Discussion
This hypoxia-related prognostic panel may be effective in predicting the clinical prognosis, immune infiltrations, immunotherapy, and chemotherapy in GC.
Methods
The GC scRNA-seq data and bulk RNA-seq data were downloaded from the GEO and TCGA databases, respectively. AddModuleScore() and AUCell() were used to calculate module scores and fractions of enrichment for hypoxia-related gene expression in single cells. Least absolute shrinkage and selection operator cox (LASSO-COX) regression analysis was utilized to build a prognostic panel, and hub RNAs were validated by qPCR. The CIBERSORT algorithm was adopted to evaluate immune infiltration. The finding of immune infiltration was validated by a dual immunohistochemistry staining. The TIDE score, TIS score and ESTIMATE were used to evaluate the immunotherapy predictive efficacy.
Results
Hypoxia-related scores were the highest in fibroblasts, and 166 differentially expressed genes were identified. Five hypoxia-related genes were incorporated into the hypoxia-related prognostic panel. 4 hypoxia-related genes (including POSTN, BMP4, MXRA5 and LBH) were significantly upregulated in clinical GC samples compared with the normal group, while APOD expression decreased in GC samples. Similar results were found between cancer-associated fibroblasts (CAFs) and normal fibroblasts (NFs). A high hypoxia score was associated with advanced grade, TNM stage, N stage, and poorer prognosis. Decreased antitumor immune cells and increased cancer-promoting immune cells were found in patients with high hypoxia scores. Dual immunohistochemistry staining showed high expression of CD8 and ACTA2 in gastric cancer tissue. In addition, the high hypoxia score group possessed higher TIDE scores, indicating poor immunotherapy benefit. A high hypoxia score was also firmly related to sensitivity to chemotherapeutic drugs.