Introduction

Preeclampsia affects at least 2–3% of all pregnancies and is a major cause of maternal and perinatal morbidity and mortality [1]. The condition is recognized clinically after 20 weeks of gestation with the new appearance of hypertension and proteinuria. In countries with limited access to medical care, it is estimated that the disorder is responsible annually for greater than 60,000 deaths worldwide [2]. In developed countries, therapeutic intervention is often concluded with early delivery. While this intervention protects the mother, it results in significant morbidity and mortality to the neonate [3]. Early diagnosis has been a goal permitting intervention at an early time point [4].

Preeclampsia is among the great obstetrical syndromes that include premature rupture of the membranes, intrauterine growth retardation, premature birth and miscarriage. Khong, Brosens and Romero have identified histologic features of inadequate spiral artery transformation that are shared by each of the great obstetrical syndromes [5, 6]. Malperfusion results in placental stress from the beginning of the second trimester. Investigators have identified shed placental components in the plasma that correlate with malperfusion. The shed placental components have been used as biomarkers, often in combination with ultrasound and history, as predictors of the development of the great obstetrical syndromes.

Sweeting et al. note that a diagnostic regimen performed during the same time window as NIPT (non-invasive prenatal testing utilizing plasma DNA) would provide clinicians with important information at the outset of pregnancy management [7]. Plasma has been studied extensively as a carrier of material shed from tumors allowing for their non-invasive assessment, the so-called “liquid biopsy” [8]. Likewise, quantification of biomarkers within plasma such as PLGF, sFLT-1 and sENG [9,10,11] have been applied to pregnancy outcome prediction during the second and third trimesters [12, 13]. However, these markers can only identify 35% of all preeclampsia and approximately 40% of preterm births, at false-positive rate of 10% [14, 15, 16]. None of these markers can predict preeclampsia during the first trimester when the process of maladaptive placentation is initiated [17, 18].

Romero et al. identify events within the placental bed that first affects the placenta during the second trimester [5]. They emphasize that inadequate remodeling of spiral arteries within the placental bed leads to placental malperfusion and consequent placental dysfunction after the initiation of intervillous blood flow at the end of the first trimester. Maternal immune cells present from the beginning of pregnancy appear to have both a direct role in this spiral artery remodeling and in subsequent guidance of trophoblast activity [7, 19]. Thus, successful spiral artery remodeling involves maternal immune cells in addition to the well-understood role of trophoblast.

Prediction of preeclampsia during the first trimester by assessment of shed placental components is impaired. Prior to the second trimester, the spiral arteries of the placental bed are plugged by trophoblasts. During the majority of the first trimester, the placenta relies on histiotrophic nutrition and exists in a state of physiologic hypoxia, a state shared by all placentas until weeks 11 through 14 of gestation. After this, the trophoblast plugs begin to dissipate [20] initiating blood flow, increasing oxygen levels, producing a mature, non-proliferative, invasive trophoblast phenotype [21]. Nevertheless, significant effort has been made at discovering serum-based biomarkers for first trimester prediction of preeclampsia [22]. However, we speculate that the placenta destined to healthy as well as compromised outcome experiences essentially the same hypoxic conditions while spiral arteries remain plugged. Thus, the placenta during the first trimester would not be expected to differentially shed biomarkers into the plasma. Quantification of peripheral blood cell microRNA may offer a new approach.

MicroRNA is a class of RNA species comprising a 22–24 base non-coding polynucleotide. They integrate disparate genetic elements into collaborative metabolic and signaling pathways. They form networks that supervise coordinated expression of mRNAs that guide and maintain cell identity and buffer cell systems against changing conditions. MicroRNA has attracted great interest in the diagnosis and monitoring of various conditions including cancer, autoimmune, inflammatory and neurologic diseases [23].

In our previous studies, we found that first trimester peripheral blood mononuclear cell (PBMC) microRNA provides sensitive and specific prediction of preeclampsia and preterm birth when sampled within a range of 4–11 weeks gestation [24, 25]. In our present study, we wished to widen our first trimester microRNA sample collection to include specimens collected at a later time-point and to demonstrate that buffy coat samples may also be used. Samples from a biorepository of buffy coat specimens maintained at King’s College London were used in contrast to the PBMC sample collection interrogated in our prior studies. The sample archive utilized was collected at a later time point, at a range of 11–13 weeks gestation. Together, we hoped to demonstrate that quantification of first trimester peripheral blood cell microRNA, prepared as either PBMCs or as buffy coat is predictive of various disorders involving the placental bed throughout the first trimester.

Materials and methods

Study population and consent

MicroRNA quantification was performed on maternal buffy coat blood specimens obtained between 11–13 weeks gestation from women with singleton pregnancies attending their routine first hospital visit in pregnancy. Patients represented a non-selected patient population within the patient catchment region around King’s College London Hospital. “Normal delivery” was defined as the delivery of a singleton, normal karyotype baby with the following pregnancy criteria: delivery at 38–42 weeks gestation, baby weight within the normal range for gestational age and maternal BMI <30. Preeclampsia was defined according to the guidelines of the International Society for the Study of Hypertension in Pregnancy [26]. See Table 1 for population details.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study population.

https://doi.org/10.1371/journal.pone.0190654.t001

Venous blood was obtained from women who had given written informed consent to provide samples for research as approved by the National Research Ethics Service of the National Health Service. Blood was collected in EDTA Vacutainer tubes (Becton Dickinson UK Ltd, Oxfordshire, United Kingdom) and processed within 15 minutes of collection. The tubes were centrifuged at 2000g for 10 minutes. Buffy coat was collected by pipette aspiration guided visually and frozen immediately at -80°C without an RNA preservative, labeled with a unique patient identifier and maintained for up to nine years. Specimens were shipped from King's College London (KCL), UK directly to the Stanford Human Immune Monitoring Center (Stanford, California, USA) on dry ice where they remained blinded as to clinical outcome through testing, identified only by a unique identification number. The study was a retrospective analysis using clinical data from patient charts and specimens frozen and stored as buffy coat. Institutional Review Board approval for retrospective analysis of first trimester peripheral blood specimens was obtained (Western Institutional Review Board (WIRB), Study Number 1151255, Pro. Number 20142368). Patient identifying information was maintained in accordance with HIPAA requirements.

Development of the microRNA scoring system using the training set

MicroRNA Ct values were analyzed by Medcalc^® software generating AUC-ROC (Area Under the Curve-Receiver Operating Characteristic) values, their associated p values and Youden Index J Associated Criterion values for each microRNA in the training set (Table 2). The Youden Index J Associated Criterion Value for each microRNA was used as its positive/negative cut-off. A panel of microRNAs was established that included only those microRNAs where p ≤ 0.05. The panel and calculated cut-off points were then applied to the validation set. Ct values less than the cut-off value for each of the microRNAs in the panel were assigned a score of “1”. The sums of the individual microRNA scores for each sample in the validation set were designated the “Risk Scores”. An AUC-ROC was calculated from the Risk Scores for the validation set.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Defining the final microRNA panel and positive/negative cutoffs using the training set.

https://doi.org/10.1371/journal.pone.0190654.t002

Results

Application of the 8 microRNA panel Risk Score to the 48 patient population

The panel and cutoff criteria were then applied to each of the 48 patients (Table 3). A ROC curve was calculated using microRNA Risk Scores. AUC-ROC for preeclampsia risk was 0.91 (p<0.0001). When women with normal delivery and high-risk background (those with SLE/APS, chronic hypertension and/or Type 2 Diabetes) were compared to women who developed preeclampsia but with a normal risk background (without these listed risk factors), preeclampsia was still predicted with an AUC-ROC of 0.92 (p<0.0001) (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Preeclampsia risk assessment using 8 microRNAs to calculate Risk Scores in 48 patients.

https://doi.org/10.1371/journal.pone.0190654.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. AUC-ROC calculations for preeclampsia study populations.

https://doi.org/10.1371/journal.pone.0190654.t004

Discussion

Our study is the first successfully to use microRNA in peripheral blood to predict preeclampsia outcome using buffy coat specimens during the first trimester. Our present data extend our previous findings in several ways. First, we have shown that we can identify patients suffering preeclampsia risk from normal risk population when sampled at a mean of 12.7±0.5 weeks gestation. Second, we have shown that we can distinguish preeclampsia and a high-risk population comprising autoimmune disease, type 2 diabetes or chronic hypertension (Table 4, column B). Finally, samples were prepared as buffy coats rather than as PBMCs as in our previous studies demonstrating that buffy coat specimens are suitable for assessment.

Our current study is supported by our previous published findings [24, 25, 27]. Samples were collected at various time-points through the first trimester. The same analytic methods were used throughout the four studies and provided consistent results. AUC-ROC values were previously calculated at 0.90 or greater for preeclampsia, preterm birth and miscarriage using PBMC-prepared specimens [24, 25, 27].

Several members of the microRNA panel that we found predictive of preeclampsia risk may have functions associated with preeclampsia’s pathogenesis. For example, miR-210 can down regulate PTPN2 mRNA contributing to the pathogenesis of preeclampsia and has a central role in the switch from trophoblast proliferation to an invasive phenotype [28]. MicroRNA-148a has been shown to downregulate HLA-G expression which may prevent inhibition of NK cell killing through the LILRB1 inhibitory receptor [29]. MicroRNA-582-3p has been demonstrated to suppress EG-VEGF expression and inhibit trophoblast invasion and migration. From this, we are tempted to speculate that decidual immune cells may have a unique role prior to the recruitment of trophoblast to the spiral artery during the post-implantation period [30].

Immune cells within the decidua participate in the regulation of immune tolerance as well as providing essential support in the remodeling of spiral arteries within the placental bed [31]. Interrogation of immune cells within the decidua might provide predictive information about the future course of a pregnancy. However, acquisition of the decidual tissue is invasive, limiting any such immune investigation to selected, high risk patients. Our studies quantifying microRNA within peripheral blood cells during the first trimester offer a promising alternative [24, 25, 27].

The study had some limitations. Because of the small sample set, following validation, the criteria established with the training set were then applied to the combined data set introducing potential overfitting of the data. A larger population size is needed to confirm clinical utility. Also, most of the buffy coat specimens were frozen for many years without an RNA preservative.

In conclusion, our data suggest that the identification of a maternal immune cell surrogate provides a promising new path to early prediction and intervention. While permitting concurrent assessment along with NIPT, such a tool would also permit assessment of pregnancy risk both in the first trimester in the obstetric clinic and following embryo transfer at IVF clinic. As therapies directed to prevention of the various disorders of the placental bed become available, earlier intervention and treatment becomes possible. We suggest further studies to confirm and extend our findings.

Supporting information

Acknowledgments

We thank Argyro Syngelaki at King’s College Hospital, London, UK for her assistance collecting and compiling patient data.