Analysis of nearly 50,000 brain scans has revealed five different patterns of brain atrophy linked to aging and neurodegenerative diseases. The analysis also linked the patterns to lifestyle factors such as smoking and alcohol consumption, as well as to genetic and blood-based markers associated with health status and disease risk.

The work is a "methodological masterpiece" that could significantly advance researchers' understanding of aging, says Andrei Irimia, a gerontologist at the University of Southern California in Los Angeles who was not involved in the work. "Before this study, we knew that brain anatomy changes with aging and disease. But our ability to capture this complex interaction was far more modest."

The study was conducted on August 15 inNature Medicinepublished.

Wrinkles in the brain

Aging can cause not only gray hair but also changes in brain anatomy that are visible on magnetic resonance imaging images, with some areas shrinking or undergoing structural changes over time. However, these transformations are subtle. “The human eye is unable to perceive patterns of systematic brain changes” associated with this degradation, says Christos Davatzikos, a biomedical imaging specialist at the University of Pennsylvania in Philadelphia and an author of the paper.

Previous studies have shown that machine learning can extract the subtle fingerprints of aging from MRI data. However, these studies were often limited in scope and usually included data from a relatively small number of people.

To identify broader patterns, Davatzikos' team began a study that took about eight years to complete and publish. They used a deep learning method called Surreal-GAN, which was developed by first author Zhijian Yang while he was a student in Davatzikos' lab. The scientists trained the algorithm using brain MRIs from 1,150 healthy people ages 20 to 49 and 8,992 older adults, including many with cognitive impairments. This taught the algorithm to recognize recurring features of aging brains, allowing it to build an internal model of anatomical structures that change simultaneously, as opposed to those that tend to change independently.

The researchers then applied the resulting model to MRI scans of nearly 50,000 people participating in various studies on aging and neurological health. This analysis provided five discrete patterns of brain atrophy. The scientists linked different types of age-related brain degeneration to combinations of the five patterns, although there were some differences between people with the same condition.

Patterns of aging

For example, dementia and its precursor, mild cognitive impairment, had links to three of the five patterns. Interestingly, the researchers also found evidence that the patterns they identified could potentially be used to reveal the likelihood of further brain degeneration in the future. “If you want to predict the transition from a cognitively normal state to mild cognitive impairment, one thing was the most prescient,” says Davatzikos. “In later stages, adding a second [pattern] enriches your prediction, which makes sense because this captures the spread of the pathology.” Other patterns were associated with diseases such as Parkinson's and Alzheimer's, as well as a combination of three patterns that were strongly predictive of mortality.

The authors found clear links between specific patterns of brain atrophy and various physiological and environmental factors, including alcohol consumption and smoking, as well as various genetic and biochemical signatures associated with health. Davatzikos says these results likely reflect the impact of general physical well-being on neurological health, as damage to other organ systems can have consequences for the brain.

However, Davatzikos cautions that the study “does not mean that everything can be reduced to five numbers,” and his team intends to work with datasets that include a broader range of neurological diseases and have greater ethnic and cultural diversity.