Association Of Gut Microbiota During Early Pregnancy With Risk Of Incident Gestational Diabetes Mellitus: A Nested Case-Control Study

doi:10.21203/rs.3.rs-48405/v1

Download PDF

Research

Association Of Gut Microbiota During Early Pregnancy With Risk Of Incident Gestational Diabetes Mellitus: A Nested Case-Control Study

https://doi.org/10.21203/rs.3.rs-48405/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: We aimed to assess gut bacterial biomarkers during early pregnancy and subsequent risk of gestational diabetes mellitus (GDM) in Chinese pregnant women.

Methods: Based on the Tongji-Shuangliu Birth Cohort study, we conducted a nested case-control study among 201 incident GDM cases and 201 matched controls individual matched on age, gestational week, and date of fecal sample collection. Fecal samples were collected during early pregnancy, and GDM was diagnosed at 24-28 weeks of pregnancy. Community DNA isolated from fecal samples and V3-V4 region of 16S rRNA gene amplicon libraries were sequenced.

Results: In GDM cases versus controls, Rothia, Actinomyces, Bifidobacterium, Adlercreutzia, and Coriobacteriaceae, and Lachnospiraceae spp. were significantly reduced, while Enterobacteriaceae, Ruminococcaceae spp. and Veillonellaceae were over-represented. It was also found that the abundance of Staphylococcus relative to Clostridium, Roseburia and Coriobacteriaceae as reference microorganisms were positively correlated with fasting blood glucose, 1-h postprandial glucose, and 2-h postprandial glucose levels. Conditional Logistic regression showed that 4 microbial taxa during early pregnancy were associated with subsequent GDM risk, including Actinobacteria, Coriobacteriaceae, genus 26 of Coriobacteriaceae and an unknown specie of the Coprococcus.

Conclusion: Gut microbiota during early pregnancy was associated with subsequent risk of GDM. Several beneficial and commensal gut microorganisms showed inverse relations with incident GDM, while opportunistic pathogenic members involving in GDM development, where they positively correlated with clinical measurements on OGTT. Our study provided promising biological markers that may be used in monitoring pregnant women’s health and developing therapeutic interventions on GDM at early stage of gestation.

General Microbiology

gestational diabetes mellitus

gut microbiota

pregnancy

China

Gestational diabetes mellitus (GDM) is a leading pregnancy-related complication worldwide. While the prevalence of GDM varies globally from 1% to > 30% [1], an estimate 15% of pregnant women have GDM in China based on the criteria from the International Association of the Diabetes and Pregnancy Study Groups (IADPSG) [2]. GDM can increase risk of immediate pregnancy complications such as preterm birth, macrosomia, and Cesarean section, and risk of future cardio metabolic disease in both mothers and their offspring [1]. Thus, elucidating the etiology of GDM is essential to tackle GDM and its adverse consequences.

There is accumulated evidence that type 2 diabetes is related to profound dysbiosis that acts through underlying mechanisms such as immune metabolic effects from short-chain fatty acids (SCFAs) and metabolic inflammation due to endotoxemia [3]. Although less evidence on the role of gut microbiota in GDM has been provided, several metagenomics and metabolomics studies observed correlations between gut microbiome and GDM. Gut bacteria such as Prevotella, Bacteroides thetaiotaomicron, Staphylococcus aureus, and Bacteroides vulgatus were enriched in pregnant women with overweight, which was a risk factor for GDM [4]. The enrichment of several gut microorganisms correlated with metabolites that interact with GDM, including Ruminococcaceae, Coriobacteriaceae, Collinsella, Lachnospiraceae, Prevotellaceae, and Bacteroidaceae. Although there have been studies showing aberrant gut microbiota associated with GDM at middle [5] and late pregnancy [6], for GDM prevention purposes, it is critical to focus on gut microbiota of pregnant women during the 1st semester of gestation before GDM is diagnosed. Since gut microbiota can be influenced by many factors including diet, lifestyles, and medications etc., it is a big challenge to identify microbial biomarkers masked by these impacting factors, especially during pre-disease state. Several prospective studies reported some gut microorganisms changed their relative abundances in GDM women [7–10], although the results were confusing in that no common biomarkers were identified across various studies, which may due to limitations in sample sizes and study design.

To estimate the role of gut microbiota in GDM development, we prospectively examined the associations between gut microbiota characteristics during early pregnancy and subsequent risk of GDM in a large birth cohort in China. We selected 201 pregnant women with GDM and 201 matched non-GDM pregnant women from the Tongji-Shuangliu Birth Cohort (TSBC), and explored gut microbial biomarkers during early pregnancy for prediction of GDM diagnosed at mid pregnancy.

Study population

The TSBC is an ongoing birth cohort that was designed to enroll over 10,000 pregnant women starting from the early pregnancy to identify risk factors for health outcomes of mothers and their offspring. Briefly, pregnant women aged 18-40 years old were recruited during the early pregnancy when they presented to antenatal care clinics in Shuangliu Maternal and Child Health Hospital. The SMCHH provides comprehensive health services to women and children of a population of 1.3 million in the catchment area of Shuangliu District of Chengdu, China [11]. We excluded pregnant women if 1) they had received infertility treatments such as in-vitro fertilization and intrauterine insemination; 2) they reported severe chronic or infectious diseases (e.g., cancer, HIV, or tuberculosis); or 3) they were unable to or refused to complete the baseline questionnaire.

From March 2017 to August 2019, a total of 4370 pregnant women were recruited. 3995 received 75-g oral glucose test (OGTT) during 24-28 weeks of pregnancy, and 262 (6.6%) were diagnosed of GDM using the following IADPSG cutoffs for 75-g oral glucose tolerance test during 24-28 weeks of pregnancy: fasting glucose, 5.1 mmol/L; 1-hour glucose, 10.0 mmol/L; and 2-hour glucose, 8.5 mmol/L [12]. Of the confirmed GDM cases, 61 did not provide either fecal samples (27 cases) or sufficient information (34 cases) at enrollment for current analyses. We conducted a 1:1 nested case-control study among the remaining 201 GDM cases and 201 non-GDM controls individually matched on age (±3 years), gestational weeks (±3 weeks), and date of fecal sample collection (±4 weeks).

Sample collection, 16S rRNA gene sequencing and data analyses

Briefly, all participants completed a set of structured questionnaires. For this study, a total of 402 fecal samples (201 GDM cases and 201 matched controls) were collected and used to isolate genomic DNA. Community bacterial 16S rRNA gene sequences (V3-V4 region) were PCR amplified from the extracted DNA with the primers 341F and 805R [13]. We applied a two-step nested PCR procedure to generate community 16S rRNA gene amplicons and implemented 300bp×2 paired-end sequencing on the Illumina MiSeq platform (Illumina, USA). Bioinformatics analyses were performed using R Statistical Software version 3.5.3 (Foundation for Statistical Computing, Vienna, Austria) and the Quantitative Insights into Microbial Ecology version 2 (QIIME2) software suite (version 2019.4) software. Detailed information on data analysis etc. were illustrated in Fig. 1 and described in additional file 1: Supplementary Methods and Materials.

Baseline characteristics of study population

Demographic characteristics of 201 GDM cases and 201 controls are shown in Table 1. The mean age of women in this study was 28.0 (standard deviation [SD], 4.2) years old, with gestational age of 11.6 (SD, 2.0) weeks. No significant differences were found between cases and controls in smoking, family history of GDM, and intakes of meat, vegetables, and fruits at enrollment in early pregnancy. Compared with non-GDM controls, women with GDM had significantly higher BMI, and were more likely to be current or former alcohol drinkers. A total of 13 GDM cases had a history of GDM, while none of the controls had a history of GDM.

Sequence processing and quality control

We employed a community 16S rRNA gene amplicon based sequencing approach to compare the gut microbiota of GDM cases and controls. For each sample, community 16S rRNA gene were amplified and sequenced yielding 21,914,351 high quality joined reads after quality filtering and chimera check (54,513 reads per sample on average). We clustered all quality sequences into 9,852 features. Rarefaction curves demonstrated that the sequencing depth of each sample at 13,000 was deep enough to represent its microbial diversity (Additional file 2: Figure S1).

Gut microbiota differences between GDM cases and controls

The overall gut microbial α-diversity indexes including observed ASVs, Shannon diversities, evenness, and Simpson diversity based on randomly subsampled 13,000 sequences per fecal sample showed no significant difference between GDM cases and controls (Additional file 2: Figure S2). Results on β-diversities by PCoA analyses suggested that the overall gut microbial community structures in GDM cases were significantly different (P<0.05, PERMANOVA) from controls during early pregnancy (P>0.05, PERMDISP) (Fig. 2). Then we followed the reference protocol of QIIME2 to identify the differentiated ASVs by Ancom analysis from the remaining 208 ASVs after filtered from 9852 ASVs. However, only 2 genera (Coprococcus and Anaerostipes) were identified to successfully differentiate GDM cases from controls during early pregnancy (Additional file 2: Figure S3), and both belonged to the Lachnospiraceae family of the Clostridiales order and were significantly reduced in the GDM group (Additional file 3: Table S1).

To further investigate the taxonomic differences of gut microbiota, we compared the relative abundances of bacterial groups at multiple taxonomic levels (Additional file 3: Table S2). At phylum level, Actinobacteria (case: 4.5%, control: 6.9%), Cyanobacteria (case: 0.19%, control: 0.21%) and TM7 (case: 0.39%, control: 0.55%) were significantly under-represented, while Proteobacteria that contained multiple opportunistic pathogens (case: 4.5%, control: 3.2%) was significantly over-represented in GDM cases (Fig. 3). We further compared Actinobacteria at lower taxonomic levels and found SCFA producing Actinobacteria members including Rothia, Actinomyces, Bifidobacterium, Adlercreutzia, and Coriobacteriaceae were all significantly reduced in GDM cases versus controls, while inflammation related members of Collinsella were not significantly different (Additional file 3: Table S2, Fig. 3). There were no significant differences in relative abundances on other dominant phyla of Firmicutes, Bacteroidetes, or their ratios (Additional file 2: Figure S4).

At family level, 14 abundant and prevalent families (relative abundance >1%, prevalence >50%) that added up to >95% of the gut microbiota were identified (Fig. 3, Additional file 3: Table S2). Among these abundant families, 4 were found significantly different between GDM cases and controls. Veillonellaceae in the order of Clostridia was significantly enriched in GDM cases versus controls (9.4% versus 8.4%), while potential beneficial bacterial group Bifidobacteriaceae (2.9% versus 4.3%) and Coriobacteriaceae (1.4% versus 2.4%) belonging to Actinobacteria were reduced. Streptococcaceae (3.8% versus 5.3%) in the order of Lactobacillales were also reduced in GDM cases. Enterobacteriaceae belonging to Proteobacteria that contains multiple pathogenic microorganisms was slightly higher in GDM cases (3.4% versus 2.2%), despite no difference of statistical significance (Additional file 3: Table S2).

Biomarker identification through multiple approaches

1) Genera differences between GDM cases and controls identified by Random Forestclassifier and LEfSe analysis

Random Forest classifier selected 96 from the total 274 genera included in the model (Fig. 4). The model has a relatively high robust diagnostic accuracy (areas under the ROC curve [AUC]=0.77) for detecting GDM. In the ranks of importance for the model, we found several taxa showed high impact constructing the model (Additional file 3: Table S3). The genus 227 of Enterobacteriaceae, which has multiple members of opportunistic pathogens, was significantly elevated in GDM cases (Additional file 3: Table S3). Beneficial Actinobacteria members showed high importance in the model. Those include Actinomyces and Coriobacteriaceae spp. (genus 26 and genus 27). Moreover, Gemellaceae that were significantly under-represented in GDM cases had high importance for GDM model construction. Other microorganisms included Ruminococcus, Clostridium and other several taxa in Lachnospiraceae (genus 103, Anaerostipes and Blautia).

LEfSe analysis identified 17 biomarkers at genus level that differentiate GDM cases from controls (Fig. 5). Some overlapped with the ones of high importance identified by random forest analysis as Gemellaceae and multiple beneficial Actinobacteria members including Adelercreutzia, genus26 of Coriobacteriaceae, Bifidobacterium, Rothia, and Actinomyces were found significantly reduced, while the genus67 of Cyanobacteria was significantly over-represented in GDM cases. Interestingly, there were 4 other genera enriched in GDM cases. These genera included Lachnospira, SCFA producing bacteria Phascolarctobacterium, Cardiobacterium that associated with inflammation, and opportunistic pathogen Paraprevotella (Fig. 5).

2) Gut microbiota correlated to GDM and glucose levels in the DR analysis

We applied 396 features at species level to the DR analysis, and screened from 197 features after regression to identify which taxa changed the most relative to each other in GDM cases versus controls. The top 10 most differential taxa were identified to discriminate GDM cases from controls, and added information compared with relative abundance-based analysis (Fig. 6). Interestingly, the SCFA producer Ruminococcus was on the top list of shifting towards GDM. The top 5 most reduced taxa in GDM cases included opportunistic pathogen Erysipelatoclostridium, beneficial SCFA producing Lactobacillus and Weissella.

We also highlighted several well-known gut microorganisms that tend to have high or low ranks and were considered associated with disease or healthy status. We found in that Clostridium, Roseburia, and Coriobacteriaceae tended to have high ranks, and Staphylococcus tended to have low ranks. The difference in ranks suggested similar findings in relative abundances that Clostridium, Roseburia, and Coriobacteriaceae were more abundant relative to other taxa in controls, while Staphylococcus was more abundant relative to other taxa in the GDM group. When we investigated the log-ratio of the relative abundance on the above two taxa between GDM cases and controls, we observed the increase of Staphylococcus relative to Clostridium, Roseburia, and Coriobacteriaceae in GDM cases. We also investigated the correlation of log ratios between the above selected taxa and OGTT values measured during the middle of pregnancy (Table 1). The log ratios between Staphylococcus and the other taxa were significantly different between GDM cases and controls, and correlated with OGTT values (Fig. 6).

3) GDM associated microbial interactions in co-occurrence network analyses

The microbial association network analysis was first performed based on all 274 features identified at genus level (Additional file 2: Figure S5). GDM cases had much few microorganisms (136 nodes for GDM cases, 155 nodes for controls) and significant correlations (891 edges for GDM cases, 1073 edges for controls) in the network compared with the control group (Additional file 3: Table S4). Each of the genera selected for both networks was calculated for the degree of differences between GDM cases and controls based on its importance (with how many nodes significantly correlated) in the network. Correlation analysis identified 177 genera that were significantly different in GDM cases compared with controls (Additional file 3: Table S5).

Besides microorganisms identified in other analyses, the roles of additional members were identified as changing dramatically in the network between GDM cases and controls. The top changing genera included Proteobacteria members living in extreme environments (Psychrobacter), pathogenic microorganisms of Fusobacterium, short chain fatty acid producers of Anaerostipes, Coprococcus, Blautia, Faecalibacterium, Roseburia, and Sutterella, beneficial bacteria Akkermansia of Verrucomicrobia, Atopobium and Eggerthella of Actinobacteria, and Paraprevotella of Bacteroidetes (Additional file 3: Table S6). To examine how the roles of discriminatory taxon differentiate between GDM cases and controls, we selected 27 genera of interest and conducted co-occurrence network analysis on a smaller scale. The resulting networks (P<0.05, |r|>0.2) showed that the correlations among these microbes were different between GDM cases and controls (Fig. 7). Consistent with what we found in the overall network analysis, few nodes (25 nodes for GDM cases, 26 nodes for controls) and edges (69 edges for GDM cases, 98 edges for controls) were noted for GDM cases compared with controls (Additional file 3: Table S7). Compared with controls, several taxa including genus 261 of TM7, SCFA producers Actinomyces and Roseburia, and the genus 127 of the GDM-related group Ruminococcaceae lost the most significant correlations with other taxa in the GDM network (Additional file 3: Table S8).

Gene prediction suggested little functional differences

The PICRUSt program was used to predict functional capacities of the gut microbiota from community 16S rRNA gene amplicons based on 9852 features, and 328 predicted functional categories were identified. Although we found the overall gut microbial community structures were significantly different between GDM cases and controls in earlier analyses, we did not find significant differences in the overall predicted functional structures indicated by NMDS (Additional file 2: Figure S6). Nevertheless, we further investigated the possibility of biomarkers in functional genes by going through individual biomarker identification analysis pipeline and implemented LEfSe to identify predicted functional categories at KEGG level 1, 2 and 3 that could discriminate GDM cases from controls. From the categories that made up KEGG level 1, genetic information processing showed significantly lower abundance in GDM cases versus controls. At KEGG level 2, we detected 39 functional categories in total. Of them, 3 were significantly over-represented in GDM cases, including functional categories of metabolism of cofactors and vitamins, cellular processes and signaling, and metabolism of other amino acids. Meanwhile, 5 were found enriched in controls, including functional categories of replication and repair, translation, nucleotide metabolism, metabolism of terpenoids and polyketides, and signaling molecules and interaction (Additional file 3: Table S9, Additional file 2: Figure S7). Of them, 3 were over-represented in the GDM group including pores ion channels, other ion-coupled transporters and membrane and intracellular structural molecules. The rest 5 predicted functional categories enriched in controls mainly included pyrimidine metabolism, purine metabolism, ribosome, aminoacyl-tRNA biosynthesis and DNA repair and recombination proteins. Collectively, we did not find any strong evidence or clear trends in predicted functional gene profiles to discriminate GDM cases from controls.

Prospective associations between microbial taxa and risk of GDM

We compared relative abundances of 41 gut microbial taxa of interest (identified through earlier procedures) between GDM cases and controls on all taxonomic levels. Higher relative abundance of 4 microbial taxa showed statistical associations with lower risk of GDM (Table 2). We found that greater abundance of Actinobacteria was associated with decreased risk of GDM (multivariable-adjusted OR per percent increase of relative abundance, 0.94 and 95% CI [0.90 to 0.98]). The family of Coriobacteriaceae was also associated with decreased risk of GDM (OR, 0.79; 95% CI, 0.68 to 0.91). In addition, genus 26 (OR, 0.47; 95%, 0.29 to 0.77) of Coriobacteriaceae and an unknown specie of the Coprococcus (OR, 49%; 95%, 0.30, 0.81) was respectively 53% and 51% less likely as to confer GDM risk for each 1 percent increase of relative abundance.

In this study, we prospectively examined whether the gut microbiota in pregnant women during early pregnancy already showed signs of aberrance before GDM was diagnosed. β-diversity analyses showed that the gut microbiota already showed signs of abnormality during early pregnancy for GDM cases, with overall gut microbial structures significantly different from controls. Although several recent studies reported reduction of α-diversity in GDM women during early pregnancy[7, 9, 10], while no or only marginal significant β-diversity differences were observed. Since the gut microbiota can be influenced by multiple factors including diet, lifestyles, and medication etc., it is critical to get a well study design, recruit comparatively large number of samples, and explore comprehensive analyses along with clinical data. According to [14], the sample size in our study can yield 0.95 statistical power.

We found gut SCFA producing Actinobacteria significantly decreased in GDM cases and were associated with GDM during early pregnancy at multiple taxonomic levels (Actinobacteria, Coriobacteriaceae, and an unknown genus of Coriobacteriaceae). The role of Actinobacteria in GDM cases could be complex since it included multiple notable SCFA producing bacteria as well as some inflammation related taxa, which were frequently reported to be elevated in the gut of patients with diabetes or related metabolic disorders. At family level, Actinobacteria includes beneficial bacteria (eg. SCFA producing bacteria Bifidobacteriales and Coriobacteriaceae), and their abundance was under-represented in GDM cases in our study. We also found reduced abundance of Actinobacteria members in GDM women at genus level: Rothia, Actinomyces, Bifidobacterium, and Adlercreutzia, most of which were well known SCFA producing and anti-inflammatory microbial groups [15–18], and were previously found to be reduced in patients with type 2 diabetes mellitus [19, 20] or GDM women in the second trimester [5]. Similarly, another study also reported an overall decrease of Actinobacteria in GDM cases compared with controls during early pregnancy, although with a very small sample size (n = 15) [21]. Since Actinobacteria was reported to correlate positively with inflammatory cytokines interleukin-6 (IL-6) and tumor necrosis factor alpha (TNF-α) levels [22], as a whole phylum it could be a good target for treatment to prevent host from developing GDM. However, other conflicting results from previous studies reported over-represented Actinobacteria in aberrant gut microbiota of pregnant women [6, 21] and type 2 diabetes mellitus patients [23, 24]. This may due to elevated Collinsella of Actinobacteria that have potential deleterious effects on regulation of glucose metabolism in middle and late stage of pregnancy [6, 21, 25], while such deleterious members were not significantly different between GDM cases and controls in our study. Nevertheless, the gut microbiota was for early pregnancy in our study when inflammation was not as obvious as during middle and late pregnancy. The role of Actinobacteria interacting with other microbial groups may be changing dramatically throughout the whole pregnancy.

Another gut microbial member that was associated with protective effect for GDM was an unknown species of Coprococcus, which was considered a SCFA producing microbial group in the family of Lachnospiraceae. Although there was not much information on whether Coprococcus or Lachnospiraceae were related to GDM pathogenesis, previous studies showed multiple hormones during early pregnancy and inflammatory markers in diabetic patients strongly correlated with them. For instance, adipokine levels were reported to strongly correlate with Lachnospiraceae, while gastrointestinal polypeptide was positively correlated with the Coprococcus [25].

Besides the main results from final conditional Logistic regression analyses, we also found signs of gut microbial dysbiosis indicated by significantly differentiated relative abundances in other taxa. For instance, SCFA producing Anaerostipes was under-represented in GDM group and has been previously found nominally associated with type 2 diabetes mellitus [26], which was elevated along with alleviation of inflammation in animal models [27, 28]. On the other hand, we found several taxa including Ruminococcaceae spp., Veillonellaceae and opportunistic pathogens Enterobacteriaceae of Proteobacteria were significantly over-represented in GDM cases. Nevertheless, such differences discovered at such a high taxonomic level (family) still requires further investigation to find clinical meanings.

While there were other microbial groups considered discriminatory between diabetic and healthy controls in other studies, such as human pathogen Staphylococcus aureus and SCFA-producing taxa Clostridium, Roseburia, and Akkermansia, none of them were discriminatory based on relative abundance data between cases and controls in our study. Staphylococcus aureus was frequently reported to correlate with obesity and overweight during pregnancy [4, 32, 35, 36], while Clostridium and Roseburia were commonly seen reduced in GDM and aberrant gut microbiota [5, 37]. To eliminate the influence of absolute load of total microbes on health, we used Differential Ranking [38] analysis to compensate for analyses relying solely on relative abundance data, which could lead to false negative results [38]. We found the log ratios of relative abundances on several lactate and butyrate-producing microbial groups (Clostridium, Roseburia and Coriobacteriaceae) to Staphylococcus in GDM cases were significantly lower compared with controls and significantly positively correlated with OGTT values during middle pregnancy. These findings suggested that it may be important to include additional estimators besides relative abundance to discriminate cases and controls. However, our study had a relatively low prevalence of Staphylococcus across samples, and thus future investigations are needed on this potential target for GDM prevention.

Besides the findings in aberrant structure of gut microbiota, we investigated the interactions among microorganisms through co-occurrence network analysis, which may yield additional information on how gut microbiota of GDM women differentiated from the healthy controls during early pregnancy. We found the GDM network had much less significant correlations (edges) among microorganisms. In the healthy human gut, different microbial groups tended to interact more closely with each other, either positively or negatively, which is important for maintaining the stability of ecological equilibrium. However, in dysbiosis state, a lot of interactions among gut microorganisms tend to be lost, which was also reported in other metabolic disorders [39]. Our results indicated that women who developed GDM later already held less stable gut microbiota during early pregnancy, which could be an additional important signature for pre-disease state.

Our study has certain limitations while also important strengths in discovering the association of gut microbiota with GDM before the disease is diagnosed. The limitations include the use of 16S rRNA gene amplicon sequencing technology rather than shortgun metagenomics sequencing, which may give us direct functional gene analyses in addition to microbial structural characteristics. Another limitation is inherent with the population of Tongji-Shuangliu Birth Cohort, in which our findings may only well represent for the local residents thus further investigation is needed to validate its applicability to other population. It would also be good to include more time points at middle and late pregnancy to explore the changes in microbiota during pregnancy. On the other hand, our study has a number of strengths: 1) the prospective cohort was well designed and managed with adequate background information of the participants, and the nested case-control gut microbiota research was carefully implemented; 2) our study included so far the largest sample size (201 cases and 201 matched controls) with high statistical power; 3) the field of GDM associated gut microbiota at early pregnancy has just emerged with confusing results, and our well designed study with comprehensive analyses can add promising information in understanding GDM etiology and discovering biomarkers for disease prevention.

In conclusion, our data suggested that as early as during early pregnancy, the gut microbiota of pregnant women who subsequently developed GDM already showed significant signs of aberrancy. Actinobacteria members of Adelercreutzia, Coriobacteriaceae sp., Bifidobacterium, Rothia, and Actinomyces, and an unknown species in Coprococcus of Firmicutes were strongly associated with less risk of GDM. The abundance of Staphylococcus relative to selective SCFA microorganisms (Clostridium, Roseburia and Coriobacteriaceae) could also be potential clinical characteristics targeting GDM, considering their associations with OGTT values at second trimester of gestation. Our study provided promising biological markers that may be used in monitoring pregnant women’s health and developing therapeutic interventions on GDM at early stage of gestation. Future studies throughout the whole gestation targeting microorganisms at species level and community functional structures are still needed to elucidate the functions of gut microbiota in the development of GDM.

Ethics approval and consent to participate

The TSBC study was approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology ([2017]伦审字（S225）-1号), and informed consent was obtained from all participants.

Availability of data and material

All 16S rDNA sequencing data has been deposited in SRA archive under project number: PRJNA637490.

Competing interests

The authors declare that they have no competing interests.

Funding

The study is supported by the National Natural Science Foundation of China (81773517) and the Fundamental Research Funds for the Central Universities (2019kfyXKJC052). PH is supported by Natural Science Foundation of Hubei Province of China (2018CFB197). AP is also supported by the National Key Research and Development Program of China (2017YFC0907504) and the National Natural Science Foundation of China (81930124). WY and JWD were supported, in part, by a grant from STINT from Sweden.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Authors' contributions

An Pan is in charge of the Tongji-Shuangliu Birth Cohort study. This gut microbiota study was designed, interpreted by Ping Hu, Xiongfei Pan, and An Pan. Ping Hu wrote the manuscript, and all other authors contributed to it. Xiuyi Chen carried out the molecular microbiological experiments and did the statistical analyses. Xufeng Chu and Bo Xiong did the bioinformatic and statistical analyses. Yi Ye, Yi Wang, and Chunxia Yang collected fecal samples and helped with project management. Xue Yang, Chunxia Yang and Xiaorong Qi helped with project management and epidemiological data analyses. Maozhen Han, Ning Kang, Justin Debelius and Weimin Ye contributed to bioinformatics analyses and interpretation of data.

Acknowledgements

We would like to thank Emily B Hollister (Texas Children's Microbiome Center, Department of Pathology, Texas Children's Hospital and Baylor College of Medicine, Houston, TX, USA) who helped us on result interpretations.

McIntyre HD, Catalano P, Zhang C, Desoye G, Mathiesen ER, Damm P. Gestational diabetes mellitus. Nat Rev Dis Primers. 2019;5:47.
Zhu WW, Yang HX, Wei YM, Yan J, Wang ZL, Li XL, et al. Evaluation of the value of fasting plasma glucose in the first prenatal visit to diagnose gestational diabetes mellitus in china. Diabetes Care. 2013;36:586-90.
Tilg H, Moschen AR. Microbiota and diabetes: an evolving relationship. Gut. 2014;63:1513-21.
Collado MC, Isolauri E, Laitinen K, Salminen S. Distinct composition of gut microbiota during pregnancy in overweight and normal-weight women. Am J Clin Nutr. 2008;88:894-9.
Kuang YS, Lu JH, Li SH, Li JH, Yuan MY, He JR, et al. Connections between the human gut microbiome and gestational diabetes mellitus. Gigascience. 2017;6:1-12.
Crusell MKW, Hansen TH, Nielsen T, Allin KH, Ruhlemann MC, Damm P, et al. Gestational diabetes is associated with change in the gut microbiota composition in third trimester of pregnancy and postpartum. Microbiome. 2018;6:89.
Zheng W, Xu Q, Huang W, Yan Q, Chen Y, Zhang L, et al. Gestational Diabetes Mellitus Is Associated with Reduced Dynamics of Gut Microbiota during the First Half of Pregnancy. mSystems. 2020;5.
Mokkala K, Tertti K, Ronnemaa T, Vahlberg T, Laitinen K. Evaluation of serum zonulin for use as an early predictor for gestational diabetes. Nutr Diabetes. 2017;7:e253.
Gao B, Zhong M, Shen Q, Wu Y, Cao M, Ju S, et al. Gut microbiota in early pregnancy among women with Hyperglycaemia vs. Normal blood glucose. BMC Pregnancy Childbirth. 2020;20:284.
Ma S, You Y, Huang L, Long S, Zhang J, Guo C, et al. Alterations in Gut Microbiota of Gestational Diabetes Patients During the First Trimester of Pregnancy. Front Cell Infect Microbiol. 2020;10:58.
Chengdu Bureau of Statistics. Chengdu Statistics Book 2019,Chengdu: Chengdu Bureau of Statistics; 2020.
Weinert LS. International Association of Diabetes and Pregnancy Study Groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy: comment to the International Association of Diabetes and Pregnancy Study Groups Consensus Panel. Diabetes Care. 2010;33:e97; author reply e98.
Herlemann DPR, Labrenz M, Jurgens K, Bertilsson S, Waniek JJ, Andersson AF. Transitions in bacterial communities along the 2000 km salinity gradient of the Baltic Sea. Isme J. 2011;5:1571-79.
Mattiello F, Verbist B, Faust K, Raes J, Shannon WD, Bijnens L, et al. A web application for sample size and power calculation in case-control microbiome studies. Bioinformatics. 2016;32:2038-40.
Bordalo Tonucci L, Dos Santos KM, De Luces Fortes Ferreira CL, Ribeiro SM, De Oliveira LL, Martino HS. Gut microbiota and probiotics: Focus on diabetes mellitus. Crit Rev Food Sci Nutr. 2017;57:2296-309.
Cani PD, Neyrinck AM, Fava F, Knauf C, Burcelin RG, Tuohy KM, et al. Selective increases of bifidobacteria in gut microflora improve high-fat-diet-induced diabetes in mice through a mechanism associated with endotoxaemia. Diabetologia. 2007;50:2374-83.
Wilson R, Willis J, Gearry RB, Hughes A, Lawley B, Skidmore P, et al. SunGold Kiwifruit Supplementation of Individuals with Prediabetes Alters Gut Microbiota and Improves Vitamin C Status, Anthropometric and Clinical Markers. Nutrients. 2018;10.
Zhao X, Zhang Z, Hu B, Huang W, Yuan C, Zou L. Response of Gut Microbiota to Metabolite Changes Induced by Endurance Exercise. Front Microbiol. 2018;9:765.
Wei X, Tao J, Xiao S, Jiang S, Shang E, Zhu Z, et al. Xiexin Tang improves the symptom of type 2 diabetic rats by modulation of the gut microbiota. Sci Rep. 2018;8:3685.
Zhang HH, Liu J, Lv YJ, Jiang YL, Pan JX, Zhu YJ, et al. Changes in Intestinal Microbiota of Type 2 Diabetes in Mice in Response to Dietary Supplementation With Instant Tea or Matcha. Can J Diabetes. 2019;44:44-52.
Koren O, Goodrich JK, Cullender TC, Spor A, Laitinen K, Backhed HK, et al. Host remodeling of the gut microbiome and metabolic changes during pregnancy. Cell. 2012;150:470-80.
Pedersen C, Gallagher E, Horton F, Ellis RJ, Ijaz UZ, Wu H, et al. Host-microbiome interactions in human type 2 diabetes following prebiotic fibre (galacto-oligosaccharide) intake. Br J Nutr. 2016;116:1869-77.
Lambeth SM, Carson T, Lowe J, Ramaraj T, Leff JW, Luo L, et al. Composition, Diversity and Abundance of Gut Microbiome in Prediabetes and Type 2 Diabetes. J Diabetes Obes. 2015;2:1-7.
Liu H, Pan LL, Lv S, Yang Q, Zhang H, Chen W, et al. Alterations of Gut Microbiota and Blood Lipidome in Gestational Diabetes Mellitus With Hyperlipidemia. Front Physiol. 2019;10:1015.
Gomez-Arango LF, Barrett HL, Wilkinson SA, Callaway LK, McIntyre HD, Morrison M, et al. Low dietary fiber intake increases Collinsella abundance in the gut microbiota of overweight and obese pregnant women. Gut Microbes. 2018;9:189-201.
Yang Q, Lin SL, Kwok MK, Leung GM, Schooling CM. The Roles of 27 Genera of Human Gut Microbiota in Ischemic Heart Disease, Type 2 Diabetes Mellitus, and Their Risk Factors: A Mendelian Randomization Study. Am J Epidemiol. 2018;187:1916-22.
Zhang Y, Tang K, Deng Y, Chen R, Liang S, Xie H, et al. Effects of shenling baizhu powder herbal formula on intestinal microbiota in high-fat diet-induced NAFLD rats. Biomed Pharmacother. 2018;102:1025-36.
Chen Y, Guo KM, Nagy T, Guo TL. Chronic oral exposure to glycated whey proteins increases survival of aged male NOD mice with autoimmune prostatitis by regulating the gut microbiome and anti-inflammatory responses. Food Funct. 2020;11:153-62.
Serena C, Ceperuelo-Mallafre V, Keiran N, Queipo-Ortuno MI, Bernal R, Gomez-Huelgas R, et al. Elevated circulating levels of succinate in human obesity are linked to specific gut microbiota. Isme J. 2018;12:1642-57.
Martinez-Medina M, Denizot J, Dreux N, Robin F, Billard E, Bonnet R, et al. Western diet induces dysbiosis with increased E coli in CEABAC10 mice, alters host barrier function favouring AIEC colonisation. Gut. 2014;63:116-24.
Raso GM, Simeoli R, Iacono A, Santoro A, Amero P, Paciello O, et al. Effects of a Lactobacillus paracasei B21060 based synbiotic on steatosis, insulin signaling and toll-like receptor expression in rats fed a high-fat diet. J Nutr Biochem. 2014;25:81-90.
Everard A, Belzer C, Geurts L, Ouwerkerk JP, Druart C, Bindels LB, et al. Cross-talk between Akkermansia muciniphila and intestinal epithelium controls diet-induced obesity. Proc Natl Acad Sci U S A. 2013;110:9066-71.
Rabol R, Larsen S, Hojberg PM, Almdal T, Boushel R, Haugaard SB, et al. Regional anatomic differences in skeletal muscle mitochondrial respiration in type 2 diabetes and obesity. J Clin Endocrinol Metab. 2010;95:857-63.
Candela M, Biagi E, Soverini M, Consolandi C, Quercia S, Severgnini M, et al. Modulation of gut microbiota dysbioses in type 2 diabetic patients by macrobiotic Ma-Pi 2 diet. Br J Nutr. 2016;116:80-93.
Dao MC, Everard A, Aron-Wisnewsky J, Sokolovska N, Prifti E, Verger EO, et al. Akkermansia muciniphila and improved metabolic health during a dietary intervention in obesity: relationship with gut microbiome richness and ecology. Gut. 2016;65:426-36.
Santacruz A, Collado MC, Garcia-Valdes L, Segura MT, Martin-Lagos JA, Anjos T, et al. Gut microbiota composition is associated with body weight, weight gain and biochemical parameters in pregnant women. Br J Nutr. 2010;104:83-92.
Allin KH, Tremaroli V, Caesar R, Jensen BAH, Damgaard MTF, Bahl MI, et al. Aberrant intestinal microbiota in individuals with prediabetes. Diabetologia. 2018;61:810-20.
Morton JT, Marotz C, Washburne A, Silverman J, Zaramela LS, Edlund A, et al. Establishing microbial composition measurement standards with reference frames. Nat Commun. 2019;10:2719.
Wang Y, Gao X, Ghozlane A, Hu H, Li X, Xiao Y, et al. Characteristics of Faecal Microbiota in Paediatric Crohn's Disease and Their Dynamic Changes During Infliximab Therapy. J Crohns Colitis. 2018;12:337-46.

Table 1

Baseline characteristics of study participants
Baseline characteristics		Total (n = 402)	GDM status		P value^*
Baseline characteristics		Total (n = 402)	Yes (n = 201)	No (n = 201)	P value^*
Age, years, mean (SD)		28.0 (4.2)	28.2 (4.4)	27.9 (4.0)
Gestational age, weeks, mean (SD)		10.7 (2.0)	10.5 (2.1)	10.8 (1.9)
Pre-pregnancy BMI, kg/m2, mean (SD)		21.5 (3.3)	21.9 (3.2)	21.2 (3.4)	0.04
Cigarette smoking, n (%)					0.64
	Current/former	20 (5.0)	11 (5.5)	9 (4.5)
	Never	382 (95.0)	190 (94.5)	192 (95.5)
Alcohol drinking, n (%)					0.02
	Current/former	72 (17.9)	45 (22.4)	27 (13.4)
	Never	330 (82.1)	156 (77.6)	174 (86.6)
Intake of meat, gram/day, n (%)					0.66
	0	47 (11.7)	21 (10.5)	26 (12.9)
	≤ 50	99 (24.6)	48 (23.9)	51 (25.4)
	≤ 100	137 (34.1)	68 (33.8)	69 (34.3)
	> 150	119 (29.6)	64 (31.8)	55 (27.4)
Intake of vegetables, gram/day, n (%)					0.94
	≤ 50	99 (24.6)	47 (23.4)	52 (25.9)
	≤ 100	69 (17.2)	35 (17.4)	34 (16.9)
	≤ 150	57 (14.2)	28 (13.9)	29 (14.4)
	> 150	177 (44.0)	91 (45.3)	86 (42.8)
Intake of fruits, gram/day, n (%)					0.63
	≤ 50	35 (8.7)	18 (9.0)	17 (8.5)
	≤ 100	45 (11.2)	23 (11.4)	22 (11.0)
	≤ 150	26 (6.5)	16 (8.0)	10 (5.0)
	> 150	296 (73.6)	144 (71.6)	152 (75.6)
Family history of diabetes, n (%)					0.35
	Yes	31 (7.7)	18 (9.0)	13 (6.5)
	No	371 (92.3)	183 (91.0)	188 (93.5)
History of GDM, n (%)					< 0.001
	Yes	13 (3.2)	13 (6.5)	0 (0.0)
	No	389 (96.8)	188 (93.5)	201 (100.0)
^*P value was estimated through univariable conditional logistic regression. P values for matching factors were not provided.

Table 2

Selected microbial taxa showing significant associations with incident GDM
Taxon*	Median relative abundance (%)		OR (95% CI)†	P value†	P value‡
	GDM status
	Yes(n = 201)	No (n = 201)
Phylum
Actinobacteria	2.34	3.76	0.94 (0.90, 0.98)	0.002	0.034
Family
Coriobacteriaceae	0.78	1.17	0.79 (0.68, 0.91)	0.001	0.034
Genus
Coriobacteriaceae;g_ unknown (genus26)	0.06	0.11	0.47 (0.29, 0.77)	0.003	0.034
Species
Coprococcus;s_unknown	0.00	0.00	0.49 (0.30, 0.81)	0.005	0.049
*Taxa that showed significant associations in 41 selected taxa from previous analyses in the study. Microbial taxa with mean relative abundance > 0.0001% and prevalence > 25%.
†ORs, 95% CIs and p values for 1% increase in relative abundance were calculated from conditional logistic regression models after controlling for age (years), gestational week at sampling (weeks), prepregnancy BMI (kg/m²), smoking status (current/ever and never in the past year), alcohol consumption status (current/ever and never in the past year), intakes of meat (0, ≤ 50 g, ≤ 100 g, and > 150 g), vegetables (≤ 50 g, ≤ 100 g, ≤ 150 g, and > 150 g), and fruits (≤ 50 g, ≤ 100 g, ≤ 150 g, and > 150 g), family history of diabetes (yes and no), and history of gestational diabetes (yes and no).
‡ False discovery rate corrected P values.

Download PDF

Version 1

posted

You are reading this latest preprint version

Association Of Gut Microbiota During Early Pregnancy With Risk Of Incident Gestational Diabetes Mellitus: A Nested Case-Control Study

Status:

Version 1

Abstract

Figures

Introduction

Research Design And Methods

Results

Discussion

Conclusion

Declarations

References

Tables

Supplementary Files

Status:

Version 1