Genomic investigation of Staphylococcus aureus recovered from Gambian women and newborns following an oral dose of intra-partum azithromycin

Abstract

Background: Oral azithromycin given during labour reduces carriage of bacteria responsible for neonatal sepsis, including Staphylococcus aureus. However, there is concern that this may promote drug resistance.

Objectives: Here, we combine genomic and epidemiological data on S. aureus isolated from mothers and babies in a randomized intra-partum azithromycin trial (PregnAnZI) to describe bacterial population dynamics and resistance mechanisms.

Methods: Participants from both arms of the trial, who carried S. aureus in day 3 and day 28 samples post-intervention, were included. Sixty-six S. aureus isolates (from 7 mothers and 10 babies) underwent comparative genome analyses and the data were then combined with epidemiological data. Trial registration (main trial): ClinicalTrials.gov Identifier NCT01800942.

Results: Seven S. aureus STs were identified, with ST5 dominant (n = 40, 61.0%), followed by ST15 (n = 11, 17.0%). ST5 predominated in the placebo arm (73.0% versus 49.0%, P = 0.039) and ST15 in the azithromycin arm (27.0% versus 6.0%, P = 0.022). In azithromycin-resistant isolates, msr(A) was the main macrolide resistance gene (n = 36, 80%). Ten study participants, from both trial arms, acquired azithromycin-resistant S. aureus after initially harbouring a susceptible isolate. In nine (90%) of these cases, the acquired clone was an msr(A)-containing ST5 S. aureus. Long-read sequencing demonstrated that in ST5, msr(A) was found on an MDR plasmid.

Conclusions: Our data reveal in this Gambian population the presence of a dominant clone of S. aureus harbouring plasmid-encoded azithromycin resistance, which was acquired by participants in both arms of the study. Understanding these resistance dynamics is crucial to defining the public health drug resistance impacts of azithromycin prophylaxis given during labour in Africa.

© The Author(s) 2019. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy.

Figures

Figure 1.

Flow chart showing the sample selection criteria from women and babies in the PregnAnZI trial. NPS, nasopharyngeal swab; AZMR, azithromycin resistant; AZMS, azithromycin susceptible. Resistance Etest cut-off values: azithromycin ≥8 mg/L.

Figure 2.

Distribution and phylogenetic relatedness of participant isolates. Illustrated are ML phylogenetic trees for each ST identified in the study population (excluding the ST1-SLV), inferred from whole-genome alignments. Branch tips are coloured based on phenotypic susceptibility to azithromycin (blue=susceptible and red=resistant, as determined by Etest). Scale bar indicates the estimated substitutions per site. Isolates recovered from the same participant are indicated by a larger circle, labelled with the participant ID and coloured based on the treatment arm to which the participant was assigned (purple=azithromycin and green=placebo). Participants with isolates representing multi-STs are connected by a dotted line. Isolates belonging to ST1 and ST152 were found to harbour the PVL toxin-encoding gene, lukFS.

Figure 3.

Individual participant timelines. Illustrated are the isolate timelines for each participant selected for the study, grouped based on the selection criteria outlined in Figure 1. Each participant is indicated by their participant ID and the adjacent circles represent the isolates recovered from samples taken prior to the intervention (day 0 for mothers) and following the intervention on days 3, 6, 14 and 28 (for mothers and babies). Isolate circles are coloured based on the phenotypic susceptibility to azithromycin (blue=susceptible and red=resistant, as determined by Etest). The internal number indicates the in silico MLST of the isolate and the superscript letter indicates the macrolide resistance gene detected, ‘m’ for msr(A) and ‘e’ for erm(C). Black lines connecting adjacent isolates indicate that the same clone has probably been maintained, with the pairwise core SNP distance provided. Alternatively, gaps between adjacent isolates indicate a potential clonal replacement event. AZMR, azithromycin resistant; AZMS, azithromycin susceptible.

Figure 4.

Schematic for the structure of pS80062MN28. Illustrated is the structure of pS80062MN28 (central sequence PRJEB31151) with genes of interest annotated. pS800262MN28 has been aligned with S. aureus plasmid pJSA01 (accession AP014922.1) and S. epidermidis plasmid pSE95_1 (accession CP024438.1). Connecting lines indicate sequences that share >99% nucleotide sequence homology, coloured red or blue if the sequence was present in the same or reverse orientation, respectively.

Figure 5.

Phylogenetic contextualization of study isolates. (a) Illustrated is an ML phylogenetic tree for a global population of 7192 S. aureus isolates. The Gambian isolates identified in this study are indicated by white circles located at the branch tips. (b) A subtree of CC5 isolates and (c) a subtree of CC15 isolates. In both subtrees, branches with <85% support (approximate likelihood ratio test) are labelled with a red circle. The Gambian isolates identified in this study are indicated by circles located at the branch tips, coloured based on phenotypic susceptibility to azithromycin (blue=susceptible and red=resistant, as determined by Etest). Adjacent to each subtree is a vertical line of circles indicating the countries from which the isolates were recovered (refer to key); the heatmap indicates the presence or absence of three macrolide resistance genes: erm(C) (1), msr(A) (2) and mph/ere (3). The scale bar indicates the estimated number of core genome substitutions.