Global Proteomics: Pharmacodynamic Decision Making via Geometric
Interpretations of Proteomic Analyses
Paul Kearney, Nathan L. Currier, Daniel Chelsky, Clarissa Desjardins, Patrice Hugo, Joanna
Hunter, Eustache Paramithiotis, Marc Riviere, Olivier Maes, Howard M. Cherkow,
Hyman M. Schipper
Disease and drugs can modulate the concentrations of hundreds of proteins in the blood which can be accurately
measured using contemporary proteomic methods. Nevertheless, it is common practice to reduce the plurality
of disease and drug effects by a few proteins for the pragmatic purposes of immunoassay development. The vast
majority of putative biomarkers discovered by this reductionist approach never reach the clinic for two reasons:
the prohibitive time and cost of assay development and the acute risk of a reduced protein panel failing when
validated on a broader cross-section of the population.
Global Proteomics is an alternate methodology where all blood proteins modulated by disease or drug are used
to resolve pharmacodynamic questions without the time, cost, and risk of developing an immunoassay. The
Global Proteomic approach was applied to an Alzheimer study where it was demonstrated that a large panel of
plasma proteins is predictive of disease severity (as measured by the Mini Mental State Examination).
Furthermore, a subset of this panel was shown to be modulated by disease treatment (donepezil), thereby providing
a means to quantify response to treatment. Finally, to establish that the Global Proteomics methodology has
broad utility, it was also applied to a Hypertension study, illustrating that a panel of plasma proteins can also be
derived that are correlated with disease severity (as measured by blood pressure). In particular, the Global
Proteomics methodology can readily distinguish patients responsive and non-responsive to hypertension therapies.
The Global Proteomics approach is based upon a bioinformatics analysis approach which clusters samples by
proteomic similarity and then uses a geometric representation of sample similarity to answer common