Exploiting genetic complexity in cancer to improve therapeutic strategies

Abstract

Advances in genome sequencing technologies are enabling researchers to make rapid progress in defining the entire repertoire of causal genetic changes in cancer. The response of patients with cancer to therapy is often highly variable and there is an increasing number of examples where mutations in cancer genomes have been shown to have a profound effect on the clinical effectiveness of drugs. An urgent challenge for the research and clinical communities is how to translate these genomic data sets into new and improved therapeutic strategies for the treatment of patients. The use of large-scale cell line-based drug screens to identify genomic 'biomarkers' of drug response for the stratification of patients has the potential to transform how patients with cancer are treated.

FIGURE 1

The complexity of somatic alterations in human cancer genomes. A Circos plot showing the whole-genome catalogue of somatic mutations from the malignant melanoma cell line COLO-829. This genome has approximately 30,000 somatic base substitutions and 1000 somatic insertions and/or deletions. In coding exons, 272 somatic substitutions are present, including 155 missense changes, 16 nonsense changes and 101 silent changes. The number and types of mutation are highly variable across different cancer genomes. Chromosome number and karyotype are indicated on the exterior of the plot. Key: blue lines, copy number across each chromosome; red lines, sites of loss of heterozygosity (LOH); green lines, intrachromosomal rearrangements; purple lines, interchromosomal rearrangements; red spots, nonsense mutations; green spots, missense mutations; black spots, silent mutations; brown spots, intronic and intergenic mutations (merged).

FIGURE 2

A schematic workflow for high-throughput cell-line screening to identify genomic features associated with drug response. (a) A library of clinical and preclinical cancer drugs is screened (b) against a panel of cancer cell lines in 384-well plate format. (c) Drug sensitivity is measured over a nine-point titration of concentration and a curve-fitting algorithm is used to derive a signature of response, including the half-maximal inhibitory concentration (IC₅₀) and slope of dose response curve. (d and e) Drug-sensitivity data are correlated with genomic features, including point mutations, gene amplification and deletion, as well with gene-expression data to identify genomic features associated with drug sensitivity. (f) In this example, the association between v-raf murine sarcoma oncogene homologue B1 (BRAF) mutations and sensitivity to a BRAF inhibitor is identified. Each circle represents the IC₅₀ (natural log) of a single cell line and red line indicates the mean IC₅₀ for BRAF-mutated or wild-type (WT) cell lines.