A proteomic signature to detect preeclampsia

Introduction: Preeclampsia (PE) is one of the most common pregnancy-related complications associated with significant maternal and fetal morbidity. Growing evidence suggests that PE and gestational diabetes mellitus (GDM) share metabolic, vascular, and inflammatory pathways. However, targeted proteomic signatures of PE have not been evaluated in longitudinal cohorts, so far.

Materials and Methods: A total of 92 women were recruited from the Department of Obstetrics at the University of Leipzig, comprising 46 women with PE and 46 healthy pregnant controls. PE was defined as gestational hypertension (systolic blood pressure (SBP)>140 mmHg or diastolic pressure (SBP)>90 mmHg, accompanied by proteinuria. Six months postpartum, fasting blood samples were collected from 86 women for re-examination. The Olink®​ Target 96 Cardiovascular III assay, including 92 biomarkers, was measured during and after pregnancy, respectively. Data reduction and biomarker identification were performed using principal component analysis (PCA), Mann–Whitney U tests, generalized linear models (GLM), linear discriminant analysis (LDA), and Partial least squares discriminant analysis (PLS-DA).

Results: At baseline, SBP, DBP, gestational age at delivery, birth weight, creatinine, markers of lipid and glucose metabolism, and C-reactive protein were significantly different in women with PE as compared to pregnant control women. Six months postpartum, body mass index, SBP, DBP, and fasting insulin remained elevated in the PE group, while creatinine, fasting glucose, and lipid profiles did not differ between the two groups. Exploratory, unsupervised multivariate PCA showed a clear separation of the PE and control group during pregnancy, but not postpartum. Based on the factor loadings, markers of coagulation (tPA, VWF), cardiovascular function/cell adhesion (NTproBNP, ICAM2, CDH5, SELP, OPN), and inflammation (ST2, NOTCH3) contributed most to the separation. PLS-DA identified NTproBNP, ST2, tPA, and NOTCH3 as the top biomarkers, which were 2.2-, 1.5-, 1.3-, and 1.2-times higher in PE compared to control women (p<0.001). These biomarkers also had the highest Odds Ratios for PE in GLM analysis: NTproBNP: 69.5 [95% Confidence interval: 9.3-519.3], tPA: 61.6 [10.9-348.5], ST2: 27.7 [7.1-107.4], NOTCH3: 8.1 [3.4-19.1]. LDA, only including tPA, ST2, NOTCH3, TRAP, BLMhydrolase, and VWF, achieved an AUC of 0.98 [0.94-1] to detect PE with tPA showing the highest discriminant coefficient.

Conclusion: GDM and PE share key pathophysiological mechanisms, including insulin resistance, endothelial dysfunction, and placental dysfunction. In this cohort, NTproBNP, ST2, tPA, and NOTCH3 involved in thrombotic, cardiovascular, and inflammatory pathways seemed to best distinguish PE from healthy, pregnant controls. Future studies on PE should focus on these biomarkers.

Publication History

Article published online:
28 May 2025