Skip to Main Content

Nearly a decade on from the 2009 influenza pandemic, scientists are still trying to solve what is proving to be an intractable medical mystery: Did some of the vaccines used to protect against the new flu virus trigger an increase in narcolepsy cases?

A major attempt to unravel the mystery — a study that’s still in the publication pipeline — did not find evidence that vaccines containing a boosting compound called an adjuvant sparked a rise in cases of narcolepsy, a serious but rare sleep disorder.


But absence of evidence is not evidence of absence, as scientists are quick to point out. Among them would be the researcher who led the effort.

Dr. Steven Black believes that the SOMNIA study, for a variety of reasons, was unable to crack the puzzle. And he thinks it’s critical to keep looking for the answer, because the use of adjuvants will likely be necessary the next time a severe flu pandemic hits — both to make the vaccine more potent and to stretch limited supplies.

“We don’t know if there’s an association or not for sure. I believe there is, even though the study I did showed that there wasn’t,” Black, an infectious diseases expert at Cincinnati Children’s Hospital, told STAT.


“The upshot of this is that this is a complicated story,” he said. “But I think the real upshot is we need to figure this out because there will be another pandemic.”

Black was the principal investigator on SOMNIA; the acronym stands for Systematic Observational Method for Narcolepsy and Influenza Immunization Assessment. The research was funded by the Centers for Disease Control and Prevention. (Black has consulted with a number of companies that make or made vaccines, including two mentioned in this article, GSK and Novartis.)

An H1N1 flu vaccination clinic in 2009 in San Francisco. Justin Sullivan/Getty Images

The United States did not use adjuvanted vaccine during the 2009 pandemic, which was mild by historic comparisons. Even if vaccines containing the boosting compounds led to or contributed to a spike in narcolepsy cases, people here were not at risk.

But the U.S. stockpiles two adjuvants as a hedge against a severe flu pandemic, meaning pandemic preparedness authorities are keen to know if they come with previously unknown risks.

Narcolepsy, a neurological disorder that affects the body’s ability to maintain normal sleep-wake patterns, would certainly be among them. People who develop the condition experience periods of profound daytime sleepiness, to the point where they fall asleep, sometimes mid-task. Some experience cataplexy, whereby they lose muscle control and fall.

It is an autoimmune disorder, caused when cells in the brain that produce a chemical called hypocretin are destroyed. People can develop narcolepsy at any stage of life, but onset most commonly occurs in childhood, adolescence, and young adulthood.

The condition is rare — about one person in 100,000 develops narcolepsy in a given year. It’s also chronic; the damage done to the hypocretin-producing neurons is irreversible. But narcolepsy can be managed, especially with early and aggressive treatment.

“I think a lot of narcoleptic patients can lead a quasi-normal life. They are never completely normal, but they are close to normal and they can usually do most jobs,” said Dr. Emmanuel Mignot, a leading sleep disorder expert who has been working on the possible link between the H1N1 pandemic and narcolepsy. Mignot was not involved in the SOMNIA study, but has followed the work.

New H1N1 viruses started circulating in 2009, and the first hints that vaccinations might have been linked to a rise in cases of narcolepsy came from Scandinavia. In the summer of 2010, sleep experts in Finland, and later in Sweden, realized they were seeing more cases of the disorder than they normally would. Suspicion focused on the pandemic vaccine.

Both countries had used Pandemrix, a vaccine made in Germany by GSK that contained the adjuvant AS03. (Pandemrix is no longer sold; after the pandemic GSK and other manufacturers reverted to making seasonal vaccines that protect against several types of flu. GSK’s seasonal flu vaccines do not contain AS03.)

Syringes with Pandemrix flu vaccine in Dresden, Germany. Matthias Rietschel/AP

As reports of the possible association started to emerge, researchers in other countries looked to see if there had been an increase in narcolepsy cases after their H1N1 vaccination campaigns were rolled out. Dozens of studies, meta-analyses, and commentaries have been published in the years since.

A review article published in April by the Finnish sleep experts who first flagged the problem concluded the association existed but was limited to Pandemrix. They suggested that for every 18,400 children and adolescents who received the vaccine, one would go on to develop narcolepsy.

Preben Aavitsland, a professor of public health at the University of Oslo, said the experiences of Finland, Sweden, Norway, France, England, and Ireland — all of which used Pandemrix — are suggestive of an increased risk from use of the vaccine.

“I think Pandemrix in rare instances contributed to causing narcolepsy in children and young adults,” said Aavitsland, who is a senior physician at the Norwegian Institute of Public Health and was deeply involved in Norway’s pandemic response in 2009, including its vaccination program with Pandemrix.

The concerns raised by the Finnish and Swedish researchers attracted a lot of media attention back in 2010, a fact that may have actually made it harder to see if the association was real. In a presentation of the SOMNIA findings at a recent meeting of the CDC’s Advisory Committee on Immunization Practices, Dr. Tom Shimabukuro — a CDC vaccine safety expert and a member of the SOMNIA team — used the analogy of an apple tree to explain the problem, known as “awareness bias.”

In the normal course of events, apples ripen and fall in their own time, but if you shake the tree, more apples fall at once. With narcolepsy, some of the people who are genetically vulnerable to developing the condition will be diagnosed with it over time. In some countries, however, the time from onset of the condition to a diagnosis can be years. But the publicity surrounding the problem — the tree shaking — may have brought forward narcolepsy diagnoses that might otherwise only have come to light later.

That could create an artificial spike in cases that appears to suggest a causal relationship between, in this example, the pandemic vaccine and incidence of narcolepsy — even if such a relationship doesn’t exist.

Conversely, the long lag between onset of narcolepsy symptoms and diagnosis could make it harder for scientists to detect a genuine phenomenon.

Black uses the metaphor of a food poisoning incident in a restaurant. “There were 10 people there and one got diagnosed a month later and one six months and one a year later and one two years later — you’re not going to be able to figure out what happened,” he said. Plotted on a graph, those cases — though actually related — don’t form a peak. “You’d see a blur.”

Not surprisingly, the studies that preceded SOMNIA paint a confusing picture, likely for a combination of reasons. The effect of awareness bias. The fact that vaccination efforts started at different times in different countries. A variety of brands of vaccines were used.

A small grouping of H1N1 influenza virus particles. NIAID

Those same factors faced the SOMNIA researchers, who were studying vaccines that contained AS03 but also MF-59, an adjuvant that was included in pandemic vaccines produced by Novartis. (Novartis has since sold its influenza vaccine operations to Seqirus.)

Sweden, which used Pandemrix, saw a sharp increase in narcolepsy cases. But Ontario, which used a different vaccine with the AS03 adjuvant, did not, according to a study there.

While both Sweden and Ontario used GSK vaccines, the vaccines were manufactured in different countries, using different manufacturing processes, complicating any comparison.

“It’s probably [an] apples to apples [comparison], but it’s McIntosh to Golden Delicious or something. They’re not identical. And since we don’t know what the mechanism is, if this association was real, that was one issue,” Black said.

That’s another huge part of the problem for people trying to solve this mystery. It’s not clear why the vaccines would have raised a recipient’s risk of developing narcolepsy.

Scientists are still trying to figure out what triggers destruction of the hypocretin-producing neurons, but attention has focused on infections, Mignot said, because there’s a seasonal pattern to narcolepsy diagnoses. They appear to be more common in the spring and summer.

“It was really fitting with the idea there was something with the winters that was triggering the auto-immune process,” said Mignot, who is the director of Stanford University’s Center for Sleep Sciences and Medicine.

It had been thought streptococcal infections — the cause of strep throat — might be a trigger for some people who are genetically vulnerable to developing narcolepsy. But after the pandemic H1N1 virus made its way around China, authorities there reported an increase in narcolepsy diagnoses.

Importantly for this investigation, China hadn’t vaccinated during the pandemic. The increase in narcolepsy cases there put influenza infection on the table as a possible trigger.

Chinese medical staff checks the body temperature of a passenger after landing in Shanghai in 2009. Joern Pollex/Getty Images

“So now our hypothesis is that the flu in some circumstances can produce an immune reaction that turns wrong, and instead of clearing the flu, is also attacking the hypocretin neurons. And then you have narcolepsy,” Mignot said.

Taiwan, which used the Novartis adjuvanted vaccine during the pandemic, also reported a spike in narcolepsy cases. But there the increase in diagnoses occurred before the country began to vaccinate, which further supports the theory that infection with the pandemic H1N1 virus may have been associated in some places with a rise in narcolepsy cases. Studies in mice have shown the virus is able to move into the brain.

Both China and Taiwan have a short interval — weeks — between onset of excessive daytime sleepiness and the diagnosis of narcolepsy. That means if there are spikes in cases, they are easier to detect.

The fact that infection rather than vaccination was linked to an increase in narcolepsy in China and Taiwan both complicates the vaccine-narcolepsy story and makes Black even more convinced the effect was real.

“If you see it with wild-type virus, why wouldn’t you believe it could occur with vaccine?” Black said.

Earlier this year, Black convened a meeting to try to sort through the data generated by the SOMNIA team, composed of experts from nine countries including Sweden, Canada, the Netherlands, Spain, and Argentina.

Instead of answers, they came up with more theories, including one that argues that the timing of vaccination efforts may have played a role in the narcolepsy story.

Under that scenario, countries that rolled out vaccine while the H1N1 virus was actively circulating would have vaccinated some people who were actually infected at the time. (Some flu infections are low grade.) The combination of virus and vaccine might have actually created an enhanced immune response, Black explained.

Canada was one of the first countries hit by the pandemic virus. And while it was also one of the earliest places to start vaccinating, by the time the vaccine had been made many people there had already been infected. But the pandemic wave struck Sweden later.

People wore protective masks in the western Ukrainian city of Lviv in 2009. Yuriy Dyachyshyn/AFP/Getty Images

This idea is of a “two-hit process,” Black said. “It’s the only thing that I’ve heard that allows you to reconcile the two sets of results.”

Finnish researchers already explored this idea, however, Aavitsland said. A study published in 2013 reported that of 45 people in Finland who had developed narcolepsy after being vaccinated with Pandemrix, only two had antibodies to the pandemic H1N1 virus.

Mignot, too, believes there was a true rise in narcolepsy cases during the pandemic, both among vaccinated and unvaccinated people. The available data suggest the problem was worse in places that used Pandemrix, he said. But he cautioned that because of the differences in timing — whether vaccination campaigns followed or were concurrent with spread of the virus — it’s not possible from these studies to know if there was something about that particular brand of vaccine, or the time at which it was used.

“It’s apples and oranges. And that’s why it’s so difficult at the end to make a picture of the epidemiology. Because every situation is a combination of factors and you don’t know which one was key,” Mignot said.

He also noted that most people who were vaccinated during the H1N1 pandemic didn’t go on to develop narcolepsy. So even if the virus itself and the adjuvanted vaccines made to protect against it triggered an increased risk, the outcome was still rare.

“It’s not like: You have the right gene, you get the vaccine, you get narcolepsy,” Mignot said. “Clearly even with people who get the vaccine there are a lot of other things that must have been confluent to produce narcolepsy.”

Whatever happened in the H1N1 pandemic, it does not appear to have continued. The rate at which new cases of narcolepsy are found appears to have returned to its baseline, Mignot said.

That means the answer to this troubling and important question may come from laboratory work, not further epidemiological studies. Mignot’s lab is studying the way in which blood from narcolepsy patients reacts to various parts of flu viruses, hoping to find answers. But he has no funding for the research and only a single member of his team is working on it.

“I love this story because it’s like a medical detective story,” Mignot said. “The only problem is: Will we find the answer?”

Aavitsland uses another word to describe this story. “The Pandemrix-narcolepsy saga is a tragedy.”