Abstract
OBJECTIVE: To explore new diagnostic markers and therapeutic targets associated with the histological types of thymoma. METHODS: The TCGA and GEO was used to gather genomic and clinical data from thymoma patients. WGCNA, ConsensusClusterPlus, five machine learning algorithms employed to screen critical histological predictive genes. The effects of different GBX2 expression levels on histological type differences, genomic instability, immune infiltration, carcinogenesis pathway correlation, singel cell expression and other functions in thymoma were examined. Autodock is used to calculate the binding site between GBX2 and potential drugs. RESULTS: Thymoma patients with low GBX2 expression likely have Type 2(B1/B2/B3) thymoma. The mutation rate of GTF2I was lower in samples with low GBX2 expression. GBX2 has a higher degree of methylation in the GTF2I SNV mutation group. GBX2 expression was correlated with microRNA in C19MC cluster and the development of an immunosuppressive tumor microenvironment. GBX2 may influences oxidative phosphorylation pathway activation in thymoma and is expressed on cTEC, mTEC, etc. Erlotinib may be effective in the treatment of thymoma through GBX2. CONCLUSION: Low expression of GBX2 is associated with the formation of Type 2 thymoma by influencing the oxidative phosphorylation pathway in and may participate in immunosuppression in tumor cells. Erlotinib may be a therapeutic drug for thymoma acts through GBX2. These findings suggest that GBX2 may be a potential therapeutic target for clinical thymoma. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s12672-026-04460-z.