Serum Metabolites Differentiate Amnestic Mild Cognitive Impairment From Healthy Controls and Predict Early Alzheimer's Disease via Untargeted Lipidomics Analysis

Abstract

Background and Aim: Alzheimer's disease (AD) is the most common type of dementia and presents with metabolic perturbations early in the disease process. In order to explore biomarkers useful in predicting early AD, we compared serum metabolites among patients suffering different stages of AD. Methods: We recruited 107 participants including 23 healthy controls (HC), 21 amnestic mild cognitive impairment (aMCI), 24 non-amnestic mild cognitive impairment (naMCI) and 39 AD patients. Via liquid chromatography-mass spectrometry based serum untargeted lipidomics analysis, we compared differences in serum lipid metabolites among these patient groups and further elucidated biomarkers that differentiate aMCI from HC. Results: There were significant differences of serum lipid metabolites among the groups, and 20 metabolites were obtained under negative ion mode from HC and aMCI comparison. Notably, 16:3 cholesteryl ester, ganglioside GM3 (d18:1/9z-18:1) and neuromedin B were associated with cognition and increased the predictive effect of aMCI to 0.98 as revealed by random forest classifier. The prediction model composed of MoCA score, 16:3 cholesteryl ester and ganglioside GM3 (d18:1/9z-18:1) had good predictive performance for aMCI. Glycerophospholipid metabolism was a pathway common among HC/aMCI and aMCI/AD groups. Conclusion: This study provides preliminary evidence highlighting that 16:3 cholesteryl ester were useful for AD disease monitoring while ganglioside GM3 (d18:1/9z-18:1) and neuromedin B discriminated aMCI from HC, which can probably be applied in clinic for early predicting of AD.