Paper

Individual-specific functional connectivity shows improved performance in detecting and predicting individualized symptoms of Alzheimer's disease in elderly people with/without APOE ε4 allele

Zhao, Yonghua, Gao, Fei, Guo, Qiwei, Yuan, Zhen, Lin, Hua, Xia, Xiaoluan

2022

Overview

To date, no reliable biomarkers are available that link individual-specific functional connectivity and patients' individualized symptoms for early detection and prediction of Alzheimer's disease (AD) in elderly people with specific genotypes. Meanwhile, functional magnetic resonance imaging (MRI) and machine learning are promising tools that can reveal the relationships between brain and behavior at individual level towards predicting the transition to AD. In this study, individual-specific functional connectivity was constructed in elderly participants with Apolipoprotein E (APOE) ε4 allele (N = 120) and without APOE ε4 allele (N = 115), respectively. In particular, machine learning based on a recursive feature selection technique was carried out to track multiple clinical symptoms among differing genotypes at individual level from normal aging (NA) and AD. It was found that the captured neuroimaging features in both APOE genotyping groups were able to distinguish the changes of clinical symptoms from NA to AD. Besides, our findings illustrated that the connections between individual-specific functional regions exhibited significantly higher correlation between estimated and observed scores in multiple clinical symptoms than those from atlas-based functional connectivity for both APOE genotyping groups, while no significant performance was detected when the data of two APOE genotyping groups were combined for the estimation models. Further, individual-specific between-network connectivity constitutes a major contributor for accessing cognitive symptoms in both APOE genotyping groups. Therefore, this study demonstrated the essential role of individual variation in cortical functional anatomy and the significance in combining brain and behavior for improving the accuracy in detection and prediction of AD in elderly people with specific genotypes.Competing Interest StatementThe authors have declared no competing interest.

• Publisher: Cold Spring Harbor Laboratory Press,Cold Spring Harbor Laboratory
• Subject: Aging / Alleles / Alzheimer's disease / Apolipoprotein E / Brain mapping / Cognitive ability / Functional anatomy / Functional magnetic resonance imaging / Genotypes / Genotyping / Geriatrics / Learning algorithms / Machine learning / Neural networks / Neurodegenerative diseases / Neuroimaging / Neuroscience / Older people