Toward a predictive model of Alzheimer's disease progression using capillary electrophoresis-mass spectrometry metabolomics.

Alzheimer's disease (AD) is the most prevalent form of dementia with an estimated worldwide prevalence of over 30 million people, and its incidence is expected to increase dramatically with an increasing elderly population. Up until now, cerebrospinal fluid (CSF) has been the preferred sample to investigate central nervous system (CNS) disorders since its composition is directly related to metabolite production in the brain. In this work, a nontargeted metabolomic approach based on capillary electrophoresis-mass spectrometry (CE-MS) is developed to examine metabolic differences in CSF samples from subjects with different cognitive status related to AD progression. To do this, CSF samples from 85 subjects were obtained from patients with (i) subjective cognitive impairment (SCI, i.e. control group), (ii) mild cognitive impairment (MCI) which remained stable after a follow-up period of 2 years, (iii) MCI which progressed to AD within a 2-year time after the initial MCI diagnostic and, (iv) diagnosed AD. A prediction model for AD progression using multivariate statistical analysis based on CE-MS metabolomics of CSF samples was obtained using 73 CSF samples. Using our model, we were able to correctly classify 97-100% of the samples in the diagnostic groups. The prediction power was confirmed in a blind small test set of 12 CSF samples, reaching a 83% of diagnostic accuracy. The obtained predictive values were higher than those reported with classical CSF AD biomarkers (Aβ42 and tau) but need to be confirmed in larger samples cohorts. Choline, dimethylarginine, arginine, valine, proline, serine, histidine, creatine, carnitine, and suberylglycine were identified as possible disease progression biomarkers. Our results suggest that CE-MS metabolomics of CSF samples can be a useful tool to predict AD progression.