Abstract B35: Integrative nomination of prostate cancer driver genes from paired-end sequencing

Abstract

Listen

Translate article

Abstract Structural variation arising from genomic instability, a hallmark of cancer, contributes to tumor progression through loss of tumor suppressor genes and dysregulation of cancer drivers. Additionally, genomic rearrangements may result in the formation of chimeras, such as the TMPRSS2-ERG fusion, in a subset of prostate cancers. However, a complete catalog of structural variants (SVs) in prostate cancers has not yet been established. Using a genome-wide, paired-end approach with massively parallel sequencing, we mapped breakpoints and copy number variation (CNVs) in 8 prostate cancer cell lines (LNCaP, PC3, DU145, VCaP, MDA-PC2b, LNCaPC4-2, LAPC4, and 22RV1) and 2 non-tumorigenic immortalized prostate epithelial cell lines (PrEC and RWPE1). Because these are cell lines established without a germline reference, we utilized comparable sequencing data from the 1000 Genomes Project to remove structural rearrangements that are present in the normal population. In addition, we removed from the set of prostate cancer cell line variants all events that were found in either of the normal cell lines as a method of correcting for cell culture artifacts. These two approaches removed between 40 and 65% of the structural variant calls from the prostate cancer cell lines. After normalization, RWPE1 was the least rearranged cell line (129 variants) and VCaP was the most heavily rearranged (784 variants). More than 40% of these structural variants fall within 2kb of a known gene, a remarkable enrichment over the 1.5% coding rate of the human genome (approximately 1 gene per 100kb). Structural variants from these 8 cell lines nominated 436 recurrently disrupted genes, in which we find significant enrichment of known transcription factor motifs (NFAT, FOXO4, and others), epigenetic marks (H3K27me3), and cancer related pathways (differentiation and cell-cell adhesion). An integrative approach was employed to nominate potential cancer driver genes from the SV- and CNV-affected genes through extensive comparison with available gene expression, aCGH, and sequencing data from human prostate tumors, and network centrality modeling. As expected, known drivers of prostate cancer, such as PTEN, MYC, and AR were significantly disrupted across multiple data types and therefore ranked highly by our Nomination Index. However, we have also identified several genes that have not previously been implicated in prostate cancer and may represent novel drivers of the disease. Citation Format: Christina Stewart Pichot, Cristian Coarfa, Mark P. Hamilton, Sean M. Hartig, Caroline Schoenherr, Adrian V. Lee, Aleksander Milosavljevic, Sean E. McGuire. Integrative nomination of prostate cancer driver genes from paired-end sequencing [abstract]. In: Proceedings of the AACR Special Conference on Advances in Prostate Cancer Research; 2012 Feb 6-9; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2012;72(4 Suppl):Abstract nr B35.