Abstract PR05: Investigating the importance of low allele frequency mutations for cancer patient management

Abstract

Abstract Genomic profiling is becoming a routine for linking targeted cancer therapy to patients likely to benefit. However, it is unclear whether it is clinically valuable to identify low allele frequency mutations that are present in heterogeneous tumor samples, either in terms of response or resistance to targeted therapies. To investigate this question, we established a T200 sequencing platform that performs routine deep targeted exon sequencing of 201 cancer related genes to identify somatic mutations that are present at low allele frequency. Through our intensively optimized protocol, we successfully obtained a 1,000-fold average sequencing depth and identified 4,794 non-synonymous mutations and over 400 high copy number alterations, from 515 formalin-fixed paraffin-embedded tumor samples across 12 disease sites in 475 patients. We found that 15.2% of the non-synonymous mutations were present at less than 10% allele frequency, which would have little chance to be identified using whole exome sequencing as currently implemented in the majority of the sequencing laboratories or other standard platforms that do not reach high sequencing depth. In addition to genomic profiling, we established a knowledgebase consisting of genes and mutations that are potentially clinically actionable in the literature and a collection of drug and clinical trial databases. We also predicted 2,316 putative driver mutations in 1,544 genes and 18 tumor types based on a cancer driver annotation (CanDrA) program and a statistical framework that comprehensively integrates functional evidences implicated by 1) mutational hotspots, 2) sequence context, 3) differential mRNA expression, 4) differential protein expression and pathway activity, 5) differential drug sensitivity, 6) mutual-exclusive occurance of mutations in cancer signaling pathways and 7) differential prevalence of mutations across tumor types, derived from the Catalog of Somatic Mutations in Cancer (COSMIC) version 67, the Cancer Genome Atlas (TCGA) and the Cancer Cell-Line Encyclopedia (CCLE). We found that a considerable fraction of low allele frequency mutations that we identified were likely drivers with potential functional and/or clinical importance. An ability to identify these mutations provided potentially clincally relevant information in 118 (24.84%) patients, among which 47 (9.8%) would otherwise not have an actionable event. Our study demonstrated the potential clinical relevance of detecting mutations with low allele frequency and potential subclonal mutations in clinical tumor samples via deep sequencing and integration of multiple sources of functional genomic evidence into decision making. Citation Format: Ken Chen, Funda Meric-Bernstam, Hao Zhao, Tenghui Chen, Kenna Shaw, John Mendelsohn, Agda Eterovic, Gordon Mills. Investigating the importance of low allele frequency mutations for cancer patient management. [abstract]. In: Proceedings of the AACR Special Conference on Computational and Systems Biology of Cancer; Feb 8-11 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 2):Abstract nr PR05.