Gut microbial-related metabolite trimethylamine-N-oxide and its precursor L-carnitine are associated with spontaneous preterm birth: a nested case-control study

Background: Gut microbiota has been proven to disease susceptibility and may lead to increased risk of preterm birth. To date, the link of gut microbial-related metabolite trimethylamine-N-oxide (TMAO), L-carnitine, and betaine, with spontaneous preterm birth (sPTB) has not been established. This study aimed to investigate the association of TMAO, L-carnitine and betaine, with sPTB risk. Methods: A nested case-control study was designed including 129 sPTB cases and 258 controls based on Guangxi Birth Cohort Study. TMAO, L-carnitine, and betaine level in maternal serum were determined by liquid chromatography with mass spectrometry. Conditional logistic regression analyses were used to examine the association between maternal serum metabolites and sPTB. Stratified analyses were further conducted according to BMI and preterm prelabor rupture of membranes. Spline analyses were performed to explore the dose-response relationship between the metabolites and sPTB. Results: Statistically significant association with decreased sPTB risk was observed for the highest L-carnitine (OR: 0.47; 95% CI: 0.23, 0.95). In risk analyses stratified by BMI, similar results were observed in normal weight gravida (BMI: 18.5~23.9 kg/cm 2 ). The significant subtype-specific association with TMAO (OR: 0.43; 95% CI: 0.20, 0.93) and L-carnitine (OR: 0.45; 95% CI: 0.21, 0.97) were observed for preterm labor but not PPROM. Spline regression analysis indicated non-linear associations with TMAO and sPTB risk ( P for nonlinearity: 0.057). Significant associations of TMAO with sPTB were observed in normal weight gravida ( P = 0.028) and preterm labor subtype ( P = 0.025). No statistically significant associations with sPTB risk were observed for betaine (P > 0.05). Conclusions: TMAO and

and sPTB may open new avenues for diagnose and therapy.

Background
Spontaneous preterm birth (sPTB), delivered at less than 37 weeks' gestation age, remains most common complication in obstetrics and gynecology [1,2]. sPTB is the leading perinatal morbidity and mortality worldwide [3]. Preterm infants have been verified to face quite complicated issues such as low birth weight, feeding difficult, high quality of neonatal intensive care, and an increased risk of neurodevelopmental impairments, respiratory and gastrointestinal complications in adulthood compared with their term counterparts [3]. Hence, effective prediction of sPTB is useful to allow us to initiate appropriate risk-specific treatment and reduce the risk as far as possible.
Recently, gut microbiota metabolites trimethylamine N-oxide (TMAO) has been received increasing attentions. The growing recognition of a contributory role of gut microbiota to health and disease promises that great efforts have devoted to define the cause-effect between gut microbiota and preterm birth [6][7][8][9]. Moreover, experimental and clinical studies have demonstration that TMAO synthesized by the intestinal microbiota is involved in the progression of atherosclerosis, cardiovascular disease, type 2 diabetes, vitamin D deficiency, chronic kidney disease, non-alcoholic fatty liver disease [10-12]. Those adverse impacts induced by microbial-related metabolite TMAO may directly affect the physical condition of pregnant women and birth outcomes [13,14]. Meanwhile, TMAO is a gut microbiota-dependent metabolite of L-carnitine [15]. L-carnitine participates in energy metabolism and is essential to fetal growth and organ development [16]. L-carnitine, similar to glycine betaine (betaine) in structure, belongs to the "betaines" group [17,18].
Betaine serves as an osmolyte to protect cell and protein from environmental stress, and a methyl group donor to homocysteine for synthesis of methionine [19,20]. Betaine insufficiency is attributable to the high risk factors of the metabolic abnormalities, obesity, diabetes, and vascular disease [18].
Despite gut-microbiota dependent L-carnitine metabolism has been linked to maternal health and fetus developments, the association of TMAO and its precursor (L-carnitine, betaine), with sPTB risk has not yet been reported. Herein, this study conducted a nested case-control study to investigate the association of TMAO, L-carnitine and betaine levels in maternal serum, with sPTB risk in a large Chinese population.

Study Design and Population
Our nested case-control study was embedded in the Guangxi Birth Cohort Study (GBXCS), an ongoing multicenter prospective cohort study in Guangxi, South China, which aims to investigate pregnancy outcome and offspring short-and long-term health consequences of hereditary, environmental and lifestyles factors in the fast-paced society. The baseline design of the GBCS has been previously reported [21]. Flow chart of the study population was shown in Figure 1 In 2016, after first follow-up survey, 5,541 (89.33%) mother-infant pairs had been observed. 206 pregnant women (delivery week < 37 weeks) were selected excluding abortion, stillbirth and birth defects, multiple pregnancies, medically premature delivery, full-term low birth weight and Macrosomia. To minimize the effect of the erratic menstrual cycle on gestation age, we carried out 129 participants delivers at < 36 weeks' gestation age as cases in this work. 258 term deliveries (delivery week ≥ 37 weeks) were selected randomly according to maternal age, enrollment hospital, enrollment gestational age at baseline blood draw and infant gender as controls. This study was approved by Medical Ethics Committee of First Affiliated Hospital of Guangxi Medical University (ID: 2015(028)).
All patients consented to participate in the research, and written informed consent was obtained from each patient.
(USA) and used as internal standards. Serum were collected and stored at -80 °C until analysis. Frozen serum samples were thawed and centrifuged at 4 °C for 5 min. Serum concentrations of TMAO, L-carnitine and betaine were quantified using stable isotope internal standards method on an Agilent 6460 triple quadrupole mass spectrometer in positive multiple reaction monitoring (MRM) mode according to the literature method with slight modifications [15,22]. Agilent ZORBAX HILIC plus column (2.1 × 50 mm, 3.5 μm, Agilent, USA) was used. 20 μL plasma was aliquoted into 1.

Statistical Analyses
The baseline characteristics of cases and controls were performed using descriptive analysis. Continuous variables were compared using t-tests and categorical variables compared using chi-square tests. Conditional logistic regression analyses were used to examine the association between maternal serum metabolites tertile concentrations and preterm delivery by calculating odds ratios (ORs) and their 95% confidence intervals (CIs), which were adjusted by maternal age, ethnic, education levels, smoking, and drinking status during pregnancy. Stratified analyses were further conducted according to BMI (< 18.5, 18.5 -23.9, ≥ 24 kg/m 2 ), and PPROM (yes, no) by using logistic regression. Linear trend P-values were derived by modeling the median value of each tertile as a continuous variable in adjusted models. A spline analysis to explore the dose-response relationship of the selected biomarkers and the risk of sPTB was performed. Knots were placed at the 25th, 50th, and 75th percentiles of the serum TMAO distribution, and the reference value was set at the 25th percentile. Adjustment factors were age, pre-pregnancy BMI (not for BMI stratified model), ethnic, education, smoking, and drinking. SPSS version 22.0 (IBM) and SAS (version 9.4; SAS Institute Inc.) were applied in statistical analysis. Two-sided P < 0.05 was considered as statistical significance.

Results
Baseline characteristics of sPTB cases (n = 129) and controls (n = 258) are shown in Table1. Compared with the controls, the cases involved in a higher BMI (P = 0.02), lower infant weight (P < 0.001) and shorter infant height (P < 0.001). No significant differences were observed in maternal age, ethnic, family income, educational levels, drinking, smoking, gestational age at sample collection, infant gender, history of gravid and parity, and infant gender (all P > 0.05). The sPTB group had a higher percentage of preterm labor (n = 63, 48.8%). As well, the percentage of PPROM was 43.4% (n = 56), and 10 cases were unknown (7.8%). Serum concentrations of TMAO, L-carnitine, and betaine did not differ between cases and controls.
Associations between maternal serum TMAO, L-carnitine and betaine level and sPTB risk are shown in Figure 2. Statistically significant associations with decreased sPTB risk were observed for the highest categories of L-carnitine. Multivariable-adjusted for age, prepregnancy BMI, ethnic, education, smoking, and drinking, women in highest tertile of Lcarnitine were 53% lower sPTB risk than those in lowerst tertile (OR= 0.47; 95% CI: 0.23, 0.95; P trend 0.035). However, no statistically significant associations with sPTB risk were observed for TMAO and betaine (all P > 0.05). Spline regression analysis indicated marginally significant non-linear associations with TMAO and sPTB risk ( Figure 3A, P for overall = 0.070, P for nonlinearity = 0.057), with a sharp decrease in slope for concentrations above 1.63 μmol/L. Other metabolites had no dose-relationship with sPTB (P for overall > 0.05).
Analyses of the other metabolites did not suggest clear associations with specific sPTB subtypes.

Discussion
As far as we know, this is the first study to assess the associations of gut microbialrelated metabolite TMAO and its precursor L-carnitine with sPTB risk. A nonlinear association between serum TMAO and sPTB risk were observed. According to prepregnancy BMI, the nonlinear association between the serum TMAO and sPTB risk were observed in normal BMI. Meanwhile, preterm birth subtypes appeared to modify the association between serum TMAO and sPTB risk with linear association observed in the preterm labor, but not in PPROM. L-carnitine levels in maternal serum were inversely associated with sPTB. The significant subtype-specific associations with L-carnitine were observed in normal BMI and preterm labor. As such, the association of maternal serum TMAO and L-carnitine with sPTB risk appears to be dependent on BMI and preterm labor availability. Additionally, no correlation was found between betaine and sPTB.
Nutrition metabolism in the context of pregnancy is crucial to maternal health and fetus development from the pre-implantation embryo to infancy [23,24]. The gut microbiome has been identified to change during gestation [7]. Shiozaki et al. demonstrated that disturbance of intestinal microbiota would result in preterm delivery in a prospective and cross-sectional study [8]. Dahl et al. provided further evidences for the impact of the low gut diversity and low relative abundance of bifidobacterium and streptococcus on sPTB [9]. The changes of gut microbiome diversity and evenness may alter bioactivation of nutrients and vitamins, normal physiology of host, and at same times, also affect immune system [25][26][27].
In this study, our data have uncovered increased sPTB risk was observed for the low categories of TMAO and L-carnitine. These findings revealed that the gut-microbiota initiated TMAO metabolism may represent a key factor contributing to sPTB. Gut microbialrelated metabolite TMAO depend on the dietary sources from fish or endogenous synthesis from L-carnitine present in eggs, milk and red meat. [15] L-cartinine can be metabolized to TMA by the anaerobic intestinal bacteria, then further oxidized to TMAO [15]. L-carnitine participates in energy metabolism as the transfer of activated long-chain fatty acids into the mitochondrial matrix, regulates coenzyme A (CoA)/acylCoA ratio and maintains energy storage in the form of acetylcarnitine [16,28]. Maternal plasma carnitine levels on delivery decrease to half of the concentrations in non-pregnant women, comparable to those found in known carnitine deficiency [29]. L-carnitine deficiency has been implicated in the pathogenesis of cardiomyopathy, skeletal weakness, diabetes mellitus, hyperammonaemia, and other disorders [15,28,30]. L-carnitine supplementation in pregnancy doses avoids insulin resistance and gestational diabetes mellitus [31]. Because the low activity of the butyrobetaine hydroxylase was to restrict the endogenous synthesis of carnitine in fetal tissue, fetal L-carnitine level mainly depended on the maternal carnitine via transplacental transfer [32,33]. L-carnitine supplementation in pregnancy is highly demand to pregnant women and their fetuses. Therefore, low TMAO and carnitine status of pregnant women may lead to malnutrition, organ dysfunction, and embryonic development restriction. In addition, TMAO synthesized by the intestinal microbiota has been demonstrated as early predictors of metabolic syndrome, and strong linked with atherosclerosis, type 2 diabetes, vitamin D deficiency and chronic kidney disease [10-12, 34]. The metabolic syndrome induced by TMAO caused increased sPTB risk [13,14].
Collectively, the nonlinear relationship between TMAO level and sPTB has been uncovered, with a significant positive association for the low level and an inverse association with high TMAO concentration.
In the stratified analysis, the effect of serum TMAO and L-carnitine levels on sPTB depended on the level of pre-pregnancy BMI. The adverse association of serum TMAO and L-carnitine, with sPTB risk became insignificant among the low or High BMI, and PPROM.
Schugar et al. found that levels of the TMAO-producing enzyme flavin-containing monooxygenase (FMO3) is linked to obesity [35]. The effect of BMI on FMO3/TMAO metabolism pathway could partially explain the improved predictive ability with BMI stratification. In addition, a growing body of evidence suggests that utero infection and vaginal microorganisms have been causally associated to PPROM [36,37]. The effect caused by the inflammatory activation on sPTB was much greater than those of TMAO and L-carnitine, which may cover the risk effect of TMAO and L-carnitine on sPTB in PPROM.
In this study, null associations between betaine and sPTB risk were noted. Betaine serves as an osmolyte. The betaine concentration in tissue is much higher than that in serum, while serum betaine is presumably mainly influenced by liver betaine level [18,38]. Thus, the serum betaine could not reflect the whole level of the pregnant women, which may weaken the effect of serum betaine on sPTB.
Key strengths of this study are that this is the first study to investigate the association of maternal serum TMAO, L-carnitine and betaine, with sPTB risk in humans. Furthermore, in this nested case-control study, cases and controls were enrolled with comparable age, ethnic, family incomes, education, drinking, smoking, gestational age at sample collection, gravid, parity and infant gender in order to minimize recall bias. Thirdly, to reduce multiple confounding factors, well-established factors (i.e., age, pre-pregnancy BMI, ethnic, education, smoking, and drinking) were included in the analyses. The main limitation of this study is that although cases and controls were recruited with complete matching by gestational age at sample collection, the concentrations of serum TMAO, Lcarnitine, and betaine have changes during gestation, one sample collection may not exactly give an accurate representation of the metabolites levels. Secondly, although we have considered some confounding factors in the analyses, there is a potential for residual confounding factors that were not recorded. Thirdly, the sample size might be relatively small, limiting the statistical power to detect associations of stratified analyses. More prospective studies are required to verify the causality between TMAO, L-carnitine and sPTB.

Conclusion
TMAO and L-carnitine were inversely linked with sPTB, while betaine level is not associated with sPTB risk. If possible, these results should be confirmed by large prospective studies in the future. It is worthwhile to make further study of the mechanism of the effect of gut-microbiota initiated L-carnitine/TMA/TMAO metabolism path way on the sPTB. Discovery of a relationship between gut-microbiota initiated nutrition metabolism and sPTB provides opportunities for the development of new diagnostic procedures and therapeutic strategy for sPTB.

Ethics approval and consent to participate
The study was reviewed and approved by the Guangxi Medical University Medical Ethics Committee (ID: 2015(028), Approved 16 July 2015). All patients consented to participate in the research, and written informed consent was obtained from each patient.

Consent for publication
All authors have read and approved the submission of the manuscript. The manuscript has not been published and is not being considered for publication elsewhere.

Availability of data and materials
The data used to support the findings of this study are included within the article or available from the corresponding author upon request.

Conflicts of Interest
The authors declare no conflict of interest.   Non-normally distributed variables were shown as median (IQR) and P value was derived from U-test. Categorical variables were shown as numbers (percentage) and P value was derived from Chi-square test. Figure 1 Flow chart of the study population. Individuals with preterm labor.