As the coronavirus pandemic continues, scientists are racing to understand the underlying causes and implications of long COVID, the umbrella term for symptoms that persist for at least 12 weeks but often last even longer and affect roughly 30 percent of individuals who contract COVID-19. Evidence for specific risk factors such as diabetes and the presence of autoantibodies is starting to emerge, but throughout the pandemic, one assumption has been that an important indicator of whether a COVID-19 survivor is likely to develop long COVID is the severity of their acute illness.

However, a preprint shared online on January 10 suggests that even mild SARS-CoV-2 infections may lead to long-term neurological symptoms associated with long COVID such as cognitive impairment and difficulties with attention and memory, a suite of symptoms often lumped together as “brain fog.” In the study, which has not yet been peer-reviewed, scientists led by Stanford University neurologist Michelle Monje identified a pathway in COVID-19–infected mice and humans that almost perfectly matches the inflammation thought to cause chemotherapy-related cognitive impairment (CRCI), also known as “chemo fog,” following cancer treatments. On top of that, the preprint shows that the neuroinflammation pathway can be triggered even without the coronavirus infecting a single brain cell.

“It is neuroinflammation that happens even in the absence of any central nervous system infection,” Monje tells The Scientist.

See “Mechanisms of Long COVID Remain Unknown but Data Are Rolling In

As far back as March 2020, Monje feared that cytokine storms caused by the immune response to SARS-CoV-2 would cause the same neuroinflammation and symptoms associated with CRCI, she tells The Scientist. But because her lab doesn’t study viral infections, she had no way to test her hypothesis until other researchers created the appropriate models. In the study, Monje and her colleagues used a mouse model for mild SARS-CoV-2 infections developed at the lab of Yale School of Medicine biologist and study coauthor Akiko Iwasaki as well as brain tissue samples taken from people who had COVID-19 when they died to demonstrate that mild infections can trigger inflammation in the brain.

“The mice [in Iwasaki’s model] have relatively mild disease, which allows us to ask exactly the question I was wondering,” Monje tells The Scientist. “All these really bad things can happen in severe COVID, but what happens in just relatively mild lung infection, when the mice look fine and don’t lose weight” or show other obvious signs of illness?

A clue as to what was happening came from her previous CRCI research, in which Monje found that immune cells in the brain known as white matter microglia seem to be “exquisitely sensitive” to infection. The inflammatory process begins when microglia become reactive and cause dysregulation of support cells in the brain called astrocytes, which assume a neurotoxic state. Together, Monje explains, they dysregulate the cells that typically produce myelin, the layer of insulation on neurons that helps signals propagate and protects cells from damage.

“When you disrupt both myelin homeostasis and myelin plasticity,” Monje explains, “you would expect that to cause a lot of the symptoms that you see in these syndromes of cognitive impairment.”

It is neuroinflammation that happens even in the absence of any central nervous system infection.

—Michelle Monje, Stanford University

The researchers measured levels of immune messengers called cytokines and other biomarkers in the cerebrospinal fluid (CSF) of the mice. Then, they sectioned their brains to measure demyelination by using a transmission electron microscope to examine neurons in a region of the brain called the corpus callosum. As expected, in the mice with mild COVID-19, “we found white matter-specific microglial reactivity, a dropout of [myelin-producing] oligodendrocytes, impairment in myelin, and inhibition of new neuron generation in the hippocampus, just as predicted,” she says. “All of these things we had previously associated with this toxic microglial activity [in CRCI].” All this occurred without the virus infecting the brain itself.

As Iwasaki explains on Twitter, cytokine levels remained elevated in mouse serum and CSF for seven weeks after the infection, which is a duration comparable to many cases of long COVID in humans. The researchers also found that a cytokine called CCL11, previously shown to tamp down the growth of new neurons in the brain’s hippocampus, was upregulated for seven weeks after infection as well.

The researchers were then able to study samples from nine human brains taken from people who died in March 2020 and were later confirmed to have had COVID-19. Their brains showed the same patterns of microglia activity that emerged in infected mice and in human CRCI studies.

The results made it clear, Monje says, that “essentially the same biology,” is at work in CRCI and in neurological symptoms of long COVID, “except instead of it being [chemotherapy], it’s peripheral inflammation” that’s to blame.

The preprint hints at the possibility of therapeutic interventions to treat or prevent neurological symptoms of the condition, Monje tells The Scientist, as some CRCI treatments are already being explored in early clinical trials that she describes as small but “promising.”

The brain samples aren’t as convincing as the mouse study, Northwestern University neuroscientist Alicia Guemez-Gamboa tells The Scientist, because much less is known about the donors’ symptoms before death, and it’s not clear whether their brains were directly infected. In addition, “dead people cannot have COVID fog!” Guemez-Gamboa points out. Monje notes that the samples probably came from people with more severe COVID-19 and concedes that the study was limited by the small number of available human samples.

See “COVID-19’s Effects on the Brain

Guemez-Gamboa says that the mouse experiment is “very compelling” because Monje’s background and the model’s specifications allow researchers to ask and answer specific questions about neural mechanisms, adding that “I think it’s a really good paper and it’s a really good start” toward revealing the risk factors and mechanisms of long COVID.

The mouse component of the study had limitations too, however. Guemez-Gamboa notes that behavioral research on mice would have allowed the team to complement their histological and cellular analyses with cognitive tests. That, she suggests, would have solidified the notion that the observed inflammation is actually linked with brain fog. But behavioral experiments weren’t possible in this case because the study was pieced together across a multitude of collaborators’ labs and facilities, Monje explains.

“What I would love to be able to do in the future, when we have access to a BSL-3 facility, is test whether any potential intervention not only rescues the cellular phenotype but also rescues the cognitive function, to the extent we can measure that,” Monje says.

The experimental setup also could not account for the effects of vaccines, prior infections, or other confounding factors, Monje notes, or reveal whether the neuroinflammation pathway and its effects differ between children and adults.

Clinical relevance

While the models and analyses seem robust, the link between the results and clinical outcomes for patients is shakier, Monica Malta, a psychiatrist and mental health policy expert at the University of Toronto, tells The Scientist. In part, that’s because there’s a wide variety of environmental factors that can cause cognitive symptoms such as impaired memory or attention span—not the least of which, Malta says, is the stress of maintaining a career, raising children, or otherwise managing during a years-long pandemic. Factors including age-related dementia or underlying mental health conditions may complicate attempts to study long COVID as well, Malta suggests, as clinicians don’t have the tools to tease the various illnesses apart.

What we’re describing here, specifically, is all potentially reversible.

—Michelle Monje, Stanford University

As a result, Malta says that she thinks drawing a line between the results of the preprint and clinical outcomes for patients “is a big jump to make.”

John Baratta, a clinician at University of North Carolina Health who runs a rehabilitation clinic for long COVID patients, tells The Scientist that studying long COVID remains difficult because SARS-CoV-2 is still relatively new, making it hard to understand its long-term implications. There simply isn’t enough information out there to draw many definitive conclusions. “We’re still collecting data, at this point, to help determine what the risk factors for long COVID might be,” Baratta says.

Baratta adds that the syndrome is particularly heterogeneous when it comes to the timing and duration of symptoms, as well as what those symptoms are. He adds that, in his experience, “it does seem that one key risk factor is the severity of illness,” but doesn’t rule out the possibility that mild cases of COVID-19 can lead to long-term symptoms as well.

See “Studies Identify Risk Factors for Long COVID

“There is not a really clear understanding in the medical community as far as why long COVID is happening,” Baratta says, “particularly why it affects certain people and not others.”

While many unanswered questions remain, Monje says her study lays the groundwork for future experiments while hinting at the possibility of clinical interventions for coronavirus-related brain fog.

“What I’m really encouraged by is that we know, in the context of chemotherapy or cancer therapy, that this kind of glial dysregulation is reversible,” Monje tells The Scientist. Because the inflammatory pathway caused by mild cases of COVID-19 doesn’t seem to kill irreplaceable neurons, Monje says that “what we’re describing here, specifically, is all potentially reversible.”