Year: 2014
Type: article
Abstract: In a landmark 2011 paper, Clifford Jack and coauthors hypothesized long-term progression curves of biomarkers, cognition, function, and memory. These curves depict the Alzheimer's Disease cascade with each marker of disease progressing in turn. Some progress has been made towards applying short-term data to estimate these long-term curves. We build on these models by incorporating covariate effects so that we can explore different subpopulations and patterns of progression. We apply the proposed method to data from the Alzheimer's Disease Neuroimaging Initiative and the Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBL). Donohue and coauthors (2014) proposed a self-modeling regression approach to estimate long-term progression curves from short-term data. The iterative algorithm estimates subject-specific time shifts that spread the short-term data over a longer interval. Long-term progression curves are then estimated over this longer interval, and the time shifts themselves are updated based on the progression curve estimates. Iteration continues until convergence. We extend this algorithm to include nonlinear effects on each outcome for age, gender, education, APOE4 genotype, A β, tau, and their interactions. Covariates are selected for each outcome at each stage of the iteration based on the Akaike Information Criterion using a generalized additive model approach. Outcome measures considered include assessments (ADAS13, MMSE, FAQ, and RAVLT), MRI brain volumetrics (hippocampus, ventricles, and entorhinal cortex), CSF (A β, tau, p-tau), and PET (PiB, Florbetapir, and FDG). Similar attempts at this estimation have stratified by covariates or ignored them altogether. Our current approach will allow closed form, as opposed to bootstrap, estimation of important covariate effects for each outcome. Covariate adjustment for age is particular important since our progression curve estimates span 20 years. The proposed method provides improved estimates of long-term progression curves, interrogation of covariate effects, and inspection of different patterns of progression.
Source: Alzheimer s & Dementia
Language: English
Subfield: Psychiatry and Mental health
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2085685392