[1885] A Target Capture Based Next Generation Sequencing Panel for Identification of Reurrent Somatic Mutations in Cancer

Eric Duncavage, David Spencer, Haley Abel, Shashikant Kulkarni, Karen Seibert, Rakesh Nagarajan, Rob Mitra, Mark Watson, John Pfeifer. Washington University, St. Louis, MO; Washington University, St Louis, MO

Background: The detection of recurrent somatic mutations in cancer has become increasingly important to both the diagnosis and treatment of many malignancies. Current molecular testing paradigms rely on disparate methods to detect underlying sequencing alterations one gene at a time. Next Generation Sequencing (NGS) allows for the comprehensive sequencing of whole genomes at low cost, however these methods are generally optimized for the discovery of constitutional single nucleotide variation (SNV). Here we demonstrate that NGS can be used to detect a spectrum of clinically-relevant DNA mutations, including SNVs, insertions/deletions (indels), gene amplification, and translocations.
Design: We identified a set of cases with known recurrent somatic mutations including SNVs (10 cases), insertions (11), deletions (1), gene amplifications (3), and translocations (1). Genomic DNA was extracted from formalin-fixed or fresh tissue, indexed sequencing libraries created, and DNA captured using a set of custom-designed cDNA capture probes targeting a 437kb region including 28 genes commonly mutated in cancer. Cases were sequenced in multiplex using 2x101bp reads and the resulting data aligned to the human genome. SNVs were identified using the freely available Genome Analysis Toolkit and indels were identified using Pindel. Translocations were identified using Breakdancer and Slope.
Results: Each case generated approximately 12 million reads, resulting in a mean capture region coverage of 3076x, with a 47% on-target reads. Using well-characterized HapMap DNA we estimated a SNV sensitivity of >99% for heterozygous alleles; for alleles present at 10% frequency, the sensitivity was 98% with a positive predictive value of 94%. Using multiple software packages we identified somatic SNVs in 10 of 10 tested cases, FLT3 insertions in 10/11 cases, KIT deletions in 1/1 cases, EGFR gene amplifications in 3/3 cases, and MLL translocations in 1/1 case.
Conclusions: We demonstrate that high-coverage, targeted NGS is well-suited for the identification of the full spectrum of somatic mutations present in cancer. Further, we show that targeted NGS can reliably detect mutations present at 10% allele frequency, allowing for the detection of mutations in dilute tumor populations. As the number and complexity of recurrent cancer mutations increases, NGS-based methods will undoubtedly become invaluable in the clinical laboratory.
Category: Special Category - Pan-genomic/Pan-proteomic approaches to Cancer

Tuesday, March 20, 2012 9:30 AM

Poster Session III # 271, Tuesday Morning

 

Close Window