Pancreatic Ductal Adenocarcinoma Arising in Young and Old Patients Displays Similar Molecular Features

Abstract

Pancreatic ducal adenocarcinoma is classically diagnosed in the 7th decade, but approximately 10% of patients are diagnosed under 55 years (y.o.). While the genomic and transcriptomic landscapes of late-onset tumors (LOT) have been described, little is known about early-onset tumors (EOT). Ageing is known to impact DNA methylation and proteome integrity through carbonylation-related oxidative damages. We therefore aimed to assess the global molecular features of EOT. We compared 176 EOT (≤55 y.o.) and 316 LOT (≥70 y.o.) from three distinct surgical cohorts at the clinical/genomic/epigenomic/transcriptomic level. Furthermore, we assessed oxidative stress responses and oxidative proteome damages using 2D gel electrophoresis followed by mass spectrometry protein identification. There was no consistent clinical difference between EOT and LOT across the three cohorts. The mutational landscape of key driver genes and the global methylation profile were similar in the two groups. LOT did display age-related features such as enriched DNA repair gene signatures and upregulation of oxidative stress defenses together with increased proteome carbonylation. However, these age-related differences were more preeminent in non-tumor tissues while tumor proteome and proteome damages were fairly comparable. In conclusion, this multi-omics comparison showed that EOT harbor a comparable molecular profile to that of LOT.

Keywords: PDAC; elderly patients; multi-omics; young patients.

Figure 1

Genomic landscape of pancreatic cancer patients. (a) Oncogrid of the top four altered genes (KRAS, TP53, SAMD4, and CDKN2A) in the Multi-centric cohort, 53 tumors from early-onset patients (EOP), 89 tumors from late onset patients (LOP); (b) Oncogrid from the TCGA data portal web application of the TCGA_PAAD project composed of 87 patients, 30 tumors from EOP (orange) and 57 tumors from LOP (green). Genes are ordered by decreasing alteration frequencies in each tumor. Each type of gene mutation is represented as a colored dot, red: Missense mutation, grey: loss of the start codon, purple: gain of the stop codon, green: coding sequence frameshift, yellow: loss of the stop codon.

Figure 2

Transcriptomic landscape of early-onset patients (EOP) and late-onset patients (LOP). Hierarchical clustering of EOP (orange) and LOP (green) from (a) Multi-centric cohort; (b) ICGC_PAN_AU cohort; (c) TCGA_PAAD cohort. (d,e) Gene set analysis (GSEA) of LOP vs. EOP from each dataset. Normalized Enrichment Score (NES) with a false discovery rate less than 0.25, and a p value less than 0.05. (d) GSEA of gene ontology (GO); (e) GSEA of Moffitt et al. PDAC subtype signature [10].

Figure 3

Oxidative stress damages in EOT vs. LOT. (a) Left panel: In-gel analysis of total protein carbonylation levels of non-tumor (NT) and tumor tissues (T) from early-onset patients (EOP) and late-onset patients (LOP) normalized by total protein input. Right panel: carbonylated protein rates of T compared to NT in EOP and LOP. Representative in-gel protein deposits are displayed in Figure S5a; (b) N(6)-Carboxyl-methyl-lysine (CML) and 4-Hydroxynonenal (HNE) levels quantified by Western blot and normalized by total protein expression (“stain-free” Biorad^® technology). Representative stain-free membranes, HNE, and CML Western blots are displayed in Figure S5b,c, (c) Thioredoxin (TXN), catalase, and superoxide dismutase (SOD) antioxidant protein expression normalized to β-Actin expression was analyzed by Western blot from EOP and LOP. Representative blots are displayed in Figure S5d, t test: * p < 0.05; ** p < 0.01.

Figure 4

Specific protein damages in tumor tissues (T) compared to non-tumor tissues (NT) in early-onset patients (EOP) and late-onset patients (LOP). (a) Principal component analysis (PCA) of total protein expression from each spot analyzed by 2D-DIGE. Lower panel: Venn diagram of specific spots from EOP and LOP, (b) PCA of carbonylated protein expression from each spot analyzed by 2D-DIGE. Lower panel: Venn diagram of specific spots from EOP and LOP. (c–f)—String software analysis of significant differential proteins (c) Down regulated in EOP; (d) Down regulated in LOP; (e) More carbonylated in EOP; (f) More carbonylated in LOP.