Identification of potential genes and microRNAs related to recurrence risk of osteosarcoma by miRNA and mRNA integrated analysis

Objective: This study aimed to identify potential risk genes and microRNAs (miRNAs) related to the recurrence risk of osteosarcoma (OS) using support vector machine (SVM) algorithm. Methods: Based on the mRNA expression profiling dataset GSE39055 and the miRNA expression profiling dataset GSE39040, differentially expressed genes (DEGs) and differentially expressed miRNAs (DE-miRNAs) indiagnostic biopsy specimens between OS patients with and without recurrence were identified. Subsequently, an integrated network of DE-miRNAs and DEGs was constructed, and the modules were identified from the network. Afterwards, the significant genes and miRNAs in the modules were identified using the neighborhood scoring algorithm. Additionally, SVM was used to establish a prediction model of recurrence risk. Another independent mRNA expression profiling dataset GSE39057 and the miRNA expression profiling dataset GSE39052 were used to evaluate the efficiency of the prediction model. Results: In total, 1118 DEGs and 63 DE-miRNAs were identified between the recurrence group and non-recurrence group. Eleven modules were identified from the integrated network. Genes and miRNAs in the modules 1, 5, 6 and 10 were predicted to be significantly associated with the survival time of OS patients. Based on the neighborhood scoring algorithm, 14 genes belonging to the four network modules with a higher score were used to establish the prediction model. Finally, five feature genes in the prediction model were identified, including NUDT21, PPP2R5A, PLK4, RPS9 and MX1. The total precision of the model was 81%. Conclusion: The genes in the prediction model were new-found to be potentially associated with the recurrence risk of OS.