Proteome profiling in cerebrospinal fluid reveals novel biomarkers of Alzheimer's disease

Neurodegenerative diseases are a growing burden, and there is an urgent need for better biomarkers for diagnosis, prognosis, and treatment efficacy. Structural and functional brain alterations are reflected in the protein composition of cerebrospinal fluid (CSF). Alzheimer's disease (AD) patients have higher CSF levels of tau, but we lack knowledge of systems-wide changes of CSF protein levels that accompany AD. Here, we present a highly reproducible mass spectrometry (MS)-based proteomics workflow for the in-depth analysis of CSF from minimal sample amounts. From three independent studies (197 individuals), we characterize differences in proteins by AD status (> 1,000 proteins, CV < 20%). Proteins with previous links to neurodegeneration such as tau, SOD1, and PARK7 differed most strongly by AD status, providing strong positive controls for our approach. CSF proteome changes in Alzheimer's disease prove to be widespread and often correlated with tau concentrations. Our unbiased screen also reveals a consistent glycolytic signature across our cohorts and a recent study. Machine learning suggests clinical utility of this proteomic signature.