Professional Documents
Culture Documents
Pregnancy Hypertension
journal homepage: www.elsevier.com/locate/preghy
A R T I C L E I N F O A B S T R A C T
Keywords: Background: Preeclampsia is a multi-system disorder in pregnancy which has no effective treatment. The di-
Preeclampsia agnosis of preeclampsia is based on clinical presentation and routine laboratory tests.
Diagnostics of preeclampsia Objective: This study aimed at identifying serum protein markers for diagnosis of preeclampsia and predicting its
Severe features of preeclampsia severe features.
Pregnancy
Study design: In total, 172 pregnant women were enrolled in this study including 110 subjects with preeclampsia
Diagnostic biomarkers
and 62 normotensive subjects. Eleven serum proteins (VEGF, sFlt-1, sEndoglin, PlGF, sEGFR, prolactin, PTX3,
PAI-1, NGAL, IL-27, COX-2) were assessed using Luminex multiplex immunoassay and ELISA.
Results: The levels of seven proteins (sFlt-1, VEGF, sEndoglin, sEGFR, PlGF, NGAL, COX-2) correlated with
preeclampsia, and 4 proteins (VEGF, sEndoglin, PlGF, sEGFR) were identified as independent factors associated
with preeclampsia. The levels of three proteins (sEndoglin, PTX3, sFlt-1) correlated with severe features of
preeclampsia, and three variables (serum creatinine, platelet count and sEndoglin) were identified as in-
dependent factors in predicting severe features of preeclampsia.
Conclusions: A combination of serum protein markers (VEGF, sEndoglin, PlGF, sEGFR) and clinical variables
(serum creatinine, platelet count and sEndoglin) could be used as analytical tool in diagnosis of preeclampsia
and its severe features, respectively. Serum sEGFR, a novel biomarker in preeclampsia, may be involved in the
pathogenesis of preeclampsia.
1. Introduction balancing the potential benefit of the further fetal development against
the maternal risk of progression of the disease [7,8]. Therefore, it is
Preeclampsia refers to new onset of hypertension and proteinuria or important to accurately assess both the maternal and fetal outcomes of
end-organ dysfunction after 20 weeks of gestation in previously nor- preeclampsia in pregnant women.
motensive women. As a multisystem disorder, it is one of the leading The pathogenesis of preeclampsia is associated with placental hy-
causes of maternal morbidity, fetal morbidity and mortality in both poxia and/or ischemia and excessive oxidative stress. Release of soluble
developing and developed countries, affecting 2–8% of the pregnant factors from the ischemic placenta into the maternal blood stream plays
women worldwide [1–4]. It may cause cerebrovascular, cardiac, he- a central role in altering the endothelial function, which is the most
patic, hematologic, and renal complications in mothers, and premature prominent feature of the disease [9]. Furthermore, the coagulation
birth, growth retardation and death in fetuses. Preeclampsia can be dysfunction, inflammation, and immunological responses are also in-
devastating and life threatening [5,6]. The only definitive treatment is volved in the pathogenesis of preeclampsia. Our previous studies [10]
to deliver the fetus and placenta. The decision to deliver involves demonstrated that many proteins are dysregulated in the preeclampsia
⁎
Corresponding author.
E-mail address: lylchina78@yahoo.com (Y. Li).
https://doi.org/10.1016/j.preghy.2018.01.009
Received 27 November 2017; Received in revised form 23 January 2018; Accepted 24 January 2018
,QWHUQDWLRQDO6RFLHW\IRUWKH6WXG\RI+\SHUWHQVLRQLQ3UHJQDQF\3XEOLVKHGE\(OVHYLHU%9$OOULJKWVUHVHUYHG
L. Cui et al. 3UHJQDQF\+\SHUWHQVLRQ[[[[[[[[[[²[[[
(COX-2, VEGF, sEndoglin, IL-27, PTX3, PIGF, PAI-1, NGAL, human level (1000 pg/mL). ELISA for sFlt-1 was performed according to the
prolactin, sEGFR) or ELISA (sFlt-1) using the bead-based multiplex manufacturer’s instructions (R&D Systems, Minneapolis, MN, USA).
immuno-assays. All 11 paired primary antibodies (capture antibody and Briefly, high-binding ELISA plates (R&D Systems, were coated with
detection antibody) and standards were purchased from R&D 100 μL/well of 2.0 μg/mL monoclonal anti-human sFlt-1. Then, 100 μL
(Minneapolis, MN, USA). serum sample (diluted at 1:5) and standards in half-log serial dilutions
per well were added and incubated for two hours at room temperature.
2.3. Bead-based multiplex immuno-assay on Luminex platform After washing, 100 μL/well of 400 ng/mL biotinylated polyclonal goat-
antihuman sFlt-1 detection antibody were added and incubated for two
2.3.1. Coupling of capture antibody to beads hours at room temperature. After washing, 100 μL of Streptavidin-HRP
Magplex microspheres (Luminex Corporation, Austin, TX, USA) diluted 1:200 were added to the wells. After 20 min of incubation, the
were used in the development of the multiplex immuno-assays which plates were washed, and 100 μL of Substrate Solution (R&D Systems,
included 10 analytes (COX-2, VEGF, sEndoglin, IL-27, PTX3, PlGF, PAI- Minneapolis, MN, USA) were added to each well and incubated on a
1, NGAL, human prolactin, and sEGFR). Microsphere activation and plate shaker for 20 min in the dark. The reaction was terminated by
conjugation reactions were accomplished according to the manu- adding 50 μL of Stop Solution (R&D Systems, Minneapolis, USA). The
facturer’s instructions. Briefly, 1x106 microspheres were washed and re- optical densities were read at 450 nm using the microplate reader set
suspended in 100 mM monobasic sodium phosphate (pH 6.2). This was (BIO-TEK instruments, INC, Winooski, VT, USA). The standard curve
followed by the sequential addition of 50 mg/mL Sulfo-NHS (Thermo was created using computer software capable of generating a four
Scientific, Waltham, MA, USA) and 50 mg/mL of 1-ethyl-2-(dimethy- parameter logistic (4-PL) curve-fit. Values were obtained by compar-
laminopropyl) carbodiimide (EDC) (Sigma, St Louis, MO, USA). This ison with the standard curves after subtracting the blanks.
mixture was then incubated for 20 min at room temperature in the dark
to yield the long-lived intermediate sulfo-NHS ester. The activated 2.5. Statistical analysis
microspheres were washed, and 2.5 µg of each capture antibody were
added to the above sulfo-NHS ester-derivatized beads. Beads were in- Chi-square tests were used to compare the categorical variables of
cubated on a rotator for two hours at room temperature wrapped in foil clinical characteristics and t test was used for the comparison of con-
paper. After incubation, the microspheres were washed twice with tinuous variables. The concentrations of 11 serum protein markers were
1xPBS containing 1% BSA and 0.05% sodium azide and stored at 4 °C converted in logarithmical transformation (log2) due to asymmetric
protected from the light for the future use. distribution. In order to determine the association of preeclampsia with
severe features, univariate and multivariate logistic regression analyses
2.3.2. Multiplex immuno-assays were performed. On the basis of the multivariate logistic regression
Coupling efficiency for each capture antibody coupled to the beads analysis, the significant markers were included in the risk score calcu-
was confirmed, and optimal concentration of detection antibody was lation, as described previously [18–20]. Receiver Operating Char-
also determined by a seven point serial 2-fold dilution of phycoerythrin- acteristic (ROC) curve was constructed for each significant serum pro-
labeled anti-mouse or rabbit IgG detection antibody (0.078–5 μg/mL). tein marker and combined with clinical variables. The logistic
Standard curves were constructed with seven serial dilutions of each regression and ROC curve construction were performed using SPSS 17.0
standard, and cross reaction was evaluated by mixing the ten standards software (SPSS, Chicago, IL). The area under the curve (AUC) was
(analytes) and capture antibodies coupled to different beads together. calculated using the MedCalc statistical software (Mariakerke, Bel-
Based on different dilution factors, the ten analytes were divided into gium). A p value < 0.05 was considered to be statistically significant.
two groups: one was a 7-plex panel including COX-2, VEGF, sEndoglin, Risk score was calculated using the following formula: Risk
IL-27, PTX3, PlGF, and PAI-1 (1:5 dilution for patient serum samples), Score = Probability × 100, where Probability = eZ/(eZ + 1), “e” is the
and another one was a 3-plex panel including NGAL, human prolactin base of natural logarithm and “Z’’ is the result of the logistic regression.
and sEGFR (1:50 dilution for patient serum samples). Z was calculated as follows: Z = B0 + B1 X1 + B2 X2 + …+ Bp Xp,
Multiplex immuno-assays (including both 7-plex panel and 3-plex where “B0” is the constant of regression coefficients, ‘‘X1…Xp’’ are the
panel) were performed as follows. Serum samples were made by 1:5 independent variables, and ‘‘B1…Bp’’ are their corresponding coeffi-
and 1:50 dilutions for each sample. 50 μL of serial-dilution standards (7- cients [18,20].
plex and 3-plex) and pre-diluted patient samples (1:5 and 1:50 dilu-
tions) were dispensed into the wells on a COSTAR™ 96-well round 3. Results
bottom-plate (Corning, NY, USA). Then, 50 μL of capture antibody
coupling beads mixture (7-plex and 3-plex, respectively) (2.5 μg/mL) 3.1. Clinical characteristics of women with preeclampsia
were added into each well and incubated at room temperature for 2 h.
Following incubation, the microspheres were washed, and then 50 μL of The clinical characteristics of the 110 preeclampsia subjects and 62
biotinylated detection antibodies were added into each well. The mix- control subjects are summarized in Table 1. There were no significant
tures were incubated for 1 h at room temperature. After washing, the differences in the age of patients (29.7 ± 5.9 years vs
beads were incubated with 100 μL of streptavidin-phycoerythrin (4 μg/ 29.7 ± 5.3 years; p = 0.967), gestational age at sampling
mL, Roche Molecular Systems, Inc. Branchburg, NJ, USA) for 30 min at (35.7 ± 2.9 weeks vs 35.0 ± 3.5 weeks; p = 0.18), percentage of
room temperature in the dark. After the incubation, the beads were nulliparity (69.4% vs 70.9%; p = 0.83) at triage between control group
washed again and re-suspended in 100 μL of PBS/1% BSA before and preeclampsia group. The preeclampsia group had a significantly
reading on Luminex-200™ bioanalyzer (Luminex Inc., Austin, TX, USA) higher body mass index (BMI) (29.4 ± 4.1 kg/m2 vs 26.1 ± 2.6 kg/
with a minimum of 100 beads reads per well. Median fluorescence in- m2; p < 0.0001), systolic blood pressure (160.6 ± 17.7 mmHg vs
tensity (MFI) values for each analyte were converted into absolute 121.8 ± 9.7 mmHg; p < 0.0001) and diastolic blood pressure
concentration values by a serial dilution of standards with known (108.3 ± 14.1 mmHg vs 78.8 ± 7.7 mmHg; p < 0.0001) than con-
concentrations for the analyte. trol group. Regarding the liver function, the preeclampisa group had a
significantly higher serum AST (56.3 ± 179.7 U/L vs 12.3 ± 7.5 U/L;
2.4. Enzyme-linked immunosorbent assay (ELISA) p = 0.0017) and lower albumin (27.2 ± 5.2 g/l vs 31.9 ± 3.8 g/l;
p < 0.0001) than the control group. Although the average serum ALT
The expression level of sFlt-1 was detected by ELISA instead of level (31.0 ± 70.7 U/L) in preeclampsia group was higher than that in
multiplex immuno-assays due to unsatisfactory minimal detectable control group (18.1 ± 5.3U/L), no statistical difference was reached
L. Cui et al. 3UHJQDQF\+\SHUWHQVLRQ[[[[[[[[[[²[[[
(p = 0.071). The preeclampsia group demonstrated a significant in- sFlt-1 were significantly higher than those in preeclampsia patients. In
crease in serum creatinine compared to the control group contrast, the levels of COX-2, PlGF, NGAL, and sEGFR were significantly
(61.1 ± 16.7 μmol/L vs 46.0 ± 10.6 μmol/L; p < 0.0001). There lower than those in preeclampsia patients. Furthermore, sEndoglin and
were no significant differences in platelet count (198.3 ± 65.1 × 109/ sFlt-1 were significantly higher in PS group than in PNS group. Even
L vs 211.4 ± 48.9 × 109/L; p = 0.148) and amniotic fluid index though no difference was found in the level of PTX3 between normo-
(11.1 ± 4.9 cm vs 10.8 ± 4.9 cm; p = 0.708), as determined by ul- tensive control group and preeclampsia group, PTX3 was significantly
trasound at triage between preeclampsia and control group. The higher in PS group than in PNS group. No differences in the serum
average gestational age at delivery in the preeclampsia group was sig- levels of PAI-1 and prolactin were found between control group and
nificant lower than that in control group (35.7 ± 2.9 weeks preeclampsia group and between PNS group and PS group.
vs36.9 ± 2.2 weeks; p = 0.0037). The placenta weight
(481.5 ± 99.0 g vs 520.0 ± 87.4 g; p = 0.015) and newborn birth 3.4. Correlation of clinical factors and biomarker levels with preeclampsia
weight (2454.3 ± 891.3 g vs 2699.5 ± 656.9 g; p = 0.048) in the
preeclampsia group were significantly lower than that in control group. To identify the proteins and the clinical variables that are associated
with preeclampsia, univariate logistic regression analysis was initially
3.2. Preeclampsia with severe features performed (Table 4). Among these clinical variables (shown in Table 1),
BMI, AST, serum albumin and serum creatinine were found to be sig-
According to the presence of one or more of the severe features, nificantly correlated with preeclampsia (p < 0.05, Table 4). In addi-
including symptoms of central nervous system dysfunction, hepatic tion, 7 of the 11 analyzed proteins (COX-2, VEGF, sEndoglin, PIGF,
abnormality, severe blood pressure elevation, thrombocytopenia, renal NGAL, sEGFR and sFlt-1) were found to be significantly correlated with
abnormality, or pulmonary edema, the preeclampsia patients were preeclampsia (p < 0.05, Table 4). Stepwise multivariate logistic re-
stratified into two subgroups: group with severe features (PS group) and gression analysis was performed for the above seven proteins and four
group without severe features (PNS group). Out of 110 preeclampsia clinical variables. After adjusting for the above clinical variables, 4 of
patients enrolled in this study, 25 patients (22.7%) had the severe the 7 proteins (VEGF, sEndoglin, PlGF and sEGFR) remained statisti-
features. Baseline characteristics of women with and without severe cally significant (p < 0.05, Table 5) as independent diagnostic markers
features are shown in Table 2. There was no significant difference in the of preeclampsia. The above four variables (Table 5) were included in
age of the patients (29.6 ± 5.6 year vs 29.7 ± 5.3 year; p = 0.972), the risk score calculation, using the formula: Z = 7.651 + 3.058
BMI (28.6 ± 3.3 kg/m2 vs 29.6 ± 4.4 kg/m2; p = 0.335) and per- (VEGF) + 3.730 (sEndoglin) + (−3.217) (PlGF) + (−4.023) (sEGFR).
centage of nulliparity (60% vs 74.1%; p = 0.172) between PS group For a given patient, the risk score can be calculated based on the values
and PNS group. However, the patients in PS group presented at initial of 4 variables using the above formula. The score ranged from 0 to 100.
triage significantly earlier (33.7 ± 3.4 weeks vs 35.4 ± 3.5 weeks; To assess the ability of these 4 proteins in identifying the pre-
p = 0.029) than the patients in PNS group. The patients in PS group eclampsia, the receiver operating characteristic (ROC) curve analysis
demonstrated significantly higher SBP (172.7 ± 17.3 mmHg vs was performed (Fig. 1) and the areas under the curves (AUC) were
157.1 ± 16.3 mmHg; p < 0.001) and DBP (115.4 ± 13.0 mmHg vs calculated. The AUCs for VEGF, PlGF, sEGFR, sEndoglin and the four-
106.2 ± 13.7 mmHg; p = 0.003) than the patients in PNS group. As to protein risk core were 0.634, 0.711, 0.724, 0.910 and 0.988, respec-
liver function test, ALT was significant higher in PS group than in PNS tively (supplemental Table 1). These results suggest that the risk score
group (82.7 ± 133.7 U/L vs 14.9 ± 7.5 U/L; p < 0.021), while no for the diagnosis of preeclampsia was the best among all the predictors
significant difference in AST level between PS and PNS groups was (p < 0.05). In addition, the ROC curve of sEGFR protein expression, a
found (166.8 ± 351.4 U/L vs 21.9 ± 6.9 U/L; p = 0.055). Both PS new valuable biomarker diagnosis factor of preeclampsia, revealed cut-
and PNS groups had similar serum albumin level (26.1 ± 4.9 g/l vs off points with the highest sum of specificity and sensitivity at
27.5 ± 5.3 g/l; p = 0.244). Patients in PS group had significantly 28.49 ng/mL, leading to specificity of 87% and sensitivity of 55%.
higher creatinine (75.2 ± 20.0 µmol/L vs 57.0 ± 13.1 µmol/L;
p < 0.001) than that in PNS group. As of the blood count, PS group 3.5. Correlation of biomarker levels and clinical variables with severe
had significantly lower platelet count (131.3 ± 59.3 × 109/L vs features of preeclampsia
217.5 ± 53.1 × 109/L; p < 0.001) than that in PNS group. No sig-
nificant difference was found on AFI between PS and PNS groups To identify the proteins and the clinical variables that are associated
(9.5 ± 4.1 cm vs 11.7 ± 5.1 cm; p = 0.002). Patients in PS group with severe features, univariate logistic regression analysis was per-
delivered the babies significantly earlier (34.2 ± 3.1 weeks vs formed (Table 6). Among the clinical variables in Table 2, SBP, DBP,
36.2 ± 2.7 weeks; p < 0.001) and had a higher percentage of preterm AST, ALT, serum creatinine and platelet count at triage were found to
delivery (88.0% vs 48.2%; p < 0.001) than those in PNS group. The be correlated with severe features of preeclampsia. Among the 11
newborns’ birth weight was significantly smaller in PS group protein markers, only sEndoglin, PTX3 and sFlt-1 were found to be
(1,838.2 ± 566.4 g vs 2,699.1 ± 849.8 g; p < 0.001) than that in correlated with the severe features of preeclampsia. Therefore, the six
PNS group, and 28% of newborns in PS group were SGA compared to clinical variables and the above 3 proteins were included in a stepwise
10.6% in PNS group (p = 0.034). The placenta weight in PS group was multivariate logistic regression analysis. Only serum creatinine, platelet
significantly smaller than in PNS group (422.3 ± 82.7 g vs count and sEndoglin still stood as independent factors of PNS (Table 7).
517.6 ± 76.0 g; p < 0.001). The above three variables (Table 7) were included in the risk score
calculation using the following formula: Z = −16.46 + 0.049 (serum
3.3. Serum biomarker levels in preeclampsia women with or without severe creatinine) + (−0.03) (platelet count) + (1.294) (sEndoglin). For a
features given patient, the risk score can be calculated based on the values of 3
variables using the above formula. The risk score ranged from 0 to 100.
As shown in Table 3, eleven protein biomarkers were screened in To further determine the optimal model to predict severe features of
110 preeclampsia and 62 normotensive pregnant women using the preeclampsia, the ROC curve analysis was applied to either single or
multiplex immunoassays and ELISA. Among the preeclampsia cases, 85 risk score model (Fig. 2). The AUCs of serum creatinine, sEndoglin, and
women had no severe features (PNS) and 25 had severe features (PS). platelet count and combination of all three markers were 0.780, 0.753,
The levels of 7 serum proteins (out of 11) were found to be significantly 0.845, and 0.920, respectively (supplemental Table 2). These results
different between preeclampsia group and normotensive group suggest that the risk score model was the best among all predictors
(p < 0.05). Among those proteins, the levels of VEGF, sEndoglin, and (p < 0.05).
L. Cui et al. 3UHJQDQF\+\SHUWHQVLRQ[[[[[[[[[[²[[[
Table 3
Serum protein levels in preeclampsia and preeclampsia with severe features.
VEGF (pg/mL) 871.10 ± 562.22 1169.76 ± 679.42 0.001* 1189.04 ± 684.04 1115.26 ± 677.60 0.940
sEndoglin (ng/mL) 3.97 ± 1.68 9.87 ± 5.44 < 0.001* 9.07 ± 5.29 12.37 ± 5.04 < 0.001#
sFlt-1 (ng/mL) 3.97 ± 4.68 22.79 ± 2.98 < 0.001* 18.38 ± 22.13 38.59 ± 44.13 0.005#
COX-2 (pg/mL) 859.00 ± 348.47 633.29 ± 309.47 < 0.001* 649.43 ± 286.30 585.08 ± 386.71 0.097
PlGF(pg/mL) 232.27 ± 156.30 142.57 ± 100.70 < 0.001* 147.17 ± 106.78 131.33 ± 82.62 0.925
NGAL (ng/mL) 335.44 ± 211.58 291.65 ± 311.63 0.007* 257.03 ± 227.82 407.00 ± 491.81 0.093
sEGFR(ng/mL) 34.62 ± 7.44 28.70 ± 7.38 < 0.001* 28.86 ± 7.23 28.24 ± 7.97 0.971
IL-27 (ng/mL) 16.24 ± 6.05 15.35 ± 6.64 0.403 15.422 ± 6.74 15.17 ± 6.40 0.750
PAI-1 (ng/mL) 12.78 ± 7.03 11.67 ± 5.78 0.615 11.79 ± 5.64 11.24 ± 6.43 0.534
PTX3(pg/mL) 691.63 ± 600.03 730.98 ± 835.08 0.229 628.64 ± 682.48 1065.18 ± 1181.75 0.003#
Prolactin (ng/mL) 78.89 ± 36.91 71.23 ± 31.82 0.189 71.05 ± 33.00 73.37 ± 28.92 0.406
PNS: preeclampsia without features of severe disease; PS: preeclampsia with severe features; * significant difference between control and preeclampsia groups; # significant difference
between PNS and PS.
Table 4
Factors associated with preeclampsia based on univariate analysis.
Lower Upper
BMI: Body Mass Index; AST: Aspartate Transaminase; B: regression coefficients; OR: Odds
Ratio; CI: Confidence Interval.
Table 5
Independent factors associated with preeclampsia based on the multivariate analysis.
Variables B p value OR 95% CI Fig. 1. Receiver operating characteristic (ROC) curves of protein markers for diagnosis of
preeclampsia. Four-protein combination, VEGF, sEndoglin, PIGF, and sEGFR were in-
Lower Upper
cluded in the analysis. Four-protein model performed better compared to individual
biomarkers (p < 0.05).
VEGF 3.058 < 0.001 21.284 4.020 112.672
sEndoglin 3.730 < 0.001 41.680 6.708 258.970
PlGF −3.217 < 0.001 0.040 0.007 0.221
Table 6
sEGFR −4.023 0.011 0.018 0.001 0.396
Factors associated with severe features of preeclampsia based on univariate analysis.
The constant (B0) in the model was 7.651; B: regression coefficients; OR: Odds ratio; CI:
Variables B p value OR 95% CI
Confidence Interval.
Lower Upper
4. Discussion
SBP 0.57 < 0.001 1.059 1.027 1.092
DBP 0.050 0.006 1.051 1.015 1.089
Traditionally, the serum marker detection has been based on the AST 0.100 0.002 1.106 1.037 1.178
enzyme-linked immunosorbent assay (ELISA), measuring only one ALT 0.078 0.008 1.081 1.021 1.145
serum marker at one sample one time. The Luminex® xMAP® Creatinine 0.067 < 0.001 1.069 1.034 1.106
Technology (Luminex Corp, Austin, TX, USA), a bead-based technology, Platelet −0.031 < 0.001 0.969 0.956 0.983
sEndoglin 1.273 0.001 3.573 1.728 7.390
enables a large numbers of protein markers to be analyzed simulta-
PTX3 0.545 0.005 1.725 1.182 2.517
neously in a single test, using very small samples. This antibody-based sFlt-1 0.480 0.008 1.616 1.134 2.303
microarray has high sensitivity and cost-efficient and has been adopted
widely in clinical diagnosis. In this study, we developed a multiplex B, regression coefficients; OR: Odds Ratio; CI: Confidence Interval; SBP: Systolic Blood
immuno-assay panel using Luminex® xMAP® Technology to evaluate Pressure; DBP: Diastolic Blood Pressure; AST: Aspartate Transaminase; ALT: Alanine
Transaminase.
serum protein levels in the preeclampsia women with or without severe
features. We identified several protein markers including human free
vascular endothelial growth factor (VEGF), human pentraxin 3 (PTX3), factor receptor 1(sFlt-1) and so on, that have potential to be used in
plasminogen activator inhibitor-1 (PAI-1), soluble human endoglin clinical settings to help with diagnostics of preeclampsia and to predict
(sEndoglin), human placenta growth factor (PlGF), soluble epidermal severe features of preeclampsia in order for clinicians to make a better
growth factor receptor (sEGFR), soluble vascular endothelial growth decision on delivery. In addition, we also evaluated the role of sEGFR as
L. Cui et al. 3UHJQDQF\+\SHUWHQVLRQ[[[[[[[[[[²[[[
L. Cui et al. 3UHJQDQF\+\SHUWHQVLRQ[[[[[[[[[[²[[[