Mixture Disease Progression Model to Predict and Cluster the Long-Term Trajectory of Cognitive Decline in Alzheimer's Disease.

Abstract

Background: Alzheimer's disease (AD) is a neurodegenerative disease for which many clinical trials failed to detect treatment effects, possibly due to the heterogeneity of disease progression among the patients. Predicting and clustering a long-term trajectory of cognitive decline from the short-term cognition data of individual patients would help develop therapeutic interventions for AD.

Methods: This study developed mixture disease progression model to predict and cluster the long-term trajectory of cognitive decline in the population. We predicted the 30-year long-term trajectories of the three cognitive scales and categorized the individuals into rapid and slow cognitive decliners by applying the method, which was based on the two-component normal mixture nonlinear mixed-effects model, to the short-term follow-up data of the Mini-Mental State Examination, the 13-item Alzheimer's Disease Assessment Scale-Cognitive, and the Clinical Dementia Rating Scale-sum of boxes collected in patients with mild cognitive impairment and AD in the Alzheimer's Disease Neuroimaging Initiative.

Results: For each cognitive scale, the models identified two distinct subpopulations, including a population of comprising approximately 10-20% of individuals experiencing rapid cognitive decline, wherein the posterior means of the differences in cognitive decline speed between the two groups ranged from 2 to 3 years. We also identified baseline background factors associated with rapid decliners for three cognitive scales.

Conclusion: Identifying the risk factors associated with rapid decline of cognition by the proposed method aids in planning eligibility criteria and allocation strategy for accounting for the varying disease progression speeds among the patients enrolled in clinical trials for AD.