Betűméret: A A A

Analysis of interactions between inflammatory and vasoregulatory pathways in chronic heart failure

Project summary:

The hypothesis that multiple risk factors and interactions between genetic and environmental factors are involved in the predisposition to multifactorial diseases is well supported by our understanding of their pathomechanisms. The development of statistical methods to analyze thousands of genetic and biochemical markers in clinical studies, however, have not kept pace with the progress of high-throughput technologies. Furthermore, biostatistical strategies are underdeveloped to analyze complex interactions in clinical studies.

The aim of the current project is to introduce a data-mining tool, described in operations-research for non-medical applications, into medical research, allowing the analysis of complex interactions. This goal will be achieved by the organization of a model clinical study and determination of large number of genetic and biochemical markers. Chronic heart failure will be investigated to build up a large database containing pathway-based genomic data together with detailed clinical characteristics, phenotypes and outcomes. Analysis of this database will be done by logical analysis of data (LAD). LAD is able to describe patterns of interacting variables with high power to predict phenotypes/outcomes. The LAD models will be validated against predictive models generated by multivariable methods. The components (molecular-, or other types of interactions) of predictive patterns obtained in LAD models will be checked and biologically validated in new studies including functional genetic, endothelial cell-based-, and immunochemical assays to delineate the molecular interactions within the models. As a result, interaction based risk-stratification models, together with mechanism of action, will be obtained for characterization of clinical phenotypes; for prediction of outcomes; and for planning of novel diagnostics/therapeutics. Furthermore, the improved LAD software may fuel follow-up research in a wide spectrum of applications where identification of interacting variables is the cardinal question.