Companion predictive biomarkers of response have become deeply embedded in early clinical trials of new oncology agents. Nevertheless, three confounding factors limit the success in the pursuit of these diagnostics and include: (i) the complex molecular biology of most common malignancies;
(ii) the multi-target inhibitory properties of many new agents such that antitumor activity is confounded by uncertainty of which target is responsible for the observed effect; and (iii) the respective platform used including whether it should focus on DNA-, RNA-, or protein-based assays. To address these factors, complex molecular genetic profiles from Sequenom and array comparative genomic hybridization assays carried out on patients entering phase I clinical studies will be presented to demonstrate the fallacy that single-agent antitumor activity can be demonstrated for all agents in phase I clinical study. Furthermore, recent examples of multi-targeted therapies and the serendipity of single-agent activity will be presented. Finally, the argument that DNA-based strategies have proven to be the most robust but are limited by the small spectrum of genomic evaluation, as well as the superficial approach are currently employed. The solution proposed is a system biology approach to genomic analysis of cancer that is necessary before improved drug development and companion biomarker strategies can be developed.