A significant challenge in treating Alzheimer’s disease (AD) is that the disease progresses silently for years before memory loss or other symptoms appear. By that time, the opportunity to prevent or slow its progress is limited. To improve interventions and treatments, we need a better method for assessing risk and predicting how the disease will unfold—a dementia “forecast,” so to speak.
Like meteorologists who analyze temperature, pressure, and other patterns to predict the weather, researchers are developing ways to use data from brain scans, spinal fluid, and memory assessments to forecast disease development. However, one challenge for researchers is that these tests are often collected at different times across individuals, which means the results are a mix of normal aging, different disease stages, and individual differences. This creates a fragmented picture of how the disease progresses, making it difficult to piece together how the disease unfolds.
To overcome this, Drs. Wang and Caffo have developed a Temporally Aligned Multimodal (TAM) framework that places each person’s results on a personalized disease timeline that separates disease progression from aging. By incorporating multiple biomarkers and other genetic information about individuals, Drs. Wang and Caffo have shown that they can align biomarker changes with a more accurate “disease time” measure than simply using factors such as age or time since diagnosis. While this TAM framework can estimate disease timelines, it was not designed for clinical risk assessment or prognosis. Here, Drs. Wang and Caffo intend to expand the capabilities of their TAM framework in the hope that it can be clinically used to support personalized Alzheimer’s care.
In this project, Drs. Wang and Caffo will take the next step with their TAM framework by linking disease progression biomarkers to clinical diagnoses, cognitive tests, and other functional assessments. They will use data from three different cohort studies—BIOCARD (Biomarkers of Cognitive Decline Among Normal Individuals), ADNI (Alzheimer’s Disease Neuroimaging Initiative), and PAC (Preclinical Alzheimer’s Consortium)—to build and validate their model. The project consists of three primary aims. In the first, they will connect individual biomarker data to clinical diagnoses (such as MCI or AD) and clinical assessments. They will then test their model’s ability to predict future clinical manifestations of disease using the integrated biomarker data. By testing their framework against patients who have already been diagnosed and have several years of data collected, Drs. Wang and Caffo can ensure it functions as intended. In their second aim, the researchers will identify patient subgroups within their TAM framework whose disease progression is slower or faster than typically seen and isolate the risk or resilience variables that drive these differences. This aim illustrates the versatility of their framework beyond risk assessment and disease prognoses. Identifying resilience factors could prompt future studies into therapies and guide treatment plans if the factors are linked to modifiable lifestyle risk factors. In the final aim, they will utilize all the analyses from Aims 1 and 2 to develop a “deep learning” model that can 1) enable accurate prognosis of AD progression and cognitive decline, and 2) uncover biomarker signatures that could point to new mechanisms of disease progression.
The goal of this project is to develop an accurate, personalized prognosis of AD and cognitive decline, and to identify new biomarkers that provide insight into how the disease progresses. Both of these things will advance personalized care strategies to guide future therapies.
