This article has been corrected to note that the mean coverage was 391 times, not 102 times as cited in the original article.
CHICAGO (GenomeWeb News) – A team of researchers led by investigators at the Dana-Farber Cancer Institute have developed a massively-parallel sequencing strategy for profiling "actionable" cancer genes from formalin-fixed paraffin-embedded tumor samples, Dana-Farber oncologist Nikhil Wagle reported at the American Society of Clinical Oncology annual meeting.
The new strategy builds on a mass spectrometry-based method known as OncoMap, which Dana-Farber's Levi Garraway and others described in PLoS ONE in 2009. OncoMap profiles around 400 mutations in 33 oncogenes or tumor suppressor genes.
As Wagle explained during an ASCO session on tumor biology yesterday, the OncoMap approach, though useful, has certain limitations. For instance, it can only detect known mutations in a relatively small number of genes. And though it can detect point mutations, he noted, it has limited potential for finding small insertions and deletions and misses chromosomal amplifications and deletions.
To overcome such issues, researchers have come up with a high-throughput sequencing-based strategy that involves isolating genomic DNA from tumors and using it to create bar-coded Illumina sequencing libraries.
Researchers then pool these libraries and nab cancer gene sequences using the Agilent SureSelect capture system before sequencing the genes.
When Wagle and his co-workers tested this approach in 10 cancer lines — assessing about 400,000 bases of DNA coding for 138 cancer-related genes that are thought to be clinically actionable — they were able to simultaneously detect SNPs, indels, amplifications, and deletions in the cells.
For example, in a breast cancer cell line known as MDA-MB-231, the team detected a CDKN1A amplification, along with deletions in JAK2 and CDKN2A — changes that they subsequently validated using microarray analysis.
From there, the team set out to do a pilot study using DNA for 10 FFPE samples for breast and colon cancer and two control DNA samples, generating sequences covering the targeted genes with a mean depth of around 391 times.
In the process, they identified a host of mutations that appear to lend themselves to targeted therapeutics. Among them: alterations in KRAS, PIK3CA, TSC1, and BRCA1.
In addition, comparison of data from the pilot studies with findings from OncoMap suggests that the new sequencing-based approach has very high sensitivity and specificity.
As such, the method has potential applications for studies of tumor biology as well as for translational analyses, Wagle said, pointing to a melanoma treatment resistance study that he and his colleagues published online in the Journal of Clinical Oncology this spring.
For that study, researchers used their targeted cancer gene sequencing approach to track down a treatment-related mutation that ups MEK1 kinase activity in a melanoma sample from an individual whose BRAF-mutant tumor had developed resistance to PLX4032 (vemurafenib), an anti-cancer drug being developed by Plexxikon and Roche/Genentech.
In a discussion talk at the session, Massachusetts General Hospital pathologist John Iafrate commended the team on their proof-of-principle efforts to apply massively parallel sequencing for multiplexed cancer gene mutation analysis.
"This type of data will open up incredible new opportunities for clinical trials," he said.
Even so, Iafrate cautioned, several issues need to be addressed before this and other high-throughput sequencing-based approaches become routine for profiling tumors in a clinical setting — from clinical validation and tumor tissue size and quality issues to regulatory compliance and health care reimbursement.