Abstract Background Delivery mode has been linked to child health, e.g., allergic disease. However, it remains unclear whether protein and metabolite differences across different delivery modes may underlie child development. Methods A cohort comprising 16 spontaneous onset vaginal delivery (VD), 16 prelabor cesarean delivery on maternal request (CS), and 8 intrapartum cesarean section (Intra_CS) women were analyzed using label-free proteomic and untargeted metabolomics assays on amniotic fluid and cord blood samples, respectively. We used weighted gene co-expression network analyses (WGCNA) to identify modules of highly correlated proteins or metabolites that associated with delivery modes and related clinical traits. KEGG enrichment analyses were performed to investigate the biological function of the identified modules. Integrative multiomics analysis was employed to examine the biological interplay between proteomic and metabolic interactions. Results Compared to the CS group, the proteomic and metabolomic profiles were similar between the Intra_CS and VD groups in our study. We did not identify any enriched protein or metabolite pathways related to immune development that could influence the risk of allergic diseases in offspring across different delivery modes. However, we identified seven protein modules correlated with the duration from the rupture of the membranes to full dilation of the cervix, with the actin cytoskeleton module significantly enriched. A metabolic module in cord blood that correlated with VD was enriched in subclasses including C21 steroids, steroid sulfates, and oxysterols. Integrative analysis of proteomic and metabolomic data suggested pathways related to mode of delivery and duration of labor, encompassing the actin cytoskeleton, NADP metabolic process, nicotinate, and nicotinamide metabolism in amniotic fluid, and the steroid hormone biosynthesis pathway in cord blood. Conclusions Differences in steroid hormones and the actin cytoskeleton pathway according to proteomics and metabolomics in amniotic fluid and cord blood were more indicative of the labor process. These findings could guide future studies on delivery-associated biochemical pathways. Supplementary Information The online version contains supplementary material available at 10.1186/s12884-024-07097-4. Keywords: Delivery mode, Cesarean section, Proteomics, Metabolomics, WGCNA Introduction The childbirth process causes short-term profound physical and mental changes in parturients. Different modes of delivery may lead to distinct biological responses in both the mother and the child [[36]1]. Understanding these molecular changes may help identify early biomarkers, optimize postnatal care and interventions, and potentially improve outcomes for both the mother and child [[37]2–[38]4]. Although a few studies have explored the effects of mode of delivery on omics profiles [[39]5, [40]6], they often focused on a single omics layer, analyzed only one type of sample (e.g., cord blood or amniotic fluid), or ignored the potential correlation between omics from samples that undergo concurrent changes during labor. Additionally, intrapartum caesarean deliveries were not included. Proteins, as the main carriers of biological activities, are directly related to phenotype. Yet, limited studies have investigated the impact of delivery modes on the label-free proteomic profile. In addition, the metabolic network is downstream from gene expression and protein synthesis. Therefore, metabolomics can reflect cell activity more closely at a more functional level. Metabolomics analyses have revealed changes in saccharides, corticosteroids, Cys, and other compounds in cord blood of the fetus that underwent labor [[41]5, [42]7, [43]8]. However, evidence regarding the relationships between omics and clinical traits is scarce. Utilizing multi-omics data as a holistic approach may provide a comprehensive understanding of the interaction between biological processes, helping to uncover the underlying mechanisms across different delivery modes in a more profound manner. The objectives of this study are twofold: (1) to conduct a comprehensive proteomic and metabolomic profiling of different modes of delivery (vaginal delivery, elective cesarean section without labor, and intrapartum cesarean section) using cord blood and amniotic fluid samples; (2) to explore the molecular composition of amniotic fluid and umbilical cord blood that associated with delivery mode and clinical traits of both mother and child. Methods Participants We conducted a prospective study including 40 pregnant women who had a singleton live birth at 39–40 gestational weeks without any of the following conditions: intrauterine infection, gestational diabetes, gestational hypertension, preeclampsia, and pre-pregnancy body mass index (BMI) ≥ 30 kg/m^2. Prior to enrollment, a written informed consent was obtained from all participants. Enrollment took place between November, 2020 and May, 2021 and follow-up continued until May, 2022. Participants were classified into three subgroups based on the mode of delivery: (1) prelabor cesarean section on maternal request (CS, reference); (2) spontaneous onset vaginal delivery (VD); (3) intrapartum cesarean section (Intra_CS). Amniotic fluid and cord blood samples were collected from all participants. In the VD and Intra_CS groups, amniotic fluid was collected from women with a descended and fixed presenting part, cervical dilation of at least 4 cm, delayed labor progression, and a willingness to expedite the labor process, informed consent was obtained to perform an artificial rupture of membranes. After thorough disinfection of the vulva, a speculum was inserted to expose the cervix and ensure a clear view of the amniotic sac. The inside of the vagina was wiped with dry gauze to reduce potential contaminants. The fetal head was carefully avoided during the procedure. Amniotic fluid was collected by puncturing the sac with a sterile syringe or gently aspirated from the rupture site using a needle-free syringe. In the CS group, after the uterine myometrium was incised to expose the amniotic sac, amniotic fluid was aseptically suctioned using a sterile syringe inserted the intact amnion membrane during cesarean section by an obstetrician. Only amniotic fluid samples collected without visible blood or meconium contamination were included in the study. At birth, umbilical cord blood samples were collected, centrifuged for plasma separation, and aliquoted for further analysis. Demographic information and delivery data were acquired through medical records. Ethics approval was obtained from both the Ethics Committee of the International Peace Maternity and Child Health Hospital and the Ethics Committee of Xinhua Hospital, both of which are affiliated with the Shanghai Jiao Tong University School of Medicine. Proteomic analysis The proteome of amniotic fluid and plasma cord blood samples were analyzed with label‑free liquid chromatography with tandem mass spectrometry (LC-MS/MS). These proteomic assays were conducted at Jingjie Bio (Hangzhou, China) according to standard procedures [[44]9]. We used 500 µL of plasma and amniotic fluid samples per sample. Briefly, before the highly abundant proteins were removed with the Pierce™ Top 14 Abundant Protein Depletion Spin Columns Kit (ThermoFisher Scientific), cellular debris was removed from the plasma and amniotic fluid samples. Then the protein concentration was quantified using the BCA kit following the manufacturer’s instructions. After trypsin digestion, the peptides were separated on an EASY-nLC 1200 UPLC system (ThermoFisher Scientific) prior to LC–MS/MS analyses. For both amniotic fluid sample and plasma sample, the tryptic peptides were gradient eluted from the column from 5 to 25% solvent B (0.1% formic acid in 90% acetonitrile) over 60 min followed by a ramp of 25–35% in 22 min, and climbing to 80% in 4 min then holding at 80% for the last 4 min (solvent A: 0.1% formic acid, 2% acetonitrile/ in water). The flowrate was set at 500 nL/min. The separated peptides were analyzed on a Q Exactive ^TM HF-X mass spectrometer (ThermoFisher Scientific) with a nano-electrospray ion source. For MS acquisition, the scan resolution was set to 60,000 for a scan range of 350–1600 m/z, and a maximum injection time of 60 ms. Up to 20 most abundant precursors were then selected for further MS/MS analyses using the Orbitrap mass analyzer with 30 s dynamic exclusion, a resolution of 30,000, and automatic gain control (AGC) set to 1e^5. The obtained MS/MS data were processed using the MaxQuant search engine (version 1.6.15.0). Metabolomics analysis The untargeted metabolomics were analyzed with liquid chromatography with tandem mass spectrometry (LC-MS/MS) in both amniotic fluid and cord blood plasma samples. These assays were conducted at Jingjie Bio (Hangzhou, China). Each 100 µL amniotic fluid or plasma were extracted by adding 400 µL of methanol extraction solution. After the protein precipitate was removed by centrifugation, the extracted metabolites were then separated using the Waters ACQUITY ultra-high-performance liquid chromatography (UPLC) system with a Waters ACQUITY UPLC BEH C18 column (1.7 μm, 2.1 mm × 100 mm). The elution was performed at a flow rate of 400 µL/min at a column temperature of 40 ℃. Mobile phase A consisted of a water solution containing 0.1% formic acid, while mobile phase B consisted of an acetonitrile-water solution containing 0.1% formic acid. The liquid gradient was set as follows: 0–11 min, 2–98% B; 11.0–12.0 min, 98% B; 12.0–12.1 min, 98–2% B; 12.1–15.0 min, 2% B. After separation, metabolites were ionized and then analyzed using the timsTOF Pro mass spectrometer (Bruker, USA). The ion source voltage was set at 4.5 kV, and both peptide ions and their fragment ions were detected and analyzed using high-resolution TOF. The mass spectrometry scanning range was set from 50 to 1300 m/z. Data acquisition was performed in Parallel Accumulation Serial Fragmentation (PASEF) mode. After acquiring a full MS spectrum, two rounds of PASEF MS/MS scans were conducted for precursor ions with charge states ranging from 0 to 1. The dynamic exclusion time for tandem mass spectrometry scans was set to 6 s to avoid repeated scanning of precursor ions. The mass spectrometry data were processed using MetaboScape 2022 for peak extraction, alignment, and retention time correction of the raw data. Metabolite structures and annotation information were obtained by comparing the spectra with databases such as NIST, HMDB, proprietary databases, and integrated public databases. Weighted protein co-expression network analysis To determine co-expressed protein correlations with the mode of delivery and related clinical variables, and to identify the underlying cellular responses or pathways, a weighted protein co-expression network was built with R package “WGCNA” [[45]10] using the protein data. To attain a scale-free R^2 fit of 0.9 with the smallest threshold, the thresholding power was set at 6 for cord blood and 5 for amniotic fluid when constructing the co-expression networks. We introduced an unsigned topological overlap measure (TOM) to identify modules of highly co-expressed proteins, and to filter out spurious or isolated connections. Network relationships among proteins were then identified using hierarchical clustering. A dynamic tree-cutting algorithm was utilized to determine the module assignments. Module eigenproteins (MEs), which were representative summaries of the protein expression profiles within a module, were derived by calculating the first principal component of the module’s protein expression data. We also calculated the module membership (MM), which measures the correlation between a protein and the module eigenprotein of a specific module. Proteins with high MM (top 10) within a module were identified as top drivers. We used Pearson correlation to examine the relationship between the eigenproteins of identified modules and clinical traits including delivery mode (CS vs. other group), sex, birth weight, parity, duration of the first stage of labor, duration of the second stage of labor, duration between rupture of the membranes (ROM) and full cervical dilation, duration between ROM and delivery, duration of labor onset to ROM, administration of labor epidural analgesia, labor analgesia as well as oxytocin use (OT). Only modules that showed associations with at least one clinical trait and passed the Benjamini-Hochberg (BH) correction (P < 0.05) were selected for further analysis. Weighted metabolites co-expression network analysis Weighted metabolite co-expression networks were constructed using the WGCNA approach. This was done separately for the cord blood and amniotic fluid metabolite data. The construction process followed the steps outlined in the “weighted protein co-expression network analysis” section. The thresholding power was set at 5 for both cord blood and amniotic fluid. Similarly, we investigated the relationship between metabolic module membership within the selected modules and clinical traits of interest, following a similar approach as in the analysis of protein modules. Enrichment analyses of protein modules and annotation of top drivers of the metabolic modules To investigate the biological function of the protein modules that are relevant to the clinical characteristics, we performed KEGG enrichment analyses using WebGestalt (WEB-based Gene SeT AnaLysis Toolkit), with the genome database serving as the reference set. Pathways with a false discovery rate (FDR) below 0.05 were considered statistically significant. The pathways were selected based on FDR significance. If more than 10 pathways were significant by FDR, the top 10 pathways were chosen. We also constructed the protein-protein interaction network for the top 10 protein drivers of modules that correlated with delivery mode, using online tool STRING ([46]http://string-db.org). Furthermore, metabolic annotation was performed in metabolic modules that were linked with at least one clinical trait related to delivery mode. MetaboAnalyst ([47]https://www.metaboanalyst.ca/) was used to define the sub-class for top drivers of metabolic modules. Multi-omics integration To gain a more comprehensive insight into the biological interplay between proteomic and metabolic interactions, we proceeded to co-enrich correlated proteomic and metabolic modules within KEGG pathways. This was accomplished through an integrated pathway analysis approach using MetaboAnalyst ([48]https://www.metaboanalyst.ca/). Furthermore, we conducted Pearson correlation analyses to evaluate the pairwise correlations among the top drivers within the identified proteomic and metabolic modules. Cytoscape software was used to visualize the correlation network and KEGG pathway involving the identified proteins and metabolites. Statistical analysis The proteomic and metabolomic data that were quantified less than 50% of all the detected subjects in a particular delivery mode group was discarded. We imputed the missing values with k-nearest neighbour (KNN) imputation function [[49]11], then log2 converted data was utilized for further analyses. Differentially expressed proteins and metabolites were identified using Student’s t-test. Differential proteins and metabolites were selected based on criteria including Fold Change (FC) and t-test p-value. A threshold of FC greater than or equal to the absolute value of log2 1.5 was applied to ascertain significant differences. P-values from the differential analysis, correlation analysis and KEGG pathway enrichment were adjusted using the Benjamini-Hochberg (BH) method to avoid Type I error rate. Results Demographics of participants Table [50]1 presents the demographics and baseline characteristics. The VD group had a slightly higher gestational age than the CS and Intra_CS groups. No significant differences were observed among the three groups regarding newborn gender, birth weight, maternal age, maternal body mass index (BMI), and parity. Table 1. Demographics and baseline characteristics of participants Characteristic Prelabor cesarean delivery on maternal request Spontaneous onset vaginal delivery Intrapartum cesarean section P N 16 16 8 Newborns Male 7 (43.8) 11 (68.8) 6 (75.0) 0.221 Gestational age 39.3 (0.7) 40.4 (1.3) 40.0 (1.0) 0.011 Birth weight 3428 (329) 3384 (358) 3442 (366) 0.904 Apgar scores 10.0 (0.0) 9.9 (0.3) 9.9 (0.4) 0.421 Mothers Age 32.8 (3.2) 30.5 (3.5) 31.1 (3.9) 0.186 Prepregnancy BMI (kg/m^2) 20.2 (2.1) 21.1 (2.5) 19.5 (1.2) 0.190 Parity 0 11 (68.8) 15 (93.8) 8 (100.0) 0.058 ≥ 1 5 (31.2) 1 (6.2) 0 (0.0) Education Lower than college 1 (6.2) 0 (0.0) 0 (0.0) 0.084 College degree 8 (50.0) 15 (93.8) 5 (62.5) Higher than college 7 (43.8) 1 (6.2) 3 (37.5) First stage of labor (hours), Median (IQR) NA 9.2 (4.0) NA Second stage of labor (minutes), Median (IQR) NA 36 (26.3) NA Total labor hours, Median (IQR) NA 9.9 (4.3) 12.5 (8.6) Duration of labor onset to ROM ^a, Median (IQR) NA 6.4 (4.2) 11.0 (8.3) Duration of ROM to cervix fully dilated (hours), Median (IQR) NA 2.4 (2.3) NA Duration of ROM to childbirth (hours), Median (IQR) NA 3.1 (1.7) 6.0 (4.4) Epidural analgesia (%) 0 (0.0) 14 (87.5) 6 (75.0) < 0.001 Analgesia (%) 16 (100.0) 0 (0.0) 8 (100.0) < 0.001 Oxytocin use (%) 0 (0.0) 12 (75.0) 5 (62.5) < 0.001 [51]Open in a new tab ROM, rupture of the membranes. Comparative proteomic and metabolic profiling of different delivery modes A total of 4593 quantifiable proteins were detected in amniotic fluid samples by spectrum search analysis. For further analysis, 3,752, 3,847, and 3,932 proteins that were detected in at least 50% of subjects in the CS, VD, and intra-CS groups, respectively, were identified. For cord blood, 4271 quantifiable proteins were detected, and 3,752, 3,847, and 3,932 proteins met the filtration criteria were included for further analysis in the CS, VD, and intra-CS groups, respectively. A total of 1336 metabolites for amniotic fluid samples and 1482 metabolites for cord blood passed quality control. The exploratory analyses of differentially expressed proteins and metabolites are presented in Figures [52]S1-4. Venn diagrams illustrating the differential proteins and metabolites among the VD, CS, and Intra_CS groups (Figures [53]S1a, [54]S3a, and [55]S4a) show limited differences between the VD and Intra_CS groups. We further compared the pathways enriched for differentially detected proteins between CS vs. VD and CS vs. Intra_CS. The results revealed a significant overlap in pathways (Fig. [56]1), indicating the similar profiles observed in proteins between the Intra_CS and VD groups. Consequently, we combined these two groups into a single group, which we referred to as the merged_labor group for subsequent analysis. Fig. 1. [57]Fig. 1 [58]Open in a new tab KEGG enrichment analysis for proteomics between groups of delivery mode. (a) KEGG pathway enriched in amniotic fluid proteins between CS vs. VD and CS vs. Intra_CS. (b) KEGG pathway enriched in cord blood proteins between CS vs. VD and CS vs. Intra_CS. VD, spontaneous onset vaginal delivery group. CS, prelabor cesarean delivery on maternal request, Intra_CS, intrapartum cesarean section In the amniotic fluid sample, the volcano plots revealed significant differences in 347 proteins (Fig. [59]2a) and 122 metabolites (Fig. [60]2c) with a fold-change greater than 1.5 and p-values < 0.05 after Benjamin-Hochberg correction. In contrast, for the cord blood sample, the volcano plots showed that 0 proteins (72 proteins with raw p values < 0.05, Fig. [61]2b) and 50 metabolites (Fig. [62]2d) exhibited statistically significant differences. Fig. 2. [63]Fig. 2 [64]Open in a new tab Volcano plot for the proteome and metabolome analysis in amniotic fluid and cord blood sample among the merged vaginal delivery (merged_labor) group and prelabor cesarean delivery on maternal request (CS) group. (a) Volcano plots of the proteins in the amniotic fluid. (b) Volcano plots of the proteins in the cord blood. (c) Volcano plots of the metabolites in the amniotic fluid. (d) Volcano plots of the metabolites in the cord blood. Benjamini-Hochberg (BH) corrected p-value < 0.05 and log2FC > 0.585 or < -0.585 were used as cutoffs to define up and down regulation Hierarchical cluster analyses were applied to the differentially proteins identified in amniotic fluid, revealing two major clusters (Fig. [65]3a). In Cluster I, concentrations of 159 proteins were found to be upregulated in the merged_labor group, including notable proteins such as thioredoxin domain-containing protein 17, dihydropteridine reductase, and cytosolic non-specific dipeptidase. On the other hand, Cluster II exhibited higher expression levels in 188 proteins in the CS group compared to the merged_labor group. Among these proteins were intercellular adhesion molecule 1, protein canopy homolog 3, and programmed cell death 1 ligand 2. KEGG annotation and enrichment analyses of the differentially expressed proteins in amniotic fluid demonstrated that most proteins were involved in the metabolic pathway (Fig. [66]3b). Fig. 3. [67]Fig. 3 [68]Open in a new tab Quantitative proteomics and metabolomics of amniotic fluid and cord blood sample among the merged vaginal delivery (merged_labor) group and prelabor cesarean delivery on maternal request (CS) group a. Hierarchical clustering based on differential proteins in amniotic fluid; b. Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis of differentially expressed proteins in amniotic fluid; c. Hierarchical clustering based on differential metabolites in amniotic fluid; d. KEGG pathway enrichment analysis of differentially expressed metabolites in amniotic fluid; e. Hierarchical clustering based on differential metabolites in cord blood; f. KEGG pathway enrichment analysis of differentially expressed metabolites in cord blood. ROM_birth, duration of rupture of the membranes (ROM) to childbirth (hours). For hierarchical clustering, the differentially expressed proteins and metabolites were identified with Benjamini-Hochberg (BH) corrected p-value < 0.05 and log2FC > absolute value of log2|1.5|. Significantly altered proteins and metabolites are highlighted in red (increased) and blue (decreased) Regarding metabolites, in amniotic fluid, the differential metabolites that were highly detected in the merged_labor group were mainly enriched in sphingolipid metabolism and the biosynthesis of unsaturated fatty acids (Fig. [69]3c, d). Notably, these differential metabolites included cortisol, 8,11,14-eicosatrienoic acid, and ropivacaine. Similarly, in cord blood, the differential metabolites that were highly detected in the merged_labor group consisted of hydrocortisone, ropivacaine, aldosterone, 2’,6’-Pipecoloxylidide, 11-deoxycortisol, and progesterone, primarily enriched in the steroid hormone biosynthesis pathway (Fig. [70]3e, f). Association of protein modules with clinical traits related to delivery mode To investigate the association between delivery mode-related clinical traits and proteins/metabolites at a systems level, we conducted an integrative multiscale network analysis using the WGCNA approach. Subsequently, we correlated the identified modules with the clinical traits in question. Amniotic fluid In amniotic fluid, we found nine protein modules that were associated with at least one trait after controlling the false discovery rate (FDR) (Fig. [71]4a, Figure [72]S5). Among these modules, turquoise, greenyellow and black modules were found to be correlated with the mode of delivery. Specifically, the turquoise module showed a positive correlation with VD, while the black module displayed a decrease in the VD group, as well as in women who received labor epidural analgesia. Notably, seven modules were observed to correlate with the duration of the rupture of the membranes (ROM) to full dilation of the cervix. Among these, three modules showed positive correlations, while four modules exhibited negative correlations. Fig. 4. [73]Fig. 4 [74]Open in a new tab Correlations among protein networks and metabolic networks and clinical traits in amniotic fluid. a. Circus plot showing the correlations among metabolite modules, protein modules and clinical traits in amniotic fluid. Only significant associations after FDR correction are shown, and the width of the lines represents the strength of the correlations. b. Protein-protein interaction plot of top drivers for turquoise module in amniotic fluid. BP, MF, and CC represent Biological Process, Molecular Function, and Cellular Component groups of gene ontology (GO), respectively. c. Top drivers of turquoise module between merged_labor and CS group (top), and correlation between top drivers of turquoise module with the duration of ROM to childbirth. d. Heat map showing the Pearson correlations and FDR-adjusted P values (in bracket) between metabolite modules (rows) and protein modules (columns) associated with clinical traits in amniotic fluid We identified top 10 protein drivers with a high module membership in module correlated with at least one clinical trait. Table [75]S1 presents these top drivers for modules with an absolute correlation coefficient greater than 0.5. The top drivers of the turquoise module including Actin-related protein 3, Actin, cytoplasmic 2, and Cofilin-1, which have been reported to be associated with regulation of actin cytoskeleton process participating in regulating uterine contractility. The protein-protein interaction and enriched pathway are shown in Fig. [76]4b. All these drivers are significantly higher in VD group and show a positive correlation with the duration of ROM to cervix fully dilated or the duration of ROM to childbirth (partially presented in Fig. [77]4c). In the black module, the top driver tumor necrosis factor receptor superfamily member 19 L (TNFRSF19L) was negatively correlated with the use of epidural analgesia. Further, we conducted over-representation analyses on these top drivers to delve into the functional roles of proteins within the identified modules. Only the turquoise module had significantly enriched pathways (Figure [78]S6). After the Benjamini-Hochberg correction, pathways including pentose phosphate pathway, the regulation of actin cytoskeleton, and bacterial invasion of epithelial cells were identified for the turquoise module. Cord blood For cord blood, we identified a total of four protein modules associated with at least one trait (Fig. [79]5a). The greenyellow module demonstrated a positive correlation with VD. And the turquoise module exhibited a positive correlation with the duration of the second stage of labor, whereas the greenyellow module displayed a negative correlation with this parameter. Furthermore, the greenyellow module also displayed an association with the utilization of labor epidural analgesia, labor analgesia, and oxytocin (Figure [80]S7). Fig. 5. [81]Fig. 5 [82]Open in a new tab Correlations among protein networks and metabolic networks and clinical traits in cord blood. (a) Circus plot showing the correlations among metabolite modules, protein modules and clinical traits in cord blood. Only significant associations after FDR correction are shown, and the dotted edges represent a significant association between the protein module and the metabolite module. The width of the lines represents the strength of the correlations. (b) Protein-protein interaction plot of top drivers for turquoise module in cord blood. CC represents Cellular Component groups of gene ontology (GO). (c) Levels of Cofilin-1, top driver of turquoise module of proteins, between the merged_labor and CS group. (d) Correlation between Cofilin-1 in cord blood with the duration of second stage of labor. (e) Levels of metabolic green module between the merged_labor and CS group. (f) Levels of metabolic magenta module between the merged_labor and CS group. (g) Heat map showing the Pearson correlations and FDR-adjusted P values (in bracket) between metabolite modules (rows) and protein modules (columns) associated with clinical traits in cord blood Table [83]S2 presents the top ten protein drivers for modules with an absolute correlation coefficient greater than 0.5. In the greenyellow module, Ephrin type-A receptor 4 and Ephrin-B1 emerged as two of its top 10 drivers. Within the turquoise module, prominent drivers include Cofilin-1, Adenylyl cyclase-associated protein 1, Coronin-1 C, and Glyceraldehyde-3-phosphate dehydrogenase, all of which were associated with actin cytoskeleton regulation or energy metabolism (Fig. [84]5b). Cofilin-1, also identified as the top driver of turquoise in amniotic fluid, was not expressed differentially in the cord blood between the merged_labor and CS group (Fig. [85]5c). However, it’s worth mentioning that Cofilin-1was positively correlated with the duration of the second labor (Fig. [86]5d). After BH correction, no pathways were significantly enriched in the over-representation analyses of top proteins in the identified modules. Association of metabolite modules with clinical traits We identified three metabolite modules that associated with at least one trait in amniotic fluid (Fig. [87]4a and [88]S8). Only the turquoise module showed absolute correlation coefficients greater than 0.5. The turquoise module showed a negative correlation with merged_labor (R^2 = -0.72) and positive correlation with using spinal anesthesia (R^2 = 0.76). The top 10 metabolite drivers of turquoise module were listed in Table [89]S3. The turquoise module was enriched in oligopeptide and peptide (Table [90]S4). In cord blood, five metabolic modules were associated with at least one trait after FDR correction (Fig. [91]5a and [92]S9). The green and magenta modules were correlated with merged_labor. The purple, black and turquoise modules were in correlation with duration of labor. Table S5 shows the top 10 metabolite drivers of each module. The metabolites in the green module were enriched in subclasses including C21 steroids, steroid sulfates, and oxysterols. The magenta module was enriched in ceramide phosphocholines, glycerophosphoserines, diacylglycerophosphoserines, PE, and SM (Figure [93]S10). Integrated omics analysis We examined the relationship between protein modules and metabolic modules that were correlated with clinical traits in amniotic fluid and cord blood, respectively. There were five and one significant associations between protein modules and metabolite modules in amniotic fluid (Fig. [94]4a, d) and cord blood (Fig. [95]5a, g), respectively, after BH correction. For amniotic fluid, integrative analyses were further launched in the correlated protein and metabolites modules with absolute correlation coefficient greater than 0.6. Joint pathway analysis for protein (turquoise) and metabolites (turquoise) modules identified 3 pathways including actin cytoskeleton, NADP metabolic process, and nicotinate and nicotinamide metabolism. The strongly correlated (r > 0.7) proteins and metabolites and pathway involved are shown in Fig. [96]6a. The integrated pathway analysis for protein (black) and metabolites (turquoise) modules identified nicotinate and nicotinamide metabolism pathway (Fig. [97]6b). Fig. 6. [98]Fig. 6 [99]Open in a new tab Pathway enrichment analysis for metabolic and proteomic top drivers of identified modules. (a) joint pathway analysis for protein (turquoise) and metabolites (turquoise) modules, borderline indicate pathway that enriched in the joint pathway analysis in amniotic fluid. (b) joint pathway analysis for protein (black) and metabolites (turquoise) modules in amniotic fluid. (c) joint pathway analysis for protein (greenyellow) and metabolites (green) modules in cord blood For cord blood, the only significant correlation between protein (greenyellow) and metabolites (green) modules was further investigated. Joint pathway analysis identified steroid hormone biosynthesis pathway (Fig. [100]6c). All top drivers are depicted in a correlation network to visualize interomic correlations (edges indicate an absolute R > 0.6, N = 40). Discussion Our study provides comprehensive resources that characterize different modes of delivery by profiling 3,788 proteins and 1,366 metabolites in amniotic fluid, as well as 3,023 proteins and 1,482 metabolites in cord blood. Overall, our study did not identify any enriched protein or metabolite pathways related to immune development that could influence the risk of allergic diseases in offspring across different delivery modes. However, we identified modules of proteins and metabolites that were strongly correlated with delivery mode and related clinical traits, such as the duration of ROM to full dilation of the cervix. Several protein and metabolite modules were observed to be correlated with each other. Finally, integrated pathway analysis for amniotic fluid and cord blood identified different significant pathways for actin cytoskeleton, NADP metabolic process, nicotinate and nicotinamide metabolism, and steroid hormone biosynthesis pathway. To the best of knowledge, this is the first study that explored the proteomics and metabolomics for individuals underwent intrapartum cesarean section. We observed similar proteomic and metabolomic profiles between the Intra_CS and VD groups. This finding is biologically plausible due to the similar physiological processes during labor such as uterine contractions, hormonal changes, and stress responses [[101]12]. Particularly, for amniotic fluid samples collected around the same stage of labor in both groups, the profiles reflect the metabolic and proteomic state associated with active labor. To examine the effect of labor, we first identified differentially expressed proteins between the merged_labor and CS groups. The identified proteins upregulated in the amniotic fluid of the merged_labor group, such as thioredoxin domain-containing protein 17, dihydropteridine reductase, and cytosolic non-specific dipeptidase, are involved in redox signaling, synthesis of neurotransmitters, and protein turnover/degradation, respectively. Vaginal delivery involves significant mechanical stress and potential hypoxia-reoxygenation events, leading to oxidative stress [[102]13]. The increased oxidative stress during VD might enhance the demand for redox signaling, necessitating higher levels of thioredoxin domain-containing protein 17 to modulate the redox state and protect cells [[103]14]. Vaginal delivery induces more significant physical exertion and stress responses compared to cesarean section, triggering the upregulation of stress-related and metabolic enzymes to cope with the physiological changes. Cytosolic non-specific dipeptidase may support efficient amino acid recycling and utilization [[104]15], while dihydropteridine reductase plays a crucial role in regenerating tetrahydrobiopterin, a cofactor for the synthesis of neurotransmitters and other metabolites [[105]16], thereby supporting the heightened metabolic activity observed in VD. Our study revealed that the differential metabolites in the merged_labor group were enriched in sphingolipid metabolism and the biosynthesis of unsaturated fatty acids in amniotic fluid, and in steroid hormone biosynthesis in cord blood. Sphingolipids are structural membrane components with crucial signaling functions. Among sphingolipids, ceramides are well-known mediators of stress signals and pro-inflammatory responses [[106]17]. Previous reports demonstrated that ceramide concentration, a well-known sphingolipid, was upregulated during labor, consistent with our results [[107]18]. The biosynthesis of unsaturated fatty acids is crucial for cell membrane fluidity and signaling, which are essential during the physical stress of vaginal delivery [[108]19, [109]20]. Furthermore, the increased synthesis of steroid hormones during labor aids in the adaptation to stress and the progression of labor [[110]21]. Studies have reported significant increases in steroid hormones, including cortisol and estrogen, during vaginal delivery [[111]22, [112]23], indicating the metabolic adaptations underline the complex physiological adjustments that occur during vaginal delivery. The protein co-expression analysis for amniotic fluid revealed correlations of three protein modules with the mode of delivery and seven protein modules correlated with the duration of ROM to full dilation of the cervix. Notably, only the turquoise module showed significant enrichment in pathways, including the pentose phosphate pathway, regulation of the actin cytoskeleton, and bacterial invasion of epithelial cells. The top drivers of the turquoise module, which are higher in the VD group, exhibit a positive correlation with the duration of ROM to cervix fully dilated and the duration of ROM to childbirth. Actin proteins are crucial for actin cytoskeleton regulation, essential for uterine contractions [[113]24, [114]25]. Vaginal delivery involves significant mechanical and physiological changes, requiring extensive actin cytoskeleton remodeling to facilitate uterine contractions, fetal descent, and cervical dilation [[115]24]. Consistent with our findings, the actin regulation pathway was enriched and correlated with labor duration in a transcriptome analysis [[116]26]. In uterine smooth muscle, there is approximately six times more actin than myosin [[117]27]. The upregulation of actin-related proteins reflects the increased demand for cytoskeletal dynamics in vaginal delivery. Longer durations from ROM to full dilation involve extended periods of uterine contractions and mechanical stress on the cervix, necessitating ongoing cytoskeletal remodeling [[118]25]. We speculate that the correlation between actin proteins and the duration of ROM to full dilation underscores the importance of cytoskeletal remodeling in response to the mechanical stress and physiological demands associated with childbirth. Notably, the turquoise module does not exhibit any correlation with the duration of labor onset to ROM (onset_ROM), which is the time of amniotic fluid collection. This indicates a potential predictive role of amniotic fluid actin cytoskeleton proteins in predicting the duration of labor. Although no pathway was significantly enriched for the top proteins in the identified protein modules of cord blood, it is noteworthy that the turquoise module exhibited a positive correlation with the duration of the second stage of labor. Key drivers within this module, such as Cofilin-1, Adenylyl cyclase-associated protein 1, Coronin-1 C, and Glyceraldehyde-3-phosphate dehydrogenase, are all involved in actin cytoskeleton regulation or energy metabolism [[119]28, [120]29]. Consistent with the findings from the proteomic analysis of amniotic fluid, the correlation between Cofilin-1 and the duration of the second stage of labor highlights the importance of cytoskeletal dynamics and energy requirements during prolonged labor. Although Cofilin-1 was not differentially expressed between the cord blood of merged_labor and CS groups, its specific association with the duration of labor underscores its potential role in labor progression. We found metabolites in cord blood of green module enriched in subclasses including C21 steroids, steroid sulfates, and oxysterols, were positively correlated with VD. C21 steroids, including cortisol and other glucocorticoids, are essential hormones produced by the adrenal cortex. They are involved in various physiological functions, including stress response and metabolic regulation [[121]30, [122]31]. Vaginal delivery involves significant hormonal changes, including elevated levels of stress hormones like cortisol and other glucocorticoids [[123]32]. These hormones are crucial for managing the physical stress and metabolic demands of labor. Gitau et al. reported that cortisol and progesterone levels were significantly higher in the normal vaginal delivery group compared to the cesarean section group, which can induce fetal stress through cord blood [[124]33]. The increase in C21 steroids and their sulfates reflects the body’s response to the intense physiological challenge of VD. Steroid sulfates, conjugated forms of steroids, serve as reservoirs for active steroids and can be converted back into their active forms when needed, providing a mechanism for rapid hormonal regulation [[125]34]. During labor, the increased demand for steroids to manage stress and metabolic needs may lead to elevated levels of steroid sulfates, ensuring a ready supply of active hormones [[126]35]. Oxysterols, oxygenated derivatives of cholesterol, have various biological functions including regulating cholesterol metabolism, and the synthesis of bile acids and steroid hormones [[127]36]. The metabolic demands of VD may stimulate increased cholesterol turnover and oxysterol production, reflecting heightened metabolic activity. Oxysterols also modulate inflammatory and immune responses [[128]37] and may act as signaling molecules in various pathways that manage inflammation and tissue repair during VD, likely triggering inflammatory responses [[129]38]. Additionally, the magenta module for cord blood was enriched in ceramide phosphocholines, glycerophosphoserines, diacylglycerophosphoserines, PE, and SM, which were positively correlated with VD. These lipid metabolites are essential components of cell membranes and play crucial roles in maintaining cell membrane integrity, facilitating cell signaling, and managing energy metabolism, ensuring fetal cells can withstand and adapt to the intense conditions of labor [[130]39]. Further, integrative analysis of proteomic and metabolomic data suggests that several pathways that are differentially regulated in amniotic fluid and cord blood in the context of mode of delivery and related clinical traits such as duration of labor. These pathways include the actin cytoskeleton, NADP metabolic process, nicotinate and nicotinamide metabolism for amniotic fluid, and the steroid hormone biosynthesis pathway for cord blood. The actin cytoskeleton pathway in amniotic fluid provides insights into the physiological and biochemical changes necessary for maintaining cell shape, enabling cell motility, and facilitating cellular responses to mechanical stress [[131]40]. Enhanced actin cytoskeletal activity may support the structural demands of labor and facilitate cellular processes involved in tissue remodeling and repair. NADP (nicotinamide adenine dinucleotide phosphate) is vital for cellular redox balance and serves as a cofactor in anabolic reactions, including lipid and nucleic acid synthesis [[132]41]. The metabolic demands of labor necessitate efficient energy production and redox regulation [[133]42]. Nicotinate (niacin) and nicotinamide are precursors of NAD + and NADP+, which are crucial for energy production, DNA repair, and cellular health [[134]43]. We speculate that enhanced NADP metabolic processes could ensure the maintenance of redox balance and produce the necessary biomolecules for energy, which is essential for sustaining the energy-intensive processes of labor and delivery. The steroid hormone biosynthesis pathway was identified for integrative analysis in cord blood. Steroid hormones, including cortisol and other glucocorticoids, play crucial roles in managing stress, regulating metabolism, and supporting fetal development [[135]32]. The elevated levels of steroid hormone biosynthesis in cord blood during VD may reflect the heightened hormonal activity required to manage the physiological stress of labor and support the newborn’s adaptation to extrauterine life. Our study has several strengths. Firstly, the prospective cohort design minimized the potential for reverse causality. Secondly, we included an intrapartum cesarean section group, providing an opportunity to explore the nuanced middle ground between VD and CS using a multiomics approach with both amniotic fluid and cord blood samples. Thirdly, the integrated omics analysis of proteins and metabolites enabled us to comprehensively investigate the mechanisms of different modes of delivery and related clinical traits. Nevertheless, it is important to acknowledge the limitations of this study. First, the sample size was small, warranting further replication, especially for the intrapartum cesarean section group. Second, we used an untargeted approach to detect the proteins and metabolites in amniotic fluid and cord blood samples. This method can lead to ambiguity in compound identification, as multiple low-molecular-weight substances may correspond to the same m/z value. The present study did not directly confirm these findings in a different population using a targeted method such as multiple reaction monitoring (MRM) or ELISA-based assays. This validation step will be important for confirming the reproducibility of our results across different cohorts. Third,, the results are based on a homogeneous population of women recruited at a single center. Future studies in a more diverse population would be needed to further test the generalizability of our findings. Conclusions Our study did not find any protein or metabolite pathways enriched for immune development between different delivery modes, suggesting that other pathways, such as the microbiome, may underlying the changed risk of allergic diseases in offspring. However, the differentially expressed steroid hormones and actin cytoskeleton pathway highlights that proteomics and metabolomics in amniotic fluid and cord blood were more focused on processes related to delivery itself, such as the length of labor. These findings, if reproducible in other populations, could provide the basis for mechanistic studies targeting specific biochemical pathways implicated in the process of delivery. Electronic supplementary material Below is the link to the electronic supplementary material. [136]Supplementary Material 1^ (1.4MB, docx) [137]Supplementary Material 2^ (7MB, xlsx) Acknowledgements