Editors' ChoiceCancer

Mapping a path for precision cancer therapies

See allHide authors and affiliations

Science Translational Medicine  20 Jul 2016:
Vol. 8, Issue 348, pp. 348ec115
DOI: 10.1126/scitranslmed.aah4512

The future of precision cancer medicine will be dictated in part by the degree to which diverse tumor genotypes and tissue lineages can be connected with specific, druggable vulnerabilities. One approach to this challenge is to perform large-scale drug sensitivity screens in genomically annotated cancer cell lines to reveal context-specific vulnerabilities, an idea that was first broadly implemented using the NCI60, a panel of established cell lines that have been subjected to thousands of drug profiling assays. In recent years, several groups have expanded this approach by genomically annotating hundreds of cell lines, then screening these lines against dozens to hundreds of carefully chosen drug and probe compounds.

Now, in an ambitious pharmacogenomic screening effort, a team of researchers reported the screening of around 1000 cancer cell lines with 265 drugs. They began by analyzing recurrent somatic mutations, copy number alterations, and hypermethylation patterns across over 11,000 human tumor samples to create a catalog of 1273 cancer functional elements (CFEs). CFEs were then mapped to cell lines and systematically investigated as possible predictors of differential drug sensitivity. Nearly 700 interactions between individual CFEs and drugs were detected, with 34% of CFEs associating with sensitivity to at least one drug and 85% of drugs associating with at least one CFE. Multi-input logic models of sensitivity often outperformed single predictor models, implying that AND/OR combinations of CFEs may more accurately predict response. Last, the authors used machine learning to demonstrate that tissue type is a dominant predictor of sensitivity in pan-cancer analyses, and genomic features dominated cancer type-specific predictions.

This and related studies suggest several important avenues for investigation. First, many newly nominated therapeutic strategies can be explored because around 50% of human tumors harbor at least one CFE or CFE combination associated with increased drug responsiveness. Second, the authors’ demonstration that increasing the number of cell lines screened enabled the discovery of substantially more CFE-drug interactions suggests that further expanding these screens to additional cell line models, including those developed using organoid and conditional reprogramming methods, will continue to yield new, actionable therapeutic vulnerabilities. Third, loss-of-function genetic screens are likely to be important complements to the pharmacological screens described here. Last, given the tendency of single agent therapies to give rise to resistance, it will be imperative to understand whether large scale screening can identify combinations of drugs with further improved activity, selectivity, and durability.

F. Iorio et al., A landscape of pharmacogenomic interactions in cancer. Cell 10.1016/j.cell.2016.06.017 (2016). [Full Text]

Stay Connected to Science Translational Medicine

Navigate This Article