Blood-based biomarker discovery for early pregnancy loss using integrative multi-omics strategies

Background
Early pregnancy loss (EPL), a spontaneous death of the embryo or foetus occurring within the first trimester, is a major challenge for human reproduction with profound adverse consequences for women’s health. Currently, reliable blood-based biomarkers for EPL remain limited. Therefore, there is an urgent need to discover novel biomarkers for EPL using a multi-omics-based approach to facilitate early detection and timely management.
Methods
In the discovery cohort, 40 patients with EPL and 40 healthy pregnancies (HP) at 7–13 weeks of gestation were enrolled. Serum proteins and metabolites were assayed by Olink® technology and ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS), respectively. Biomarkers were defined by false discovery rate (FDR) < 0.05 and fold change (FC) > 1.2. Random forest (RF) and logistic regression (LR) models incorporating selected biomarkers were employed to develop diagnostic models for EPL. In the external validation cohort, we prospectively enrolled 142 pregnancies at 7–10 gestational weeks, including 47 subjects who subsequently developed EPL and 95 pregnancies with full-term birth. Serum levels of selected biomarkers were quantified by ELISA.
Findings
The combined proteomics and metabolomics screening identified 26 proteins and 21 metabolites significantly changed in the EPL group and tightly associated with EPL-related clinical phenotypes, with functional enrichment in immunoregulation and lipid oxidation processes. Moreover, integrating serum levels of angiopoietin-like 4 (ANGPTL4), programmed death-ligand 1 (PD-L1), neutrophil%, and lymphocyte% achieved an AUC of 0.944 (95% CI: 0.835–1.000) in the random forest model and 0.954 (95% CI: 0.875–1.000) in the logistic regression model to discriminate EPL from HP. Importantly, this four-biomarker model achieved an AUC of 0.857 (95% CI: 0.747–0.968) in the random survival forest model and a C-index of 0.804 (95% CI: 0.685–0.973) in the validation cohort for EPL prediction.
Interpretation
Our integrative omics study reveals a panel of potential circulating biomarkers for EPL, which further offer mechanistic insights into EPL pathogenesis, including impaired maternal immune tolerance and dysregulated lipid metabolism pathways. Moreover, the newly identified biomarkers exhibit promising diagnostic and predictive performance for EPL, underscoring its clinical translational value for human reproduction and maternal–foetal health.