Establishment of a Prognostic Model for Pancreatic Cancer Based on Hypoxia-Related Genes

Abstract

Objectives: Pancreatic cancer presents a formidable challenge with its aggressive nature and dismal prognosis, often hampered by elusive early symptoms. The tumor microenvironment (TME) emerges as a pivotal player in pancreatic cancer progression and treatment responses, characterized notably by hypoxia and immunosuppression. In this study, we aimed to identify hypoxia-related genes and develop a prognostic model for pancreatic cancer leveraging these genes.

Methods: Through analysis of gene expression data from The Cancer Genome Atlas (TCGA) and subsequent GO/KEGG enrichment analysis, hypoxia-related pathways were identified. We constructed a prognostic model using lasso regression and validated it using an independent dataset.

Results: Our results showed that expression levels of PLAU, SLC2A1, and CA9 exhibited significant associations with prognosis in pancreatic cancer. The prognostic model, built upon these genes, displayed robust predictive accuracy and was validated in an independent dataset. Furthermore, we found a correlation between the risk score of the prognostic model and clinical parameters of pancreatic cancer patients. At the same time, we also explored the relationship between the established hypoxia-related prognostic model and the immune microenvironment at the single-cell level. RT-qPCR results showed notable differences in the expression of hypoxia pathway-related genes between normal PANC-1 and hypoxic-treated PANC-1 cells.

Conclusion: Our study provides insights into the role of the hypoxic microenvironment in pancreatic cancer and offers a promising prognostic tool for clinical application.

Keywords: hypoxia; immune infiltration; immunosuppression; pancreatic cancer; tumor microenvironment.

Figure 1.

Gene differential expression and enrichment analysis from GEPIA2, A A total of 9211 differentially expressed genes were screened, comprising 1279 genes up-regulated and 116 genes down-regulated. B Pathways related to hypoxia screened through GO/KEGG enrichment analysis. One of the pathways related to hypoxia was “response to hypoxia” (GO:0001666), with a total of 45 genes identified.

Figure 2.

Visualization of genes associated with hypoxia pathways, A Genes related to hypoxia pathway screened by the “Stress” algorithm in the STRING database. B Selected key hypoxia genes with a p-value less than 0.05 identified through univariate Cox regression analysis.

Figure 3.

Risk score analysis, prognostic performance, and survival analysis of prognostic models, A-B selected key genes related to hypoxia by Lasso regression and established the formula of prognosis model. C Divides 178 patients in the training set into high-risk group and low-risk group. D In the training set, the expression of PLAU, SLC2A1, and CA9 was higher in the high-risk group and lower in the low-risk group. F ROC time analysis showed that the prognostic accuracy of this model at 1 year, 3 years, and 5 years were 0.69011, 0.72589, and 0.78947, respectively.

Figure 4.

Analysis of the validation set prognostic model's risk score, prognostic performance, and survival analysis, A 65 patients in the verification set were divided into high risk group and low risk group. B Verification set. The expression of PLAU, SLC2A1, and CA9 was higher in the high-risk group, while lower in the low-risk group. C ROC time analysis showed that the prognostic accuracy of this model at 1 year, 3 years, and 5 years were 0.5848, 0.81829, and 0.94286, respectively. D Verifies the prediction ability of the model through calibration curve and ROC curve analysis. E The risk score of this model has a higher AUC value and a better prognosis. F Prognostic model established by combining clinical prognostic risk factors and risk score of prognostic model.

Figure 5.

Association between clinical features and risk scores in the TCGA-PAAD dataset; The relationship between the risk score of the prognostic model and T stage(A) and tumor residue (B).

Figure 6.

Immune infiltration and risk score association in the TCGA-PAAD dataset, The relationship between the risk score of the prognostic model and the degree of immune cell invasion (A) and the expression of immune-related molecules (B).

Figure 7.

Distribution of hypoxic-related genes in the immune microenvironment of pancreatic cancer, A GSA: cell scatter diagram of CRA001160. B Distribution of cell clusters in GSA: CRA001160 in tumor tissue and normal tissue. C Distribution of CA9 in cell clusters. D Distribution of PLAU in cell clusters. E Distribution of SLC2A1 in cell clusters. F Expression of CA9 in cell clusters. G Expression of PLAU in cell clusters. H Expression of SLC2A1 in cell clusters. I Single-cell RNA dataset GSA: CRA001160 Components of the pancreatic cancer immune microenvironment in CRA001160 and cellular communication in pancreatic cancer malignant cells.

Figure 8.

The gene was expressed in PANC-1 cells and hTERT-HPNE cells, Real-time quantitative PCR verified the high expression of PLAU, SLC2A1, and CA9 in pancreatic cancer cells (A,C,E), and the higher expression in hypoxia conditions (B,D,F).