You are on page 1of 12

Ultrasound Obstet Gynecol 2018; 51: 720–730

Published online 8 May 2018 in Wiley Online Library ( DOI: 10.1002/uog.18959

Genetic and non-genetic risk factors for pre-eclampsia:

umbrella review of systematic reviews and meta-analyses of
observational studies
1 Cyprus International Institute for Environmental & Public Health, Cyprus University of Technology, Limassol, Cyprus; 2 Department of

Hygiene and Epidemiology, University of Ioannina Medical School, Ioannina, Greece; 3 Department of Epidemiology and Biostatistics,
School of Public Health, Imperial College London, London, UK

K E Y W O R D S: epidemiology; meta-analysis; pre-eclampsia; risk factors; umbrella review

ABSTRACT a summary odds ratio of 4.33 (95% CI, 3.11–6.03),

was supported by 2712 cases with small heterogeneity
Objective To summarize evidence from the literature
(I2 = 26%) and 95% prediction intervals excluding the
on genetic and non-genetic risk factors associated with
null value, and without hints of small-study effects (P for
pre-eclampsia (PE), assess the presence of statistical
Egger’s test > 0.10) or excess of significance (P > 0.05).
bias in the studies and identify risk factors for which
Of the statistically significant (P < 0.05) genetic risk
there is robust evidence supporting their association
factors for PE, only PAI-1 4G/5G (recessive model)
with PE.
polymorphism was supported by strong evidence for a
Methods PubMed and ISI Web of Science were searched contribution to the pathogenesis of PE. Eleven factors
from inception to October 2016, to identify systematic (serum iron level, pregnancy-associated plasma protein-A,
reviews and meta-analyses of observational studies chronic kidney disease, polycystic ovary syndrome, mental
examining associations between genetic or non-genetic stress, bacterial and viral infections, cigarette smoking,
risk factors and PE. For each meta-analysis, the oocyte donation vs assisted reproductive technology,
summary-effect size was estimated using random-effects obesity vs normal weight, severe obesity vs normal weight
and fixed-effects models, along with 95% CIs and the and primiparity) presented highly suggestive evidence for
95% prediction interval. Between-study heterogeneity an association with PE.
was expressed using the I2 statistic, and evidence of
Conclusions A large proportion of meta-analyses of
small-study effects (large studies had significantly more
genetic and non-genetic risk factors for PE have
conservative results than smaller studies) and evidence of
caveats that threaten their validity. Oocyte donation vs
excess significance bias (too many studies with statistically
spontaneous conception and PAI-1 4G/5G polymorphism
significant results) were estimated.
(recessive model) showed the strongest consistent evidence
Results Fifty-eight eligible meta-analyses were identified, for an association with risk for PE. Copyright © 2017
which included 1466 primary studies and provided ISUOG. Published by John Wiley & Sons Ltd.
data on 130 comparisons of risk factors associated
with PE, covering a wide range of comorbid diseases, INTRODUCTION
genetic factors, exposure to environmental agents and
biomarkers. Sixty-five (50%) associations had nominally Pre-eclampsia (PE) is a severe pregnancy-associated
statistically significant findings at P < 0.05, while 16 disease characterized by the occurrence of hypertension
(12%) were significant at P < 10 –6 . Sixty-five (50%) and proteinuria in previously healthy pregnant women
associations had large or very large heterogeneity. after 20 weeks’ gestation. PE affects approximately 2–8%
Evidence for small-study effects and excess significance of all pregnancies and is associated with substantially
bias was found in 10 (8%) and 26 (20%) associations, higher maternal and fetal morbidity and mortality
respectively. The only non-genetic risk factor with worldwide1,2 . The clinical spectrum of PE ranges from
convincing evidence for an association with PE was mild PE, which is characterized by a moderate increase
oocyte donation vs spontaneous conception, which had in blood pressure and proteinuria, to the most severe

Correspondence to: Dr S. I. Papatheodorou, Cyprus International Institute for Environmental and Public Health, 95 Eirinis St, 3041 Limassol,
Cyprus (e-mail:
Accepted: 30 October 2017

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. SYSTEMATIC REVIEW
Genetic and non-genetic risk factors for PE 721

form manifesting with seizures as a sign of damage review’ OR ‘meta-analysis’). First, the title and abstract
to the cerebral vessels, and HELLP syndrome, which of each article identified through the search were exam-
significantly threatens the lives of pregnant women ined, and then the full texts of potentially eligible articles
and their fetuses3 . The true etiology of PE remains were retrieved for evaluation. PubMed was also searched
an issue of debate, which generates uncertainty in the systematically to identify genome-wide association studies
prediction, prevention and treatment of the disease, examining genetic associations with PE. Any discrepancies
occurring as interaction between genetic and non-genetic were resolved by discussion.
factors4,5 .
Numerous meta-analyses and systematic reviews have
claimed that several environmental, biological and genetic Eligibility criteria and data extraction
risk factors are associated with risk for PE. If causal, Articles were eligible for inclusion if the authors had
these associations might be useful for the accurate performed a systematic search to identify pertinent studies
prediction and diagnosis of this condition in early that examined the association between various risk factors
pregnancy, which would allow a timely allocation of and PE. The full texts of potentially eligible articles were
screening resources and prevention of maternal and fetal scrutinized independently by two investigators (K.G.,
complications6–8 . In addition, preventive measures such S.P.). Meta-analyses or systematic reviews were retained
as aspirin administration in high-risk women appear more if they included at least three studies in which information
likely to be beneficial if started earlier in pregnancy, i.e. per included study was provided on a measure of
during the first trimester or even before conception9,10 . association and its standard error (SE) between the risk
Nevertheless, it is possible that some associations observed factor and PE and on the number of cases per population
in the literature do not reflect a genuine result but include in each study included in the meta-analysis. We excluded
different types of bias in favor of positive statistically studies in which risk factors were used for screening,
significant associations11 . The pursuit of positive results diagnostic or prognostic purposes, or meta-analyses that
may be generated by several different mechanisms12 , such examined PE as a risk factor for other medical conditions.
as selective analyses, outcome bias and fabrication bias13 . No language restriction was applied. If more than one
These biases can cause either false published findings or meta-analysis on the same research question was eligible,
inflated effects12,14 . the one with the largest number of component studies
To our knowledge, this is the first attempt to summarize with data on effect size of individual studies was retained
evidence from existing meta-analyses on genetic and for the main analysis.
non-genetic risk factors for PE. We aimed to summarize Data extraction was performed independently by two
evidence on risk factors that have been associated with investigators (K.G., S.P.), and in case of disagreement, the
PE, evaluate whether biases are present in the literature final decision was reached by discussion between them or,
and how they manifest and finally identify which of the if necessary, by review by a third investigator (E.E.). From
previously studied associations are supported by robust each eligible meta-analysis, information was extracted
epidemiological evidence. on the first author, year of publication, examined risk
factors, number of studies included, study-specific relative
METHODS risk estimates (risk ratio, odds ratio) or standardized
mean differences along with the corresponding CIs. Also,
Umbrella review reported summary meta-analytic estimates were recorded
using both fixed- and random-effects methods along with
This was an umbrella review, which is a systematic col- the corresponding CIs and the number of cases and
lection and evaluation of multiple systematic reviews and controls for each study. We noted whether or not the
meta-analyses performed on a specific research topic15 . An selected meta-analyses applied any criteria to evaluate the
umbrella review brings together comparisons of a large quality of the included observational studies.
number of existing systematic reviews and meta-analyses
into one accessible and practical document15,16 . The meth-
ods of the umbrella review are standardized and in this Assessment of summary effect and heterogeneity
work we followed state-of-the-art approaches, as seen For each meta-analysis, the summary effects with 95%
in other previously published umbrella reviews on risk CIs were estimated using both fixed- and random-effects
factors and various outcomes17–20 . models21,22 . Additionally, the 95% prediction intervals
(PI) were calculated for the summary random-effects
Literature search estimates, which further account for between-study
heterogeneity and indicate the uncertainty for the effect
Two researchers (K.G. and S.P.) independently searched that would be expected in the new study observing the
PubMed and ISI Web of Science from inception to 8 Octo- same association23,24 . The SE of the effect size was
ber 2016, to identify systematic reviews and meta-analyses calculated for the largest study of each meta-analysis,
of observational studies examining associations between and we examined whether it was < 0.10 and whether the
risk factors and PE. The search strategy used the keywords largest study presented a statistically significant effect. In
(‘pre-eclampsia’ OR ‘preeclampsia’) AND (‘systematic a study with a SE of < 0.10, the difference between the

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
722 Giannakou et al.

effect estimate and the upper or lower 95% CI is < 0.20 categories, as selective reporting bias may arise in different
(i.e. this uncertainty is less than what is considered a small domains of research. The excess-of-statistical-significance
effect size). test was conducted separately for meta-analyses with I2
Heterogeneity between studies was assessed, and we values ≤ 50% and those with values > 50%, because
reported the P-value of the χ2 -based Cochran Q test values above 50% typically reflect evidence of large
and the I2 statistic for inconsistency, which could reflect heterogeneity beyond chance31 .
either diversity or bias. I2 ranges between 0% and 100%
and indicates the ratio of between-study variance over
Grading of non-genetic and genetic associations with PE
the sum of within and between-study variances25 . Values
exceeding 50% or 75% are usually considered to represent The association of non-genetic risk factors with PE was
large or very large heterogeneity, respectively. CIs were characterized as: (a) ‘convincing’ if the effect under
calculated as per Ioannidis et al.26 . the random-effects model was significant at P < 10 –6 ,
it was based on evidence arising from > 1000 cases,
Assessment of small-study effect between-study heterogeneity was not large (I2 < 50%),
the 95% PI excluded the null value and there was no
The presence of the small-study effect (i.e. if small evidence of small-study effects and excess of significance
studies tend to give higher risk estimates than large bias; (b) ‘highly suggestive’ if the effect was significant at
studies) was evaluated. The small-study effect can indicate P < 10 –6 , it was based on > 1000 cases and a nominally
publication and other selective reporting biases, but can statistically significant effect was present in the largest
also reflect genuine heterogeneity, chance or other reasons study; (c) ‘suggestive’ if the effect was significant at
for differences between small and large studies27 . The P < 10 –3 and it was based on >1000 cases; and (d) ‘weak’
regression asymmetry test proposed by Egger et al.28 was for statistically significant associations at P < 0.05 based
used for this assessment. P < 0.10 accompanied by a on the random-effects model.
more conservative effect in larger studies was considered The Venice criteria were used to evaluate the
evidence for the presence of a small-study effect. epidemiological credibility of all significant genetic
associations32 . Credibility was defined based upon the
Evaluation of excess statistical significance grade (A = strong, B = moderate or C = weak) of three
categories: amount of evidence, replication of the
The excess-of-statistical-significance test was performed association and protection from bias. The amount of
to evaluate whether there was a relative excess of evidence was graded by the sum of test alleles or genotypes
formally significant findings in the published literature among both cases and controls in the meta-analysis:
for any reason. The number of expected positive studies ‘A’ for > 1000, ‘B’ for 100–1000, and ‘C’ for < 100.
is estimated and compared with the observed number of Replication of the association was graded as ‘A’ if there
studies with statistically significant results (P < 0.05)13 . was an extensively replicated study supported by at
A binomial test was used to evaluate whether the least one well-conducted meta-analysis, ‘B’ if it was a
number of positive studies in a meta-analysis was too well-conducted meta-analysis with some methodological
large according to the power that these studies have to limitations and ‘C’, if there was no independent
detect plausible effects at α = 0.05. A comparison between replication, failed replication or flawed meta-analysis.
observed and expected studies with statistically significant Assessment of protection from bias included consideration
results is performed separately for each meta-analysis, of the magnitude of the association, heterogeneity statistic
and it is extended to groups of several meta-analyses and findings from tests for selective reporting biases (tests
after summing the observed and expected studies from for small-study effects and excess statistical significance).
each meta-analysis. The power of each component study According to these criteria, the credibility level of the
was estimated using the fixed-effects summary, the cumulative evidence was defined as high (A grades only),
random-effects summary, or the effect size of the largest low (one or more C grades) or intermediate (all other
study (smallest SE) as the plausible effect size29 . The combinations)32 .
power of each study was calculated with an algorithm All authors had full access to all data in the study.
using a non-central t distribution30 . Excess statistical Statistical analysis and power calculations were performed
significance for single meta-analyses was claimed at in STATA version 14 (STATA Corp., College Station,
P < 0.10 (one-sided P < 0.05, with observed > expected TX, USA).
as previously proposed), given that the power to detect
a specific excess will be low, especially with few positive
studies13 . RESULTS
Risk factors were classified, based on biological Description of eligible meta-analyses
pathways or types of exposure involved, into the following
categories: biomarkers, environmental factors, genetic The search identified 635 items, of which 535 were
markers, diseases and disorders, dietary supplementation, excluded after review of the title and abstract (Figure 1).
infections and other risk factors. Excess statistical Of the remaining 100 articles that were reviewed in
significance was examined separately in each of these full text, eight articles did not report the appropriate

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
Genetic and non-genetic risk factors for PE 723

References identified (n = 635)

Titles reviewed (n = 635)

References excluded (n = 403):

PE as a risk factor for other outcomes (n = 77)
Study of efficacy of treatments (n = 48)
Diagnostic, prognostic or screening study (n = 22)
Intervention study for prevention of PE (n = 119)
Outcome other than PE (n = 76)
Incidence or prevalence study (n = 8)
Editorial or narrative review (n = 39)
Methodological study (n = 14)

Abstracts reviewed (n = 232)

References excluded (n = 132):

Primary studies (n = 6)
PE as a risk factor for other outcomes (n = 22)
Study of efficacy of treatments (n = 10)
Diagnostic, prognostic or screening study (n = 4)
Intervention study for prevention of PE (n = 30)
Editorial or narrative review (n = 29)
Outcome other than PE (n = 28)
Incidence or prevalence study (n = 3)

Articles reviewed (n = 100)

References excluded (n = 29):

Pooled analysis of cohort studies (n = 1)
Meta-analysis with only two component studies (n = 2)
Duplicate data/largest systematic review or meta-analysis
investigating same risk factor (n = 18)
Appropriate information not available for required
calculations (n = 8)

Articles included (n = 71)

Systematic reviews Meta-analyses included in

included in quantitative synthesis
qualitative synthesis (n = 58)
(n = 13) (data on 130 comparisons)

Figure 1 Flowchart showing systematic reviews and meta-analyses of observational studies reporting on genetic and non-genetic factors
associated with pre-eclampsia (PE), included in this umbrella review.

information for the calculation of excess of statistical The characteristics of the included meta-analyses
significance (either because the total sample size was are shown in Table S1. Based on the study design
missing or the study-specific relative risk estimates were of the synthesized studies that examined non-genetic
missing), one article was a pooled analysis of cohort associations, seven (20%) meta-analyses synthesized
studies, two articles included only two component retrospective case–control data only, three (9%) included
studies, and 18 articles were excluded because a larger prospective data (cohort studies) and 25 (71%) included
systematic review or meta-analysis investigating the same both types of data, noted as mixed. With respect to
risk factor was available. Therefore, 71 articles were the genetic association studies, 15 (65%) meta-analyses
analyzed, of which 13 were systematic reviews without synthesized case–control data, seven (30%) used both
any quantitative component and 58 were meta-analyses. types of data (case–control and cohort data), and one
The 58 eligible meta-analyses5,33–89 included data on (4%) meta-analysis included only cohorts.
130 comparisons in seven broad areas (biomarkers (27 Three to 51 studies, with a median of eight studies,
comparisons), environmental factors (six comparisons), were included per meta-analysis. The median number
genetic markers (66 comparisons), diseases and disorders of case and control subjects in each study was 96
(eight comparisons), supplementation (one comparison), and 161, respectively. The median number of case
infections (three comparisons) and other risk factors (19 and control subjects in each meta-analysis was 1123
comparisons)). and 3598, respectively. The number of cases was

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
724 Giannakou et al.

greater than 1000 in 70 comparisons. Overall, 441 vs normal conception, obesity vs normal weight, severe
(30%) individual studies observed nominally statistically obesity vs normal weight and primiparity.
significant results. Twenty-one (36%) meta-analyses used Across the seven categories of risk factors, differences
the Newcastle–Ottawa Scale to assess qualitatively the were observed in the proportion of associations that
included primary studies. Two meta-analyses used assess- had nominally statistically significant summary effects.
ment criteria for non-randomized observational studies Based on random effects calculations at P < 0.05, 100%,
adapted from Duckitt and Harrington, two meta-analyses 75%, 63% and 59% of the meta-analyses on infections,
used the Methodological Index for Non-Randomized diseases and disorders, other risk factors and biomarkers,
Studies (MINORS) and nine meta-analyses used other respectively, found nominally statistically significant
assessment tools. Twenty-four (41%) meta-analyses summary effects. On the other hand, this was seen only in
did not perform any quality assessment. Details of the 39% and 33% of the meta-analyses on genetic markers
130 comparisons that included 1466 individual study and environmental factors, respectively.
estimates are summarized in Table S2.
Between-study heterogeneity and prediction intervals

Systematic reviews with qualitative synthesis Thirty-three (25%) comparisons had large (I2 ≥ 50%
and ≤ 75%) and 32 (25%) had very large (I2 > 75%)
We also summarized evidence of systematic reviews heterogeneity estimates (Table S1). The highest pro-
that did not include any quantitative component90–102 . portion (56%) of I2 exceeding 75% was observed
According to the findings of these studies, serum in comparisons of biomarkers. When calculating the
calprotectin and cardiac troponin-I levels were elevated 95% PIs, the null value was excluded in only 14
in women with PE compared with healthy controls, and (11%) comparisons. This included two comparisons on
cell-free fetal DNA quantification has been shown to be biomarkers (PAPP-A and Vitamin D < 50 mmol/L), five
a promising marker for PE prediction, especially for the on genetic markers (G20210A single nucleotide polymor-
development of early-onset or severe PE90–92 . PE was phism (SNP), PAI-1 4G/5G, F5 rs6025, F2 rs1799963,
more prevalent in cold and humid seasons93 , and a long AGT/T704C-Met235Thr), two on diseases and disorders
interpregnancy interval, possibly longer than 5 years, (chronic kidney disease and PCOS) and five on other
was also associated independently with an increased risk factors (oocyte donation vs ART, oocyte donation
risk for PE94 . Psychotropic drugs such as lithium for vs spontaneous conception, high vs low levels of physical
the management of antenatal psychiatric disorders have activity in early pregnancy, obese vs normal weight and
also been associated with PE95 . Pregnant women with primiparity) (Table S1).
systemic lupus erythematosus or Cushing’s syndrome
are at higher risk of developing PE compared with
Small-study effects
healthy pregnant women96,97 . Laparoscopic adjustable
gastric band (LAGB) surgery seems to improve pregnancy Evidence for statistically significant small-study effects
outcome for conditions such as PE in obese women, com- (Egger test P < 0.10 and random-effects summary
pared with those who do not have LAGB98,99 . Limited estimate larger compared with the point estimate of the
evidence was found on whether shift work, HIV infection largest study in the meta-analysis) was identified in 10
or antiretroviral therapy, and thrombophilic disorders of 130 (8%) comparisons (Table S2). These included
are associated with an increased risk for PE100–102 . two meta-analyses on biomarkers (PAPP-A, PlGF), one
on environmental factors (oxides of nitrogen), four on
genetic markers (NOS3 27 bp-VNTR in intron 4, F2
Summary-effect sizes and significant findings rs1799963, ACE rs4646994, ACE-I/D), two on diseases
and disorders (PCOS and mental stress) and one on other
Of the 130 comparisons, 65 (50%) had nominally risk factors (prepregnancy physical activity of 1 h per day).
statistically significant findings at P < 0.05 using the
random-effects model, of which 53 reported an increased
risk and 12 a decreased risk for PE. Of these, a Test of excess statistical significance
total of 28 (22%) associations presented statistically Hints of excess-statistical-significance bias were observed
significant effect at P < 0.001, while only 16 (12%) in 26 (20%) associations, with statistically significant
remained significant after the application of a more (P < 0.05) excess of positive studies under any of the
stringent P-value threshold of P < 10 –6 (Table S1). The three assumptions for the plausible effect size, i.e. the
sixteen risk factors that presented a significant effect at fixed-effects summary, random-effects summary or results
P < 10 –6 for an association with PE were: serum iron of the largest study (Table S2). Of these, 10 (38%)
level, pregnancy-associated plasma protein-A (PAPP-A), pertained to biomarkers, nine (35%) to genetic markers,
placental protein 13, placental growth factor (PlGF), F5 three (12%) to diseases and disorders and four (15%) to
rs6025, F2 rs1799963, chronic kidney disease, polycystic other risk factors. In addition, the observed and expected
ovary syndrome (PCOS), mental stress, bacterial and number of positive studies showed that, overall, the
viral infections, cigarette smoking, oocyte donation vs excess of positive results was driven by meta-analyses
assisted reproductive technology (ART), oocyte donation with small estimates of heterogeneity (I2 ≤ 50%). The

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
Genetic and non-genetic risk factors for PE 725

results of excess-of-statistical-significance bias according a summary odds ratio (OR) of 4.33 (95% CI, 3.11–6.03;
to category of risk factor are shown in Table 1. P = 3.48 × 10 –18 ) with small heterogeneity (I2 = 26%),
and was supported by 2712 cases.
Moreover, 11 risk factors (serum iron level, PAPP-A,
Risk factors with strong evidence of association
chronic kidney disease, PCOS, mental stress, bacterial and
After applying our credibility criteria, only one non- viral infections, cigarette smoking, oocyte donation vs
genetic risk factor, oocyte donation vs spontaneous con- ART, obesity vs normal weight, severe obesity vs normal
ception, presented convincing evidence for an association weight and primiparity) presented highly suggestive
with PE, i.e. it was supported by data from more than evidence for an association with PE, five risk factors
1000 cases, P < 10 –6 based on the random-effects model, were supported by suggestive evidence, and 22 by weak
there was an absence of small-study effects and excess evidence. An overall assessment of statistically significant
statistical significance, low heterogeneity (I2 < 50%) and associations of risk factors with PE is presented in
a 95% PI excluding the null value. This association had Table 2.

Table 1 Excess of statistical significance bias according to category of risk factor associated with pre-eclampsia, based on expected number
of studies with statistically significant findings estimated using fixed-effects summary, random-effects summary, effect of largest study and
most conservative of three estimates (composite)

Fixed-effects summary Random-effects summary Effect of largest study Composite

Observed Expected Expected Expected Expected
Risk factor Total positive positive positive positive positive
category studies studies studies† P* studies‡ P* studies§ P* studies¶ P*

Total 1466 479 560.3 0.00 605.9 0.00 601.3 0.00 560.3 0.00
Biomarkers 353 178 166 0.20 200 0.02 133 0.00 133 0.00
Environmental 23 4 4.9 0.80 4.4 NP 10.5 0.01 4.4 NP
Genetic markers 830 162 229.6 0.00 235.5 0.00 323.4 0.00 229.6 0.00
Diseases & disorders 59 29 37.6 0.03 45 0.00 27.4 0.70 27.4 0.70
Supplementation 3 0 0.32 NP 0.32 NP 0.3 NP 0.3 NP
Infections 47 21 27.3 0.08 28.9 0.02 23 0.66 23 0.66
Other 151 85 95 0.09 92.2 0.24 84 0.93 84 0.93

Data are n. *P of excess of statistically significant test; all statistical tests were two-sided. Expected number of statistically significant studies
using as plausible effect size: †summary fixed-effects estimate of each meta-analysis; ‡summary random-effects estimate of each
meta-analysis; §effect of largest study of each meta-analysis; ¶most conservative of three estimates (fixed-effects summary, random-effects
summary, largest study) of each meta-analysis. NP, not pertinent, because estimated is larger than observed, and there is no evidence of
excess of statistical significance based on assumption made for plausible effect size.

Table 2 Non-genetic risk factors associated significantly with pre-eclampsia, according to level of evidence for this association

Level of
evidence Criteria Risk factor

Convincing > 1000 cases, ∗P < 10 –6 , not large heterogeneity Oocyte donation vs spontaneous conception
(I2 < 50%), 95% prediction interval excluding null
value, no evidence for small-study effects† or excess
significance bias‡
Highly sug- > 1000 cases, ∗P < 10 –6 and nominally statistically Serum iron level, PAPP-A, chronic kidney disease, polycystic ovary
gestive significant effect present in largest study syndrome, mental stress, bacterial & viral infections, cigarette
smoking§, oocyte donation vs ART, obesity vs normal weight,
severe obesity vs normal weight, primiparity
Suggestive > 1000 cases, ∗P < 10 –3 Serum vitamin C§, sFlt-1, depression, periodontal disease, high vs
low PA in early pregnancy§
Weak *P < 0.05 β-hCG, serum zinc level§, PP-13, inhibin A, IFN-γ, serum
concentration of NO§, PlGF§, sENG, arterial stiffness,
anticardiolipin antibodies, NO2 , O3 , work stress, depression and
anxiety, 25 (OH) D < 75 mmol/L, 25 (OH) D < 50 mmol/L,
chronic hepatitis B infection§, high vs low prepregnancy PA§,
prepregnancy PA per 1 h per day§, early pregnancy PA per 1 h
per day§, early pregnancy walking§, donor insemination

P-value on meta-analysis random-effects model. †Small-study effect based on P-value from Egger’s regression asymmetry test (P < 0.10).
‡P < 0.05 on excess significance test using largest study (smallest standard error) in a meta-analysis as plausible effect size. §Factors that
show a protective effect against developing pre-eclampsia. ART, assisted reproductive technology; β-hCG, β-human chorionic gonadotropin;
IFN-γ, interferon gamma; NO, nitric oxide; NO2 , nitrogen dioxide; O3 , ozone; PA, physical activity; PAPP-A, pregnancy-associated plasma
protein-A; PlGF, placental growth factor; PP-13, placental protein 13; sENG, soluble endoglin; sFlt-1, soluble fms-like tyrosine kinase-1;
25 (OH) D, 25-hydroxyvitamin D.

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
726 Giannakou et al.

Cumulative epidemiological evidence for genetic genes are expressed, whereas a fetus conceived through
associations with PE was scored as strong, moderate oocyte donation is an absolute allograft, and this could
or weak based on grades of ‘A’, ‘B’ or ‘C’, as specified lead to altered or inadequate immune protection of
by the Venice criteria. Of the 26 variants with signifi- placentation and eventually result in PE107–110 . This
cant associations with risk of PE, according to P < 0.05 theory is further supported by the observation that oocyte
using the random-effects model, only the PAI-1 4G/5G donation vs other assisted reproduction techniques had
polymorphism (recessive model) was supported by strong highly suggestive evidence of epidemiological credibility.
evidence for a contribution to the pathogenesis of PE Moreover, immune dysregulation may explain the
(Table 3). highly suggestive evidence of an increased risk of PE
among primiparous women because the first successful
(non-pre-eclamptic) pregnancy may induce adaptive
Independent tool-based quality assessment of primary
changes in favor of immune tolerance in subsequent
pregnancies89 .
The quality of the included studies of the meta-analysis The genetic architecture behind PE is complex111 .
of the non-genetic risk factor that presented convincing To date, most research in this field has been focused
evidence for an association with PE was also assessed on candidate genes, primarily those with a plausible
using the Newcastle–Ottawa Scale in addition to the role in the known underlying pathophysiology112 . We
MINORS scale that the authors used in the original identified only three genome-wide association studies,
assessment103 . The methodological quality ranged from which included several genetic loci associated with
3 points to 8 points maximally, with a median of 6 PE113–115 . One study identified two loci (rs7579169 and
points, which implies a fair quality for the majority of rs12711941) near the inhibin beta B gene that satisfied
studies. Quality assessment was also performed among the genome-wide significance threshold114 , but the results
the included studies of meta-analysis of the PAI-1 4G/5G could not be replicated in two cohorts from Norway and
polymorphism using the Q-Genie tool104 . Among the Finland. Subsequent case–control studies in European
reviewed studies, eight (67%) were rated to have high and Chinese women have shown a significant association
quality (> 45) and four (33%) were rated to have between the SNP rs7579169 and PE116,117 .
moderate quality (> 35 and ≤ 45). Eleven factors (serum iron level, PAPP-A, chronic
kidney disease, PCOS, mental stress, bacterial and viral
infections, cigarette smoking, oocyte donation vs ART,
DISCUSSION obesity/severe obesity vs normal weight, primiparity)
Main findings achieved highly suggestive evidence for an association
with PE. There are several mechanisms that support these
Overall, 130 associations have been studied as risk findings. With respect to biomarkers, iron is considered a
factors for PE, including biomarkers, genetic markers, significant etiologic factor of the endothelial cell damage
environmental factors, supplementation, diseases and in PE because of its effects on the formation of oxygen
disorders, infections and other risk factors. Of those, free radicals and subsequent lipid peroxidation118–120 .
oocyte donation vs spontaneous conception provided Reduced levels of PAPP-A, being an important regulator
convincing evidence for an association with PE. PAI-1 of insulin-like growth factor, can play a role in the devel-
4G/5G (recessive model) polymorphism was supported opment of PE in pregnancies with normal karyotype121 .
by strong evidence for a contribution to the pathogenesis Renal insufficiency, maternal hypertension, proteinuria
of PE, as specified by the Venice criteria. Eleven risk and recurrent urinary tract infection, which often coexist
factors from various categories achieved highly suggestive in women with chronic kidney disease, may contribute
evidence for an association with PE. individually and cumulatively to PE122–124 . Insulin
resistance and/or associated hyperglycemia, which is/are
Interpretation often present in PCOS and obese patients, could be a
possible explanation of a higher risk for PE, since they
PE remains a ‘disease of theories’, as a large number might directly predispose women to hypertension through
of factors and a genetic component are probably increased renal sodium re-absorption and stimulation
involved in its pathogenesis, but to date none has been of the sympathetic nervous system and/or may impair
established clearly. From a biological standpoint, oocyte endothelial function125 . Increased levels of androgens in
donation can act as an independent risk factor for the PCOS pregnancies have also been associated with the
development of PE. During normal pregnancy, various development of PE126 .
immunosuppressive mechanisms maintain a diminished Cigarette smoking during pregnancy seems to have
innate immune response in order to prevent fetal rejection, a protective effect against the development of PE.
as the fetal tissue is exposed directly to the maternal Experimental studies have demonstrated that carbon
blood and, hence, is at risk of being attacked by monoxide decreases the levels of antiangiogenic factors
components of both the innate and the acquired immune such as soluble fms-like tyrosine kinase-1 and soluble
systems105,106 . A fetus conceived spontaneously is a endoglin, or increases the levels of angiogenic factors like
semi-allograft, in which both maternal and paternal vascular endothelial growth factor127 , which are thought

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
Genetic and non-genetic risk factors for PE 727

Table 3 Assessment of cumulative evidence on significant (P < 0.05 on random-effects model) associations of genetic variants with risk of

Cases/ Excess Cumulative

Gene or Studies controls Random statistical Venice evidence of
Study variant Comparison (n) (n) effects* P random Egger† I2 (P)‡ signif§ criteria¶ association**

Zeng G894 T TT vs 26 3241/6419 1.45 0.0118 0.65 41 (0.02) No BAA ++

(2016)48 GT + GG (1.09–1.94)
Zhang rs18001133 in Carriers vs 49 13 356/23 082 1.17 5.89 × 10 –3 0.32 75 (< 0.01) No AAB ++
(2016)49 MTHFR non-carriers (1.05–1.31)
Zhang rs6025 in F5 Carriers vs 28 8210/9834 1.53 0.0239 0.61 74 (< 0.01) No BAB ++
(2016)49 gene non-carriers (1.06–2.21)
Zhang rs1137101 in Carriers vs 28 8210/9834 1.53 0.0239 0.61 74 (< 0.01) No BAB ++
(2016)49 LEPR non-carriers (1.06–2.21)
Li A1675G of GG vs 5 972/3072 1.58 0.0269 0.47 50 (0.09) No BAB ++
(2015)50 AT2R AG + AA (1.05–2.37)
Yang IL-10 -819 C/T C vs T 5 729/1146 1.28 0.0248 0.86 41 (0.15) No AAB ++
(2014)51 (1.03–1.51)
Yang IL-10 -592 C/A C vs A 3 459/926 1.28 0.0264 0.39 0 (0.46) No BAA ++
(2014)51 (1.03–1.59)
Wang G20210A SNP GG vs GA/AA 16 2296/3262 1.79 2.55 × 10 –3 0.96 0 (0.92) No AAB ++
(2014)52 (1.23–2.61)
Wang V G1691A GG vs GA/AA 23 3131/4036 1.60 2.44 × 10 –4 < 0.01 15 (0.25) No AAB ++
(2014)52 SNP (1.25–2.06)
Li MTHFR CT + TT vs 47 6238/11 771 1.12 5.19 × 10 –3 0.16 14 (0.21) Yes AAB ++
(2014)53 C677T CC (1.04–1.22)
Li TGF-β TT vs 4 466/618 0.70 6.05 × 10 –4 0.93 0 (0.84) No BAA ++
(2014)55 1869 T > C TC + CC (0.57–0.86)
Buurma CTLA4 Carriers vs 4 353/536 1.25 0.0434 0.82 14 (0.32) No BAA ++
(2013)5 rs231775 non-carriers (1.01–1.56)
Buurma LPL rs268 Carriers vs 4 530/933 2.43 0.0081 0.66 20 (0.29) No BAA ++
(2013)5 non-carriers (1.26–4.68)
Cheng VEGF +936 T vs C 8 805/1033 1.52 0.0144 0.58 69 (< 0.01) No BAC +
(2013)56 C/T (1.09–2.12)
Song VEGF - 634 C vs G 6 408/479 1.35 6.67 × 10 –3 0.86 12 (0.34) No BAB ++
(2013)57 C/G (1.09–1.67)
Morgan PAI-1 4G/5G or 12 1511/3492 1.28 2.65 × 10 –3 0.56 0 (0.63) No AAA +++
(2013)58 5G/5G (1.09–1.50)
Zhao SERPINE1 4G/4G vs 11 1297/1791 1.37 5.11 × 10 –3 0.42 20 (0.25) No BAB ++
(2013)59 -675 4G/5G + (1.10–1.71)
Staines-Urias F5 rs6025 Carriers vs 41 4499/15 188 1.74 2.90 × 10 –13 0.56 0 (0.53) Yes AAB ++
(2012)60 non-carriers (1.50–2.02)
Staines-Urias F2 rs1799963 Carriers vs 30 3546/11 712 1.72 3.21 × 10 –7 0.03 0 (0.55) Yes BAB ++
(2012)60 non-carriers (1.40–2.12)
Staines-Urias ACE Carriers vs 30 3101/5134 1.17 0.0171 0.06 68 (< 0.01) Yes AAC +
(2012)60 rs4646994 non-carriers (1.03–1.34)
Staines-Urias AGT rs699 Carriers vs 27 2329/4896 1.26 0.0111 0.32 70 (< 0.01) No AAB ++
(2012)60 non-carriers (1.05–1.51)
Lin AGT M235 T TT vs MM 29 5053/11 578 1.61 9.99 × 10 –4 0.47 45 (< 0.01) Yes AAC +
(2012)61 (1.21–2.14)
Zhong ACE D/I D vs I 11 1600/1898 1.93 7.83 × 10 –3 0.26 91 (< 0.01) Yes AAC +
(2012)63 (1.19–3.12)
Medica AGT/T704C CC + TT vs 15 1146/2276 1.66 2.24 × 10 –3 0.77 6 (0.38) No BAB ++
(2007)67 (Met235Thr) TT (1.20–2.29)
Serrano ACE-I/D Carriers vs 22 2596/3828 1.23 0.0174 0.01 57 (< 0.01) No AAC +
(2006)68 non-carriers (1.04–1.45)
Lin FVL Carriers vs 11 1135/1471 2.25 4.61 × 10 –3 0.43 57 (< 0.01) No BAA ++
(2005)69 (1691 G-A) non-carriers (1.28–3.94)

Only first author of each study is given. *Summary random effects odds ratio (95% CI) of each meta-analysis. †P-value from Egger regression
asymmetry test for evaluation of publication bias. ‡I2 metric of inconsistency and P-value of Cochran Q test for evaluation of heterogeneity.
§Based on P-value (P < 0.05) of excess significance test using largest study (smallest standard error) in a meta-analysis as plausible effect size.
¶Venice criteria grades are in order of amount of evidence, replication of association and protection from bias. **Cumulative epidemiological
evidence as graded by Venice criteria as strong (+++), moderate (++) or weak (+) for association with pre-eclampsia risk. ACE,
angiotensin; AGT, angiotensinogen; AT2R, angiotensin type 2 receptor; FVL, factor V Leiden; IL-10, interleukin-10; LEPR, leptin receptor;
LPL, lipoprotein lipase; MMP-9, matrix metallopeptidase-9; MTHFR, methylene tetrahydrofolate reductase; PAI-1, plasminogen activator
inhibitor-1; SNP, single-nucleotide polymorphism; TGF-β, transforming growth factor beta; VEGF, vascular endothelial growth factor.

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
728 Giannakou et al.

to be involved in the pathogenesis of PE128–130 . Infection 3. Chaiworapongsa T, Chaemsaithong P, Yeo L, Romero R. Pre-eclampsia part 1:
current understanding of its pathophysiology. Nat Rev Nephrol 2014; 10: 466–480.
may play an important role in the pathogenesis of PE, 4. Uzan J, Carbonnel M, Piconne O, Asmar R, Ayoubi JM. Pre-eclampsia:
pathophysiology, diagnosis, and management. Vasc Health Risk Manag 2011;
either through initiation of the condition by increasing 7: 467–474.
the risk of acute uteroplacental atherosclerosis and/or 5. Buurma AJ, Turner RJ, Driessen JH, Mooyaart AL, Schoones JW, Bruijn JA,
Bloemenkamp KWM, Dekkers OM, Baelde HJ. Genetic variants in pre-eclampsia:
its enhancement by magnifying the maternal systemic a meta-analysis. Hum Reprod Update 2013; 19: 289–303.
inflammatory response131 , or through a direct effect on 6. Sibai B, Dekker G, Kupferminc M. Pre-eclampsia. Lancet 2005; 365: 785–799.
7. Askie LM, Duley L, Henderson-Smart DJ, Stewart LA. Antiplatelet agents for
trophoblast cells by destruction or impairment, resulting prevention of pre-eclampsia: a meta-analysis of individual patient data. Lancet
in shallow invasion of maternal spiral arteries132 . 2007; 369: 1791–1798.
8. Duley L, Henderson-Smart DJ, Meher S, King JF. Antiplatelet agents for preventing
pre-eclampsia and its complications. Cochrane Database Syst Rev 2007; CD004659.
9. Bujold E, Roberge S, Lacasse Y, Bureau M, Audibert F, Marcoux S, Forest JC,
Strengths and limitations Giguere Y. Prevention of preeclampsia and intrauterine growth restriction with
aspirin started in early pregnancy: a meta-analysis. Obstet Gynecol 2010; 116:
Both Egger and excess-of-significance tests offer hints of 402–414.
10. Groeneveld E, Lambers MJ, Lambalk CB, Broeze KA, Haapsamo M, De Sutter P,
bias, not definitive proof thereof, while the Egger test is Schoot BC, Schats R, Mol BWJ, Hompes PGA. Preconceptional low-dose aspirin
difficult to interpret when between-study heterogeneity is for the prevention of hypertensive pregnancy complications and preterm delivery
after IVF: a meta-analysis with individual patient data. Hum Reprod 2013; 28:
large. The frequency of meta-analyses with small-study 1480–1488.
asymmetry effects was not high (8%), and this rate is 11. Rifai N, Altman DG, Bossuyt PM. Reporting bias in diagnostic and prognostic
studies: time for action. Clin Chem 2008; 54: 1101–1103.
commensurate with chance. Nevertheless, our estimates 12. Ioannidis JP. Why most published research findings are false. PloS Med 2005; 2:
are likely to be conservative, as a negative test result does e124.
13. Ioannidis JP, Trikalinos TA. An exploratory test for an excess of significant findings.
not exclude the potential for bias. Clin Trials 2007; 4: 245–253.
The majority of the included studies for non-genetic 14. Ioannidis JP. Why most discovered true associations are inflated. Epidemiology
2008; 19: 640–648.
associations were retrospective, which is indicative of 15. Ioannidis JP. Integration of evidence from multiple meta-analyses: a primer on
a higher inherent potential for bias. However, by umbrella reviews, treatment networks and multiple treatments meta-analyses. Can
Med Assoc J 2009; 181: 488–493.
performing a standardized methodological process for 16. Tsagris M, Fragkos KC. Umbrella reviews: evidence synthesis with overviews of
the assessment of the epidemiological credibility of reviews and meta-epidemiologic studies. In Umbrella Reviews, Biondi-Zoccai (ed.).
Springer International Publishing: Switzerland, 2016; 43–54.
the findings using a variety of tests, we succeeded in 17. Tsilidis KK, Papatheodorou SI, Evangelou E, Ioannidis JP. Evaluation of excess
combining all these biases, and provide a complete statistical significance in meta-analyses of 98 biomarker associations with cancer
risk. J Natl Cancer Inst 2012; 104: 1867–1878.
picture of the totality of the evidence as it stands today. 18. Belbasis L, Bellou V, Evangelou E, Ioannidis JP, Tzoulaki I. Environmental risk
The findings of the excess-of-statistical-significance test factors and multiple sclerosis: an umbrella review of systematic reviews and
meta-analyses. Lancet Neurol 2015; 14: 263–273.
for the results of a single meta-analysis, especially one 19. Bellou V, Belbasis L, Tzoulaki I, Evangelou E, Ioannidis JP. Environmental risk
with few studies, should be interpreted with caution factors and Parkinson’s disease: an umbrella review of meta-analyses. Parkinsonism
Relat Disord 2016; 23: 1–9.
because a negative test does not exclude the potential for 20. Belbasis L, Bellou V, Evangelou E. Environmental risk factors and amyotrophic
bias13 . Furthermore, quality assessment of the primary lateral sclerosis: an umbrella review and critical assessment of current evidence from
systematic reviews and meta-analyses of observational studies. Neuroepidemiology
studies was highly heterogeneous, reflecting the lack of 2016; 46: 96–105.
standardized quality assessment methodologies. 21. Lau J, Ioannidis JP, Schmid CH. Quantitative synthesis in systematic reviews. Ann
Intern Med 1997; 127: 820–826.
22. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;
7: 177–188.
Conclusion 23. Riley RD, Higgins JP, Deeks JJ. Interpretation of random effects meta-analyses.
BMJ 2011; 342: d549.
24. Higgins JP, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects
Oocyte donation vs spontaneous conception was sup- meta-analysis. J R Stat Soc Ser A Stat Soc 2009; 172: 137–159.
ported by convincing evidence, and 11 risk factors 25. Higgins J, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med
2002; 21: 1539–1558.
achieved highly suggestive evidence for an association 26. Ioannidis JP, Patsopoulos NA, Evangelou E. Uncertainty in heterogeneity estimates
with PE. PAI-1 4G/5G (recessive model) polymorphism in meta-analyses. BMJ 2007; 335: 914–916.
27. Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J, Carpenter J, Rucker
was supported by strong evidence for a contribution to the G, Harbord RM, Schmid CH, Tetzlaff J, Deeks JJ, Peters J, Macaskill P, Schwarzer
pathogenesis of PE. The use of standardized definitions G, Duval S, Altman DG, Moher D, Higgins JPT. Recommendations for examining
and interpreting funnel plot asymmetry in meta-analyses of randomised controlled
and protocols for exposures, outcomes and statistical trials. BMJ 2011; 343: d4002.
analyses133,134 , adoption of reporting guidelines (e.g. 28. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a
simple, graphical test. BMJ 1997; 315: 629–634.
Strengthening the Reporting of Observational Studies in 29. Ioannidis JP. Clarifications on the application and interpretation of the test for
Epidemiology (STROBE) and STrengthening the REport- excess significance and its extensions. J Math Psychol 2013; 57: 184–187.
30. Lubin JH, Gail MH. On power and sample size for studying features of the relative
ing of Genetic Association Studies (STREGA)) and regis-
odds of disease. Am J Epidemiol 1990; 131: 552–566.
tration of hypothesis-testing observational studies135–138 , 31. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in
may help to improve the evidence in the future, dimin- meta-analyses. BMJ 2003; 327: 557–560.
32. Ioannidis JP, Boffetta P, Little J, O’brien TR, Uitterlinden AG, Vineis P, Balding
ish the risk of biases and improve the reliability of this DJ, Chokkalingam A, Dolan SM, Flanders WD, Higgins JPT, McCarthy MI,
important literature. McDermott DH, Page GP, Rebbeck TR, Seminara D, Khoury MJ. Assessment of
cumulative evidence on genetic associations: interim guidelines. Int J Epidemiol
2008; 37: 120–132.
33. Fan Y, Kang Y, Zhang M. A meta-analysis of copper level and risk of preeclampsia:
evidence from 12 publications. Biosci Rep 2016; 36: e00370.
REFERENCES 34. Song QY, Luo WP, Zhang CX. High serum iron level is associated with an increased
risk of hypertensive disorders during pregnancy: a meta-analysis of observational
1. Duley L. Pre-eclampsia, eclampsia, and hypertension. BMJ Clin Evid 2011 Feb 14; studies. Nutr Res 2015; 35: 1060–1069.
2011. pil:1402. 35. Cohen JM, Beddaoui M, Kramer MS, Platt RW, Basso O, Kahn SR. Maternal
2. Steegers EA, von Dadelszen P, Duvekot JJ, Pijnenborg R. Pre-eclampsia. Lancet antioxidant levels in pregnancy and risk of preeclampsia and small for gestational
2010; 376: 631–644. age birth: A systematic review and meta-analysis. PloS One 2015; 10: e0135192.

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
Genetic and non-genetic risk factors for PE 729

36. Liu HQ, Wang YH, Wang LL, Hao M. Predictive Value of Free β-hCG Multiple 66. Rodger MA, Betancourt MT, Clark P, Lindqvist PG, Dizon-Townson D, Said J,
of the Median for Women with Preeclampsia. Gynecol Obstet Invest 2015; 81: Seligsohn U, Carrier M, Salomon O, Greer IA. The association of factor V leiden
137–147. and prothrombin gene mutation and placenta-mediated pregnancy complications: a
37. Ma Y, Shen X, Zhang D. The Relationship between Serum Zinc Level and systematic review and meta-analysis of prospective cohort studies. PLoS Med 2010;
Preeclampsia: A Meta-Analysis. Nutrients 2015; 7: 7806–7820. 7: e1000292.
38. Allen RE, Rogozinska E, Cleverly K, Aquilina J, Thangaratinam S. Abnormal blood 67. Medica I, Kastrin A, Peterlin B. Genetic polymorphisms in vasoactive genes and
biomarkers in early pregnancy are associated with preeclampsia: a meta-analysis. preeclampsia: a meta-analysis. Eur J Obstet Gynecol Reprod Biol 2007; 131:
Eur J Obstet Gynecol Reprod Biol 2014; 182: 194–201. 115–126.
39. Yang Y, Su X, Xu W, Zhou R. Interleukin-18 and interferon gamma levels in 68. Serrano NC, Dı́az LA, Páez MC, Mesa CM, Cifuentes R, Monterrosa A, Gonzalez
preeclampsia: a systematic review and meta-analysis. Am J Reprod Immunol 2014; A, Smeeth L, Hingorani AD, Casas JP. Angiotensin-converting enzyme I/D
72: 504–514. polymorphism and preeclampsia risk: evidence of small-study bias. PLoS Med
40. Lashley EE, Meuleman T, Claas FH. Beneficial or harmful effect of antipaternal 2006; 3: e520.
human leukocyte antibodies on pregnancy outcome? A systematic review and 69. Lin J, August P. Genetic thrombophilias and preeclampsia: a meta-analysis. Obstet
meta-analysis. Am J Reprod Immunol 2013; 70: 87–103. Gynecol 2005; 105: 182–192.
41. Dai B, Liu T, Zhang B, Zhang X, Wang Z. The polymorphism for endothelial 70. Saccone G, Berghella V, Sarno L, Maruotti GM, Cetin I, Greco L, Khashan AS,
nitric oxide synthase gene, the level of nitric oxide and the risk for pre-eclampsia: a McCarthy F, Martinelli D, Fortunato F, Martinelli P. Celiac disease and obstetric
meta-analysis. Gene 2013; 519: 187–193. complications: a systematic review and metaanalysis. Am J Obstet Gynecol 2016;
42. Kleinrouweler CE, Wiegerinck MM, Ris-Stalpers C, Bossuyt PM, van der Post 214: 225–234.
JA, von Dadelszen P, Mol BWJ, Pajkrt E. Accuracy of circulating placental 71. Zhang JJ, Ma XX, Hao L, Liu LJ, Lv JC, Zhang H. A systematic review and
growth factor, vascular endothelial growth factor, soluble fms-like tyrosine kinase meta-analysis of outcomes of pregnancy in CKD and CKD outcomes in pregnancy.
1 and soluble endoglin in the prediction of pre-eclampsia: a systematic review and Clin J Am Soc Nephrol 2015; 10: 1964–1978.
meta-analysis. BJOG 2012; 119: 778–787. 72. Hu R, Li Y, Zhang Z, Yan W. Antenatal depressive symptoms and the risk
43. Hausvater A, Giannone T, Sandoval YH, Doonan RJ, Antonopoulos CN, Matsoukis of preeclampsia or operative deliveries: a meta-analysis. PloS One 2015; 10:
IL, Petridou ET, Daskalopoulou SS. The association between preeclampsia and e0119018.
arterial stiffness. J Hypertens 2012; 30: 17–33. 73. Qin JZ, Pang LH, Li MJ, Fan XJ, Huang RD, Chen HY. Obstetric complications
44. do Prado AD, Piovesan DM, Staub HL, Horta BL. Association of anticardiolipin in women with polycystic ovary syndrome: a systematic review and meta-analysis.
antibodies with preeclampsia: a systematic review and meta-analysis. Obstet Reprod Biol Endocrinol 2013; 11: 56.
Gynecol 2010; 116: 1433–1443. 74. Zhang S, Ding Z, Liu H, Chen Z, Wu J, Zhang Y, Yu Y. Association between
45. Clark P, Wu O. ABO(H) blood groups and pre-eclampsia. A systematic review and mental stress and gestational hypertension/preeclampsia: a meta-analysis. Obstet
meta-analysis. Thromb Haemost 2008; 100: 469–474. Gynecol Surv 2013; 68: 825–834.
46. Hu H, Ha S, Roth J, Kearney G, Talbott EO, Xu X. Ambient air pollution and 75. Grigoriadis S, VonderPorten EH, Mamisashvili L, Tomlinson G, Dennis CL, Koren
hypertensive disorders of pregnancy: a systematic review and meta-analysis. Atmos G, Steiner M, Mousmanis P, Cheung A, Radford K, Martinovic J, Ross LE. The
Environ 2014; 97: 336–345. impact of maternal depression during pregnancy on perinatal outcomes: a systematic
47. Pedersen M, Stayner L, Slama R, Sørensen M, Figueras F, Nieuwenhuijsen MJ, review and meta-analysis. J Clin Psychiatry 2013; 74: e321–e341.
Raaschoo-Nielsen O, Dadvand P. Ambient air pollution and pregnancy-induced 76. Schoenaker DA, Soedamah-Muthu SS, Mishra GD. The association between dietary
hypertensive disorders. Hypertension 2014; 64: 494–500.
factors and gestational hypertension and pre-eclampsia: a systematic review and
48. Zeng F, Zhu S, Wong MC, Yang Z, Tang J, Li K, Su X. Associations between nitric
meta-analysis of observational studies. BMC Med 2014; 12: 157.
oxide synthase 3 gene polymorphisms and preeclampsia risk: a meta-analysis. Sci
77. Wei SQ, Qi HP, Luo ZC, Fraser WD. Maternal vitamin D status and adverse
Rep 2016: 6: 23 407.
pregnancy outcomes: a systematic review and meta-analysis. J Matern Fetal
49. Zhang G, Zhao J, Yi J, Luan Y, Wang Q. Association Between Gene Polymorphisms
Neonatal Med 2013; 26: 889–899.
on Chromosome 1 and Susceptibility to Pre-Eclampsia: An Updated Meta-Analysis.
78. Huang QT, Chen JH, Zhong M, Hang LL, Wei SS, Yu YH. Chronic Hepatitis B
Med Sci Monit 2016; 22: 2202–2214.
Infection is Associated with Decreased Risk of Preeclampsia: A Meta-Analysis of
50. Li Y, Zhu M, Hu R, Yan W. The effects of gene polymorphisms in angiotensin
Observational Studies. Cell Physiol Biochem 2016; 38: 1860–1868.
II receptors on pregnancy-induced hypertension and preeclampsia: a systematic
79. Sgolastra F, Petrucci A, Severino M, Gatto R, Monaco A. Relationship between
review and meta-analysis. Hypertens Pregnancy 2015; 34: 241–260.
periodontitis and pre-eclampsia: a meta-analysis. PloS One 2013; 8: e71387.
51. Yang W, Zhu Z, Wang J, Ye W, Ding Y. Evaluation of association of maternal
80. Rustveld LO, Kelsey SF, Sharma R. Association between maternal infections and
IL-10 polymorphisms with risk of preeclampsia by a meta-analysis. J Cell Mol Med
preeclampsia: a systematic review of epidemiologic studies. Matern Child Health J
2014; 18: 2466–2477.
2008; 12: 223–242.
52. Wang X, Bai T, Liu S, Pan H, Wang B. Association between thrombophilia gene
81. Xu Y, Ren L, Zhai S, Luo X, Hong T, Liu R, Ran L, Zhang Y. Association Between
polymorphisms and preeclampsia: A meta-analysis. PloS One 2014; 9: e100789.
Isolated Single Umbilical Artery and Perinatal Outcomes: A Meta-Analysis. Med
53. Li X, Luo YL, Zhang QH, Mao C, Wang XW, Liu S, Chen Q. Methylenetetrahy-
Sci Monit 2016; 22: 1451–1459.
drofolate reductase gene C677T, A1298C polymorphisms and pre-eclampsia risk:
82. Wei J, Liu CX, Gong TT, Wu QJ, Wu L. Cigarette smoking during pregnancy and
a meta-analysis. Mol Biol Rep 2014; 41: 5435–5448.
preeclampsia risk: a systematic review and meta-analysis of prospective studies.
54. Gong LL, Liu H, Liu LH. Lack of association between matrix metalloproteinase-9
gene-1562C/T polymorphism and preeclampsia: a meta-analysis. Hypertens Oncotarget 2015; 6: 43 667–43 678.
Pregnancy 2014; 33: 389–394. 83. Masoudian P, Nasr A, de Nanassy J, Fung-Kee-Fung K, Bainbridge SA, El
55. Li X, Shen L, Tan H. Polymorphisms and plasma level of transforming growth Demellawy D. Oocyte donation pregnancies and the risk of preeclampsia or
factor-beta 1 and risk for preeclampsia: a systematic review. PloS One 2014; 9: gestational hypertension: a systematic review and metaanalysis. Am J Obstet
e97230. Gynecol 2016; 214: 328–339.
56. Cheng D, Hao Y, Zhou W, Ma Y. Vascular endothelial growth factor +936C/T, 84. Aune D, Saugstad OD, Henriksen T, Tonstad S. Physical activity and the risk
-634G/C, -2578C/A, and -1154G/A polymorphisms with risk of preeclampsia: a of preeclampsia: a systematic review and meta-analysis. Epidemiology 2014; 25:
meta-analysis. PloS One 2013; 8: e78173. 331–343.
57. Song GG, Kim JH, Lee YH. Associations between vascular endothelial growth factor 85. González-Comadran M, Urresta Avila J, Saavedra Tascón A, Jimenéz R, Solà I,
gene polymorphisms and pre-eclampsia susceptibility: a meta-analysis. Immunol Brassesco M, Carreras R, Checa MÁ. The impact of donor insemination on the
Invest 2013; 42: 749–762. risk of preeclampsia: a systematic review and meta-analysis. Eur J Obstet Gynecol
58. Morgan JA, Bombell S, McGuire W. Association of plasminogen activator Reprod Biol 2014; 182: 160–166.
inhibitor-type 1 (-675 4G/5G) polymorphism with pre-eclampsia: systematic review. 86. Wang Z, Wang P, Liu H, He X, Zhang J, Yan H, Xu D, Wang B. Maternal
PloS One 2013; 8: e56907. adiposity as an independent risk factor for pre-eclampsia: a meta-analysis of
59. Zhao L, Bracken MB, DeWan AT, Chen S. Association between the SERPINE1 prospective cohort studies. Obes Rev 2013; 14: 508–521.
(PAI-1) 4G/5G insertion/deletion promoter polymorphism (rs1799889) and 87. Kasawara KT, Nascimento SL, Costa ML, Surita FG, Silva E, Pinto JL. Exercise
pre-eclampsia: a systematic review and meta-analysis. Mol Hum Reprod 2013; and physical activity in the prevention of pre-eclampsia: systematic review. Acta
19: 136–143. Obstet Gynecol Scand 2012; 91: 1147–1157.
60. Staines-Urias E, Paez MC, Doyle P, Dudbridge F, Serrano NC, Ioannidis JP, Keating 88. Başaran A, Başaran M, Topatan B, Martin JN Jr. Effect of chorionic villus
BJ, Hingorani AD, Casas JP. Genetic association studies in pre-eclampsia: systematic sampling on the occurrence of preeclampsia and gestational hypertension: An
meta-analyses and field synopsis. Int J Epidemiol 2012; 41: 1764–1775. updated systematic review and meta-analysis. J Turk Ger Gynecol Assoc 2016; 17:
61. Lin R, Lei Y, Yuan Z, Ju H, Li D. Angiotensinogen gene M235T and T174M 65–71.
polymorphisms and susceptibility of pre-eclampsia: a meta-analysis. Ann Hum 89. Luo ZC, An N, Xu HR, Larante A, Audibert F, Fraser WD. The effects and
Genet 2012; 76: 377–386. mechanisms of primiparity on the risk of pre-eclampsia: a systematic review.
62. Zhao L, DeWan AT, Bracken MB. Association of maternal AGTR1 polymorphisms Paediatr Perinat Epidemiol 2007; 21: 36–45.
and preeclampsia: a systematic review and meta-analysis. J Matern Fetal Neonatal 90. Pergialiotis V, Prodromidou A, Frountzas M, Perrea DN, Papantoniou N. Maternal
Med 2012; 25: 2676–2680. cardiac troponin levels in pre-eclampsia: a systematic review. J Matern Fetal
63. Zhong WG, Wang Y, Zhu H, Zhao X. Meta-analysis of angiotensin-converting Neonatal Med 2016; 29: 3386–3390.
enzyme I/D polymorphism as a risk factor for preeclampsia in Chinese women. 91. Pergialiotis V, Prodromidou A, Pappa E, Vlachos GD, Perrea DN, Papantoniou N.
Genet Mol Res 2012; 11: 2268–2276. An evaluation of calprotectin as serum marker of preeclampsia: a systematic review
64. Shaik AP, Sultana A, Bammidi VK, Sampathirao K, Jamil K. A meta-analysis of of observational studies. Inflamm Res 2016; 65: 95–102.
eNOS and ACE gene polymorphisms and risk of pre-eclampsia in women. J Obstet 92. Martin A, Krishna I, Badell M, Samuel A. Can the quantity of cell-free fetal DNA
Gynaecol 2011; 31: 603–607. predict preeclampsia: a systematic review. Prenat Diagn 2014; 34: 685–691.
65. Xie C, Yao MZ, Liu JB, Xiong LK. A meta-analysis of tumor necrosis factor-alpha, 93. Poursafa P, Keikha M, Kelishadi R. Systematic review on adverse birth outcomes
interleukin-6, and interleukin-10 in preeclampsia. Cytokine 2011; 56: 550–559. of climate change. J Res Med Sci 2015; 20: 397–402.

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
730 Giannakou et al.

94. Conde-Agudelo A, Rosas-Bermúdez A, Kafury-Goeta AC. Effects of birth spacing 116. Kuśmierska-Urban K, Rytlewski K, Huras H, Wybrańska I. Association of single
on maternal health: a systematic review. Am J Obstet Gynecol 2007; 196: 297–308. nucleotide polymorphism rs7579169 with hypertension disorders during pregnancy
95. Oyebode F, Rastogi A, Berrisford G, Coccia F. Psychotropics in pregnancy: safety and perinatal outcome. Neuro Endocrinol Lett 2015; 36: 282–287.
and other considerations. Pharmacol Ther 2012; 135: 71–77. 117. Guo LF, Wang ZH, Wang YF. Common variant rs7579169 is associated with
96. Smyth A, Oliveira GH, Lahr BD, Bailey KR, Norby SM, Garovic VD. A systematic preeclampsia in Han Chinese women. Genet Mol Res 2015; 15.
review and meta-analysis of pregnancy outcomes in patients with systemic lupus 118. Fenzl V, Flegar-Meštrić Z, Perkov S, Andrišić L, Tatzber F, Žarković N, Duić Ž.
erythematosus and lupus nephritis. Clin J Am Soc Nephrol 2010; 5: 2060–2068. Trace elements and oxidative stress in hypertensive disorders of pregnancy. Arch
97. Caimari F, Valassi E, Garbayo P, Steffensen C, Santos A, Corcoy R, Webb SM. Gynecol Obstet 2013; 287: 19–24.
Cushing’s syndrome and pregnancy outcomes: a systematic review of published 119. Casanueva E, Viteri FE. Iron and oxidative stress in pregnancy. J Nutr 2003; 133:
cases. Endocrine 2017; 55: 555–563. 1700S–1708S.
98. Vrebosch L, Bel S, Vansant G, Guelinckx I, Devlieger R. Maternal and neonatal 120. Rayman MP, Barlis J, Evans RW, Redman CW, King LJ. Abnormal iron parameters
outcome after laparoscopic adjustable gastric banding: a systematic review. Obes in the pregnancy syndrome preeclampsia. Am J Obstet Gynecol 2002; 187:
Surg 2012; 22: 1568–1579. 412–418.
99. Maggard MA, Yermilov I, Li Z, Maglione M, Newberry S, Suttorp M, Hilton 121. Spencer K, Cowans NJ, Nicolaides KH. Low levels of maternal serum PAPP-A in
L, Santry HP, Morton JM, Livingston EH, Shekelle PG. Pregnancy and fertility the first trimester and the risk of pre-eclampsia. Prenat Diagn 2008; 28: 7–10.
following bariatric surgery: a systematic review. JAMA 2008; 300: 2286–2296. 122. Imbasciati E, Gregorini G, Cabiddu G, Gammaro L, Ambroso G, Del Giudice A,
100. Bonzini M, Palmer KT, Coggon D, Carugno M, Cromi A, Ferrario MM. Shift Ravani P. Pregnancy in CKD stages 3 to 5: fetal and maternal outcomes. Am J
work and pregnancy outcomes: a systematic review with meta-analysis of currently Kidney Dis 2007; 49: 753–762.
available epidemiological studies. BJOG 2011; 118: 1429–1437. 123. Jones DC, Hayslett JP. Outcome of pregnancy in women with moderate or severe
101. Adams JW, Watts DH, Phelps BR. A systematic review of the effect of HIV infection renal insufficiency. N Engl J Med 1996; 335: 226–232.
and antiretroviral therapy on the risk of pre-eclampsia. Int J Gynaecol Obstet 2016; 124. Williams D, Davison J. Chronic kidney disease in pregnancy. BMJ 2008; 336:
133: 17–21. 211–215.
102. Kujovich JL. Thrombophilia and pregnancy complications. Am J Obstet Gynecol 125. Tehrani FR, Behboudi-Gandevani S. Polycystic ovary syndrome. In Contemporary
2004; 191: 412–424. Gynaecologic Practice, Darwish A (ed.). InTech Publishers: Rijeka, Croatia, 2015;
103. Stang A. Critical evaluation of the Newcastle–Ottawa scale for the assessment of 79–102.
the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 2010; 25: 126. Troisi R, Potischman N, Johnson CN, Roberts JM, Lykins D, Harger G, Markovic
603–605. N, Siiteri P, Hoover RN. Estrogen and androgen concentrations are not lower in the
104. Sohani ZN, Meyre D, de Souza RJ, Joseph PG, Gandhi M, Dennis BB, Norman umbilical cord serum of pre-eclamptic pregnancies. Cancer Epidemiol Biomarkers
G, Anand SS. Assessing the quality of published genetic association studies in Prev 2003; 12: 1268–1270.
meta-analyses: the quality of genetic studies (Q-Genie) tool. BMC Genet 2015; 16: 127. Choi YK, Kim CK, Lee H, Jeoung D, Ha KS, Kwon YG, Kim KW, Kim YM.
50. Carbon monoxide promotes VEGF expression by increasing HIF-1α protein level
105. Genest DS, Falcao S, Gutkowska J, Lavoie JL. Impact of exercise training via two distinct mechanisms, translational activation and stabilization of HIF-1α
on preeclampsia; potential preventive mechanisms. Hypertension 2012; 60: protein. J Biol Chem 2010; 285: 32 116–32 125.
1104–1109. 128. Levine RJ, Maynard SE, Qian C, Lim KH, England LJ, Yu KF, Schisterman EF,
106. Girardi G, Yarilin D, Thurman JM, Holers VM, Salmon JE. Complement activation Thadhani R, Sachs BP, Epstein FH, Sibai BM. Circulating angiogenic factors and
induces dysregulation of angiogenic factors and causes fetal rejection and growth the risk of preeclampsia. N Engl J Med 2004; 350: 672–683.
restriction. J Exp Med 2006; 203: 2165–2175. 129. Levine RJ, Lam C, Qian C, Yu KF, Maynard SE, Sachs BP, Sibai BM, Epstein FH,
107. van der Hoorn ML, Scherjon SA, Claas FH. Egg donation pregnancy as an Romero R, Thadhani R, Karumanchi SA. Soluble endoglin and other circulating
immunological model for solid organ transplantation. Transpl Immunol 2011; 25: antiangiogenic factors in preeclampsia. N Engl J Med 2006; 355: 992–1005.
89–95. 130. Cudmore M, Ahmad S, Al-Ani B, Fujisawa T, Coxall H, Chudasama K, Devey
108. Chernyshov VP, Tumanova LE, Sudoma IA, Bannikov VI. Th1 and Th2 in LR, Wigmore SJ, Abbas A, Hewett PW, Ahmed A. Negative regulation of soluble
human IVF pregnancy with allogenic fetus. Am J Reprod Immunol 2008; 59: Flt-1 and soluble endoglin release by heme oxygenase-1. Circulation 2007; 115:
352–358. 1789–1797.
109. Lashley LE, van der Hoorn ML, Haasnoot GW, Roelen DL, Claas FH. 131. Von Dadelszen P, Magee LA. Could an infectious trigger explain the differential
Uncomplicated oocyte donation pregnancies are associated with a higher incidence maternal response to the shared placental pathology of preeclampsia and
of human leukocyte antigen alloantibodies. Hum Immunol 2014; 75: 555–560. normotensive intrauterine growth restriction? Acta Obstet Gynecol Scand 2002;
110. van der Hoorn ML, Lashley EE, Bianchi DW, Claas FH, Schonkeren CM, Scherjon 81: 642–648.
SA. Clinical and immunologic aspects of egg donation pregnancies: a systematic 132. Arechavaleta-Velasco F, Ma Y, Zhang J, McGrath CM, Parry S. Adeno-associated
review. Hum Reprod Update 2010; 16: 704–712. virus-2 (AAV-2) causes trophoblast dysfunction, and placental AAV-2 infection is
111. Cnattingius S, Reilly M, Pawitan Y, Lichtenstein P. Maternal and fetal genetic associated with preeclampsia. Am J Pathol 2006; 168: 1951–1959.
factors account for most of familial aggregation of preeclampsia: a population-based 133. Tzoulaki I, Siontis KC, Evangelou E, Ioannidis JP. Bias in associations of emerging
Swedish cohort study. Am J Med Genet A 2004; 130A: 365–371. biomarkers with cardiovascular disease. JAMA 2013; 173: 664–671.
112. Williams PJ, Pipkin FB. The genetics of pre-eclampsia and other hypertensive 134. Tzoulaki I, Siontis KC, Ioannidis JP. Prognostic effect size of cardiovascular
disorders of pregnancy. Best Pract Res Clin Obstet Gynaecol 2011; 25: biomarkers in datasets from observational studies versus randomised trials:
405–417. meta-epidemiology study. BMJ 2011; 343: d6829.
113. Zhao L, Triche EW, Walsh KM, Bracken MB, Saftlas AF, Hoh J, Dewan AT. 135. Von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP;
Genome-wide association study identifies a maternal copy-number deletion in Iniciativa STROBE. The Strengthening the Reporting of Observational Studies in
PSG11 enriched among preeclampsia patients. BMC Pregnancy Childbirth 2012; Epidemiology [STROBE] statement: guidelines for reporting observational studies.
12: 61. Gac Sanit 2008; 22: 144–150.
114. Johnson MP, Brennecke SP, East CE, Göring HH, Kent JW Jr, Dyer TD, Said JM, 136. Little J, Higgins J, Ioannidis J, Moher D, Gagnon F, Von Elm E, Khoury MJ,
Roten LT, Iversen AC, Abraham LJ, Heinonen S, Kajantie E, Kere J, Kivinen K, Cohen B, Davey-Smith G, Grimshaw J, Scheet P. STrengthening the REporting of
Pouta A, Laivuori H, Austgulen R, Blangero J, Moses EK. Genome-wide association Genetic Association studies (STREGA) – an extension of the STROBE statement.
scan identifies a risk locus for preeclampsia on 2q14, near the inhibin, beta B gene. Eur J Clin Invest 2009; 39: 247–266.
PloS One 2012; 7: e33666. 137. Dal-Ré R, Ioannidis JP, Bracken MB, Buffler PA, Chan AW, Franco EL, Vecchia C,
115. Zhao L, Bracken MB, DeWan AT. Genome-wide association study of preeclampsia Weiderpass E. Making prospective registration of observational research a reality.
detects novel maternal single nucleotide polymorphisms and copy-number variants Sci Transl Med 2014; 6: 224cm1.
in subsets of the hyperglycemia and adverse pregnancy outcome (HAPO) study 138. Ioannidis JP. The importance of potential studies that have not existed and
cohort. Ann Hum Genet 2013; 77: 277–287. registration of observational data sets. JAMA 2012; 308: 575–576.


The following supporting information may be found in the online version of this article:
Table S1 Quantitative synthesis and assessment of bias for association of 130 genetic and non-genetic risk
factors with pre-eclampsia
Table S2 Detailed description of 130 risk factors associated with pre-eclampsia, including observed and
expected number of studies with statistically significant findings

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 51: 720–730.
Ultrasound Obstet Gynecol 2018; 51: 720–730
Published online 8 May 2018 in Wiley Online Library ( DOI: 10.1002/uog.18959

Factores de riesgo gen éticos y no gen éticos para la preeclampsia: resumen de revisiones sis-
tem áticas y metaan álisis de estudios observacionales
Objetivo Resumir la evidencia de la literatura sobre factores de riesgo genéticos y no genéticos asociados con la preeclampsia
(PE), evaluar la existencia de sesgos estadı́sticos en los estudios e identificar los factores de riesgo para los cuales existe evidencia
sólida que respalde su asociación con la PE.
Métodos Se realizaron búsquedas en PubMed e ISI Web of Science desde su inicio hasta octubre de 2016, a fin de identificar
revisiones sistemáticas y metaanálisis de estudios observacionales que examinaron las asociaciones entre factores de riesgo
genéticos o no genéticos y la PE. Para cada metaanálisis, el tamaño del efecto resumen se estimó mediante modelos de efectos
aleatorios y de efectos fijos, con un IC del 95% y un intervalo de predicción del 95%. La heterogeneidad entre estudios se
expresó utilizando el estadı́stico I2 y la evidencia de efectos de estudios pequeños (los estudios grandes tuvieron resultados
significativamente más conservadores que los estudios más pequeños) y se estimó la evidencia de sesgo por exceso de significación
(demasiados estudios con resultados estadı́sticamente significativos).

Resultados Se identificaron cincuenta y ocho posibles metaanálisis, que incluı́an 1466 estudios primarios y proporcionaban
datos sobre 130 comparaciones de factores de riesgo asociados con la PE, que cubren una amplia gama de enfermedades
concomitantes, factores genéticos, exposición a agentes ambientales y biomarcadores. Sesenta y cinco (50%) asociaciones
mostraron nominalmente hallazgos estadı́sticamente significativos con una P<0,05, mientras que 16 de ellas (12%) fueron
significativas con una P<10-6. Sesenta y cinco (50%) asociaciones mostraron una heterogeneidad grande o muy grande. Se
encontró evidencia de efectos de estudios pequeños y sesgos por exceso de significación en 10 (8%) y 26 (20%) asociaciones,
respectivamente. El único factor de riesgo no genético con evidencia convincente de una asociación con la PE fue la donación de
ovocitos en comparación con la concepción espontánea, que tuvo una razón de momios resumen de 4,33 (IC 95%, 3,11–6,03),
fue respaldada por 2712 casos con una heterogeneidad pequeña (I2 = 26%) e intervalos de predicción del 95% excluyendo el
valor nulo, y sin indicios de efectos de estudios pequeños (P para la prueba de Egger > 0,10) o sin un exceso de significación
(P > 0,05). Entre los factores de riesgo genéticos estadı́sticamente significativos (P<0.05) para la PE, tan solo el polimorfismo
PAI-1 4G/5G (modelo recesivo) fue respaldado por una evidencia fuerte de contribución a la patogenia de la PE. Once factores
presentaron evidencia que sugerı́a en gran medida una asociación con la PE (nivel de hierro sérico, proteı́na plasmática A
asociada al embarazo, enfermedad renal crónica, sı́ndrome de ovario poliquı́stico, estrés mental, infecciones bacterianas y virales,
tabaquismo, donación de ovocitos frente a la tecnologı́a de reproducción asistida, obesidad frente a peso normal, obesidad severa
frente a peso normal y primiparidad).
Conclusiones Una gran proporción de metaanálisis de los factores de riesgo genéticos y no genéticos para la PE tienen salvedades
que ponen en duda su validez. La donación de ovocitos, en comparación con la concepción espontánea y el polimorfismo 4G/5G
del gen PAI-1 (modelo recesivo) mostró de forma constante la evidencia más robusta de una asociación con el riesgo de PE.

Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd. SYSTEMATIC REVIEW