OBJECTIVE

Preeclampsia (PE) is a multifactorial systemic pregnancy disease, in which iron metabolism and ferroptosis play significant roles during its pathogenesis. The diagnosis and prevention of PE remain urgent clinical issues that need to be addressed. Therefore, finding molecular diagnostic targets for PE through bioinformatics and machine learning methods is crucial for the diagnosis and prevention of patients with PE.

METHODS

Data sets for PE were obtained from the GEO database, and gene differential expression analysis was conducted along with enrichment analysis annotations. Subsequently, WGCNA was used to screen for genes associated with PE. Functional annotations were performed for the intersection of differentially expressed genes (DEGs), two key modules, and iron metabolism-related genes. Lasso-Cox, SVM, and XGboost machine learning methods were utilized to identify hub genes related to iron metabolism in PE, followed by GSEA analysis. The diagnostic value of the hub genes was assessed using ROC curves, and the correlations of hub genes with ferroptosis were evaluated based on ssgsea scores. Finally, the immune cell infiltration in PE was assessed, along with the relationship between hub genes and infiltrating immune cells.

RESULTS

A total of 355 differentially expressed genes in PE were identified. The functional enrichment analysis indicated that the genes were primarily associated with extracellular matrix, inflammatory response, immune response, iron ion binding, transport, and homeostasis, endoplasmic reticulum lumen, and hypoxic response. Pathway enrichment analysis revealed associations primarily with metabolic pathways, PI3K-Akt signaling pathway, cAMP signaling pathway, JAK-STAT signaling pathway, oxidative phosphorylation, HIF-1 signaling pathway, and pathways related to iron absorption and transport. Through WGCNA analysis and machine learning, five hub genes associated with PE were finally identified: LTF, PLOD2, CP, NR1D2, and P3H2. LTF, PLOD2, and CP were highly expressed in the PE group, while NR1D2 and P3H2 were lowly expressed. ROC curve analysis demonstrated that all hub genes had good diagnostic value. The ssgsea scores indicated that hub genes were significantly associated with ferroptosis. The immune infiltration results revealed that resting CD4+ memory T cells and regulatory T cells participated in the pathogenesis of PE.

CONCLUSION

LTF, PLOD2, CP, NR1D2, and P3H2 may serve as diagnostic biomarkers for PE, and the occurrence of PE is related to iron metabolism responses.