Construction and Validation of an Autophagy-Related Prognostic Model for Osteosarcoma Patients

Abstract

While the prognostic value of autophagy-related genes (ARGs) in OS patients remains scarcely known, increasing evidence is indicating that autophagy is closely associated with the development and progression of osteosarcoma (OS). Therefore, we explored the prognostic value of ARGs in OS patients and illuminate associated mechanisms in this study. When the OS patients in the training/validation cohort were stratified into high- and low-risk groups according to the risk model established using least absolute shrinkage and selection operator (LASSO) regression analysis, we observed that patients in the low-risk group possessed better prognosis ( P < 0.0001 ). Univariate/Multivariate COX regression and subgroup analysis demonstrated that the ARGs-based risk model was an independent survival indicator for OS patients. The nomogram incorporating the risk model and clinical features exhibited excellent prognostic accuracy. GO, KEGG, and GSVA analyses collectively indicated that bone development-associated pathway mediated the contribution of ARGs to the malignance of OS. Immune infiltration analysis suggested the potential pivotal role of macrophage in OS. In summary, the risk model based on 12 ARGs possessed potent capacity in predicting the prognosis of OS patients. Our work may assist clinicians to map out more reasonable treatment strategies and facilitate individual-targeted therapy in osteosarcoma.

A Novel Glutamine Metabolism-Related Gene Signature in Prognostic Prediction of Osteosarcoma

“…Based on the identified glutamine metabolism-related genes in osteosarcoma, a sixteen-gene signature was discovered by LASSO regression analysis. Unlike studies of previously reported prognostic models for osteosarcoma, such as models using autophagy-related genes, 17 super enhancer-related genes, 18 and immune-related genes, 19 , 20 our study was the first to use two external validation datasets rather than only one validation cohort to evaluate the model’s predictive accuracy, which is expected to substantially improve the model’s reproducibility and generalizability to new and different patients. Most importantly, the risk score calculated from the sixteen-gene signature had excellent performance in predicting the three- and five-year OS probabilities for all of the available osteosarcoma datasets in the GEO database.…”