Objective
The present study utilized bioinformatics techniques and data from the GEO, TARGET, and ArrayExpress databases to compare gene expression in INSS4 and INSS1 neuroblastomas (NBs), thereby identifying metabolites with different levels of expression and predicting the prognosis of patients with NB.
METHODS
Genes of patients with INSS4 and INSS1 NBs from the GEO database were screened, with those having ཛྷlog2fold change (FC)ཛྷ>3 and adjusted P < 0.05 defined as being differentially expressed. These differentially expressed genes (DEGs) were screened to obtain clinical data and RNA sequence datasets from NB patients in the TARGET database. Univariate Cox proportional hazards regression analysis identified prognosis-related genes, which were incorporated into a prognosis model. Based on median risk scores, these patients were divided into high and low-risk groups. Their survival rates were compared, and ROC curves were used to analyze predictive values for NB. NB patients were also divided into two clusters by consensus clustering based on levels of POLR2H and DYNC1I2 expression. Immune infiltration analyses were performed using GSEA, ESTIMATE, CIBERSORT, and ssGSEA. Tumor tissue of 17 NB patients was used for experimental verification and their survival was compared.
Result
Analysis of three datasets identified 62 up-regulated genes and 163 down-regulated genes. The prognostic model predicted that the areas under the 3-year and 5-year survival curves were 0.786 and 0.817, respectively. Levels of expression of POLR2H and DYNC1I2 accounted for the highest percentage of risk scores and were included in follow-up analysis. Samples were consistently clustered according to their expression matrix. POLR2H was more highly expressed in cluster 2, whereas DYNC1I2 was more highly expressed in cluster 1. The survival rate of cluster 1 was significantly higher than that of cluster 2. Experimental verification in 17 NB patients showed that these patients could also be divided into two groups, which differed significantly in mortality hazard ratio (HR 9.37 P < 0.05).
Conclusion
The expression of POLR2H and DYNC1I2 affects the immune microenvironment of NB and can affect patient prognosis. These factors can be used to refine clinical groupings, guide personalized treatment, and suggest new methods for the diagnosis and treatment of NB.