Significant biomarkers for predicting depression in patients with Alzheimer’s disease

Abstract

AbstractBackgroundDepression is considered to be one the most common psychiatric co‐morbidities in Alzheimer’s disease (AD). Besides its prevalence, depression is associated with faster cognitive decline, physical aggression toward caregivers, earlier admission to nursing homes and lower quality of life. Therefore, early prediction of depression and selecting patients with higher risks for developing depression in the course of AD can provide an opportunity for earlier intervention and higher quality of life. This study’s primary intent was to explore the ability of feature selection algorithms to indicate baseline features that can predict the development of depression in AD patients.MethodWe have selected 28 AD patients from The Alzheimer’s Disease Neuroimaging Initiative (ADNI) that developed probable depression in 3‐5 years of follow‐up and 56 age‐ and sex‐matched AD patients without probable depression developed in the course of the disease. We implemented statistical analyses to identify the top significant markers offering the highest f values.ResultOur results indicated that P‐tau csf level, tau csf level, APOE4 genotype, ECogPtLang score (representing the interference of language impairment reported by the participant) and The Mini‐Mental State Exam (MMSE) score were the best features for predicting the development of depression in AD patients (f values: 7.38, 6.17, 3.96, 3.03 and 2.37, respectively). The first 3 markers possess p values < 0.05.ConclusionThese results provide primary evidence in favor of a distinguishable baseline profile in AD patients with higher risk for depression in the future. Therefore, caregivers might be able to classify AD patients prone to developing depression based on these features and provide earlier interventions.