Construction of a Pyroptosis-Related Genes Signature to Improve the Prognostic Prediction and Therapeutic Drugs Selection in Patients with Pancreatic Cancer

Abstract

Background: Effective prognostic assessment and appropriate drug selection are important for the clinical management of pancreatic cancer (PaC). Here, we aimed to establish a pyroptosis-associated genes (PRGs) signature to predict the prognostic outcomes of PaC and guide clinical drug therapy. Methods: We identified the differentially expressed PRGs between pancreatic adenocarcinoma (n = 178) and control pancreas samples (n = 171) obtained from different databases, and performed Lasso and Cox regression analysis to create a prognosis signature. Kaplan–Meier (K-M) survival curves and time-dependent receiver operating characteristics were further constructed to assess the utility of the risk model. The International Cancer Genome Consortium (ICGC) PACA-AU cohort (n = 95) was used as a validation dataset to examine the validity of this prognostic model. The correlations of risk score (RS) with clinical features, immune cell infiltration, tumor mutation burden and half-maximal inhibitory concentrations (IC50) of chemotherapeutic drugs were analyzed, and the expression levels of PRGs in cell lines were detected. Results: A prognostic signature was constructed, which consisted of 4 PRGs (AIM2, IL18, GSMDC and PLCG1). K-M analysis demonstrated a remarkable difference in overall survival (OS) time between low-risk (LR) and high-risk (HR) groups (P < 0.001). The RS contributed to the progression of PaC, and could be a significant independent factor for prognostic prediction. The validation of the ICGC cohort confirmed the effectiveness of the proposed signature. The patients with a HR score in the TCGA cohort had higher tumor mutation burden and more sensitivity to paclitaxel, gemcitabine, 5-fluorouracil and cisplatin than those with a LR score. The differential expression levels of signature genes were verified in vitro. Conclusion: The PRGs signature can be applied for predicting the prognosis of PaC, and may provide useful information for selection of therapeutic drugs.