Mapping Alzheimer's

Advanced analysis of brain structure shape may track progression to dementia

A computational model of brain structures that form the basis of BrainPrint, a system for representing the whole brain based on the shape, rather than the size, of structures. Image: Martin Reuter and Christian Wachinger/Martinos Center for Biomedical Imaging, Mass General
A computational model of brain structures that form the basis of BrainPrint, a system for representing the whole brain based on the shape, rather than the size, of structures. Image: Martin Reuter and Christian Wachinger/Martinos Center for Biomedical Imaging, Mass General

A novel approach to analyzing brain structure that focuses on the shape, rather than the size, of particular features may allow identification of individuals who are in the early, pre-symptomatic stages of Alzheimer’s disease.

A team of Harvard Medical School investigators at Massachusetts General Hospital used advanced computational tools to analyze data from standard MRI scans.

They found that people with Alzheimer’s disease, including those diagnosed partway through a multiyear study, had greater levels of asymmetry in key brain structures: differences in shape between the left and right sides of the brain. Their study has been published in the journal Brain.

Get more HMS news here.

"Our results show for the first time that asymmetry of the hippocampus and amygdala increases with disease severity, above and beyond age-associated effects,” said Christian Wachinger, formerly an HMS research fellow in neurology at Mass General and lead author of the report. “By studying the progression of asymmetry from mild cognitive impairment to dementia, we demonstrated that greater asymmetry in those and a few other structures can predict disease progression and could be a biomarker allowing early detection of dementia.”

Wachinger is part of a team led by Martin Reuter, HMS assistant professor of radiology at Mass General, that developed BrainPrint, a computer-aided system for representing the whole brain based on the shape, rather than the size or volume, of structures. Originally described in a 2015 article in NeuroImage, BrainPrint appears to be as accurate as a fingerprint in distinguishing among individuals. In a recent paper in the same journal, Wachinger and Reuter demonstrated the use of BrainPrint for automated diagnosis of Alzheimer’s disease.

Structural asymmetries

The current study used BrainPrint to analyze structural asymmetries in a series of MR images of almost 700 participants in the National Institutes of Health-sponsored Alzheimer’s Disease Neuroimaging Initiative. Participation in that study involves MR brain imaging taken upon enrollment and repeated every 6 to 12 months, along with cognitive and genetic testing. The Mass General study analyzed data from participants who had had at least three MRI scans.

Participants were divided into four groups: those diagnosed with probable Alzheimer’s when entering the study, healthy controls with no sign of dementia, individuals with mild cognitive impairment that remained stable over the two to three years for which scans were available, and those with mild cognitive impairment that progressed to Alzheimer’s disease during the study.

BrainPrint analysis of the data revealed that initial, between-hemisphere differences in the shapes of the hippocampus and amygdala—structures known to be sites of neurodegeneration in Alzheimer’s disease—were highest in individuals with dementia and lowest in healthy controls. Among those originally classified with mild cognitive impairment, baseline asymmetry was higher in those that progressed to Alzheimer’s dementia and became even greater as symptoms developed. Increased asymmetry was also associated with poorer cognitive test scores and with increased cortical atrophy.

Several studies have indicated that Alzheimer’s has different effects in different substructures of the hippocampus and amygdala, said Reuter, the senior author of the Brain paper, explained.

Predicting future progression

“Since the shape descriptors of BrainPrint are more sensitive to subtle changes within a structure than are standard volume-based measures, they are better suited to quantify early disease effects and predict future progression, which opens up new research directions into the mechanisms that cause these asymmetries,” Reuter said. “For example, in addition to asymmetric distribution of amyloid beta, which has been reported, the differences could reflect disease subtypes that affect hemispheres differently.”

Now a professor of neurobiological research in the Department of Child and Adolescent Psychiatry, Psychosomatics, and Psychotherapy at Ludwig Maximilian University of Munich, Wachinger said that in collaboration with colleagues at Mass General, he is planning to explore further the relationship between shape asymmetries and established Alzheimer’s disease biomarkers to better understand the underlying biological mechanisms.

“Differentiating between those with stable mild cognitive impairment and those who will progress to Alzheimer’s is of great clinical relevance, as it could help select individuals appropriate for clinical trials of disease-modifying therapies,” Wachinger said.

Reuter is director of the Laboratory for Computational Longitudinal Neuroimaging at the Martinos Center at Mass General. He also holds a research affiliation at MIT.

Support for the study includes National Institutes of Health grant 1K25CA181632, Massachusetts Alzheimer’s Disease Research Center grant 5P50AG005134, and grants from the Humboldt Foundation, MGH Neurology Clinical Trials Unit, the Harvard NeuroDiscovery Center, and the Genentech Foundation.

Adapted from a Mass General news release.