A signature of 18 immune-related gene pairs to predict the prognosis of pancreatic cancer patients

Abstract

Pancreatic cancer is one of the most lethal malignancies. With the promising prospects conveyed by immunotherapy in cancers, we aimed to construct an immune-related gene pairs (IRGPs) signature to predict the prognosis of pancreatic cancer patients. We downloaded clinical and transcriptional data of pancreatic cancer patients from The Cancer Genome Atlas data set as the training group and GSE57495 data set as the verification group. We filtered immune-related transcriptional data by IMMPORT. With the assistance of lasso penalized Cox regression, we constructed our prognostic IRGPs signature and divided all samples into high-/low-risk groups by receiver operating characteristic curve for further comparisons. The comparisons between high- and low-risk groups including survival rate, multivariate, and univariate Cox proportional-hazards analysis, infiltration of immune cells, and Gene Set Enrichment Analysis (GSEA). Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) are facilitated to analyze the proceedings in which our IRGPs signature may involve in. The results revealed that 18 IRGPs were defined as our prognostic signature. The prognostic value of this IRGPs signature was verified from the GSE57495 data set. We further demonstrated the independent prognostic value of this IRGPs signature. The contents of six immune cells between high-/low-risk groups were different, which was associated with the progression of diverse cancers. Results from GO, KEGG, and GSEA revealed that this IRGPs signature was involved in extracellular space, immune response, cancer pathways, cation channel, and gated channel activities. Evidently, this IRGPs signature will provide remarkable value for the therapy of pancreatic cancer patients.

Keywords: immune-related gene pair; pancreatic cancer; prognostic signature.

Figure 1

Construction of the immune‐related gene pairs (IRGPs) signature. (A) Time‐dependent receiver operating characteristic (ROC) curve analysis (1 year) for IRGPs signature in the training group, the optimal cutoff is 1.057 to classify patients into high‐/low‐risk groups. (B) Area under receiver operating characteristic curve (AUC) is 0.843

Figure 2

Validation of the immune‐related gene pairs (IRGPs) signature. (A) Divide the training‐group patients into high‐/low‐risk groups. (B) The comparison of overall survival rate between high‐/low‐risk groups in patients from The Cancer Genome Atlas (TCGA) data set. (C) Univariate Cox proportional‐hazards analysis of the risk factors in the training group. (D) Multivariate Cox proportional‐hazards analysis of the risk factors in the training group. (E) Divide the validation‐group patients into high‐/low‐risk groups. (F) The comparison of overall survival rate between high‐/low‐risk groups in patients from GEO (GSE57495) data set

Figure 3

Correlation between the immune‐related gene pairs (IRGPs) model with infiltration of immune cells. (A) Summarize the difference of infiltration of immune cells between high‐/low‐risk groups (*p < .05, **p < .01, and ***p < .001). (B) The content of B cells memory (p = .036) was higher in the high‐risk group. (C) The content of B cells naïve (p < .001) was lower in the high‐risk group. (D) The content of macrophages M0 (p = .035) was higher in the high‐risk group. (E) The content of macrophages M1 (p = .015) was higher in the high‐risk group. (F) The content of natural killer cells activated (p = .036) was higher in the high‐risk group. (G) The content of T cells CD8 (p = .010) was lower in the high‐risk group

Figure 4

Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA) of the immune‐related gene pairs (IRGPs) model. (A–F) GSEA analysis between the high‐/low‐risk groups. The results revealed six gene sets that the patients in the high‐risk group may involve in (results of FDR < 0.05 were listed). (G) GO analysis including biological process, molecular function and cell component of the genes in the IRGPs signature. (H) KEGG analysis predicted the biological pathways which our IRGPs signature may involve in

Figure 5

The mRNA expression of each IRGP in pancreatic cancer. Results revealed that CD1C_MUC5AC, CD1D_DKK1, CXCL9_APLNR, CXCL11_CD79A, ERAP2_SSTR1, EREG_RARB, GMFB_TGFA, ICAM1_MET, IRF3_MET, OAS1_AGT, PLAU_ZYX, and PPARG_FGR were overexpressed in pancreatic cancer. *p < .05. IRGP, immune‐related gene pair; mRNA, messenger RNA

Figure 6

Correlation between the expression profile of each IRGP with pancreatic cancer stage. (A–C) Results revealed that mRNA expressions of AGT_OAS1, ICAM1_MET, and CHGA_IL22RA1 correlated with pancreatic cancer stage. IRGP, immune‐related gene pair; mRNA, messenger RNA

Figure 7

Mutation analysis of the IRGPs signature in pancreatic cancer. (A–C) The mutation rate of CD1D_DKK1 (15%), IRF3_MET (14%), and ZYX_PLAU (12%) was the top three highest in pancreatic cancer. The most common mutational category of the three IRGPs was mRNA high. Moreover, nine patients have CD1D_DKK1 genetic amplification, which is the highest among all IRGPs. (D) Results revealed that 112 (67%) of 168 pancreatic cancer samples have genetic alterations of the genes in the IRGPs signature, the most common mutational category was mRNA high. IRGP, immune‐related gene pair; mRNA, messenger RNA