Background
Pregnancy is a fascinating biological process that involves simultaneous changes physiologically, some of which have well been established, such as metabolic and hormonal alterations. However, only in the last decade has the importance of the gut microbiota in pregnancy been recognized [
1].
There are millions of bacteria present in the gut, the majority of which are commensals. Although the actual composition of the gut microbiota is unclear, current research has revealed that 80–90% of bacteria morphologies belong to two phyla: Bacteroides and Firmicutes [
2]. In addition to nutritional consumption, antibiotics, stress, and obesity; pregnancy has been proven to cause alteration in the gut microbiota composition.
The alteration in gut microbiota composition in pregnancy is accompanied by weight gain, insulin insensitivity, and increased cytokines that suggest inflammation. All of these changes are similar to those reported in people with metabolic syndrome [
3]. These modifications were thought to be essential to accommodate the normal pregnancy demand.
To date, there is still a lack of information on gut microbiota profile among the pregnant Malaysian women population. With the rising trend of obesity among women of reproductive age, it is crucial to understand the composition of the gut microbiota as they are at risk of developing gestational diabetes later.
Knowledge on gut microbiota composition will allow it to be used as a platform to explore the role of modulation of the gut microbiota as a preventive and therapeutic tool in the treatment of gestational diabetes. Even though gut microbiota pattern has been reported in other countries, contrasting ethnic, cultural and dietary practices have been associated with different gut microbiota profile [
4]. Hence, this study is crucial to observe whether there is any discrepancy with the published findings.
Therefore, this study aimed to determine the gut microbiota composition in the T1 and T3 among pregnant Malaysian women, to demonstrate its composition between women with gestational diabetes mellitus (GDM) and non-GDM (NGDM) and in different BMI categories.
Methods
Study design
This was a prospective observational study involving 38 women in a tertiary medical centre in Malaysia. All pregnant women who attended the antenatal clinic as outpatient that met the inclusion criteria were offered to participate. The inclusion criterias were: (i) Pregnant patients in the first trimester (T1); (ii) Malaysian; (iii) Be willing to be followed up until the third trimester (T3) and (iv) Agreeable to undergo Oral Glucose Tolerance Test (OGTT). The exclusion criteria were: (i) any known case of preexisting diabetes mellitus, metabolic syndrome or any other endocrine disorders and (ii) on any antibiotics /prebiotics/ probiotics during or in the past four weeks prior to recruitment.
The first trimester was defined as any pregnancy less than 13 weeks of gestation and the third trimester was any pregnancy beyond 27 weeks of gestation. GDM was diagnosed based on the OGTT result, a diagnostic test for GDM recommended by the national guideline. It was performed in the antenatal clinic using 75 g oral glucose. A fasting blood sample will be taken, followed by another blood sample taken two hours after consuming the oral glucose drink prepared (which they need to complete it within five minutes). If either fasting blood glucose is more than 5.1 mmol/L or 2-h post-prandial glucose is more than 7.8 mmol/L, they were diagnosed as GDM. Others were classified as non-GDM (NGDM).
Sample size calculation
The sample size is determined based on the study by Collado et. al. (2008), who found the Bacteroides-Prevotella group count in fecal samples at first trimester was 9.74 (9.62, 9.87) log fecal cells/g and the Bacteroides-Prevotella group count in fecal samples at third trimester was 10.36 (10.27,10.45) log fecal cells/g. By taking α = 0.05, 80% power of the study, the standard deviation for T1 was 0.12, the standard deviation for T3 was 0.09, and estimated mean difference of 0.62, the sample size required for this study is 28 using the following formula:
$$\begin{array}{l}\mathrm{n}\hspace{0.17em}=\hspace{0.17em}{({\mathrm{Z}}_{\mathrm{\alpha }}\hspace{0.17em}+\hspace{0.17em}{\mathrm{Z}}_{\upbeta })}^{2} \frac{{\left({\upsigma }_{1}+{\upsigma }_{2}\right)}^{2}}{\mathrm{d}}\\ {(9.74\hspace{0.17em}+\hspace{0.17em}10.36)}^{2} \frac{{\left(0.12+0.09\right)}^{2}}{0.62}\\ \begin{array}{l}=\hspace{0.17em}404.01 (0.0711)\\ \hspace{0.17em}=\hspace{0.17em}28\end{array}\end{array}$$
By adjusting the 10% attrition rate, the minimum sample size in this study is 31.
Data collection
The participants were asked to fill in a study proforma enquiring the participants’ basic demographic details. Anthropometric measurements were taken by trained nursing staff. Body Mass Index (BMI) was calculated and participants were categorised based on the World Health Organisation recommendation; underweight (below 18.5 kg/m2), normal (18.5–24.9 kg/m2), pre-obese (25.0–29.9 kg/m2) and obese (30 kg/m2 and above). The participants were then asked to give their stool samples during the first trimester. Sample collection, preservation and storage were performed using Stool Nucleic Acid Collection and Preservation Tube (NORGEN, Canada). A total of 2 g samples were collected and filled into the collection tubes, gently mixed until the stool is well submerged under the liquid preservative. They were required to have an OGTT test at least once during this pregnancy. Once they reach the third trimester, they were again asked to give another fecal sample using the same kit. Patients were followed up until delivery, and delivery details was be obtained including the mode of delivery and the baby’s anthropometric measurement.
Total DNA of the stool samples was extracted from approximately 400 µl of liquid samples by Stool DNA Isolation Kit (NORGEN, Canada) following the manufacturer's instruction. The final DNA concentration and purity were determined by SpectraMax QuickDrop Micro-Volume Spectrophotometer (Molecular Devices, USA). The ratio of sample absorbance at 260 and 280 nm was used to assess the purity of the DNA. The DNA integrity was assessed by running a 1% agarose gel electrophoresis (Sigma-Aldrich, USA) and stained with SYBR Safe DNA Gel Stain (Invitrogen, USA). Extracted DNA was stored at -20˚C pending sequencing analysis.
16S ribosomal ribonucleic acid metagenome analysis
The V3-V4 hypervariable regions of the bacteria 16S rRNA gene were amplified with a set of primers 338F (5′-ACTCCTACGGGAGGCAGCA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) by thermocycler PCR system (27 cycles for each sample) (GeneAmp 9700, ABI, USA) according to the standard protocols by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The PCR reactions were conducted using the following conditions: three minutes of denaturation at 95 °C, 27 cycles of 30 s at 95 °C, 30 s for annealing at 55 °C, and 45 s for elongation at 72 °C, and a final extension at 72 °C for 10 min.
PCR amplification was performed using TransStart Fastpfu DNA Polymerase (TransGen AP221-02) under 20 μl reaction containing 4 μL of 5 × FastPfu Buffer, 2 μL of 2.5 mM dNTPs, 0.8 μL of each primer (5 μM), 0.4 μL of FastPfu Polymerase, 0.2 μL BSA and 10 ng of template DNA. The PCR products were detected by gel electrophoresis in 2% agarose gel. Amplicons were extracted from the agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (AxygenBiosciences, USA) and quantified using QuantiFluor™-ST (Promega, USA) following the manufacturer's protocol.
The sample libraries were pooled in equimolar and paired-end sequenced (2 × 300) on an Illumina MiSeq platform (Illumina, USA) according to the standard protocols by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). Assembly, binning, and annotation of DNA sequences were performed. Raw fastq files were demultiplexed, quality-filtered using QIIME (version 1.9.1) with the following criteria: The 300 bp reads were truncated at any site receiving an average quality score < 20 over a 50 bp sliding window, discarding the truncated reads that were shorter than 50 bp, exact barcode matching, two nucleotide mismatch in primer matching, reads containing ambiguous characters were removed and only sequences that overlap longer than 10 bp were assembled according to their overlap sequence. Reads which could not be assembled were discarded. The taxonomy of each 16S rRNA gene sequence was analyzed by RDP Classifier (
http://rdp.cme.msu.edu/) against the Silva (SSU123) 16S rRNA database using a confidence threshold of 0.7. OTU-level species accumulation curve was used to assess the sequencing depth and species richness from the result of sampling. Alpha diversity indices, including Chao 1 richness, Abundance-based Coverage Estimator (ACE) metric, Shannon-Weiver curve and Simpson Index were calculated using Mothur.
Statistical and comparative metagenomics analysis
Clinical baseline characteristics are presented as mean ± standard deviation. Spearman's rank correlation coefficient analysis was carried out. All statistical analyses were carried out using SPSS version 22 (IBM Corp., Armonk NY, USA). Statistical significance was defined as a
P‐value < 0.05. Comparative metagenomics analysis of the 16S rRNA gut microbiota profiles were performed between different gestational trimesters (T1 vs T3), GDM status (GDM vs NGDM), BMI (Normal vs Abnormal) and also among the BMI subgroups (Underweight, Normal, Pre-obesity, Obesity) using METAGENassist [
5].
Row-wise normalization by sum was performed on the bacterial relative abundance data matrix to normalize the inherent differences within metagenomes (sequencing depth). Column-wise normalization by log10 transformation was employed to obtain a more normal/Gaussian distribution of each bacterial taxa before statistical analysis is performed. Univariate statistics such as Student T-test and Anova with Post hoc Fisher’s LSD test with the significant P value less than 0.05 were used to determine any significant differences in the abundance of each phylum and genera between trimesters.
Multivariate analysis using the supervised model PLSDA of
β-diversity was used to reveal any similarity or clustering pattern in the community structure between the gestational trimester, GDM status and BMI groups. The performance of the discriminant pattern from the PLSDA model was evaluated based on R
2 values (less than 0.33, weak; 0.33–0.67, moderate; 0.67 and above, substantial model [
6]. Loading plots from PLSDA and variable importance in projection (VIP) were used to determine the importance of each phylum and genus in each community profile.
Discussion
Pregnancy is often associated with an increase in bacterial load and dramatic changes in the taxonomic composition of the gut microbiota [
3,
5,
6]. These substantial changes are manifested by decreased individual richness (α-diversity), increased inter-subject diversity (β-diversity), and shifts in the abundance of certain species [
3].
The majority of the alterations were seen between the non-pregnant women or pregnant women in early pregnancy and those from advanced pregnancy. Hence, we believe our study comparing the pregnant women in T1 and T3 would add valuable findings in this area. However, in general, our findings indicate that pregnancy progression from T1 to T3 was not related to a substantial alteration in the variety and composition of pregnant women's gut microbiota biodiversity. The mean number of the observed OTUs and α-diversity (individual richness) indices (Ace, Chao, Shannon and Simpson) between the trimester of pregnancy was not significantly different. This contradicts the findings by Koren O. et al. (2012) that reported the presence of a significant reduction in within-subject diversity [
3]. However, DiGiulio D. B. et. al. (2015) also found no significant trend in the Shannon diversity index of the gut microbiota composition during pregnancy [
7].
The prevalence of GDM among our participants was 31.6%, similar to the prevalence of GDM in Malaysia which was reported to be 27.9% in 2017 [
8]. Similarly, there was no statistically significant difference in gut microbiota α-diversity between pregnant women with GDM and NGDM, and between pregnant women with different BMI groups. However, a trend of relatively lower α-diversity indices (Ace, Chao, Shannon and Simpson) was observed in the gut microbiota profiles of GDM than in NGDM pregnant women. Likewise, there was a trend of lower α-diversity indices in pre-obese and obese pregnant women than women with normal BMI.
Our study also demonstrates that there was no statistically significant difference between the abundances of gut microbiota phyla and genera between T1 and T3. Similar to previously reported in the literature, the most dominant phyla were
Firmicutes,
Bacteroidetes, Proteobacteria and
Actinobacteria [
3]
Actinobacteria and
Proteobacteria both shows an increasing trend, and
Faecalibacterium shows a decreasing trend, despite it was not statistically significant. Whereas
Bacteroides, Alistipes, Faecalibacterium and
Collinsella were identified as the dominant bacterial genus in both community structures of gut microbiota in T1 and T3.
Proteobacteria are often associated with inflammatory conditions [
9]. Interestingly, with the findings of significantly higher levels of the proinflammatory cytokines IFN-g, IL-2, IL-6, and TNF-a in T3 than in T1, this would suggest that the T3 mucosal surfaces of the gastrointestinal tract present low-grade inflammation in advanced gestation [
3].
Faecalibacterium, a butyrate producer with anti-inflammatory properties that is deficient in inflammatory bowel disease [
10], and in patients with metabolic syndrome [
11] is less prevalent in women with a normal pregnancy in the third trimester [
3]. Even though the reduction of Faecalibacterium in this study was not statistically significant, it did show a reducing pattern.
This aberrant gut microbiota dysbiosis toward the third trimester of pregnancy was reported to be related to adiposity, low-grade inflammation, insulin resistance, and hyperglycemia independent of GDM status [
3]. Its similarity with the changes associated with metabolic syndrome, however, seems to be a requirement of a normal healthy pregnancy.
Our study also did not find any significant difference at phylum level between women with GDM and NGDM; which similar to previously published studies [
12,
13]
Bacteroidetes was the dominant phyla in the GDM group throughout their pregnancy, and there is no predominantly Firmicutes trend as demonstrated in overall subjects or in NGDM group.
On the contrary, specific differences between GDM and normoglycemic women were reported by a few studies. The increased gut abundance of
Parabacteroides distasonis, Klebsiella variicola, Ruminococcus, Eubacterium, Prevotella, Collinsella, Rothia, Desulfovibrio, Actinobacteria, Firmicutes and reduced gut richness of
Methanobrevibacter smithii, Alistipes species,
Bifidobacterium species,
Eubacterium species,
Akkermansia, Bacteroides, Parabacteroides, Roseburia, and
Dialister were reported in GDM patients compared to normoglycemic controls [
14].
However, at the genus level, apart from demonstrating Bacteroides and Faecalibacterium were the dominant genera representing more than 50% of the gut microbiota community structure in both GDM and NGDM groups, there was a discriminant pattern observed between GDM-associated and NGDM-associated gut microbiota (R2 = 0.59).
Among the 15 key genera with the highest VIP score (> 1.5) that contributed to the observed discriminant pattern of gut microbiota associated with gestational diabetes,
Acidaminococcus, Clostridium, Megasphaera and
Allisonella were found significantly higher, while
Barnesiella and
Blautia were found significantly lower in women with GDM. An elevation of bacterial genera from the class of Negativicutes such as
Acidaminococcus, Megasphaera and
Allisonella in GDM is also seen abundant in type 2 diabetes mellitus patients [
15,
16]. The difference in gut microbiota could presumably be related to the metabolic changes during pregnancy, or perhaps, it could be due to distinct lifestyle and eastern dietary habits such as high carbohydrate and fat intake, and low fiber intake during the pregnancy. This suggestion was made as all the subjects had no known pre-pregnancy case of diabetes, metabolic syndrome or any other endocrine disorders. The findings may suggest the metabolic roles of these bacteria in adiposity, low-grade inflammation, insulin resistance, and hyperglycemia independent which required functional study confirmation.
Further analysis looking at the BMI of the participants in this study demonstrated a lower α- diversity among women who were obese, followed by pre-obese compared to women with normal BMI, even though this observation was not statistically significant. This was also observed by Koren O. et. al. (2012) who found that the women who were obese prior to pregnancy had the lowest within-subject α-diversity at both T1 and T3, although this was not significantly different from normal-weight women [
3].
The gut microbiota's role in the pathogenesis of obesity has been clarified through studies in both humans and animal models [
17]. However, no statistically significant differences in the relative abundances of each phylum between the BMI groups in this study.
Bacteroidetes was found to be the most dominant phyla in obese (54.7%) and pre-obese (46.7%) groups. Bacteroidetes is a gram-negative bacteria, which is the largest contributor to lipopolysaccharides (LPS) production. As a result, increasing Bacteroidetes abundances during pregnancy may cause higher inflammation [
18]. LPS can trigger inflammation through the Toll-like receptor 4 (TLR4) signalling pathway in preeclampsia [
19], hence this could explain why obesity increases the risk of pre-eclampsia. This finding is in accordance with findings reported by Zhang et al. (2009) where they found there were more Bacteroidetes in the obese subjects than subjects with normal BMI 20]. This contradicts the earlier findings reported by Ley et. al. (2006) that obese people had lower Bacteroidetes and more Firmicutes than did lean control subjects [
21]. Whereas Duncan et al. (2008) did not detect any differences between obese and non-obese individuals in terms of the proportion of Bacteroidetes measured in the fecal samples, and no significant changes in the percentage of Bacteroidetes occurred in feces from obese subjects even upon weight loss [
22].
The generated PLSDA score plot, according to the BMI, showed a discriminant pattern of gut microbiota between two overlapping clusters. It demonstrated that in the obese group,
Megamonas, Succinatimonas and
Dialister were elevated whereas
Oscillibacter, Oscillospira, Butyricimonas, Alistipes, and
Prevotella were reduced. Two genera which are
Barnesiella and
Blautia were found reduced in both obese and pre-obese profiles. We could suggest that obese BMI gut microbiota during pregnancy is enriched with bacteria from a class of Negativicutes and Proteobacteria such as Megamonas, Succinatimonas and Dialister. An elevation of Negativicutes is also seen in the GDM profile from this study and observed in type 2 diabetes mellitus patients [
15].
In a study of 81 stool samples from Taiwanese for analysis of the association between the gut flora and obesity, they found the most abundant genera of bacteria in cases with a BMI ≥ 27 were Bacteroides (29%), Prevotella (21%), Escherichia (7.4%), Megamonas (5.1%), and Phascolarctobacterium (3.8%) [
23]. Similar dominance of Megamonas was demonstrated, however other bacterial dominance pattern was not the same. Megamonas also was found to be significantly higher in a study among obese children [
24].
Whereas the normal and mild underweight BMI gut microbiota during pregnancy are enriched with bacteria from the class of Clostridia (
Papillibacter, Oscillibacter, Oscillospira, Blautia, Dorea) and Bacteroidia (
Alistipes, Prevotella, Paraprevotella). Some genus from Clostridia such as
Oscillospira has been associated with leanness and low BMI as seen in 6376 participants from the Guangdong Gut Microbiome Project, China [
25]. Several other animal and human studies also found a correlation between the high abundance of
Oscillospira with lower BMI in lean mice and human subjects from several populations such as Colombia, USA, and Europe [
26‐
30]. Nevertheless, our findings suggest there are potential metabolic links between Negativicutes, Clostridia and Proteobacteria with host parameters such as body weight which required further investigation.
Limitation
This study has a few limitations that need to be considered. It has been reported that gut microbiota also influenced primarily by dietary intake [
31,
32]. Lack of dietary information in this study, except for the absence of dietary modification by prebiotics, probiotics or antibiotics for 4 weeks as part of recruitment criteria, leads to an inability to correlate the gut microbiota profile with the dietary intake. This study also was not designed to match a disease and control, but rather an observational study to explore the gut microbiota profile among Malaysian pregnant women which has not yet been reported till date. Finally, the recruitment of the participant and faecal sample was done in both the first and trimester. Despite it covered the whole period of trimester which spanned three months, this was an acceptable method of data collection due to the variation of the gut microbiome within a trimester is negligible [
3].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.