BackgroundMalignant mesothelioma (MM) is a highly aggressive cancer with a poor prognosis. Despite the use of several well-known markers, the diagnosis of MM is still challenging in some cases. we applied bioinformatics to identify key genes and screen for diagnostic and prognostic markers of MM.MethodsThe expression profiles of GSE2549 and GSE112154 microarray datasets from the Gene Expression Omnibus database contained 87 cases of MM tissue and 8 cases of normal mesothelial tissue in total. The GEO2R tool was used to detect differentially expressed genes (DEGs). Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of DEGs were performed using DAVID Bioinformatics Resources. The DEGs protein-protein interaction networks were constructed from the STRING database. Cytoscape was used to identify significant modules and hub genes. The GEPIA database was used to explore relationships between hub genes and prognosis of MM. Immunohistochemistry was used to analyze protein expression in tissue microarrays with 47 Chinese MM tissues. Statistical analyses diagnostic and prognostic values.Results346 DEGs were identified: 111 genes upregulated, and 235 downregulated. GO analysis showed that the primary biological processes of these DEGs were cell adhesion, leukocyte migration, and angiogenesis. The main cellular components included the extracellular space, extracellular exosome, and extracellular region. The molecular functions were integrin binding, heparin binding, and calcium ion binding. KEGG pathway analysis showed that DEGs are primarily involved in PPAR signaling pathway, extracellular matrix–receptor interactions, and regulation of lipolysis in adipocytes. Survival analysis showed that seven genes—AURKA, GAPDH, TOP2A, PPARG, SCD, FABP4, and CEBPA—may be potential prognostic markers for MM. Immunohistochemical studies showed that Aurora kinase A (AURKA gene encode, Aurora-A) and GAPDH were highly expressed in MM tissue in comparison with normal mesothelial tissue. Kaplan-Meier analysis confirmed a correlation between Aurora-A protein expression and overall survival but did not confirm a correlation with GAPDH. The receiver operating characteristic curves of Aurora-A protein expression suggested acceptable accuracy (AUC = 0.827; 95% CI [0.6686 to 0.9535]; p = 0.04). The sensitivity and specificity of Aurora-A were 83.33% and 77.78%, respectively.ConclusionAurora-A could be an optimal diagnostic biomarker and a potential prognostic marker for MM.