Abstract

BackgroundPlacental dysfunction and immune dysregulation are central to disorders such as fetal growth restriction (FGR) and preeclampsia (PE). Although aberrant placental glycosylation has been implicated in various pregnancy complications, the specific expression patterns of glycosylation-related genes (GRGs) associated with placental pathology and their potential role in driving immune imbalance at the maternal−fetal interface remain poorly understood.MethodsFirst, differentially expressed gene (DEG) identification and gene set enrichment analysis based on two GEO datasets (GSE114691 and GSE203507) revealed disease-associated pathways. Weighted gene co-expression network analysis (WGCNA) further identified glycosylation-related gene modules significantly associated with disease status. Three machine learning algorithms (LASSO, random forest, and support vector machine) were employed to screen diagnostic key genes, and a predictive nomogram was constructed. The efficacy of the diagnostic model was evaluated using receiver operating characteristic curves, calibration curves, and decision curve analysis. CIBERSORT was used to assess immune cell infiltration characteristics in the placenta and decidua of FGR, PE, and PE+FGR. Furthermore, single-cell RNA sequencing validated the expression patterns of key diagnostic genes across different trophoblast subsets, and cell-cell communication networks were analyzed using CellChat. In addition, this study validated the diagnostic efficacy of key genes in real-world samples.ResultFollowing batch correction and integration of two datasets, 104 samples were analyzed, including controls (n=26), PE (n=28), FGR (n=23), and PE+FGR (n=27). Glycosylation-related pathways were enriched in disease groups. Intersection of WGCNA modules with glycosylation genes yielded 97 overlapping genes, from which five core genes were identified using three machine learning algorithms. A diagnostic model based on these genes showed good predictive p