blaNDM-1–positive Klebsiella pneumoniae from Environment, Vietnam

To the Editor: The blaNDM-1 gene, which produces the New Delhi metallo-β-lactamase (NDM-1) enzyme, confers resistance to the carbapenem class of antimicrobial drugs and can be transferred among different types of bacteria. NDM-1 was identified in 2008 in Sweden from a patient from India who had been hospitalized in New Delhi (1). Since that report, blaNDM-1–positive bacteria have been identified from patients in several countries; most of these patients had a direct link with the Indian subcontinent (2). The spread of blaNDM-1 among bacterial pathogens is of concern not only because of resistance to carbapenems but also because such pathogens typically are resistant to multiple antimicrobial drug classes, which leaves few treatment choices available (3–5). In 2011, spread of blaNDM-1–positive bacteria in an environmental setting in New Delhi was reported (6). 
 
The possible appearance of bacteria harboring blaNDM-1 in Vietnam is of concern because cultural and economic links between Vietnam and India are strongly established, including extensive person-to-person exchanges that could enable easy exchange of pathogens. In addition, Vietnam faces a serious problem of antimicrobial drug resistance because drugs are freely available and used in an indiscriminate fashion. Thus, once blaNDM-1–positive bacteria colonize persons in Vietnam, they would be able to spread easily and pose a serious public health threat. 
 
During September 2011, we collected paired swab samples (1 for PCR, 1 for culture) of seepage water from 20 sites (rivers, lakes, and water pools in streets) within a 10-km radius of central Hanoi, Vietnam. Samples were transported in Transystem (COPAN Italia S.p.A, Brescia, Italy) to preserve bacteria and DNA. The 20 PCR swab specimens were squeezed out into 0.5-mL volumes of sterile water and centrifuged at 3,000 × g for 30 seconds; 1 μL of the resulting suspension was then used as PCR template to detect blaNDM-1 as described (7). Two samples were positive for blaNDM-1; these 2 samples were collected from the same river (Kim Nguu River) but at sites 3 km apart. 
 
To isolate and identify the phenotype and genotype of blaNDM-1–positive bacteria, we repeatedly spread the 20 culture swab specimens onto Muller-Hinton agar (Nissui, Tokyo, Japan) containing 100 mg/L vancomycin (Nakalai, Kyoto, Japan) plus 0.5 mg/L meropenem (LKT Laboratories, St. Paul, MN, USA) until single colonies were obtained. Each colony was then subcultured by plating onto MacConkey agar (Nihon Seiyaku, Tokyo, Japan) containing 0.5 mg/L meropenem to ensure culture purity; colonies were identified by using API 20E strips (bioMerieux, Basingstoke, UK). MICs of these isolates for 13 antimicrobial drugs were calculated by using Etest (bioMerieux), and susceptibility data were interpreted by using Clinical and Laboratory Standards Institute guidelines (www.clsi.org). 
 
We harvested several species of bacteria from the 2 seepage samples positive for blaNDM-1: Acinetobacter baumannii, Klebsiella pneumoniae, Pseudomonas aeruginosa, P. fluorescens/putida, and P. luteola. These isolates were placed onto media containing 0.5 mg/L meropenem, and bacterial DNA was extracted and used for the template for PCR analysis to detect blaNDM-1 as described (7). blaNDM-1 was detected in 3 K. pneumoniae isolates from each of the 2 positive samples (6 isolates total); this result was confirmed by sequencing. All 6 isolates were highly resistant to all β-lactam antimicrobial drugs, including carbapenems (Table). To detect another β-lactamase, multiplex PCRs were carried out as described (8); genetic variants blaTEM, blaSHV, blaOXA, blaCTX-M, blaIMP, blaVIM, and blaKPC were not detected in any of the isolates other than K. pneumoniae. All 6 K. pneumoniae isolates were positive for blaTEM and blaCTX-M variants by PCR; these variants were confirmed as blaTEM-1 and blaCTX-M-3 by sequencing. 
 
 
 
Table 
 
Resistance to 13 antimicrobial drugs of blaNDM-1–positive Klebsiella pneumoniae isolates from the Kim Nguu River, Hanoi, Vietnam* 
 
 
 
Aminoglycosides are often used in the management of severe infectious diseases caused by gram-negative pathogens. 16S rRNA methylases were found to confer high levels of resistance to aminoglycosides such as amikacin, tobramycin, and gentamicin. The 6 K. pneumoniae isolates we found were highly resistant to gentamicin (MIC >1,024 mg/L) and tobramycin (MIC 256–>1,024 mg/L) (Table). Therefore, we screened genetic elements of 16S rRNA methylases (rmtB, rmtC, and armA) by PCR and detected rmtB in all 6 isolates (9). Multilocus sequence typing was applied for these 6 isolates; all were identified as K. pneumoniae sequence type 283 (10), which had not been reported as harboring blaNDM-1. The azide-resistant Escherichia coli strain J53 has been used as recipient for conjugation assay, which had been reported previously (6), but we found no transconjugant strain with blaNDM-1 on MacConkey agar containing 100 mg/L sodium azide and 0.5 mg/L meropenem. 
 
Our results show that blaNDM-1–positive K. pneumoniae sequence type 283 is present in the Kim Nguu River, which flows through the central part of Hanoi at 2 sites. The isolates we obtained were also positive for 2 other β-lactamases, blaTEM-1 and blaCTX-M-3, were highly resistant to aminoglycosides related to rmtB, and showed mild elevation of MIC against ciprofloxacin up to 1.5 mg/L. Wide-scale surveillance of environmental and clinical samples in Vietnam and establishment of a strategy to prevent further spread of blaNDM-1 are urgently needed.

bla NDM-1 -positive Klebsiella pneumoniae from Environment, Vietnam To the Editor: The bla NDM-1 gene, which produces the New Delhi metallo-β-lactamase (NDM-1) enzyme, confers resistance to the carbapenem class of antimicrobial drugs and can be transferred among different types of bacteria. NDM-1 was identifi ed in 2008 in Sweden from a patient from India who had been hospitalized in New Delhi (1). Since that report, bla NDM-1 -positive bacteria have been identifi ed from patients in several countries; most of these patients had a direct link with the Indian subcontinent (2). The spread of bla NDM-1 among bacterial pathogens is of concern not only because of resistance to carbapenems but also because such pathogens typically are resistant to multiple antimicrobial drug classes, which leaves few treatment choices available (3)(4)(5). In 2011, spread of bla NDM-1 -positive bacteria in an environmental setting in New Delhi was reported (6).
The possible appearance of bacteria harboring bla NDM-1 in Vietnam is of concern because cultural and economic links between Vietnam and India are strongly established, including extensive person-to-person exchanges that could enable easy exchange of pathogens. In addition, Vietnam faces a serious problem of antimicrobial drug resistance because drugs are freely available and used in an indiscriminate fashion. Thus, once bla NDM-1positive bacteria colonize persons in Vietnam, they would be able to spread easily and pose a serious public health threat.
During September 2011, we collected paired swab samples (1 for PCR, 1 for culture) of seepage water from 20 sites (rivers, lakes, and water pools in streets) within a 10-km radius of central Hanoi, Vietnam. Samples were transported in Transystem (CO-PAN Italia S.p.A, Brescia, Italy) to preserve bacteria and DNA. The 20 PCR swab specimens were squeezed out into 0.5-mL volumes of sterile water and centrifuged at 3,000 × g for 30 seconds; 1 μL of the resulting suspension was then used as PCR template to detect bla NDM-1 as described (7). Two samples were positive for bla NDM-1 ; these 2 samples were collected from the same river (Kim Nguu River) but at sites 3 km apart.
To isolate and identify the phenotype and genotype of bla NDM-1 -positive bacteria, we repeatedly spread the 20 culture swab specimens onto Muller-Hinton agar (Nissui, Tokyo, Japan) containing 100 mg/L vancomycin (Nakalai, Kyoto, Japan) plus 0.5 mg/L meropenem (LKT Laboratories, St. Paul, MN, USA) until single colonies were obtained. Each colony was then subcultured by plating onto MacConkey agar (Nihon Seiyaku, Tokyo, Japan) containing 0.5 mg/L meropenem to ensure culture purity; colonies were identifi ed by using API 20E strips (bioMérieux, Basingstoke, UK). MICs of these isolates for 13 antimicrobial drugs were calculated by using Etest (bioMérieux), and susceptibility data were interpreted by using Clinical and Laboratory Standards Institute guidelines (www.clsi.org).
We harvested several species of bacteria from the 2 seepage samples positive for bla NDM-1 : Acinetobacter baumannii, Klebsiella pneumoniae, Pseudomonas aeruginosa, P. fl uorescens/putida, and P. luteola. These isolates were placed onto media containing 0.5 mg/L meropenem, and bacterial DNA was extracted and used for the template for PCR analysis to detect bla NDM-1 as described (7). bla NDM-1 was detected in 3 K. pneumoniae isolates from each of the 2 positive samples (6 isolates total); this result was confi rmed by sequencing. All 6 isolates were highly resistant to all β-lactam antimicrobial drugs, including carbapenems (Table). To detect another β-lactamase, multiplex PCRs were carried out as described (8); genetic variants bla TEM , bla SHV , bla OXA , bla CTX-M , bla IMP , bla VIM , and bla KPC were not detected in any of the isolates other than K. pneumoniae. All 6 K. pneumoniae isolates were positive for bla TEM and bla CTX-M variants by PCR; these variants were confi rmed as bla TEM-1 and bla CTX-M-3 by sequencing.
Aminoglycosides are often used in the management of severe infectious diseases caused by gram-negative pathogens. 16S rRNA methylases were found to confer high levels of resistance to aminoglycosides such as amikacin, tobramycin, and gentamicin. The 6 K. pneumoniae isolates we found were highly resistant to gentamicin (MIC >1,024 mg/L) and tobramycin (MIC 256->1,024 mg/L) (Table). Therefore, we screened genetic elements of 16S rRNA methylases (rmtB, rmtC, and armA) by PCR and detected rmtB in all 6 isolates (9). Multilocus sequence typing was applied for these 6 isolates; all were identifi ed as K. pneumoniae sequence type 283 (10), which had not been reported as harboring bla NDM-1 . The azide-resistant Escherichia coli strain J53 has been used as recipient for conjugation assay, which had been reported previously (6), but we found no transconjugant strain with bla NDM-1 on MacConkey agar containing 100 mg/L sodium azide and 0.5 mg/L meropenem.
Our results show that bla NDM-1positive K. pneumoniae sequence type 283 is present in the Kim Nguu River, which fl ows through the central part of Hanoi at 2 sites. The isolates we obtained were also positive for 2 other β-lactamases, bla TEM-1 and bla CTX-M-3 , were highly resistant to aminoglycosides related to rmtB, and showed mild elevation of MIC against ciprofl oxacin up to 1.   (1). Many fl eas are capable of transmitting the following pathogens to their hosts: bacteria (e.g., Rickettsia typhi, R. felis, Yersinia pestis, and many Bartonella spp.); viruses (e.g., myxoma virus); protozoa (e.g., Trypanosoma spp.); or helminths (e.g., Hymenolepis spp.) (2). Ctenocephalides spp. fl eas are of special interest as main reservoirs and vectors of R. felis, because this agent causes an emerging disease, fl eaborne rickettsiosis. The distribution and prevalence of this disease have not been well studied. Symptoms of this disease range from mild to moderate and include fever, cutaneous rash, and sometimes an inoculation eschar (3,4). R. felis can also infect at least 10 other species of arthropods, including P. irritans fl eas, trombiculid and mesostygmata mites, hard and soft ticks, and booklice (5,6).
In Africa, the presence of R. felis in fl eas has been documented in Algeria, Tunisia, Egypt, Ethiopia, Gabon, Côte d'Ivoire, and the Democratic Republic of Congo (5). Recent studies conducted in Senegal (3) and Kenya (4) have shown that as much as 4.4% and 3.7%, respectively, of acute febrile diseases in these regions may be caused by R. felis infections. We conducted a study to determine the distribution and prevalence of R. felis in fl eas in Ethiopia.
In our study, 55 fl eas were collected in 2010 in 2 villages in Ethiopia; 25 fl eas were collected from Tikemit Eshet (6°51′837″N and 35°51′348″E; altitude2,121 m), and 30 fl eas were collected from Mizan Teferi (6°59′640″N and 35°35′507″E; altitude 1,700 m). The specimens were collected by using a plate fi lled with soapy water with a candle in the middle of the plate. Because fl eas are thermotropic, they jumped toward the candle and fell onto the plate, where they rapidly drowned in the soapy water. The fl eas were identifi ed by morphologic features and stored in 90% ethanol until DNA extraction.
To confi rm the phenotypic identifi cation, we designed primers and probes for quantitative real-time PCR (qPCR) that were specifi c for 2 species of fl ea (P. irritans and C. felis) based on the sequences of mitochondrial cytochrome oxidase gene published in GenBank (Table). All of the identifi cations made by morphologic appearance were confi rmed by qPCR because some specimens were damaged and diffi cult to identify. We found that most (52/55) of the fl eas collected in human dwellings were P. irritans, and 3 specimens were C. felis. A screening by amplifi cation using primers and probes specifi c for the 16S-23S internal transcribed spacer of Bartonella spp. (7) produced no positive results.
We screened rickettsial DNA by using qPCR with a Rickettsia-specifi c, gltA gene-based RKND03 system (8) and a bioB-based qPCR system specifi c for R. felis. We found that the 3 specimens of C. felis fl eas contained the DNA of R. felis; however, 23 (43%) of 53 P. irritans specimens also contained DNA of R. felis. We amplifi ed and sequenced nearly the entire rickettsial gltA gene from 3 C. felis and 10 P. irritans specimens and found that the sequence was identical to that of R. felis (GenBank accession no. NC_007111).
During the fi eld collection of the fl eas, the conservation of specimens may be diffi cult. Degradation of specimens may pose a problem for the ensuing morphologic identifi cation. For fl eas, a specifi c preparation is required that destroys internal organs and produces a chitin complex of the insect. This type of preparation makes it diffi cult, and sometimes impossible, to use the insect later for molecular studies. The development of qPCR specifi c for P. irritans and C. felis fl eas facilitated the identifi cation of damaged samples and also precluded the laborious and time-consuming procedure of identifi cation by morphologic features.
We conclude that the reservoirs of R. felis in Ethiopia include both C. felis and P. irritans fl eas. In Ethiopia, P. irritans fl eas have been reported to be prevalent (9). P. irritans fl eas have been shown to be infected by R. felis in several locations, notably in the Democratic Republic of the