Systematic review

DOI: 10.1111/j.1471-0528.2012.03311.x
www.bjog.org

Accuracy of circulating placental growth factor,

vascular endothelial growth factor, soluble
fms-like tyrosine kinase 1 and soluble endoglin
in the prediction of pre-eclampsia: a systematic
review and meta-analysis
CE Kleinrouweler,a MMJ Wiegerinck,a C Ris-Stalpers,a,b PMM Bossuyt,c JAM van der Post,a
P von Dadelszen,d BWJ Mol,a E Pajkrta for the EBM CONNECT Collaboration
a

Department of Obstetrics and Gynaecology b Reproductive Biology Laboratory c Department of Clinical Epidemiology, Biostatistics and
Bioinformatics, Academic Medical Center, Amsterdam, the Netherlands d Department of Obstetrics and Gynaecology, University of British
Columbia, Vancouver, BC, Canada
Correspondence: Ms CE Kleinrouweler, Department of Obstetrics and Gynaecology, Academic Medical Center, Room H4-232, Meibergdreef 9,
1105 AZ Amsterdam, the Netherlands. Email c.e.kleinrouweler@amc.uva.nl
Accepted 8 February 2012. Published Online 20 March 2012.

Background Biomarkers have been proposed for identification of

women at increased risk of developing pre-eclampsia.

Objectives To investigate the capacity of circulating placental

growth factor (PlGF), vascular endothelial growth factor (VEGF),

soluble fms-like tyrosine kinase-1 (sFLT1) and soluble endoglin
(sENG) to predict pre-eclampsia.
Search strategy Medline and Embase through October 2010 and

reference lists of reviews, without constraints.

Selection criteria We included original publications on testing of

PlGF, VEGF, sFLT1 and sENG in serum or plasma of pregnant

women at <30 weeks of gestation and before clinical onset of
pre-eclampsia.
Data collection and analysis Two reviewers independently

identified eligible studies, extracted descriptive and test accuracy

data and assessed methodological quality. Summary estimates of
discriminatory performance were obtained.

who developed pre-eclampsia: standardised mean differences

(SMD) )0.56 (95% CI )0.77 to )0.35) and )1.25 (95% CI
)2.73 to 0.23). Concentrations of sFLT1 (19 studies) and sENG
(ten studies) were higher: SMD 0.48 (95% CI 0.210.75) and
SMD 0.54 (95% CI 0.240.84). The summary diagnostic odds
ratios were: PlGF 9.0 (95% CI 5.614.5), sFLT1 6.6 (95% CI
3.113.7), sENG 4.2 (95% CI 2.47.2), which correspond to
sensitivities of 32%, 26% and 18%, respectively, for a 5% falsepositive rate.
Authors conclusions PlGF, sFLT1 and sENG showed modest but
significantly different concentrations before 30 weeks of gestation
in women who developed pre-eclampsia. Test accuracies of all
four markers, however, are too poor for accurate prediction of
pre-eclampsia in clinical practice.
Keywords Biomarkers, placental growth factor, pre-eclampsia,

soluble endoglin, soluble fms-like tyrosine kinase-1, vascular

endothelial growth factor.

Main results We included 34 studies. Concentrations of PlGF

(27 studies) and VEGF (three studies) were lower in women

Please cite this paper as: Kleinrouweler C, Wiegerinck M, Ris-Stalpers C, Bossuyt P, van der Post J, von Dadelszen P, Mol B, Pajkrt E, for the EBM CONNECT Collaboration. Accuracy of circulating placental growth factor, vascular endothelial growth factor, soluble fms-like tyrosine kinase 1 and soluble
endoglin in the prediction of pre-eclampsia: a systematic review and meta-analysis. BJOG 2012;119:778787.

Introduction
Pre-eclampsia, a multisystem disorder of pregnancy defined
as hypertension and proteinuria, is a major obstetric problem. It contributes substantially to maternal and perinatal

778

morbidity and mortality worldwide, especially in developing countries.1

A singular problem in the management of pre-eclampsia
is that by the time symptoms occur the only definitive
treatment of the underlying disorder is delivery, often

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

Biomarkers for prediction of pre-eclampsia

preterm. Expectant management of preterm pre-eclampsia

focuses on safely prolonging pregnancy through intensive
monitoring to prevent maternal and fetal complications.1
Screening for pregnant women to identify those at risk
of developing pre-eclampsia should substantially improve
the quality, focus, resource use and efficacy of antenatal
care, with the prospect of improving maternal and perinatal outcomes.2,3 Moreover, preventive treatment such as
aspirin would probably be more beneficial when started in
early pregnancy.4
The pathogenesis of pre-eclampsia, especially that of
early onset, begins at the time of trophoblast invasion and
remodelling of the spiral arteries during the first 12 weeks
of pregnancy.1 Inadequate placentation and subsequent
hypoxia are thought to be followed by an increased release
of the placenta-produced anti-angiogenic factor soluble
fms-like tyrosine kinase-1 (sFLT1) into the maternal circulation.5 The sFLT1 binds to the angiogenic proteins placental growth factor (PlGF) and vascular endothelial growth
factor (VEGF), thereby blocking their actions through the
plasma-membrane-bound form of the receptor, that contains the tyrosine kinase domain essential to its biological
activity.69 Soluble endoglin (sENG), the extracellular
domain of the co-receptor endoglin, impairs binding of
transforming growth factor-b1 to cell surface receptors and
decreases endothelial nitric oxide signalling, hence inhibiting angiogenesis and promoting vascular dysfunction.5
Dysfunction of the maternal endothelium ultimately results
in the clinical syndrome of pre-eclampsia.10,11
A number of studies have been performed to investigate
circulating levels of these factors in pre-eclamptic pregnancies and to compare them with uncomplicated pregnancies.
A recent Health Technology Assessment report that investigated the accuracy of predictive tests for pre-eclampsia and
their possible cost-effectiveness expressed the need for systematic reviews on new tests not considered in the report,
including biomarkers.12 In a systematic review from 2007,
sFLT1 levels were found to be elevated and PlGF levels to
be reduced in pre-eclamptic pregnancies, but absolute levels
varied markedly between studies.13 Because the number of
studies on sFLT1 and PlGF has increased substantially since
that review and the evidence on VEGF and sENG has not
been summarised previously, we undertook a systematic
review of the literature on the accuracy of the biomarkers
PlGF, VEGF, sFLT1 and sENG in the prediction of preeclampsia.

Methods
Data sources
We performed an electronic search on 26 October 2010 in
Medline (from 1948) and Embase (from 1980) without language or publication date restrictions to identify all articles

reporting on the prediction of pre-eclampsia using one or

more of the markers PlGF, VEGF, sFLT1 and sENG. The
electronic search strategy was based on MeSH terms and
keywords related to pre-eclampsia and to each of the four
markers, combined with methodological filters, allowing
efficient identification of studies on diagnostic and prognostic tests (see Appendix S1). Reference lists of review
articles13 and eligible primary studies were checked to identify cited articles not captured by the electronic search.
This systematic review and meta-analysis was conducted
according to the Meta-analysis Of Observational Studies in
Epidemiology (MOOSE) guidelines.14

Eligibility criteria
Eligible studies were those that reported on testing of PlGF,
VEGF, sFLT1 or sENG in serum or plasma of pregnant
women with blood sampling before clinical onset of preeclampsia and before 30 weeks of gestation. To be
included, studies should describe the occurrence of preeclampsia conditional on the test result in such a way that
2 2 classification tables could be (re-)constructed, or
should describe the test results conditional on the occurrence of pre-eclampsia as means and standard deviations in
pregnancies before pre-eclampsia and uncomplicated pregnancies.

Study selection
Studies were selected in a staged process. First, two reviewers
(EK and MW) independently scrutinised titles and abstracts
of all retrieved references to select potentially eligible articles. Full text papers of references selected by at least one
reviewer were obtained. Second, both reviewers independently examined these full text papers to see whether they
met the inclusion and exclusion criteria. In case of multiple
publications of one dataset we included only the most
recent or most complete paper. Disagreements about inclusion were resolved by consensus or by consulting a third
reviewer (BM). We did not contact authors for further
information.

Data extraction
For each included study, data on clinical characteristics of
the women (age, obstetric history), characteristics of the
index test (marker, medium, test kit and manufacturer),
reference standard and test accuracy were extracted independently by two experienced reviewers (EK and MW)
using standardised data extraction forms.

Quality assessment
Both reviewers assessed the methodological quality of the
included studies using the quality assessment of diagnostic
accuracy studies (QUADAS) criteria.15 In addition, we
assessed whether the study participants had received

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

779

Kleinrouweler et al.

preventive treatment. Acceptable reference standards for

pre-eclampsia were persistent high systolic (140 mmHg)
or diastolic (90 mmHg) blood pressure and proteinuria
(0.3 g/24 hours or a dipstick result of 1+, equivalent to
30 mg/dl in a single urine sample or spot urine protein/
creatinine ratio 30 mg protein/mmol creatinine) of new
onset after 20 weeks of gestation, according to the International Society for the Study of Hypertension in Pregnancy
criteria.16

Data synthesis
For the studies with marker concentration reported as a
continuous variable, we assessed the differences in marker
concentration between women who did and did not
develop pre-eclampsia and expressed the results in standardised mean differences. For uniform presentation in the
tables, all reported marker concentrations were converted
into pg/ml, as this is the most frequently reported unit. For
pooling of the results, we used an inverse-variance
weighted random effect approach in Review Manager
5.0.17
Results of studies reporting sensitivities and specificities
were plotted in receiver operating characteristics spaces
with summary receiver operating characteristics curves that
correspond to summary diagnostic odds ratios. Analogous
to the odds ratio for expressing the strength of association
between exposure and disease, the diagnostic odds ratio
can be applied to express the strength of the association
between test result and disease. The diagnostic odds ratio
describes the odds of positive test results in patients with
disease compared with the odds of positive test results in
those without disease. Higher diagnostic odds ratios represent higher test accuracies: a test with a sensitivity and
specificity of 90% has a diagnostic odds ratio of 81. A diagnostic odds ratio value of 1 indicates that a test does not
discriminate between women with the disease and those
without the disease. Values <1 point to improper test interpretation with more negative tests among the diseased.18,19
We pooled the diagnostic odds ratios with a random effects
model using the DerSimonian Laird method in Meta-DiSc
1.4 software.20
Sensitivity analyses were performed to assess the impact
of gestational age at testing, study design and study population, selection of controls in casecontrol studies, type of
test kit used and commercial funding on the pooled estimates.

Results
The searches in Medline and Embase provided a total of
4072 citations after removing duplicates. Of these, 3968
studies had to be discarded after reviewing titles and
abstracts. The full text of the remaining 104 papers was

780

examined in more detail. Of these, 70 studies did not meet

the inclusion criteria as described. These studies are listed
in the Appendix S2. Thirty-four studies met the inclusion
criteria and were included in this systematic review.2154
No eligible unpublished studies on the subject were identified (Figure 1).
There were 27 studies reporting on PlGF, three studies
on VEGF, 19 studies on sFLT1 and ten studies on sENG.
Fifteen studies reported on one marker, 13 on two, and six
on three different markers. The most common combination was PlGF and sFLT1. The set contained 12 cohort
studies and 22 casecontrol studies. Markers had been evaluated from 5 to 28 weeks of gestation. Study sample size
ranged from 29 to 3098 women. Several studies investigated testing of markers at multiple gestational ages, mostly
in the same women. Some studies reported data on preeclamptic pregnancies for subgroups only, for example in
women with clinical onset of pre-eclampsia before and after
32, 34 or 37 weeks of gestation. We provide an overview of
all studies in Tables S1S7.

Quality
Figure 2 summarises the results of the quality assessment.
All studies met the following QUADAS quality criteria: use
of an acceptable and independent reference test, appropriate
time period between tests, avoidance of partial and differential verification bias, withdrawals explained and relevant
clinical information available. More than 70% of studies
also met the items for representative spectrum of patients,
selection criteria described, blinding of the reference test,
and reporting of uninterpretable results (and instrument
variation and free of commercial funding).
Studies scored poorly in terms of adequately describing
details of the reference test and, more importantly, the
index test: only four studies (three of the 27 on PlGF and
one of the three on VEGF) accurately described whether
the instrument measured the free subset or total pool of
the marker. Four studies reported on the application of
preventive treatment (aspirin, calcium, vitamins C and E).
For PlGF, five out of 15 studies reporting sensitivity and
specificity did not report corresponding cut-off points. For
sFLT1 and sENG, these were one of eight and none of five
studies, respectively.

PlGF
A total of 27 studies reported concentrations of PlGF in
women who developed pre-eclampsia and in women with
uncomplicated pregnancies. In women who developed preeclampsia, concentrations of PlGF were significantly lower:
standardised mean difference (SMD) )0.56 (95% CI )0.78
to )0.35). In studies that performed the test before
16 weeks of gestation, the difference was less apparent
(SMD )0.51, 95% CI )0.95 to )0.07) than in studies that

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

Biomarkers for prediction of pre-eclampsia

Figure 1. Flow diagram of identification and selection of studies of PlGF, VEGF, sFLT1 and sENG for prediction of pre-eclampsia, for inclusion in this
systematic review.

tested from 19 weeks onward (SMD )0.85, 95% CI )1.52

to )0.18) but greater than in studies that tested in the
overlap of both periods, 626 weeks (SMD )0.45, 95% CI
)0.69 to )0.22), see Appendix S3A.
We identified 15 eligible studies that estimated the sensitivity and specificity of PlGF testing (see Appendix S4A).
The summary diagnostic odds ratio was 9.0 (95% CI 5.6
14.5). This corresponds to both sensitivities and specificities
of 0.75, or a sensitivity of 0.32 for a 5% false-positive rate.

VEGF
There were three studies reporting on VEGF testing.
Women who would develop pre-eclampsia had lower concentrations of VEGF, but this was not significant: SMD
)1.25 (95% CI )2.73 to 0.23) (see Appendix S3B). We did

not identify any studies reporting sensitivity and specificity

of VEGF testing.

sFLT1
Concentrations of sFLT1 were reported in 19 studies. These
were higher in pregnancies before pre-eclampsia than in
uncomplicated pregnancies: SMD 0.48 (95% CI 0.210.74).
Here also, differences were most pronounced in studies
that tested women 19 weeks of gestation (SMD 0.85, 95%
CI 0.361.33) compared with 16 weeks of gestation (SMD
0.60, 95% CI )0.11 to 1.31) or 726 weeks (SMD 0.24,
95% CI 0.050.43) (see Appendix S3C).
We identified eight studies that estimated the sensitivity
and specificity of sFLT1 testing (see Appendix S4B). The
summary diagnostic odds ratio for these studies was 6.6

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

781

Kleinrouweler et al.

Yes or not applicable

No

Unclear

Figure 2. Summary of quality assessment.

(95% CI 3.113.7). This corresponds to a 0.72 sensitivity

and specificity, or a sensitivity of 0.26 for a 5% false-positive rate.

sENG
In the ten studies that reported concentrations of sENG,
higher concentrations were found in pregnancies before
pre-eclampsia: SMD 0.54 (95% CI 0.240.84). In studies
testing before 16 weeks of gestation, these differences were
smaller (SMD 0.18, 95% CI )0.02 to 0.38) compared with
studies testing between 7 and 26 weeks of gestation (SMD
0.70, 95% CI 0.301.10) (see Appendix S3D).
We identified four studies that reported sensitivity and
specificity of sENG testing (see Appendix S4C). The summary diagnostic odds ratio was 4.2 (95% CI 2.47.2). This
corresponds to a sensitivity and specificity of 0.67, or a
sensitivity of 0.18 for a 5% false-positive rate.
Highly significant between-study heterogeneity was
recorded for all markers, with I2 statistics of 84% (PlGF),
96% (VEGF), 93% (sFLT1) and 91% (sENG). Sensitivity
analyses taking into account differences in gestational age
at testing (see Appendix S3), study design and selection of
study population, and selection of eligible controls in case
control studies (see Appendix S5) did not explain this
heterogeneity. In a sensitivity analysis with only studies that
used the most frequently used ELISA kits from R&D Systems (Abingdon, UK), I2 statistics decreased to 80% for
PlGF (15/18 studies, SMD )0.60, 95% CI )0.84 to )0.30)
and 70% for sFLT1 (14/17 studies, SMD 0.33, 95% CI

782

0.160.49) but remained 91% for sENG (eight of nine

studies, SMD 0.42, 95% CI 0.080.76). All studies on VEGF
used R&D Systems assays. We could not find any evidence
that commercial participation in a study influenced the
predictive ability of the markers.

Discussion
This systematic review provides an overview of the discriminatory performance and predictive capacity of the proangiogenic and anti-angiogenic biomarkers PlGF, VEGF,
sFLT1 and sENG for pre-eclampsia before 30 weeks of gestation. We included studies that investigated differences in
marker concentration between women who developed preeclampsia and those who remained healthy throughout
pregnancy and studies that provided details on the accuracy
of such tests. In 34 such studies with overall good quality,
we found that mean concentrations of PlGF were modestly
but significantly lower before pre-eclampsia and concentrations of sFLT1 and sENG were higher. In studies that compared the timing of testing (16 weeks, 19 weeks or
both), differences between the groups were largest from
19 weeks of gestation onwards and significant for all three
markers, whereas for sFLT1 and sENG, the differences in
marker concentration were not significantly different
16 weeks of gestation. However, the test accuracy of the
markers PlGF, sFLT1 and sENG in terms of sensitivity and
specificity was too poor for accurate identification of preeclampsia cases in clinical practice, and we recommend that

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

Biomarkers for prediction of pre-eclampsia

these markers should not be used alone for the prediction

of pre-eclampsia.
Some methodological aspects of this study require comment. This systematic review revealed a considerable
amount of heterogeneity between studies, which is reflected
in the high values of I2 in the meta-analyses. We propose
several mechanisms through which this might have
occurred. First, eligible studies reported on testing in dissimilar and overlapping periods of gestational age, which
hampered the distinction between first-trimester and second-trimester assessments, and concentrations of the markers are known to change with gestational age as the result
of biological variability.55,56 In addition, the majority of
studies24 of 27 for PlGF and two out of three for
VEGFdid not report exactly what was measured by the
test kit, i.e. whether it concerned the total circulating level
of the marker, the free fraction or a subset of either one of
these. Also, different ELISA test kits were used. A combination of these factors may have resulted in varying levels of
the same circulating factor, even in similar populations and
at similar gestational ages, which makes it possible to
express the results only as SMD.
Although we maintained high quality standards for the
conduct of this systematic review and meta-analysis, we
realise that the results and their implications are only as
good as the source of the data. In the absence of a sufficient number of well-designed marker evaluation studies,
we had to include studies that differed in design: nested or
matched casecontrol studies as well as cohort studies, with
widely ranging sample sizes. Furthermore, selection of the
study population (at lower or higher risk) in cohort studies
and selection of eligible controls (completely healthy or
without pre-eclampsia) in casecontrol studies differed
between studies. Despite these likely sources of heterogeneity, sensitivity analyses did not point to a single most
important factor. Other factors that may partly explain the
heterogeneity are the endpoint of disease and the definition
of pre-eclampsia used in the original studies, although we
minimised this by only including studies that defined preeclampsia according to the International Society for the
Study of Hypertension in Pregnancy criteria,16 and we
found that almost all studies used all pre-eclampsia as the
endpoint of disease. Because only a few studies have provided results for subgroups with onset of disease before or
after a certain (varying) gestational age, we did not perform
sensitivity analyses taking this factor into account. The endpoint of disease and selection criteria used in the original
studies can be found in Tables S1S7.
We identified several studies that reported the results in
medians and ranges or multiples of the medians, which
cannot be compared directly with the included studies, so
we were not able to include all existing evidence in this
paper.

We also investigated the test accuracy of the markers

PlGF, sFLT1 and sENG in terms of their sensitivity and
specificity. Because of different or unreported positivity
thresholds, a meta-analysis of sensitivities and specificities
was inappropriate because this would ignore threshold differences and underestimate diagnostic performance. Instead
we used the diagnostic odds ratio as a single measure of
test performance. A disadvantage of this method is that it
does not result in unique summary estimates of sensitivity
and specificity: summary estimates of one value can only
be obtained by specifying the value of the other.18,19 Hence,
for each of the biomarkers, we reported two possible estimates of sensitivity and specificity that corresponded to the
pooled diagnostic odds ratios: one in which sensitivity and
specificity were equal, and the other in which there was a
false-positive rate of 5% (or a specificity of 95%), which is
common in screening studies.
Identifying women at risk for pre-eclampsia remains an
important aspect of antenatal care57 because it can not only
contribute to the development and evaluation of preventive
treatments, but can also guide the structure of antenatal
care. A recent meta-analysis showed that aspirin started
at 16 weeks or earlier in pregnancy was associated with a
significant and greater reduction in pre-eclampsia in women
at moderate or high risk than aspirin started after 16 weeks
of gestation.4 In addition, early estimation of patient-specific risks for pregnancy complications could shift antenatal
care from a series of routine visits for everyone to an
approach in which care, in terms of schedule and content,
is given to those who need it and others are reassured at an
early stage. Women identified as being at high risk can have
close surveillance, but the great majority of women are safe
with a substantially reduced number of visits.58
Our findings that concentrations of PlGF, sFLT1 and
sENG are significantly different before the onset of preeclampsia and that these differences are greatest from
19 weeks of gestation onwards would be unfavourable in
the clinical situation because it would not allow for early
intervention. Moreover, using these markers as a discriminative test proved to be more difficult and had only limited
accuracy.
In addition, all pre-eclampsia cases do not appear to
have the same origin.1 While early-onset pre-eclampsia is
most closely associated with inadequate placentation and
may well be associated with alterations in angiogenic balance, as suggested by this review, term pre-eclampsia is
most commonly associated with normal placental development and likely to be predicted by factors associated with
long-term cardiovascular risk, such as obesity, diabetes and
chronic hypertension.59 However, most studies included in
this review did not describe subgroups of early-onset and
late-onset pre-eclampsia and we did not make this distinction in our analyses.

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

783

Kleinrouweler et al.

It is unlikely that any single biomarker or related pairing

of biomarkers will be able to predict all forms of preeclampsia. An effective test is likely to require the assessment of the various pathways that lead to the clinical
phenotype both remote from and at term. The large number of relatively small studies that we identified is of
concern in this area of research. There are as yet no rules
for the early registration of cohort studies on prediction
models similar to the registration of randomised clinical
trials. As a consequence, there is the chance of many small
studies in which multiple potential markers will be tested.
The predictive capacities of the markers in these types of
studies tend to be overestimated, whereas larger studies will
generally report less extreme results.
Although we found that the accuracy of the biomarkers
PlGF, sFLT1 and sENG is too poor to allow their routine
use for prediction of pre-eclampsia, they might be useful
when incorporated in multivariable prediction models. The
recent SCOPE study showed that the ability of clinical
characteristics to predict pre-eclampsia in healthy nulliparous women is modest. Clinical characteristics included
age, mean arterial blood pressure, body mass index, family
history of pre-eclampsia, family history of coronary heart
disease, maternal birthweight, vaginal bleeding for at least
5 days, previous single miscarriage with the same partner,
time to conception, intake of fruit, cigarette smoking and
alcohol use in the first trimester. To improve overall accuracy and detection of cases, the clinical algorithm will
require the addition of predictors. The addition of 1921week uterine artery Doppler indices did not improve the
predictive performance, but the authors suggest instead
that biomarkers could be added to the model and if externally validated, such models could provide a personalised
clinical risk profile.60 This is supported by some of the
included studies and other cohort studies that have shown
that combinations of patient factors and (multiple) biomarkers result in better prediction,21,22,49,61 although these
models have to be externally validated as well. In contrast,
the bulk of relatively small studies, such as we found in this
meta-analysis, are unlikely to clarify this matter.
Before a biomarker can be implemented in clinical practice, its clinical validity and clinical utility have to be evaluated. Ultimately, it is the influence on patient outcome
provided by the test that matters. In the decision as to
whether these markers should be used, one should not only
be informed about the test accuracy, but also about its
ability to influence medical decision making; for example
the intensity of monitoring or application of preventive
treatment, and the resulting change in patient outcome.
Currently, we are only at the start of this process. No
randomised trials have been conducted that compare prenatal care including biomarker testing to standard prenatal
care. Future studies should also consider the cost-effective-

784

ness of such a screening strategy as a whole. Prior tests

with high costs, such as biomarker testing, would need
substantially improved sensitivities to be able to improve
cost-effectiveness, but such levels of sensitivity have rarely
been achieved.12 Realising that most studies on biomarkers
use scarce public resources, we may consider a global,
collaborative approach to evaluate the role of biomarkers
in the prediction of pre-eclampsia.

Disclosure of interests
PD receives salary support from the Canadian Institutes for
Health Research, Michael Smith Foundation for Health
Research, and the Child and Family Research Institute. He
also reported receiving salary support for consultancy and
lectures from Alere International and for expert testimony
from the Canadian Medical Protective Agency.

Contribution to authorship
BWM conceived the study, supervised the analyses and
critically revised the manuscript for important intellectual
content. EK and MW carried out the literature search and
drafted the manuscript. EK performed the analyses. PB
advised on and supervised the analyses and critically
revised the manuscript for important intellectual content.
CRS, EP, JP and PD assisted in data interpretation and
critically revised the manuscript for important intellectual
content. All authors approved the final version of the manuscript that was submitted.

Details of ethics approval

Not required.

Funding
EK is supported by a PhD Scholarship from the AMC
Graduate School. PD receives salary support from the
Canadian Institutes for Health Research, Michael Smith
Foundation for Health Research, and the Child and Family
Research Institute. The authors received funding from the
European Union made available to the EBM-CONNECT
Collaboration through its Seventh Framework Programme,
Marie Curie Actions, International Staff Exchange Scheme
(Proposal number: 101377; Grant Agreement number:
247613); EBM-CONNECT Canadian Collaborators received
funding from the Canadian Institutes of Health Research.
The EBM-CONNECT (Evidence-Based Medicine COllaboratioN: NEtwork for systematic reviews and guideline
development researCh and dissemination) Collaboration
(in alphabetical order by country) includes: L Mignini,
Centro Rosarino de Estudios Perinatales, Argentina; P von
Dadelszen, L Magee, D Sawchuck, University of British
Columbia, Canada; E Gao, Shanghai Institute of Planned
Parenthood Research, China; BW Mol, K Oude Rengerink,
Academic Medical Center, the Netherlands; J Zamora,

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

Biomarkers for prediction of pre-eclampsia

Ramon y Cajal, Spain; C Fox, J Daniels, University of

Birmingham, UK; and KS Khan, S Thangaratinam, C
Meads, Barts and the London School of Medicine, Queen
Mary University of London, UK. No funders were involved
in the design and conduct of the study; collection, management, analysis and interpretation of the data; or in the
preparation, review or approval of the manuscript.

Acknowledgement
We thank Faridi van Etten, clinical librarian, for her help
with the electronic search.

Supporting Information
Additional Supporting Information may be found in the
online version of this article.
Tables S1S7. Study characteristics, characteristics of
women and test results of all 34 included studies, grouped
by biomarker and presentation of results.
Appendix S1. Search strategy for MEDLINE and
EMBASE.
Appendix S2. References of studies excluded after
screening of full text, as described in Figure 1.
Appendix S3. Forest plots showing standardised mean
differences in marker concentration between women
who would develop pre-eclampsia and women with
uncomplicated pregnancies in all studies reporting data as
continuous variables. Negative values indicate lower concentrations; positive values indicate higher concentrations
before pre-eclampsia. (A) PlGF; (B) VEGF; (C) sFLT1; and
(D) sENG.
Appendix S4. Results of studies on PlGF (A), sFLT1 (B)
and sENG (C) reporting sensitivity and specificity of the
test, plotted in receiver operating characteristics (ROC)
spaces with summary ROC curves. Area of circles is proportional to study sample size.
Appendix S5. Sensitivity analyses that are not reported
in the paper.
Please note: Wiley-Blackwell are not responsible for the
content or functionality of any supporting information
supplied by the authors. Any queries (other than missing
material) should be directed to the corresponding
author. j

References
1 Steegers EA, von Dadelszen P, Duvekot JJ, Pijnenborg R. Pre-eclampsia. Lancet 2010;376:63144.
2 Duley L, Henderson-Smart DJ, Meher S, King JF. Antiplatelet agents
for preventing pre-eclampsia and its complications. Cochrane Database Syst Rev 2007;:CD004659.
3 Askie LM, Duley L, Henderson-Smart DJ, Stewart LA. Antiplatelet
agents for prevention of pre-eclampsia: a meta-analysis of individual
patient data. Lancet 2007;369:17918.

4 Bujold E, Roberge S, Lacasse Y, Bureau M, Audibert F, Marcoux S,

et al. Prevention of preeclampsia and intrauterine growth restriction
with aspirin started in early pregnancy: a meta-analysis. Obstet
Gynecol 2010;116:40214.
5 James JL, Whitley GS, Cartwright JE. Pre-eclampsia: fitting together
the placental, immune and cardiovascular pieces. J Pathol
2010;221:36378.
6 Gu Y, Lewis DF, Wang Y. Placental productions and expressions of
soluble endoglin, soluble fms-like tyrosine kinase receptor-1, and
placental growth factor in normal and preeclamptic pregnancies.
J Clin Endocrinol Metab 2008;93:2606.
7 Maynard SE, Min JY, Merchan J, Lim KH, Li J, Mondal S, et al.
Excess placental soluble fms-like tyrosine kinase 1 (sFlt1) may contribute to endothelial dysfunction, hypertension, and proteinuria in
preeclampsia. J Clin Invest 2003;111:64958.
8 Tsatsaris V, Goffin F, Munaut C, Brichant JF, Pignon MR, Noel A,
et al. Overexpression of the soluble vascular endothelial growth factor receptor in preeclamptic patients: pathophysiological consequences. J Clin Endocrinol Metab 2003;88:555563.
9 Zhu X, Wu S, Dahut WL, Parikh CR. Risks of proteinuria and hypertension with bevacizumab, an antibody against vascular endothelial
growth factor: systematic review and meta-analysis. Am J Kidney Dis
2007;49:18693.
10 Roberts JM, Hubel CA. The two stage model of preeclampsia: variations on the theme. Placenta 2009;30(Suppl.):327.
11 Huppertz B. Placental origins of preeclampsia: challenging the current hypothesis. Hypertension 2008;51:9705.
12 Meads CA, Cnossen JS, Meher S, Juarez-Garcia A, ter Riet G, Duley
L, et al. Methods of prediction and prevention of pre-eclampsia:
systematic reviews of accuracy and effectiveness literature with
economic modelling. Health Technol Assess 2008;12:iiiiv, 1270.
13 Widmer M, Villar J, Benigni A, Conde-Agudelo A, Karumanchi SA,
Lindheimer M. Mapping the theories of preeclampsia and the role
of angiogenic factors: a systematic review. Obstet Gynecol
2007;109:16880.
14 Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D,
et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283:200812.
15 Whiting PF, Weswood ME, Rutjes AW, Reitsma JB, Bossuyt PN, Kleijnen J. Evaluation of QUADAS, a tool for the quality assessment of
diagnostic accuracy studies. BMC Med Res Methodol 2006;6:9.
16 Brown MA, Lindheimer MD, de SM, Van AA, Moutquin JM. The
classification and diagnosis of the hypertensive disorders of pregnancy: statement from the International Society for the Study of
Hypertension in Pregnancy (ISSHP). Hypertens Pregnancy 2001;20:
IXXIV.
17 Review Manager (RevMan) [Computer program] Version 5.0. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration,
2008.
18 Deeks JJ. Systematic reviews in health care: systematic reviews
of evaluations of diagnostic and screening tests. BMJ 2001;323:
15762.
19 Glas AS, Lijmer JG, Prins MH, Bonsel GJ, Bossuyt PM. The diagnostic
odds ratio: a single indicator of test performance. J Clin Epidemiol
2003;56:112935.
20 Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A. MetaDiSc: a software for meta-analysis of test accuracy data. BMC Med
Res Methodol 2006;6:31.
21 Akolekar R, Zaragoza E, Poon LCY, Pepes S, Nicolaides KH. Maternal
serum placental growth factor at 11 + 0 to 13 + 6 weeks of gestation in the prediction of pre-eclampsia. Ultrasound Obstet Gynecol
2008;32:7329.

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

785

Kleinrouweler et al.

22 Audibert F, Boucoiran I, An N, Aleksandrov N, Delvin E, Bujold E,

et al. Screening for preeclampsia using first-trimester serum markers
and uterine artery Doppler in nulliparous women. Am J Obstet
Gynecol 2010;203:3838.
23 Baumann MU, Bersinger NA, Mohaupt MG, Raio L, Gerber S, Surbek
DV. First-trimester serum levels of soluble endoglin and soluble fmslike tyrosine kinase-1 as first-trimester markers for late-onset preeclampsia. Am J Obstet Gynecol 2008;199:266.
24 Bersinger NA, Odegard RA. Serum levels of macrophage colony
stimulating, vascular endothelial, and placenta growth factor in relation to later clinical onset of pre-eclampsia and a small-for-gestational age birth. Am J Reprod Immunol 2005;54:7783.
25 Bills VL, Varet J, Millar A, Harper SJ, Soothill PW, Bates DO. Failure
to up-regulate VEGF165b in maternal plasma is a first trimester predictive marker for pre-eclampsia. Clin Sci (Lond) 2009;116:26572.
26 Bosio PM, Wheeler T, Anthony F, Conroy R, OHerlihy C, McKenna
P. Maternal plasma vascular endothelial growth factor concentrations in normal and hypertensive pregnancies and their relationship
to peripheral vascular resistance. Am J Obstet Gynecol 2001;184:
14652.
27 Chaiworapongsa T, Romero R, Kim YM, Kim GJ, Kim MR, Espinoza
J, et al. Plasma soluble vascular endothelial growth factor receptor-1
concentration is elevated prior to the clinical diagnosis of preeclampsia. J Matern Fetal Neonatal Med 2005;17:318.
28 Crispi F, Llurba E, Dominguez C, Martin-Gallan P, Cabero L, Gratacos E. Predictive value of angiogenic factors and uterine artery
Doppler for early- versus late-onset pre-eclampsia and intrauterine
growth restriction. Ultrasound Obstet Gynecol 2008;31:3039.
29 Diab AE, El-Behery MM, Ebrahiem MA, Shehata AE. Angiogenic factors for the prediction of pre-eclampsia in women with abnormal
midtrimester uterine artery Doppler velocimetry. Int J Gynaecol
Obstet 2008;102:14651.
30 Espinoza J, Romero R, Nien JK, Gomez R, Kusanovic JP, Goncalves
LF, et al. Identification of patients at risk for early onset and/or
severe preeclampsia with the use of uterine artery Doppler velocimetry and placental growth factor. Am J Obstet Gynecol 2007;196:
326.e113.
31 Koga K, Osuga Y, Tajima T, Hirota Y, Igarashi T, Fujii T, et al. Elevated serum soluble fms-like tyrosine kinase 1 (sFlt1) level in women
with hydatidiform mole. Fertil Steril 2010;94:3058.
32 Kusanovic JP, Romero R, Chaiworapongsa T, Erez O, Mittal P, Vaisbuch E, et al. A prospective cohort study of the value of maternal
plasma concentrations of angiogenic and anti-angiogenic factors in
early pregnancy and midtrimester in the identification of patients
destined to develop preeclampsia. J Matern Fetal Neonatal Med
2009;22:102138.
33 Lim JH, Kim SY, Park SY, Yang JH, Kim MY, Ryu HM. Effective prediction of preeclampsia by a combined ratio of angiogenesis-related
factors. Obstet Gynecol 2008;111:14039.
34 Lynch AM, Murphy JR, Gibbs RS, Levine RJ, Giclas PC, Salmon JE,
et al. The interrelationship of complement-activation fragments and
angiogenesis-related factors in early pregnancy and their association
with pre-eclampsia. BJOG 2010;117:45662.
35 Madazli R, Kuseyrioglu B, Uzun H, Uludag S, Ocak V. Prediction of
preeclampsia with maternal mid-trimester placental growth factor,
activin A, fibronectin and uterine artery Doppler velocimetry. Int J
Gynaecol Obstet 2005;89:2517.
36 Park CW, Park JS, Shim SS, Jun JK, Yoon BH, Romero R. An elevated
maternal plasma, but not amniotic fluid, soluble fms-like tyrosine
kinase-1 (sFlt-1) at the time of mid-trimester genetic amniocentesis
is a risk factor for preeclampsia. Am J Obstet Gynecol 2005;193:
9849.

786

37 Parra M, Rodrigo R, Barja P, Bosco C, Fernandez V, Munoz H, et al.

Screening test for preeclampsia through assessment of uteroplacental blood flow and biochemical markers of oxidative stress and
endothelial dysfunction. Am J Obstet Gynecol 2005;193:148691.
38 Polliotti BM, Fry AG, Saller DN, Mooney RA, Cox C, Miller RK. Second-trimester maternal serum placental growth factor and vascular
endothelial growth factor for predicting severe, early-onset preeclampsia. Obstet Gynecol 2003;101:126674.
39 Powers RW, Jeyabalan A, Clifton RG, Van DP, Hauth JC, Klebanoff
MA, et al. Soluble fms-Like tyrosine kinase 1 (sFlt1), endoglin and
placental growth factor (PlGF) in preeclampsia among high risk pregnancies. PLoS ONE 2010;5:e13263.
40 Rana S, Karumanchi SA, Levine RJ, Venkatesha S, Rauh-Hain JA, Tamez H, et al. Sequential changes in antiangiogenic factors in early
pregnancy and risk of developing preeclampsia. Hypertension 2007;
50:13742.
41 Robinson CJ, Johnson DD. Soluble endoglin as a second-trimester
marker for preeclampsia. Am J Obstet Gynecol 2007;197:1745.
42 Savvidou MD, Noori M, Anderson JM, Hingorani AD, Nicolaides KH.
Maternal endothelial function and serum concentrations of placental
growth factor and soluble endoglin in women with abnormal
placentation. Ultrasound Obstet Gynecol 2008;32:8716.
43 Schmidt M, Dogan C, Birdir C, Kuhn U, Gellhaus A, Kimmig R, et al.
Placental growth factor: a predictive marker for preeclampsia? Gynakol Geburtshilfliche Rundsch 2009;49:949.
44 Shokry M, Bedaiwy MA, Fathalla MM, Alsemary A, Elwakil S, Murphy A. Maternal serum placental growth factor and soluble fms-like
tyrosine kinase 1 as early predictors of preeclampsia. Acta Obstet
Gynecol Scand 2010;89:1436.
45 Sibai BM, Koch MA, Freire S, Pinto e Silva JL, Rudge MV, MartinsCosta S, et al. Serum inhibin A and angiogenic factor levels in pregnancies with previous preeclampsia and/or chronic hypertension: are
they useful markers for prediction of subsequent preeclampsia? Am
J Obstet Gynecol 2008;199:2689.
46 Stepan H, Geipel A, Schwarz F, Kramer T, Wessel N, Faber R. Circulatory soluble endoglin and its predictive value for preeclampsia in
second-trimester pregnancies with abnormal uterine perfusion. Am J
Obstet Gynecol 2008;198:1756.
47 Stepan H, Unversucht A, Wessel N, Faber R. Predictive value of
maternal angiogenic factors in second trimester pregnancies with
abnormal uterine perfusion. Hypertension 2007;49:81824.
48 Su YN, Lee CN, Cheng WF, Shau WY, Chow SN, Hsieh FJ.
Decreased maternal serum placenta growth factor in early second
trimester and preeclampsia. Obstet Gynecol 2001;97:898904.
49 Thadhani R, Mutter WP, Wolf M, Levine RJ, Taylor RN, Sukhatme
VP, et al. First trimester placental growth factor and soluble fms-like
tyrosine kinase 1 and risk for preeclampsia. J Clin Endocrinol Metab
2004;89:7705.
50 Thadhani R, Ecker JL, Mutter WP, Wolf M, Smirnakis KV, Sukhatme
VP, et al. Insulin resistance and alterations in angiogenesis: additive
insults that may lead to preeclampsia. Hypertension 2004;43:988
92.
51 Tidwell SC, Ho HN, Chiu WH, Torry RJ, Torry DS. Low maternal
serum levels of placenta growth factor as an antecedent of clinical
preeclampsia. Am J Obstet Gynecol 2001;184:126772.
52 Tjoa ML, van Vugt JM, Mulders MA, Schutgens RB, Oudejans CB,
van WI. Plasma placenta growth factor levels in midtrimester pregnancies. Obstet Gynecol 2001;98:6007.
53 Wortelboer EJ, Koster MP, Cuckle HS, Stoutenbeek PH, Schielen PC,
Visser GH. First-trimester placental protein 13 and placental growth
factor: markers for identification of women destined to develop
early-onset pre-eclampsia. BJOG 2010;117:13849.

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

Biomarkers for prediction of pre-eclampsia

54 Yu J, Shixia CZ, Wu Y, Duan T. Inhibin A, activin A, placental growth

factor and uterine artery Doppler pulsatility index in the prediction of
pre-eclampsia. Ultrasound Obstet Gynecol 2011;37:52833.
55 Levine RJ, Maynard SE, Qian C, Lim KH, England LJ, Yu KF, et al.
Circulating angiogenic factors and the risk of preeclampsia. N Engl J
Med 2004;350:67283.
56 Levine RJ, Lam C, Qian C, Yu KF, Maynard SE, Sachs BP, et al.
Soluble endoglin and other circulating antiangiogenic factors in preeclampsia. N Engl J Med 2006;355:9921005.
57 National Collaborating Centre for Womens and Childrens Health.
Antenatal Care: Routine Care for the Healthy Pregnant Woman
[Clinical guideline]. London: National Institute for Health and Clinical
Excellence, 2008.

58 Nicolaides KH. A model for a new pyramid of prenatal care based

on the 11 to 13 weeks assessment. Prenat Diagn 2011;31:36.
59 Kenny LC, Broadhurst DI, Dunn W, Brown M, North RA, McCowan
L, et al. Robust early pregnancy prediction of later preeclampsia
using metabolomic biomarkers. Hypertension 2010;56:7419.
60 North RA, McCowan LM, Dekker GA, Poston L, Chan EH, Stewart
AW, et al. Clinical risk prediction for pre-eclampsia in nulliparous
women: development of model in international prospective cohort.
BMJ 2011;342:d1875.
61 Poon LC, Stratieva V, Piras S, Piri S, Nicolaides KH. Hypertensive disorders in pregnancy: combined screening by uterine artery Doppler,
blood pressure and serum PAPP-A at 1113 weeks. Prenat Diagn
2010;30:21623.

2012 The Authors BJOG An International Journal of Obstetrics and Gynaecology 2012 RCOG

787