(430h) From Genomes to Tissue-Level Metabolic Models- A Step towards Pathogenesis and Personalized Medicine

Renewed interest in metabolic changes in cancer has been spurred on due to a decreasing number of newly released anticancer drugs. In constraint-based analysis, functional states of network can be limited by various physicochemical constraints. Integration of OMICS data into a genome scale network of human metabolism can help to predict the tissue specific metabolic activity in normal and tumor cells. Gene expression data delineates components in a gene list for a specific tissue and suggests information about gene activity for a particular condition. In our modeling approach, we use gene expression data to constrain the maximum flux in the previously published model (Recon1) for human metabolism. Using this approach we are able to predict changes in flux capacity of every enzyme (coded by a gene) in the human metabolic network. We experimentally study changes in metabolism in A549 lung adenocarcinoma cells using phenotypic metabolite arrays and further use mass spectrometry to identify a spectrum of metabolites as predicted by the in silico cell at high growth rates.

We have identified diagnostic/prognostic measures (like Positron Emission Tomography with glucose, amino acid analogs) when combined with patient blood plasma microarray data and allowed us to predict clinically relevant tumor parameters like growth and response to therapy including stereotactic radiation therapy. A better understanding of links between cell metabolism and growth in the tumor cell can thus be achieved using this systems approach. Applied to patient models this can facilitate identification of biomarker panels and unique signatures and help drive individualized therapy and personalized medicine.