Modeling MRI Brain Aging in Autosomal Dominant Alzheimer's Disease
MentorEric McDade, DO Washington University, School of Medicine
We will measure how “old” a person’s brain appears, and test whether these age predictions are older than expected in participants with a rare genetic mutation that causes Alzheimer disease.
Aim 1: We will train a machine learning model to predict age from functional connectivity MRI in participants with autosomal dominant Alzheimer disease (ADAD) mutations. We predict that age predictions from this model will change as ADAD progresses. Aim 2: We will apply a similar method as Aim 1, using volumetric MRI instead of functional connectivity. We predict that age predictions from this model will also change with ADAD progression. Aim 3: We will compare the age predictions from Aims 1 and 2 to established MRI markers of AD progression.
Brain-age gap (BAG) models may yield novel MRI biomarkers of AD progression, but most BAG studies have focused on sporadic Alzheimer disease (AD). Thus, applying BAG to preclinical autosomal dominant AD (ADAD) is a relatively novel approach. We will extend upon recent work in this area by using more comprehensive, multimodal MRI features, which capture more variance in healthy age differences. This approach also allows us to compare functional and structural age predictions to established MRI markers of AD progression. Brain age gap (BAG) estimates may reflect comprehensive, easily-interpretable, non-invasive biomarkers of brain health and may improve sensitivity to Alzheimer disease (AD) over established MRI measures (e.g., hippocampal volume). Thus, they may offer clinical utility as an AD screening or staging tool or as a clinical trial endpoint. This proposal aims to validate BAG by testing whether BAG patterns previously demonstrated in sporadic AD are consistently observed in autosomal dominant AD participants, who are younger and lack most confounding age-related pathologies.