Evidence and potential clinical significance of changes in gene network interactions in ovarian cancer

Abstract

The molecular basis of cancer is not merely the consequence of structural and/or regulatory mutations in genes, but additionally to disruptions in networks of regulatory interactions existing among these genes and other components of the genome. Disruptions in network relationships may manifest as the loss, gain or reversal of functionally significant interactive gene relationships in cancer cells. In this study, we first employ an unsupervised (Pearson correlation) approach to quantitatively estimate the overall change in network relationships between precursor (control) ovarian surface epithelial cells and ovarian cancer epithelial cells. We find that ovarian cancer cells display a significant overall reduction in correlated gene network interactions relative to normal precursor cells reflective of an overall loss of regulatory control.  We next focus on gene relationships that qualitatively change between normal and cancer samples. We find that biological processes significantly over represented among differentially expressed genes are substantially different from those associated with genes involved in qualitatively disrupted network interactions. Our findings provide novel insights into the processes underlying ovarian cancer and identify a potential new class of genes for targeted therapy.