Development and Validation of Novel Prognostic Models for Immune-Related Genes in Osteosarcoma

Immunotherapy has shown excellent therapeutic effects on various malignant tumors; however, to date, immunotherapy for osteosarcoma is still suboptimal. In this study, we performed comprehensive bioinformatic analysis of immune-related genes (IRGs) and tumor-infiltrating immune cells (TIICs). Datasets of differentially expressed IRGs were extracted from the GEO database (GSE16088). The functions and prognostic values of these differentially expressed IRGs were systematically investigated using a series of bioinformatics methods. In addition, CCK8 and plate clone formation assays were used to explore the effect of PGF on osteosarcoma cells, and twenty-nine differentially expressed IRGs were identified, of which 95 were upregulated and 34 were downregulated. Next, PPI was established for Identifying Hub genes and biology networks by Cytoscape. Six IRGs (APLNR, TPM2, PGF, CD86, PROCR, and SEMA4D) were used to develop an overall survival (OS) prediction model, and two IRGs (HLA-B and PGF) were used to develop a relapse-free survival (RFS) prediction model. Compared with the low-risk patients in the training cohort (GSE39058) and TARGET validation cohorts, high-risk patients had poorer OS and RFS. Using these identified IRGs, we used OS and RFS prediction nomograms to generate a clinical utility model. The risk scores of the two prediction models were associated with the infiltration proportions of some TIICs, and the activation of memory CD4 T-cells was associated with OS and RFS. CD86 was associated with CTLA4 and CD28 and influenced the infiltration of different TIICs. experiments showed that the knockdown of PGF inhibited the proliferation and viability of osteosarcoma cells. In conclusion, these findings help us better understand the prognostic roles of IRGs and TIICs in osteosarcoma, and CD86 and PGF may serve as specific immune targets.