Whole-transcriptome paired-end sequencing and the pancreatic cancer genetic landscape.

Abstract

4048 Background: A deeper knowledge of the pancreatic cancer (PDAC) biology is needed to improve the prognosis of the disease. Methods: 17 PDAC samples were collected by ultrasound-guided biopsy used for DNA and RNA extraction. 14 samples were analyzed by high resolution copy number analysis (CNA) on Affymetrix SNP array 6.0 and with segmentation algorithm against a reference of 270 Ceu HapMap individuals (Partek Genomic Suite). 17 samples were analyzed by whole transcriptome massively parallel sequencing, performed at 75x2 bp on a HiScanSQ Illumina platform. An average of 7, 3x107 reads per sample were generated, with a mean read depth of 50X. Single nucleotide variants (SNVs) were detected with SNVMix2 and compared with genetic variation databases (dbSNP, 1000genomes, Cosmic). Non-synonimous SNVs were analyzed with the predictors SNPs and GO and PROVEAN. Results: CNA results in 9/14 samples exhibited both macroscopic and cryptic cytogenetic alterations, with a mean of 10 CNA per patient. Most frequent gains were observed in 18q11.2 involving GATA6 (3/14) and 19q13 targeting AKT2 (3/14) while hotspot deletions were found on 18q21 (7/14), 17p13 (6/14), 9p21.3 (6/14), 15q (5/14) and 1q35 (4/14). RNAseq showed that samples exhibited a mean of 145 (range: 61-240) non-synonimous SNVs, of which 16 on average are potentially disease-related. Merging copy number and RNAseq data we highlighted the major oncogenic hits of PDAC, confirming the prevalence (14/17) of KRAS mutations, in one case also NRAS (G13D), and the three oncosuppressor CDKN2A (mutated in 3 cases and deleted in 6 cases, in hetero- or homozygosity), SMAD4 (altered by point mutation or gene deletion in 7/14), and TP53 (lost in 6/14 and mutated in 5/17). The signaling pathways affected were: KRAS/MAPK, TGFbeta and integrin signaling, proliferation and apoptosis, DNA damage response, and epithelial to mesenchymal transition. Moreover we found new oncogenic alterations, such as HMGCR, that displayed mutations in 17% of the analyzed patients (3/17). Conclusions: NGS combined with high resolution cytogenetic analysis can improve the understanding of pancreatic carcinogenesis.