A risk score model for the prediction of osteosarcoma metastasis

Abstract

Osteosarcoma is the most common primary solid malignancy of the bone, and its high mortality usually correlates with early metastasis. In this study, we developed a risk score model to help predict metastasis at the time of diagnosis. We downloaded and mined four expression profile datasets associated with osteosarcoma metastasis from the Gene Expression Omnibus. After data normalization, we performed LASSO logistic regression analysis together with 10‐fold cross validation using the GSE21257 dataset. A combination of eight genes ( RAB1 , CLEC3B , FCGBP , RNASE3 , MDL1 , ALOX5 AP , VMO1 and ALPK3 ) were identified as being associated with osteosarcoma metastasis. These genes were put into a gene risk score model, and the prediction efficiency of the model was then validated using three independent datasets ( GSE33383, GSE66673, and GSE49003) by plotting receiver operating characteristic curves. The expression levels of the eight genes in all datasets were shown as heatmaps, and gene ontology gene annotation and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis were performed. These eight genes play a role in cancer‐related biological processes, such as apoptosis and biosynthetic processes. Our results may aid in elucidating the possible mechanisms of osteosarcoma metastasis, and may help to facilitate the individual management of patients with osteosarcoma after treatment.