Prevalence of carbapenem resistance and its potential association with antimicrobial use in humans and animals in rural communities in Vietnam

Abstract Background Vietnam and Southeast Asia are hotspots for antimicrobial resistance; however, little is known on the prevalence of carriage of carbapenem resistance in non-hospitalized humans and in animals. Carbapenem-resistant Enterobacteriaceae (CRE), particularly Escherichia coli (CREC) and Klebsiella pneumoniae (CRKP) and also Acinetobacter baumannii (CRAB) are emerging threats worldwide. Methods We investigated healthy humans (n = 652), chickens (n = 237), ducks (n = 150) and pigs (n = 143) in 400 small-scale farms in the Mekong Delta of Vietnam. Samples (rectal swabs, faecal swabs) were investigated for carriage of CRE/CRAB and were further characterized phenotypically and genotypically. Results In the Mekong Delta of Vietnam, the prevalence of CRE isolates in human rectal swabs was 0.6%, including 4 CREC and 1 CRKP. One pig was infected with CREC (prevalence 0.7%). CRAB was isolated from chickens (n = 4) (prevalence 2.1%) and one duck (prevalence 0.7%). CRKP was isolated from a human who was also colonized with CREC. The CRKP strain (ST16), from an 80 year-old person with pneumonia under antimicrobial treatment, genetically clustered with clinical strains isolated in a hospital outbreak in southern Vietnam. The prevalence of CRE was higher among humans that had used antimicrobials within 90 days of the sampling date than those had not (4.2% versus 0.2%) (P = 0.005). All CRE/CRAB strains were MDR, although they were susceptible to colistin and neomycin. The carbapenemase genes identified in study strains were blaNDM and blaOXA. Conclusions The finding of a CRKP strain clustering with previous hospital outbreak raises concerns about potential transmission of carbapenem-resistant organisms from hospital to community settings or vice-versa.


Introduction
Carbapenems are β-lactam antimicrobials used for the treatment of infections caused by MDR Gram-negative bacteria. 1 Currently they are classified by WHO as high priority, critically important antimicrobials; 2 carbapenem-resistant Enterobacteriaceae (CRE) and Acinetobacter baumannii (CRAB) (alongside Pseudomonas aeruginosa) are regarded as 'critical, high priority pathogens' by WHO. 3 Globally, the incidence of infections with both types of pathogen has been steadily increasing. 4,5 CRE infections are now being widely reported in Southeast Asian hospital settings. 6,7 Data from Vietnamese hospitals have documented that this leads to increased mortality and associated health care costs. 8,9 Prevalence of infection with CRE among patients correlates with length of hospitalization (from 13% on admission to 89% at day 15). 9 Laboratory data indicate increased prevalence of carbapenem resistance between 2012 and 2016 among Escherichia coli (CREC) (from 6% to 8%) and Klebsiella pneumoniae (CRKP) (from 17% to 24%). 10 Recently, two nosocomial CRKP outbreaks caused by distinct lineages of sequence type (ST) 16 have been reported in Vietnam. CRKP strains from Vietnam are typically MDR, and are resistant to colistin. 11,12 Studies in Ho Chi Minh City (southern Vietnam) (2010-12) indicated that resistance to carbapenems among Acinetobacter spp. from ventilator-associated pneumonia patients was 84%-86%, and these had been steadily increasing since 2000. 13 In another Vietnam-wide hospital study, the prevalence of carbapenem resistance among A. baumannii was seen to increase from 70% to 78% between 2012 and 2016. 14 Carbapenemase production, often encoded by genes located on plasmids, is the most common carbapenem resistance mechanism among Enterobacteriaceae. 5 Carbapenemases have been further classified into Class A (KPC); Class B (NDM, IMP, VIM); and Class D (OXA-type). 15 Previous studies in Vietnam have reported a range of carbapenem resistance genes among CREC (bla KPC , bla KPC-2 , bla NDM-1 , bla NDM-4 , bla NDM-5 , bla OXA , bla OXA-48 ), CRKP isolates (bla KPC-2 , bla NDM-1 , bla NDM-4 , bla OXA , bla OXA-48 and bla VIM ), 6,16 and CRAB isolates (bla OXA-51 , bla OXA-23 , bla OXA-58 , and bla NDM-1 ). 17 Few studies have investigated carriage of CRE among nonhospitalized human subjects. These were carried out in Spain (prevalence 0.4%); 18 China (3.6% in children; 19 2.3% in the general population) 20 and Cambodia (1%). 21 In the latter study, bla OXA-48 was identified both in E. coli and K. pneumoniae. In addition, a number of studies have demonstrated CRE in animal reservoirs (domestic, wild, companion) and food. 22 CRE was not detected among 285 livestock samples (including ruminants, pigs and poultry) investigated in Cambodia. 21 In contrast, a study in China identified CRE in 10.6% and 3.9% of pigs and chickens, respectively. 20 Another study from China detected CREC, CRKP and Enterobacter cloacae in, respectively, 21.8%, 7.4% and 3.9% of poultry samples along the production chain, with bla NDM detected in 33.2% samples. 23 Studies on healthy populations in Netherlands and the USA have identified carriage of A. baumannii in 0.9% (faeces) and 10.4% (hands) individuals, respectively. 24,25 Studies in Germany and Switzerland identified A. baumannii in 2.7%-45.7% of poultry samples investigated (choana and raw meat). However, none was identified as CRAB. 26,27 Previous studies identified exposure to several classes of antimicrobials as a key explanatory factor for colonization with CRE. [28][29][30] In Vietnam, information on the prevalence of carriage of CREC/ CRKP/CRAB in healthy livestock and in-contact human communities is limited. Vietnam and Southeast Asia are considered hotspots for antimicrobial usage (AMU)/antimicrobial resistance (AMR). 31 It is not known to what extent this may affect the colonization with CREC/CRKP/CRAB. This knowledge is essential for effective risk management of carbapenem resistance in the country. Using a One Health approach involving co-sampling of animals and human residents, we aimed to investigate: (1) the prevalence of carriage of carbapenem-resistant bacteria among livestock and in-contact humans in the Mekong Delta (Vietnam) and its potential relationship with antimicrobial use; and (2) the genetic determinants of carbapenem resistance in CREC, CRKP and CRAB in this area.

Ethics
The project was conducted in accordance with the Declaration of Helsinki following institutional standards. The study was granted ethics approval by Oxford University Ethics Committee (OxTREC No. 503-20).

Sample and data collection
The study was carried out in Dong Thap province, considered to be representative of the Mekong Delta region of Vietnam in terms of human and animal demographics. The province has a census population of 1.6 million of whom 80.9% are classed as urban and a human population density of 494.1 per km 2 (versus 70.4% rural and a population density of 426.8 per km 2 in the region as a whole). Poultry and pig farm owners in Dong Thap province were randomly selected from the official farm census held by the veterinary authority (Dong Thap Sub-Department of Animal Health, Production and Aquaculture, SDAH-DT). We aimed to recruit 400 farms using a cluster sampling technique (i.e., based on random selection of 20 out of 141 communes in the province, and an average of 20 farms from each commune). Farms raising poultry (chickens or ducks) in flocks with .20 birds, and/or pigs (.2) were eligible. Farmers that consented to the study were enrolled, and their farms were visited during June and July 2020 by SDAH-DT (to collect animal data and samples) and Dong Thap Center for Disease Control (CDC-DT) staff (human data and samples). Data on AMU were collected using structured questionnaires aimed at the person with primary responsibility for animal husbandry. Farmers were also asked to provide all packages (bottles, sachets, etc.) of any antimicrobial-containing products used by humans or animals over the previous 90 and 7 days, respectively. Rectal swabs were collected from 1-3 consenting individuals living in each household. Pooled faecal samples were collected from each type of food-producing animal (chicken, duck and pig) present in the farm. This was achieved by swabbing with a cotton swab, three visibly fresh droppings from each of the target species. Swab samples were placed in 1 mL of sterile brain heart infusion broth (Oxoid, UK) plus 20% glycerol (Sigma, USA).

Isolation of carbapenem-resistant bacteria
All swab samples were vortexed thoroughly, then a loop of the corresponding suspension was plated onto Chromagar-carbapenem agar supplemented with meropenem 2 mg/L (Nam Khoa, Vietnam) to screen for non-susceptible E. coli, K. pneumoniae and A. baumannii. The plates were incubated at 35 + 2°C for 20 h. Up to three suspected E. coli (reddish), K. pneumoniae (metallic blue) and A. baumannii (white) colonies from each sample were confirmed using MALDI-TOF (Bruker, Germany). E. coli ATCC 25922 was used as a negative control, two CRKP and CRAB were used as positive controls.

Antimicrobial susceptibility testing
Phenotypic AST was performed using disc diffusion and Etest (meropenem) methods for A. baumannii. For CRE and CRAB isolates VITEK 2 (bioMerieux, France) and Sensititre AST was used (Thermo Fisher Scientific, UK). The AST panel included 42 (CRE) and 26 antimicrobials (CRAB) (belonging to 13 classes and 10 classes, respectively) (Table S1, available as Supplementary data at JAC-AMR Online). CLSI breakpoints were used for susceptibility interpretation. 32 E. coli ATCC 25922 was used for quality control. In this study, isolates that were intermediately resistant to carbapenems were regarded as resistant.

WGS
Genomic DNA was extracted from carbapenem-resistant isolates using Wizard genomic DNA extraction kit (Promega, US). Genome library preparation was carried out using Nextera XT library preparation kit and WGS was performed on the HiSeq2500 Illumina platform to generate 100 paired-end reads (Macrogen, Korea).

Data analyses
The prevalence of CREC, CRKP and CRAB in human/animal samples was compared between those individuals/animals using and not using Yen et al.
antimicrobials using Fisher's exact test. Acquired AMR and virulence genes as well as plasmid replicons were identified using SRST2 v0.2.0 33 with ARG-ANNOT antimicrobial resistance, 34 BIGSdb virulence genes (https://bigsdb.web.pasteur.fr) and PlasmidFinder 35 databases, respectively. We used SRST2 with the corresponding MLST scheme downloaded from PubMLST (https://pubmlst.org/mlst). SRST2 used Bowtie2 36 to map out raw reads against the reference database and SAMtools v1.3 37 to identify genes and alleles. We assembled all Illumina reads using the de novo assembler Unicycler v0.4.8 with the default settings. 38 Prokka v.1.5 was used to annotate the assembled contigs. 39 To investigate the phylogenetics of the K. pneumoniae ST16 isolate, we combined its genomic data with that from nine isolates obtained from a previous study in Vietnam. 11 Raw Illumina reads were mapped to the reference genome MGH78578 (CP000647.1) using RedDog pipeline v1.10b (https://github.com/katholt/RedDog). In brief, RedDog used Bowtie2 v2.2.3 to map raw reads and single nucleotide polymorphisms (SNPs) were identified with SAMTools v1.3. Gubbins v1.4.5 was used to remove recombinant regions from the resulting alignment file; SNPs identified in the recombinant regions were subsequently removed, resulting in a final alignment of 142 SNPs. Randomized Axelerated Maximum Likelihood (RaxML) was used to construct a maximum likelihood (ML) phylogenetic tree using GTR + G model of nucleotide substitution with 100 bootstrap replicates. 40 Sequence data of all CRE and CRAB isolates described in this study have been deposited in the European Nucleotide Archive (ENA) under study accession number PRJEB49509 (ERS8847840 to ERS8847852).

Prevalence of CRE and CRAB in human and animals
A total of 652 human rectal swabs were collected from 381 households (19 subjects did not consent to rectal swabbing). About half (49.9%) human subjects were male, with a median age of 50 years (IQR 39-59). The prevalence of CRE isolates from human rectal swabs was 0.6%, including four CREC and one CRKP; CRAB was not found ( Table 1). The prevalence of CRE was higher among individuals who had used antimicrobials in the previous 90 days than among those who had not (4.2% versus 0.2%, Fisher exact test, P = 0.005).
A total of 530 pooled faecal swabs were collected from animals in 400 households (237 chickens, 143 pigs and 150 ducks). The prevalence of carbapenem-resistant CRE/CRAB in chicken samples (2.1%) was greater than in pig/duck samples (0.7% each). In animals, a total of seven CRAB (six from chickens, one from ducks) and one CREC (from pig) were identified. Among chicken samples, the prevalence of CRAB was higher among flocks not recently treated with antimicrobials (3.4% versus 0.8%). In pigs and ducks, carbapenem-resistant bacteria were only detected in (one each) herd/flock that had not been treated with antimicrobials over the previous 7 days (Table 1).
Most carbapenem-resistant isolates came from different farms and samples, except one CREC and one CRKP isolate from the same human sample, and two distinct CRAB isolates from the same chicken sample by MLST (Table S2). CRE/CRAB bacteria were found in 6 out of 8 districts investigated, but 6/11 (55%) of the CRE/CRAB-positive samples were collected from Lai Vung district ( Figure 1) (Tables S3 and S4).

Antimicrobial consumption and demographic features
Antimicrobial consumption and demographic features of CRE/ CRAB-positive hosts are described in Tables S3 and S4. Three out of four human individuals whose rectal swab cultures were positive with CRE had a recent history of AMU within 90 days of the sampling date, including amoxicillin/clavulanic acid (n = 2), cefuroxime (n = 1), cefixime (n = 1), and clarithromycin (n = 1). The individual detected with both CREC and CRKP had pneumonia and was being treated with cefixime at the time of sample collection. In one CRAB-positive chicken farm raising fighting cockerels, tilmicosin and gentamicin had been used during the previous week. Carbapenem resistance in rural communities, Vietnam

Sequence types (STs)
Based on the MLST profile, we identified five distinct E. coli STs: ST2705, ST1638, ST38, ST405 and ST398 (one for each isolate). The K. pneumoniae isolate was ST16. Out of the seven A. baumannii isolates, two (29%) were identified as ST762, and five (71%) were classified as STNF, (i.e. the ST was not identified based on the current MLST scheme). However, two out of five STNF isolates (from two chickens located on different farms) shared the same MLST profile (Table S2).

Discussion
This is, to the best of our knowledge, the first report describing carriage of CRE and CRAB in animals and humans living in close contact in rural (Mekong Delta) Vietnam. Our study confirms the presence of CRE in non-hospitalized human subjects and pigs at a relatively low prevalence (0.6%-0.7%). Although we did not find evidence of CRE in poultry species, CRAB was detected in 2.1% and 0.7% chickens and ducks. Carbapenem resistance was encoded by bla NDM-1-like (4/5 CREC) and bla OXA-181 (1/5 CREC and 1/1 CRKP) genes. One CREC contained both bla NDM-1-like and bla OXA-1 .
The observed prevalence of carriage of CRE (0.6%) was of similar magnitude to a previous study in rural Cambodia ( 1%); however, in that study CRE was not identified in any of 285 livestock faecal samples. In that study, both CREC and CRKP isolates detected in humans harboured bla OXA-48 . 21 Previous studies have detected bla OXA-48 -positive E. coli in 0.09% (1/1086) and 1.6% (3/183) healthy humans from Switzerland and Lebanon,respectively. 41,42 In contrast, no CRE was detected in 433 and 320 non-hospital human samples from India or Spain. 43,44 In China, carriage of CREC was 2.3% of 735 non-hospitalized humans, 3.9% of 305 chickens and 10.6% of 417 pigs (all encoded by bla NDM carbapenemase genes). 20 Another study in Egypt revealed high levels of CRKP carriage in chickens with signs of respiratory diseases (15%) as well as in workers and veterinarians (10% of 49 faecal samples). The CRKP strains carried bla KPC , bla OXA-48 and bla NDM genes. 41 A previous study demonstrated a higher prevalence of CRE among Vietnamese hospital patients on day of admission (13%), 9 a much higher figure than our 0.6% among healthy individuals. It is likely that this difference reflects previous antimicrobial use or exposure to healthcare facilities.
The finding of a CRKP strain in a province identical to that found in a previous nosocomial outbreak in a crowded city in the country 11 suggests that transmission of CRKP from hospital facilities to the community may occur.
The finding of a CRKP strain in these rural settings identical to that found in a previous nosocomial outbreak in a hospital the country 11 and the high prevalence of colonization in hospital settings confirms transmission from hospital facilities to the community.
We only detected CRAB in poultry faecal samples, all of which harboured bla OXA-51-like genes, and most (6/7) bla OXA-24-like genes. Carbapenem resistance in rural communities, Vietnam A study in Germany identified bla OXA-51-like -producing A. baumannii in choanal swab samples of chickens (2.7%) and geese (7.5%). 26 CRAB was detected in 5.8% (3/52) and 11.2% (112/ 1000) poultry meat in Iran and Turkey, respectively. 45,46 In contrast, a study in Switzerland demonstrated that poultry was the most frequently A. baumannii-contaminated type of meat (45.7% of 94 samples); however, none of them were carbapenem resistant. 27 Meat is suspected to be a potential source of MDR A. baumannii, presumably resulting from faecal contamination. 47 Although we did not investigate meat samples, our study suggests that poultry (including its meat) may potentially be a source of infection of CRAB and therefore this merits further study.
Although none of the four CRE-carrying individuals investigated had been recently treated with carbapenems, we found a strong association between antimicrobial use in the last 90 days and carriage of CRE. A previous study identified antimicrobial usage as the single most important explanatory factor for colonization with CRE. 28 Further interview data (data not shown) revealed that all CRE-positive individuals had visited health care facilities. It is not known to what extent individuals may have been colonized in these settings. We did not, however, investigate to what extent farming practices and exposure to manure may have contributed to colonization with CRE in human subjects, as shown in the Cambodian study. 21 Our data was suggestive of geographical clustering for CRAB and CREC, four chicken CRAB and two human CREC isolates came from the same district (Lai Vung).
Our study confirmed the presence of bla NDM-1-like and bla OXA genes in CREC strains. For meropenem and imipenem, the highest MICs (.16 mg/L) were observed among strains carrying bla NDM-1-like genes. We did not, however, find evidence of colistin resistance among any of the tested strains. In addition to colistin, amikacin and neomycin (aminoglycosides) were the three antimicrobials to which all CREC strains were susceptible; in the case of CRKP, only neomycin, streptomycin and colistin had inhibitory activity.

Conclusions
We demonstrated faecal carriage of E. coli, K. pneumoniae and A. baumannii harbouring carbapenemase genes in humans and  Yen et al.
animals in the Mekong Delta of Vietnam. The highest prevalence of colonization with CREC corresponded to human subjects previously treated with β-lactams and/or in contact with health care facilities. Our results suggest One Health genomic surveillance for CRE/CRAB to detect potential transmission from hospital settings; this could be implemented by longitudinal follow-up sampling of individuals and their animal contacts after being discharged from hospitals. It would also be important to investigate short-versus long-term fitness of carbapenemase gene-encoding plasmids in these individuals.