Alzheimer's disease (AD) is the most common cause of dementia among the elderly and affects over 5 million Americans. Causes are not known and therapies are not optimal. Increasing evidence now suggests that the early pathogenesis of Alzheimer's disease is frequently characterized by progressive isolated memory deficits - referred to in research settings as amnestic mild cognitive impairment (MCI). It is important to study the different biochemical pathways that may be affected in AD and MCI. Metabolomics is a new but rapidly growing field that offers analytical instruments that can simultaneously quantitate thousands of substances present in a biological sample of interest, such as cerebrospinal fluid (CSF). In this collaborative project between Duke, University of Pennsylvania and centers of excellence in metabolomics and informatics we plan to identify differences in banked CSF of a well characterized cohort of 90 subjects (30 AD patients, 30 MCI and 30 controls). This cohort is unique since thorough baseline testing is already completed and the cohort will have been followed for about two years before this funding period starts. Our long-term goal is to then confirm these findings using the national multicenter ADNI biobank and clinical data. Future studies will connect these central changes with peripheral metabolic changes to enable identification of biomarkers that can be measured easily in plasma and that are also disease specific.