Attributions

Identifying Groups of Alzheimer's Disease Patients with Slower Disease Progression

Justin Miller, PhD Brigham Young University

Mentor

John S.K. Kauwe, PhD

Summary

Millions of Americans watch themselves or their loved ones lose their ability to retain memories as they advance through the devastating stages of Alzheimer's disease with little guidance on how quickly the disease will progress. We carefully designed an approach to use machine learning to group individuals with similar health trajectories based on their genetics, clinical tests, and neuroimages, and we will use these subtypes to assess differences in the rate of cognitive decline, the age of disease onset, and the age of death for each proposed subtype using a longitudinal dataset spanning 20 years. We anticipate that identifying Alzheimer's disease subtypes will allow future studies to improve diagnoses for patients, identify subtype-specific drug targets, calculate disease trajectories for each subtype, focus clinical trials on specific subtypes, and eventually develop subtype-specific treatment plans. By better understanding differences in Alzheimer's disease, we will provide patients and caregivers with more information about the underlying causes of the disease and the projected health outcomes associated with cognitive decline. In a disease with so many unknowns, having clearer health trajectories and diagnoses gives all of us with hope that we might find a cure for at least some patients, and immediately provides patients and caregivers currently dealing with the effects of Alzheimer's disease with more information to make informed end-of-life decisions.

Project Details

We aim to identify Alzheimer's disease subtypes with different health trajectories. First, we will use machine learning to cluster Alzheimer's disease patients based on the results of common lab tests such as blood draws and MRIs. We will use several datasets to ensure that the identified clusters are broadly applicable. Next, we will link genetic variants to each cluster so that we can predict the odds of a person belonging to a given cluster based solely on their DNA. We will assess the genetic basis of each cluster to evaluate the original classifications that were made based on common lab tests. After we are confident that we have identified Alzheimer's disease clusters with unique genetic signatures, we will determine the rate at which Alzheimer's disease progresses in people belonging to each cluster. We will also evaluate the age at onset and other health markers that may precede the development of Alzheimer's disease within each cluster. Currently, most Alzheimer's disease research focuses on Alzheimer's disease as a single disease with many different causes, although some distinct clusters have recently emerged that indicate Alzheimer's disease may be more complex than previously thought. Our research takes an innovative statistical approach to determine if the different causes of Alzheimer's disease result in different disease trajectories that may require a different mode of treatment. We anticipate that this research will improve our understanding of Alzheimer's disease, provide patients and caregivers with more information about the trajectory of their specific Alzheimer's disease subtype, and facilitate the future development of subtype-specific treatments.