Prospective associations of breastfeeding parents’ postpartum dietary intake with infant gut microbiome at 6 months in the Pregnancy Eating Attributes Study

Abstract

Background:

Oligosaccharides in breast milk facilitate colonization of infant gut microbiota that reduce the risk of metabolic disorders. Although diet influences human milk composition, no study to date has examined the association of breastfeeding parents’ dietary intake, exclusively during the postpartum period, with infant gastrointestinal microbiome.

Objective:

To examine the relationship of postpartum diet quality of the breastfeeding parent, as measured by Healthy Eating Index-2015 (HEI-2015), with 6-month infant gut microbiota.

Design:

A secondary data analysis of a prospective pregnancy cohort participating in the Pregnancy Eating Attributes Study in North Carolina.

Participants/settings:

Of 458 participants enrolled from November 2014 through October 2016, this study included 103 breastfeeding parent-infant dyads. Dietary recalls collected at 4–6 weeks and 23–31 weeks postpartum estimated diet quality. Infants were classified into one of the following groups based on their feeding exposures at 6 months: 1) breastmilk only; 2) breastmilk and solids; 3) breastmilk and formula (with/without solids).

Main outcome measures:

Infant rectal swabs, collected 23–31 weeks following delivery, were used for deoxyribonucleic acid extraction and sequencing. The paired-end FASTQ files were input into Just A Microbiology System pipeline.

Statistical analyses:

Multivariate linear models examined relationships between HEI-2015 components and abundances of infant microbial taxa in the full sample and by feeding groups.

Results:

In the overall sample, higher breastfeeding parents’ total HEI-2015 score was associated with lower abundance of Campylobacter hominis (β = −0.0012, SE = 0.0003, p < .001, FDR q-value=0.002) and Acidaminococcus (Unclassified) (β = −0.0012, SE = 0.0003, p < .001, FDR q-value=0.002). Among infants exclusively fed breastmilk, higher HEI-2015 total protein foods score was associated with lower abundance of Streptococcus (unclassified) (β = −0.078, SE = 0.012, p < .001, FDR q-value=<0.001) and Anaerococcus tetradius (β = −0.014, SE = 0.003, p < .001, FDR q-value=0.043). Among infants fed breastmilk and solid foods, higher HEI-2015 refined grain score was associated with lower abundance of Clostridiaceae (β = −0.002, SE = 0.0004, p < .001, FDR q-value=0.013). Among infants fed breastmilk and formula, higher HEI-2015 total protein foods score was associated with lower abundance of Atopobium (Unclassified) (β = −0.01, SE = 0.004, p < .001, FDR q-value= 0.04).

Conclusions:

Higher breastfeeding parents’ HEI-2015 scores were associated with lower abundance of gut microbial genera that have been previously implicated in inflammation. Findings suggest the potential of the parent’s dietary intake during breastfeeding to support the development of infant gut microbiome associated with favorable short- and long-term health outcomes.

Introduction

Infancy represents a crucial period for the complex assembly of gut microbiota^1,2, which in turn are essential for metabolic, immune, trophic, endocrine and neural functions^3,4, and the gut microbiome is highly dynamic during this period⁵. Dysbiosis of the infant gut microbiota, characterized by overrepresentation of pathobiont bacteria, such as Enterobacteriaceae, and underrepresentation of Bifidobacterium, has been linked to increased risk of immune and metabolic disorders in later life⁶. Microbiota transmission from childbearing parent to child, such as via lactation, shapes the development and maturation of the infant gut microbiome^7,8. Since childbearing parents’ diet during lactation is a critical determinant of breast milk composition⁹, an understanding of how childbearing parents’ dietary exposures influence the infant gut microbiota could inform efforts to promote short- and long-term health of the child.

Breastfeeding is among the critical factors that influence the infant gut microbiota^10,11, the effects of which are attributed to the presence of oligosaccharides and probiotic microorganisms in human milk ^12,13. The resulting colonization of gut microbiota has been associated with reduced risk of metabolic disorders in later life ¹⁴. Previous research suggests that diet during lactation could influence human milk oligosaccharide (HMO) ⁹ and milk microbiota composition ^15,16. These studies found that higher total poly- and mono-unsaturated fatty acids (PUFA and MUFA), ꞷ−3 PUFA, linoleic acid, fiber, fruit and vegetable intakes are related to greater microbial diversity ¹⁶, higher abundance of Bifidobacterium ¹⁷ and several HMOs ^18,19 in human milk, whereas higher saturated fat, cholesterol and added sugar intakes are associated with lower HMO abundance ^18,20. The relationship of breastfeeding parents’ diet with human milk composition suggests that diet during lactation could influence the infant gut microbiota²¹. To date, only one observational study has investigated the association of breastfeeding parents’ diet during lactation with the infant gut microbiome ¹⁵. However, this study included only exclusively breastfed infants and combined estimates of diet during pregnancy with that of postpartum, thereby limiting inference regarding the impact of diet during lactation on infant gut microbiota. The goal of this research is to examine the relationship of breastfeeding parents’ diet quality as measured by the Healthy Eating Index-2015 (HEI-2015) total score and its component scores with 6-month infant gut microbiota in a sample of breastfeeding parent-infant dyads, stratified by type of feeding (i.e., breastfeeding only; breast- and solid-feeding; and breast- and formula-feeding, with or without solid feeding).

Methods

Study participants

The current analyses utilized data from the Pregnancy Eating Attributes Study (PEAS)²², a prospective observational cohort followed from first trimester of pregnancy to 1 year postpartum along with their infants from birth to 1 year of age. Patients presenting for pregnancy care from two university-based obstetrics clinics in Chapel Hill, North Carolina, were eligible to participate in the study if they met the following inclusion criteria: 1) gestational age of ≤12 weeks at enrollment; 2) BMI ≥18.5 kg/m²; 3) aged between 18–45 years; 4) anticipated an uncomplicated, singleton pregnancy; 5) had access to internet with an associated email; 6) had ability to complete questionnaires/assessments in English; and 7) intended to deliver at the University of North Carolina Women’s Hospital and planned to remain in the vicinity for 1 year following childbirth. Patients were excluded if they had a multiple gestation pregnancy or any medical/psychosocial condition (such as eating disorder, diabetes, any illness or use of medication that could affect diet or weight).

Identification and recruitment of potential participants were facilitated by use of clinical appointments and medical records. Participants provided signed informed consent for their participation and consent for their infant’s participation. Enrollment spanned from November 2014 through October 2016. Participants completed one study visit each trimester and three postpartum visits corresponding to 4–6 weeks, 23–31 weeks and 50–58 weeks after delivery. All study procedures were approved by the University of North Carolina Institutional Review Board.

While the primary goal of PEAS was to examine behavioral and environmental influences on eating behaviors and weight changes during pregnancy and postpartum, biospecimens for infant gut microbiome processing were collected about 23–31 weeks following delivery for secondary analyses. Data for the current analysis includes infant gut microbiome data, childbearing parent dietary recall, and infant feeding mode at 4–6 weeks and 23–31 weeks. Of the 458 participants enrolled, 91 withdrew during pregnancy, primarily due to unexpected pregnancy termination, unwillingness to continue participation, or relocation. Of the 367 participants retained through pregnancy, 172 participants submitted an infant microbiome specimen at 6 months postpartum. Of these, 10 were excluded for preterm delivery (i.e., gestational age less than 37 weeks) or low birth weight (i.e., birth weight less than 2.5 kg) due to possible differences in infant growth and microbiome composition^23–25. Additionally, 41 were excluded due to lack of postpartum dietary intake, 1 was excluded due to lack of infant feeding information and 13 were excluded due to low quality of the microbiome specimen (less than 50% assembly rate). Of the resulting sample of 107 childbearing parent-infant dyads, only 4 infants had been exclusively formula-fed at 6 months. Since the study associations were examined separately by feeding mode at 6 months and this group was too small for meaningful analysis, these participants were excluded. Thus, the analytic sample consisted of 103 dyads.

Feeding mode at 6 months

At each postpartum visit, participants reported infant intake of breastmilk, formula, and complementary foods. From these data, infants were classified into one of the following groups based on their feeding exposures at age 6 months: 1) breastmilk only; 2) breastmilk and solids; or 3) breastmilk and formula (with/without solids).

Parents’ diet quality

Participants were asked to complete a 24-hour dietary recall during each study visit. For this analysis, dietary recalls collected at 4–6 weeks and 23–31 weeks postpartum were used to estimate diet quality. Dietary recalls were collected using the Automated Self-Administered 24-Hour Recall (ASA24) method, a web-based tool developed by the National Cancer Institute²⁶. The ASA24 uses an online interface to prompt participants to list all foods consumed, including information on food preparation, portion size and brands, for the specified period. Food codes from the US Department of Agriculture Food and Nutrient Database for Dietary Surveys (FNDDS)²⁷ are assigned to food items reported in the diet recalls and estimates of macronutrient, micronutrient, food categories, and Food Patterns Equivalents Database²⁸ food groups are obtained. Details on the diet recall and estimation of food group/nutrient intake have been published previously²⁹. Diet quality was assessed using the HEI-2015³⁰, a metric that assesses dietary adherence to the 2015–2020 Dietary Guidelines for Americans³¹. The HEI-2015 score was calculated as the sum of scores for nine adequacy dietary components (i.e., total fruits, whole fruits, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant proteins, and fatty acids) and four moderation components (i.e., refined grain, sodium, added sugars, and saturated fats). Of note, moderation components are scored so that lower intakes receive higher scores; therefore, higher scores reflect greater adherence to guidelines for all the HEI-2015 dietary components. The maximum possible total score for HEI-2015 is 100. Overall postpartum HEI-2015 was calculated by pooling dietary recalls across postpartum visits using HEI simple scoring algorithm per person³².

Infant microbiome biospecimen collection and storage

At the second postpartum visit (23–31 weeks following delivery), parents were asked to collect their child’s rectal swab using a study-supplied collection kit (BD BLL^™ CultureSwab). They were asked to gently insert the swab ½ inch into their child’s rectum, then gently rub the swab 4 – 5 times around the wall of the rectum. The biospecimen was then processed for storage until sequencing of all biospecimens could be completed. The swab end of biospecimen was inserted into a microcentrifuge tube with 0.5 mL of Deoxyribonucleic acid (DNA) Catch all Reagent (Thermofisher). The swab was rotated within the tube prior to a series of vortex procedures and heated to 60°C followed by 98°C. Final processed biospecimens were stored in a labeled box at a temperature ranging between −70 and −80°C.

DNA extraction and shotgun metagenomic sequencing

Microbiome biospecimens were transferred to the Microbiome Core at the National Cancer Institute (Bethesda, Maryland) for Deoxyribonucleic acid (DNA) extraction, library preparation, and shotgun metagenomic sequencing. Library preparation was conducted using Illumina Nextera DNA Flex Library Prep kit with Qiagen MagAttract PowerMicrobiome DNA/RNA kit (Cat No./ID: 27500–4-EP) with Qubit quantification was used for automated DNA extraction and quality control. After library preparation, the DNA samples were sequenced on the Illumina NovaSeq 6000 platform.

Taxonomic classification

The paired-end FASTQ files, generated by the Illumina NovaSeq 6000 platform, were input into the Just A Microbiology System (JAMS) pipeline (https://github.com/johnmcculloch/JAMS_BW)³³. The first phase of the pipeline, JAMSalpha, was used to determine the quantity and quality of the taxonomic sequences. All contigs and k-mer assigned reads were assigned to a last known taxon (LKT), using Kraken³⁴, which represents the most specific taxonomic classification available within JAMS. The LKT label represents a single microbe classification. JAMS was executed on the National Institutes of Health high-performance computer, Biowulf. This pipeline has been used and described in previously published studies^35–40. Microbiome taxa relative abundances were normalized by arc sin square root⁴¹.

Demographic and clinical data

Race/ethnicity and education were reported by childbearing parents at baseline. At this time, participants also were measured for height and weight, which were used to calculate early pregnancy BMI (kg/m²). They were classified as underweight (<18.5 kg/m²), healthy weight (18.5 to 24.9 kg/m²), overweight (25 to 29.9 kg/m²) or obese (≥30 kg/m²) based on their early pregnancy BMI⁴². Information on parents’ age at baseline and infants’ mode of delivery were obtained from medical records. At each postpartum visit, participants reported their child’s general health status, including information on whether the child had any infections. Reports of child infections were reviewed to determine whether the child had an infection that could have potentially required antibiotic treatment.

Statistical analysis

Descriptive statistics including mean and standard deviation were calculated for sociodemographic, dietary, and clinical data for the three feeding groups. Differences between the feeding groups were examined using ANOVA and chi-square tests for continuous and categorical variables, respectively. Descriptive analyses were performed using SAS (version 9.4; SAS Institute, Inc., Cary, NC, USA) software.⁴³

Alpha and beta diversity metrics were calculated using the vegan package (version 2.6–2) in Rstudio.⁴⁴ Several alpha diversity indices⁴⁵ were calculated including, 1) the inverse Simpson index, assessing the richness or the total number of distinct species as well as their abundances in a community; 2) the Shannon-Weiner index, evaluating both richness and evenness of microbiota in the biospecimen; and 3) the Chao1 richness index, quantifying the total number of species, accounting for the distribution of rare species, in a biospecimen. Statistical analysis of the average alpha diversity indices across different infant feeding groups was performed using the non-parametric Kruskal-Wallis test. Beta diversity, which reflects the dissimilarity between biospecimens, was determined using the Bray-Curtis dissimilarity. ANCOM-BC2 was used to detect taxa that are differentially abundant between the feeding groups. ANCOM-BC2 accounts for the compositionality of microbiome data by correcting sample- and taxa-specific biases. For this analysis, taxa that were present in less than 10% of samples were removed. Parents’ age at the time of enrollment and the sequencing depth of the biospecimen were included into the model as covariates. P-values were adjusted using Benjamini–Hochberg procedure. Taxa with adjusted p-value less than 0.05 were considered differentially abundant. Data visualizations were used to perform exploratory analysis (Supplementary Figure 1) and to support model variables (Supplementary Figures 2 and 3).

An outlier analysis and multivariate linear model using previous work conducted by Bolte et al⁴¹ was implemented to examine the relationship between HEI-2015 components and abundances of infant microbial taxa for each feeding group. Based on previous research^46–51, parents’ age, infant mode of delivery and infant antibiotic administration were hypothesized to influence infant gut microbiota. However, mode of delivery and infant diagnosis of a condition with likely antibiotic administration were not temporally aligned with the independent and dependent variables and did not yield clear beta diversity groupings in exploratory analysis (Supplementary Figures 2 and 3). Therefore, they were not included as covariates. Parents’ age at the time of enrollment and sequencing depth of the biospecimen were included as covariates. Given that the HEI-2015 components used to derive the score are standardized for energy, total energy intake was not considered a covariate. All analyses were adjusted for multiple testing corrections using the Benjamini-Hochberg method via the p.adjust function in Rstudio. For determining statistical significance, false discovery rate (FDR) threshold was set at less than 0.10. All microbiome-related statistics were completed in Rstudio (version 4.2.2).⁵² Additionally, linear regression analyses were conducted to examine the relationship of HEI-2015 total and component scores with the Shannon-Weiner diversity index. These analyses were adjusted for parents’ age at time of enrollment.

Results

Breastfeeding parent and Infant Characteristics

Table 1 indicates the demographic and clinical characteristics of parents and infants by type of feeding exposure. No significant differences in demographic and clinical characteristics were observed between the three feeding groups. However, all parents in the ‘breastmilk only’ feeding group had a college degree or above.

Table 1.

Demographic and clinical characteristics^a of breastfeeding parents and infants in the Pregnancy Eating Attributes dataset, by type of feeding exposure

	Breastmilk only (n=24)	Breastmilk and solids (n=21)	Breastmilk, formula and/or solids (n=58)	p-value
Breastfeeding parents’ characteristics
Age in years, mean (SD) b	32.6 (4.2)	31.2 (3.3)	31.7 (3.8)	0.43
Race/ethnicity, n (%)
Non-Hispanic White	19 (79.2)	16 (76.2)	40 (69)	0.80c
Non-Hispanic Black	1 (4.2)	1 (4.8)	7 (12.1)
Hispanic	2 (8.3)	2 (9.5)	3 (5.2)
Other race	2 (8.3)	2 (9.5)	8 (13.8)
Educational attainment, n (%) d
High school or less	0 (0)	0 (0)	2 (3.6)	0.07c
Some college/associate degree	0 (0)	1 (4.8)	10 (18.2)
College graduate or above	24 (100)	20 (95.2)	43 (78.2)
Early pregnancy weight status, n (%)
Healthy weight	14 (58.3)	13 (61.9)	30 (51.7)	0.59c
Overweight	6 (25)	6 (28.6)	13 (22.4)
Obese	4 (16.7)	2 (9.5)	15 (25.9)
Postpartum HEI-2015 score e,f , mean (SD)	63.4 (11.5)	66.5 (13.6)	60 (12.4)	0.10
Infant characteristics Delivery method, n (%) d
Vaginal	22 (91.7)	20 (95.2)	43 (76.8)	0.07c
C-section	2 (8.3)	1 (4.8)	13 (23.2)
Infection requiring antibiotic treatment, n (%)
No	21 (87.5)	20 (95.2)	52 (90)	0.66c
Yes	3 (12.5)	1 (4.8)	6 (10)
Alpha diversity index variables, mean (SD) b
Inverse Simpson	6.6 (3.8)	6.1 (3.9)	8.1 (5.8)	0.21
Shannon	2.4 (0.6)	2.3 (0.7)	2.5 (0.7)	0.34
Chaol	240.3 (43)	242.9 (36.7)	240 (39.7)	0.96

^aAnalysis of variance and chi-square tests were used for continuous and categorical variables, respectively, with p<0.05 used to indicate statistical significance

^bSD, standard deviation

^cDue to cell counts of less than 5 for categorical variables, p-values of chi-square tests must be interpreted with caution

^dDue to missing data, cell sizes do not add to the total for the group, ‘Breastmilk, formula and/or solids’

^eHEI-2015, Healthy Eating Index-2015

^fHEI-2015 score calculated from all postpartum visits are presented to indicate overall diet quality for the four groups of interest

As shown in Figure 4, there were significant pairwise differences in taxa abundances between infants in the ‘breastmilk only’, ‘breastmilk and solids’ and ‘breastmilk, formula and/or solids’ feeding groups.

Figure 4.

Pairwise differences in taxa abundances between feeding groups (i.e., breastmilk only, breastmilk and solids, and breastmilk, formula and/or solids) of infants in the Pregnancy Eating Attributes dataset (n=103), using ANCOM-BC2

*p<0.05; **p<0.01; ***p<0.001

lfc represents the difference in bias-corrected abundances between groups

Association between breastfeeding parents’ diet and the infant microbiota in the overall sample (data not shown in tables)

In the overall sample of breastfeeding parent-infant dyads, higher parents’ total HEI-2015 score was associated with lower abundance of Campylobacter hominis (β = −0.0012, SE = 0.0003, p < .001, FDR q-value=0.002) and Acidaminococcus (Unclassified) (β = −0.0002, SE = 0.00003, p < .001, FDR q-value=0.002). At an FDR cutoff of <0.10, higher total HEI-2015 score was associated with lower abundance of Firmicutes bacterium (β = −0.0003, SE = 0.00008, p < .001, FDR q-value=0.08). Additionally, higher HEI-2015 total protein foods score was associated with lower abundance of Atopobium (Unclassified) (β = −0.015, SE = 0.004, p < .001, FDR q-value=0.04). There were no significant associations between any other component of the HEI-2015 score and the individual microbial taxa. Associations of HEI-2015 total and component scores with Shannon-Weiner diversity index were not statistically significant.

Sub-analysis by infant feeding group (data not shown in tables)

Exclusive breastmilk

Among infants exclusively fed breastmilk, higher HEI-2015 total protein foods score was associated with lower abundance of Streptococcus (unclassified) (β = −0.078, SE = 0.012, p < .001, FDR q-value=<0.001) and Anaerococcus tetradius (β = −0.014, SE = 0.003, p < .001, FDR q-value=0.043). Higher HEI-2015 total protein foods score also was associated with lower abundance of Peptoniphilius harei (β = −0.076, SE = 0.019, p < .001, FDR q-value=0.058) and Streptococcus anginosus (β = −0.07, SE = 0.018, p < .001, FDR q-value=0.058). Associations of HEI-2015 total and component scores with Shannon-Weiner diversity index were not statistically significant.

Breastmilk and solid food

Among infants fed breastmilk and solid foods, higher HEI-2015 refined grain score was associated with lower abundance of Clostridiaceae (β = −0.002, SE = 0.0004, p < .001, FDR q-value=0.013). The HEI-2015 total and component scores were not significantly associated with Shannon-Weiner diversity index.

Breastmilk and formula

Among infants fed both breastmilk and formula (and who may or may not have received solid food), higher HEI-2015 total protein foods score was associated with lower abundance of Atopobium (Unclassified) (β = −0.01, SE = 0.004, p < .001, FDR q-value= 0.04). At an FDR cutoff of <0.10, higher HEI-2015 added sugar score was associated with lower abundance of several taxa. These include, Enterobacteriaceae (β = −0.01, SE = 0.003, p < .001, FDR q-value=0.09), Subdoligranulum (Unclassified) (β = −0.003, SE = 0.001, p < .01, FDR q-value=0.09), Sutterellaceae (β = −0.001, SE = 0.0004, p < .01, FDR q-value=0.09), Sutterella megalosphaero (β = −0.0035, SE = 0.001, p < .01, FDR q-value=0.09), Sutterella faecalis (β = −0.001, SE = 0.0003, p < .01, FDR q-value=0.09), Sutterella parvirubra (β = −0.003, SE = 0.0008, p < .001, FDR q-value=0.09), Bacteroidea ovatus (β = −0.01, SE = 0.004, p < .01, FDR q-value=0.09), Sutterella wadsworthensi (β = −0.002, SE = 0.0007, p < .01, FDR q-value=0.09), Coprobacillus (Unclassified) (β = −0.003, SE = 0.001, p < .01, FDR q-value=0.097), Duodenibacillus massilie (β = −0.001, SE = 0.0004, p < .01, FDR q-value=0.097). HEI-2015 total and component scores were not significantly associated with Shannon-Weiner diversity index.

Discussion

Findings suggest that breastfeeding parents’ dietary intake is significantly associated with the gut microbial composition of their infants. In this analysis of the PEAS breastfeeding parent-infant dyads, infant gut microbial composition differed by the type of feeding and was associated with parents’ postpartum dietary quality. Between the three feeding groups of interest, significant pairwise differences were also observed in differential abundance at the genus and species level. Across feeding types, higher breastfeeding parents’ HEI-2015 scores on several dietary components were associated with lower abundance of infant gut microbial genera that have been previously implicated in inflammation^53–57. Additionally, associations of parents’ dietary intake with abundance of infant gut microbial taxa differed based on the type of feeding at 6 months (i.e., whether infants were exclusively breastfed, breastfed with solids or mixed fed with/without solids). Findings herein contrast previous research which found no significant relationship between breastfeeding parents’ dietary intake and infant gut microbiome¹⁵; this discrepancy might be attributable to the use of a food frequency questionnaire, a less precise measure of dietary intake, in the previous study. Another explanation could be that the previous study captured dietary intake representing the pregnancy and postpartum periods, whereas the current research focused exclusively on nutrition during postpartum. However, in the current study, breastfeeding parents’ HEI-2015 total and component scores were not significantly associated with Shannon-Weiner diversity index in all the feeding groups. With previous studies suggesting dynamic changes in infants’ microbial diversity in response to solid food intake^58,59, it is possible that infants’ dietary intake could more strongly influence microbial diversity than breastfeeding parents’ diet intake.

In the overall sample of breastfed infants, higher postpartum diet quality was significantly associated with lower abundance of Campylobacter hominis and Acidaminococcus (a butyric acid-producing bacteria within the Firmicutes phylum), as well as the Firmicutes phylum. While Campylobacter hominis is an emerging bacterial species that has been detected in pediatric patients with Crohn’s disease⁵⁷, the phylum Firmicutes has been shown to be more abundant in individuals with obesity possibly due to its role in energy harvesting^60,61. Peptides and amino acids derived from dietary proteins can be fermented by gut bacterial species belonging to genera, such as Bacteroides, Prevotella, Clostridium, Megasphaera and Acidaminococcus⁶². In the current study, higher total protein foods scores were significantly associated with lower abundance of Atopobium. Although Atopobium genus was unclassified, species within the Atopobium genus have been associated with bacterial vaginosis⁶³, Crohn’s disease⁵⁴ and colorectal cancer⁶⁴.

In the sub-sample of exclusively breastfed infants, higher postpartum total protein foods scores were significantly associated with lower abundance of Anaerococcus tetradius, Peptoniphilius harei and Streptococcus anginosus. Although Anaerococcus tetradius and Peptoniphilius harei have less established roles as gut microflora, higher abundance of gut Streptococcus species, including Streptococcus anginosus, has been associated with higher levels of systemic inflammation markers⁵⁶. In the sub-sample of infants who were breastfed and introduced to solids by 6 months, lower intake of refined grains (as indicated by higher HEI-2015 refined grain score) was significantly associated with lower abundance of Clostridiaceae. While some species of Clostridiaceae could reduce inflammation via butyrate production⁶⁵, other species within this bacterial family characterize the guts of patients with inflammatory bowel disease (IBD) and rheumatoid arthritis⁵⁵. In the group of infants who were breast- and formula-fed, higher total protein foods scores were associated with lower abundance of Atopobium. At an FDR cutoff of <0.10, lower added sugar intake was associated with lower abundance of Enterobacteriaceae; the proliferation of this bacterial family has been shown to increase inflammation and implicated in the pathogenesis of IBD^66,67. Greater added sugar scores also were associated with lower abundance of several species within Sutterella, a genus that includes members with inflammatory potential⁵³, and Bacteroides ovatus. Although Bacteroides ovatus is known to produce short chain fatty acids with anti-inflammatory potential, such as acetate and propionate,⁶⁸ one study has shown that this bacterial species causes an antibody response in IBD⁶⁹.

The results of this study must be interpreted in view of the limitations. The observational nature of the study design limits causal inferences. Although infants were classified by feeding group, we were not able to quantify the extent of exposure to breastmilk, formula or solids. Given the independent effect of breastmilk, formula and solids on infant gut microbiome,⁷⁰ lack of adjustment for the extent of exposure to breastmilk, formula and solids could have confounded some of the study associations. The study analyses also did not account for probiotic supplement use of the breastfeeding parent, which also could potentially influence the infant gut microbiome.⁷¹ The geographic restriction of the study sample limits generalizability of the study results to other populations with different dietary patterns. Since the study utilized infant rectal swabs collected by parents rather than feces, the sample could represent a combination of gastrointestinal, fecal, and rectal microbial communities. Nevertheless, the study findings contribute to the existing literature on postpartum diet and infant gut microbiota. Further, dietary intake was assessed via 24-hour recalls, which has been shown to be the least biased method of assessing dietary intake⁷², and high quality metagenomic sequencing was used to assess composition of infant gut microbiota.

Conclusions

These findings suggest the potential of breastfeeding parents’ dietary intake to support the development of infant gut microbiome associated with favorable short- and long-term health outcomes. While the human gut microbiome is highly dynamic in the early years of life, our study extends the current understanding of factors associated with development of infant gut microbiota. Longitudinal studies that characterize changes in infant gut microbiome in response to breastfeeding parents’ dietary intake are required to build on the current study findings.

Research snapshot.

Research question:

What is the association of postpartum diet quality of the breastfeeding parent, as measured by Healthy Eating Index-2015 (HEI-2015), with the infant gut microbiome?

Key findings:

Higher postpartum HEI-2015 scores on several dietary components were associated with lower abundance of infant gut microbial genera that have been previously implicated in inflammation.