In the battle against cancer, knowing what you're up against is crucial. Two recent Nature papers provide elegant examples of how genetically characterizing tumours can help to predict progression and drug sensitivity, and show that patient stratification will be essential to evaluate the efficacy of targeted anticancer drugs.

The use of 'gene signatures' — specific gene-expression profiles for particular tumour types — is well established in cancer diagnosis, target discovery and drug screening. Joseph Nevins and colleagues took a fresh approach to this concept and generated signatures that represent the activation status of individual oncogenic pathways. They used recombinant adenoviruses to express various oncogenic signals in quiescent human mammary epithelial cells (HMECs) so that the subsequent gene-expression signature represents the downstream effects of that particular oncogenic event. Principal components analysis was then used to select a set of genes that represents the signature for that pathway.

The authors successfully used the signatures to distinguish between cells expressing each oncogenic activity and the control cells, both in HMECs and in tumour samples derived from standard mouse cancer models. In non-small-cell lung carcinoma samples, they found that a Ras pathway signature distinguished lung adenocarcinoma from lung squamous cell carcinoma, confirming that the former has a higher probability of Ras mutation than the latter.

However, real predictive power was achieved by combining pathway signatures. Analysis of lung tumours showed that adenocarcinomas predicted to have low-level Ras activity generally also had elevated Myc, E2F3, -catenin and Src activity. Integrating these signatures identified a patient subpopulation with poor survival. Perhaps most importantly, the authors could also predict drug sensitivity: growth inhibition measured in cancer cell lines exposed to Ras or Src inhibitors correlated with predicted status of the corresponding pathway, indicating the potential of these signatures as a guide to therapy in patients.

Genetic characterization of tumours was also used by Neal Rosen et al. in a study of the effects of Braf and Raf mutations on the mitogen-activated protein kinase kinase (MEK) and extracellular signal-related kinase (ERK) pathway. Mutations in Ras and Braf are often mutually exclusive and therefore might exert their oncogenic activity through common downstream signals that could be exploited as drug targets.

The authors used a selective MEK inhibitor in cells with Ras or Braf mutations and found that tumours with Braf mutations were exquisitely sensitive to MEK inhibition compared with Ras mutant and wild-type cells. Indeed, using the National Cancer Institute's chemical sensitivity database (NCI60), the authors ranked compounds according to growth inhibition of tumour cells with Braf mutations and found that the most effective compounds were all known to inhibit MEK. Furthermore, in mice bearing wild-type and Braf-mutant xenograft tumours, growth of the Braf-mutant tumours was completely suppressed.

The finding that tumours containing mutations in Braf are much more dependent on MEK inhibition than tumours with mutant Ras has several implications for drug development. First, MEK inhibitors have a selectivity that could be exploited in the treatment of Braf-mutation-dependent cancers; however, in certain genetic contexts, MEK inhibitors might need to be used in combination with other targeted therapies. Second, the authors speculate that their results could explain the mixed success of Braf inhibitors in clinical trials, and that patients should be stratified according to the mutation status of their tumours in future clinical studies.

Author: Joanna Owens