Abstract
Osteosarcoma (OS) is a highly heterogeneous malignant bone tumor, and its tendency to metastasize leads to a poor prognosis. TGFβ is an important regulator in the tumor microenvironment and is closely associated with the progression of various types of cancer. However, the role of TGFβ-related genes in OS is still unclear. In this study, we identified 82 TGFβ DEGs based on RNA-seq data from the TARGET and GETx databases and classified OS patients into two TGFβ subtypes. The KM curve showed that the Cluster 2 patients had a substantially poorer prognosis than the Cluster 1 patients. Subsequently, a novel TGFβ prognostic signatures (MYC and BMP8B) were developed based on the results of univariate, LASSO, and multifactorial Cox analyses. These signatures showed robust and reliable predictive performance for the prognosis of OS in the training and validation cohorts. To predict the three-year and five-year survival rate of OS, a nomogram that integrated clinical features and risk scores was also developed. The GSEA analysis showed that the different subgroups analyzed had distinct functions, particularly, the low-risk group was associated with high immune activity and a high infiltration abundance of CD8 T cells. Moreover, our results indicated that low-risk cases had higher sensitivity to immunotherapy, while high-risk cases were more sensitive to sorafenib and axitinib. scRNA-Seq analysis further revealed that MYC and BMP8B were strongly expressed mainly in tumor stromal cells. Finally, in this study, we confirmed the expression of MYC and BMP8B by performing qPCR, WB, and IHC analyses. To conclude, we developed and validated a TGFβ-related signature to accurately predict the prognosis of OS. Our findings might contribute to personalized treatment and making better clinical decisions for OS patients.