Placental growth factor testing at 19–23 weeks of gestation as a guide to subsequent care in pregnancy: A prospective observational study

To determine whether serum placental growth factor (PlGF) at 19–23 weeks of gestation can improve the identification of risk for adverse outcomes.


| I N TRODUC TION
The placental disorders of pre-eclampsia (PE) and fetal growth restriction (FGR) are leading causes of maternal and fetal/newborn mortality and morbidity, worldwide.A key objective of antenatal care is to identify women and babies at increased risk for the development of these conditions, to provide preventative therapy (for PE) and enhanced maternal and fetal surveillance (for PE and FGR).
National organisations in the UK recommend that pregnancies at increased risk of PE or FGR be identified using clinical risk factors.The National Institute for Health and Care Excellent (NICE) recommends that women with one 'major' or at least two 'moderate' risk factors for PE be identified at antenatal care booking and offered low-dose aspirin. 1 In women screened using an alternative, 'competing risks' strategy, low-dose aspirin reduces the risk of preterm PE by more than 60%. 2 The Royal College of Obstetricians and Gynaecologists (RCOG) recommends that pregnancies with one 'major' or at least three 'minor' risk factors for FGR be offered enhanced ultrasonographic surveillance of fetal growth and well-being. 3However, each of these screening strategies is known to detect less than half of pregnancies that will end in adverse placental outcomes. 4,5s low serum placental growth factor (PlGF) has been associated with both PE and FGR, we evaluated whether it could be measured at 19-23 weeks of gestation, in conjunction with the routine ultrasound assessment for fetal anomalies, as a 'contingency screening tool' to improve the predictive performance of clinical risk factors for the subsequent development of PE and/or FGR. 6

| Study design and participants
The study data for this secondary analysis were derived from a prospective screening study for adverse pregnancy outcomes in women attending routine pregnancy care at 19-23 weeks of gestation at two maternity hospitals in England (King's College Hospital and Medway Maritime Hospital), from October 2011 to March 2020, inclusive.Women gave written informed consent to participate, and the study was approved by the National Health Service Research Ethics Committee (ref.02-03-033).There was no patient involvement in the study.
At the routine hospital visit at 19-23 weeks of gestation, women underwent an assessment that included the recording of maternal demographics and medical history and the measurement of serum PlGF (as described below).Gestational age was determined by the measurement of the fetal crown-rump length at 11-13 weeks of gestation. 7ncluded in the study were singleton pregnancies delivering a nonmalformed live birth or stillbirth at ≥24 0/7 weeks of gestation.Excluded were pregnancies with aneuploidy or major fetal abnormalities.

| Risk of PE or FGR
Pregnancies were considered at increased risk of PE or FGR according to contemporaneous UK national guidance, as summarised in Table S1.
According to the 2019 NICE guidelines, 1 women are considered at high risk of PE if they have at least one high or two moderate risk factors (Table S1).The five high risk factors are: hypertensive disease in previous pregnancy; chronic hypertension; diabetes mellitus; chronic kidney disease; and autoimmune disease.The five moderate risk factors are: first pregnancy; age of >40 years; body mass index (BMI) at first visit of >35 kg/m 2 ; interpregnancy interval of >10 years; and family history of PE.
According to the 2014 RCOG guidance, 3 pregnancies are considered at increased risk of FGR if they have at least one major or three or more minor risk factors (Table S1).The major risk factors are: maternal age >40 years; BMI ≥ 35 kg/ m 2 ; chronic hypertension; type 1 or 2 diabetes mellitus with vascular disease; chronic kidney disease; antiphospholipid antibody syndrome; prior small for gestational age (SGA) fetus; prior stillbirth; smoker of ≥11 cigarettes/day; cocaine use; maternal or paternal SGA; daily vigorous exercise; heavy bleeding; uterine artery pulsatility index (UtA-PI) of >95th percentile; and serum pregnancy-associated plasma protein A (PAPP-A) of <0.4 MoM (multiple of medians) in the first trimester.The minor risk factors are: maternal age of ≥35 years; a BMI of <20 or 25-34.9kg/m 2 ; PE during a previous pregnancy; smoking 1-10 cigarettes/day; nulliparity; an interpregnancy interval of <6 or ≥60 months; conception by in vitro fertilisation; and low fruit intake before pregnancy.

| Serum PlGF
Placental growth factor was considered low if the level was below the fifth percentile (<108 pg/mL) for the study population, as measured by BRAHMS Kryptor compact PLUS (ThermoFisher Scientific, Waltham, MA, USA).

| Outcome measures
Adverse outcomes of interest were ascertained by health record review of electronic hospital maternity records or records held by the women's general medical practitioners.
PE and gestational hypertension (GH) were defined by the American College of Obstetricians and Gynaecologists 2013 criteria. 8GH was defined as a systolic blood pressure (BP) ≥140 mmHg or a diastolic BP of ≥90 mmHg, measured twice, at least 4 h apart.PE was chronic or gestational hypertension with the development of at least one of the following: new-onset proteinuria; serum creatinine of >97 mmol/L (in the absence of underlying renal disease); serum transaminases more than twice the upper limit of normal (i.e.≥65 IU/L for our laboratory); platelet count of <100 000/mL; headache or visual symptoms; or pulmonary oedema.The maternity records of all women with pregnancy hypertension were examined to determine the diagnosis of PE or GH.The contemporaneous management of hypertension was to initiate antihypertensive therapy for a BP of 150/100 mmHg, 9 before the guidance was changed to target a BP of ≤135/85 mmHg. 1 Birthweights below the third percentile for gestational age were based on the Fetal Medicine Foundation (FMF) fetal and neonatal weight charts. 10Other outcomes were stillbirth (defined as the birth of a fetus at ≥24 +0 weeks of gestation with no signs of life) and neonatal intensive care unit (NICU) admission for ≥48 h.A core outcome set was not used, as the core outcome set for obstetric studies is currently in development.

| Statistical analysis
Descriptive analysis was undertaken for baseline data, PlGF measured at 19 +0 -23 +6 weeks of gestation and pregnancy outcomes for the study population overall, for those with PlGF below the fifth percentile and for those deemed at risk of PE or FGR according to clinical risk factors.Continuous variables were summarised by medians and interquartile ranges, and categorical variables were presented as numbers (percentages).
Diagnostic test properties were determined for the following approaches: (i) clinical risk factor screening with one or more PE or FGR clinical risk factors; (ii) low PlGF at 19-23 weeks of gestation (for comparison); and (iii) a twostage approach, first with clinical risk factor screening and then, for screen-positive individuals, with PlGF testing.Detection rates for each adverse outcome were calculated, overall and according to gestational age at birth, as <32, <34, <37 and ≥37 weeks of gestation.The way in which testing could inform care was interpreted as the ability to provide meaningful reassurance, taken as a negative likelihood ratio (−LR) of <0.20, or meaningful concern, taken as a positive LR (+LR) of ≥5.0, for adverse pregnancy outcomes. 11No correction was made for multiple testing.

| R E SU LTS
Our cohort comprised 30 013 unselected pregnancies at 19-23 weeks of gestation.Table 1 shows that, on average, pregnant women were in their early 30s, with a BMI near the threshold of normal and overweight.The population was ethnically diverse, with 21% from non-white ethnicities.A small proportion were smokers.Few pregnancies were complicated by chronic medical conditions.Few women had a family history of PE.Just over half of pregnant women were parous, among whom 2.1% had experienced prior PE and 6.7% had delivered a baby with a birthweight below the tenth percentile.Prior pregnancies occurred 2-3 years previously.Few women conceived by assisted means.Only 4.2% of the women were taking lowdose aspirin.
Placental growth factor was low, as defined above, in 1501 (5.0%) pregnancies (Table 1).One-third of pregnancies had one or more clinical risk factors for PE or FGR; 10.5% had a risk factor PE (usually two or more moderate risk factors) and 29.3% had a risk factor for FGR (usually a major risk factor).Women with low PlGF or women with one or more clinical risk factors for PE or FGR appeared to differ from the population overall, with respect to many characteristics associated with the risk of placental disease, including maternal age, BMI, ethnicity, medical and family history, parity, previous obstetric complications, assisted conception and aspirin therapy.
On average, pregnancies ended at almost 40 weeks of gestation, with 5.8% preterm deliveries and 21.9% labour inductions.The mode of birth was usually vaginal.PE developed in 2.9%, GH developed in 2.8%, stillbirth occurred in 0.2%, birthweights below the third percentile occurred in 4.7% and NICU admission for ≥48 h occurred in 8.5% of pregnancies.Of the 69 stillbirths, 36 (52.2%) were associated with PE and/or with birthweights below the tenth percentile.Women with low PlGF or with one or more clinical risk factors for PE or FGR appeared to have more complicated pregnancies, including more interventions, PE, GH, stillbirth, birthweight below the third percentile and prolonged NICU admission.
Table 2 shows that a primary screening strategy using clinical risk factors for PE or FGR had a screen-positive rate of 33%, but an associated detection rate of no better than approximately 50% for any of the adverse outcomes evaluated, with detection rates of approximately 60% for preterm disease.Importantly, for none of the adverse outcomes at any gestational age could clinical risk factors raise the level of concern or reassure women and their healthcare providers.
Table 3 shows that a primary screening strategy using PlGF at 19-23 weeks of gestation had a screen-positive rate of 5% (as defined).The associated detection rate was no better than 17% for any of the adverse outcomes of interest, although it varied and was much higher for delivery with adverse outcomes at <34 weeks of gestation.However, low PlGF could meaningfully identify women at increased risk of adverse outcomes, with +LR values of at least 5.0 for delivery with PE at <37 weeks of gestation, GH at <34 weeks of gestation, stillbirth at <37 weeks of gestation and birthweight below the third percentile at <37 weeks of gestation.
Table 4 shows the results of a two-stage screening approach, first by screening with clinical risk factors, and then for the 9941 pregnancies with one or more relevant risk factors, screening by PlGF at 19-23 weeks of gestation, and considering only those with PlGF below the fifth percentile to be screen-positive.For the 547 (1.8%) pregnancies that were screen-positive, detection rates were <10% for all adverse outcomes examined, at any gestational age.Although screen-positive pregnancies were at a meaningfully increased risk of preterm adverse outcomes, no women with clinical risk factors could be reassured that adverse outcomes were unlikely to develop.S1.

| Main findings
In this cohort of more than 30 000 unselected singleton pregnancies at 19-23 weeks of gestation, clinical risk factor screening had a high screen-positive rate (of 33%).Although this was primarily because of risk factors for FGR, there was substantial overlap in risk factors with PE, such that twothirds of women with risk factors for FGR were also identified as being at high risk of PE.Clinical risk factors for PE or FGR identified only about half of those destined to develop an adverse outcome of PE, GH, stillbirth, birthweight below the third percentile or NICU admission for ≥48 h.Importantly, among those who were screen-positive, the risk of adverse outcome was not meaningfully increased compared with those without any clinical risk factors.
As an alternative primary screening strategy, PlGF testing at 19-23 weeks of gestation had low detection rates (under 20%) for adverse outcomes, but pregnancies with low PlGF were indeed at a meaningfully higher risk of complications at preterm, but not term, gestational age.
Two-stage screening, first with clinical risk factors and then, if present, with PlGF testing at 19-23 weeks of gestation, could not address the high screen-positive rate associated with clinical risk factor screening in early pregnancy, as the large number of women who screened positive with clinical risk factors could not be reassured if their PlGF levels were then normal at 19-23 weeks of gestation.

| Interpretation
Currently, UK national guidelines recommend clinical risk factor screening to identify PE and FGR risk, in which risk factors are treated independently.Similar to our findings, others have identified that this approach is associated with a low detection rate for preterm (approx.40%) or term (approx.35%) PE, 12 or preterm (55%) or term (47%) SGA (birthweight below tenth percentile). 13However, PE and FGR risk factors have been evaluated separately, despite their substantial T A B L E 2 Primary screening by NICE and RCOG guidelines of the total population of 30 013 pregnancies, from which 9941 (33.1%) women were screen-positive.overlap, as well as the overlap in PE and FGR clinical presentations and the evolution of disease during antenatal surveillance.Our findings confirm that one-third of pregnancies are screen-positive for PE and/or FGR risk factors, but despite this very high screen-positive rate, the detection rates for adverse outcomes are poor.Alternatives to clinical risk factor screening are needed.Our findings do not suggest that taking a two-stage screening approach, by adding PlGF testing at 19-23 weeks of gestation for women who are screen-positive with clinical risk factors, could improve the screening performance of clinical risk factors.Similarly, our findings do not support the implementation of routine PlGF screening at 19-23 weeks of gestation, based on the poor detection rates for placental disorders.These results are consistent with a systematic review that demonstrated that PlGF testing alone (16 studies) is inferior to PlGF-based models (six studies) for the prediction of PE. 14 Superior approaches are available for the detection of women at risk of delivery with PE or FGR at preterm or term gestational ages.The FMF 'competing risks' model uses the multivariable modelling of clinical, ultrasonographic and laboratory assessment of uteroplacental perfusion and function to identify those who may benefit from low-dose aspirin, to decrease the risk of preterm PE, or enhanced surveillance and timed birth, to optimise outcomes related to PE and/or FGR.The FMF model for PE detects approximately 75% of women who will develop preterm PE (when screened at 11-13 weeks of gestation) or term PE (when screened at 35-36 weeks of gestation) for screen-positive rates of 10%. 12At 19-23 weeks of gestation, at a screen-positive rate defined by the RCOG guideline (22.5% in the relevant publication), the detection rates for SGA below the tenth percentile at <32, <37 and ≥37 weeks of gestation were 81%, 72% and 56%, respectively, using the FMF model, compared with 45%, 44% and 36%, respectively, following the RCOG guidelines. 15The model has been updated to provide an effective personalised continuous stratification of pregnancy care pertinent to SGA, defining the appropriate timing of a subsequent  ultrasonographic examination of fetal growth and well-being at 24-36 weeks of gestation, depending on the individual characteristics and the biophysical marker levels at the mid-gestation assessment. 16

| Strengths and limitations
Strengths of our study include the very large sample size recruited prospectively in an unselected fashion from a diverse clinical population, using comprehensive and standardised data collection.We evaluated the collective risk of PE or FGR (given the clinical overlap in risk factors and clinical presentations), the use of PlGF as a marker of placental dysfunction and a spectrum of relevant outcomes.All hypertensive pregnancies were reviewed to distinguish between chronic hypertension, GH and PE, using a broad definition. 8We evaluated the detection rates and diagnostic test properties, which are prevalence independent, and evaluated whether the screening strategy meaningfully increases or decreases the detection of risk for adverse outcomes.
Limitations include the enrolment of only singleton pregnancies, so our results may not apply to multiple pregnancies.At first we evaluated serum concentrations of PlGF alone, not MoM using maternal characteristics, and then included MoM, as this is how PlGF is most commonly assessed in clinical practice.

| CONCLUSION
Clinical risk factor screening for PE and FGR results in a high screen-positive rate but a poor detection rate of adverse outcomes, and the high false-positive rate cannot be reduced by PlGF testing at 19-23 weeks of gestation.Importantly, clinical risk factor screening and/or the results of PlGF testing at 19-23 weeks of gestation should not be used to guide the timing of birth at term.Future research should address the optimal follow-up of women

T A B L E 1
Baseline characteristics, details of screening and pregnancy outcomes for the study population of 30 013 unselected pregnancies at 19-23 weeks of gestation.

T A B L E 3
Primary screening by serum PlGF below the fifth percentile of the total population of 30 013 pregnancies, from which 1501 (5.0%) women were screen-positive.a 3-35.6) 31.7 (27.6-35.1)34.6 (28.9-38.1) Note: data are expressed as N pregnancies (%) or median value (IQR).Abbreviations: APAS, antiphospholipid antibody syndrome; FGR, fetal growth restriction; IQR, interquartile range; NICU, neonatal intensive care unit; PI, pulsatility index; PlGF, placental growth factor; UtArt, uterine artery.a For the clinical risk factors for PE and FGR, see Table Two-stage screening for adverse outcomes.a T A B L E 4