AUTHOR=Xu Zhen , Yang Mei , Ou Chenghao , Mao Liming , Liu Zhaoxiu 

TITLE=Identification of ferroptosis-genes associated with pediatric inflammatory bowel disease bioinformatics and machine learning approaches

JOURNAL=Frontiers in Immunology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2025.1619944

DOI=10.3389/fimmu.2025.1619944

ISSN=1664-3224

ABSTRACT=BackgroundPediatric inflammatory bowel disease (PIBD) is increasingly common, and early diagnosis remains challenging due to unclear etiology. Ferroptosis, an iron-dependent form of cell death, may be involved in intestinal inflammation, but its expression and role in PIBD are poorly understood.ObjectiveTo identify ferroptosis-related genes as candidate biomarkers for early diagnosis of PIBD and validate their role in ferroptosis.MethodsRNA-seq data of PIBD from GEO datasets were analyzed using DESeq2, WGCNA, and functional enrichment analysis. Ferroptosis-related diagnostic genes were screened through LASSO, Random Forest, and mSVM-RFE algorithms, and validated in GSE57945 and GSE117993 datasets. In vitro experiments using NCM460 cells were performed to validate the roles of PML and CHAC1 in LPS-induced ferroptosis, including siRNA-mediated gene knockdown, western blotting of ferroptosis-related proteins (ACSL4, SLC7A11, GPX4, FTH), and measurement of lipid peroxidation (MDA levels). CIBERSORT was used to assess immune cell infiltration, and DGIdb was used to predict potential targeted drugs. A ceRNA network was further constructed to explore miRNA-lncRNA interactions regulating these genes.ResultsPML and CHAC1 were identified as potential biomarkers for early diagnosis of PIBD, showing high diagnostic performance (AUC > 0.7) in training, validation, and external datasets. In vitro experiments confirmed that knockdown of PML or CHAC1 significantly alleviated LPS-induced ferroptosis in NCM460 cells, as evidenced by restored ferroptosis-related protein expression and reduced MDA accumulation. Consistent with immune infiltration results, both genes were associated with immune-related pathways, and a ceRNA network revealed their potential involvement in complex regulatory mechanisms. DGIdb predicted several candidate drugs targeting these genes.ConclusionPML and CHAC1 are promising biomarkers for early PIBD diagnosis. These findings, supported by both bioinformatic analyses and experimental validation, may improve diagnostic accuracy and provide insights into the immune microenvironment and therapeutic strategies.