Improvement in rates of maternal morbidity and mortality has been limited by the inability to assess fetal and maternal tissues before symptoms develop. Thomas McElrath, MD, PhD, an obstetrician in the Division of Maternal-Fetal Medicine at Brigham and Women’s Hospital, Michal A. Elovitz, MD, of the University of Pennsylvania, and colleagues have shown in a publication in Nature that the patterns of cell-free RNA (cfRNA) from maternal plasma progress in a predictable gestational-age dependent fashion. As such, these molecular signatures can be used to survey the gestational age of the pregnancy as well as the overall wellbeing of mother, placenta and baby. The team demonstrated this concept by predicting the risk of preeclampsia, a hypertensive condition unique to pregnancy, on average 14 weeks prior to the development of symptoms.
The researchers measured cfRNA in 2,539 plasma samples drawn from 1,840 pregnant women of multiple ethnicities, nationalities, geographic locations and socioeconomic circumstances, while covering a range of gestational ages. This was the largest and most diverse dataset of maternal transcriptomes ever analyzed.
Measuring Gestational Age
The researchers first restricted their study to plasma samples from healthy pregnancies, building a machine learning model of normal cfRNA patterns in uncomplicated pregnancies. By design, these models also served to predict gestational age. A cfRNA signature was as accurate as second-trimester ultrasound and superior to third-trimester ultrasound while also providing insights into the biology of pregnancy progression.
Thus, the model could offer alternative dating for women who start prenatal care later in pregnancy. Its predictions were driven almost entirely by information from the cfRNA transcripts, as body mass index, maternal age and race accounted for less than 1% of the variance.
A Window into Maternal–Fetal Development
cfRNA profiles also made it possible to assess the molecular status of the placenta, fetal organs, cervix and uterus. Hundreds of independently identified gene sets in maternal blood mirrored the maternal and fetal physiological changes expected during pregnancy.
A study of three independent cohorts verified that multiple gene sets were uniquely associated with specific tissues of origin. These included the uterus and cervix as well as the placenta. Most intriguingly, however, the team also detected the unique signature of fetal tissues including fetal heart, GI tract and kidneys. Thus, the use of cfRNA signatures could reveal and characterize molecular changes in the maternal–fetal dyad during gestation.
Early Prediction of Preeclampsia
As an example of predicting adverse outcomes, the team evaluated the ability of cfRNA signatures in maternal blood during the second trimester to identify women at risk of preeclampsia. A case–control study included 72 women with preeclampsia and 452 without, selected from two independent cohorts.
Correlation tests identified signatures that separated the cases and controls and identified seven genes consistently associated with preeclampsia. A screening test based on those genes achieved a sensitivity of 75%, with an area under the receiver operating curve of 0.82.
The inclusion of maternal body mass index, age and race had no effect on the new test’s performance.
Women who tested positive delivered significantly earlier during gestation than women who tested negative. In addition, a positive test correctly identified 73% of women who had a medically indicated preterm birth—more than three months ahead of clinical symptoms or delivery.
Hypertension vs. Normotension
In pregnant women who have preexisting hypertension, it can be quite challenging to distinguish superimposed preeclampsia from exacerbation of hypertension. The difference is important because one requires delivery for a cure and the other usually doesn’t.
One of the cohorts used to estimate preeclampsia risk included 50 women with hypertension (chronic hypertension, n=31; gestational, n=19) and 263 normotensive women. Genes identified by comparing the two groups showed no overlap with genes significant for preeclampsia. Also, none of the genes were differentially expressed. Thus, the molecular signal for preeclampsia was specific to hypertension driven by a placental disorder.
Considering the large, diverse study population, it’s noteworthy that race had almost no effect on gestational age estimates or preeclampsia risk evaluation. Including race in clinical risk assessment has been shown in several fields to be problematic, have little or no utility, and in fact represent a source of latent bias against members of underrepresented minority communities. In contrast, the cfRNA signature directly exposes the development of preeclampsia.
A better understanding of the maternal–fetal–placental transcriptome should lead to precision therapeutic interventions that can target molecular subtypes of preeclampsia and preterm birth.