Identification of Prognosis-Related Molecular Subgroups and Construction of a Prognostic Prediction Model Using Immune-Related Genes in Pancreatic Cancer

BACKGROUND: Pancreatic cancer patients with similar clinicopathological status exhibit substantially different therapeutic responses, which might be caused by the vast molecular heterogeneity of tumors. In this study, we attempted to identify specific molecular subgroups and construct a prognostic prediction model based on the expression level of immune-related genes in pancreatic cancer. The transcriptome profiling, single nucleotide variation, copy number variation, clinicopathological information, and follow-up data of pancreatic cancer patients were obtained from The Cancer Genome Atlas database. Thereafter, the immune-related genes with prognostic significance were identified for further consensus cluster analysis. The molecular characteristics and clinicopathological information were compared between the identified subgroups, and a weighted correlation network analysis was performed to identify the hub genes associated with the subgroups. Finally, the prognostic prediction model based on immune-related genes was established using the least absolute shrinkage and selection operator (LASSO) analysis. RESULTS: A total of 67 immune-relevant genes with prognostic significance were selected and used for the consensus cluster analysis. The total samples were divided into two groups, C1 and C2. The subgroup C1 had a significantly worse prognosis than C2, as well as lower levels of immune cell infiltration, which indicate an immunosuppressed state. The mutational rate of the cancer-related genes including KRAS, TP53, and RNF43 was higher in the C1 subgroup. The C1 subgroup was associated with more advanced tumor grade and T stage and with higher mortality. Using LASSO regression, we developed a prognostic prediction model based on the expression levels of 19 immune-related genes, which we validated in three external data sets. In addition, we identified four potential therapeutic and prognostic biomarkers (TNNT1, KCNN4, SH2D3A, and PHLDA2). CONCLUSION: We identified two novel molecular subgroups of pancreatic cancer and developed a prognostic prediction model based on the expression levels of immune-related genes, which could be used in a clinical setting and could aid in unraveling the molecular processes leading to the development of pancreatic cancer.