Declined Maternal GDCA is Correlated With Insulin Resistance and β-Cell Compensation in GDM: A Cross-Sectional Study in Chinese Pregnant Women

We organized GDM patient (n = 67) and normal pregnant women (n = 48) around the oral glucose tolerance test (OGTT) screening period, fasting serums were collected for the measurement of Bas. Clinical data were collected and Bas metabolism proles were analyzed in the GDM and normal glucose tolerance (NGT). Delivery characteristics, delivery gestational age and infant birthweight were abstracted from medical record to illustrate the differences among the groups.

Insulin resistance (IR) progressed with the gestational advance in order to supply adequate energy for the growing fetus. The increased IR promotes endogenous glucose production and the breakdown of fat stores, resulting in an elevation in glycemia and free fatty acid concentrations [8]. The abnormal metabolic adaptations of IR in pregnancy always consequently lead to GDM. Bile acids (Bas) have also emerged as crucial signaling molecules in glucose metabolism via nuclear hormone farnesoid X receptor (FXR) and Takeda G protein receptor 5 (TGR5) signaling pathways [9]. Studies in rodents have shown that induction of Bas secretion postprandial, leads to stimulation of glycogen storage and inhibition of hepatic glycolytic and lipogenic gene expression in a FXR dependent manner [10,11]. Bas activation of FXR causes repression of enzymes involved in hepatic gluconeogenesis such as phosphoenolpyruvate carboxy kinase (PEP-CK) and glucose 6-phosphatase [12,13]. In addition to the direct regulation, Bas are also involved in glucose metabolism through insulin secretion. TGR5 is differentially activated by (lithocholic acid) LCA > (deoxycholic acid) DCA > chenodeoxycholic acid (CDCA) > cholic acid (CA), activation of TGR5 promotes intestinal L cells to secrete glucagon like peptide-1 (GLP-1) which acts on the pancreatic β-cells to stimulate insulin secretion [14]. FXR is differentially activated by CDCA > DCA > LCA > CA [15], although GLP-1 secretion is negatively regulated by FXR, TGR5 activation in L cells occurs rapidly postprandial, whereas activation of FXR induces a more delayed response that requires transcriptional activation [16,17]. The research of interaction between Bas and insulin illustrated that insulin regulates Bas composition by regulating Bas 12 α-hydroxylase cyp8b1 through FoxO1, serum insulin level is positive correlated with 12 α-hydroxylated Bas (CA, DCA, and their conjugated forms) concentration [18]. In healthy subjects, IR was associated with increased 12 αhydroxylated Bas, ratios of 12 α-hydroxylated/non-12 α-hydroxylated Bas were associated with key features of IR. In type 2 diabetes mellitus (T2DM) patients, TBA were nearly twofold increased compared with healthy subjects, although no disproportionate increases in 12 α-hydroxylated Bas was spotted [19].
We have recently show that Bas metabolism pro le changed with gestation advance, speci cally, unconjugated bile acids dominate during the second trimester, conjugated bile acids overwhelmed in the third trimester [20]. As Bas individuals exerts distinctive a nity with FXR and TGR5 and they play the different role in glucose metabolism, it is necessary to clarify whether the Bas pro le changed in GDM and the clinical signi cance of such change.
The main purpose of this work was to verify the hypothesis that Bas metabolism pro le changed in GDM and there exist a correlation between insulin metabolism and speci c Bas components. To address this issue, we analyzed the fasting Bas metabolism pro le in GDM and normal pregnant women during OGTT screening, and identi ed key Bas individuals changed in GDM, which laid the foundation for the assessment of insulin resistance and β-cell compensation using Bas.

Study design and population
A total of 67 GDM patients and 48 normal pregnant women with a maternal age of 23-44 and a gestational age 21-29 weeks were enrolled between May 2019 to October 2019 in Women's Hospital of Zhejiang University School of Medicine, China. Pregnant Women with preconceptional diabetes, infections, abnormal liver or kidney function and positive for HIV and hepatitis C antibodies were excluded from participation. Blood samples were obtained from all the participants after 8-14 hours of fasting. All the women underwent a fasting 75-g oral glucose tolerance test (OGTT). Based on OGTT results, we defined GDM according to International Association of the Diabetes and Pregnancy Study Groups (IADPSG) criteria [21]. Insulin and glucose levels during the OGTT were used to estimate HOMA-IR, HOMA-β, insulin secretion (using the Stumvoll first-phase estimate) and insulin sensitivity (using the Matsuda index) [22][23][24]. Insulin secretion index and Matsuda index were multiplied to calculate the oral disposition index (DIo), which assesses β-cell compensation for insulin resistance [25]. This study was carried out according to the declaration of Helsinki guidelines, all the participants provided informed consent to participate in the follow-up study, the research protocol was approved by the ethics committee of the Women's Hospital of Zhejiang University School of Medicine (grant number: IRB-20200015-R).

Bile acids measurement
The bile acids measurement procedure was referred in previously published methods with minor modi cations [20]. Brie y, reference standards were commercially available from TRC lnc. (Toronto, Canada) and Sigma-Aldrich (St. Louis, USA). 100 μl of the serum specimen, standard solutions and quality control was vortex mixed with 300 μl of the stable isotope-labeled internal standard (IS) stock solutions. A total of 150 µl of the supernatant was aspirated by the autosampler for LC-MS/MS analysis after centrifuging 10 min at 13,000g. Bas were separated using a UPLC BEH C18 column (2.1 mm × 100 mm, 1.7 µm; Waters) with the 40 • C column temperature. Multiple reaction monitoring (MRM) measurements of the bile acids were performed using individually optimized cone voltage and collision energy (Table S1) Statistical methods Mass spectrometric data was analyzed using Analyst software v1.6.0 (SCIEX, Framingham, USA). Student's t test or Mann-Whitney U test were used to evaluated the difference among groups for continuous variables. One-way analysis of variance (ANOVA) or Kruskal-Wallis test was conducted to compare the different for categorical variables. Bas data were log transformed for orthogonal partial least square discriminant analysis (OPLS-DA) using SIMCA-P (version 13.0, Umetrics, Sweden). All the statistical analyses were performed using GraphPad Prism 7.0. P < 0.05 was considered to be statistically signi cant.
Although there was no statistical signi cance in insulin secretion within the NGT group, fasting insulin and insulin in OGTT over 1 hour and 2 hours in GDM patients was signi cantly increased (P = 0.0003, P = 0.0048, P < 0.0001). In addition, GDM patients had a de ciency in insulin sensitivity (P < 0.0001) and worse β-cell compensation (P < 0.0001) than that of NGT (Table 1). We used all the signi cantly changed parameters in table 1 for OPLS-DA calculations to evaluate differences in GDM and NGT groups. There was a distinct clustering pattern between samples from GDM and NGT individuals ( Figure 1A). The variable importance in projection (VIP) score for the mentioned indicator showed that DIo Matsuda index, HOMA-IR, Stumvoll Phase I and insulin levels in OGTT contributed signi cantly as a principal component to separate the two groups ( Figure 1B). Bile acids metabolism pro les in GDM and NGT As glucose and insulin metabolism discriminations exist between GDM and NGT, and Bas are known to participate in the glycemia homeostasis via direct and indirect pathways, we postulated that GDM and NGT could be distinguished by Bas pro les. However, from the OPLS-DA results, the two cohorts could not be separated (Data not show), even though total bile acid (TBA) decreased slightly in the GDM group (Figure 2A). This nding suggested that speci c types of Bas may be more intimately involved in glycemia regulation. Indeed, 12α-hydroxylated Bas decreased obviously (P = 0.04), Non-12α-hydroxylated Bas remained stable in GDM ( Figure 2B). Furtherly, in 12α-hydroxylated Bas individuals, we veri ed that the concentrations of the TDCA and GDCA in fasting serum decreased signi cantly in the GDM cohort (TDCA, P = 0.03; GDCA, P = 0.0261), Non-12α-hydroxylated Bas individuals changed not signi cantly ( Figure 2C). To further illustrate the relationship between Bas components and the obviously changed insulin related indicators, we analyzed the correlation of them. Fasting GDCA is positive correlated with insulin sensitivity in both NGT and GDM, the correlation coe cient decreased slightly in GDM. GDCA is also positively related to DIo in GDM, but negatively with HOMA-IR and fasting insulin (FIN). TDCA has no correlation with insulin related indicators in NGT and GDM, although changed obviously in GDM. In non-12α-hydroxylated Bas individuals, CDCA is positively related to HOMA-IR and FIN, TUDCA is negatively related DIo (Figure 3).

Correlation of bile acids individuals and insulin indexes
Clinical outcome of GDM and GDM with higher HOMA-IR and lower DIo As GDCA declined in GDM and it is correlated with insulin sensitivity, IR, DIo and FIN. In order to assess the relationship between GDCA and clinical outcome including delivery gestational age, fetus birthweight and cesarean delivery ratio, GDM patients were subdivided into four quartiles (Q1: <25%, Q2: 25%~50%, Q3: 50%~75% and Q4: >75%) according to the Matsuda index, HOMA-IR, DIo and FIN respectively. GDCA changed not obviously among Q1 and Q4 groups divided by Matsuda index and FIN ( Figure 4A, B), but decreased in HOMA-IR Q4 group and increased in DIo Q4 groups ( Figure 4C, D). In clinical outcome assess, GDM patients gained no variations compared with NGT, however, GDM with high HOMA-IR and decreased GDCA (HOMA-IR Q4) group exerts increased cesarean delivery ratio.
GDM with low DIo gained earlier delivery gestational age compared with NGT (Table2).   GDM and GDM subgroups. GDM patients were GDM patients were subdivided into four quartiles Q1, Q2, Q3, Q4) according to HOMA-IR and Dio, § represents no statistically significance with NGT group, P # represents the difference across Q1 and Q4 groups. P< 0.05 was considered to be statistically significant.

Discussion
In this study, we learned that the abnormalities of insulin-related indicators played the main role to distinguish GDM and NGT. Combined with the bile acids metabolism pro le, we clari ed the correlation between speci c bile acids and insulin related indicators. Overall, GDM patients exert higher levels of BMI and HbAc1 during the OGTT screening period, in addition, fasting insulin and insulin at 1 and 2 hours post 75-g glucose intake increased signi cantly in GDM. In terms of bile acid metabolism pro le, TDCA and GDCA were the most signi cantly changed individuals in GDM. Moreover, GDCA was negatively correlated with the insulin resistance index (HOMA-IR) and positively correlated with the β-cell compensation index (DIo). In clinical outcomes assay, there was no signi cant difference between GDM and NGT, but GDM patients in HOMA-IR Q4 group possessed elevated ration of cesarean section delivery, and patients in DIo Q1 group gained earlier delivery gestational ages. Therefore, our ndings suggest that fasting GDCA levels during OGTT may be a useful indicator to evaluate insulin resistance and β-cell compensation, furtherly to predict adverse clinical outcomes of GDM.
GDM shares the similar pathogenesis with T2DM, β-cell function was insu cient to compensate for the need for glycemic homeostasis in pregnancy combined with reduced insulin sensitivity, nally result to increased serum glucose [1]. In addition to glucose and insulin metabolic disorders, lipids metabolism also changed in GDM, total triglycerides, total cholesterol, low-density lipoprotein cholesterol increase gradually throughout pregnancy in GDM [26]. Our results were consistent with previous report in clinical characteristic of GDM, including advanced age, higher BMI , TG and HbA1c.
Bile acids metabolism variation has been partially proved in T2DM, elevation of TBA increased the T2DM incidence, TBA almost doubled in T2DM patients compared with healthy subject [19,27]. Intrahepatic cholestasis of pregnancy (ICP) was characterized by an increased TBA, ICP women with higher TBA concentration are much more vulnerable to suffering GDM [28]. These studies suggest that TBA probably related to the occurrence of T2DM and GDM. Indeed, in Chinese prospective cohort studies, the incidence of GDM in the group with highest TBA (≥ 4.0 µM) in early pregnancy had a 6.72-fold increased risk of GDM compared with the lowest level. Even after adjusting potential confounders, TBA ≥ 2.0 µM still presented an increased risk for developing GDM [29,30]. In our research, TBA changed slightly in GDM compared with normal glucose tolerance, this may be partially caused by the difference between the TBA detected by enzymatic cycling assay and the TBA consist of 15 bile acids individuals detected mass spectrometry. This provokes us that more attentiveness should payed to the bile acids components related to glucose metabolism.
Research conducted on a large cohort of Chinese pregnant women in early pregnancy found that fasting serum levels of GUDCA ≤ 0.07 nmol/mL and DCA ≤ 0.28 nmol/mL were independently associated with an increased risk for development of GDM [31].Untargeted metabolomics administrated in pregnant women's fasting serum during OGTT screening period spot that CA, ios-DCA and dehydio-LCA decreased in GDM [32]. In this research, GDCA and TDCA declined obviously in GDM. The difference between our research and the previous studies may be caused by the difference between targeted and non-targeted detection platforms and the samples collected in different trimesters, as our previous research has illustrated that bile acids metabolism pro les changed periodically with gestational age [20].
GDCA has no correlation with HOMA-IR and fasting insulin in normal glucose tolerance. In GDM, declined GDCA combined with the increased fasting insulin and fasting glucose result in the signi cant negative correlation between GDCA and HOMA-IR. Studies in vivo and in vitro demonstrated that GDCA is indeed related to insulin secretion and insulin resistance, more importantly, insulin resistance can be alleviated by GDCA administration. For example, fasting for 40 hours induce insulin resistance, while this insulin resistance model has no affect in bile acids metabolism pro le. However, postprandial GDCA and insulin concentration changed in a signi cant positive correlation pattern. Increased GDCA triggered the secretion of insulin in a GLP-1 dependent manner [33]. To some extent, this is helpful to explain why even if GDCA elevated the post glucose intake in GMD patients, but declined baseline of GDCA still make it insu cient to promote insulin secretion via GLP-1, nally failed in the glycemia regulation. Polycystic ovary syndrome (PCOS) is usually accompanied by features of insulin resistance, HOMA-IR of PCOS patients is signi cantly negatively correlated with GDCA. Mechanistically, the signi cant increased Bacteroides vulgatus in the gut microbiota of PCOS individuals reduced the GDCA. GDCA combined with GATA-3 in the intestinal group 3 innate lymphoid cell to facilitate the interleukin-22 secretion, which in turn improves the PCOS phenotype including insulin resistance and ovarian function recovery [34]. However, whether variations occurred in GDM gut microbiome declined the GDCA and result in insulin resistance, the con rmation need further research.
Hyperglycemia burdened severely clinical outcome to both maternal and fetus outcome with GDM, but the clinical outcome of the whole GDM was not differed signi cantly from that of NGT in this study, which may be due to the early lifestyle intervention and even pharmacological therapy. Of note, GDM patients with serious de ciency of insulin compensation and high HOMA-IR subgroups gained worse clinical outcome.
Our study design, however, has several limitations. Firstly, the GLP-1 baseline was not measured in the fasting serum, and any variation in GLP-1 baseline in GDM and NGT may have in uenced fasting insulin levels. Secondly, GLP-1 and Ba changes in 1-h and 2-h period after the glucose loading were also not detected, which limited the elucidation of the dynamic mechanism of Bas metabolism regulating insulin secretion through GLP-1, moreover, the fewer participants failed us to establish the cut-off point of GDCA, under which may increase the worse clinical outcome risk.

Conclusions
GDCA and TDCA declined signi cantly in GDM. GDCA was inversely correlated with insulin sensitivity and positively related with insulin compensation. which make it a valuable candidate biomarker for the assessment of insulin sensitivity and β-cell compensation. Declined GDCA is correlated to adverse pregnancy outcomes of GDM.

Declarations
Ethics approval and consent to participate This study was carried out according to the declaration of Helsinki guidelines, the research protocol was approved by the ethics committee of the Women's Hospital of Zhejiang University School of Medicine (grant number: IRB-20200015-R). All the participants provided informed consent to participate in this study and the future follow-up study.

Consent for publication
Not applicable.

Availability of data and materials
All the data are available to interested researchers upon reasonable request. Requests for access to data should be made to the rst author with e-mail:5202054@zju.edu.cn.
Competing interests "#" these authors have the same contribution to this article, none of the authors has any con ict in relation to the study Zhu and Dajing Xia contributed to drafting the manuscript. All authors reviewed and edited the manuscript to produce the nal version. OPLS-DA score plot of NGT and GDM according to the index changed obviously in table 1, the black and red square represent the NGT and GDM individuals respectively, the black circle represents the 95% con dence interval (A). A taxon with a VIP score of > 1 was considered important in the group discrimination (B). OPLS-DA score plot of NGT and GDM according to the index changed obviously in table 1, the black and red square represent the NGT and GDM individuals respectively, the black circle represents the 95% con dence interval (A). A taxon with a VIP score of > 1 was considered important in the group discrimination (B).    FIN, 1HIN, 2HIN. The correlation strength is shown by the color bar, red represents a positive association, blue represents a negative association and white no association, * P < 0.05, ** P < 0.01.