Bacterial Genomics for National Antimicrobial Resistance Surveillance in Cambodia

Abstract

Background: Antimicrobial resistance (AMR) surveillance in low- and middle-income countries (LMICs) often relies on poorly resourced laboratory processes. Centralized sequencing was combined with cloud-based, open-source bioinformatics solutions for national AMR surveillance in Cambodia.

Methods: Blood cultures growing gram-negative bacteria were collected at 6 Cambodian hospitals (January 2021 to October 2022). Isolates were obtained from pure plate growth and shotgun DNA sequencing performed in country. Using public nucleotide and protein databases, reads were aligned for pathogen identification and AMR gene characterization. Multilocus sequence typing was performed on whole-genome assemblies and haplotype clusters compared against published genomes.

Results: Genes associated with acquired resistance to fluoroquinolones were identified in 59%, trimethoprim/sulfamethoxazole in 45%, and aminoglycosides in 52% of 715 isolates. Extended-spectrum β-lactamase encoding genes were identified in 34% isolates, most commonly blaCTX-M-15, blaCTX-M-27, and blaCTX-M-55 in Escherichia coli sequence types 131 and 1193. Carbapenemase genes were identified in 12% isolates, most commonly blaOXA-23, blaNDM-1, blaOXA-58, and blaOXA-66 in Acinetobacter species. Phylogenetic analysis revealed clonal strains of Acinetobacter baumannii, representing suspected nosocomial outbreaks, and genetic clusters of quinolone-resistant typhoidal Salmonella and extended-spectrum β-lactamase E. coli cases suggesting community transmission.

Conclusions: With accessible sequencing platforms and bioinformatics solutions, bacterial genomics can supplement AMR surveillance in LMICs.

Keywords: Cambodia; bacterial resistance; genomic surveillance; outbreak investigation; resistance genes.

Figure 1.

Flowchart representation of national antimicrobial resistance surveillance in Cambodia. Existing processes are denoted by grey arrows. Novel workflows using bacterial genomics to inform resistance identification and reporting are denoted by orange arrows. Abbreviations: AMR, antimicrobial resistance; AST, antibiotic susceptibility testing; ID, identity; QC, quality control.

Figure 2.

A, Geographic map of Cambodia demonstrating home location of sampled patients (n = 641). Filled circles represent individual patients, colored by bloodstream pathogen. Contributing hospitals (n = 6) are denoted by a cross within a bordered circle. Inset in bottom left represents Phnom Penh and surrounding provincial lines. Distance scales are included in both inset and larger map. B, Left, bar charts of proportion of isolates representing priority pathogens for which AST was performed and reported (indicated by purple fill) for specific pathogen-drug combinations. Combinations recommended by World Health Organization GLASS and National Cambodian guidelines on antimicrobial resistance surveillance (Supplementary Table 1) are indicated by an asterisk and bold font. Right, spider charts displaying prevalence of resistance (% isolates with intermediate or full resistance indicated by distance from polygon center) as determined by AST among 5 key antibiotic classes for priority pathogens corresponding to bar charts. Abbreviations: 3GCeph, third-generation cephalosporins; AST, antibiotic susceptibility testing; GLASS, Global Antimicrobial Resistance and Use Surveillance System; GNR, gram-negative rods; TMP/SMX, trimethoprim/sulfamethoxazole.

Figure 3.

A, Geographic map of 7 Salmonella enterica serovar Paratyphi A isolates collected across 6 Cambodian hospitals, denoted by a red cross within a red bordered circle. Filled circles denote individual cases, with fill color scaled by date of sample collection (ranging from 3 January 2021 to 21 February 2022). B, Phylogenetic tree of all 7 S. enterica serovar Paratyphi A isolates in this cohort and closely related isolates (<25 allelic difference) in the Pathogen Detection repository, reconstructed from single-nucleotide polymorphisms within whole-genome multilocus sequence type clusters using maximum compatibility criteria. Branch ends indicate individual genomes, with the 7 cohort isolates denoted by circles with blue-purple spectrum fill colors corresponding to the time scale (legend shown alongside geographic map). Other isolates within the cluster are represented by orange filled circles (if from Cambodia) and unlabeled branch ends (other locations). Metadata bars indicate presence (orange) of gyrase A mutations gyrA-D87G and gyrA-S83F, along with reported isolate location.

Figure 4.

A, Geographic map of 142 Escherichia coli isolates collected across 6 Cambodian hospitals, denoted by a red cross within a red bordered circle. Filled circles denote individual cases, with fill color representing sequence type (ST; PubMLST Achtman scheme) for dominant STs; less represented STs with frequency <5 in the cohort are labeled “other.” B, Phylogenetic tree of 2 whole genome near-identical cohort E. coli isolates (ST131) within a cluster (ie, genomes with <25 allelic differences) shared by >50 isolates in the Pathogen Detection repository. Cohort isolates are denoted by black filled circles. Metadata bars indicate presence (orange) of extended-spectrum β-lactamase genes blaCTX-M-15, bla-CTX-M-27, bla-CTX-M-55,  blaTEM-30, and blaSHV-12, along with reported isolate location.

Figure 5.

Phylogenetic tree of 13 Acinetobacter baumannii ST571 isolates collected at a single hospital in September 2022 (black filled circles), and other isolates in the Pathogen Detection repository from Cambodia (single orange filled circle) and elsewhere (unlabeled branch ends) sharing the same cluster (<25 allelic differences). Metadata bars indicate presence (orange) of genes associated with carbapenem resistance, along with reported isolate location.