Alzheimer's disease (AD) is a neurodegenerative disorder lacking early biochemical diagnosis and treatment. Lipids have been implicated in neurodegenerative disorders including AD. A shotgun lipidomic approach was undertaken to determine if lipid biomarkers exist that can discriminate AD cases from controls. The discovery study involved sera from 29 different stage AD cases and 32 controls. Lipid extraction was performed using organic solvent and the samples were directly infused into a time-of-flight mass spectrometer. Differences between AD cases and controls were detected with 87 statistically significant lipid candidate markers found. These potential lipid markers were reevaluated in a second confirmatory study involving 27 cases and 30 controls. Of the 87 candidates from the first study, 35 continued to be statistically significant in the second confirmatory set. Tandem MS studies were performed and almost all confirmed markers were characterized and classified. Using a Bayesian lasso probit regression model on the confirmed markers, a multi-marker set with AUC = 0.886 was developed comparing all stages of AD with controls. Additionally, using confirmed biomarkers, multi-marker sets with AUCs >0.90 were developed for each specific AD Clinical Dementia Rating versus controls, including the earliest stage of AD. More conservative and likely more realistic statistical analyses still found multi-marker sets that appeared useful in diagnosing AD. Finally, using ordinal modeling a set of markers was developed that staged AD accurately 70% of the time, p = 0.0079. These results suggest that these serum lipidomic biomarkers may help diagnose and perhaps even stage AD.
Keywords: Alzheimer’s disease; biomarkers; diagnosis; disease staging; lipidomics; mass spectrometry.