Background: Osteosarcoma (OS) is a common pediatric malignancy with high mortality and disability rates. Autophagy is an essential process in regulating the apoptosis and invasion of tumor cells, so constructing a risk score model of OS autophagy-related genes (ARGs) will bring benefit to the evaluation of both treatment and prognosis.
Methods: We downloaded a dataset of OS from the Therapeutically Applicable Research To Generate Effective Treatments (TARGET) database, and found OS-related ARGs through the Human Autophagy Database (HADb). Five hub ARGs (CCL2, AMBRA1, VEGFA, MYC and EGFR) were obtained using a multivariate Cox regression model. We then generated the risk scores and constructed a prediction model. Another dataset obtained from the Gene Expression Omnibus (GEO) was used to test accuracy and validity. The role of immune cell infiltration was systematically explored, and prediction of response to targeted drugs was assessed. Immunohistochemistry was carried out to verify the expression of the key ARGs.
Results: Based on the five hub ARGs, we established a risk score model related to OS. High accuracy and validity were demonstrated by datasets downloaded from the GEO. The five ARGs played a role in the PI3K and MAPK pathways. Results from targeted drug sensitivity analyses were consistent with pathway analyses. Immunohistochemistry demonstrated that the expression differences of the five ARGs were significant between the OS group and the paracancerous group.
Conclusions: We constructed a risk score model related to autophagy of OS, explored the diagnostic value of ARGs, and present possible therapeutic targets.
Keywords: Autophagy genes; osteosarcoma (OS); risk prediction model.
2022 Annals of Translational Medicine. All rights reserved.