Non-Invasive Biomarker Detection for Pregnancy Complications Using Cell-Free RNA

Abstract

Pregnancy complications pose a significant threat to maternal health as they may result in a higher risk for issues during pregnancy or labor relative to the risk for these issues in a typical pregnancy. Many cases of maternal deaths and complicated pregnancies can be avoided with a richer understanding of maternal health early on in pregnancy. Genetic analysis of fetal DNA in maternal blood is becoming increasingly common1, and while genetic screening efforts have progressed substantially in recent years, they have focused on fetal health rather than the health of the mother2. This work focuses on the detection of common complications of pregnancy including preeclampsia, gestational diabetes, and chronic hypertension using non-invasive circulating cell-free RNA data. We developed interpretable supervised machine learning methods that had high performance in identifying pregnancy complications from healthy pregnancies (AUC = 0.86). Using our models, we found various relevant transcripts, related to pregnancy biology. These included S100A9, which encodes for a protein involved in inflammation and was elevated in complicated pregnancies, as well as two small RNAs involved in cell proliferation and body mass, RNY4 and RNY3, which were reduced in preeclampsia and GDM and have previous roles in pregnancy. Our findings highlight several promising non-invasive biomarkers for the early diagnosis of complications of pregnancy that have the potential to be easily integrated into existing clinical workflows.