Abstract
Background
Recurrent pregnancy loss (RPL) is a common reproductive complication, and the specific pathogenesis is still unclear. This study aimed to investigate RPL-related biomarkers and molecular mechanisms from the transcriptome of RPL decidua tissue using modern bioinformatics techniques, providing new perspectives for the etiology and clinical diagnosis and treatment of RPL.
Methods
Three gene expression profiles of RPL decidua tissue were retrieved and downloaded from the GEO database. Differential analysis, WGCNA analysis, and functional enrichment analysis were performed on the merged data. Subsequently, three machine learning methods (LASSO, SVM-RFE, and RF) were used to select the optimal feature genes for RPL, which were experimentally validated by RT-qPCR. The immune cell infiltration in RPL was evaluated using the ssGSEA algorithm, and the biological functions of the optimal feature genes were explored. Lastly, a heatmap was constructed to assist clinical physicians.
Results
10 key differentially expressed genes were identified: CFHR1, GPR155, TIMP4, WAKMAR2, COL15A1, LNCOG, C1QL1, KLK3, XG, and XGY2. Enrichment analysis showed associations with complement and coagulation cascade pathways. The three machine learning algorithms identified CFHR1 as the optimal feature gene for RPL, and RT-qPCR confirmed its high expression in RPL. ROC curve and nomogram demonstrated its diagnostic efficacy for RPL. Immune infiltration analysis revealed increased macrophages and γδT cells in RPL decidua tissue, with a significant positive correlation between CFHR1 and macrophages.
Conclusion
Transcriptomic abnormalities exist in RPL decidua tissue, with key genes closely related to complement and coagulation cascade pathways; CFHR1 is identified as the optimal feature gene for RPL. Abnormal immune infiltration and correlation with CFHR1 are observed in RPL decidua tissue.