Aging people lose variation in brain oxygen levels—a sign of declining cognitive flexibility. A new drug study probes whether that loss can be reversed.
To eavesdrop on a brain, one of the best tools neuroscientists have is the fMRI scan, which helps map blood flow, and therefore the spikes in oxygen that occur whenever a particular brain region is being used. It reveals a noisy world. Blood oxygen levels vary from moment to moment, but those spikes never totally flatten out. “Your brain, even resting, is not going to be completely silent,” says Poortata Lalwani, a PhD student in cognitive neuroscience at the University of Michigan. She imagines the brain, even at its most tranquil, as kind of like a tennis player waiting to return a serve: “He’s not going to be standing still. He’s going to be pacing a little bit, getting ready to hit the backhand.”
Many fMRI studies filter out that noise to find the particular spikes researchers want to scrutinize. But for Lalwani, that noise is the most telling signal of all. To her, it’s a signal of cognitive flexibility. Young, healthy brains tend to have signals with a lot of variability in blood oxygen levels from moment to moment. Older ones vary less, at least in certain regions of the brain.
About a decade ago, scientists first showed the link between low neural signal variability and the kind of cognitive decline that accompanies healthy aging, rather than specific dementias. A brain’s noisiness is a solid proxy for details that are more abstract, Lalwani says: “How efficient information transfer is, how well-connected the neural networks are, in general how well-functioning the underlying neural network is.”
But why that change happens with age has been a mystery. So has the question of whether it’s reversible.
In results published in November in the Journal of Neuroscience, Lalwani’s team showed that a small dose of Lorazepam, an anti-anxiety medicine, could reverse the dip in signal variability, at least momentarily. The drug dials up inhibitory messages in the brain but makes it more dynamic, ready to react and respond quickly. In the study, the brain signals of older participants who had previously performed poorly on cognitive tasks returned to noise levels that looked more like those of young people.
“Ten or so years ago, most people thought that variability in the brain was a bad thing,” says Cheryl Grady, a cognitive neuroscientist at the Rotman Research Institute who has studied brain signal variability but was not involved in Lalwani’s study. But now, she feels, more people realize the potential of this new metric. “I’m very much in favor of this whole approach.”
Around 2008, researchers began to suspect that the so-called noise in fMRI signals had a deeper meaning. By 2010, Douglas Garrett, then a PhD student, had shown that variability in blood oxygen fMRI signals predicted a person’s age better than the size of the spikes in those readings. His hunch was that standard deviation—a measure of how similar or different the signals are in a raw dataset—could tell stories that simply averaging spike sizes couldn’t.
Time and time again, he and his collaborators have shown a strong link between this “noise,” age, and cognitive processing speed. When you perform cognitive tasks, like having to quickly pick two images that match, your brain signals become more variable. But in 2012, Garrett showed that young people get significantly bigger variability increases than older people when doing matching tasks. And in 2017, his team proved that age-related changes in variability are not some consequence of individual differences in blood flow.
Variability among those blood oxygen level signals consistently beat averagesignal size as the best predictor of cognitive decline. “The same task, the same subjects, same data,” he says. “It won that race, and continues to win that race, every time we have it.” Garrett, now a senior scientist at the Max Planck Institute for Human Development, coauthored the new paper with Lalwani.
This notion—that the noise could be a signal—wasn’t always well-received. “People were actually thinking about that as just nonsense,” says Greg Samanez-Larkin, a cognitive neuroscientist at Duke University who was not involved in this study. (He joined the study of signal variability as a graduate student around the same time as Garrett.) In his own experiments, he has studied the correlation between levels of this noise and the likelihood of older adults making worse financial decisions.
In their new work, the team focused on the brain chemistry that controls this change in levels of neural noise. The neurotransmitter GABA seemed like an obvious target. GABA is a tiny molecule, just about twice as heavy as a single molecule of carbon dioxide. Yet it serves an enormous role as the brain’s primary inhibitor. When molecular receptors in neurons come across GABA, they’re less likely to fire. Too little GABA can make a brain overexcited. Too much … well, that’s how anesthesia works. GABA is the wet blanket that keeps the brain’s party balanced, and its concentrations drop in aging brains.
Lalwani’s team recruited 25 young adults (between 18 and 25 years old) and 21 older ones (ages 65 to 85) for an fMRI study. Before being scanned, each person completed seven tasks designed to test cognition. One asked them to match images by shape or color, or to spot whether two images were different, as quickly as possible. Another tested their ability to read aloud. Still another asked them to memorize the photos of animals they saw on an iPad, then list them in order of size. The tasks gave Lalwani’s team an objective way of categorizing the participants’ normal cognitive levels.
Then, each person underwent two fMRI scanning sessions, during which they rested quietly while a snapshot of their brain’s blood oxygen levels was collected every two seconds. For one scan, the team gave each person a low dose of Lorazepam, a benzodiazepine known to enhance the effect of GABA. For the other, everyone took a placebo that had no effect. (The order for each session was randomized, so participants did not know which pill they had taken first.)
The placebo sessions revealed a familiar story: Older participants showed less brain signal variability in their resting state than the younger ones, particularly in frontal regions of the brain responsible for memory and language, and others linked to processing sight, sound, and touch. But would the researchers see a difference during the Lorazepam sessions?
Lalwani remembers waiting for her computer to churn through the enormous data set of brain images. “I was seeing the progress bar go,” she recalls. Waiting. More waiting. Then, her answer loaded: Yes, people had more signal variability when taking Lorazepam. And even better, she could see who had the biggest change: the people who had performed the worst on the cognitive testing tasks. “They were the ones who needed it the most. They were the ones getting it,” says Lalwani. “It’s rare to see these sorts of effects as clean as they look.”
Making the brain more inhibited made it more dynamic. “You might think that’s kind of counterintuitive. Shouldn’t it just shut the whole thing down?” says Garrett. “But that’s not really how the brain is built.”
Neural networks stay nimble by finding the right balance between inhibition and excitation. If GABA’s effect is too small and inhibition is too weak, the brain gets hyperexcited—neurons fire redundantly, and the network falls into a too-stable state from which it can’t easily stray. Enhancing the inhibitory effect of GABA brings the brain to a less excited, more flexible state. Altering the brain’s GABA levels has previously been shown to help against some symptoms of Alzheimer’s, like brain cell death and memory loss.
But while the study shows that the brain did indeed get noisier, and while noise is associated with better cognitive performance, Samanez-Larkin points out that one key piece of evidence is missing from the work: proof that a GABA-enhancing drug actually improves cognition. The team didn’t show that Lorazepam made poor performers do better on the tests later, or show how long the changes in brain noise level last. It’s also not clear if the drug affects the parts of the brain that need a boost in order to perform better. Lorazepam’s effect is diffuse. “It’s not like it goes to one specific part of the brain,” says Samanez-Larkin. “You could be basically helping certain parts of the brain but overdosing other parts of the brain.” A drug tested for real-world treatment would have to be more specific in its action than Lorazepam.
In fact, Lalwani guesses that subjects might actually perform worse after taking Lorazepam because the side effects from increasing GABA levels throughout the brain include drowsiness and memory loss. “This in no way means that people should be out there popping GABA pills,” Lalwani says.
That said, targeting GABA pathways doesn’t have to mean popping pills. Daily aerobic exercise can enhance those levels, too, and increase signal variability, according to Aga Burzynska, a cognitive neuroscientist at Colorado State University who was not involved in the work. “[Lalwani’s study] actually opens another avenue for studying what causes—or what are the ‘players’—in cognitive decline.” says Burzynska. She is especially interested in how this could help us better treat healthy aging. More than 50 million Americans are 65 or older. Most won’t be diagnosed with a cognitive disease, but they may still deal with normal decline. “That is the main reason for loss of independence and everyday life,” she says. “There’s a lot to win there.”
Neural noise looks like it may be a sort of biomarker for how the brain responds to GABA-enhancing drugs. A study from October linked irregular signal variability to attention deficit hyperactivity disorder, bipolar disorder, and borderline personality disorder. And new results from Garrett in a separate study show that signal variability predicts treatment outcomes for social anxiety disorders.
Having a metric like signal variability that could predict whether a psychiatric drug is working or not is a big deal. “That would be an incredible breakthrough,” says Grady. “People have struggled to understand who will respond to treatment, and who won’t.”
Garrett envisions a standardized fMRI test for signal variability that doctors could use to predict how well the brains of people with certain psychiatric conditions or Alzheimer’s might respond to drug treatments. It would consist of a minute-long task that a person would do while being scanned and software that would help a clinician make sense of the fMRI data. He imagines it being “just a canned, easy thing.”
Finding a drug to relieve cognitive decline by tinkering with neural noise is far off. And a clinical test that uses noise measurements to decide on treatments isn’t coming right away, either. Still, Lalwani says, their team feels they have the right concept and be narrowing in on the right neurotransmitter to decipher and control noise in the brain. “GABA seems to definitely be a promising target,” she says. “We know that we are moving in the correct direction.”
All Rights Reserved for Max G. Levy