Radioresistance-related gene signatures identified by transcriptomics characterize the prognosis and immune landscape of pancreatic cancer patients.

BACKGROUND

Radiotherapy (RT) is an important means of local treatment of solid tumors, and radioresistance is the main reason for RT failure for tumors, especially pancreatic cancer (PC). It is urgent to distinguish key genes and mechanisms of radioresistance in PC.

METHODS

We acquired the data from The Cancer Genome Atlas (TCGA), obtained the gene modules associated with radioresistance by weighted gene coexpression network analysis (WGCNA), and identified differentially expressed genes (DEGs) between normal and tumor samples. Radioresistance-related genes (RRRGs) were determined with the intersection of WGCNA and DEGs. The hub RRRGs associated with prognosis were distinguished by the least absolute shrinkage and selection operator (LASSO) regression. We established a risk score model using multivariate Cox regression. Immune cell infiltration and drug sensitivity were evaluated through the CIBERSORT algorithm and the "OncoPredict" software package, respectively. The association of the key gene RIC3 and PC clinical features was verified in public databases, and its biological behaviors were explored in vitro.

RESULTS

The intersection of DEGs and WGCNA confirmed 14 RRRGs, then six hub RRRGs were identified using LASSO. A key four genes (DUSP4, ADORA2B, SCGB2A1, and RIC3) risk score model was constructed and proved to be capable of independently estimating the prognosis of PC. There is no significant difference between risk score groups in various immune cell infiltration and response to immunotherapy. Although the low-risk group seemed to exhibit greater sensitivity to antitumor drugs, the four drugs (5-fluorouracil [5-FU], leucovorin, irinotecan, and oxaliplatin) currently used for PC patients had no statistical difference for the low- and high- group. The overexpression of RIC3 had a synergy effect with irradiation on inhibited malignant biological properties of PC cells, which was verified by detecting the proliferation ability, apoptosis rate, cell cycle distribution, and migration ability of PANC-1 cells.

CONCLUSIONS

We herein presented signature genes correlated with radioresistance in PC and established a risk score model competent in estimating patients' clinical outcomes and response to antitumor drugs. The above evidence could contribute to comprehending the mechanisms of radioresistance and identifying the underlying therapy targeting.