Pregnant women with regular peripartum coverage at the Obstetrics Outpatient Clinic of the Third Affiliated Hospital of Zhengzhou University from March 2024 to June 2024 were selected; 30 pregnant women with hypothyroid and 30 healthy pregnant women who met the inclusion criteria were recruited and included in the hypothyroidism group (H) and control group (C), respectively.

Inclusion criteria: 1) all pregnant women were ≤20 weeks of gestation; 2) the thyroid function of pregnant women in the hypothyroidism group was in accordance with the 2022 Guidelines for the Prevention and Management of Thyroid Diseases in Pregnancy and Childbirth and the reference range for hypothyroidism in pregnancy (TSH >4.0 mIU/L) of the Clinical Laboratory of the Third Affiliated Hospital of Zhengzhou University. Pregnant women in the control group had normal thyroid function and no other obstetric complications.

Exclusion criteria: 1) age <18 years or ≥35 years; 2) combined with other pregnancy complications; 3) artificial insemination, assisted reproduction technology/assisted conception; 4) twin and above pregnancies; 5) severe anxiety or depression; 6) currently taking antithyroid medication or thyroid medication substitution; 7) severe systemic diseases such as liver diseases or malignant tumors; 8) current infection or history of combined chronic inflammation and other autoimmune diseases; 9) pregnancy combined with hyperlipidemia or other comorbidities affecting the metabolic level of the body.

Ethical statement: This study was approved by the Medical Ethics Committee of the Third Affiliated Hospital of Zhengzhou University (Ethics No. 2024–284). All enrolled participants were aware of the risks and rights associated with the experiment, participated voluntarily, and signed informed consent forms.

Blood samples were collected from all pregnant women on the day of enrollment. All pregnant women fasted for 8–12 h before sample collection. Approximately 5 mL of blood was drawn from the elbow vein using a sterile syringe needle and placed in ethylenediaminetetraacetic acid (EDTA) and sodium heparin anticoagulant tubes (2.5ml each). Randomly selected 10 women from both group and collected their cord blood using EDTA tubes (3ml) during birth. Immediately after sampling, the blood samples were stored in a refrigerator at 4 °C. The blood samples in EDTA anticoagulated tubes were centrifuged (4 °C, 2000 rpm, 10 min) within 2 h and after centrifugation, the upper serum layer was aspirated using a sterile pasteurized pipette, placed in a freezing tube, and stored in a -80 °C refrigerator.Maternal blood collected in EDTA anticoagulant tubes was used for metabolomic and proteomic analyses, while umbilical cord blood was used for ELISA validation. Maternal blood collected in sodium heparin tubes was utilized for flow cytometry and cytokine experiments.

Data on gestational week, age, body mass index (BMI), serum-free T4 (FT4), thyrotropin (TSH), antithyroid peroxidase antibodies (TPOAb), C-reactive protein (CRP), fasting blood glucose, and hemoglobin levels were collected from all pregnant women at enrollment. Upon enrollment, patients in the hypothyroidism group first provided blood samples and clinical data. In accordance with clinical guidelines and ethical guidelines, they then subsequently received oral levothyroxine therapy. Patients were monitored through follow-up, and clinical data on pregnancy-related comorbidities and delivery complications were collected. This included the incidence of conditions such as gestational diabetes mellitus, preterm labor, gestational hypertension, gestational anemia, and intrahepatic cholestasis during pregnancy.

For all participants in this study (30 pregnant women in the hypothyroidism group and 30 in the control group) maternal peripheral blood was collected in EDTA anticoagulant tubes at enrollment for metabolomic analysis. A Waters UPLC I-Class Plus (Waters, USA) in tandem with a Q Exactive high-resolution mass spectrometer (Thermo Fisher Scientific, USA) was used for metabolite separation and detection.

Chromatographic conditions: The chromatographic column used was a BEH C18 column (1.7 μm 2.1*100 mm, Waters, USA). The mobile phases in positive ionization mode were an aqueous solution containing 0.1% formic acid (liquid A) and methanol containing 0.1% formic acid (liquid B), and the mobile phases in negative ionization mode were aqueous solutions containing 10 mM ammonium formate (liquid A) and 95% methanol containing 10 mM ammonium formate (liquid B).

Mass spectrometry conditions: A Q Exactive mass spectrometer (Thermo Fisher Scientific, USA) was used for primary and secondary mass spectrometry data acquisition. The mass spectrometry was performed in the range of 70~1050, with a primary resolution of 70, 000, AGC of 3e6, and a maximum injection time (IT) of 100 ms. In accordance with the parent ion intensity, Top3 was selected for fragmentation, and the secondary information was collected with a secondary resolution of 17, 500, AGC of 1e5, a maximum IT of 50 ms, and fragmentation energy (IT, fragmentation energy) of 50 ms. The secondary resolution was 17, 500, the AGC was 1e5, the maximum IT was 50 ms, and the fragmentation energy (stepped once) was set to 20, 40, and 60 eV. The ion source parameters were set to 40 for the sheath gas flow rate, 10 for the auxiliary gas flow rate, and 10 for the spray voltage (spray voltage [|KV|]). |KV| was set to 3.80 for positive ion mode and 3.20 for negative ion mode. The capillary temp was set to 320 °C, and the aux gas heater temp was set to 350 °C. The parameters were set as follows: sheath gas flow rate at 40, aux gas flow rate at 10, |KV| at 3.80 for positive ion mode and 3.20 for negative ion mode, capillary temp at 320 °C, and aux gas heater temp at 350 °C.

Mass spectrometry data were imported into Compound Discoverer 3.3 (Thermo Fisher Scientific, USA) software and analyzed in conjunction with the BMDB (UW Metabolome Database, BGI Metabolome Database), mzCloud database, and ChemSpider online database. After analyzing the mass spectrometry data, a data matrix containing information on metabolite peak areas and identification results was obtained, and the table was further analyzed and processed.

Eighteen patients from each group were randomly selected for proteomic analysis. The random selection procedure was conducted as follows: During the screening within the hypothyroidism group, a statistical analysis of the enrolled cohort was performed using Excel and assigned sequential identification numbers according to the order of enrollment. The RAND function was used to generate unique random numbers between 0 and 1, which were assigned to all participants in the hypothyroidism group. These random numbers were then sorted, leading to the selection of 18 corresponding random values. The INDEX function was subsequently applied in combination with the RAND function to display the identification numbers associated with the selected random values in Excel. The same procedure was used to randomly select 18 patients from the control group.Mass spectrometry data were collected for 36 samples using a TimsTOF Pro instrument in the data-independent acquisition (DIA) mode. The experimental procedure was as follows: protein extraction and enrichment were performed using a C18 column after extraction. After enrichment and quality control, an equal amount of peptide was extracted from each sample and mixed into one tube with 20 μg of peptide. A Shimadzu LC-20AD liquid chromatography system was used to take the mixed samples. Peptide separation was carried out using a 5 μm, 20 cm x 180 μm Gemini C18 column as a separator, and further separation was carried out using a 5 μm 4.6 x 250 mm Gemini C18 column as a separator. High-pH RP separation was performed to obtain DDA mode data. Downstream DDA data were analyzed using MaxQuant’s integrated Andromeda engine to create a spectral library. For large-scale DIA data, the mProphet algorithm was used to analyze the data for quality control. After completing the correction, MSstats derived from the bioconductor repository were used for difference analysis. Differential proteins were screened according to the criteria of multiplicity of differences >1.2 and Qvalue <0.05 as significant differences, and a series of functional analyses were performed on the differentially expressed proteins.

Eighteen pregnant women with hypothyroidism and ≤20 weeks gestation and 18 normal pregnant women were selected as the validation cohort, and blood samples from the validation cohort were collected and stored using the same procedure. The validation cohort does not overlap with the main study cohort, and shares the same inclusion and exclusion criteria. From the main study cohort, ten pregnant women with hypothyroidism who delivered at our hospital were randomly selected to provide umbilical cord blood at delivery, using the same randomization method as applied in the proteomics screening. The same selection procedure was used for the control group participants.

The cord blood samples and validation cohort’s blood samples were thawed at room temperature and subsequently validated for selected metabolites and proteins using ELISA. The ELISA kits were from MEIMIAN Crop (Jiangsu, China). Human sphingosine ELISA kit (Catalog: MM-61057H1, sensitivity 120ng/L-4400ng/L) and Human BH3 interacting domain death agonist ELISA kit (Catalog: MM-0379H1, sensitivity 3pg/ml-150pg/ml) were used. The serum was diluted within the linear range of each assay following the manufacturer’s recommendations. The absorbance was measured at 450 nm using a Bioteck EPOCH enzyme-labeling instrument. Standard curves were drawn, and protein concentrations were calculated using ELISACALC software.

At enrollment, maternal peripheral blood was collected in sodium heparin tubes from all participants in both the hypothyroidism group (n=30) and the control group (n=30). Th17 was quantitatively measured using intracellular cytokine staining: after incubation at 37 °C 5% CO2 for 4 h the cell surface labeling antibody APC-CY7-CD3 (Clone:SP34-2) with FITC-CD4 (Clone:RPA-T4) was added. After incubation for 20 min in dark, erythrocyte lysate was added and incubated further for 25 min. Subsequently, centrifugation was performed at 500 g at 4 °C, and the supernatant was discarded. A fixed membrane-breaking solution was added, and the cells were incubated for 30 min in the dark. After washing, the APC-IL17 (Clone:SCPL1362) antibody was added for intracellular staining, and the cells were incubated for 30 min in the dark. Th17 cells were defined as CD3⁺CD4⁺IL-17⁺ lymphocytes. Treg cells were quantitatively measured using cell surface staining, and cell surface-labeling antibodies APC-CY7-CD3, FITC-CD4, APC-CD127 (Clone:HIL-7R-M21), and PE-CD25 (Clone:M-A251) were added to the peripheral blood and incubated in the dark. Erythrocyte lysates were incubated, centrifuged, washed, and resuspended in phosphate buffer. Treg cells were defined as CD3⁺CD4⁺CD25+CD127- lymphocytes. The samples were analyzed using flow cytometry. Data analysis was performed using the FlowJo software (Tree Star, Ashland, OR, USA). All antibodies were purchased from BD Biosciences (Franklin Lakes, NJ, USA).

The levels of IL-2, IL-6, IL-10, and TNF-a were measured using a Human Cytokine Kit (Jiangxi Sage Biotechnology Co., Ltd.) based on the flow fluorescence technique. The kit contains microspheres encapsulated with antibodies specific to IL-2, IL-6, IL-10, and tumor necrosis factor-alpha (TNF-α). The capture microsphere mixture was mixed with 25 μL of serum, and then the mixture was incubated with 25 μL of fluorescently labeled detection antibody for 2.5 h at 20–25 °C in the dark. The beads were washed and resuspended in phosphate buffer saline, and the samples were analyzed using flow cytometry.

Detection limits of the Kit is as followed: IL-2(0.2-5000pg/ml), IL-6(2.4-5000pg/ml), IL-10(1.22-5000pg/ml), TNF-α(0.5-5000pg/ml).

All statistical analyses of metabolomic and proteomic data were performed using metaX (21). Graphpad Prism 8.2 (GraphPad, La Jolla, CA, USA) was used for the statistical analysis of clinical indicators and flow cytometry. Measurement information that conformed to normal distribution was described as mean ± standard deviation, and a t-test was used for comparison between groups. Measures that did not meet the normal distribution were described as medians and quartiles, and the Mann–Whitney t-test was used for comparisons between groups. Categorical variables are described as frequencies, and differences between groups were determined using the chi-square test. Correlations were analyzed using Spearman’s analysis. Significance was set at P < 0.05.

Overall, 30 pregnant women in the hypothyroidism group and 30 pregnant women in the control group were included in this study. As shown in Table 1, no significant differences were observed between the hypothyroidism and control groups in terms of age, BMI, gestational week, hemoglobin level, and fasting blood glucose level (P >0.05).

The metabolite categorization loop (Figure 1A) showed that lipids accounted for 25.25% of all detected metabolites, making them the metabolite species with the largest percentage. Partial least squares discriminant analysis (PLS-DA) showed that the hypothyroidism group could be separated from the control group, indicating differences in metabolic profiles between the two groups (Figure 1B). Fold change (FC) ≥1.2, VIP ≥1, and P value <0.05 were used as parameters to screen the metabolites and plot the volcano diagrams, and the results shown in Figure 1C indicated a total of 183 differential metabolites between the two groups, with 75 metabolites significantly upregulated in the hypothyroidism group, 108 metabolites significantly downregulated, and sphingosine (Sph) was significantly upregulated.The permutation test results show no signs of overfitting (Figure 1D).

As shown in Figure 2A, the PLS-DA plot showed only a small overlap between the hypothyroidism and control groups, with most samples being separated, suggesting that the proteomic characteristics of the hypothyroidism group differed from those of the control group. FC ≥1.2 and P-value <0.05 were the conditions for the screening of differential proteins. Subsequently, a volcano plot of the differentially expressed proteins was drawn, and the results are shown in Figure 2B. Seventy-nine differentially expressed proteins were identified between the two groups, with 35 significantly upregulated, 44 significantly downregulated, and BH3 interacting domain death agonist (BID) significantly upregulated.

Combined metabolomics and proteomics were performed for the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (Figure 3A), and the results showed that the sphingolipid signaling pathway (SSP) and ameobiasis pathway were significantly enriched in both histologies. Combined with the biological significance, the SSP was selected to draw a multi-omics pathway co-annotation map. As shown in Figure 3B, the upregulated differential metabolite Sph and the differential protein BID were co-enriched in the SSP, suggesting a regulatory relationship between Sph and BID. In addition, Figure 3A shows that the differential metabolites were significantly enriched in the necroptosis and apoptosis pathways.

As shown in Figure 4, Sph and BID levels in the serum samples of the validation cohort and the cord blood samples were measured using ELISA, and both Sph and BID levels were significantly higher in the hypothyroidism group than in the control group.

Peripheral blood Th17 and Treg cells were detected in both groups of pregnant women using flow cytometry. The gating strategy is shown in Figure 5A. Lymphocytes co-expressing CD3 and CD4 were identified as CD3⁺CD4⁺ T cells. Th17 cells were defined as IL-17⁺ cells within the CD3⁺CD4⁺ T cell population, whereas Treg cells were identified as CD25⁺CD127^- cells within the same parent population. As shown in (Figures 5B, C) and Table 2, TSH, TPOAb, CRP, percentage of Th17 cells, Th17/Treg ratio, IL-2, IL-6, IL-10 and TNFα levels increased significantly in the hypothyroidism group compared to the control group, and FT4 levels and percentage of Treg cells decreased significantly.

As shown in Table 3, the probability of developing gestational diabetes mellitus(GDM) in the hypothyroidism group was significantly higher than that in the control group, and the probabilities of developing gestational hypertension, preterm delivery, anemia, intrahepatic cholestasis of pregnancy and placental abruption during pregnancy were not significantly different from those of the control group. Neonatal weight in the hypothyroidism group was significantly lower than in the control group. There were no significant differences between the two groups in the incidence of intrauterine growth restriction or neonatal length.

Correlations between the potential biomarkers screened against the flow cytometry results, cytokine levels, and pregnancy complication were statistically analyzed using Spearman correlation analysis. As shown in Figure 6; Table 4, BID is positively correlated with CRP. Sph and BID positively correlated with Th17/Treg ratio, and Sph positively correlated with Th17, IL-6 levels and gestational diabetes.