Identification of longitudinally dynamic biomarkers in Alzheimer's disease cerebrospinal fluid by targeted proteomics

Background: Alzheimer's disease (AD) is the leading cause of dementia affecting greater than 26 million people worldwide. Although cerebrospinal fluid (CSF) levels of A beta(42), tau, and p-tau(181) are well established as diagnostic biomarkers of AD, there is a need for additional CSF biomarkers of neuronal function that continue to change during disease progression and could be used as pharmacodynamic measures in clinical trials. Multiple proteomic discovery experiments have reported a range of CSF biomarkers that differ between AD and control subjects. These potential biomarkers represent multiple aspects of the disease pathology. The performance of these markers has not been compared with each other, and their performance has not been evaluated longitudinally. Results: We developed a targeted-proteomic, multiple reaction monitoring (MRM) assay for the absolute quantitation of 39 peptides corresponding to 30 proteins. We evaluated the candidate biomarkers in longitudinal CSF samples collected from aged, cognitively-normal control (n = 10), MCI (n = 5), and AD (n = 45) individuals (age > 60 years). We evaluated each biomarker for diagnostic sensitivity, longitudinal consistency, and compared with CSF A beta(42), tau, and p-tau181. Four of 28 quantifiable CSF proteins were significantly different between aged, cognitively-normal controls and AD subjects including chitinase-3-like protein 1, reproducing published results. Four CSF markers demonstrated significant longitudinal change in AD: Amyloid precursor protein, Neuronal pentraxin receptor, NrCAM and Chromogranin A. Robust correlations were observed within some subgroups of proteins including the potential disease progression markers. Conclusion: Using a targeted proteomics approach, we confirmed previous findings for a subset of markers, defined longitudinal performance of our panel of markers, and established a flexible proteomics method for robust multiplexed analyses.