ABOVE: MRI images of a human brain © ISTOCK.COM, SUDOK1

Research published last month (October 25) in JAMA Neurology represents the latest evidence for the controversial notion that the onset of Parkinson’s disease can be triggered by a viral infection. But experts tell The Scientist that the viral Parkinson’s hypothesis still lacks convincing evidence.

For the study, researchers analyzed medical records from the Danish National Patient Registry dating between 1977 to 2016 and found a correlation between a Parkinson’s diagnosis and an influenza diagnosis from at least 10 years prior. In total, the scientists included the records of 61,626 people, 10,271 of whom had been diagnosed with Parkinson’s. They found that people who were diagnosed with the flu were 73 percent more likely to be diagnosed with Parkinson’s more than a decade later than were people who’d never had a flu diagnosis.

The study is “certainly intriguing,” University of Utah neurobiologist Jason Shepherd tells The Scientist in an email. He adds that he finds a connection between viral infections and neurodegenerative conditions such as Parkinson’s to be plausible, because “anything that can trigger an inflammatory response in the brain can be a trigger for subsequent neurodegeneration.”

See “Can the Flu and Other Viruses Cause Neurodegeneration?

Previous research on seasonal and epidemic flu has indicated that the viral infections can cause symptoms similar to Parkinson’s in the short term. Infections can trigger inflammation and perhaps injury in regions of the brain such as the substantia nigra that are affected by the neurodegenerative disease. However, there have been no data indicating that a case of the flu directly increases one’s risk of developing Parkinson’s years or decades later. The new study credibly indicates that there is a relationship between the flu and Parkinson’s disease that ought to be further probed in subsequent research, experts tell The Scientist.

Thomas Jefferson University Parkinson’s researcher Richard Smeyne tells The Scientist that he sees a connection between the study’s results and previous animal research by his group and others. In a 2017 study published in npj Parkinson’s Disease, he and colleagues infected mice with either the H5N1 bird flu virus or the H1N1 virus that caused the 2009 swine flu pandemic, allowed them to completely recover, and then treated them with a drug that attacks dopaminergic cells in the substantia nigra and causes Parkinson’s–like symptoms. The drug induced the expected effect in the mice, but those that had been infected with the flu suffered a greater degree of dopaminergic loss than the control group. However, mice that had been inoculated with a flu vaccine, treated with Tamiflu, or both appeared to be protected against neurological damage.

As part of the current study, the team also looked for an association between an infection of any type and a Parkinson’s diagnosis more than 10 years later, and did not find one. In addition, the researchers identified other infections including pneumonia and septicemia that seemed to be linked to a Parkinson’s diagnosis within five years, but the correlations disappeared over longer timeframes so the researchers assumed that they were flukes. Conversely, the team found no association between flu and Parkinson’s disease within five years. If there is a longer-term relationship between the flu and Parkinson’s, the scientists conclude, it plays out over at least a decade.

“There is some compelling evidence for a viral connection to Parkinson’s,” study coauthor and Harvard Medical School population medicine researcher Noelle Cocoros tells The Scientist by email. But acknowledging the controversy, she adds that “this question is quite difficult to study and it is important not to overemphasize any single analysis.”

“Overall, the study is interesting but only suggestive,” Icahn School of Medicine at Mount Sinai geneticist Joel Dudley, who was not involved in the work, tells The Scientist in an email. He adds that the combination of small effect sizes and the use of observational data give him some pause and suggests that other viral infections might lead to similar issues, even though the new paper didn’t find such a link. “So, it’s not clear at all if this is influenza-specific, or just generally related to activation of innate and adaptive immune systems.”

University of Toronto neurologist and movement disorder specialist Anthony Lang tells The Scientist he isn’t entirely convinced that the researchers have found much at all, pointing out what he sees as issues with the data. For instance, he notes, “the diagnosis was a clinical diagnosis, there was no biomarker for influenza here,” meaning that their diagnoses may have been incorrect in the first place. He adds that the diagnosis numbers could be further muddled by the fact that health records only account for those who sought medical attention. “Especially during the influenza season, if you know influenza is going around and you feel like crap, do you go see your doctor?” he asks. “Most people know there isn’t a real treatment for it, so they stay home and grin and bear it.”

Overall, the study is interesting but only suggestive.

—Joel Dudley, Icahn School of Medicine at Mount Sinai

When the team ran a separate analysis that only counted patients with flu diagnosis in the months when influenza was circulating among the population at a higher rate—an attempt to ensure that the recorded diagnoses they counted were more likely to be correct—their calculated correlation between infection and a subsequent Parkinson’s diagnosis grew weaker. That detracts from the strength of the team’s conclusions, Smeyne says, but it may just be an oddity of the sort of retroactive population study that’s required to tease out correlations that might emerge over the course of decades. He tells The Scientist that any one of the endless variables that influence health outcomes—whether or not they’re recorded in the database—could have an unexpected effect, necessitating further study before any definitive conclusions can be drawn.

Cocoros and coauthor Victor Henderson, a neurologist and epidemiologist at Stanford University, acknowledge that drawing conclusions from historical health records will always result in imperfect answers, but they argue that large health record databases like Denmark’s are the best tools available to conduct research over such a long period of time. Another concern, though, is that the Danish registry’s population isn’t very diverse, experts tell The Scientist, so it would be nice to see the study replicated on a broader data set.

Lang says he would have been more convinced if the study included an analysis of hospitalized patients in addition to diagnosis. A higher incidence of Parkinson’s among patients who’d been hospitalized for the flu than among those who had milder infections would have strengthened the argument for a causal link, he argues, as it would suggest that disease severity plays a role in triggering neurological symptoms over longer periods of time.

That said, the correlation the authors found “is one of the bits of information that may provide us some insight into some factors or factor that might influence the development or the course of Parkinson’s,” Lang says.

Gleaning clinically relevant information from animal research is notoriously difficult, but Smeyne tells The Scientist that he feels the Danish population study has validated his animal model, which he can now go on to use in an attempt to identify the sort of casual links that can’t be discerned from retrospective human studies.

“To see this epidemiological study support almost completely what we had shown in experimental animals . . . it gives me the impetus to go on and keep doing the work we’re doing,” says Smeyne.