Effect of intrapartum azithromycin on early childhood gut mycobiota development: post hoc analysis of a double-blind randomized trial

Abstract

Intrapartum azithromycin prophylaxis reduced maternal infections but showed no effect on neonatal sepsis and mortality. Although antibiotic exposure may indirectly alter the mycobiota (community of fungi that live in a given environment), there is no data available on how intrapartum azithromycin impacts gut mycobiota development. We hereby assess the impact of intrapartum azithromycin on gut mycobiota development from birth to the age of three years, by ITS2 gene profiling of rectal samples from 102 healthy Gambian infants selected from a double-blind randomized placebo-controlled clinical trial (PregnAnZI-2 - ClinicalTrials.org NCT03199547). In the trial, women received 2 g oral azithromycin or placebo (1:1) during labour with the intension of assessing effect on neonatal sepsis or mortality. Secondary objectives included effects on bacterial carriage and resistance, puerperal infections, and infant growth. Our analysis show that season and parity were key factors that influenced gut mycobiota development. Intrapartum azithromycin increased the abundance of Candida orthopsilosis but only in the wet season and did not show different effects by sex of the child. These data suggest that season and parity can be key factors influencing gut mycobiota development and may inform strategies for a wider implementation of intrapartum azithromycin intervention.

Fig. 1. Factors that influenced Shannon diversity and species richness.

Effect of treatment and other covariates on Shannon diversity (a) and species richness (b) estimated by multiple multivariate analysis using a linear mixed-effects model. References for comparisons: Treatment: placebo, Sampling season: dry season, Age: day_0, Parity: primipara, Ethnicity: Mandinka, Sex: female. In both figures a and b, error bars show the mean and 95% confidence intervals of standard deviation for each covariate. A total of 427 samples from 125 individuals were included. Random variations were averaged on individuals to control for repeated measurements. c Variation in Shannon diversity between azithromycin and placebo arms at each time-point. d Variation in Shannon diversity between azithromycin and placebo arms at each time-point stratified by season. e Variation in Shannon diversity between azithromycin and placebo arms at each time-point stratified by parity. For all analyses of temporal variation in Shannon diversity between trial arms, unpaired Wilcoxon’s Rank Sum test was used. The box and whiskers denote the distribution of Shannon diversity scores. The box shows the median and lower and upper quartiles (middle 50% of the scores), while the whiskers show the upper and lower 25% scores not excluding outliers. Para parity, Azi Azithromycin.

Fig. 2. Overall community composition.

a Community composition between trial arms at each time-point estimated by PERMANOVA. b Community composition by age in azithromycin arm. c Community composition by age in placebo arm. Age comparisons were made across all time-points with permutations restricted within individuals to control for repeated measurement, and pairwise between day 0 and each of the subsequent time-points. Azi Azithromycin.

Fig. 3. Gut mycobiota community profiles by treatment over time.

Community profiles showing relative abundance of the top 28 fungal taxa between azithromycin and placebo arms at each time-point. Azi Azithromycin.

Fig. 4. Fungal clusters (community types) based on community similarity generated by Dirichlet’s multinomial mixtures model.

a Heatmap showing the most abundant fungal species in the samples grouped by community type. b Frequencies of fungal clusters at each time-point grouped by treatment. c Individual fungal community transitions over time by treatment. Azi Azithromycin.

Fig. 5. Study profile summarizing sample selection process.

a Sample selection for mycobiota study. b Distribution of samples included in mycobiota analysis between trial arms at each time-point.