Biochemical and Biophysical Research Communications 485 (2017) 119e125

Contents lists available at ScienceDirect

Biochemical and Biophysical Research Communications

journal homepage: www.elsevier.com/locate/ybbrc

Biomarker identication and pathway analysis of preeclampsia based

on serum metabolomics
Tingting Chen a, b, 1, Ping He a, 1, Yong Tan c, Dongying Xu a, *
a
Department of Gynaecology and Obstetrics, Tongren Hospital, Shanghai Jiaotong University School of Medicine, 1111XianXia Road, Shanghai 200336,
China
b
School of Medicine, Suzhou University, Suzhou 215021, China
c
Institute of Basic Research in Clinical Medicine, China Academy of Chinese Medical Science, Beijing 100700, China

a r t i c l e i n f o a b s t r a c t

Article history: Preeclampsia presents serious risk of both maternal and fetal morbidity and mortality. Biomarkers for the
Received 4 February 2017 detection of preeclampsia are critical for risk assessment and targeted intervention. The goal of this study
Accepted 6 February 2017 is to screen potential biomarkers for the diagnosis of preeclampsia and to illuminate the pathogenesis of
Available online 7 February 2017
preeclampsia development based on the differential expression network. Two groups of subjects,
including healthy pregnant women, subjects with preeclampsia, were recruited for this study. The
Keywords:
metabolic proles of all of the subjects' serum were obtained by liquid chromatography quadruple time-
Biomarker
of-ight mass spectrometry. Correlation between metabolites was analyzed by bioinformatics technique.
Metabolomics
Preeclampsia
Results showed that the PC(14:0/00), proline betaine and proline were potential sensitive and specic
biomarkers for preeclampsia diagnosis and prognosis. Perturbation of corresponding biological path-
ways, such as iNOS signaling, nitric oxide signaling in the cardiovascular system, mitochondrial
dysfunction were responsible for the pathogenesis of preeclampsia. This study indicated that the
metabolic proling had a good clinical signicance in the diagnosis of preeclampsia as well as in the
study of its pathogenesis.
2017 Elsevier Inc. All rights reserved.

1. Introduction low predictive ability. The exact pathogenesis of PE is still unde-

Preeclampsia (PE) is a complex syndrome affecting about 3% of
pregnancies in approximately 2% of women worldwide, which is a
major cause of 50,000e60,000 maternal deaths annually [1e3]. It is
characterized by hypertension and large amounts of urine protein
after 20 weeks of gestation [4]. Apart from perinatal problems, PE is
also associated with substantial health problems later in life. Both
women who suffered from PE and their children have a substan-
tially elevated risk of chronic hypertension, cardiovascular disease
and diabetes mellitus type-2 [5]. It is important to diagnose women
at high risk of preeclampsia, and then it might enable potential
prophylactic treatment to reduce or avoid the onset of symptoms.
As of today, PE lack simple and reliable diagnostic indicators, and
their pathogenesis remain unclear. The diagnosis of PE has been
currently focused on blood pressure and urinary protein levels with
* Corresponding author.
E-mail address: xudongyingtongren@163.com (D. Xu).
1
Both authors contributed equally to this work.
ned though much of the pathophysiology has been explained
[6e10]. Generally, PE starts in early pregnancy with poorly devel-
oped placental vascularization, giving rise to placental oxidative
stress and imbalanced interaction between maternal and fetal cells.
Then, inappropriate and exaggerated maternal responses to the
placental stress are established, involving endothelial activation
and systemic inammation [11,12]. In a word, methods for its early
diagnosis or prediction have not been adequately developed.
Increasing evidences demonstrated that PE is caused by in-
teractions between complex pathophysiological mechanisms, in-
dividual genes and environmental factors [13,14]. There is currently
no complete view on the pathophysiology of PE. With the devel-
opment of high-throughput omics detection and analysis tech-
nique, metabolomics is widely used in discovery of biomarkers and
better understanding of pathogenesis. Metabolomics is integrated
system-based method which study the dynamic multi-parametric
metabolic response of living systems to pathophysiological stimuli
or genetic modications [15]. By detecting and semi-quantitation of
low molecular weight metabolites present in cells, tissues or body
uids, using high throughput analysis platforms such as proton

http://dx.doi.org/10.1016/j.bbrc.2017.02.032
0006-291X/ 2017 Elsevier Inc. All rights reserved.
120 T. Chen et al. / Biochemical and Biophysical Research Communications 485 (2017) 119e125
nuclear magnetic resonance (1HNMR) spectroscopy or Mass
Spectrometry (MS), metabolomics offers fresh insight into disease
[16,17]. Recent interest has mounted in the metabolomics approach
to predict and characterize preeclampsia [18e23]. So far, no single
biomarker has been deemed suitable for clinical application at
present, and it is not yet clear how molecules are mutual inuence
in the process of PE [24].
In the present study, we applied a metabolomics-based liquid
chromatography quadruple time-of-ight mass spectrometry (LC-
Q-TOF-MS) technique to characterize the metabolic proles of
preeclampsia, subsequently employed bimolecular network tech-
nology to identify corresponding pathogenesis. We aimed to deci-
pher the reliable and sensitive biomarkers for preeclampsia and
corresponding pathogenesis.
2. Materials and methods
2.1. Study population
Women with pregnancies complicated by preeclampsia were
included as cases. Healthy pregnant women with no previous his-
tory of pregnancies with preeclampsia or FGR were included as
controls. Preeclampsia was dened as persistent hypertension
(systolic/diastolic blood pressure 140/90 mmHg) plus proteinuria
(
0.3 g/24 h or
1 by dipstick) developing after 20 weeks of
gestation [25]. The participants were interviewed about their
health and pregnancy. All participants were weighed, and BMI was
calculated in kg/m2. The characteristics of study participants was
shown in Table 1. The women were asked to fast for 1 h before their
visit to avoid immediate effects of meals on the metabolic proles.
All women gave written informed consent at study entry. The study
was approved by the Regional Committee for Medical Research
Ethics in Tongren Hospital, School of Medicine, Shanghai Jiaotong
University.
2.2. Chemicals and reagents
LC-MS grade acetonitrile and methanol were purchased from
Honeywell Burdick and Jackson (Muskegon, MI, USA). Mass Spec-
troscopic grade formic acid was purchased from Fluka (Buchs,
Switzerland). Ultra-pure water was obtained in a Milli-Q system
from Millipore (Bedford, MA, USA). Reference standards were
purchased from Sigma-Aldrich (St. Louis, MO, USA).
2.3. Serum sample collection
Venous blood was drawn into non-heparinized tubes, left to clot
for 30 min at room temperature, and centrifuged at 3500 rpm for
10 min. A serum sample (0.6 mL) was separated and stored
at 80  C, thawed once and 100 mL of the serum was added to
300 mL methanol (with Internal standard 2-chlorophenylalanine,
20 mg/ml), and then the mixture was shaken vigorously for 30 s.
After centrifugation at 12 000 rpm for 10 min at 4  C, 200 mL of
supernatant and the mixture was stored at 80  C until analysis.
2.4. LC-Q-TOF/MS
Serum metabolite proling was performed on an Agilent 1290
Innity Liquid Chromatography System quipped with an Agilent
6538 Accurate-Mass QTOF/MS (Agilent Technologies, Santa Clara,
CA, USA). Separation of all samples was performed on an Eclipse
plus C18 column (1.8 mm, 3.6 mm  100 mm, Agilent) with a col-
umn temperature set of 45  C. The ow rate was 0.25 mL/min and
the mobile phase consisted of ultrapure water with 0.1% formic acid
and acetonitrile. The following gradient program was used: 2%
acetonitrile for 0e1.5 min; 2e100% acetonitrile for 1.5e13 min;
washed with 100% acetonitrile for 13e16 min; re-equilibration step
for 5 min. The sample injection volume was 5 mL.
Mass detection was operated in both positive and negative ion
modes with the following setting: drying gas (N2) ow rate, 8 L/
min; gas temperature, 330  C; pressure of nebulizer gas, 35 psig;
Vcap, 4000 V; fragmentor, 160 V; skimmer, 65 V; scan range, m/z
80e1000. All analyses were acquired using the instrument mass
spray to ensure accuracy and reproducibility. Leucine enkephalin
was used as the instrument reference mass (m/z 556.2771 in ES
and 554.2615 in ES) at a concentration of 100 ng/ml for all ana-
lyses. The MS/MS analysis was acquired in targeted MS/MS mode
with collision energy from 10 EV to 40 EV.
2.5. Sequence analysis
Serum samples run was randomization and the pooled quality
control (QC) sample was analyzed every 6 serum samples. If dif-
ferences were detected between QC control results (either previous
results or within the analysis), the analysis was stopped, the ioni-
zation source and the column were cleaned, and the mass spec-
trometer recalibrated. Additional samples were not analyzed until
the QC analyses were satisfactory.
2.6. Data processing and statistical analysis
The LCeMS raw data were exported by Agilent Mass Hunter
Qualitative Analysis Software (Agilent Technologies, Palo Alto, CA,
USA). The lter parameters were used as follows: restrict retention
time, 0e21 min; restrict m/z, 80e1000 amu; peak relative height,

1.5%; Mass tolerance, 0.05 Da; Retention time windows, 0.05 min;
A peak table was then created including the information of the
retention time, m/z and ion intensity of all ions. The data of each
sample were normalized to the total area to correct for the MS
response shift between injections due to any possible intra- and
inter-day variations. Then the data was exported into SIMCA-P13.0
software (Umetrics AB, Umea, Sweden) for multivariate analysis,
that is, PCA, PLS-DA and OPLS-DA. Principle component analysis
(PCA) and orthogonal partial least squares-discriminant analysis
(OPLS-DA) were carried out to discriminate the metabolic patterns
between PE and healthy controls after mean centering and unit
variance scaling. To guard against model over-tting, the default 7-
fold cross-validation was applied. Those variables with VIP >1.0
were selected as relevant for group discrimination [26]. Then the
Student's t-test was applied to all metabolites. A classical one stage
method of false discovery rate (FDR) was performed to adjust the p-
value [27]. Differentiating metabolites with VIP >1 and p < 0.05
(adjusted p-value) were selected as potential biomarkers. Those
markers were identied with the aid of available reference stan-
dards in our lab and the web-based resources such as the Human
Metabolome Database (http://www.hmdb.ca/), Metlin(https://
metlin.scripps.edu/). The MS/MS analysis was acquired in tar-
geted MS/MS mode with collision energy 10, 20, 40 eV to identify
potential biomarkers. Diagnostic model was constructed with the
markers, using linear discrimination analysis method with the
MetaboAnalyst 3.0 (http://www.metaboanalyst.ca/), a compre-
hensive, Web-based tool designed for processing, analyzing, and
interpreting metabolomic data. The classication performance
(specicity and sensitivity) was assessed by AUC value of the
receiver operating characteristic (ROC) curves. The corresponding
up- and down-regulated trend metabolites varied between PE and
healthy controls were used for subsequent metabolic pathway
analysis.
 T. Chen et al. / Biochemical and Biophysical Research Communications 485 (2017) 119e125
2.7. IPA analysis
Molecular networks for the candidate metabolites were built
and analysis of canonical pathways and bio-functions were con-
ducted by using the Ingenuity Pathway Analysis system (IPA, In-
genuity Systems, http://www.ingenuity.com), to gain insight into
the typical metabolic alterations associated with the biomarkers
and the mechanisms relevant to PE.
3. Results
3.1. Identication of the differential metabolites
Typical total ion current (TIC) chromatograms of serum samples
were obtained from both healthy and Preeclampsia. PCA scores plot
revealed a trend of separation between 20 PE patients and 20
controls (Fig. 1A). An OPLS-DA model was obtained with one pre-
dictive component and two orthogonal components. The scores
plot demonstrated a separation between PE patients and 20 con-
trols (Fig. 2A). A 999-time permutation test was performed to
validate the OPLS-DA model. The result showed that the intercept
for the Q2 to the y-axis is below zero (Fig. 2B), which indicated
models is not over-tting. 49 metabolites were selected with VIP >1
and p < 0.05 (adjusted p-value) (Table 2). Those differentiating
metabolites included 1 lysophosphatidylcholines (LysoPE (0:0/
16:0)), 8 glycerophosphocholines (2PC, 2PEs, 2PI, 2PS), and 1
acylcarnitine. Among that 21 metabolites were up regulated in PE
patients with fold change >2 and 4metbolites were down regulated
in PE patients with fold change <0.5. ROC curve analysis indicated
that PC(14:0/00), Proline betaine and Proline, with an area under
the curve value of 0.935,0.9275,0.9225 respectively, were potential
sensitive and specic biomarkers for preeclampsia diagnosis and
prognosis (Fig. 3).
3.2. Metabolic pathway analysis with IPA
In order to further understand the correlation between the
metabolites, bioinformatics analyses were performed using the IPA
software, leading to the identication of biological association
networks. As shown in Fig. 4, the network was built based on the
differentiated metabolites between the PE and healthy control. The
established network functions in PE include lipid metabolism,
Fig. 1. Principle component analysis (PCA) scores plot discriminating the metabolic proles in serum of PE and those in healthy controls dataset. A, POS(7 components model:
R2X 0.424, Q2 0.216). B, NEG (7 components model: R2X 0.427, Q2 0.198). (:) QC group; (-) Healthy Control group; (C) PE group.
121
nucleic acid metabolism, carbohydrate metabolism and cardiovas-
cular system development. These metabolites were correlated with
nine canonical pathways, such as glycine degradation (creatine
biosynthesis), iNOS signaling, nNOS Signaling in skeletal muscle
cells, mitochondrial dysfunction, nitric oxide signaling in the car-
diovascular system, citrulline-nitric oxide cycle, as well as eNOS
signaling.
4. Discussion
Preeclampsia (PE) is a severe pregnancy complication, which is a
leading cause of 50,000e60,000 maternal deaths annually. Until
now the diagnosis of PE has been the focus on blood pressure and
urinary protein levels with low predictive ability. Therefore, it is
need to identify novel biomarkers to improve diagnosis and deepen
the understanding of the pathogenesis of PE. In order to establish
highly efcient and specic biomarkers for screening of PE patients,
in this study we applied LC-MS based metabolomics to screen the
metabolites with high predict efciency of PE from maternal serum.
Biomarker Analysis was assessed by AUC value of the receiver
operating characteristic (ROC) curves with the MetaboAnalyst 3.0.
The results suggested that Phosphatidylcholine- PC(14:0/00), Pro-
line betaine and Proline were potential sensitive and specic bio-
markers for preeclampsia diagnosis and prognosis. And then we
rstly build and analyzed molecular networks for the candidate
metabolites to gain insight into the typical metabolic alterations
associated with the biomarkers and the mechanisms relevant to PE.
According to our pathway analysis with IPA, glycine degradation
(Creatine Biosynthesis), iNOS Signaling, nNOS Signaling in Skeletal
Muscle Cells, Mitochondrial Dysfunction, Nitric Oxide Signaling in
the Cardiovascular System, Citrulline-Nitric Oxide Cycle, as well as
eNOS Signaling, were very interesting concerning the pathophysi-
ology of PE.
PC is a major component of biological membranes in higher
eukaryotes, and it can be secreted by specialized tissues for
important extracellular tasks. In PE patients, PC and cholesterol
content of placental tissue have been observed higher than control
by enzymatic detection [28,29], and there were differences in lipid
mass spectral ion intensity between healthy and PE pregnancies
[30,31]. In our study, PC(14:0/00) was higher in PE patients, which
was consistent with previous studies. Other PCs were almost higher
in PE patients with fold change >2, which were synthesized with
122 T. Chen et al. / Biochemical and Biophysical Research Communications 485 (2017) 119e125

Fig. 2. Orthogonal partial least squares-discriminant analysis (OPLS-DA) scores plot and permutation test for the model discriminating serum samples from PE patients and healthy
controls. (A1) POS-OPLS-DA scores plot. The model parameters were: R2Xcum 0.264, R2Ycum 0.977, Q2 0.819. (A2) NEG-OPLS-DA scores plot. The model parameters were:
R2Xcum 0.384, R2Ycum 0.904, Q2 0.856. (B1) A 999-times permutation test for the corresponding model. The Y-axis intercepts were:R2 (0, 0.651), Q2 (0, 0.253). (B2) A 999-
times permutation test for the corresponding model. The Y-axis intercepts were:R2 (0, 0.622), Q2 (0, 0.257).

Table 1 caused by the lipotoxicity [33]. Inappropriate treatment of fatty

Characteristics of study participants. acids in PE patients may lead to vascular dysfunction, increased
Data PE C p-value insulin resistance, and defects of long chain fatty acid mobilization
n (samples) 20 20 e
[34,35]. Indeed, lipid storage droplets in placenta have been
Age (years) 30.5 (23e38) 33.2 (29e38) >0.05 observed in a rodent model of PE using histochemical staining [36].
GA at sampling (week) 25.2 (21.7e27.9) 26 (18.6e27.4) >0.05 In addition, elevated lipid and low-density lipoprotein levels in
GA at onset (week) 21.2 (20.4e26.9) N/A e maternal serum may induce endothelial dysfunction secondary to
BP sys. (mmHg) 163 (143e174) 120 (100e191) 0.000
oxidative stress [37]. According to our research, it is important to
BP dia. (mmHg) 104 (96e111) 78 (60e96) 0.000
Proteinuria 3 (1e4) 0.1 (0e1) 0.000 detect the changes of lipids to nd the risk of PE. It may be that we
Maternal weight (kg) 64 (62.0e70.0) 66.5 (63.0e76.0) >0.05 can reduce the content of these lipids in the clinical treatment of
Maternal BMI (kg/m2) 24.7 (22.3e27.5) 22.8 (20.7e23.5) 0.01 patients with preeclampsia, to improve the patient's symptoms.
Values are given as median (min-max). PE: Women with preeclampsia. C: Pregnant Although several etiologies have been connected with the
controls; GA: Gestational age. BP: Blood pressure. Dia: Diastolic. Sys: Systolic. N/A: development of PE, the most common, endothelial dysfunction
Not applicable. Statistical p-values computed by Kruskal-Wallis independent sam- resulting from an abnormal poorly perfused placenta [38,39], is
ples test.
thought to be the primary cause in the pathogenesis of the disease
[40,41]. Poor perfusion is thought to result in the secretion of agents
fatty acids and glycerol. In healthy pregnancy, the mother stores into the maternal circulation, leading to activation of the vascular
fatty acids in adipose tissue, as a result of the effect of pregnancy endothelium, and ultimately, to its dysfunction [39]. However, the
hormones in mid to late gestation [32]. In PE patients the excessive attack of reactive oxygen species (ROS), and oxidative stress, is
lipid synthesis, reducing the fatty acid in adipose tissue storage thought to be the most important contributing factor to the path-
capacity, may lead to abnormal fatty accumulation in liver and ogenesis of PE. Until now, there were no uniform conclusion for the
other tissues as well as a series of pathological consequences pathogenesis of PE. And it also need to search for a molecular
 T. Chen et al. / Biochemical and Biophysical Research Communications 485 (2017) 119e125 123

Table 2
Identied differential metabolites in the serum of healthy and PE group.

No RT (min) mass Name Vip P PE/C

1 8.81 317.2933 Phytosphingosine 1.6017 8.98E-05 8.81

2 0.81 143.0945 Proline betaine 2.2163 4.67E-05 0.81
3 4.82 121.0891 Phenylethylamine 1.6921 1.36E-03 4.82
4 0.83 129.0425 Pyroglutamic acid 1.1305 6.00E-03 0.83
5 9.72 301.2984 Sphinganine 1.5260 3.64E-04 9.72
6 0.78 113.0587 Creatinine 1.0230 5.71E-03 0.78
7 10.95 183.0661 Phosphocholine 1.3829 3.18E-03 10.95
8 0.77 161.105 Carnitine 2.1350 1.21E-05 0.77
9 4.05 117.0581 Indole 1.1843 4.12E-02 4.05
10 1.14 129.0789 pipecolic acid 1.5774 2.07E-03 1.14
11 9.45 250.1207 Ubiquinone-1 1.6439 1.36E-02 9.45
12 1.15 152.0336 Xanthine 1.0878 3.39E-02 1.15
13 1.14 168.0285 Uric acid 2.1422 2.43E-05 1.14
14 0.78 131.0692 Creatine 1.6950 8.13E-04 0.78
15 1.14 136.0385 Hypoxanthine 1.8084 1.73E-04 1.14
16 0.78 147.0531 Glutamate 1.3899 1.35E-02 0.78
17 0.83 117.0789 Valine 1.9897 2.47E-05 0.83
18 0.83 164.0472 Phenylpyruvic acid 1.3435 2.03E-02 0.83
19 1.15 192.0271 Citric acid 1.2781 2.18E-02 1.15
20 0.79 115.0632 Proline 1.9428 1.34E-04 0.79
21 0.69 155.0695 Histidine 1.4490 2.61E-02 0.69
22 0.91 131.0945 Leucine 1.4357 9.77E-04 0.91
23 2.26 165.079 Phenylalanine 2.3769 2.14E-07 2.26
24 0.70 174.1117 Arginine 1.1610 7.80E-03 0.70
25 8.08 978.7179 PI(22:0/22:0) 1.3299 1.58E-03 8.08
26 12.37 620.3118 PI(20:4 (5Z,8Z,11Z,14Z)/0:0) 1.3698 3.14E-02 12.37
27 11.47 437.2907 PE (P-16:0/0:0) 1.2193 5.19E-03 11.47
28 11.44 479.3017 PE (18:1 (9Z)/0:0) 1.2782 6.85E-03 11.44
29 10.05 467.3012 PC(14:0/0:0) 1.7064 1.73E-03 10.05
30 10.90 453.2856 PC(13:0/0:0) 1.4753 2.75E-03 10.90
31 11.03 399.335 Palmitoylcarnitine 1.4361 1.08E-03 11.03
32 13.38 255.2564 Palmitic amide 1.1939 5.15E-03 13.38
33 13.28 354.2759 MG (0:0/18:2 (9Z,12Z)/0:0) 1.2456 1.47E-02 13.28
34 2.26 330.2686 MG (0:0/16:0/0:0) 1.5130 1.88E-03 2.26
35 10.72 423.3349 Linoleyl carnitine 1.5997 8.75E-04 10.72
36 5.80 675.6653 Cer(d18:1/26:1 (17Z)) 1.6762 7.39E-04 5.80
37 9.09 287.2828 C17 Sphinganine 1.2677 1.56E-02 9.09
38 8.74 273.2672 C16 Sphinganine 2.2330 8.28E-06 8.74
39 14.37 282.256 Oleic Acid 2.07 7.83E-03 0.69
40 0.81 168.0293 Uric acid 1.41 2.79E-04 0.64
41 0.81 244.07 Uridine 1.07 1.40E-02 0.69
42 11.07 256.2401 Palmitic acid 1.82 7.94E-03 0.34
43 0.81 192.0278 Citric acid 1.63 1.18E-04 0.53
44 0.80 150.0534 Ribose 2.47 6.88E-04 0.39
45 1.20 181.074 Tyrosine 1.81 8.34E-03 0.80
46 0.80 180.064 D-Glucose 1.38 1.40E-03 1.06
47 10.88 453.2851 LysoPE (0:0/16:0) 1.92 1.13E-04 0.60
48 11.09 521.2722 PS(18:2 (9Z,12Z)/0:0) 2.63 2.62E-04 0.40
49 12.29 549.3034 PS(20:2 (11Z,14Z)/0:0) 1.35 3.54E-03 0.48

Fig. 3. ROC curve analysis of potential serum biomarker levels for differentiating the PE group from the control group.

predictor of the PE. According to our pathway analysis with IPA, Signaling in Skeletal Muscle Cells, Mitochondrial Dysfunction, Ni-
these potential makers found in our study were related to Glycine tric Oxide Signaling in the Cardiovascular System, Citrulline-Nitric
Degradation (Creatine Biosynthesis), iNOS Signaling, nNOS Oxide Cycle, as well as eNOS Signaling, as shown in Fig. 4, All the
124 T. Chen et al. / Biochemical and Biophysical Research Communications 485 (2017) 119e125

Fig. 4. Biological network, canonical pathways and functions related to the identied metabolites. In the network, molecules are represented as nodes, and the biological rela-
tionship between two nodes is represented as a line. Red symbols represent up-regulated metabolites; blue symbols represent down-regulated metabolites; while the green
symbols represent canonical pathways that are related to the identied specic metabolites. Solid lines between molecules show a direct physical relationship between molecules,
while dotted lines show indirect functional relationships. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

metabolites are very interesting concerning the pathophysiology of References

PE, and there are many pathways associated with the pathogenesis
of PE. It conrm the underlying mechanisms include endothelial
dysfunction, metabolic abnormalities and increased oxidative
stress [42]. These data indicate that PE is a multifaceted disorder
associated with the metabolic syndrome. In clinical treatment, it is
promising to regulate these metabolic pathways to improve the
symptoms of the patient.
Conicts of interest
The authors declare no conict of interest.
Acknowledgments
We thank all the participation in Tongren hospital for their
willingness to complete the data of this study.
Abbreviations
PE, Preeclampsia; RRLC, high-performance liquid chromatog-
raphy; LC/MS, high-pressure liquid chromatography combined
mass spectrometry; LC-Q-TOF-MS, liquid chromatography
quadruple time-of-ight mass spectrometry; EIC, extracted ion
chromatograms; ESI, electrospray ionization; PCA, principal com-
ponents analysis; PLS-DA, partial-least-squares discriminate anal-
ysis; OPLS, orthogonal partial least square; ROC, receiver operating
characteristic curve; IPA, ingenuity pathway analysis; ANOVA,
analysis of variance; PC, phosphocholine.
[1] J. Uzan, M. Carbonnel, O. Piconne, R. Asmar, J.M. Ayoubi, Pre-eclampsia:
pathophysiology, diagnosis, and management, Vasc. health risk Manag. 7
(2011) 467e474.
[2] B.W. Mol, C.T. Roberts, S. Thangaratinam, L.A. Magee, C.J. de Groot,
G.J. Hofmeyr, Pre-eclampsia, Lancet 387 (2016) 999e1011.
[3] WHO Recommendations for Prevention and Treatment of Pre-eclampsia and
Eclampsia, World Health Organization., Geneva, 2011.
[4] E. Eiland, C. Nzerue, M. Faulkner, Preeclampsia 2012, J. pregnancy 2012 (2012)
586578.
[5] L. Bellamy, J.P. Casas, A.D. Hingorani, D.J. Williams, Pre-eclampsia and risk of
cardiovascular disease and cancer in later life: systematic review and meta-
analysis, BMJ 335 (2007) 974.
[6] A. Kumar, N. Begum, S. Prasad, S. Agarwal, S. Sharma, IL-10, TNF-alpha & IFN-
gamma: potential early biomarkers for preeclampsia, Cell. Immunol. 283
(2013) 70e74.
[7] I. Caniggia, J. Winter, S.J. Lye, M. Post, Oxygen and placental development
during the rst trimester: implications for the pathophysiology of pre-
eclampsia, Placenta 21 (Suppl A) (2000) S25eS30.
[8] K. Verdonk, W. Visser, A.H. Van Den Meiracker, A.H. Danser, The renin-
angiotensin-aldosterone system in pre-eclampsia: the delicate balance be-
tween good and bad, Clin. Sci. Lond. 126 (2014) 537e544.
[9] A.C. Staff, S.J. Benton, P. von Dadelszen, J.M. Roberts, R.N. Taylor, R.W. Powers,
D.S. Charnock-Jones, C.W. Redman, Redening preeclampsia using placenta-
derived biomarkers, Hypertension 61 (2013) 932e942.
[10] S.R. Hansson, A. Naav, L. Erlandsson, Oxidative stress in preeclampsia and the
role of free fetal hemoglobin, Front. Physiol. 5 (2014) 516.
[11] P. Gathiram, J. Moodley, Pre-eclampsia: its pathogenesis and pathophysiolgy,
Cardiovasc. J. Afr. 27 (2016) 71e78.
[12] A.M. Borzychowski, I.L. Sargent, C.W. Redman, Inammation and pre-
eclampsia, Semin. Fetal Neonatal Med. 11 (2006) 309e316.
[13] A. Var, Y. Yildirim, E. Onur, N.K. Kuscu, B.S. Uyanik, K. Goktalay, Y. Guvenc,
Endothelial dysfunction in preeclampsia, Semin. Reprod. Endocrinol. 16
(1998) 69e74.
[14] C.W. Redman, I.L. Sargent, Latest advances in understanding preeclampsia,
Science 308 (2005) 1592e1594.
[15] J.K. Nicholson, J.C. Lindon, E. Holmes, 'Metabonomics': understanding the
metabolic responses of living systems to pathophysiological stimuli via
multivariate statistical analysis of biological NMR spectroscopic data,
 T. Chen et al. / Biochemical and Biophysical Research Communications 485 (2017) 119e125
xenobiotica, fate foreign Compd. Biol. Syst. 29 (1999) 1181e1189.
[16] V. Fanos, L. Atzori, K. Makarenko, G.B. Melis, E. Ferrazzi, Metabolomics
application in maternal-fetal medicine, Biomed Res. Int. 2013 (2013) 720514.
[17] J.C. Lindon, J.K. Nicholson, E. Holmes, J.R. Everett, Metabonomics: metabolic
processes studied by NMR spectroscopy of biouids, Magnetic Reson. Educ. J.
12 (2000) 289e320.
[18] M. Austdal, L.H. Tangers, R.B. Skrstad, K. Salvesen, R. Austgulen, A.C. Iversen,
T.F. Bathen, First trimester urine and serum metabolomics for prediction of
preeclampsia and gestational hypertension: a prospective screening study,
Int. J. Mol. Sci. 16 (2015) 21520e21538.
[19] R.O. Bahado-Singh, R. Akolekar, R. Mandal, E. Dong, J. Xia, M. Kruger,
D.S. Wishart, K. Nicolaides, First-trimester metabolomic detection of late-
onset preeclampsia, Am. J. Obstetrics Gynecol. 208 (2013) 1e7.
[20] R.O. Bahado-Singh, R. Akolekar, R. Mandal, E. Dong, J. Xia, M. Kruger,
D.S. Wishart, K. Nicolaides, Metabolomics and rst-trimester prediction of
early-onset preeclampsia, J. maternal-fetal neonatal Med. ofcial J. Eur. Assoc.
Perinat. Med. 25 (2012) 1840 the Federation of Asia and Oceania Perinatal
Societies, the International Society of Perinatal Obstetricians.
[21] S.O. Diaz, A.S. Barros, B.J. Goodfellow, I.F. Duarte, E. Galhano, C. Pita,
M.D.C. Almeida, I.M. Carreira, A.M. Gil, Second trimester maternal urine for
the diagnosis of trisomy 21 and prediction of poor pregnancy outcomes,
J. proteome Res. 12 (2013) 2946e2957.
[22] J. Pinto, E. Maciel, T.S. Melo, M.R.M. Domingues, E. Galhano, C. Pita,
M.D.C. Almeida, I.M. Carreira, A.M. Gil, Maternal plasma phospholipids are
altered in trisomy 21 cases and prior to preeclampsia and preterm outcomes,
Rapid Commun. Mass Spectrom. 28 (2014) 1635e1638.
[23] R.O. Bahado-Singh, A. Syngelaki, R. Mandal, S.F. Graham, R. Akolekar, B. Han,
T.C. Bjondahl, E. Dong, S. Bauer, Z. Alpay-Savasan, Metabolomic determination
of pathogenesis of late-onset preeclampsia, J. maternal-fetal neonatal Med.
ofcial J. Eur. Assoc. Perinat. Med. (2016) 1 the Federation of Asia and Oceania
Perinatal Societies, the International Society of Perinatal Obstetricians.
[24] Y. Zhong, M. Tuuli, A.O. Odibo, First-trimester assessment of placenta function
and the prediction of preeclampsia and intrauterine growth restriction, Pre-
nat. Diagn. 30 (2010) 293e308.
[25] M.A. Brown, M.D. Lindheimer, S.M. De, A.A. Van, J.M. Moutquin, The classi-
cation and diagnosis of the hypertensive disorders of pregnancy: statement
from the International Society for the Study of Hypertension in Pregnancy
(ISSHP), Hypertension in Pregnancy 20 (2001). IXeXIV.
[26] J. Jansson, B. Willing, M. Lucio, A. Fekete, J. Dicksved, J. Halfvarson, C. Tysk,
P. Schmitt-Kopplin, Metabolomics reveals metabolic biomarkers of Crohn's
disease, PLoS One 4 (2009) e6386.
[27] D. Zhu, A.O. Hero, Z.S. Qin, A. Swaroop, High throughput screening of co-
expressed gene pairs with controlled false discovery rate (FDR) and mini-
mum acceptable strength (MAS), J. Comput. Biol. A J. Comput. Mol. Cell Biol. 12
(2005) 1029e1045.
[28] X. Huang, A. Jain, M. Baumann, M. Ko rner, D. Surbek, P. Btikofer, C. Albrecht,
Increased placental phospholipid levels in pre-eclamptic pregnancies, Int. J.
125
Mol. Sci. 14 (2012) 3487e3499.
[29] A.C. Staff, T. Ranheim, J. Khoury, T. Henriksen, Increased content of phos-
pholipids, cholesterol, and lipid peroxides in decidua basalis in women with
preeclampsia, Am. J. Obstet. Gynecol. 180 (1999) 587e592.
[30] H.A. Korkes, N. Sass, A.F. Moron, N.O. Ca ^mara, T. Bonetti, A.S. Cerdeira, S.I. Da,
O.L. De, Lipidomic assessment of plasma and placenta of women with early-
onset preeclampsia, PLoS One 9 (2014) e110747.
[31] S.H. Brown, S.R. Eather, D.J. Freeman, B.J. Meyer, T.W. Mitchell, A lipidomic
analysis of placenta in preeclampsia: evidence for lipid storage, PLoS One 11
(2016).
[32] S.S. Huda, N.A. Sattar, D.J. Freeman, Lipoprotein metabolism and vascular
complications in pregnancy, Clin. Lipidol. 4 (2009) 91e102.
[33] E. Jarvie, S. Haugueldemouzon, S.M. Nelson, N. Sattar, P.M. Catalano,
D.J. Freeman, Lipotoxicity in obese pregnancy and its potential role in adverse
pregnancy outcome and obesity in the offspring, Clin. Sci. 119 (2010)
123e129.
[34] F.M. Stewart, D.J. Freeman, J.E. Ramsay, I.A. Greer, M. Caslake, W.R. Ferrell,
Longitudinal assessment of maternal endothelial function and markers of
inammation and placental function throughout pregnancy in lean and obese
mothers, J. Clin. Endocrinol. Metab. 92 (2007) 969e975.
[35] J.E. Ramsay, W.R. Ferrell, L. Crawford, A.M. Wallace, I.A. Greer, N. Sattar,
Divergent metabolic and vascular phenotypes in pre-eclampsia and intra-
uterine growth restriction: relevance of adiposity, J. Hypertens. 22 (2004)
2177e2183.
[36] M. Steinlin, P. Baxter, E. Boltshauser, V. Brankovic, C.E. Catsman-Berrevoets,
E. Bertini, C. Kennedy, F. Mancini, A. Nemeth, L. Scho ls, Effect of high-fat diet
on liver and placenta fatty inltration in early onset preeclampsia-like mouse
model, Chin. Med. J. 125 (2012) 3532e3538.
[37] D.A. Enquobahrie, M.A. Williams, C.L. Butler, I.O. Frederick, R.S. Miller,
D.A. Luthy, Maternal plasma lipid concentrations in early pregnancy and risk
of preeclampsia *, Am. J. Hypertens. 17 (2004) 574e581.
[38] C.A. Hubel, Oxidative stress in the pathogenesis of preeclampsia, Proc. Soc.
Exp. Biol. Med. Soc. Exp. Biol. Med. 222 (1999) 222e235.
[39] L.C. Kenny, W.B. Dunn, D.I. Ellis, J. Myers, P.N. Baker, D.B. Kell, G. Consortium,
Novel biomarkers for pre-eclampsia detected using metabolomics and ma-
chine learning, Metabolomics 1 (2005) 227e234.
[40] S. Kumru, A. Godekmerdan, S. Kutlu, Z. Ozcan, Correlation of maternal serum
high-sensitive C-reactive protein levels with biochemical and clinical pa-
rameters in preeclampsia, Eur. J. Obstetrics Gynecol. Reproductive Biol. 124
(2006) 164e167.
[41] C.A. Hubel, J.M. Roberts, R.N. Taylor, T.J. Musci, G.M. Rogers, M.K. Mclaughlin,
Lipid peroxidation in pregnancy: new perspectives on preeclampsia, Am. J.
Obstet. Gynecol. 161 (1989) 1025e1034.
[42] I. Craici, S. Wagner, V.D. Garovic, Preeclampsia and future cardiovascular risk:
formal risk factor or failed stress test? Ther. Adv. Cardiovasc. Dis. 2 (2008)
249e259.