Establishment Of A Three-Gene Prognostic Model In Osteosarcoma Based On GEO And TARGET Databases

Background and objective:Osteosarcoma is the most common bone tumor among children,adolescents and young adults.It is characterized by high malignancy,high recurrence rate and easy to metastasis,which often lead to poor prognosis.With advances in multidisciplinary treatment for osteosarcoma,the 5-year survival rate for most patients with osteosarcoma has improved to nearly 70%.However,in the past 40 years,there has been no further breakthrough in the prognosis of osteosarcoma.The aim of this study was to further improve the prognosis of patients with osteosarcoma by combining RNA microarray and sequencing data from GEO and TARGET databases to identify potential genetic biomarkers for osteosarcoma.Methods and results:Based on the GEO database,we identified 730 differentially expressed genes in18 pairs of osteosarcoma/healthy bone tissue samples in the GSE99671 dataset,and explored the potential function of these genes by enrichment analysis and protein-protein interaction network analysis.In addition,we screened 94 osteosarcoma patients from the TARGET database and extracted their differential gene expression matrices and clinical information,resulting in 703 gene expression matrices and complete clinical data for 70 patients.Three genes(GCA,GFRA3 and MOCOS)were screened by univariate Cox regression analysis,LASSO regression,survival analysis and multivariate Cox regression analysis,and a prognostic model was constructed.Further analysis showed that the prognostic model is independent of conventional clinical features.The AUC value of ROC curve is 0.812,indicating that the model has a good prediction performance.We have constructed a nomogram based on the three-gene model and conducted internal verification using the TARGET database.The C-index is 0.813 and the 95% confidence interval is 0.722-0.904,indicating that the nomogram has good predictive power.The 3-and 5-year calibration curves showed good predictive accuracy.Conclusion:Differentially expressed genes in osteosarcoma were mainly enriched in biological processes such as neutrophil-mediated immunity,neutrophil-degranulation,neutrophil-activation involved in immune response,and extracellular matrix-receptor(ECM-receptor)interaction pathway.The risk score of the three-gene prognostic model(GCA,GFRA3 and MOCOS)is an independent prognostic factor for osteosarcoma.The nomogram combined with the three-gene prognostic model can provide a basis for clinicians to select a personalized treatment plan for patients with osteosarcoma.