Early Prediction of Preeclampsia Based on Transposable Elements signature in cell-free RNA

Abstract

Preeclampsia (PE) is a pregnancy-associated hypertension disorder affecting 5-10% of pregnant women each year worldwide, which leads to adverse maternal and child outcomes. PE remains inadequately predicted, prevented and treated. Here, we comprehensively investigated altered transcriptomic and epigenetic changes in the PE placenta and maternal cell-free RNA (cfRNA). We show that many transposable elements (TEs) in sub-families are deregulated in the gestational age-matched PE placenta. Increased expression of endogenous retrovirus (ERV) and long interspersed nuclear element 1 (L1, LINE1) subfamilies of TEs is associated with higher histone acetylation levels at their regulatory elements. A higher TE transcript level correlates with the type I interferon (IFN-I) pathway, suggesting inflammation associated with PE could be due to the sensing of TE transcripts by the antiviral innate immune system. Consistent with higher TE transcript levels in the PE placentas, analysis of maternal cfRNA data revealed differential levels of TE subfamilies in PE compared to healthy controls. Machine learning-based training for TE transcripts for early prediction of PE showed a robust performance in the validation cohort, with an area under the curve (AUC) of 0.88, 81% sensitivity, and 74% positive predictive value (PPV). Overall, we show TE deregulation in the placenta is associated with PE, and maternal cfRNA TE signature accurately predicts early diagnosis of PE, which can improve prophylaxis and obstetric outcomes.