Genomics

Combining Genomic Approaches to Decrypt the Cancer Code

While some investigators are mapping the structural alterations in the cancer genome, others are searching for new oncogenes that fuel tumors.

Mapping Cancer's Altered Genome

The most comprehensive of the efforts to map structural alterations is The Cancer Genome Atlas (TCGA), a nationwide research network of more than a dozen institutions, including Dana-Farber and the Broad Institute. Its goal is to fully characterize the architecture of the cancer genome. The first published paper of the TCGA pilot project, led by Lynda Chin, MD, of the Department of Medical Oncology, and Matthew Meyerson, MD, PhD, focused on an integrative analysis of DNA copy number variations, gene expression, and DNA methylation aberrations in 206 glioblastoma tumor samples, as well as nucleotide sequence aberrations in 91 samples.

The epidermal growth factor receptor gene, EGFR, was analyzed in glioblastoma samples using SNP array and DNA sequence data. Each sample is a thin horizontal line. The EGFR copy number is indicated by the intensity of line: amplification as dark red lines, internal deletions within the gene by lighter shades of red or white stretches within the line. Mutations are shown as blue dashes. Courtesy of Matthew Meyerson, Wendy Winckler, and Gad Getz. 

Interim analysis revealed alterations in three core pathways (RB, p53, and RTK/RAS/PI3K) and somatic mutations in three genes (ERBB2, NF1, and PIK3R1) not previously associated with glioblastoma. "No one had looked at this cancer in enough specimens and in high enough resolution before to derive that information," says Meyerson. Moreover, because PI3K inhibitors are already in development, adds Chin, "the PIK3R1 discovery may soon lead to new clinical trials in glioblastoma and enable stratification of patients based on their PIK3R1 status."

One of the most exciting insights from the interim analysis came from integrating the sequencing and DNA methylation data with treatment information. Combining these data sparked a new hypothesis for the mechanism of resistance to temozolomide (Temodar, a standard treatment for glioblastoma): the tumor cell bypasses the drug by inactivating DNA repair machinery.

Their ultimate goal is to sequence and map every base pair of the cancer genome across a large collection of samples from ovarian, lung, and other tumors and disseminate the data to the research community.

Finding Mutations that Matter

Not all genetic mutations play an essential role in carcinogenesis. The goal of Project Achilles, led by William Hahn, MD, PhD, in collaboration with Meyerson, Levi Garraway, MD, PhD, Kornelia Polyak, MD, PhD, of the Department of Medical Oncology, and Todd Golub, MD, of Pediatric Oncology, is to find and exploit cancer's vulnerabilities using integrative genomic approaches.

"Each genomic approach is powerful, but gives you only one dimension of information," explains Hahn. "By combining different approaches, you quickly find a small set of genes – and in some cases, just one gene – that satisfy all criteria" for the development of cancer.

In one study, Hahn and colleagues, including postdoctoral fellow Ron Firestein, MD, PhD, in Hahn's lab and clinical fellow Adam Bass, MD, in Meyerson's, set out to find the genes required for full transformation to colon cancer. At Dana-Farber's RNAi facility and at the Broad, investigators used RNA interference to conduct two loss-of-function screens: one for genes essential for proliferation, the other for those that regulate the Wnt/beta-catenin pathway (implicated in virtually all colon cancers), resulting in a subset of genes at the intersection of the two screens. Next, they determined which genes of this subset were also amplified in colon cancer specimens. Using high-density SNP arrays - originally designed for genome-wide detection of single nucleotide polymorphisms (SNPs) associated with disease, but modified by Meyerson to assess copy number variations – a single gene emerged from the SNP arrays, CDK8, a new oncogene mutated in up to 50 percent of colon cancers.

The investigators applied a similar rationale to hunt for breast cancer oncogenes. They added another tool from the genomic arsenal, thanks to Marc Vidal, PhD, of the Department of Cancer Biology and director of Dana-Farber's Center for Cancer Systems Biology. Vidal had compiled a library of open reading frames (ORFs), the coding sequences of genes. While short interfering RNAs allow investigators to knock down genes of interest, ORFs allow them to over-express genes to screen for gain-of-function mutations, a common occurrence in breast cancer. Integrating RNAi, the ORF library, and high-density SNP arrays, they discovered the oncogene IKBKE, which is mutated in 30 percent of breast cancers. Since the products of CDK8 and IKBKE are kinases, which lend themselves to small-molecule inhibition, these two discoveries alone could have an impact on patients.

"We're able to do this research because we have resources at Dana-Farber and the Broad that do not exist anywhere else," notes Hahn. "It feels almost effortless, because we're all collaborating."