Dietary amino acids, macronutrients, vaginal birth, and breastfeeding are associated with the vaginal microbiome in early pregnancy

Abstract

The vaginal microbiome is a key player in the etiology of spontaneous preterm birth. This study aimed to illustrate maternal environmental factors associated with vaginal microbiota composition and function in pregnancy. Women in healthy pregnancy had vaginal microbial sampling from the posterior vaginal fornix performed at 16 weeks gestation. After shotgun metagenomic sequencing, heatmaps of relative abundance data were generated. Community state type (CST) was assigned, and alpha diversity was calculated. Demography, obstetric history, well-being, exercise, and diet using food frequency questionnaires were collected and compared against microbial parameters. A total of 119 pregnant participants had vaginal metagenomic sequencing performed. Factors with strongest association with beta diversity were dietary lysine (adj-R² 0.113, P = 0.002), valine (adj-R² 0.096, P = 0.004), leucine (adj-R² 0.086, P = 0.003), and phenylalanine (adj-R² 0.085, P = 0.005, Fig. 2D). Previous vaginal delivery and breastfeeding were associated with vaginal beta diversity (adj-R² 0.048, P = 0.003; adj-R² 0.045, P = 0.004), accounting for 8.5% of taxonomy variation on redundancy analysis. Dietary fat, starch, and maltose were positively correlated with alpha diversity (fat +0.002 SD/g, P = 0.025; starch +0.002 SD/g, P = 0.043; maltose +0.440 SD/g, P = 0.013), particularly in secretor-positive women. Functional signature was associated with CST, maternal smoking, and dietary phenylalanine, accounting for 8.9%-11% of the variation in vaginal microbiome functional signature. Dietary amino acids, previous vaginal delivery, and breastfeeding history were associated with vaginal beta diversity. Functional signature of the vaginal microbiome differed with community state type, smoking, dietary phenylalanine, and vitamin K. Increased alpha diversity correlated with dietary fat and starch. These data provide a novel snapshot into the associations between maternal environment, nutrition, and the vaginal microbiome.

Importance: This secondary analysis of the MicrobeMom randomized controlled trial reveals that dietary amino acids, macronutrients, previous vaginal birth, and breastfeeding have the strongest associations with vaginal taxonomy in early pregnancy. Function of the vaginal niche is associated mainly by species composition, but smoking, vitamin K, and phenylalanine also play a role. These associations provide an intriguing and novel insight into the association between host factors and diet on the vaginal microbiome in pregnancy and highlight the need for further investigation into the complex interactions between the diet, human gut, and vaginal microbiome.

Keywords: alpha diversity; beta diversity; community state type; diet; environmental; exercise; metagenomic sequencing; microbiota; pregnancy; vagina.

Fig 1

Vaginal microbial composition at 16 weeks’ gestation in a healthy pregnancy cohort. Heatmap of vaginal microbial composition. Graphical representation of the dominant species in each vaginal sample shown on heatmap to form nine distinct vaginal community state type groups based on dominant species.

Fig 2

Vaginal species composition at 16 weeks’ gestation in healthy pregnancy cohort. (A) NMDS ordination plot of vaginal microbial species using non-metric multidimensional scaling based on Bray-Curtis dissimilarity. (A) is annotated with the stress of the model (amount of variability unexplained by the NMDS ordination). (B) Envfit model plotted on functional ordination, displaying variables with significant (P < 0.05) association with species plot, and adjusted-R values of significant covariates annotated. (C) Redundancy analysis of maternal meta-data against vaginal species composition, showing variation in the data attributable specifically to covariates identified in envfit model on the x-axis. (D) Envfit barchart of covariate adjusted R squared and P value when comparing difference in centroids of each covariate relative to the total variation, with significant covariates denoted with asterisk.

Fig 3

Analysis of functional gene ontology of the vaginal microbiome at 16 weeks’ gestation. (A–C) NMDS ordination of functional data sets BP, CC, and MF annotated with the stress of the model (amount of variability unexplained by the NMDS ordination). (D–F) Envfit covariates plotted on functional ordination plots, displaying variables with significant (P < 0.05) association with species plot, and adjusted-R values of significant covariates annotated. (G–I) Redundancy analysis of meta-data against function, showing variation in the data attributable specifically to covariates identified in envfit model on the x-axis. (J–L) Envfit barchart of covariate adjusted R squared and P value when comparing difference in centroids of each covariate relative to the total variation within the function, with significant covariates denoted with asterisk. CC, cell cycle functions; RDA, redundancy analysis; VD, vaginal delivery; BF, breastfeeding; NSP, non-starch polysaccharides; MET, metabolic equivalent of tasks.

Fig 4

Previous vaginal delivery (A), breastfeeding history (B), and maternal smoking status (C) and vaginal Shannon alpha diversity in early pregnancy.

Fig 5

Macronutrients daily nutritional intake and (A–F) vaginal Shannon alpha diversity and (G–I) vaginal community state type in early pregnancy. All regression models are adjusted for maternal confounders, including maternal age, weight, parity, ethnicity, education status, and deprivation index.