between groups?

Given narcolepsy (type 1 at least) is caused by lack of hypocretin, and hypocretin stimulates histamine in the brain, is Wakix/pitolisant the best narcolepsy treatment?

Pitolisant is interesting because it’s a wake-promoting medication that doesn’t primarily work through dopamine, unlike modafinil or conventional stimulants. It does appear to have some favorable properties. Pitolisant can reduce cataplexy, and it is supposed to have low abuse potential, unlike oxybate/GHB, which was the only drug approved to reduce cataplexy for years. Still, the best medication is whatever works sustainably for the individual patient. Some people with narcolepsy may find pitolisant intolerable or ineffective. Especially as a PhD non-clinician, I would not endorse any single drug or even a combination as “the best.”

Many in the narcolepsy community have been following research on hypocretin receptor agonists, such as those being developed by Takeda. For people with NT1, these compounds carry the promise of being able to comprehensively replace the missing signals from hypocretin. Whether they will be better than other medications has to be tested thoroughly.

I have seen researchers say that hypocretin does not cross the blood brain barrier. However, there have been studies in monkeys taking intranasal hypocretin. Why is intranasal hypocretin not studied more in humans? How hard would it be to get human consumable hypocretin as a nasal spray? What would be the risks?

There was a burst of hype around intranasal hypocretin several years ago. A group in Germany has published studies on intranasal hypocretin in humans. Hypocretin’s co-discoverer Luis de Lecea recently published an open-access review on the topic.

One major concern: if narcolepsy type 1 does occur through an autoimmune mechanism, which looks increasingly solid, supplementary hypocretin would be like adding fuel to a smoldering fire. It might even hasten hypocretin cell loss. Another issue is uneven distribution of a snort of hypocretin across different parts of the brain, compared with natural localized production. Plus, there are two peptides, with different pharmacological properties, and two hypocretin/orexin receptors, which impart different signals.

With hypocretin replacement therapies, there also may be risks of heightened anxiety, or addictive behaviors, as de Lecea points out. Also, some researchers have speculated that chronic elevated hypocretin may accelerate amyloid-beta accumulation (hastening Alzheimer’s?). Here we see the advantages of small molecule receptor agonists. It may be possible to select for effects on sleepiness/alertness while tuning down signals that carry addiction risk.

UCLA’s Jerry Siegel has made the point that “as is”, hypocretin can’t be patented, although a pharmaceutical company could probably find a way over that hurdle. The German researchers who studied intranasal hypocretin in humans obtained it from a commercial supplier: Bachem. Let me emphasize: I do NOT advise do-it-yourself activity.

With pitolisant, solriamfetol, armodafinil/modafinil, amphetamines, GHB, and Tak-925 what are the prospects for polypharmacy in treatment? What are the risks? Is there a model for patients to get the most out of their medications?

As a non-clinician, I am hesitant to advise on combining medications. That said, a couple studies have been published on combining pitolisant with other medications. It is also common practice to combine modafinil or conventional stimulants with oxybate/GHB.

I think a general rule of thumb might be: don’t lean on any one neurotransmitter too hard — especially dopamine or norepinephrine. Unpleasant side effects such as irritability, anxiety, nausea, headache or loss of appetite, which occur sometimes with single drugs, may be more likely with more than one.

We know that Type 1 Narcolepsy is caused by lack of hypocretin. What establishes the cutoffs for normal levels? If someone were in the 25th percentile would he still function normally?

The standards for hypocretin CSF were set by this 2002 Archives of Neurology paper from Mignot, Nishino and colleagues. Low (NT1) = less than 110 picograms per milliliter, while normal = more than 200 pg/ml, with intermediate levels seen in a few cases of brain injury or other neurological disorders. Typical is around 350; 25 percent of that is in the NT1 range.

GHB is used for increasing slow wave sleep in people with narcolepsy. Increased slow wave sleep would mean less REM sleep. Does a decrease in REM sleep also decrease learning and memory formation which are associated with it? Are there other side effects?

More than 40 years after GHB emerged as a narcolepsy treatment, sleep researchers don’t really know how it works. And don’t worry about a reduction of REM sleep. Lots of people take REM-suppressing antidepressants for depression (how they work is another matter) and those drugs don’t seem to interfere with learning and memory. People can, somewhat surprisingly, go without REM sleep. See this review by Siegel.

What we do know is that GHB doesn’t work simply by suppressing REM sleep. It is thought to consolidate nighttime sleep by enforcing sleep continuity — it often takes days or weeks to build up to a substantial effect. Increasing slow wave sleep could be part of how it works, although Swiss researcher Mehdi Tafti and his colleagues found that the slow wave sleep induced by GHB looks artificial, not physiological. GHB is naturally present at low levels in the undrugged brain, but when GHB is given as a fast-acting sedative, it reaches levels that are hundreds of times higher.

Many sleep researchers would say that GHB acts as a weak GABA-B agonist. But if that were the only way it was exerting its effects, then the less expensive and less dangerous GABA-B agonist baclofen could substitute for it. A study in mouse models suggested that baclofen could be effective for narcolepsy; most clinicians would not agree. I’ve heard from members of the hypersomnia community who’ve tried baclofen for themselves, with mixed results.

Besides being a hypnotic, does melatonin change the makeup of sleep (deep sleep, REM, etc)?

Melatonin can shift someone’s circadian phase forward, but it is not considered a hypnotic in the same way as antihistamines or benzodiazepines, and is not thought to change sleep architecture. It can have an indirect effect on REM/slow wave, depending on when it is given and the dose. This Swiss study showed melatonin can increase REM in healthy young adults, for example.

Dr David Rye argues that endozepines, chemicals that function like benzodiazepines, cause idiopathic hypersomnia. Has he found this chemical?

Not yet — and Rye and his collaborator at Emory, Andrew Jenkins, would say that the term “endozepine” is misleading. Jenkins once said it should be abolished in reference to IH. To be clear, EEGs from people with IH don’t look like they’re on benzos, and the CSF peptides Rye and Jenkins have been looking for are not the same as benzodiazepines.

Over the last three years, occasional visits to Jenkins’ laboratory gave me the impression that he was close to his prize, although recently COVID-19 has slowed things down. Jenkins and others at Emory were investigating CSF peptides via proteomics, but they have not come to the point of being able to announce any results publicly.

For a time, Jenkins’ patch clamp assays were used to confirm diagnosis of “GABA-related hypersomnia”, although his laboratory wasn’t dedicated for this purpose. At the 2018 Hypersomnia Foundation conference, he apologized for delays and pledged to catch up. Since then, he’s acquired some sophisticated equipment, allowing his assays to be performed more quickly and with less CSF material.

It was found that the Pandemrix vaccine caused an increase in Narcolepsy cases. I recently attended a Wake Up Narcolepsy online conference where HPV vaccine was discussed as a possible Narcolepsy cause. Although the benefits of vaccines far outweigh the costs, have there been any population wide studies of vaccine side effects? Have there been any other autoimmune problems associated with vaccines?

The proposed mechanism for H1N1 influenza-associated narcolepsy involves a specific viral peptide and molecular mimicry. It is very unlikely that a non-flu vaccine would trigger narcolepsy in the same way. Also, a Danish study found that narcolepsy was not linked to HPV vaccination.

Another example of an autoimmune problem associated with a vaccine is the 1976 swine flu vaccine and Guillain-Barre syndrome. A full catalog of vaccine side effects is beyond the scope of this Q+A. As a participant in a vaccine clinical trial, I agree that the benefits of vaccines outweigh the costs, and am looking forward to availability of a SARS-CoV-2 vaccine.



Quinn Eastman

Quinn was trained as a biochemist and then as a science journalist. His book on idiopathic hypersomnia will be published by Columbia University Press in 2023.