Researchers have created a new mathematical method for analyzing standard brain scans to predict the onset of Alzheimer’s disease long before symptoms appear. By rating how closely an individual’s brain matches the structural patterns typical of the disease, this new tool can track the hidden impacts of genetic and cardiovascular risks in healthy adults. The research was published in the journal Molecular Psychiatry.
Alzheimer’s disease is the leading cause of cognitive decline in older adults. The gradual changes that lead to the condition begin in the brain decades before the first signs of memory loss or confusion. This long preclinical period provides a window of opportunity for medical interventions aimed at delaying or preventing the disease.
Existing methods for detecting these early changes often rely on specialized and expensive techniques. Doctors look for specific proteins using radioactive tracers in positron emission tomography scans, or they test spinal fluid and blood. While these methods are highly accurate, they can be invasive and are not always practical for screening the broader public.
Standard magnetic resonance imaging offers a noninvasive and widely available brain scanning alternative. However, typical visual signs of the disease on these common scans, such as the shrinking of specific brain folds or the widening of fluid cavities, are not very sensitive. These visible changes usually show up only after memory problems have already started.
A team of scientists led by Peter Kochunov and L. Elliot Hong at the University of Texas Health Science Center at Houston sought an earlier warning sign. They developed a software measurement called the Regional Vulnerability Index. This analytical tool evaluates the entire brain structure all at once.
To create the index, the researchers first established a universal blueprint of how the disease physically alters the brain over time. They analyzed brain scans from people diagnosed with the disease who had a confirmed buildup of toxic proteins. They compared these scans to those of healthy adults to map out the typical regional deficits across the organ.
The resulting index scores the mathematical similarity between any individual’s brain scan and the established disease blueprint. Instead of just looking at the size of the hippocampus, the brain’s central memory structure, the formula looks at widespread structural relationships. A higher score means an individual’s brain pattern closely resembles the pattern expected in dementia.
The researchers wanted to see if this index could capture the lifelong impacts of two major risk factors for cognitive decline. They looked at a gene associated with cholesterol transport, known as the apolipoprotein E gene. People who inherit the variant of this gene known as E4 have a significantly higher risk of developing dementia as they age.
They also examined cardiovascular health, which plays a major role in brain aging. Conditions like high blood pressure, high cholesterol, and diabetes can slowly damage the blood vessels that supply the brain with oxygen and nutrients. To measure this burden, the team calculated a standard cardiovascular risk score for each study participant.
The scientists tested their approach on two large groups of neurologically healthy adults. The goal was to see if the index could detect the influence of the genetic and cardiovascular risk factors before any cognitive issues developed. The first group served as an initial discovery sample to test the mathematical concept.
This initial group included 343 healthy adults from the Amish Connectome Project. This population is highly uniform in both genetics and environment. They share a rural farming lifestyle with very low rates of alcohol or tobacco use, which helps researchers isolate the effects of the specific disease risk factors being studied.
To replicate their findings, the team then evaluated a massive secondary sample from the UK Biobank dataset. This collection included more than 31,000 healthy participants from various urban and suburban environments. Testing the index across vastly different living conditions helped prove the robustness of the structural imaging measurement.
In both study groups, healthy adults who carried the high-risk gene variant had significantly higher brain index scores than noncarriers. Their brains were already displaying subtle structural patterns associated with the disease. This was true even though these participants had no outward neurological symptoms and performed normally on cognitive tests.
When the researchers checked individual brain structures the traditional way, they found very few structural differences between the people with the high-risk gene and those without it. Simple volume measurements of the memory centers or the outer folds of the brain did not reveal the underlying genetic risk. The mathematical index proved highly sensitive to hidden patterns that a basic volume check missed.
The study also revealed an interaction between genetic inheritance and cardiovascular health. In participants carrying the high-risk genetic variant, a higher cardiovascular risk score strongly correlated with an elevated brain index score. The two risk factors seemed to combine to push the brain’s physical structure closer to a disease-like state.
In contrast, higher cardiovascular risk scores did not significantly raise the brain index in participants without the genetic risk factor. The researchers noted that simply carrying the gene variant did not cause poor cardiovascular metrics like high blood pressure. Instead, poor cardiovascular health appeared to affect the brain much more severely in those with a localized genetic vulnerability.
After proving the index worked in healthy adults, the scientists investigated whether it could predict future cognitive decline in a higher-risk population. They utilized medical records from the Alzheimer’s Disease Neuroimaging Initiative. This long-term database tracks the neurological health of older adults over many years.
The researchers focused on nearly 2,000 older adults with an average age of 74. About half of these participants had mildly impaired cognition at the start of the assessment. Mild cognitive impairment represents a state of slight memory or thinking decline that often serves as a transitional stage between normal aging and full dementia.
The team tracked these participants for up to a decade. They found that individuals with mild cognitive impairment who eventually worsened and developed full dementia had significantly higher baseline index scores. The index successfully differentiated the patients who would experience rapid decline from those who would remain stable over the years.
The predictive power of the mathematical score was strongest in the short term. High baseline index scores reliably predicted new transitions to dementia within the first three years after the initial brain scan. As the time frame stretched further into the future beyond three years, the accuracy of predicting a patient’s outcomes based on a single baseline scan gradually declined.
Importantly, participants with mild cognitive impairment who did not progress to dementia had lower index scores. Their brain patterns were statistically similar to those of completely healthy older adults. A low index score suggested a much safer neurological trajectory for the decade ahead.
The researchers noted a few boundaries to their findings. The three participant groups had vastly different environmental backgrounds, which introduces some statistical noise into the data. While the Amish participants lived in rural settings with few harmful habits, the other groups represented modern urban populations with more typical health variations.
Additionally, the basic anatomical maps used to calculate the vulnerability index were standard medical tools. They were not explicitly built to highlight the exact regions of the brain that shrink during an abnormal protein buildup. By developing specialized structural maps in the future, scientists might be able to make the index even more sensitive to hidden tissue changes.
The team did not directly compare the accuracy of their index against the current standard screening tool, the positron emission tomography scan. Future studies will need to place these two screening methods head-to-head to determine which is more reliable. They will also need to test how the new brain index performs alongside advanced blood tests.
If validated by other laboratories, this mathematical approach could transform standard medical imaging for older adults. A routine hospital brain scan could be fed into a software program to evaluate a patient’s hidden neurological risks. Such noninvasive screening would help doctors find vulnerable patients early enough to offer preventative treatments before memory loss becomes permanent.
The study, “Alzheimer’s disease-like brain pattern biomarker: capturing risks and predicting disease onset,” was authored by Peter Kochunov, Si Gao, Lauren E. Salminen, Neda Jahanshad, Talia M. Nir, Paul M. Thompson, Xiaoming Du, Bhim M. Adhikari, Alice Kochunov, Ryan Cassidy, Yizhou Ma, Joshua Chiappelli, Seth Ament, Yezhi Pan, Shuo Chen, Alan R. Shuldiner, Braxton D. Mitchell, L. Jair Soares, and L. Elliot Hong.