Abstract
PURPOSE: To explore the association of ferroptosis with repeated implantation failure (RIF) and prognostic capability of ferroptosis-related genes. METHODS: Data in GSE106602 from the GEO database were used for gene co-expression network construction to confirm ferroptosis-related genes compared to gene sets that were downloaded from FerrDB. Then these genes were analyzed for functional enrichment and validated using endometrium samples from our center. ImplantScore and ROC curve were constructed for prognostic correlation analysis. RESULTS: We observed that ferroptosis probably participated in RIF according to bioinformatics analysis on a gene set which exhibited a strong association with RIF from WGCNA. Fifty-four ferroptosis-related genes in the gene set were subsequently verified, and the PPI network was established for underlying interactions among them. There were 23 hub genes with differential expression in RIF and six of them (PML, LCN2, PRKAA1, BACH1, SLC7A11, and CAMKK2) showed significant correlation with implantation outcomes using samples collected from our center. Therefore, we combined the six genes and constructed an ImplantScore whose AUC reached 0.891, higher than the AUC of each single gene, respectively. ImplantScore of six genes with down-regulated expression in the group with failed implantation were much lower than that with successful outcome. CONCLUSION: Our results demonstrated the potential prognostic functions of ferroptosis-related biomarkers in RIF, which will provide novel perspectives for further research and clinical applications.