Spread of NDM-1-Producing Enterobacteriaceae in a Neonatal Intensive Care Unit in Istanbul, Turkey

TEXT

Carbapenem-hydrolyzing β-lactamases belonging to Ambler classes A, B, and D have been reported worldwide among Enterobacteriaceae (1, 2). In Turkey, the wide dissemination of OXA-48-producing isolates (from Klebsiella pneumoniae, Escherichia coli, Citrobacter freundii, and Enterobacter cloacae) has been demonstrated, defining this country as being endemic for OXA-48 (3–5). Apart from OXA-48 producers, isolates producing other types of carbapenemases (NDM-1, IMP-1, and KPC-2) were recently identified in Turkey as well. The non-OXA-48-producing carbapenemase producers reported from Turkey are the following: six NDM-1-producing K. pneumoniae isolates (6, 7), four IMP-1-producing K. pneumoniae isolates (7, 8), and a single KPC-2-producing K. pneumoniae isolate (9).

The spread of NDM-1-producing Enterobacteriaceae is now considered to be almost global (10). However, only a few outbreaks of NDM-1 producers have been reported. So far, most of the reports from many countries have been scattered and corresponded mainly to importation cases (1).

Our study was designed to evaluate retrospectively the occurrence of carbapenemase-producing Enterobacteriaceae in a university hospital in Istanbul, Turkey. An increased number of carbapenem-resistant isolates recovered from patients hospitalized in the neonatal intensive care unit (NICU) during a short period prompted us to perform such a study.

All enterobacterial isolates presenting with reduced susceptibility to imipenem (MIC ≥ 0.5 μg/ml) were collected between 1 January and 1 May 2013 (a 4-month period), consisting of a total of 22 isolates. The MICs of the carbapenems were determined by the Etest (AB bioMérieux, La Balme-les-Grottes, France) on Mueller-Hinton agar plates at 37°C, and the results of susceptibility testing were interpreted according to CLSI guidelines (11). Those isolates were recovered mostly from urine samples, and also from rectal swabs (Table). All isolates were considered to be colonizers. These were K. pneumoniae (n = 12), E. cloacae (n = 8), and E. coli (n = 2) (Table 1). The isolates were resistant to all β-lactams, including carbapenems for most isolates (Table 2). However, some isolates were still categorized in the susceptible range for carbapenems (Table 2). In addition, they were mostly resistant to aminoglycosides, fluoroquinolones, chloramphenicol, sulfonamides, fosfomycin, and nitrofurantoin. All except two of them had low MIC values (<0.5 μg/ml) for colistin, as measured by the Etest.

Carbapenemase detection was performed using the Carba NP test (12), and positive results were obtained for the 22 isolates. Thus, PCR assays were carried out with a series of primers designed to detect Ambler class A, B, and D carbapenemase genes, i.e., bla_KPC, bla_IMP, bla_VIM, bla_NDM, and bla_OXA-48 (13), followed by sequencing of the PCR amplicons. Two isolates were positive for bla_KPC-2, 12 for bla_NDM-1 gene, and eight for bla_OXA-48 (Table 1). None of the isolates coharbored two carbapenemase genes. Note that the two KPC-2-producing isolates were K. pneumoniae, the 12 NDM-1-producing isolates were K. pneumoniae (n = 4) and E. cloacae (n = 8), and the eight OXA-48-producing isolates were K. pneumoniae (n = 6) and E. coli (n = 2) (Table 1).

Genotyping was performed to evaluate the clonal relationship of the K. pneumoniae and E. cloacae isolates by pulsed-field gel electrophoresis (PFGE) (14), and K. pneumoniae isolates were additionally genotyped by multilocus sequence typing (MLST), as described previously (15). The two KPC-2-producing K. pneumoniae isolates were clonally related and belonged to the sequence type 307 (ST307). The four NDM-1-producing K. pneumoniae isolates were clonally unrelated, belonging to ST15, ST45, ST278, and ST1059. The eight NDM-1-producing E. cloacae isolates were clonally undistinguishable by PFGE, thus corresponding to a single clone. Finally, the six OXA-48-positive K. pneumoniae isolates were all clonally related and belonged to ST101 (Table 1).

Interestingly, looking at the hospitalization wards from which the patients originated, it appeared that a cluster of colonizing NDM-1-producing E. cloacae isolates occurred in the NICU, with a total of eight neonates harboring this same resistant strain. Similarly, the two patients colonized with the KPC-2-producing K. pneumoniae isolates had been hospitalized in the same ward (plastic surgery). Regarding the dissemination of the ST101 and OXA-48-producing K. pneumoniae isolates (five patients colonized in four distinct wards), a nosocomial route of dissemination might be also suspected. However, this hypothesis is not obvious considering that OXA-48-producing K. pneumoniae are very widespread in Turkey, and taking into account the heterogeneity observed in terms of antibiotic resistance for those isolates, the MICs of carbapenems, and coresistance markers (Table 2). Interestingly, ST101 and OXA-48-producing K. pneumoniae isolates have been reported in Libya (16) and were at the origin of an outbreak in Spain (17). In contrast, and even if OXA-48-producing K. pneumoniae isolates of many different STs have been identified in Turkey (18), that specific clone was never identified in that country. Considering the recent and frequent transfers of Libyan patients to Turkey, and in particular in the hospital where those samples have been recovered, this might explain the emergence of that specific clone.

Note that the majority of the NDM-1-positive isolates, regardless of their clonal lineage and of the species to which they belonged, were multidrug resistant. In particular, most of them coexpressed the extended-spectrum β-lactamase CTX-M-15 and the narrow-spectrum β-lactamase OXA-1, as identified by PCR and sequencing (19) (Table 1). In addition, since all NDM-1-positive isolates were resistant at a high level to all aminoglycosides, a search of 16S rRNA methylase-encoding genes was performed as described previously (20). The results showed that all those isolates (eight E. cloacae and the four K. pneumoniae isolates) were positive for the rmtC gene (21).

In order to evaluate whether the cooccurrence of the bla_NDM-1 and the rmtC genes might be related to the spread of a specific plasmid, mating-out assays were performed using all those isolates as donors and with E. coli J53 as the recipient, as described previously (22). Interestingly, E. coli transconjugants coproducing NDM-1 and RmtC were obtained for all the isolates, and a single plasmid of ca. 150 kb was identified in all transconjugants. That plasmid additionally conferred resistance to rifampin to all transconjugants. However, it did not harbor the bla_CTX-M-15 gene, in accordance with the susceptibility to aztreonam (a substrate spared by NDM-1) (23) observed for all E. coli transconjugants. Attempts to type this plasmid using the PCR-based replicon typing (24) remained unsuccessful. PCR mapping was performed to identify the genetic sequences surrounding the bla_NDM-1 gene in all positive isolates. The same structure was found, with the bla_NDM-1 gene preceded by a truncated version of ISAba125, followed by the ble_MBL gene encoding resistance to bleomycin (25). Similar structures were identified previously in different species from different countries (19).

Our study reports an outbreak of NDM-1-producing E. cloacae in a NICU setting and describes the emergence of plasmid-mediated 16S rRNA methylases in Turkey. Here, the RmtC enzyme was involved and was encoded by a gene located on the same plasmid carrying the bla_NDM-1 gene. Such an association between those two genes was previously reported in a single K. pneumoniae isolate from India (19) and in a series of K. pneumoniae isolates from Kenya (26). A 150-kb and untypeable plasmid coharboring the two genes was similarly identified here and in the Indian isolate. The wide spread of multidrug-resistant NDM-1 producers in a NICU setting is extremely worrisome, since infections developed by those immunocompromised patients are very difficult to treat. While this work was in progress, an outbreak of NDM-1-producing K. pneumoniae was also reported in a neonatal unit located in Bogotá, Colombia (27). In that study, two out of the six infected newborns died. The strain involved corresponded to ST1043, which is unrelated to the STs identified in our study.

In many countries, the occurrence of NDM-1 producers in hospitalized patients is mostly related to previous patient hospitalization on the Indian subcontinent. Nevertheless, the first case of NDM-1-producing K. pneumoniae in Turkey was related to a transfer from Iraq (8). While this work was in progress, another study showed the emergence of NDM-1-producing K. pneumoniae isolates in a Turkish hospital located in Kayseri, the center of Turkey, which is 800 km from Istanbul (7), further emphasizing the actual NDM-1 producer problem in Turkey and therefore identifying a possible hidden reservoir for those multidrug-resistant isolates. In that study, as in ours, the origin of the NDM-1-producing isolates is unknown. Altogether, those studies indicate that the spread of NDM-1 producers in the Middle East may be more important than was suspected and is now likely out of control.

This clinical experience was interesting since an outbreak was suspected as soon as two similar carbapenem-resistant E. cloacae isolates were recovered from two different patients in the NICU, which is a 30-bed unit. Nasal and rectal swabs were obtained from all 30 patients, and 10 more swabs were obtained from the incubators and environment (a total of 70 surveillance cultures). None of the swabs from the incubators and environment grew carbapenem-resistant Enterobacteriaceae (CRE). In order to detect the flaws in work practices causing this colonization, a hospital video surveillance system was used, and significant compromises in the infection control procedures were identified, especially during night shifts. A detailed and rigorous infection control training program was therefore performed, in which all staff participated. Contact isolation precautions were strictly implemented with all the patients who were positive for CRE, and after 30 to 45 days, there were no patients in the NICU who were infected or colonized with any CRE organism, and there has not been a single case since then (8 months). This further shows that complementary works between the clinical and microbiological units may contribute significantly to preventing the spread of CRE.