Direct measurement of adhesive forces in amyloid beta peptide structures

possible AD samples are collected from a well-characterized clinical population in New York. Using protein two-dimensional electrophoresis, fluorescence staining and laser fluorescence scanning, we are able to visualize approximately 1200 protein spots from typical CSF samples. The Random Forest multivariate statistical analysis is applied to spot patterns from 68 samples to segregate AD and nonAD patient samples. Conclusions: A panel of 23 protein spots is most effective at segregating AD and nonAD patient samples. Matrix assisted laser desorption ionization tandem time of flight mass spectrometry has permitted us to characterize and sequence many of these 23 proteins. Based on these identifications, the group of protein biomarkers include three general functions: Abeta transport, inflammation, and proteolytic enzyme inhibitors. These markers have 94% sensitivity, 94% specificity, and a predicted error classification rate of 5.9% for the 68 samples studied. In an ROC type analysis, the AUC is 0.965, which compares favorably to other reported AD biomarkers and suggest the potential for clinical diagnostic applications.