Multiple Introductions of Salmonella enterica Serovar Typhi H58 with Reduced Fluoroquinolone Susceptibility into Chile

Salmonella enterica serovar Typhi H58, an antimicrobial-resistant lineage, is globally disseminated but has not been reported in Latin America. Genomic analysis revealed 3 independent introductions of Salmonella Typhi H58 with reduced fluoroquinolone susceptibility into Chile. Our findings highlight the utility of enhanced genomic surveillance for typhoid fever in this region.


S almonella enterica serovars Typhi, Paratyphi A, and
Paratyphi B are the etiologic agents of typhoid and paratyphoid fever. Each year, ≈11-21 million cases and 128,000-161,000 typhoid-related deaths occur, making typhoid a continued health concern in many low-and middle-income countries, particularly among populations without access to clean water or improved sanitation (1). Salmonella Typhi H58 lineage, genotype 4.3.1, commonly is associated with multidrug resistance, including resistance to chloramphenicol, ampicillin, and trimethoprim/ sulfamethoxazole. In addition, isolates exhibiting resistance to fluoroquinolones have been linked to emergent clades of genotype 4.3.1 in South Asia (2), the spread of which could cause major challenges for disease management.
Salmonella Typhi H58 4.3.1 is the dominant genotype in many parts of Southeast and South Asia and in East Africa (3) and has spread globally but has not been reported in Latin America. Recent data on typhoid fever in South America are limited, and little is known about the population structure and antimicrobial susceptibility profiles of Salmonella Typhi on the continent. However, a report of 402 Salmonella Typhi isolates collected in Colombia during 2012-2015 showed that only 2.2% were resistant to fluoroquinolones (4). In 2016, Colombia reported collecting 204 Salmonella Typhi isolates, 12.7% of which exhibited decreased susceptibility to fluoroquinolones (5). Because these reports did not include whole-genome sequencing (WGS) data, determining whether isolates were genotype 4.3.1 is not possible.
Before the 1970s, typhoid fever was endemic in parts of South America and hyperendemic in Chile. However, water quality and sanitation improvements across the continent, partly in response to a major cholera epidemic in 1991, likely have contributed to a steep decline in the incidence of typhoid fever (6). During 1982-1992, Chile implemented interventions to reduce typhoid fever, including immunizing schoolchildren, prohibiting use of untreated sewage to irrigate crops, and detecting and treating chronic carriers. These interventions drastically reduced transmission and typhoid incidence has declined to 0.2 cases/100,000 persons (7), including in the greater Santiago metropolitan region (8).
Chile's epidemiologic surveillance system tracks suspected typhoid fever. Two thirds of cases are confirmed by pathogen isolation from ordinarily sterile body fluids, such as blood or bone marrow. Salmonella Typhi isolates from Chile typically are susceptible  Table 2). We used the non-4.3.1 genotypes and a Salmonella Paratyphi A sequence as an outgroup for phylogenetic tree rooting. We produced clean and filtered SNP alignments (Appendix 1) and used these alignments to infer maximum likelihood phylogenies and specified a generalized time-reversible model and a Gamma distribution to model site-specific rate variation by using GTRGAMMA in RAxML version 8.2.9 (https://github.com/stamatak/standard-RAxML) and 100 bootstrap pseudoreplicates to assess branch support. SNP distances were calculated by using snp-dists (https://github.com/tseemann/ snp-dists; Appendix 2). Raw Illumina reads were assembled by using either Velvet version 1.2 (European Bioinformatics Institute, https://www.ebi. ac.uk/~zerbino/velvet) or Unicycler version 0.4.7 (10). Assembled reads were input into Pathogenwatch (https://pathogen.watch) to detect nonsynonymous mutations in the quinolone-resistance determining region of gyrA and parC genes responsible for reduced fluoroquinolone susceptibility. We also used this approach to look for known antimicrobial resistance (AMR) genes. We further screened the sequences from Chile and close genetic relatives to determine molecular determinants of AMR and known plasmid replicon genes (Appendix 1 Table).
Phylogenomic and SNP analyses confirmed 7 Salmonella Typhi genotype 4.3.1 isolates from Chile. Contact tracing implies that 4/5 isolates from 2015 were part of a localized outbreak. We found that the 7 isolates were members of 2 different sublineages, lineage I (4.3.1.1) and lineage II (4.3.1.2), suggesting multiple introductions into Chile. The 2 isolates of lineage I carried a single gyrA-S83F mutation predicted to confer reduced susceptibility to fluoroquinolones. The 5 lineage II isolates carried 3 quinolone-resistance determining region mutations, 2 in gyrA genes, S83F and D87N, and 1 in parC-S80I. Genotype 4.3.1 triple mutants were predicted to be resistant to fluoroquinolones, and isolates of this sublineage with identical mutations have been observed on the subcontinent of India and have been associated with treatment failure (2,11,12). None of the lineage II triple mutants in Chile carried detectable horizontally acquired AMR genes.
To provide a global contextualization of Salmonella Typhi genotype 4.3.1 in Chile, we analyzed the novel sequences alongside 2,326 existing sequences from 31 countries (Figure 1). The 4.3.1.2 triple Multiple Introductions of Salmonella Typhi, Chile mutants from Chile formed a closely related phylogenetic cluster (median distance 2 SNPs) with sequences that have the same antimicrobial susceptibility profile isolated from India during 2012-2014, indicating an introduction from South Asia (Figure 2, panel A).
The two 4.3.1.1 isolates from 2015 and 2016 in Chile were in distinct subclades of the tree and were separated by 19 SNPs, suggestive of 2 separate introductions. Of these, 1 introduction was closely related to a 2015 isolate from India (5 SNPs apart) (Figure 2, panel B) and the other was nested in a cluster of sequences from Southeast and South Asia and most closely related (median distance of 20 SNPs) to sequences from India and Bangladesh (Figure 2, panel C).

Conclusions
Our study confirmed Salmonella Typhi H58 genotype 4.3.1 in South America. Phylogenomic and SNP analyses indicate >3 separate genotype introductions into Chile; 5/7 isolates carried 3 distinct mutations, 2 in the gyrA gene, at D87N and S83F, and 1 in the parC gene at S80I, which are associated with ciprofloxacin resistance. For a high-income country with adequate surveillance, like Chile, the presence of fluoroquinolone-resistant genotype 4.3.1 Salmonella Typhi has no immediate implications. However, if this genotype is transferred to low-or middle-income countries in South America, it could have major consequences. Therefore, these data should be of concern to other countries in the region where potential typhoid fever transmission remains high and adequate sanitation might be lacking (5,6,10). Ciprofloxacin is a first-line drug for typhoid fever in much of Latin America, and fluoroquinolone-resistant genotype 4.3.1 would reduce its long-term efficacy.
Most diagnostic laboratories across South America are using pulsed-field gel electrophoresis to study Salmonella Typhi epidemiology (13), but efforts are underway to implement WGS for epidemiologic surveillance in several countries (14,15). However, WGS-based approaches for detecting genotype 4.3.1 and understanding trends in genotype population, circulating lineages, and AMR dynamics have not been adopted widely across the region. Our work highlights the need for a uniform WGS platform for global Salmonella Typhi monitoring and the need to elucidate the current epidemiology of typhoid fever in South America.