Gut microbiome perturbation, antibiotic resistance, and Escherichia coli strain dynamics associated with international travel: a metagenomic analysis

Abstract

Background: Culture-based studies have shown that acquisition of extended-spectrum β-lactamase-producing Enterobacterales is common during international travel; however, little is known about the role of the gut microbiome before and during travel, nor about acquisition of other antimicrobial-resistant organisms. We aimed to identify (1) whether the gut microbiome provided colonisation resistance against antimicrobial-resistant organism acquisition, (2) the effect of travel and travel behaviours on the gut microbiome, and (3) the scale and global heterogeneity of antimicrobial-resistant organism acquisition.

Methods: In this metagenomic analysis, participants were recruited at three US travel clinics (Boston, MA; New York, NY; and Salt Lake City, UT) before international travel. Participants had to travel internationally between Dec 8, 2017, and April 30, 2019, and have DNA extractions for stool samples both before and after travel for inclusion. Participants were excluded if they had at least one low coverage sample (<1 million read pairs). Stool samples were collected at home before and after travel, sent to a clinical microbiology laboratory to be screened for three target antimicrobial-resistant organisms (extended-spectrum β-lactamase-producing Enterobacterales, carbapenem-resistant Enterobacterales, and mcr-mediated colistin-resistant Enterobacterales), and underwent DNA extraction and shotgun metagenomic sequencing. We profiled metagenomes for taxonomic composition, antibiotic-resistant gene content, and characterised the Escherichia coli population at the strain level. We analysed pre-travel samples to identify the gut microbiome risk factors associated with acquisition of the three targeted antimicrobial resistant organisms. Pre-travel and post-travel samples were compared to identify microbiome and resistome perturbation and E coli strain acquisition associated with travel.

Findings: A total of 368 individuals travelled between the required dates, and 296 had DNA extractions available for both before and after travel. 29 travellers were excluded as they had at least one low coverage sample, leaving a final group of 267 participants. We observed a perturbation of the gut microbiota, characterised by a significant depletion of microbial diversity and enrichment of the Enterobacteriaceae family. Metagenomic strain tracking confirmed that 67% of travellers acquired new strains of E coli during travel that were phylogenetically distinct from their pre-travel strains. We observed widespread enrichment of antibiotic-resistant genes in the gut, with a median 15% (95% CI 10-20, p<1 × 10^-10) increase in burden (reads per kilobase per million reads). This increase included antibiotic-resistant genes previously classified as threats to public health, which were 56% (95% CI 36-91, p=2 × 10^-11) higher in abundance after travel than before. Fluoroquinolone antibiotic-resistant genes were aquired by 97 (54%) of 181 travellers with no detected pre-travel carriage. Although we found that visiting friends or relatives, travel to south Asia, and eating uncooked vegetables were risk factors for acquisition of the three targeted antimicrobial resistant organisms, we did not observe an association between the pre-travel microbiome structure and travel-related antimicrobial-resistant organism acquisition.

Interpretation: This work highlights a scale of E coli and antimicrobial-resistant organism acquisition by US travellers not apparent from previous culture-based studies, and suggests that strategies to control antimicrobial-resistant organisms addressing international traveller behaviour, rather than modulating the gut microbiome, could be worthwhile.

Funding: US Centers for Disease Control and Prevention and National Institute of Allergy and Infectious Diseases.

Figure 1:. Travel-associated loss of diversity

(A) Shannon diversity of pre-travel and post-travel samples, with sample pairs from the same traveller linked by a line. (B) The absolute change in Shannon diversity observed in travellers with and without travellers’ diarrhoea, and those reporting antibiotic treatment for travellers’ diarrhoea. (C) Absolute change in Shannon diversity associated with travel to the eight most common travel destinations in the study. All box plots denote the median, IQR, and 95% quantiles. Significance was tested by t tests (paired except for between-group comparisons).

Figure 2:. Surge in Enterobacteriaceae associated with international travel

(A) For each genus present in at least 10% of samples, the proportion of travellers with an observed decrease vs increase in relative abundance. Red labelled points denote genera with significant skew. (B) The travel-associated log fold change in Enterobacteriaceae observed in travellers visiting each of the eight most common destination regions. All box plots denote the median, IQR, and 95% quantiles. Significance determined by paired t tests.

Figure 3:. Travel-acquired Escherichia coli strains are phylogenetically distinct from those present pre-travel

(A) For each E coli phylogroup with more than five observations, the proportion of travellers carrying a strain before travel is compared with the proportion of travellers acquiring a strain during travel. Acquisition is defined here as a strain detected in the post-travel sample that was not detected in the pre-travel sample. (B) The proportion of travellers visiting each travel destination who acquired a targeted AMR organism (based on culture; red) or at least one E coli strain (based on metagenomic analyses; blue). Error bars represent the standard errors of the proportions. AMR=antimicrobial resistant.

Figure 4:. Acquisition of diverse AMR-associated gene during travel

Box plots denote the median, IQR, and central 95% quantile. (A) Change in total and high-risk antibiotic-resistant gene abundance associated with travel. Significance was evaluated by Wilcoxon signed rank tests. (B) For seven common antibiotic-resistance classes, the pre-travel and post-travel abundance of AMR-associated genes are shown in RPKM. (C) For all AMR-associated genes detected, the proportions of travellers losing vs acquiring the AMR-associated genes are plotted against each other. Labelled AMR-associated genes are both significant and high risk (appendix 2 p 7). AMR=antimicrobial resistance. MLS=macrolides, lincosamides, streptogramines. RPKM=reads per kilobase per million reads. *Representatives of multi-gene groups based on identity clustering.