Machine learning and multi-omics-based identification of hub paternal imprinted genes PEG11/RTL1, PEG9/DLK1, PEG6/NDN, and PEG5/NNAT in sperm of couples experiencing idiopathic recurrent pregnancy loss.

BACKGROUND

Idiopathic recurrent pregnancy loss (IRPL), defined as two or more consecutive pregnancy losses without an identifiable cause, affects 1-2 % of couples in their reproductive years. While maternal factors have been extensively studied, paternal contributions remain underexplored. Paternally imprinted genes are essential for spermatogenesis, embryonic development, and placental function, and their epigenetic dysregulation may contribute to IRPL.

RESULTS

Single-cell RNA-seq analysis highlighted RTL1, DLK1, NDN, and NNAT as hub paternal imprinted genes specifically enriched in spermatogenic lineages. In sperm samples, PEG11/RTL1, PEG9/DLK1, and PEG5/NNAT showed significantly reduced expression, whereas PEG6/NDN was upregulated in the IRPL group. DNA methylation assays revealed hypermethylation of PEG11/RTL1, PEG9/DLK1, and PEG5/NNAT and hypomethylation of PEG6/NDN in IRPL cases. Machine learning-based consensus models achieved high predictive accuracy, with the Random Survival Forest model performing best (average AUC = 0.747). Functional enrichment analyses linked these genes to key biological processes including glucose transport, VEGF receptor signaling, and placental development.

CONCLUSIONS

Our findings demonstrate that aberrant expression and methylation of paternal imprinted genes-particularly PEG11/RTL1, PEG9/DLK1, PEG6/NDN, and PEG5/NNAT-play a critical role in IRPL. By integrating multi-omics data with machine learning, this study provides novel insights into paternal genetic and epigenetic contributions to recurrent pregnancy loss and identifies potential biomarkers for diagnosis and therapeutic targets for intervention.