Recruitment to and clinical assessment of the prospective post-term reference group (GW ≥40⁺²)

Patients were recruited to the “Predelivery Placental Biomarkers–Pregnancy and Delivery Outcome (PREPPeD)” study from September 2016 to December 2017 at the Department of Obstetrics, Oslo University Hospital, Ullevål, delivering approximately 7100 women annually. Women with singleton pregnancies (GW ≥40⁺²), referred for routine clinical post-term evaluation, were included. Recruited women gave informed written consent. Exclusion criteria were: non Norwegian or English language, HIV and/or hepatitis, and age <18 years. According to Department protocol and national guidelines, women with low-risk pregnancies were referred to the outpatient unit between GW 41⁺² and 41⁺⁴, and seen every 2–3 days until delivery or induction of labor. Induction of labor was offered to women with fetal or maternal complications, or at GW 42⁺⁰. Women ≥40 years of age were routinely offered an earlier appointment around GW 40⁺² and induction of labor within GW 41⁺². At each visit, fetal well-being was routinely assessed by cardiotocography and an ultrasound biophysical profile [17,18]. Gestational age was calculated based on routine ultrasound screening at GW 17–20 according to Norwegian national pregnancy care routine, or when not available, from last menstrual period. For in vitro fertilization, gestational age was calculated from the date of embryonal transfer. Birth weight percentile was calculated according to Norwegian population-based sex- adjusted reference ranges [19].

Blood sampling and laboratory analyses for the prospective post-term reference group (GW ≥40⁺²)

A venous blood sample was taken at study inclusion and, if possible, daily until labor onset. The blood samples were left at room temperature for at least 30 min (max 2 hours), then centrifuged at 1800xg for 10 minutes. Serum samples were stored at -80° C until analysis (mean storage time 7.8 months). At inclusion, all women were assessed for glucose homeostasis (blood glucose and HbA1c) and kidney function (plasma creatinine).

In women with longitudinal post-term samples, the last one before labor was used to calculate the reference ranges. All samples were analyzed postpartum, blinded for clinical information at the Department of Medical Biochemistry, Oslo University Hospital, on a cobas e 801. The PIGF and sFlt-1 concentrations were quantified using the fully automated Elecsys system, according to the manufacturer´s instructions. All concentrations were within the measuring ranges of the Elecsys PlGF and sFlt-1 assays (3–10,000 pg/mL and 10–85,000 pg/mL, respectively). The coefficients of variation were ≤2.1% for PlGF and ≤1.8% for sFlt-1.

Selection of pregnancies contributing to the construction of the reference ranges for the prospective post-term group (GW ≥40⁺²)

Only post-term pregnancies with apparently well-functioning placentas contributed to the final maternal PlGF and sFlt-1 reference ranges. A “Diagnostic Advisory Group” (DAG), consisting of two senior consultant obstetricians, not affiliated to the PREPPeD study, and blinded for any biomarker results was established to ensure objectivity. In case of dissent regarding a pregnancy outcome categorization, a third senior consultant obstetrician reviewed and concluded the case. The DAG oversaw the process of patient grouping of all prospectively included post-term pregnancies (N = 501). Firstly, all pregnancies belonging to the "Uncomplicated group" were identified, which were the pregnancies without predefined complications and/or adverse outcomes (N = 339). Assignment to this group required thus a live born infant, no predefined pregnancy complication (birth weight <10^th percentile (SGA), preeclampsia (PE)/ gestational hypertension (GH)/ chronic hypertension [20], obstetric catastrophes (e.g. uterine rupture, cord prolapse or placental abruption), clinical FGR, (pre)gestational diabetes mellitus (DM) [21]), or any of the primary or secondary adverse PREPPeD delivery outcomes (as detailed in S1 Table). The pregnancies that had been recruited prior to gestational week 40⁺² were removed from the study (N = 10; all approved by the DAG). Among the remaining 152 pregnancies with "Predefined complications and/or adverse outcomes”, 54 were excluded due to either SGA, PE, GH, gestational DM, cord prolapse, or uterine rupture (all approved by the DAG). All remaining pregnancies were assigned to the “Complicated group” (N = 98) (Fig 1). These 98 pregnancies and outcomes were scrutinized in detail by the DAG that reviewed all available clinical information including partogram details, umbilical cord blood gases and placental histology, if available. If the DAG concluded that the pregnancy outcome in the “Complicated group” was likely associated with placental dysfunction (“placental cause for the adverse outcome”; N = 11), this pregnancy was excluded from the final group contributing to the biomarker reference ranges. On the other hand, if the DAG evaluated the adverse outcome being most likely of non-placental cause (n = 61) or undetermined (n = 26), this pregnancy was included in the final group contributing to the reference values (Fig 1: “Final uncomplicated group”; N = 426).

Download:

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

Fig 1. Participant flow chart of the prospective post-term group (GW 40⁺²–42⁺²), resulting in the “Final uncomplicated group” of clinically healthy pregnancies without adverse outcomes*.

*PREPPeD study primary and/or secondary outcome(s). DAG, Diagnostic Advisory Group; GW, Gestational week; PE, Preeclampsia; GDM, Gestational diabetes mellitus; GH, Gestational hypertension; SGA, Small for gestational age.

https://doi.org/10.1371/journal.pone.0240473.g001

Comparison with data from retrospective term group (37⁺⁰ to 40⁺⁰)

Previously unpublished PlGF and sFlt-1 data from our retrospective term group (recruited to “Oslo Pregnancy Biobank” studies) served as a term comparison group to our post-term group. All women had given informed written consent. Serum from 146 apparently healthy, normotensive and euglycemic women with uncomplicated pregnancies (GW 37⁺⁰–40⁺⁰) was obtained prior to elective cesarean delivery (due to maternal request, breech presentation or repeated cesarean section) at our hospital (S2 Table). Mean sample storage time was 5.9 years before analysis for PlGF and sFlt-1, on a Roche, Elecsys 2010 Modular Analytics E170 or a cobas e601.

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp. The biomarker data were skewed and thus Log transformed prior to analyses. Means were compared by two-sample T-tests and categorical variables by Chi-square tests. Differences in percentile values between two independent groups and corresponding confidence intervals (corrected for multiple testing using Bonferroni correction) were calculated by the quantreg package for quantile regression in the statistical software R [22].

Ethics

National research ethical and institutional bodies have approved the PREPPeD (PREdelivery Placental biomarkers–Pregnancy and Delivery outcome) study (The Regional Committee for Medical and Health Research Ethics in South-Eastern Norway, ref. 2016/652), which the present study is a part of. The PREPPeD biobank is coordinated as a thematic biobank within the Oslo Pregnancy Biobank (OPB; The Regional Committee for Medical and Health Research Ethics in Eastern Norway, ref. 529–02162).

Trial Registration: ClinicalTrials.gov, reference number NCT03100084 (URL access on https://clinicaltrials.gov/ct2/show/NCT03100084)

Main findings

To the best of our knowledge, this is the first report of reference ranges for maternal circulating PlGF and sFlt-1 concentrations, as well as sFlt-1/PlGF ratio, in a prospectively collected, large group of healthy post-term pregnancies without adverse pregnancy outcomes. Two prior studies that included fewer post-term data did either not separately analyze the post-term groups [9,16], or only analyzed PlGF. Further, those studies did not exclude pregnancies with adverse pregnancy outcomes likely due to placental stress (and therefore likely altered biomarkers).

We observed a trend towards lower 70^th to 95^th percentiles of maternal circulating PlGF and significantly higher sFlt-1 10^th to 80^th percentiles in our post-term pregnancy group as compared to our retrospective term group. The antiangiogenic sFlt-1/PlGF ratio percentile was significantly higher for 30th percentile in the post-term group, but lower for the 95^th percentile, possibly indicating a less heterogeneous, but still elevated antiangiogenic state in clinically healthy post-term pregnancy.

Interpretation

According to our hypothesis, cellular syncytiotrophoblast stress [3,8] and senescence [7] increase at and beyond term, with lower PlGF and higher sFlt-1 representing markers of cellular stress [3] even in pregnancies that appear to be uncomplicated. In line with this concept, we observe reduced maternal PlGF and increased sFlt-1 concentrations and sFlt-1/PlGF ratios post term in clinically healthy pregnancies without adverse fetal and maternal outcomes, when compared to term values.

The course of maternal circulating free PlGF concentrations throughout pregnancy is well described, with rising levels towards GW 29–32, and decreasing towards term [9,23], as well as a sharp increase in third trimester and towards term for sFlt-1 in a cross-sectional data set for each gestational age group [9]. Both findings were reproduced by Verlohren et al. in a longitudinal study in a demographically similar patient group to ours, applying the same analytical biomarker system [16] as us. Both Levine et al. and Verlohren et al. included fewer pregnancies ≥ GW 40⁺² [9,16], as compared to our large post term group (n = 426), and neither of them specifically addressed post-term placenta-associated biomarker alterations in healthy pregnancies recruited in an ordinary clinical setting. The recent paper by Dunn et al reported decreasing PlGF levels by gestational age towards term and post-term [24], but did not exclude pregnancies with an adverse outcome most likely associated with placental dysfunction from their cohort contributing to their reference percentiles.

When comparing clinical characteristics, as well as sFlt-1 and PlGF concentrations from our retrospective term group (GW 37⁺⁰–40⁺⁰) with the Verlohren term group (GW > 37⁺⁰) [16], the data sets are comparable clinically, as were the 5^th, 50^th, and 95^th percentiles of sFlt-1, PlGF and sFlt-1/PlGF ratio. We therefore allowed us to compare our retrospective term group (GW 37⁺⁰–40⁺⁰) with the results from our prospective post-term (GW ≥40⁺²) group, assuming a reliable reflection of the longitudinal pattern changes of the angiogenic biomarkers from term to post-term in clinically healthy pregnancies. The narrower span in our post-term group between the 5^th and 95^th PlGF percentiles and observed lower PlGF levels for the 70^th through 95^th percentiles compared to the term group, although not significant after a conservative Bonferroni’s correction, are also in line with our hypothesis of an increasingly stressed placenta over the last weeks of a clinically uncomplicated pregnancy, with additional syncytiotrophoblast stress in post-term pregnancy. Our finding of even higher sFlt-1 values for all percentiles in healthy post-term pregnancies compared to term pregnancies supports our concept of increasing syncytiotrophoblast stress towards post-term pregnancy, even in those with a healthy clinical outcome.

Compared to our retrospective term pregnancy group, the 30^th percentile for the sFlt-1/PlGF ratio was significantly increased in the post-term group, whereas the 95^th percentile ratio decreased. Interestingly, we observed a trend towards higher values for the lower PlGF percentiles in the post term compared to the term group. These findings, together with the similar rate of post-term pregnancies with low antiangiogenic ratio (sFlt-1/PlGF <38) as for our retrospective term group may be due to the implicit selection bias in our study, since only the healthiest pregnancies at term (with assumedly less dysregulated angiogenic biomarkers than those already delivered) are allowed to proceed, such as the post-term women recruited to the present study. Further, there is progressive depletion of women from our study by delivery.

Our observation of a slightly shorter time to delivery in post-term healthy pregnancies with PlGF values below the 5^th percentile or sFlt-1/PlGF ratio above the 95^th is novel and not reported for healthy pregnancies before. However, further analyses in larger cohorts dichotomized into spontaneous and induced labor are warranted. Whether dysregulated angiogenic proteins at post-term might reflect increasing placental syncytiotrophoblast stress and placental membrane inflammation, promoting imminent labor onset, is an exciting concept that merits further investigation.

Our biomarker findings are consistent with increasing syncytiotrophoblast stress beyond term and imply that all pregnancies may eventually develop placental dysfunction syndromes, had the pregnancy and offspring not been “rescued” by delivery [3].

Study strengths and limitations

Our post-term study population from a real-world unselected obstetric hospital setting was prospectively recruited and extensively clinically phenotyped, and larger than previous post-term studies [9,16,24]. Our results may therefore serve as a reference for other post-term pregnant populations. All clinical decisions were made according to Department protocol, blinded for the biomarker results. The biomarkers were analyzed collectively postpartum, blinded for clinical outcomes. All pregnancies and delivery clinical outcomes were reviewed by an obstetrical expert group blinded for biomarker results. Differences in mean storage time for the term and the post-term study groups before biomarker analyses (mean storage time 5.9 years versus 7.8 months) may be viewed as a limitation. However, PlGF and sFlt-1 serum protein levels have been proven to be stable for many years [25]. Our biobank follows strict operating procedures and samples are handled and stored in a standardized manner by a limited number of experienced study personnel. Limitations for external validity include a low ethnic heterogeneity and a large percentage of highly educated women, partly explained by the inclusion criteria (Norwegian or English language).