Screening of prognostic risk markers for pancreatic ductal adenocarcinoma based on bioinformatics approaches
Objective: To identify the prognostic risk markers for pancreatic ductal adenocarcinoma (PDAC) through bioinformatics approaches. Methods: The clinical information and data of miRNA and gene expression profiles of PDAC patients were downloaded from TCGA website. Then, the miRNAs and genes significantly related to the prognostic risk of PDAC were screened successively by Elastic Net Cox’s proportional risk regression hazards model (EN-Cox), the receiver operating characteristic (ROC) curve and survival analyses. Finally, literature mining and function analyses were conducted on the significant prognostic risk genes and the potential target genes of the significant prognostic risk miRNAs. Results: After data preprocessing, the complete clinical records and data of expression profiles of total of 797 miRNAs and 19 969 genes in 137 PDAC patients were obtained. Based on the parameter λ (0.107), 59 potential prognostic risk factors that included 54 genes and 5 miRNAs were screened via EN-Cox analysis. After grouping of the patients according to the cutoff values derived from the ROC curves and then drawing of Kaplan-Meier curves, 17 significant prognostic risk markers were finally identified (all P<0.05), including 16 genes and 1 miRNA (miRNA-125a). Among the 16 prognostic risk genes, glutathione S-transferase mu 4 (GSTM4), inducible T-cell co-stimulator ligand (ICOSLG) and spermatogenesis associated 2 (SPATA2) were simultaneously the target genes of miRNA-125a; GATA binding protein 1 (GATA1) was the only one transcription factor encoding gene. Conclusion: The functions of these screened candidates in PDAC still need to be elucidated, and they may probably be used as prognostic risk indicators and even therapeutic targets of PDAC.