Identification of Novel Fluid Biomarkers for Alzheimer's Disease.

摘要

Clinicopathological studies suggest that Alzheimer's disease (AD) pathology begins to appear ∼10--20 years before the resulting cognitive impairment draws medical attention. Biomarkers that can detect AD pathology in its early stages and predict dementia onset and progression would, therefore, be invaluable for patient care and efficient clinical trial design. To discover such biomarkers, we measured AD-associated changes in the cerebrospinal fluid (CSF) using an unbiased proteomics approach (two-dimensional difference gel electrophoresis with liquid chromatography tandem mass spectrometry). From this, we identified 47 proteins that differed in abundance between cognitively normal (Clinical Dementia Rating [CDR] 0) and mildly demented (CDR 1) subjects. To validate these findings, we measured a subset of the identified candidate biomarkers by enzyme linked immunosorbent assay (ELISA); promising candidates in this discovery cohort (N=47) were further evaluated by ELISA in a larger validation CSF cohort (N=292) that contained an additional very mildly demented (CDR 0.5) group. Levels of four novel biomarkers were significantly altered in AD, and Receiver-operating characteristic (ROC) analyses using a stepwise logistic regression model identified optimal panels containing these markers that distinguished CDR 0 from CDR>0 (tau, YKL-40, NCAM) and CDR 1 from CDR1 (tau, chromogranin-A, carnosinase-I). Plasma levels of the most promising marker, YKL-40, were also found to be increased in CDR 0.5 and 1 groups and to correlate with CSF levels. Importantly, the CSF YKL-40/Abeta42 ratio predicted risk of developing cognitive impairment (CDR 0 to CDR>0 conversion) as well as the best CSF biomarkers identified to date, tau/Abeta42 and p-tau181/Abeta42. Additionally, YKL-40 immunoreactivity was observed within astrocytes near a subset of amyloid plaques, implicating YKL-40 in the neuroinflammatory response to Abeta deposition. Utilizing an alternative, targeted proteomics approach to identify novel biomarkers, 333 CSF samples were evaluated for levels of 190 analytes using a multiplexed Luminex platform. The mean concentrations of 37 analytes were found to differ between CDR 0 and CDR>0 participants. ROC and statistical machine learning algorithms identified novel biomarker panels that improved upon the ability of the current best biomarkers to discriminate very mildly demented from cognitively normal participants, and identified a novel biomarker, Calbindin, with significant prognostic potential.