Pathogen diversity and antimicrobial resistance transmission of Salmonella enterica serovars Typhi and Paratyphi A in Bangladesh, Nepal, and Malawi: a genomic epidemiological study

Abstract

Background: Enteric fever is a serious public health concern. The causative agents, Salmonella enterica serovars Typhi and Paratyphi A, frequently have antimicrobial resistance (AMR), leading to limited treatment options and poorer clinical outcomes. We investigated the genomic epidemiology, resistance mechanisms, and transmission dynamics of these pathogens at three urban sites in Africa and Asia.

Methods: S Typhi and S Paratyphi A bacteria isolated from blood cultures of febrile children and adults at study sites in Dhaka (Bangladesh), Kathmandu (Nepal), and Blantyre (Malawi) during STRATAA surveillance were sequenced. Isolates were charactered in terms of their serotypes, genotypes (according to GenoTyphi and Paratype), molecular determinants of AMR, and population structure. We used phylogenomic analyses incorporating globally representative genomic data from previously published surveillance studies and ancestral state reconstruction to differentiate locally circulating from imported pathogen AMR variants. Clusters of sequences without any single-nucleotide variants in their core genome were identified and used to explore spatiotemporal patterns and transmission dynamics.

Findings: We sequenced 731 genomes from isolates obtained during surveillance across the three sites between Oct 1, 2016, and Aug 31, 2019 (24 months in Dhaka and Kathmandu and 34 months in Blantyre). S Paratyphi A was present in Dhaka and Kathmandu but not Blantyre. S Typhi genotype 4.3.1 (H58) was common in all sites, but with different dominant variants (4.3.1.1.EA1 in Blantyre, 4.3.1.1 in Dhaka, and 4.3.1.2 in Kathmandu). Multidrug resistance (ie, resistance to chloramphenicol, co-trimoxazole, and ampicillin) was common in Blantyre (138 [98%] of 141 cases) and Dhaka (143 [32%] of 452), but absent from Kathmandu. Quinolone-resistance mutations were common in Dhaka (451 [>99%] of 452) and Kathmandu (123 [89%] of 138), but not in Blantyre (three [2%] of 141). Azithromycin-resistance mutations in acrB were rare, appearing only in Dhaka (five [1%] of 452). Phylogenetic analyses showed that most cases derived from pre-existing, locally established pathogen variants; 702 (98%) of 713 drug-resistant infections resulted from local circulation of AMR variants, not imported variants or recent de novo emergence; and pathogen variants circulated across age groups. 479 (66%) of 731 cases clustered with others that were indistinguishable by point mutations; individual clusters included multiple age groups and persisted for up to 2·3 years, and AMR determinants were invariant within clusters.

Interpretation: Enteric fever was associated with locally established pathogen variants that circulate across age groups. AMR infections resulted from local transmission of resistant strains. These results form a baseline against which to monitor the impacts of control measures.

Funding: Wellcome Trust, Bill & Melinda Gates Foundation, EU Horizon 2020, and UK National Institute for Health and Care Research.

Figure 1

Distribution of pathogen variants and disease severity by age group (A) Barplots show frequency distributions of pathogen genotypes among age groups, stratified by serovar and location, for the main STRATAA dataset (n=622 Salmonella enterica serovar Typhi and n=109 Salmonella enterica serovar Paratyphi A; table 1). Simpson’s diversity index is shown under each bar (calculated from genotype counts, excluding isolates from the extended surveillance period in Blantyre to ensure comparability between locations), along with the frequencies of each serovar (S Typhi or S Paratyphi A) as a percentage of all sequenced isolates from the given location and age group. (B) Breakdown of patient sex and pathogen serovar, within each age group at each site, for the main STRATAA dataset. (C) Breakdown of patient sex and disease severity (severe was defined as a symptom duration >10 days or requirement for hospitalisation), within each age group at each site, for the main STRATAA dataset. ∗4.3.1.1 in Blantyre is sublineage 4.3.1.1.EA1.

Figure 2

Features of zero-SNV clusters of Salmonella enterica serovars Typhi and Paratyphi A Each panel summarises clusters for S Typhi in Dhaka (A), S Paratyphi A in Dhaka (B), S Typhi in Kathmandu (C), and S Typhi in Blantyre (D). Data are shown for common zero-SNV clusters (comprising at least five cases). Left panels show monthly counts for each cluster (upper panel) above a timeline, with one row per cluster, as labelled (lower panel). Darker shaded vertical areas in the plots represent the rainy season in each location. Tables summarise key information per cluster (rows as per timeline figure); colour key indicates colour code per cluster, which applies across the whole panel. Time p value and geographical p values are from Kolmogorov–Smirnov tests comparing the pairwise distribution within a given cluster to the distribution between all unclustered isolates from that site. PWD=pairwise distance. ∗Age breakdown shows ratios of participants aged <5 years to those aged 5–15 years to those aged ≥15 years.

Figure 2

Features of zero-SNV clusters of Salmonella enterica serovars Typhi and Paratyphi A Each panel summarises clusters for S Typhi in Dhaka (A), S Paratyphi A in Dhaka (B), S Typhi in Kathmandu (C), and S Typhi in Blantyre (D). Data are shown for common zero-SNV clusters (comprising at least five cases). Left panels show monthly counts for each cluster (upper panel) above a timeline, with one row per cluster, as labelled (lower panel). Darker shaded vertical areas in the plots represent the rainy season in each location. Tables summarise key information per cluster (rows as per timeline figure); colour key indicates colour code per cluster, which applies across the whole panel. Time p value and geographical p values are from Kolmogorov–Smirnov tests comparing the pairwise distribution within a given cluster to the distribution between all unclustered isolates from that site. PWD=pairwise distance. ∗Age breakdown shows ratios of participants aged <5 years to those aged 5–15 years to those aged ≥15 years.