Effects of early-life antibiotics on the developing infant gut microbiome and resistome: a randomized trial

Abstract

Broad-spectrum antibiotics for suspected early-onset neonatal sepsis (sEONS) may have pronounced effects on gut microbiome development and selection of antimicrobial resistance when administered in the first week of life, during the assembly phase of the neonatal microbiome. Here, 147 infants born at ≥36 weeks of gestational age, requiring broad-spectrum antibiotics for treatment of sEONS in their first week of life were randomized 1:1:1 to receive three commonly prescribed intravenous antibiotic combinations, namely penicillin + gentamicin, co-amoxiclav + gentamicin or amoxicillin + cefotaxime (ZEBRA study, Trial Register NL4882). Average antibiotic treatment duration was 48 hours. A subset of 80 non-antibiotic treated infants from a healthy birth cohort served as controls (MUIS study, Trial Register NL3821). Rectal swabs and/or faeces were collected before and immediately after treatment, and at 1, 4 and 12 months of life. Microbiota were characterized by 16S rRNA-based sequencing and a panel of 31 antimicrobial resistance genes was tested using targeted qPCR. Confirmatory shotgun metagenomic sequencing was executed on a subset of samples. The overall gut microbial community composition and antimicrobial resistance gene profile majorly shift directly following treatment (R² = 9.5%, adjusted p-value = 0.001 and R² = 7.5%, adjusted p-value = 0.001, respectively) and normalize over 12 months (R² = 1.1%, adjusted p-value = 0.03 and R² = 0.6%, adjusted p-value = 0.23, respectively). We find a decreased abundance of Bifidobacterium spp. and increased abundance of Klebsiella and Enterococcus spp. in the antibiotic treated infants compared to controls. Amoxicillin + cefotaxime shows the largest effects on both microbial community composition and antimicrobial resistance gene profile, whereas penicillin + gentamicin exhibits the least effects. These data suggest that the choice of empirical antibiotics is relevant for adverse ecological side-effects.

Fig. 1. α-diversity.

Differences in Shannon diversity of the faecal microbiota between infants treated with the three studied antibiotic regimens and controls plotted per time point. Group differences were calculated using one-sided Wilcoxon tests. Boxplots with medians are shown; the lower and upper hinges correspond to the first and third quartiles (the 25^th and 75^th percentiles); the upper and lower whiskers extend from the hinge to the largest and smallest value no further than 1.5*IQR from the hinge; outliers are plotted individually by opaque circles; translucent circles visualize all data points (yellow = amoxicillin + cefotaxime [n = 17, n = 46, n = 46, n = 45 and n = 44 for the consecutive time points], pink = co-amoxiclav + gentamicin [n = 20, n = 45, n = 47, n = 46 and n = 44], purple = penicillin + gentamicin [n = 23, n = 48, n = 47, n = 46 and n = 45], teal = controls [n = 66, n = 80, n = 80, n = 78 and n = 74]. D = day, w = week, BA = before antibiotics, AA = after antibiotics, m = month. We observed a significant difference in Shannon diversity between each antibiotic regimen and the controls directly after antibiotic treatment (p = 4.915e-09 for amoxicillin + cefotaxime vs. controls, p = 4.056e-10 for co-amoxiclav + gentamicin vs. controls and p = 9.128e-07 for penicillin + gentamicin vs. controls). Source data are provided as a Source Data file.

Fig. 2. NMDS plots of overall gut microbiota composition stratified per regimen and time point.

Non-metric multidimensional scaling (nMDS), based on Bray–Curtis (BC) dissimilarity between samples, visualizing the overall gut microbial community composition stratified for antibiotic-treated infants and controls (a, b) and per regimen compared with controls (d, e), for the time points before and immediately after cessation of antibiotic treatment. Each data point represents the microbial community composition of one sample. The ellipses represent the standard deviation of data points belonging to each group, with the center points of the ellipses calculated using the mean of the coordinates per group. The stress of the ordination, effect sizes (R²) calculated by permutational multivariate analysis of variance (PERMANOVA)-tests and corresponding adjusted p-values (p.adj, calculated using the Benjamini-Hochberg method) are shown in the plots. In panels c and f the results of the PERMANOVA-tests for the later time points are summarized. Ac = amoxicillin + cefotaxime (yellow), cg = co-amoxiclav + gentamicin (pink), pg = penicillin + gentamicin (purple). Data points of controls are colored teal and those of antibiotic-treated infants (not-stratified per regimen) orange. Source data are provided as a Source Data file.

Fig. 3. Mean relative abundance of most abundant OTUs.

Mean relative abundances of the 15 most abundant OTUs are depicted for all samples per time point, stratified by antibiotic regimen. BA = before antibiotics, AA = after antibiotics, d = day, w = week, m = month. In some cases, multiple OTUs of individual bacterial species were identified, so OTUs are referred to by their taxonomical annotations and a rank number (shown in parentheses), which is based on the abundance of each given OTU in the overall dataset. Source data are provided as a Source Data file.

Fig. 4. NMDS plots of AMR gene profile stratified per regimen and time point.

Non-metric multidimensional scaling (nMDS) plots, based on Jaccard index of binary (presence/absence) data, visualizing the antimicrobial resistance (AMR) gene profiles stratified for antibiotic-treated infants and controls (a, b) and per regimen compared with controls (d, e), for the time points before and immediately after cessation of antibiotic treatment. Each data point represents the composition of AMR genes of one sample. The ellipses represent the standard deviation of data points belonging to each group, with the center points of the ellipses calculated using the mean of the coordinates per group. The stress of the ordination, effect sizes (R²) calculated by permutational multivariate analysis of variance (PERMANOVA)-tests and corresponding adjusted p-values (p.adj, calculated using the Benjamini–Hochberg method) are shown in the plots. In panels c and f the results of the PERMANOVA-tests for the later time points are summarized. Ac = amoxicillin + cefotaxime (yellow), cg = co-amoxiclav + gentamicin (pink), pg = penicillin + gentamicin (purple). Data points of controls are colored teal and those of antibiotic-treated infants (not-stratified per regimen) orange. Source data are provided as a Source Data file.