P1‐269: DEVELOPMENT AND TECHNICAL VALIDATION OF A DATA‐INDEPENDENT ACQUISITION METHODOLOGY FOR ANALYSIS OF HUMAN ALZHEIMER'S DISEASE CEREBROSPINAL FLUID

Abstract

Current diagnosis of Alzheimer's disease (AD) relies primarily on cognitive assessments supported by imaging and a small number of cerebrospinal fluid (CSF) biomarkers. These limited biomarkers fail to capture the complexity of the disease and may be inadequate predictors of treatment response. Novel biomarkers are required to better diagnose and assess AD disease activity and treatments efficacy. Given its intimate connection with the brain's interstitial fluid, CSF is an ideal biofluid for AD biomarker discovery. In order to identify possible biomarkers for disease presence, progression, and treatment response, we are developing a data-independent acquisition (DIA) liquid chromatography mass spectrometry (LC-MS/MS) approach to monitor hundreds of peptides simultaneously. Human CSF samples were concentrated with a 3K filter and digested with Lys-C followed by trypsin. Samples were analyzed by LC-MS/MS on an Orbitrap Fusion using a DIA method with non-overlapping 25m/z windows. To develop a comprehensive and reproducible library to detect proteins of interest, we used Scaffold-DIA software and a fractionated human CSF library of 1500 proteins refined by four single-shot technical replicates. Stability and reproducibility of peptides were assessed across digests, between instrument runs, and between Scaffold and Skyline analytical software. In a cohort of 90 human CSF samples, spanning diagnostic criteria from cognitively unimpaired to AD dementia, our assay reliably quantified over 200 proteins and 2300 peptides with a 2 peptide per protein minimum (median 7 peptides per protein) with a false discovery rate less than 1%. This search was able to identify ∼90% of proteins in our manually curated list of AD proteins of interest. Quantification was highly correlated between Scaffold and Skyline for all commonly identified peptides from the two libraries and search methodologies. Identification of peptides between instrument re-runs was largely stable and within sample correlations were high. DIA methodology is poised to become a powerful tool for biofluid biomarker discovery in AD, but in order to build a robust method performance characteristics of individual peptides must first be carefully established.