Identification and Diagnostic Significance of MPO, PRTN3, and CTNND1 as Biomarkers in Acute Hematogenous Osteomyelitis in Children: A Comprehensive Analysis Using Machine Learning Algorithms

Jiafei Yang^1*Tianjiu Zhang²kezhi Chen³Jihang Luo³Song Yu¹

• ¹Affiliated Hospital of Guizhou Medical University, Guiyang, China
• ²Affiliated Hospital of Zunyi Medical University, Zunyi, Guizhou Province, China
• ³Zunyi Medical University, Zunyi, Guizhou Province, China

Abstract Introduction: Acute hematogenous osteomyelitis (AHO) is a severe bacterial bone infection predominantly affecting children. Early diagnosis is crucial to prevent the progression to chronic osteomyelitis. However, current diagnostic methods are limited in sensitivity and specificity, underscoring the need for reliable biomarkers. Materials and methods: This study utilized gene expression data from the Gene Expression Omnibus (GEO) to identify differentially expressed genes (DEGs) associated with AHO. We employed three machine learning algorithms—The Least Absolute Shrinkage and Selection Operator (LASSO) regression, Support Vector Machine-Recursive Feature Elimination (SVM-RFE), and Random Forest (RF)—to screen for potential diagnostic markers. The expression levels of key genes were validated using clinical samples from pediatric AHO patients. Receiver operating characteristic (ROC) curve analysis was performed to assess the diagnostic accuracy of these biomarkers. Results: Our analysis identified five candidate genes, among which Myeloperoxidase (MPO), Serum proteinase 3 (PRTN3), Catenin delta 1 (CTNND1) were significantly associated with AHO, MPO and PRTN3 were upregulated, while CTNND1 was downregulated in AHO samples compared to healthy controls. ROC curve analysis demonstrated that CTNND1 (AUC = 0.8832), MPO (AUC = 0.9803) and PRTN3 (AUC = 0.9767) exhibited strong diagnostic potential. Importantly, the expression levels of MPO and PRTN3 positively correlated with disease severity as classified by the Cierny-Mader staging system, whereas CTNND1 expression showed a negative correlation. Corresponding author:Tianjiu Zhang. Email:873019524@qq.com and Song Yu: Email:yusongabc@126.com *These authors have contributed equally to this work and share first authorship Conclusion: MPO, PRTN3, and CTNND1 are promising biomarkers for the diagnosis and monitoring of AHO in children. Their expression levels correlate with disease severity, making them valuable tools for assessing the progression and treatment efficacy in pediatric AHO. Further research is warranted to explore their potential in clinical applications.