Infant Gut Microbiota and Childhood Blood Pressure: Prospective Associations and the Modifying Role of Breastfeeding

Abstract

Background: Germ-free mice experiments indicate that human gut microbiota influence blood pressure (BP), but no studies have prospectively examined if infant gut microbiota affects their future childhood BP. We aim to investigate prospective associations of infant gut microbiota diversity and composition with childhood BP, examining effect measure modification by breastfeeding and mediation by a child's body mass index.

Methods and results: In the Copenhagen Prospective Studies on Asthma in Childhood 2010 cohort, we measured infant gut microbiota (16S rRNA V4) at 1 week, 1 month, and 1 year and child BP at 3 and 6 years. We assessed α diversity-BP, β diversity-BP, and microbe abundances-BP associations using linear regression, permutational multivariate analysis of variance, and beta-binomial count regression, respectively. Data from 526 children showed that α diversity and several Bifidobacterium spp. had protective associations with BP but only in children breastfed for ≥6 months. For instance, a 1-unit increment in 1 month Shannon index was associated with 1.86 mm Hg (95% CI, 0.66-3.05) lower 6-year systolic BP in children breastfed ≥6 months but a 0.73 (95% CI, -1.00 to 2.45) higher 6-year systolic BP in those breastfed <6 months (P-interaction=0.02). Greater abundance of 2 Bifidobacterium microbes at 1 week was negatively associated with 6-year systolic BP when breastfeeding ≥6 months (P-interaction<0.1). Further, abundance of 8 microbes at 1week or 1 month was linked to 3-year or 6-year BP (false discovery rate P<0.05), with 5 of them independent of a child's body mass index. Lastly, 1-week unweighted UniFrac distance and 1-year weighted UniFrac distance were associated with BP after adjustment (P<0.05).

Conclusions: Gut microbiota features at 1 week and 1 month of life were associated with BP at 6 years. Breastfeeding duration modified key associations including those for α diversity and Bifidobacteria.

Keywords: Bifidobacteria; breastfeeding; childhood blood pressure; early‐life gut microbiota.

Figure 1. Prospective associations between infant gut microbiome α diversity indices and child BP at 3 and 6 years, overall and by breastfeeding duration.

The effect sizes and 95% CIs are interpreted as the changes in BP per 1‐unit increment in α diversity indices. Models were adjusted for maternal diet score during pregnancy (continuous), diagnosis of preeclampsia (yes vs no), infant birth weight (kg), gestational age (wks), and breastfeeding duration (d) at the time of microbiome measurement (for overall estimates only). P values shown on figure indicated statistical significance for interaction. ASVs indicates amplicon sequencing variants; BP, blood pressure; DBP, diastolic blood pressure; and SBP, systolic blood pressure.

Figure 2. Proportions of variance in infant gut microbiome β diversity matrices explained by child BP.

Variances explained are corresponded to R² values in % scale estimated using permutational multivariate analysis of variance with 9999 permutations. Models were adjusted for maternal diet score during pregnancy (continuous), diagnosis of preeclampsia (yes vs no), infant birth weight (kg), gestational age (wks), and breastfeeding duration (d) at the time of microbiome measurement. Star symbols denote statistically significant results (P<0.05). BP indicates blood pressure; DBP, diastolic blood pressure; SBP, systolic blood pressure; and UniFrac, unique fraction metric.

Figure 3. Infant gut microbial ASVs that are differentially abundant according to child BP levels.

The effect sizes are interpreted as the change in BP per 1‐unit increment in logit transformed ASV abundance estimated using beta‐binomial count regression. Models were adjusted for maternal diet score during pregnancy (continuous), diagnosis of preeclampsia (yes vs no), infant birth weight (kg), gestational age (wks), and breastfeeding duration (d) at the time of microbiome measurement. Symbols denoted results with false discovery rate adjusted P<0.05 using the Benjamini–Hochberg method. Sample sizes for the analysis of microbiome at 1 wk, 1 mo, and 1 y with BP at 3 y were 315, 346, and 354, respectively; sample sizes for the analysis with BP at 6 y were 405, 444, and 454, respectively. See Table S1 for the full taxonomy of the ASVs shown in this figure. ASV indicates amplicon sequence variant; BP, blood pressure; DBP, diastolic blood pressure; FDR, false discovery rate; and SBP, systolic blood pressure.

Figure 4. Interaction effects of Bifidobacterium ASVs and breastfeeding status/duration on SBP at 3 (A1, A2) and 6 (B1, B2) years.

Estimates are based on linear regression models with the interaction terms of Bifidobacterium ASVs abundance and absence/presence with breastfeeding status/duration. Models were adjusted for maternal diet score during pregnancy (continuous), diagnosis of preeclampsia (yes vs no), infant birth weight (kg), and gestational age (wks). See Table S1 for the full taxonomy of the ASVs shown in this figure. ASV indicates amplicon sequence variant; CLR, centered log‐ratio transformation; and SBP, systolic blood pressure.