The role of animals as a source of antimicrobial resistant nontyphoidal Salmonella causing invasive and non-invasive human disease in Vietnam

Abstract

Background: Nontyphoidal Salmonella (NTS) are associated with both diarrhea and bacteremia. Antimicrobial resistance (AMR) is common in NTS in low-middle income countries, but the major source(s) of AMR NTS in humans are not known. Here, we aimed to assess the role of animals as a source of AMR in human NTS infections in Vietnam. We retrospectively combined and analyzed 672 NTS human and animal isolates from four studies in southern Vietnam and compared serovars, sequence types (ST), and AMR profiles. We generated a population structure of circulating organisms and aimed to attribute sources of AMR in NTS causing invasive and noninvasive disease in humans using Bayesian multinomial mixture models.

Results: Among 672 NTS isolates, 148 (22%) originated from human blood, 211 (31%) from human stool, and 313 (47%) from animal stool. The distribution of serovars, STs, and AMR profiles differed among sources; serovars Enteritidis, Typhimurium, and Weltevreden were the most common in human blood, human stool, and animals, respectively. We identified an association between the source of NTS and AMR profile; the majority of AMR isolates were isolated from human blood (p < 0.001). Modelling by ST-AMR profile found chickens and pigs were likely the major sources of AMR NTS in human blood and stool, respectively; but unsampled sources were found to be a major contributor.

Conclusions: Antimicrobial use in food animals is hypothesized to play role in the emergence of AMR in human pathogens. Our cross-sectional population-based approach suggests a significant overlap between AMR in NTS in animals and humans, but animal NTS does explain the full extent of AMR in human NTS infections in Vietnam.

Keywords: Antimicrobial resistance; Bacteremia; Diarrhea; Nontyphoidal Salmonella; Zoonosis.

Fig. 1

The population structure of NTS isolated from humans and animals in Vietnam. Minimum spanning tree of the 672 different NTS isolates subjected to MLST. The sequence type (ST) of each allelic profile is labelled as are the major inferred serovars. Branch lengths are associated with the number of allelic variations between the STs and clonal complexes are shaded in grey. The source of each isolate (and contribution to each ST) is colour coded and the the size of each ST circle corresponds to the number of isolates per ST profile (scale shown).

Fig. 2

Source attribution of human NTS isolates in Vietnam by sequence type. Violin plots showing the results of the source attribution model for NTS infections in human blood (A) and human stool (B). Each plot represents the mixture coefficient (α), which is an estimated proportion of NTS cases attributed to each source according to sequence type using Bayesian multinomial mixture modelling with sampled sources. The centre of each violin represents the median, the length represents credibility interval and the shape displays frequencies of values. Each of the sampled sources is labelled on the x axis and the proportional contribution (source attribution) is labelled on the y axis.

Fig. 3

Source attribution of human NTS isolates in Vietnam by AMR profile. Violin plots showing the results of the source attribution model for NTS infections in human blood (A) and human stool (B). Each plot represents the mixture coefficient (α), which is an estimated proportion of NTS cases attributed to each source according to AMR profile using Bayesian multinomial mixture modelling with sampled sources. The centre of each violin represents the median, the length represents credibility interval and the shape displays frequencies of values. Each of the sampled sources is labelled on the x axis and the proportional contribution (source attribution) is labelled on the y axis.

Fig. 4

The number and proportion of unique and shared ST-AMR profiles among human and animal NTS isolates in Vietnam. Venn diagrams showing the number of unique and shared ST-AMR profiles between organisms isolated from human blood and animals (A) and human stools and animals (C). Pie charts showing the unique and shared ST-AMR profiles between organisms isolated from human blood and animals (B) and human stools and animals (D).

Fig. 5

Source attribution of human NTS isolates in Vietnam by ST-AMR profile. Violin plots showing the results of the source attribution model for NTS infections in human blood (A) and human stool (B). Each plot represents the mixture coefficient (α), which is an estimated proportion of NTS cases attributed to each source according to combined ST-MR profile using Bayesian multinomial mixture modelling with sampled sources. The centre of each violin represents the median, the length represents credibility interval and the shape displays frequencies of values. Each of the sampled sources is labelled on the x axis and the proportional contribution (source attribution) is labelled on the y axis.