Establishing an individualized model of conversion from normal cognition to Alzheimer's disease after 4 years, based on cognitive, brain morphology and neuropsychiatric characteristics

Objectives: The impact of neuropsychiatric symptoms (NPS) on cognitive performance has been reported, and this impact was better defined in the aging population. Yet the potential of using the impact of NPS on brain and cognitive performance in a longitudinal setting, as prediction of conversion - have remained questionable. This study proposes to establish a predictive model of conversion to Alzheimer's disease (AD) and mild cognitive impairment (MCI) based on current cognitive performance, NPS and their associations with brain morphology.

Methods: 156 participants with MCI from the Alzheimer's Disease Neuroimaging Initiative database cognitively stable after a 4-year follow-up were compared to 119 MCI participants who converted to AD. Each participant underwent a neuropsychological assessment evaluating verbal memory, language, executive and visuospatial functions, a neuropsychiatric inventory evaluation and a 3 Tesla MRI. The statistical analyses consisted of 1) baseline comparison between the groups; 2) analysis of covariance model (controlling demographic parameters including functional abilities) to specify the variables that distinguish the two subgroups and; 3) used the significant ANCOVA variables to construct a binary logistic regression model that generates a probability equation to convert to a lower cognitive performance state.

Results: Results showed that MCI who converted to AD in comparison to stable MCI, exhibited a higher NPS prevalence, a lower cognitive performance and a higher number of involved brain structures. Functional abilities, memory performance and the sizes of inferior temporal, hippocampal and amygdala sizes were significant predictors of MCI to AD conversion. We also report two models of conversion that can be implemented on an individual basis for calculating the percentage risk of conversion after 4 years.

Conclusion: These analytical methods might be a good way to anticipate cognitive and brain declines.