Detection and Prediction of Mild Cognitive Impairment in Alzheimer's Disease Mice.

Background: The recent failure of clinical trials to treat Alzheimer's disease (AD) indicates that the current approach of modifying disease is either wrong or is too late to be efficient. Mild cognitive impairment (MCI) denotes the phase between the preclinical phase and clinical overt dementia. AD mouse models that overexpress human amyloid-β (Aβ) are used to study disease pathogenesis and to conduct drug development/testing. However, there is no direct correlation between the Aβ deposition, the age of onset, and the severity of cognitive dysfunction.

Objective: To detect and predict MCI when Aβ plaques start to appear in the hippocampus of an AD mouse.

Methods: We trained wild-type and AD mice in a Morris water maze (WM) task with different inter-trial intervals (ITI) at 3 months of age and assessed their WM performance. Additionally, we used a classification algorithm to predict the genotype (APPtg versus wild-type) of an individual mouse from their respective WM data.

Results: MCI can be empirically detected using a short-ITI protocol. We show that the ITI modulates the spatial learning of AD mice without affecting the formation of spatial memory. Finally, a simple classification algorithm such as logistic regression on WM data can give an accurate prediction of the cognitive dysfunction of a specific mouse.

Conclusion: MCI can be detected as well as predicted simultaneously with the onset of Aβ deposition in the hippocampus in AD mouse model. The mild cognitive impairment prediction can be used for assessing the efficacy of a treatment.