Robust and clinically relevant prediction of response to anti-cancer drugs via network integration of molecular profiles

In order to tackle heterogeneity of cancer samples and high data space dimensionality, we propose a method NEAmarker for finding sensitive and robust biomarkers at the pathway level. In this method, scores from network enrichment analysis transform the original space of altered genes into a lower-dimensional space of pathways, which is then correlated with phenotype variables. The analysis was first done on in vitro anti-cancer drug screen datasets and then on clinical data. In parallel, we tested a panel of state-of-the-art enrichment methods. In this comparison, our method proved superior in terms of 1) universal applicability to different data types with a possibility of cross-platform integration, 2) consistency of the discovered correlates between independent drug screens, and 3) ability to explain differential survival of treated patients. Our new in vitro screen validated performance of the discovered multivariate models. Finally, NEAmarker was the only method to discover predictors of both in vitro response and patient survival given administration of the same drug.