Early maternal plasma protein signatures identify preeclampsia risk in twin pregnancies

Background

Twin pregnancies have a two- to three-fold higher risk of preeclampsia (PE), but early prediction remains poor because current biomarkers and prediction models are largely derived from singleton pregnancies. Defining the biological changes that precede PE in twins may facilitate the development of more effective risk assessment approaches.

Objective

We aimed to characterize maternal plasma protein changes preceding PE in twin pregnancies and to explore whether early circulating proteins could serve as improved risk stratification for twin PE. STUDY

Design

We performed a nested case-control study within a prospective twin pregnancy cohort at Peking University Third Hospital. A total of 84 women with twin pregnancies were included, comprising 40 pregnancies complicated by PE and 44 normotensive controls. Longitudinal maternal plasma samples were collected in the first, second, and third trimesters. In the discovery cohort, maternal plasma proteins were profiled by mass spectrometry to identify trimester-specific protein changes associated with PE. Candidate proteins were then validated in an independent validation set. We further assessed the functional relevance of selected proteins in trophoblast and endothelial cell models and constructed prediction models integrating biomarkers with maternal clinical characteristics.

Results

Twin pregnancies complicated by PE showed distinct trimester-dependent protein alterations, with first- and second-trimester changes mainly involving immune regulation and glycolipid metabolism, and later changes enriched in cardiovascular-related pathways. Several proteins demonstrated early discriminatory patterns, including increased annexin A3, chitotriosidase-1, and ferritin light chain (FTL), together with decreased annexin A2 in the first trimester, whereas the conventional singleton markers (sFlt-1 and PAPP-A) did not differ. ELISA validation of these candidate biomarkers yielded an AUC of 0.93 for twin preeclampsia prediction; integration with standard maternal clinical factors further improved discriminatory performance to an AUC of 0.97. Accordingly, a multivariable model combining annexin A2, annexin A3, chitotriosidase-1, FTL, and maternal clinical factors achieved strong internal performance for early discrimination of PE in twin pregnancies (AUC, 0.97; 95% CI, 0.93-1.00). Moreover, annexin A2 knockdown impaired trophoblast migration and invasion and reduced inflammatory cytokine secretion in endothelial cells, supporting a role in placental development and maternal vascular adaptation.

Conclusion

Twin pregnancies that later develop PE exhibit early and progressive maternal plasma protein alterations that can be detected before clinical disease onset. These findings identify candidate biomarkers and biologically relevant pathways that may help explain why PE in twins differs from singleton disease and may support future twin-specific risk stratification strategies. These results require external validation before clinical application.