J Matern Fetal Neonatal Med. 2026 Dec;39(1):2651458. doi: 10.1080/14767058.2026.2651458. Epub 2026 Apr 6.
ABSTRACT
BACKGROUND: Multiple omics studies on patients with recurrent pregnancy loss (RPL) have deepened the understanding of its pathogenesis. However, few studies have combined multi-omics techniques to provide a more accurate characterization of RPL. This study aims to identify biomarkers with RPL through proteomic and transcriptomic analyses, providing new insights for its diagnosis and treatment.
METHODS: Endometrial tissue samples were collected from RPL patients (n = 34) and normal controls (n = 22) for proteomic analysis to identify differentially expressed proteins (DEPs). Protein-protein interaction network analysis and functional enrichment analysis were performed to explore the biological functions of the DEPs. LASSO regression was used to screen for hub proteins, which were further validated using transcriptomic data from the GSE165004 dataset (24 RPL patients and 24 controls). An artificial neural network (ANN) model was constructed to assess the classification performance of the key DEPs.
RESULTS: A total of 275 DEPs were identified between the RPL group and the normal groups. Function enrichment analyses revealed significant involvement of these DEPs in immune and inflammatory responses. LASSO analysis identified 23 hub proteins. By combining transcriptomic data, five proteins, FOSB, HPS4, MRPL34, LCAT, and TMSB10 were ultimately identified as key DEPs. The ANN model demonstrated high accuracy in distinguishing between RPL patients and normal controls, with an accuracy rate of 81.25%.
CONCLUSION: Our study identified five key DEPs closely associated with RPL and revealed their promising diagnostic potential. Future validation in independent cohorts and functional studies is warranted to confirm their value as diagnostic biomarkers.
CLINICAL TRIALS REGISTRY: Not applicable.
PMID:41942343 | DOI:10.1080/14767058.2026.2651458
