The association between the maternal diet and the maternal and infant gut microbiome: a systematic review

Abstract Abstract During pregnancy, changes occur to influence the maternal gut microbiome, and potentially the fetal microbiome. Diet has been shown to impact the gut microbiome. Little research has been conducted examining diet during pregnancy with respect to the gut microbiome. To meet inclusion criteria, dietary analyses must have been conducted as part of the primary aim. The primary outcome was the composition of the gut microbiome (infant or maternal), as assessed using culture-independent sequencing techniques. This review identified seven studies for inclusion, five examining the maternal gut microbiome and two examining the fetal gut microbiome. Microbial data were attained through analysis of stool samples by 16S ribosomal RNA gene-based microbiota assessment. Studies found an association between the maternal diet and gut microbiome. High-fat diets (% fat of total energy), fat-soluble vitamins (mg/d) and fibre (g/d) were the most significant nutrients associated with the gut microbiota composition of both neonates and mothers. High-fat diets were significantly associated with a reduction in microbial diversity. High-fat diets may reduce microbial diversity, while fibre intake may be positively associated with microbial diversity. The results of this review must be interpreted with caution. The number of studies was low, and the risk of observational bias and heterogeneity across the studies must be considered. However, these results show promise for dietary intervention and microbial manipulation in order to favour an increase of health-associated taxa in the gut of the mother and her offspring.


Introduction
Advancements in the past decade in next-generation sequencing and associated bioinformatics analyses have facilitated a more in-depth study of the human gut 'microbiome'; a word coined to describe the overall community of microorganisms in the gastrointestinal tract (1) . Links between the microbiome and many physiological conditions of the associated host have been made (2; 3; 4) . The various components contributing and modulating the microbiome are yet to be truly defined, however, environmental factors such as lifestyle and diet have come to the fore (5; 6) .
Diet and dietary patterns have been shown to rapidly alter microbial diversity and in turn influence host physiology (7; 8) . In non-pregnant cohorts the dietary macronutrients fat and fibre have most commonly been demonstrated to be able to cause a shift in microbial diversity, with fibre consumption associated with beneficial effects (9; 10; 11) .
With respect to dietary patterns, the Mediterranean diet, the Western diet, low-fat and high fibre diets have been examined in greatest detail, with some research showing a Western diet to influence the gut microbiome more considerably than BMI (8; 12) . Diets high in fibre have been shown to have the ability to increase the relevant abundance of SCFA-producing bacteria (13) . This is in contrast to diets rich in animal fats, high in saturated fat and protein, which have been shown to have a negative impact (14) . The blueprint for the optimal gut microbiome is still unknown but the negative association of decreased diversity is commonly observed. Decreased diversity is linked to a phenomenon called dysbiosis (a disruption of normal gut microbiota); diversity is involved in the survival and adaptability of any ecosystem, the microbiome being no exception (15) . Furthermore, diets such as the Western diet are associated with decreased microbial diversity (12; 16) .

Accepted manuscript
Diversity is the method used to assess the gut microbiome. Alpha diversity (also described as the intra-personal variation) is the individual's diversity in the microbiota. It has been suggested that a higher alpha diversity correlates with a healthier microbiome (17; 18) . As for many ecosystems, a high species diversity is linked with greater resistance to dysbiosis (disruption of microbiota composition from outside normal ranges) and an overall health within the host (19) . Beta diversity on the other hand, describes the interpersonal variation of microbial composition and can be based on collapsing all microbial data to a single co-ordinate point and measuring the distance (using various metrics e.g., Bray-Curtis, un-weighted and weighted UniFrac, Euclidean) between this point and another, usually another participant, person or collection site.
In pregnancy, the gut microbiome is thought to be dynamic with a change seen in first trimester diversity compared to that of the third trimester (20) .
Mode of delivery, pre-term birth, breastfeeding and maternal diet have been identified as important factors that directly influence the composition of the neonatal gut microbiota (21) . Likewise, the presence of furry pets in the home has been shown to influence the composition of the gut microbiota of newborns (22) .
There is limited literature examining the association between maternal macronutrient and micronutrient intake and infant and maternal gut microbiome. Without this knowledge, it is impossible to develop a therapeutic improvements in infant and maternal health.
The aim of this systematic review was to summarise current evidence relating to the association between maternal diet in pregnancy and both the maternal and neonatal gut microbiome.

Protocol and registration
PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) Statement for reporting on systematic review was followed (23) .
Search checklist of items to include, methods, strategy, study selection process, a risk of bias tool and summary measures were used and reported.

Eligibility criteria
To be included in the review the studies had to be observational or crosssectional in design, subjects needed to be pregnant women and/or infants within the first 6 weeks post-partum. The study needed to include a formal dietary analysis during pregnancy and use a culture-independent sampling technique to assess the gut microbiome. Studies had to include details of ages, ethnicities, and demographic characteristics of the women/infants. Studies that evaluated the effect of dietary supplementation or probiotic use only, without formal dietary assessment, were excluded, as were book chapters, online abstracts and conference proceedings were not included.
Articles had to be published in English and no time restrictions was imposed

Outcomes
The main outcomes examined in this review were the maternal or neonatal gut microbiome composition and diversity, as assessed by culture independent sequencing techniques. These outcomes are expressed as microbial diversity in terms of both alpha (intra-individual variation) and beta (inter-individual variation) diversity and relative abundance of specific microbes. Indices such as Shannon's index, whole tree phylogenetic diversity and Simpson's index which measure diversity within microbial communities or Unifrac distances, Bayesian models or PCoA which measure diversity between microbial communities.

Information Sources:
The following five electronic databases were searched; MEDLINE (PubMed), Cochrane Library, Web of Science, CINAHL and Ovid. The last search was conducted 7 th of October 2019.
Search terms were identified by initial scoping searches, and then adjusted depending on the electronic database searched, to better match the key words and indexing terms of each database, and align with MeSH terms.

Summary measures
It was not possible to carry out a summary analysis or meta-analysis for this systematic review due to heterogeneity across the included studies. This included differences in stage of pregnancy of participants, the stool sample analysed, the dietary assessment tool used, and the method of microbiota analysis. An overall description of individual results is therefore provided in the results section, separated into two sections: maternal gut microbiota; neonatal gut microbiota.

Results
Identified articles were added to a reference manager software package (EndNote Version 7.7.1), and duplicates removed. A new file was created minus the duplicates. Studies were then screened based on the study title. A flow chart created based on the PRISMA guidelines can be seen below ( Figure 1).

Study characteristics
The study characteristics are described in Table 2.

Risk of bias in individual studies
The seven studies were assessed for risk of bias using the 2016 ROBINS-1 ("Risk Of Bias In Non-randomised Studies -of Interventions") assessment tool (24) . The ROBINS-1 consists of an assessment and a scoring algorithm

Risk of bias assessment
All studies were subject to a varying level of bias due to the observational nature of the analysis, and potential confounders. Four studies were found to be at serious risk of bias in at least 1 domain, with three studies at moderate risk of bias (Table 1). No study was judged to be at a critical risk of bias in any domain. Therefore, the seven studies were included in this review (25; 26; 27; 28; 29; 30; 31) .

Maternal diet and the maternal gut microbiota
The association between maternal diet and the maternal gut microbiome composition in pregnancy was investigated in five studies. All five studies reported that the maternal gut microbiome in pregnancy is influenced by maternal diet to varying degrees. In addition, specific macronutrients are associated with distinct bacterial compositions and relative abundances, and can modulate, either positively or negatively, the diversity of the gut microbiome.
Three studies identified an association between dietary fat intake and gut microbiome composition (26; 27; 29) . Two of these studies reported a negative correlation between alpha diversity and intakes of cholesterol (26) , total fat and saturated fatty acid (27) (SFA). The third study (29) reported a difference in beta diversity, although alpha diversity did not differ. Furthermore, microbial composition differed by type of fat. Intakes of cholesterol and Accepted manuscript monounsaturated fatty acids (MUFA) were associated with relative increases in Proteobacteria composition (26) . In contrast, SFA intake was linked to relative decreases in this Phylum and also negatively associated with the genus Roseburia (rho=-0.4,p=0.038) (29) . The study by Barrett

Accepted manuscript
Shannon's diversity, only Vitamin D was significantly associated (-5.1% change in diversity per unit increase in Vitamin D intake, p < 0.001. The authors report that associations between dietary components and beta diversity did not show any effects [UniFrac (weighted and unweighted;data not shown)]. Furthermore, multiple regression modelling was used to assess associations between microbial composition and one standard deviation of nutrient intake for several dietary components. Vitamin D was associated with relative increases in Actinobacteria and Proteobacteria. Retinol was also associated with relative increases in Proteobacteria composition. Conversely, protein and vitamin E correlated with relative decreases in Proteobacteria.

Maternal diet and the neonatal gut microbiome
Two studies investigated the effect of maternal diet in pregnancy on the neonatal gut microbiome. Both studies reported that maternal diet in pregnancy is associated with distinct changes in the neonatal gut microbiome.
Chu et al identified an association between maternal dietary fat intake and distinct changes in the neonatal gut microbiota, at birth and 4-6 weeks of age.
Participants were grouped by extremes of dietary fat intake (one SD greater or less than the cohort mean), to produce a high-fat maternal diet group (n=13, 43.1% fat intake) and low-fat group (n = 13, 24.4% fat intake).
Significant differences in neonatal microbiome clusters were detected between groups (PcoA unweighted UniFrac: p=0.04). There was an inverse

Main findings in this study
Pregnancy is a unique time-point during which improvement to the health of the woman can also benefit the immediate and long-term health of the child.
Manipulating the gut microbiome during pregnancy may be beneficial to the health of both mother and baby (21) . Indeed, each of the studies included in this review demonstrate the important influence of maternal diet in pregnancy in modulating the gut microbiome of mother and infant, both beneficially and detrimentally. They provide evidence that diet quality, determined by factors including amount of fibre, fat, fat-soluble vitamins, fruit and vegetables, and fish and meat consumed, is associated with distinct gut microbiota profiles and diversity of the gut microbiota. Interestingly, the findings from Lundgren et al demonstrate that the influence of maternal diet on gut microbiota profiles differ by delivery mode.
The findings from this review align with those of the prevailing literature.
Recent studies have shown the influence of diet and the gut-brain axis in the prenatal period, with the gut microbiome potentially playing a role in neurodevelopment (32) In addition, diet has been shown to change the composition and metabolism of gut microbes (33) . Fibre and to a lesser degree fat have been identified as important modulators of the human gut microbiome (6; 11) . It is estimated that approximately 20-60g of undigested carbohydrate Accepted manuscript reaches the large intestine (the area with the highest density of gut microbes) daily (34) . This is larger than the amount of fat and protein that reach the colon, which are both readily digested in the upper GI tract (10) , and thus are more likely to impact on the small intestinal microbiota. In high fat diets (>35% of total energy intake), a greater proportion of fat will reach the colon, and it is hypothesised that this causes reduction of bacteria usually used for carbohydrate degradation, causing a shift in the microbiome as a whole (35) .
In contrast, high fibre diets (>25g per day (36) ) are associated with greater relative abundances of SCFA producing bacteria (such as Holdemania and Roseburia) and relative depletion of lactate producers (such as Collinsella), with the former considered directly associated with beneficial metabolic profiles (37) .
In addition, probiotics have emerged as another promising means by which to manipulate the maternal gut microbiota with a view to improve health and clinical outcomes (11) . However, the research behind their use in pregnancy has not shown clear reduction of adverse outcomes such as preterm birth or secondary outcomes such as gestational diabetes or reduction in glucose level (38; 39)  Accepted manuscript neonatal) gut microbiota may offer more readily available opportunities in the immediate term for improving the health of mother and child.
Environmental determinants have been demonstrated as important mediators of the human gut microbiota, including the shared home environment. Factors such as having other children at home, or having furry pet animals, have been shown to directly influence the composition of the maternal and neonatal gut microbiota (6; 40) . None of the studies in this review explored these variables.
Significant heterogeneity pervades multiple domains of the studies included in this review. Consequently, the findings of this review should be interpreted with caution, and considered in the context of the wider literature. Four of the five studies focusing on maternal gut outcomes studied a cohort of women with overweight and obesity. Although this could be considered a representative sample in the context of rising overweight and obesity rates, a comprehensive well-designed study examining normal weight and overweight/obese women in pregnancy, nutrients and the microbiome must be conducted first for comparison. BMI was self-reported by participants in the study by Lundgren et al. It has been shown that self-reported BMI underestimates actual BMI in pregnancy (41) .
In addition, the method of dietary assessment varied considerably across the studies. Five studies assess diet by FFQ, one by three-day food diaries, and one by IDQ. Roytio et al. used three-day food diaries as well as providing participants with oral and written instruction and a portion picture booklet. This would allow for a more accurate correlation between diet and the microbiome. Of the five studies that employed FFQs, there were differences in the period of time assessed (from four to 16 weeks), and the time-point in pregnancy it was administered (two in first trimester, two in second trimester, one in third).
As pregnancy progresses, diet may vary considerably due to increased early satiety, reflux, and constipation. There is also potential for misclassification of food groups using FFQs. In the Willets FFQ used in Lundgren et al, fruit and fruit juices are both in the fruit food group. Fruit juices contain high amounts of free sugar and lower amounts of fibre, and therefore the effect on the gut microbiota could be considerably different (42) . Likewise, differences in the temperature at which collected stool samples were stored, and the time-point at which they were collected across the studies could influence the comparability of the results.
A major strength of this systematic review is the techniques used in the search strategy and the analysis of bias. The PRISMA guidelines recommended by the Cochrane group were used (23) .
Another strength of this review is that all seven studies used culture independent analytical techniques. The use of culture specific sampling technique is now seen as a major risk of bias in the microbiological research.
The benefit of culture independent analytical techniques is that all microbial species present in the microbiome can be identified and therefore analysed (1) .

Future Directions of Studies
The examination of detailed dietary data in pregnancy and its influence on the microbiome must be conducted in detail in a cohort representative of a normal Accepted manuscript obstetric population. Without this, findings from subgroups are difficult to interpret. Dietary analysis should be conducted in a systematic manor. Food diaries most accurately capture intake within the last week and therefore may be most appropriate compared with food-frequency questionnaires that capture intake in the last few months. With this said, there is emerging evidence to suggest that long-term food patterns have a stronger role in the metabolism and composition of the human gut microbiome than short-term dietary changes (16) . Therefore, perhaps both FFQ and food diaries methodologies should be used for each analysis.

Conclusion
In summary, this review demonstrates the important influence of maternal diet in pregnancy in modulating the gut microbiome of mother and infant, both beneficially and detrimentally. The findings provide evidence that diet quality, determined by factors including amount of fibre, fat, fat-soluble vitamins, fruit and vegetables, and fish and meat consumed, is associated with distinct gut microbiota profiles and diversity of the gut microbiota. However, confidence in the quality of this evidence is limited due to methodological limitations within the studies, and variability between studies. Pregnancy is a unique time-point during which benefits to the health of the mother can also benefit that of the child. Hence, further high quality research is required in this area to elucidate the relationship between diet quality and the gut microbiota of mother and child.