In Nature Biotechnology this week, a multi-national group of scientists publish the results of a retrospective study designed to use cancer genomic and proteomic data across multiple tumor types to predict patient survival. Using somatic copy-number alteration, DNA methylation and mRNA, microRNA, and protein expression data from 953 samples of four cancer types, the researchers found that the incorporation of molecular data with clinical variables yielded significantly improved predictions for three cancers, although those qualitative gains were limited. Additional analyses, meantime, revealed little predictive power across tumor types except for one case. The study provides a starting point for the use of molecular data to improve prognosis and therapeutic intervention.
Also in Nature Biotechnology, a Harvard Medical School-led team reports on the use of the gene-editing technology CRISPR-Cas9 to overcome a key hurdle in the use of mouse models for cancer research. Noting that genome sequencing studies have shown that malignancies frequently have mutations in four or more driver genes, but that recapitulating such genetic complexity in animals using conventional breeding has been difficult, they used lentiviral vectors to deliver into mice small guide RNAs and Cas9 that could modify up to five genes in a single hematopoietic stem cell. This resulted in clonal outgrowth and myeloid malignancy, and allowed the scientists to generate models of acute myeloid leukemia with cooperating mutations in genes encoding epigenetic modifiers, transcription factors, and mediators of cytokine signaling, reflecting the combinations of mutations observed in patients.