Abstract
BACKGROUND: Keloid (KD) disease is a skin lesions caused by abnormal wound healing that involve complex cellular and molecular mechanisms. The aim of this study was to screen diagnostic markers of KD via bioinformatics methods and evaluate their clinical application value. METHODS: The GSE44270, GSE145725, GSE7890 and GSE83286 datasets were analyzed in combination with difference analysis and weighted gene coexpression network analysis (WGCNA) and machine learning algorithms, candidate genes related to KD were screened and verified via receiver operating characteristic (ROC) curves and external datasets. KD samples were classified by consistent clustering, and the infiltration of immune cells was investigated. Simultaneously, a diagnostic biomarker-related ceRNA network was constructed. Drug small molecules and compounds were predicted online, and molecular docking was performed. Finally, RT‒qPCR and WB were used to verify the expression of the markers. RESULTS: In this study, two upregulated genes, SMURF2 and CCDC80, which are significantly associated with a variety of immune cells, were screened. KD was divided into the C1 and C2 subtypes, SMURF2 was highly expressed in C1, and CCDC80 was highly expressed in C2. Drug prediction and molecular docking analysis suggest that bisphenol A may have a potential effect on KD therapy. RT‒qPCR and WB revealed that the mRNA and protein expression levels of SMURF2 and CCDC80 in KD samples were significantly increased. CONCLUSION: Our study identified two genes that may be used as diagnostic markers of KD, providing new perspectives and potential molecular targets for the study of the molecular mechanisms and clinical diagnosis of KD.