The number of somatic tumor mutations with potential utility for predicting treatment response is rapidly growing due to increasing numbers of targeted therapies. Clinical validation of these potential biomarkers has been slowed by both issues with tumor samples and the current paradigm underlying cancer clinical trials. Tumor DNA samples can be heterogeneous due to invasion into stroma, infiltration by immune cells, and clonal evolution. Efforts to overcome this heterogeneity have, to date, required highly focused testing of no more than a few dozen known mutations, significantly limiting progress. Additionally, cancer drug development has traditionally focused on a tissue of origin model, where efficacy studies are focused on cancers arising from one tissue type. As molecular subtyping has emerged, molecularly defined trials have been restricted to common DNA alterations (for example, Imatinib and KIT gene mutations in gastro-intestinal stromal tumors) [1] or uncommon alterations in very common tumors (for example, Erlotinib and EGFR-L858R in non-small cell lung cancer) [2]. The identification of an increasing number of somatic tumor mutations common across cancers arising from different tissues has begun to encourage molecularly
