Recent biomarker research on neurodegenerative diseases has focused on combinations of multiple risk factors to predict development of cognitive impairment and dementia. We combine (1) global brain atrophy as measured by brain parenchymal fraction (BPF), (2) leading AD genetic risk marker (APOE), and (3) commonly examined AD biomarkers to understand the influence of risk factors in relation to differential patterns of atrophy in AD.  

We used a ~2-year longitudinal sample (followed annually) of diagnosed AD patients (baseline N=187; mean age=70.57 years; range=37-89 years) from the Sunnybrook Dementia Study. We used (1) latent growth modeling and class analyses to estimate the number of classes that best represented BPF ([(normal appearing gray matter +normal appearing white matter +white matter hyperintensities)/(total intracranial volume) x 100]) over 2 years, and (2) Random Forest Analysis to test relative predictive importance of AD biomarkers from six different domains (demographic, dementia history, IQ, depressive symptoms, clinical dementia severity, cognitive markers) on BPF in the whole sample and as stratified by APOE ?4 status.