Abstract Background The hyperglycemia in pregnancy (HIP) is classified as the gestational diabetes mellitus (GDM) and pre-gestational diabetes mellitus (PGDM). Metabolomic changes during pregnancy have been suggested to underlie the etiology of HIP, which influence fetal organogenesis and placental development. On one hand, the PGDM women suffer from hyperglycemic exposure in earlier gestation than GDM. On the other, type 2 diabetes mellitus (T2DM) is the most common type of PGDM. Therefore, it is of clinical implication to detect the metabolic alterations of T2DM pregnant women, especially for those with refractory and uncontrolled hyperglycemia. We aimed to figure out the metabolic profile of umbilical cord blood among pregnant women with T2DM and its influences on the fetal growth. Methods We included 48 uncontrolled T2DM singleton pregnancies and 52 matched healthy singleton pregnancies. Metabolites in umbilical cord blood were measured by liquid chromatography‒mass spectrometry analysis. Results We observed significant increases in the levels of several metabolites in umbilical cord blood samples from patients with uncontrolled T2DM, including alisol A, glycodeoxycholic acid, 3-quinolinecarboxylic acid- 7,8-dichloro-1,4-dihydro-4-oxo-, fraxetin, sphingosine, alpha-D-glucose, L-glutamine, isoalantolactone, L-leucine, and voriconazole. The abundance of alisol A was negatively correlated with the level of glucose in umbilical cord blood, and the abundance of L-leucine was positively correlated with the level of insulin in umbilical cord blood. The pathways of biosynthesis of amino acids and butanoate metabolism were enriched in the uncontrolled T2DM group. Further, the abundance of alisol A and L-leucine were positively correlated with the fetal weight in the uncontrolled T2DM group. Conclusion The metabolic profile of umbilical cord blood among pregnant women with uncontrolled T2DM was different from that of the healthy pregnant women. The metabolic profile of umbilical cord blood from the uncontrolled T2DM women was characterized by the enrichment of the pathways of biosynthesis of amino acids and butanoate metabolism and the metabolites of alisol A and L-leucine. Keywords: Metabolomics, Type 2 diabetes mellitus, Umbilical cord blood, Pregnant women __________________________________________________________________ what is known. Uncontrolled T2DM is characterized by a high incidence of unfavorable fetal outcomes, such as birth defects and perinatal morbidity [[30]1, [31]2]. Early studies about metabolomics in GDM indicated that the fluctuations of metabolites, including fatty acid, amino acid and glucose, had been detected in the cord blood of GDM pregnant women [[32]3]. The disturbed metabolism in the cord blood not only reflects the simultaneous alterations of metabolites in fetus, but also indicates potential threats to fetal growth and development [[33]4]. Although T2DM is associated with higher rates of unhealthy infants because of the influences on the maternal-fetal dyad by HIP in an earlier gestational age, few studies concerning metabolomics have been conducted among the T2DM mothers, especially uncontrolled T2DM. what is new. We found the levels of 140 metabolites fluctuated significantly in umbilical cord blood of uncontrolled T2DM pregnant women compared with the healthy mothers, such as alisol A, glycodeoxycholic acid, 3-quinolinecarboxylic acid- 7,8-dichloro-1,4-dihydro-4-oxo-, fraxetin, sphingosine, alpha-D-glucose, L-glutamine, isoalantolactone, L-leucine, voriconazole, Met-Ser-Arg, artemisinin, deoxypeganine, D-erythro-imidazolylglycerol phosphate, and 6-hydroxymelatonin. The pathways of biosynthesis of amino acids and butanoate metabolism were enriched in the uncontrolled T2DM group. The levels of alisol A and L-leucine were positively correlated with the fetal weight in the uncontrolled T2DM group. Introduction Hyperglycemia is one of the commonest medical conditions in metabolism affecting pregnant women, which has an increasing incidence during the recent decades [[34]5]. Worldwidely, in 2021, approximately 17% of live births were affected by hyperglycemia in pregnancy (HIP) [[35]6]. The pregnant women with diabetes before pregnancy would be diagnosed as pre-gestational diabetes mellitus (PGDM), when the diabetes mellitus attacks before pregnancy and persists during the whole gestation. In contrast, when the de novo hyperglycemia occurs during pregnancy, the gestational diabetes mellitus (GDM) would be diagnosed. Accordingly, there are 16% and 84% women with HIP are classified as PGDM and GDM, respectively [[36]7]. As GDM accounts for the majority of HIP, previous studies in HIP mainly involved GDM. However, metabolomic changes during pregnancy have been suggested to underlie the etiology of HIP, which influence fetal organogenesis and placental development. And the earlier hyperglycemic exposure is associated with an elevated risk of serious complications, such as macrosomia, neonatal hypoglycemia, and preterm delivery [[37]8]. Additionally, compared with GDM, PGDM is associated with a higher incidence rate of perinatal adverse events, for instance cesarean section and preeclampsia [[38]8]. Because type 2 diabetes mellitus (T2DM) is the most common type of PGDM, we focus on PGDM patients with T2DM and use the abbreviation T2DM to refer to PGDM in the present study. Metabolomics is a technology to detect small-molecule chemical entities involved in metabolism, which is frequently used for disease diagnosis and prediction, as well as to explore the active drivers of biological processes [[39]9]. On one hand, HIP influences the nutrient metabolism of pregnant women, which would transmit to the fetus through the umbilical cord and placenta. On the other hand, the disturbed metabolic condition influences the development of fetus. Therefore, the metabolome has been frequently utilized in maternal-fetal medicine to identify biological changes occurring in the fetus [[40]4]. A serial of studies have conducted to record the alterations in metabolism and its influences on fetus in pregnant women with HIP [[41]10–[42]13]. For instance, studies have suggested that HIP could disturb the maternal as well as fetal metabolism. For T2DM pregnant women, the uncontrolled hyperglycemia during pregnancy not only influenced the metabolic condition of maternal-fetus, but also enhanced the risks of increased birthweight, congenital malformation, and even perinatal mortality [[43]1, [44]2]. Therefore, it is of clinical implication to detect the metabolic alterations of T2DM pregnant women, especially for those with refractory and uncontrolled hyperglycemia. In this study, we utilized untargeted metabolomics to meticulously examine umbilical cord blood samples and discern distinctive metabolic patterns between pregnant women with uncontrolled T2DM and the matched healthy cases without diabetes. Additionally, we investigated the correlation between the alterations of metabolites and fetal birth weight under different maternal blood glucose patterns. Methods and materials Subjects This study was approved by the Peking University First Hospital review board (2021[488]). All participants provided informed consent before participating. Initially, the study involved 52 singleton pregnancies affected by uncontrolled T2DM. In order to exclude the disturbs on metabolic condition beyond uncontrolled T2DM, we included 52 healthy pregnant women with normal glycemia and matched the factors including the body mass index (BMI) at pre-pregnancy, the BMI at late pregnancy, gestational age and the birth weight of the fetus. Four uncontrolled T2DM cases were further excluded from the study for the umbilical cord blood less than 100 µl and failed to have metabolomics. Finally, the present study comprised 48 cases of uncontrolled T2DM and 52 healthy controls (Fig. [45]1). The 48 pregnant women with uncontrolled T2DM were managed with insulin. The T2DM was diagnosed according to the published literature [[46]14]. All pregnant women with T2DM were treated with insulin at 33–78 units per day. The normal range was 2.6–24.9 µU/ml for insulin in maternal serum. We defined Hemoglobin A1c (HbA1c) more than 6.0% and glycated albumin (GA) more than 16.0% as uncontrolled glycemia [[47]15–[48]17]. HbA1c and GA were detected three times during early (before 14 week), middle (14-28week) and late (after 28 week) pregnancy respectively. All of the above participants came from the pregnant women who registered at the beginning of pregnancy, had monthly follow-up at outpatient department, and were admitted for inpatient department for prenatal care and delivery at Peking University First Hospital. The participants were recruited at the time of delivery when the uncontrolled T2DM was documented during pregnancy or the matched factors were met for the control group. We excluded patients contemporarily have other perinatal adverse events, including hypertension during pregnancy, cardiovascular diseases, autoimmune disorders, infection, congenital fetal diseases and birth abnormalities. Because the process of labor delivery and its associated stress might influence nutrient transport and component in the umbilical cord to some extent, we only included pregnant women with elective cesarean delivery [[49]18]. The indications for cesarean delivery of the T2DM group were: primipara breech presentation (20 cases, 41.7%), prior cesarean section (19 cases, 39.5%), failure of labor induction (7 cases, 14.6%), and funnel-shaped pelvis (2 cases, 4.2%). Meanwhile, the indications for cesarean delivery in the normal control group were: primipara breech presentation (32 cases, 61.6%), prior cesarean section (14 cases, 26.9%), failure of labor induction (4 cases, 7.7%), high myopia (1 case, 1.9%), and funnel-shaped pelvis (1 case, 1.9%). Fig. 1. [50]Fig. 1 [51]Open in a new tab The enrollment flow of participants. a. The definition of uncontrolled T2DM: Hemoglobin A1c > 6.0% and glycated albumin > 16.0% during early (before 14 week), middle (14-28week) and late (after 28 week) pregnancy respectively. b. At least 100 µl umbilical cord blood was necessary for the untargeted metabolomics Metabolite extractions The process begins with the collection of umbilical cord blood plasma samples. Metabolites are then extracted using a cold solvent extraction method, typically employing methanol, acetonitrile, and water [[52]19, [53]20]. Following extraction, the samples undergo vacuum centrifugation to remove impurities. Subsequently, liquid chromatography-mass spectrometry analysis is conducted to identify and quantify the metabolites present. Finally, the samples are reconstituted and stored in LC vials for further analysis or storage. Liquid chromatography‒mass spectrometry analysis Untargeted metabolomics analysis of polar metabolites extracted from samples was performed using hydrophilic interaction chromatography coupled to a quadrupole time-of-flight mass spectrometer (Sciex TripleTOF 6600). The gradient protocol consisted of an initial 1-minute run with 85% acetonitrile, which was linearly reduced to 65% over 11 min. It was further reduced to 40% over 0.1 min and maintained for 4 min before being increased to 85% over 0.1 min, followed by a 5-minute re-equilibration period. The flow rate was set at 0.4 mL/minute, with the column temperature and auto-sampler temperature maintained at 25 °C and 5 °C, respectively. A 2 µL injection volume was employed for the analysis. The ionization modes utilized are both negative and positive. Mass spectrometry acquisition involves scanning the m/z range from 60 to 1000 Da, with a TOF MS scan accumulation time of 0.20 s per spectra. Data analysis Peak grouping parameters, including a bandwidth of 5, an m/z width of 0.025, and a minimum fraction of 0.5, are applied during data processing. Retention criteria for extracted ion features require more than 50% non-zero measurement values in at least one group. Metabolite identification involves matching MS/MS spectra to an in-house database of authentic standards. Subsequently, multivariate data analysis was conducted using SIMCA-P (version 14.1, Umetrics, Umea, Sweden), including Orthogonal Partial Least Squares-Discriminant Analysis (OPLS-DA). Model evaluation was conducted through seven-fold cross-validation and response permutation testing. Variable importance assessment utilized the Variable Importance in Projection (VIP) value. Statistical significance was determined using an unpaired Student’s t-test (VIP > 1, p < 0.05). To measure the strength and direction of the linear relationship between metabolite concentration and fetal birth weight, we computed the Pearson correlation coefficient for each sample. This statistical measure allowed us to quantify the extent of their association. Fold change (FC) was calculated by the the abundance of meatabloties in the umbilical cord blood from the pregnant women with uncontrolled T2DM in contrast to that from the control group. Bioinformatics analysis We performed Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation by matching metabolites against the online KEGG database to retrieve their compound IDs. IDs are mapped to pathways in the KEGG database, followed by pathway extraction. KEGG pathway enrichment analysis is then conducted using Fisher’s exact test methodology. The significance threshold is set at p < 0.05 for identifying enriched pathways. Statistical analysis Statistical analysis was performed using IBM SPSS Statistics v.20.0 (IBM Corp., Armonk, NY, USA). The normality of continuous variables was assessed using the Kolmogorov-Smirnov test. Normally distributed continuous variables were presented as means and standard deviations, while non-normally distributed continuous variables were presented as medians. Categorical variables and ranked variables were presented as frequency and percentile. Student t-test was used to compare differences for normally distributed continuous variables, and the Mann-Whitney U test was used to compare differences for non-normally distributed continuous variables. The chi-square test was used for categorical variables to determine differences. Binary multivariable logistic regression analysis was conducted using the forward stepwise method. A significance threshold of p < 0.05 was applied for determining statistical significance. Results Characteristics of the study cohorts Metabolomics analysis was performed on umbilical cord blood samples collected from 48 pregnant women with uncontrolled T2DM and 52 healthy controls with normal glucose tolerance (Fig. [54]1; Table [55]1). The mean birth weights were 3388 g ± 453 g for the uncontrolled T2DM group and 3345 g ± 338 g for the normal control group, with no significant difference (p = 0.176). Additionally, there were no significant differences in the pre-pregnant BMI, the late-pregnant BMI, or gestational age between the uncontrolled T2DM and normal mothers. Table 1. Demographic characteristics of T2DM and control group Characteristics Uncontrolled T2DM (N = 48) Control (N = 52) P value Gender, N (%) Male Female 24 (50) 24 (50) 21 (40) 31 (60) 0.392^a Birth weight (g), Means ± SDs 3388 ± 453 3345 ± 338 0.176^b Gestational age (weeks), Median 38 39 0.109^c Late-pregnancy BMI (kg/m^2), Means ± SDs 28.9 ± 5.0 28.6 ± 3.5 0.860^b Late-pregnancy BMI category, N (%) Lean/normal Overweight/obese 8(17) 40(83) 6(12) 46(88) 0.688^a Pre-pregnancy BMI (kg/m^2), Means ± SDs 25.0 ± 5.0 24.3 ± 3.7 0.501^b Pre-pregnancy BMI category, N (%) Lean/normal Overweight/obese 22(46) 26(54) 28(54) 24(46) 0.277^a HbA1c (%), Means ± SDs 6.5 ± 0.3 5.1 ± 0.4 0.001^b GA (%), Means ± SDs 17.0 ± 1.3 13.1 ± 1.9 0.001^b [56]Open in a new tab p value refers to a Chi-square test, b Two-sample t test, and c Mann–Whitney U test BMI = body mass index; HbA1c = glycated hemoglobin; GA = glycated albumin Untargeted metabolomics analysis for the metabolites in the umbilical cord blood Based on untargeted liquid chromatography‒mass spectrometry, we detected 972 metabolites in the umbilical cord blood samples, including 588 metabolites in negative ion mode and 384 metabolites in positive ion mode. To identify differential metabolites between the two groups from the comprehensive set of detected metabolites, we employed OPLS-DA to establish a discriminant model. Through this approach, the metabolic composition of umbilical cord blood in pregnant women with uncontrolled T2DM was distinctly separated from the healthy control group (Fig. [57]2a): in negative ion mode, R2Y = 0.768 and Q2 = 0.593; in positive ion mode, R2Y = 0.936 and Q2 = 0.665. It showed that this model had good predictive performance [[58]21]. Among the 972 metabolites, 140 metabolites had significant differences between the two groups (see methods and materials), including 78 metabolites in negative ion mode and 62 metabolites in positive ion mode. With regard to the unsupervised hierarchical clustering analysis, these 140 metabolites exhibited clear separation between the uncontrolled T2DM group and the control group (Fig. [59]2b). The pathways of mineral absorption, biosynthesis of amino acids, central carbon metabolism in cancer, steroid hormone biosynthesis, ABC transporters, and butanoate metabolism ranked in the top five among the altered pathways between the umbilical cord blood of pregnant women with uncontrolled T2DM and the controls (Table [60]2). Fig. 2. [61]Fig. 2 [62]Open in a new tab Liquid chromatography‒mass spectrometry analysis. a. Score plot of the OPLS-DA model. Green squares denote cases, and blue circles denote controls. Negative ion mode, R2Y = 0.768 and Q2 = 0.593. Positive ion mode, R2Y = 0.936 and Q2 = 0.665. b. Heatmap with unsupervised hierarchical clustering analysis of the differential metabolites. The different colors represent different intensities, of which blue indicates low intensity and red indicates high intensity. c. The bar chart represents the fold change difference. The red bar charts at right denote the increased metabolites in the T2DM group. The green bar charts at left denote the decreased metabolites in the T2DM group. FC = fold change; T2DM = type 2 diabetes mellitus; NC = normal control Table 2. Pathways and metabolites of the pathway enrichment analysis Pathway P value FDR Hits Metabolites Mineral absorption 0.015 0.002 3 L-Threonine, Phosphoric acid, L-Glutamine Biosynthesis of amino acids 0.028 0.012 6 S-Adenosylmethionine, L-Threonine, S-adenosyl-l-homocysteine, Histidine, L-Glutamine, D-erythro-imidazolylglycerol phosphate Central carbon metabolism in cancer 0.029 0.005 3 Histidine, L- (+)-lactic acid, L-Glutamine Steroid hormone biosynthesis 0.034 0.129 5 Estrone glucuronide, 3-dehydroepiandrosterone sulfate, Pregnenolone sulfate, 5-androsten-3.beta.,16.alpha.-diol-17-one, Beta-estradiol ABC transporters 0.034 0.017 6 L-Threonine, Histidine, Phosphoric acid, L-Glutamine, Maltose, Betaine Butanoate metabolism 0.038 0.048 3 Succinic semialdehyde, Dl-malic acid, 3-hydroxybutyric acid [63]Open in a new tab FDR = False Discovery Rate According to the FC, the top 15 metabolites enriched in the umbilical cord blood of the uncontrolled T2DM group included 10 metabolites with the elevated levels and 5 metabolites with the decreased levels (Fig. [64]2c). The abundance of alisol A was the most obviously increased metabolite in the umbilical cord blood of the pregnant women with uncontrolled T2DM, followed by glycodeoxycholic acid, 3-quinolinecarboxylic acid- 7,8-dichloro-1,4-dihydro-4-oxo-, fraxetin, sphingosine, alpha-D-glucose, L-glutamine, isoalantolactone, L-leucine, and voriconazole. Meanwhile, the abundance of Met-Ser-Arg, artemisinin, deoxypeganine, D-erythro-imidazolylglycerol phosphate, and 6-hydroxymelatonin were obviously decreased in the umbilical cord blood of the pregnant women in uncontrolled T2DM group. Multivariable analysis for the levels of glucose, C-peptide, and insulin in umbilical cord blood In order to explore the association between metabolites in cord blood and glucose, insulin, and C-peptide levels under the condition of gestational hyperglycemia, we conducted multivariable logistic regression analysis in the T2DM group to figure out the independent influence factors for umbilical cord blood with glucose levels exceeding 6.1 mmol/L, insulin levels surpassing 24.9 µU/ml, and C-peptide levels surpassing 4.4 ng/ml, which were the upper limit of normal values for the corresponding examination. First, the basic model (model A) was built with the variables of gestational age and pre-pregnancy BMI. The cutoff values were 39 weeks for the gestational age and 24 kg/m^2 for the BMI in the model A, which were the median values of the whole study cohort. Then, we also selected the top 15 metabolites enriched in the uncontrolled T2DM group. Each of the 15 kinds of metabolites was added to model A to generate a new model. According to the outcomes of the multivariable analysis, the abundance of alisol A was negatively correlated with the level of glucose in umbilical cord blood (p = 0.041, OR [95%] = 0.504 [0.262–0.971])(Table [65]3). Meanwhile, the abundance of glycodeoxycholic acid and L-leucine were nearly statistically significant in its negatively correlated association with the level of glucose levels in umbilical cord blood without reaching statistical significance (p = 0.054 and 0.058) (Table [66]3). The abundance of L-leucine was positively correlated with the level of insulin in umbilical cord blood (p = 0.010, OR [95%] = 3.036 [1.306–7.058]) (Table [67]4). Table 3. Multivariable logistic regression analysis for the high level of glucose in umbilical cord blood Models OR 95% CI P Model A: gestational age + prepregnancy BMI gestational age^a 0.039 0.092–1.662 0.203 prepregnancy BMI^b 0.823 0.241–2.807 0.755 Model A + one of the following metabolites^c Alisol a 0.504 0.262–0.971 0.041 Glycodeoxycholic acid 0.537 0.285–1.011 0.054 3-quinolinecarboxylic acid, 7,8-dichloro-1,4-dihydro-4-oxo- 1.223 0.689–2.170 0.492 1 h-indole-3-propanoic acid 1.073 0.614–1.877 0.804 L-Glutamine 0.853 0.490–1.486 0.576 L-Leucine 0.557 0.304–1.019 0.058 Met-Ser-Arg 0.681 0.387–1.199 0.183 Artemisinin 0.794 0.453–1.392 0.420 Deoxypeganine 0.747 0.427–1.307 0.307 [68]Open in a new tab a. The gestational age was classified into two categories with a cutoff value of 39 weeks b. Prepregnancy BMI was classified into two categories with a cutoff value of 24 kg/m^2 c. The fold change values of the metabolites were classified into four categories by quartile BMI = body mass index; OR = odds ratio; CI = confidence interval Table 4. Multivariable logistic regression analysis for the high level of insulin in umbilical cord blood Models OR 95% CI P Model A: gestational age + prepregnancy BMI gestational age^a 0.420 0.097–1.812 0.245 prepregnancy BMI^b 1.514 0.429–5.350 0.519 Model A + one of the following metabolites^c Alisol a 0.946 0.523–1.711 0.855 Glycodeoxycholic acid 1.081 0.607–1.926 0.792 3-quinolinecarboxylic acid, 7,8-dichloro-1,4-dihydro-4-oxo- 0.960 0.543–1.699 0.890 1 h-indole-3-propanoic acid 0.794 0.449–1.404 0.428 L-Glutamine 1.365 0.764–2.437 0.293 L-Leucine 3.036 1.306–7.058 0.010 Met-Ser-Arg 1.214 0.689–2.141 0.502 Artemisinin 0.843 0.475–1.494 0.558 Deoxypeganine 1.271 0.722–2.238 0.406 [69]Open in a new tab a. The gestational age was classified into two categories with a cutoff value of 39 weeks b. Prepregnancy BMI was classified into two categories with a cutoff value of 24 kg/m^2 c. The fold change values of the metabolites were classified into four categories by quartile BMI = body mass index; OR = odds ratio; CI = confidence interval Correlations between the levels of metabolites and fetal weight We further analyzed the correlations between the abundance of alisol A, L-leucine, and glycodeoxycholic acid in umbilical cord blood and fetal weight. It was found that the abundance of alisol A had a positive correlation with fetal weight in the uncontrolled T2DM group (r = 0.33, p = 0.024), but no significant correlation was observed in the normal control group (r = 0.16, p = 0.326) (Fig. [70]3a). Additionally, in the uncontrolled T2DM group, positive correlations were observed between the abundance of L-leucine and fetal weight (r = 0.32, p = 0.026) (Fig. [71]3b). However, in normal group, L-leucine was not obviously associated with fetal weight (r = 0.20, p = 0.153). Glycodeoxycholic acid had no relationship with fetal weight in both two groups (Fig. [72]3c). These findings suggest that the relationship between metabolism and fetal weight may differ between individuals with uncontrolled T2DM and those who are healthy. Fig. 3. [73]Fig. 3 [74]Open in a new tab Correlations between the levels of metabolites and Fetal Weight in T2DM and NC groups. The correlations between the levels of alisol A (a), L-leucine (b), and glycodeoxycholic acid (c) in umbilical cord blood and fetal weight in T2DM and NC groups. T2DM = type 2 diabetes mellitus; NC = normal control Discussion In our study, we examined different metabolites in umbilical cord blood between uncontrolled T2DM pregnant women and normal mothers. We found that the levels of 78 metabolites in negative ion mode and 62 metabolites in positive ion mode differed significantly between the uncontrolled T2DM and normal groups. According to the outcomes of the multivariable analysis, alisol A was negatively correlated with the level of glucose in umbilical cord blood, and L-leucine was positively correlated with the level of insulin in umbilical cord blood. The pathways of mineral absorption, biosynthesis of amino acids, central carbon metabolism in cancer, steroid hormone biosynthesis, ABC transporters and butanoate metabolism pathways were enriched in the uncontrolled T2DM group. In the uncontrolled T2DM group, the amount of alisol A and L-leucine in umbilical cord blood were positively correlated with fetal weight. Alisol A, belonging to protostane-type tetracyclic triterpenoid, serves as one of the main components in Alismatis Rhizoma [[75]22]. This study demonstrates a significant correlation between alisol A and umbilical cord blood glucose levels. Ho et al. discovered that alisol A can markedly reduce body weight and abdominal fat content in obese mice induced by a high-fat diet, while also decreasing elevated insulin levels caused by the high-fat diet, thereby improving glycol metabolism and insulin sensitivity [[76]23]. The study by Xu et al. highlights the efficacy of alisol-based compounds, especially alisol A, in modulating lipoprotein lipase activity to lower triglyceride levels [[77]24]. Alisol A shows promise in attenuating obesity induced by a high-fat diet, inhibiting hepatic steatosis, and improving both lipid and glucose metabolism in mice [[78]23]. In 1971, Pedersen proposed the “hyperglycemia-hyperinsulinemia theory” [[79]25], which states that hyperglycemia in women with diabetes mellitus has the potential to trigger hyperglycemia, pancreatic β-cell hypertrophy, hyperinsulinemia, organ immaturity, and excessive weight gain in the fetus. However, in our study, the fetuses were not large-for-dates infants. This phenomenon may be closely related to the elevated level of alisol A. Our results indicated that L-leucine, which was classified as the branched-chain amino acids (BCAAs) [[80]26], was increased in the cord blood of the uncontrolled T2DM group. Furthermore, L-leucine was positively correlated with the levels of insulin in cord blood in the uncontrolled T2DM group. In cord blood, BCAAs can be used for protein metabolism and fetal growth and can be transported from the maternal side to the fetal side [[81]13]. In late pregnancy, BCAA transport is enhanced to meet the increased nutrient demand from fetal [[82]13]. Pregnant women with diabetes mellitus encounter an increased risk of delivering infants who either exceed expected size or have macrosomia (weighing ≥ 4000 g) [[83]25]. Elevated L-leucine in cord blood in the uncontrolled T2DM group may be a reason that explains the Pedersen hypothesis. Some observations reflected that elevated BCAAs may be linked to obesity and metabolic syndrome [[84]27]. Elevated BCAA levels, linked with high p-BMI, emerged as a shared characteristic in both maternal and cord blood samples [[85]26]. Explanations were provided that showed that BCAAs were a consequence of obesity. Some authors have indicated that elevated BCAA levels increase dietary protein intake or excessive protein breakdown in skeletal muscle due to obesity-related insulin resistance (IR) [[86]28]. Compared with the normal group, glycodeoxycholic acid (GDCA) in cord blood was increased in the uncontrolled T2DM group. GDCA is a kind of bile acid that is a vital signaling molecule in glucose metabolic regulation and has been associated with the regulation of cholesterol catabolism and the homeostasis of triglycerides and glucose [[87]29]. The research demonstrates a connection between GDCA and both insulin secretion and resistance. Elevated levels of GDCA stimulate insulin secretion in a manner dependent on GLP-1 [[88]30]. Interestingly, the administration of GDCA has the potential to alleviate insulin resistance, possibly elucidating the negative correlation between GDCA levels and glucose levels [[89]31]. In addition, in the multivariable analysis, the variables of gestational age and pre-pregnancy BMI appeared to have few influences on the levels of glucose, C-peptide, or insulin. We supposed that the influences of gestational age and pre-pregnancy BMI on the levels of glucose or insulin might be weakened under the condition of gestational hyperglycemia, which need further studies to explore. The relatively small sample size limits the generalizability of some of the conclusions from the present study. To eliminate the influence from the way of delivery, we only included pregnant women with cesarean delivery and excluded those with labor delivery in the study. Moreover, only untargeted metabolic analysis was performed. However, the targeted approach in metabolomics involves analyzing specific types of metabolites and their associated metabolic pathways, building upon pre-existing information as a basis. We will conduct the targeted approach in a larger sample size in our future study. Last but not least, given the association between fetal birth weight and its metabolic profile, matching on fetal birth weight may potentially constrain the identification of genuine differences in umbilical cord blood metabolic profiles between the two groups. Conclusion The metabolic profile of umbilical cord blood among pregnant women with uncontrolled T2DM were different from that of the healthy pregnant women. The metabolic profile of umbilical cord blood from the uncontrolled T2DM women was characterized by the enrichment of the pathways of biosynthesis of amino acids and butanoate metabolism and the metabolites of alisol A and L-leucine. Acknowledgements