Measurements of Brain Age May Help Target Therapy

The genesis of modern brain aging research can be traced to a 2010 study by researchers at the University of Jena in Germany. They used over 600 MRI scans taken from healthy adults aged 18 to 65 to build a composite portrait of a typical adult brain across the lifespan. The researchers next developed an algorithm that could use this composite data to assess an anonymous brain scan and guess how old that person was at the time of the scan. They found that the computer algorithm was fairly accurate—the program came within five years of accurately predicting the age of every anonymous sample.

The Jena researchers next tested the algorithm on a set of MRI scans taken from people with mild Alzheimer’s disease. This analysis showed that a person’s actual biological age was about 10 years younger than their predicted age based on brain MRI data; this difference was termed the brain age gap estimate, or Brain AGE.

The initial excitement over Brain AGE was centered around the use of this tool to identify individuals at risk of dementia before symptoms appeared. Soon, this technique was applied to other areas in mental health. For example, studies found that smoking and drinking accelerated the brain aging process. Likewise, people with schizophrenia had a Brain AGE that was about eight years older than their biological age.

“From a clinical standpoint, boiling down a complex brain into one number is a powerful way to communicate with patients about their disorder,” said Lisa Eyler, Ph.D., a professor in residence of psychiatry at the University of California, San Diego. Eyler studies the effects of bipolar disorder on brain physiology.

For researchers, however, these differences in brain and biological age raise the question of what these numbers reflect. “Is schizophrenia-related brain aging due to lifestyle features or part and parcel with the disorder?” Eyler posed. For instance, while cigarette smoking and obesity may contribute to brain aging in people with schizophrenia, the results of a study published last year suggest there may also be something about the disorder itself that is accelerating brain aging.

That study assessed MRI scans specifically from younger adults aged 18 to 35 who recently experienced their first episode of psychosis; previous research used data from adults of all ages. Even among these young people with a recent onset of symptoms, their Brain AGE was about eight years older compared with that of healthy adults, just as in previous studies.

That same study also looked at young adults with recently diagnosed bipolar disorder. The analysis did not find any significant Brain AGE differences between healthy young adults and those with bipolar disorder, but the sample size was small. Eyler noted that with algorithms like the one that predicts Brain AGE, having robust data to scour is critical.

In her studies, Eyler has looked at specific features of the brain and found several age-associated changes in bipolar patients relative to similarly aged controls. For example, white matter (the nerve bundles that transmit information across the brain) seems to degrade more rapidly over time in people with bipolar disorder. Her group has collected comprehensive MRI data on over 100 bipolar patients, but that is not enough to quantify age differences.

Eyler recently joined ENIGMA, an international consortium that looks to merge imaging and genetics data to better understand brain diseases. ENIGMA researchers have also become interested in Brain AGE and recently analyzed 77 aging biomarkers in a sample of over 8,000 MRI scans from people with and without major depression. This large dataset enabled them to uncover that people with depression have a Brain AGE about 1.5 years older than that of a similarly aged healthy person on average. The study also found that people with a greater difference between their biological and brain ages had greater depressive symptoms on average.

Eyler is part of a team conducting a similar large-scale analysis in bipolar patients. “I think we will find that bipolar disorder will fall in between depression and schizophrenia in terms of aging effects,” she said. Of more interest to her is how different symptom histories relate to aging differences. That might identify what aspects of bipolar disorder are the most damaging, Eyler said. “Is it the total accumulation of manic episodes, the severity of depressive episodes, or even how frequently someone cycles back and forth between depression and mania?”

That answer might help identify which patients are at higher risk of long-term cognitive problems. Identifying patients with higher Brain AGE numbers can also help physicians provide more tailored treatments aimed at reducing that number.

“The promising trend we are seeing about Brain AGE is that it’s dynamic—this is not a number written in stone,” Paulus said. Studies have shown that lifestyle changes like quitting smoking or losing weight can lower Brain AGE just as they can do for a measure like blood pressure. Paulus’s lab recently completed a pilot study that showed that a single dose of ibuprofen lowered the average brain age of healthy participants by 1.1 years after 45 minutes. That finding supports the idea that inflammation plays some role in accelerated brain aging while also showing pharmacology can help reduce brain aging in addition to lifestyle changes.

“[R]ight now Brain AGE can give you an indication of a brain’s health in the moment, but it doesn’t tell us how someone got to that moment,” he said.More prospective studies that track individuals over long periods and take numerous MRI scans along the way will help paint a more fluid picture of how a brain ages over time. Since many psychiatric disorders emerge by young adulthood, Brain AGE studies would require data from children and adolescents in addition to adults.

As Brain AGE programs acquire more MRI data and become more sophisticated, Eyler stressed it’s important that the algorithm remain “explainable.”

“Many of these machine learning programs are quite complex and can collate tremendous amounts of data to find a pattern, but when you look under the hood, you can’t figure out how you got the answer,” she said. An explainable algorithm means that programmers can easily change one variable and see how the prediction changes.

“It’s similar to applying for a loan,” she said. “Bank programs process a lot of information to determine if you qualify, but they can also tell you the one or two things you need to change to turn a no into a yes.” Eyler envisions a similar Brain AGE program in which individuals would know which of their regions are the most aged and what specific intervention would work for them.

“A Brain AGE measurement is applicable to everyone, not just people with psychiatric disorders,” Paulus said. “We can consider it a marker of general health. I could see these programs making their way to clinics in the near future.” ■