Project Summary:
Developing means to identify Alzheimer’s disease (AD)-dementia at an early stage is a burning medical health priority. Even though neuropathologic signs of AD (i.e. amyloid and tau pathophysiology) arise years before cognitive decline, they are not good predictors of dementia because many people who present with these signs never experience major cognitive decline. Identifying early signs of AD-related cognitive dysfunction and determining how some neuropathology-affected brains manage to compensate for the pathophysiology is therefore crucial to forecasting dementia. A critical barrier obstructing progress is that there is no established way to identify early dysfunction at the neural level. This project will initiate work on a new approach to estimate neural dysfunction using direct measures of neural information processing in story understanding.
Developing methods to diagnose early signs of dementia targets the second focus area of the Wisconsin State Dementia Plan, namely “Improving how health care providers diagnose and care for people with dementia.” Benefits of early diagnosis, include enabling the early opportunity to administer drugs that may change disease progression or help to mitigate symptoms of memory-loss or confusion. Secondly, early diagnoses can give people extra time to plan to manage their condition and make financial, legal, caregiving and family arrangements for the future, in case they lose this ability. At the state-level diagnostic advances may help reduce the estimated $924 million cost of Alzheimer’s to the Medicaid program and estimated $5.5billion in an estimated 297 million hours of unpaid care provided by 205,000 family caregivers who bear the burden of AD in Wisconsin.
In seeking to introduce brain-based indices to help forecast Alzheimer’s-dementia, this proposal targets people at risk of Alzheimer’s dementia. Demographically, age is the biggest risk factor for AD, and in Wisconsin the Alzheimer’s Association estimates that 10.6% of adults (111,000 people) over 65 live with AD in 2020. Furthermore, it has been estimated that two-thirds of Americans with Alzheimer’s are female, and Black and Hispanic adults are more likely develop Alzheimer’s (prevalence in over 65s is 19% Black, 14% Hispanic, 10% White).
The team’s long-term vision is to forecast dementia at an early stage by distinguishing brains with high pathophysiology (amyloid and tau) that will remain cognitively intact from those that will succumb to dementia. To do this, they hope to identify when brain networks successfully “rewire” to accommodate local amyloid and tau. They propose to leverage recent language AI models to map semantic information processing networks during fMRI scans of story comprehension.
Processing semantics is reliant on long-term memories that become damaged in Alzheimer’s disease in relation to a buildup of amyloid and tau in associated brain regions. The new AI-based methods may provide a means to evaluate how well the diseased brain regions encode meaning and critically whether brain networks rewire to enable less diseased brain regions to take on the role of diseased regions.
The team’s immediate goal is to validate the fMRI language mapping methods on persons at risk from AD before they can move on to relate semantic brain maps to maps of amyloid and tau, which require collection of expensive PET scans.
This project will provide a foundation for a future longitudinal study forecasting cognitive health by combining brain maps of story processing with brain maps of PET amyloid and tau. The team hypothesizes that cognitive impairment will be reflected by distinctive patterns of information loss in brain regions that are vulnerable to early-stage pathophysiology and that the new fMRI measures will better estimate cognitive health than blood amyloid and tau measures.
Collaborator: Stanford University