Profiling Cytokines According to the Presence or Absence of Small-for-Gestational Age Using Amniotic Fluid Retrieved During Cesarean Section From Pregnant Women With Early-Onset Preeclampsia

Abstract

Background

Preeclampsia (PE) is a hypertensive disorder and a major cause of maternal and fetal mortality. We aimed to investigate the molecular properties of early-onset PE, which requires delivery before 34 weeks’ gestation by analyzing the molecular cytokine profile of amniotic fluid obtained during cesarean section from pregnant women with early-onset PE, based on the presence or absence of small-for-gestational age (SGA).

Methods

This study included 73 pregnant women with early-onset PE among which 21 women had SGA infants, whose birth weight was less than the 10th percentile of the gestational age-specific birth weight. Amniocentesis was performed after exposing the amniotic sac during cesarean delivery. Twenty-five cases of appropriate-for-gestational age (AGA) infants, who had birth weights between the 25th and 75th percentile of the gestational age-specific birth weight, were arbitrarily selected as a control group. Potential protein biomarkers were analyzed using the Olink® Explore 384 Inflammation panel with a Proximity Extension Assay technique. The biological implications of the differentially expressed proteins (DEPs) were assessed using the web-based tool Database for Annotation, Visualization, and Integrated Discovery 2021. Enrichment analysis of hub genes was performed using the Metascape Database.

Results

Although the mean birth weight was significantly lower in the SGA group than that in the AGA group (945.2 ± 302.3 vs. 1,590.0 ± 393.2, respectively; P < 0.001), no difference was observed in the mean gestational age at delivery (P > 0.05). Sixteen DEPs (EPO, WFIKKN2, CLSTN2, CSF3, COL9A1, SCG3, CCL23, SKAP2, CCL20, GZMB, TIMP3, FIS1, IL17C, PON3, VEGFA, and CXCL8) were found to be upregulated in the SGA group compared with the AGA group. Six hub genes (CCL20, CSF3, EPO, VEGFA, IL17C, and GZMB), which are mainly involved in cytokine-cytokine receptor interactions, were overexpressed in the SGA group.

Conclusion

We found six upregulated hub genes with potential as novel biomarkers for early-onset PE with SGA. Although further investigation is warranted to validate our results, our findings may contribute to a better understanding of the pathogenesis of early-onset PE with SGA.

Graphical Abstract

INTRODUCTION

Preeclampsia (PE) is a hypertensive disorder, with or without proteinuria, that generally occurs after 20 weeks of pregnancy1 and is a major cause of maternal and fetal mortality.2 PE also affects maternal organs such as the liver, heart, lungs, and kidneys3 and is one of the primary causes of fetal growth restriction (FGR) and preterm birth.4 The incidence of PE, which currently affects 3–8% of all pregnancies, is increasing worldwide.5,6 Abnormal placental implantation is a known cause of PE.7 During placental implantation in a normal pregnancy, trophoblasts of anchoring villi penetrate deep into the inner third layer of the myometrium and the spiral arteries of the maternal body. These structural changes are associated with the functional modification of lowering spiral arterial resistance.8 However, abnormal placental implantation leads to increased maternal arterial resistance, placental ischemia, and fetal complications.9 Additionally, it manifests as clinical features such as maternal symptoms, hematologic, and fetal monitoring abnormalities, leading directly to delivery. In particular, early-onset PE, which requires delivery before 34 weeks’ gestation, is acknowledged to have placental dysfunction10 and is associated with worse maternal and fetal outcomes compared with late-onset PE.11 Recent studies have suggested that alterations in molecular factors play a pathogenic role in PE12,13,14; however the molecular mechanisms underlying the pathophysiology of early-onset PE remain unknown. Therefore, elucidation of the molecular features of early-onset PE may contribute to a better understanding of its diagnosis, prevention, and treatment.

Small-for-gestational age (SGA) infants are defined as those with a birth weight below the 10th percentile for a given gestational age. They are categorized into two main groups: constitutionally normal infants who are genuinely SGA and infants with FGR. FGR is a condition wherein a fetus fails to reach its biologically determined growth potential. FGR has multifactorial etiologies including impaired placental development and function, PE and maternal hypertension, poor maternal cardiovascular adaptation to pregnancy, and multiple gestation.15 Most infants with early-onset FGR are SGA, conversely, infants with late-onset FGR have birth weights within the normal range.16 According to previous cohort studies,17,18 SGA is associated with increased risk of neonatal mortality as well as major morbidities including bronchopulmonary dysplasia, necrotizing enterocolitis, and neurodevelopmental impairments. Therefore, further research is warranted on early-onset PE, which leads to FGR, as well as preterm birth.

Proteomics has contributed to the identification of clinical biomarkers as well as therapeutic targets. In terms of PE, proteomics-based research has documented differentially expressed proteins (DEPs) from serum, plasma, maternal urine, and placenta.19 Protein biomarkers have also been identified in the amniotic fluid (AF) obtained through diagnostic amniocentesis in the second trimester of pregnancy.20,21,22 Additionally, proteomics-based studies used maternal blood or placenta collected at delivery to assess FGR.23,24 AF provides information about fetal genotype and reflects the adaptations of maternal-fetal physiology during pregnancy.25 However, no study has investigated AF obtained during delivery from pregnant women with PE.26,27,28 Therefore, we aimed to investigate the molecular properties of early-onset PE, based on the presence or absence of SGA, by analyzing the molecular profile of cytokines in AF obtained during cesarean section of pregnant women with early-onset PE.

METHODS

Patient and sample collection

AF was extracted via amniocentesis after exposing the amniotic sac during cesarean delivery in pregnant women who experienced preterm delivery, at less than 34 weeks’ gestation, and was stored in the Keimyung Human Resource Bank. AF and maternal hospital registration numbers were provided by the Keimyung Human Resource Bank. The clinical characteristics of the pregnant women and their infants were retrospectively reviewed.

As shown in Fig. 1, among the 343 cases wherein AF was collected between January 2017 and May 2022, 110 had early-onset PE. Among them, we excluded 13 cases with multiple chorionicities, either monochorionic or dichorionic, which are a known major cause of FGR through mechanisms different from fetal growth. We also excluded 12 cases exposed to preterm premature rupture of membrane or histologic chorioamnionitis considering the bacterial inflammatory response of AF. Additionally, we excluded cases with other risk factors including placenta abruption (n = 10) and congenital anomaly (n = 2). Finally, 73 cases with early-onset PE were investigated in the study. SGA infants were observed in 21 of the 73 enrolled cases. Twenty-five cases with appropriate-for-gestational age (AGA) infants, who had birth weights between the 25th and 75th percentile of the gestational age-specific birth weight, were arbitrarily selected as the control group.

Fig. 1. Flowchart of case selection.

SGA = small for gestational age, AGA = appropriate-for-gestational age.

The following maternal and neonatal data were retrospectively collected: maternal age, maternal body mass index (BMI), gestational age at delivery, birth weight, Apgar scores as well as the presence or absence of nulliparity, assisted reproductive technology, pre-gestational hypertension, gestational diabetes mellitus (GDM), and the emergence of hemolysis, elevated liver enzymes, and low platelets (HELLP) syndrome.

Proteomic analysis

AF was collected from 46 women with PE at the time of delivery and stored at −80°C. The samples were sent to Olink (Olink Proteomics, Watertown, MA, USA) and analyzed using the Olink Explore 384 Inflammation Panel. This panel was selected because of its unique inclusion of various inflammatory cytokines. The samples were processed using the Olink Explore platform, which comprises a Proximity Extension Assay (PEA) with next generation sequencing. Potential protein biomarkers were evaluated using PEA as described on the manufacturer’s website (https://www.olink.com), which provides an extensive list of potential protein biomarkers, comprising a total of 368 entries. Additionally, the website offers comprehensive assay validation data, including details on the limit of detection, lower and upper limits of quantification, and within- and between-run precision coefficients of variation. The assays ultimate protein concentration output is presented in terms of normalized protein expression (NPX) values. NPX is a logarithmic unit on a base 2 scale, where elevated NPX values correspond to higher protein concentrations. Protein expression data were obtained from 46 patients. Based on an adjusted P < 0.05, 35 potential protein biomarkers were identified. Among them, protein biomarkers with an absolute NPX value > 1 were considered as DEPs. Subsequently, 16 DEPs were selected and utilized for further analysis.

Gene ontology (GO) analysis of DEPs

The biological implications of the DEPs were assessed using the web-based tool Database for Annotation, Visualization, and Integrated Discovery (DAVID) 2021 (DAVID, http://david.ncifcrf.gov/, accessed on April 10, 2023). The GO terms were categorized into three distinct groups: biological processes (BP), cellular components (CC), and molecular functions (MF). P < 0.05 was considered significant.

Protein-protein interaction (PPI) network analysis

We utilized the Cytoscape software version 3.9.129 and Search Tool for the Retrieval of Interacting Genes version 11.5 (STRING, http://string-db.org, date last accessed on 10 April 2023) to analyze and visually represent the functional PPI network of the identified DEPs.30 The criteria for selecting interactions were a maximum number of interactors equal to 0, and a confidence score of ≥ 0.9. The Cytoscape plugin Molecular Complex Detection (MCODE) was used to identify clustered modules within the PPI network using the following parameters: degree cutoff, 2; node score cutoff, 0.2; k117 score, 2; and maximal depth, 100.

Enrichment analysis of hub genes

The Metascape Database (Metascape, https://metascape.org, date last accessed on April 11, 2023) was used to perform enrichment analysis of hub genes. The Metascape Database consolidates various authoritative data resources, including GO, Kyoto encyclopedia of genes and genomes (KEGG), UniProt, and DrugBank. This integration significantly enhances the pathway enrichment and annotation of BPs, providing comprehensive and reliable insights into the functional significance of genes and proteins.

Statistical analysis

All data analyses were performed using IBM SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Differences in clinical information between the groups were analyzed using Pearson’s chi-square test for categorical variables and Student’s t-test for continuous variables. Statistical significance was defined as P < 0.05. Receiver operating characteristic (ROC) curve analysis was performed using SPSS and an area under the curve (AUC) > 0.7 was regarded as indicative of a reliable diagnostic model.

Ethics statement

The study was approved by the Institutional Review Board (IRB) of Keimyung University Dongsan Medical Center (IRB approval number: 2022-04-041-002). All patients were educated about the study purpose and informed consent was obtained from each individual before their participation in the study.

RESULTS

Patients and demographic characteristics

Among a total of 46 pregnant women with PE, 21 women comprised the SGA group and 25 women the AGA group. As shown in Table 1, no differences were observed in maternal age, BMI, nulliparity, assisted reproductive technology, pre-gestational hypertension, GDM, and the emergence of HELLP syndrome between the two groups. Although mean birth weight was significantly lower in the SGA group than that in the AGA group (945.2 ± 302.3 vs. 1,590.0 ± 393.2, respectively; P < 0.001), no difference was observed in the mean gestational age at delivery (P > 0.05). As an indicator of the neonatal condition at birth, the 1 minute APGAR score was significantly lower in the SGA group than that in the AGA group (5.8 ± 1.3 vs. 6.7 ± 1.3, respectively; P = 0.019); however, no significant difference was observed in the 5 minutes APGAR score between the two groups.

Table 1. Maternal and neonatal characteristics.

Clinical characteristics		SGA group (n = 21)	AGA group (n = 25)	P value
Maternal factors
	Age, yr	32.8 ± 3.4	33.9 ± 3.8	0.284
	BMI	28.2 ± 7.3	30.4 ± 5.0	0.228
	Nulliparity	7 (33.3)	9 (36.0)	0.850
	ART	3 (14.3)	2 (8.0)	0.495
	Pre-gestational hypertension	1 (4.8)	1 (4.0)	0.900
	GDM	1 (4.8)	2 (8.0)	0.658
	HELLP syndrome	5 (23.8)	5 (20)	0.516
Neonatal factors
	Gestational age at delivery, wk	30.3 ± 2.6	31.4 ± 1.9	0.112
	Birth weight, g	945.2 ± 302.3	1,590.0 ± 393.2	< 0.001
	1 min APGAR score	5.8 ± 1.3	6.7 ± 1.3	0.019
	5 min APGAR score	7.9 ± 0.8	8.2 ± 0.7	0.135

Data are presented as the mean ± standard deviation or number (%).

SGA = small for gestational age, AGA = appropriate-for-gestational age, BMI = body mass index, ART = assisted reproductive technology, GDM = gestational diabetes mellitus, HELLP = hemolysis, elevated liver enzyme and low platelet, APGAR = appearance, pulse, grimace, activity and respiration.

Identification of DEPs

Sixteen DEPs were identified, all of which were upregulated in the SGA group compared with the AGA group. The gene symbols, description, log₂ fold change, and P values are listed in Table 2, and a heatmap showing the identified DEPs is shown in Fig. 2.

Table 2. Differential expression of proteins in women with preeclampsia according to the presence of small for gestational age newborn.

Gene symbols	Description	Log2 FC	Adjusted P value
EPO	Erythropoietin	1.98	0.037
WFIKKN2	Wap, Kazal, immunoglobulin, Kunitz and NTR domain-containing protein 2	1.67	0.006
CLSTN2	Calsyntenin-2	1.46	0.017
CSF3	Granulocyte colony-stimulating factor	1.44	0.017
COL9A1	Collagen alpha-1(IX) chain	1.33	0.031
SCG3	Secretogranin-3	1.30	0.028
CCL23	C-C motif chemokine 23	1.26	0.038
SKAP2	Src kinase-associated phosphoprotein 2	1.25	0.037
CCL20	C-C motif chemokine 20	1.15	0.006
GZMB	Granzyme B	1.14	0.047
TIMP3	Metalloproteinase inhibitor 3	1.14	0.029
FIS1	Mitochondrial fission 1 protein	1.06	0.028
IL17C	Interleukin-17C	1.03	0.017
PON3	Serum paraoxonase/lactonase 3	1.02	0.006
VEGFA	Vascular endothelial growth factor A	1.02	0.006
CXCL8	Interleukin-8	1.02	0.017

Log₂ FC = log₂ fold change.

Fig. 2. Heatmap showing the identified differentially expressed genes according to the presence or absence of small for gestational age using amniotic fluid retrieved during cesarean section from pregnant women with early-onset preeclampsia. The data are presented in matrix format, where rows correspond to individual genes and columns represent different tissues. Each cell in the matrix indicates the relative expression level of a genetic characteristic within a specific tissue. High expression levels are denoted by red and low expression levels are denoted by green, with intensity variations indicated by a scale bar.

AGA = appropriate-for-gestational age, SGA = small for gestational age.

DEP enrichment analysis

GO term analysis was performed using the DAVID software to analyze the functional profiles of the DEPs. In the BP, the DEPs were mainly involved in the positive regulation of the extracellular signal-regulated kinase (ERK) 1 and ERK2 cascade, chemokine-mediated signaling pathway, neutrophil chemotaxis, cellular response to interleukin (IL)-1, inflammatory response, chemotaxis, cellular response to tumor necrosis factor, immune response, induction of positive chemotaxis, cell-cell signaling, signal transduction, calcium-mediated signaling using an intracellular calcium source, lymphocyte chemotaxis, monocyte chemotaxis, negative regulation of neuronal death, and negative regulation of cell proliferation. In the CC, these proteins were mainly distributed in the extracellular region, space, and matrix. In the MF analysis, DEPs were mainly involved in cytokine activity, chemokine activity, heparin binding, metalloendopeptidase inhibitor activity, and CCR chemokine receptor binding (Table 3).

Table 3. GO terms of differential expressed genes among the groups.

Categories	Term	Gene symbols	No. of genes	P value
GO_BP	GO:0070374 Positive regulation of ERK1 and ERK2 cascade	CCL20, CCL23, EPO, VEGFA	4	< 0.05
	GO:0070098 Chemokine-mediated signaling pathway	CCL20, CCL23, CXCL8	3	< 0.05
	GO:0030593 Neutrophil chemotaxis	CCL20, CCL23, CXCL8	3	< 0.05
	GO:0071347 Cellular response to interleukin-1	CCL20, CCL23, CXCL8	3	< 0.05
	GO:0006954 Inflammatory response	CCL20, CCL23, CXCL8, IL17C	4	< 0.05
	GO:006935 Chemotaxis	CCL20, CCL23, CXCL8	3	< 0.05
	GO:0071356 Cellular response to tumor necrosis factor	CCL20, CCL23, CXCL8	3	< 0.05
	GO:0006955 Immune response	CCL20, CCL23, CXCL8, CSF3	4	< 0.05
	GO:0050930 Induction of positive chemotaxis	CXCL8, VEGFA	2	< 0.05
	GO:0007267 Cell-cell signaling	CCL20, CCL23, IL17C	3	< 0.05
	GO:0007165 Signal transduction	CCL20, CCL23, CXCL8, EPO, SKAP2	5	< 0.05
	GO:0035584 Calcium-mediated signaling using intracellular calcium source	CCL20, FIS1	2	< 0.05
	GO:0048247 Lymphocyte chemotaxis	CCL20, CCL23	2	< 0.05
	GO:0002548 Monocyte chemotaxis	CCL20, CCL23	2	< 0.05
	GO:1901215 Negative regulation of neuron death	CSF3, EPO	2	< 0.05
	GO:0008285 Negative regulation of cell proliferation	CCL23, CXCL8, SKAP2	3	< 0.05
GO_CC	GO:0005576 Extracellular region	CCL20, CCL23, CXCL8, TIMP3, COL9A1, CSF3, EPO, GZMB, IL17C, PON3, SCG3, VEGFA	12	< 0.05
	GO:0005615 Extracellular space	CCL20, CCL23, CXCL8, TIMP3, WFIKKN2, COL9A1, CSF3, EPO, IL17C, PON3, VEGFA	11	< 0.05
	GO:0031012 Extracellular matrix	TIMP3, COL9A1, VEGFA	3	< 0.05
GO_MF	GO:0005125 Cytokine activity	CSF3, EPO, IL17C, VEGFA	4	< 0.05
	GO:0008009 Chemokine activity	CCL20, CCL23, CXCL8	3	< 0.05
	GO:0008201 Heparin binding	CCL23, CXCL8, VEGFA	3	< 0.05
	GO:0008191 Metalloendopeptidase inhibitor activity	TIMP3, WFIKKN2	2	< 0.05
	GO:0048020 CCR chemokine receptor binding	CCL20, CCL23	2	< 0.05

GO = gene ontology, BP = biological process, CC = cellular component, MF = molecular function.

PPI network and hub gene analysis

We obtained a PPI network composed of 16 nodes and 19 edges, with an average node degree of 2.38 (P = 2.36E-10) to identify hub genes, which are highly interactive genes that play a role in early-onset PE with SGA (Fig. 2). Further analysis performed using the MCODE plugin identified six hub genes: C-C motif chemokine 20 (CCL20), granulocyte colony-stimulating factor 3 (CSF3), erythropoietin (EPO), vascular endothelial growth factor A (VEGFA), IL-17C (IL17C), and granzyme (GZMB) (Fig. 3).

Fig. 3. The protein-protein interaction network analysis of differentially expressed genes using SRING online database. Sixteen nodes and 19 edges with an average node degree of 2.38 (P = 2.36E-10) were identified.

Enrichment analysis of the hub genes

CCL20, CSF3, EPO, VEGFA, IL17C, and GZMB were subjected to additional enrichment analysis using the Metascape database, which combines GO function and KEGG pathway analyses. These genes were predominantly enriched in processes related to the positive regulation of peptidyl-tyrosine phosphorylation, cytokine-mediated signaling pathways, and positive regulation of the mitogen-activated protein kinase cascades. KEGG pathway enrichment analysis showed that hub genes were significantly enriched in cytokine-cytokine receptor interactions (Fig. 4).

Fig. 4. The six identified hub genes. Clustered differentially expressed genes were identified using Molecular Complex Detection in the protein-protein interaction network. Six nodes and nine edges were identified within this cluster.

ROC curve analysis of the hub genes

An ROC curve analysis was performed and the corresponding AUC, P value, and 95% confidence interval (CI) for each gene were calculated as follows to validate the diagnostic value of the six hub genes: CCL20, AUC = 0.821 (95% CI, 0.695–0.947; P < 0.001), with a sensitivity and specificity of 67 and 92%, respectively; CSF3, AUC = 0.794 (95% CI, 0.667–0.922; P = 0.001), with a sensitivity and specificity of 81 and 64%, respectively; EPO, AUC = 0.733 (95% CI, 0.580–0.886; P = 0.007), with a sensitivity and specificity of 57 and 88%, respectively; VEGFA, AUC = 0.840 (95% CI, 0.725–0.955; P < 0.001), with a sensitivity and specificity of 86 and 76%, respectively; IL17C, AUC = 0.787 (95% CI, 0.657–0.916; P = 0.001), with a sensitivity and specificity of 67 and 76%, respectively; and GZMB, AUC = 0.733 (95% CI, 0.589–0.878; P = 0.007), with a sensitivity and specificity of 62 and 80%, respectively (Fig. 5).

Fig. 5. ROC curve analysis was performed on the six hub genes to discriminate the occurrence of small for gestational age in early-onset preeclampsia. Each curve represents a specific hub gene. (A) CCL20; (B) CSF3; (C) EPO; (D) VEGFA; (E) IL17C; (F) GZMB. The calculated area under the ROC curve values for the respective genes were as follows: CCL20, 0.821; CSF3, 0.794; EPO, 0.733; VEGFA, 0.840; IL17C, 0.787; and GZMB, 0.733.

AUC = area under the curve, ROC = receiver operating characteristic.

DISCUSSION

PE is a major cause of both maternal and fetal morbidity and mortality and is traditionally considered a consequence of abnormal placental implantation.31 Inadequate remodeling of the spiral arteries during placental implantation results in failure to lower the resistance of the uterine circulation leading to high maternal blood pressure and hypo-perfusion of the placenta.32 Additionally, mechanical stress may result in maternal organ damage, including systemic endothelial damage and FGR.33 To date, PE treatment has mainly focused on regulating maternal blood pressure.34 However, accumulating evidence reveals that PE is a metabolic rather than a mechanical disorder.35,36 Therefore, recent studies have focused on aberrant cytokine production in PE, with evidence supporting the hypothesis that abnormal cytokine expression plays a key role in the development and manifestation of PE.37,38,39,40 However, the molecular mechanisms underlying PE remain unclear.

In the present study, we compared the molecular profile of cytokines, based on the presence or absence of SGA, using AF obtained during cesarean section in pregnant women with early-onset PE, and identified six hub genes (CCL20, CSF3, EPO, VEGFA, IL17C, and GZMB) from 16 overexpressed proteins in the AF collected from the SGA group. The proteins encoded by these genes are mainly involved in cytokine-cytokine receptor interactions. CCL20 is a chemokine ligand41 which regulates the inflammatory reaction by recruiting regulatory T cells and Th17 cells.42,43 Whereas CCL20 is upregulated in the late and early pregnancy plasma of patients with PE and thus, may be a novel potential predictive and diagnostic biomarker of PE.44 Meanwhile, EPO regulates red blood cell production.45 Hypoxic events induce renal EPO-producing cells, leading to increased plasma EPO levels to promote erythropoiesis.46 Moreover, EPO levels are elevated in AF collected from cases with PE or FGR.47,48 Conversely, placental insufficiency and impaired blood flow may cause placental hypoxia, which ultimately leads to changes in EPO levels. IL-17 plays a key role in T-cell activation and mainly regulates innate immunity cells against pathogens by promoting neutrophilic inflammation.49 Conversely, excessive IL-17 activity leads to autoimmune and inflammatory diseases.50 Lu et al.51 reported that IL-17 levels were elevated in both the maternal serum and placenta of women with PE. VEGF is produced by various cells and is mainly involved in angiogenesis.52 Notably, Sahay et al.53 reported that VEGF levels were higher in the control group than those in the PE group in central maternal, central fetal, peripheral maternal, and peripheral fatal regions of the placenta. Additionally, He et al.54 reported that downregulation of VEGF was observed in the placenta of women with PE. However, other studies found that VEGF levels are increased in the maternal serum and AF of patients with PE.55,56 These results indicate that VEGF plays an important role in the pathogenesis of PE but suggests that further studies are required. CSF3, known as granulocyte CSF, regulates the production and function of granulocytes57 and plays a key role in embryo implantation and placental development.58 Cai et al.59 reported that treatment with human recombinant CSF3 induced enhanced porcine embryonic development capacity in vitro. Moreover, a previous study reported that placental microvesicles from PE upregulated CSF3 in peripheral blood monocuclear cells.60 However, to the best of our knowledge, no study has evaluated CSF3 level changes in the AF of patients with PE.61 Du et al.62 reported that GZMB was upregulated in natural killer cells in early- and late-onset PE. Collectively, these studies showed that the six hub genes identified in our study were overexpressed in early-onset PE with SGA compared with normal conditions (Fig. 3), suggesting that each gene has high diagnostic value for PE (Fig. 5). Although direct experimental evidence is scarce, our study provides valuable information on AF collected from cases with early-onset PE according to the presence or absence of SGA.

To the best of our knowledge, this is the first study to use AF obtained during cesarean section from pregnant women with early-onset PE with SGA. Meanwhile, previous studies have performed diagnostic amniocentesis using ultrasound in the second trimester of pregnancy and used proteomic analysis to predict PE,20,21,22 suggesting that proteomic analysis in AF may be an important predictor of PE. In contrast, some studies report that cytokines in AF during the second trimester of pregnancy do not predict adverse pregnancy outcomes such as PE.26,27 However, these studies may differ in their expressed proteins compared with the present study, which analyzed AF retrieved during cesarean section after early-onset of PE with SGA. Moreover, labor is considered a sterile inflammatory event, wherein major changes in inflammatory cytokine expression occur irrespective of gestational age.63,64,65 Therefore, cases with suspected inflammatory reactions such as preterm labor, exposure to preterm premature rupture of membrane, and histologic chorioamnionitis were excluded from the study considering the impact of inflammatory reactions in proteomics analysis. In addition, by excluding multiple chorionicities, which are one of the main causes of FGR, we were able to conduct a comparative analysis based on the presence or absence of SGA using AF retrieved only from singleton pregnancies.

Nevertheless, our study has some limitations. First, this was a single center study with a limited sample size, therefore, the results are of a preliminary nature. The results of the power analysis confirmed that the final sample size was adequate to detect meaningful differences between the groups, with a statistical power of 98.8%. However, the limited sample size and the exclusion of 27 cases may restrict the generalizability of the findings. Additionally, we used strict criteria to avoid confounding factors, which may further limit the applicability of our findings. Therefore, larger cohort studies with multi-center collaborations, including diverse cases of early-onset PE with multiple pregnancies, preterm premature rupture of membranes (PPROM), and histologic chorioamnionitis for validation and broader applicability are required. Second, GO and PPI network analyses are generally performed using differentially expressed genes; however, we used the corresponding gene, which was derived from DEPs. In general, protein expression can be inferred based on the expression levels of mRNA; however, since we did not directly analyze the expression levels of mRNA, which may reflect altered proteins in AF, interpreting the results of GO and PPI network analysis as being related to PE with SGA has no scientific rationale. In addition, future studies should investigate mRNA expression levels using Wharton’s jelly from the umbilical cord and placenta, the tissues most closely related anatomically to AF because our results provide minimal direct experimental evidence for epigenetic regulatory mechanisms of protein expression, such as post-translational modifications, or for tissues or cells involved in DEP expression. Third, the exclusion criteria, which limit the inclusion of women with multiple pregnancies, PPROM, or chorioamnionitis may affect the external validity of the clinical study findings. Additionally, the cytokine evaluations performed in this study focused on inflammatory cytokines, therefore, it is important to consider that false positive results may occur in mothers with infected or inflamed placental and amniotic conditions. This is because inflammatory cytokines may increase in case of infection or inflammation, which implies that the cytokine expression observed in the study may not correspond to actual pathological changes. Moreover, multiple pregnancies may result in different cytokine expression between the fetuses, which could affect the results of the study.66 Thus, although we found differences in cytokine expression based on the presence of FGR in preeclamptic mothers, this study was conducted under conditions that excluded infection and inflammation. Therefore, potential biases due to infection, multiple pregnancies, and maternal inflammatory conditions must be considered when applying these findings to real-world clinical populations. Furthermore, the cytokine expression observed in infected or inflamed mothers may differ from the results reported in the study, which may limit the external validity of the findings and their applicability in clinical settings and could lead to suboptimal or inappropriate clinical decisions, particularly for high-risk patients. Consequently, future studies may need to include these higher-risk groups or conduct separate studies focusing specifically on them. Nonetheless, our findings may provide valuable information on the molecular characteristics involved in the pathogenesis of early-onset PE with SGA.

Taken together, these cytokines, encoded by six hub genes from 16 overexpressed proteins in the SGA group, were upregulated when exposed to certain conditions such as hypoxia, oxidative stress, and inflammatory stress. Likewise, analysis of AF obtained during cesarean section from patients with early-onset PE with SGA, in which the fetus is in a hypoxic state with exposure to oxidative stress and excessive inflammation, showed amplification of cytokines such as CCL20, CSF3, EPO, VEGFA, IL17C, and GZMB. Moreover, this is the first study to report that CSF3 levels are increased in the AF collected from cases with early-onset PE with SGA. Although, large-scale studies and extensive experimental validation are needed to verify our findings, changes in expression levels of CCL20, CSF3, EPO, VEGFA, IL17C, and GZMB in AF might be novel potential biomarkers of early-onset PE with SGA and provide new insights into the pathophysiology of PE with SGA.

In conclusion, although the results are of a preliminary nature, we identified 16 DEPs using AF obtained during cesarean section from pregnant women with early-onset PE and SGA. Moreover, six hub genes (CCL20, CSF3, EPO, VEGFA, IL17C, and GZMB) were overexpressed in the SGA group, which were mainly involved in cytokine-cytokine receptor interactions. These six upregulated hub genes might serve as novel biomarkers in early-onset PE with SGA. While further investigation is required to validate our results in women during the first or second trimester of pregnancy, our study may contribute to a better understanding of the pathogenesis of early-onset PE with SGA.