Maternal fecal microbiome predicts gestational age, birth weight and neonatal growth in rural Zimbabwe

Abstract

Background: Preterm birth and low birth weight (LBW) affect one in ten and one in seven livebirths, respectively, primarily in low-income and middle-income countries (LMIC) and are major predictors of poor child health outcomes. However, both have been recalcitrant to public health intervention. The maternal intestinal microbiome may undergo substantial changes during pregnancy and may influence fetal and neonatal health in LMIC populations.

Methods: Within a subgroup of 207 mothers and infants enrolled in the SHINE trial in rural Zimbabwe, we performed shotgun metagenomics on 351 fecal specimens provided during pregnancy and at 1-month post-partum to investigate the relationship between the pregnancy gut microbiome and infant gestational age, birth weight, 1-month length-, and weight-for-age z-scores using extreme gradient boosting machines.

Findings: Pregnancy gut microbiome taxa and metabolic functions predicted birth weight and WAZ at 1 month more accurately than gestational age and LAZ. Blastoscystis sp, Brachyspira sp and Treponeme carriage were high compared to Western populations. Resistant starch-degraders were important predictors of birth outcomes. Microbiome capacity for environmental sensing, vitamin B metabolism, and signalling predicted increased infant birth weight and neonatal growth; while functions involved in biofilm formation in response to nutrient starvation predicted reduced birth weight and growth.

Interpretation: The pregnancy gut microbiome in rural Zimbabwe is characterized by resistant starch-degraders and may be an important metabolic target to improve birth weight.

Funding: Bill and Melinda Gates Foundation, UK Department for International Development, Wellcome Trust, Swiss Agency for Development and Cooperation, US National Institutes of Health, and UNICEF.

Keywords: birth weight; gestation; maternal; metagenome; microbiome; pregnancy.

Figure 1

Flow of participants through the SHINE microbiome substudy.

Figure 2

Relationships between infant birth weight in Kg, epidemiologic variables, and maternal gut microbiome species relative abundance. The top 20 predictors of infant birth weight by variable importance score are shown. For microbiome abundances, the x-axis represents the percentile of the abundance distribution. Epidemiologic and microbiome diversity variables are on the original scale. Tick marks on the x-axis are a rug plot of individual feature abundance percentiles. ALEs were generated using the ALEplot package and were plotted using ggplot2. Standard deviations (sd) were calculated per increment in microbiome feature and were used to calculate and plot increment-wise 95% confidence intervals as the average change in the outcome ±1.96(sd/sqrt(n)), where n is the number of observed feature values, and sd is the standard deviation of the change in the outcome variable in an interval. gaw_final, gestational age; mom_height, maternal height in centimeters; mom_muac, maternal mid-upper arm circumference in millimeters; pct_human, percent human reads; pct_unknown, percent unknown reads.

Figure 3

Relationships between infant WAZ at 1-month, epidemiologic variables, and maternal gut microbiome species relative abundance. The top 20 predictors of infant WAZ at 1-month by variable importance score are shown. For microbiome abundances, the x-axis represents the percentile of the abundance distribution. Epidemiologic and microbiome diversity variables are on the original scale. Tick marks on the x-axis are a rug plot of individual feature abundance percentiles. ALEs were generated using the ALEplot package and were plotted using ggplot2. Standard deviations (sd) were calculated per increment in microbiome feature and were used to calculate and plot increment-wise 95% confidence intervals as the average change in the outcome ±1.96(sd/sqrt(n)), where n is the number of observed feature values, and sd is the standard deviation of the change in the outcome variable in an interval. gaw_final, gestational age; mom_height, maternal height in centimeters; mom_muac, maternal mid-upper arm circumference in millimeters; pct_human, percent human reads; pct_unknown, percent unknown reads.