Global spread of antibiotic resistance: the example of New Delhi metallo-β-lactamase (NDM)-mediated carbapenem resistance

Discovery of NDM

In the winter of 2007, a 59-year-old male patient of Indian descent who had lived in Sweden for many years travelled to India where he was hospitalized, initially in the Punjab, but then in New Delhi, for the management of a gluteal abscess. In January 2008 he was repatriated to a hospital in Orebro, Sweden, where, on the day after admission, a urine culture yielded an isolate of Klebsiella pneumoniae that was resistant to multiple antibiotics including carbapenems (ertapenem, imipenem and meropenem). This strain was not isolated from any subsequent cultures, but stool samples tested following transfer of the patient to a nursing home in March 2008 yielded a carbapenem-resistant strain of Escherichia coli. Phenotypic testing of both isolates suggested that the carbapenem resistance was due to the production of a metallo-β-lactamase (MBL), but PCR analysis failed to detect known MBL genes. Cloning and sequencing studies subsequently indicated that the resistance was due to a novel type of enzyme, which shared very little identity with other known MBLs, the most closely related being VIM-1/2, with which it shared only 32 % identity. The novel MBL was designated NDM-1, as the authors of the report believed the resistance originated from India (Yong et al., 2009). Occurrence of the same novel resistance gene in two different genera suggested that it was transferable, and conjugation experiments coupled with molecular studies confirmed that the bla_NDM-1 gene was located on transferable plasmids of 180 and 140 kb in the K. pneumoniae and Escherichia coli isolates, respectively.

A variant of NDM-1 (designated NDM-2) which differed by a single amino acid was reported in 2011 (Kaase et al., 2011), and subsequently, a series of further variants (designated NDM-3–NDM-7) have been reported on the Lahey Clinic β-lactamase website ().

Epidemiological link of NDM with the Indian subcontinent

The putative epidemiological link between NDM-1 and the Indian subcontinent was further strengthened by a subsequent study which documented the isolation of NDM-1-positive Enterobacteriaceae from patients in India, Pakistan, Bangladesh and the UK in 2008–2009 (Kumarasamy et al., 2010). NDM-positive Enterobacteriaceae were found to be geographically widespread in the Indian subcontinent, being recovered from ten areas in India, eight areas in Pakistan and one area of Bangladesh. Meanwhile, in the UK, the national reference laboratory of the Health Protection Agency had been independently investigating a growing number of unusual carbapenem-resistant isolates from UK patients. These isolates of Enterobacteriaceae displayed MBL phenotypes but, like the two ‘Swedish’ isolates, were negative for known carbapenemase genes. These had been sampled from patients in many UK hospitals, with the first received in August 2008. Cloning and DNA sequencing identified a novel MBL gene, which was subsequently found to be identical to bla_NDM-1. Of particular interest was the finding that at least 17 of the first 29 UK patients with NDM-positive bacteria (including isolates of Escherichia coli, K. pneumoniae, Enterobacter spp., Citrobacter freundii, Morganella morganii and Providencia spp.) had a history of travel to India or Pakistan within the previous year, with 14 having been hospitalized for a range of indications. Isolates positive for NDM-1 continued to be identified in the UK, and by May 2011, more than 100 such isolates had been received by the reference laboratory, with many patients from whom isolates had been obtained still having epidemiological links to India or Pakistan (Nordmann et al., 2011).

The discovery of the likely importation of NDM-producing Enterobacteriaceae into the UK resulted in the release of a National Resistance Alert by the Department of Health in England, which highlighted the potential threat to public health and the need to isolate and screen patients with a history of travel to, and particularly hospitalization in, the Indian subcontinent. By way of contrast, the official response in India was to play down the extent of the problem, with some claiming that the study and the name of the enzyme were malicious propaganda aimed at undermining the subcontinent’s medical tourism industry (Palmer, 2010; Walsh & Toleman, 2011, 2012). Despite this reaction, data have continued to accumulate, a fact which clearly indicates a substantial problem with NDM-positive bacteria in the Indian subcontinent. An investigation into the occurrence and characterization of carbapenem-resistant Enterobacteriaceae isolated in Indian hospitals in 2006–2007 recovered NDM-1-positive isolates from hospitals in New Delhi, Mumbai and Pune (Castanheira et al., 2011), with these isolates pre-dating the hitherto first reported case of NDM-1 infection (Yong et al., 2009). Subsequently, there were reports of NDM-1-positive Acinetobacter spp. and Pseudomonas spp. in a hospital in Pune in 2010 (Bharadwaj et al., 2012), with NDM-1-positive Acinetobacter spp. also being found the same year in a hospital in Chennai (Karthikeyan et al., 2010). Another study, also undertaken in 2010 at a tertiary referral hospital in Varanasi in north India, found that 54 (6.9 %) of 780 consecutive, non-duplicate clinical isolates of Enterobacteriaceae (comprising 30 Escherichia coli, 12 K. pneumoniae and 12 Citrobacter species) were positive for the bla_NDM-1 gene (Seema et al., 2011). NDM-positive Enterobacteriaceae have also been seen in the neonatal setting in Indian hospitals, with two cases of neonatal sepsis due to K. pneumoniae (Roy et al., 2011b) and a cluster of bloodstream infections due to Escherichia coli in a neonatal unit being reported (Roy et al., 2011a). International surveillance of intra-abdominal infections in 2009 (comprising centres in Europe, North America, Latin America, the South Pacific, the Middle East and Asia) undertaken as part of the Study for Monitoring Antimicrobial Resistance Trends programme found NDM-1-positive isolates only in India. As in the other studies, the bla_NDM-1 gene was found in a range of species including Escherichia coli, K. pneumoniae, Enterobacter cloacae, Providencia rettgeri and M. morganii (Lascols et al., 2011).

While the above reports provide evidence of the occurrence of NDM-1-positive bacteria in Indian hospitals, a finding of arguably greater public health importance was provided from an environmental study carried out in New Delhi in late 2010. This study showed the presence (by direct PCR) of the bla_NDM-1 gene in 51 of 171 seepage samples (water pools in streets or rivulets) and in two of 50 samples of drinking water. The two positive drinking-water samples and 12 of the 171 seepage samples yielded growth of a range of bla_NDM-1-positive bacteria including Escherichia coli, K. pneumoniae, C. freundii, Shigella boydii, Vibrio cholerae and Aeromonas caviae (Walsh et al., 2011). This clearly showed for the first time that the problem of NDM-1 was not confined to hospital strains of bacteria, but was widespread in the community environment in India, highlighting the need for improvements in sanitary conditions as a key public health intervention. Interestingly, a recent report from Vietnam (described by the authors as a country with strong cultural and economic links with India) also documented environmental contamination with NDM, with two water samples from the Kim Nguu river, which flows through the centre of Hanoi, giving positive PCR results for bla_NDM-1 (Isozumi et al., 2012). Both PCR-positive samples, which were obtained from river sites 3 km apart, yielded growth of NDM-positive K. pneumoniae of ST283, indicating a likely high level of contamination of the river with this carbapenem-resistant opportunist pathogen. It is noteworthy that a history of travel to Vietnam (but not involving hospitalization) was noted in one of five patients affected during an outbreak of carbapenem-resistant NDM-1-producing Enterobacteriaceae reported from Canada, suggesting yet again inter-continental transmission of this resistance determinant (Borgia et al., 2012).

Although a study in Mumbai failed to detect intestinal carriage of NDM-1-positive Enterobacteriaceae (Deshpande et al., 2012), gut colonization was reported from Bangladesh and Pakistan (Islam et al., 2012; Perry et al., 2011). In the Bangladesh study, screening of consecutive clinical samples over a 1-month period in late 2010 yielded 403 Gram-negative isolates, of which 14 (3.5 %) were positive for NDM-1. The study in Pakistan comprised an investigation of the prevalence of faecal carriage of Enterobacteriaceae with NDM-1 at two military hospitals in Rawalpindi. In total, 64 NDM-1-positive isolates of Enterobacteriaceae, belonging to seven species, were recovered from 37 (18.5 %) of the stool samples taken from 200 patients. In terms of different patient populations, the rates of intestinal carriage in inpatients and outpatients were 27 % and 14 %, respectively (Perry et al., 2011).

In addition to the widespread occurrence of NDM-1 in the Indian subcontinent, reports have continued to be published from many parts of the world, documenting isolation of NDM-1-positive bacteria from patients with epidemiological links to that part of the world. Such reports have emanated from geographically diverse regions of the globe including Australasia, the Far East, the USA, Canada, the Middle East and many countries in Europe (Fig. 1). While many of the patients had a history of hospitalization in India, Pakistan or Bangladesh, others had simply travelled in this region (Table 1), which may indicate community acquisition of NDM-positive bacteria through ingestion of contaminated water, with resulting gut carriage.

International transmission of NDM-positive bacteria from regions other than the Indian subcontinent

Although much work on NDM has been focussed on the Indian subcontinent, there are now many documented cases of international transmission involving movement of infected or colonized individuals from countries in other regions of the world. In particular, the Balkans has been highlighted as a possible secondary reservoir for the spread of NDM, based on the considerable numbers of reports of patients from whom NDM-positive bacteria have been isolated following medical repatriation from this geographical area (Table 2). Transmission of NDM between Balkan states has also been documented (Mazzariol et al., 2012). Routine analysis of carbapenemase-producing Gram-negative bacteria isolated in the Belgrade Military Medical Academy in 2010 identified seven isolates of Pseudomonas aeruginosa that were positive for bla_NDM-1 (Jovcic et al., 2011). Interestingly, none of the patients had a history of travel to the Indian subcontinent or to Europe, raising the possibility that such NDM-positive strains may be endemic in Serbia. However, other investigators commenting on the possible epidemiological picture of NDM in the Balkans (Livermore et al., 2011) noted a published report that stated patients from the Balkans travelled to Pakistan for commercial kidney transplants and that infections in this patient group were not uncommon (Ivanovski et al., 2011), leading these authors to speculate that such medical tourism could have introduced NDM to the Balkans. This issue remains the subject of contention, but what can undoubtedly be said for the present is that, irrespective of their origin, NDM-positive bacteria pose a significant public health threat in both the Indian subcontinent and the Balkans, with such strains being onwardly disseminated to diverse geographical regions around the globe.

The same consideration applies to the Middle East and North or Central Africa where NDM-positive bacteria have been reported from a range of countries including Afghanistan, Algeria, Cameroon, Egypt, Iraq, Israel, Kuwait, Lebanon, Morocco, the Sultanate of Oman and the United Arab Emirates (Fig. 1). In most cases, the literature comprises reports of patients being transferred from the Middle East or North or Central Africa to other parts of the world (Table 2). However, the converse is also known to have occurred, with importation of NDM-1-positive strains of K. pneumoniae from the Indian subcontinent into Kuwait and Oman (Dortet et al., 2012b; Jamal et al., 2012; Poirel et al., 2011a). The complex epidemiological picture that can be seen with the inter-country transfer of resistant organisms is exemplified by a recent report that documented two patients with NDM-1-positive organisms associated with Reunion Island. In the first case, in late 2011, a patient who was transferred to a hospital in France was found upon rectal screening to be colonized with NDM-1-positive K. pneumoniae. Three months later, a patient transferred from a hospital in India to the same unit in Reunion Island yielded growth of NDM-1-producing Salmonella enterica subsp. enterica serotype Westhampton from a urine culture (Cabanes et al., 2012).

Local spread of NDM following importation

Although there have been many reports of inter-country transmission of NDM-positive bacteria related to medical repatriation of hospitalized patients or patients returning home after a period of foreign travel, it is striking and fortunate that, with just a few exceptions (Hrabák et al., 2012; Kumarasamy et al., 2010; Poirel et al., 2011b), most do not mention subsequent cross-infection. However, despite the paucity of documented instances of spread following importation of NDM-positive bacteria, it seems likely that local dissemination of these organisms in different countries has occurred, at least as gut colonization. Several lines of evidence support this. Firstly, reports from disparate parts of the world, including Canada (Kus et al., 2011), China (Fu et al., 2012; Ho et al., 2012; Yang et al., 2012), France (Arpin et al., 2012), Guatemala (Pasteran et al., 2012), Israel (Espinal et al., 2011), Oman (Poirel et al., 2011a), Kenya (Poirel et al., 2011f), Kuwait (Jamal et al., 2012), South Africa (Brink et al., 2012), South Korea (Kim et al., 2012) and Thailand (Rimrang et al., 2012) have described the isolation of NDM-positive bacteria from patients with no history of foreign travel, implying that the organisms must have been acquired locally. In one study, isolation of NDM-1-producing Acinetobacter pittii was reported in 27 patients in an intensive care unit in China over a period of 13 months (June 2008–June 2009), none of whom had epidemiological links to South-West Asia (although links to the Balkans or other regions were not mentioned) (Yang et al., 2012). In another report, this time from France, two patients who denied any foreign travel in the previous 5 years, and who shared the same hospital room, were both colonized in the gut with the same strain of NDM-1-positive Escherichia coli, which was also isolated from the urine of one of the patients (Denis et al., 2012). Interestingly, both faecal and urine specimens from this patient remained positive when the patient was followed up 7 months later, indicating the potential for NDM-positive bacteria to persist at sites of colonization for prolonged periods of time. This has been confirmed in other studies which described gut carriage of NDM-positive Escherichia coli for periods of 13 (Poirel et al., 2011e) and 10 months (D’Andrea et al., 2011), while another report documented carriage of NDM-1-positive K. pneumoniae for more than 7 months (Kim et al., 2012). Secondly, most NDM-positive bacteria reported to date have been isolated from patients who were clinically ill and consequently subjected to microbiological investigation following admission to hospital. Clearly, travellers to high-risk areas who become asymptomatically colonized with NDM-positive bacteria would not be subjected to such investigations and may act as undetected reservoirs of carbapenem-resistant bacteria on returning home. The lack of surveillance data on rates of asymptomatic gut carriage of NDM-positive bacteria, particularly in community settings in different countries, means that our current views of the extent of the spread of NDM may well be an underestimate.

The contribution of clonal expansion and gene transfer to the spread of NDM

While the epidemiology of many infectious diseases can be described solely in terms of the spread of the causative pathogens, the epidemiology of antibiotic resistance is significantly more complex in many bacteria, not least in the Enterobacteriaceae. Dissemination of many types of resistance involves not just the spread of the resistant organisms, but also the inter-strain, inter-species or even inter-genus spread of the resistance genes. Gene spread among bacteria can be mediated by a range of genetic mechanisms including transformation, transduction and conjugative plasmid transfer (Sykes, 2010), although observations to date only implicate plasmid transfer in the spread of bla_NDM genes.

Some insight into the relative roles of strain spread versus plasmid spread in India was provided by Kumarasamy et al. (2010) in their paper on NDM from India, Pakistan and the UK. These workers found that isolates of NDM-positive K. pneumoniae from Haryana in northern India were clonal and contained plasmids that were non-conjugative, while isolates from Chennai in South India and those from the UK were clonally diverse and contained plasmids that were readily transferable. It was noteworthy, however, that among 21 isolates of K. pneumoniae from the UK, there were two pairs of related isolates (designated on the basis of their PFGE profiles) that were from epidemiology-linked patients, and hence, thought likely to represent cases of cross-infection (Kumarasamy et al., 2010).

Molecular investigations involving both the characterization of isolates of NDM-positive bacteria and the characterization of the plasmids containing bla_NDM genes show a highly complex picture. Firstly, bla_NDM has been found both in a wide range of species and genera of Gram-negative bacteria, and in a diverse range of clones and strains within individual species, as indicated by the variation in multi-locus sequence types (STs) and PFGE profiles, respectively (Table 3). For example, bla_NDM has been reported in at least 11 different STs of both Escherichia coli and K. pneumoniae to date, indicating a high level of inter-lineage and inter-species gene transfer. The plasmids encoding NDM also appear highly heterogeneous on the basis of molecular size, incompatibility type and linked antibiotic resistance genes (Table 3). While bla_NDM has commonly been found on plasmids in Enterobacteriaceae, it is notable that there has only been one report of plasmid-mediated NDM in Acinetobacter baumannii (Chen et al., 2011), although diverse plasmids encoding NDM have been found in other species of Acinetobacter (Fu et al., 2012; Yang et al., 2012). The former study reported four different strains of Acinetobacter baumannii containing plasmids of different sizes (30–50 kb) encoding NDM, and although transferable to Escherichia coli in vitro, the plasmids appeared unstable and were readily lost after subculture in antibiotic-free medium. In all other reported isolates of Acinetobacter baumannii, the bla_NDM gene was located on the chromosome (Bogaerts et al., 2012; Bonnin et al., 2012b; Boulanger et al., 2012; Espinal et al., 2011; Hrabák et al., 2012; Kaase et al., 2011; Karthikeyan et al., 2010; Pfeifer et al., 2011a; Poirel et al., 2012a). Nonetheless, there is still evidence of gene spread in Acinetobacter baumannii, as several studies have found the bla_NDM gene located between two direct repeats of the ISAba125 element, thus forming a composite transposon (Tn125) (Bogaerts et al., 2012; Boulanger et al., 2012; Espinal et al., 2011; Hrabák et al., 2012; Pfeifer et al., 2011a; Poirel et al., 2012a). Further investigation of the immediate genetic environment of the bla_NDM gene in isolates of Acinetobacter baumannii and Enterobacteriaceae revealed the presence of a novel bleomycin resistance gene designated ble_MBL (ble gene associated with the metallo-B-lactamase gene NDM) (Dortet et al., 2012a). The ble_MBL and bla_NDM genes were co-expressed, being under the control of the same promoter located upstream of bla_NDM at the extremity of ISAba125. Bleomycin refers to a family of structurally related glycopeptides produced by Streptomyces verticillus that have antibacterial properties but which are also used in cancer chemotherapy. Bleomycin(s) may also be found in the environment. It was therefore postulated that selective pressure promoting the spread of NDM-positive isolates might be due not only to use of β-lactam or other antibiotics to which NDM-positive isolates might be resistant but also to the use of anti-cancer drugs and to naturally occurring bleomycin molecules in the environment (e.g. water seepage samples) (Dortet et al., 2012a). In terms of the initial emergence of NDM in human pathogens, it has been hypothesized that the bla_NDM–ble_MBL pairing may have been integrated first into the chromosome of Acinetobacter baumannii from an unknown environmental species, where it became associated with ISAba125, and then was transposed onto plasmids capable of replication and conjugative transfer in Enterobacteriaceae, with the downstream copy and most of the upstream copy of ISAba125 subsequently being lost from some isolates (Nordmann et al., 2012b). In this regard, it is noteworthy that plasmid-encoded bla_NDM-1 and ble_MBL have also recently been reported in isolates of Acinetobacter pittii in China, with the bla_NDM-1 gene in this instance being flanked by two insertion sequences, namely ISAba125 and ISAba11, the latter having 99 % identity with an insertion sequence found in Acinetobacter baumannii ATCC 17978 (Yang et al., 2012).

Prospects for the future spread of NDM

The rapidity with which a new type of resistance can emerge in bacteria able to cause infections in humans and disseminate to become a global public health threat is clearly exemplified by NDM carbapenemase. The earliest known NDM-positive organism (an Escherichia coli strain isolated in New Delhi) was collected in 2006 (Castanheira et al., 2011), since then, NDM-positive isolates have been reported from 40 countries covering all continents except South America and Antarctica. What is notable about the global transmission of NDM is that although this has involved both strain spread and gene spread, so far, the latter appears to have been the dominant mechanism of dissemination. However, it is possible that the epidemiology of NDM may change, as the bla_NDM-1 gene has been found in bacterial strains belonging to lineages with known epidemic or pandemic potential. These include, for example, Escherichia coli of ST101 (Mushtaq et al., 2011; Nielsen et al., 2012; Peirano et al., 2011a; Poirel et al., 2010b; Williamson et al., 2012), which has spread widely in Spain, and ST131(Rogers et al., 2011) which has spread globally, with both lineages being associated with the spread of cephalosporin resistance mediated by CTX-M-type extended-spectrum β-lactamases. Moreover, it is noteworthy that another type of carbapenemase, designated KPC (for Klebsiella pneumoniae carbapenemase), has spread widely due predominantly to dissemination of a particular clone of K. pneumoniae of ST258 (Woodford et al., 2011). Hence, it is not unreasonable to be concerned that NDM may increasingly adopt a similar mode of transmission. The epidemiology of strain spread and the patient populations affected may vary, however, depending on the species and strains in which the bla_NDM gene is found. For example, NDM-positive Acinetobacter baumannii may be more likely to infect hospitalized patients, particularly those in high-dependency units, as these are typically the patient groups at greatest risk of infection or colonization with acinetobacters. In contrast, NDM-positive Escherichia coli strains, particularly those causing gut colonization, may be a cause of lower urinary tract infections, in either the community or hospital setting. Clearly, ongoing surveillance will be critical in monitoring future trends in the spread of NDM. It may be the case however, that surveillance will need to be expanded from monitoring infection and colonization in humans to encompass animals, as a recent report has documented isolation of NDM-positive Escherichia coli from companion animals in the USA (Shaheen et al., 2012).

Concluding remarks

It is now evident that globalization plays a major role in the rapid dissemination of antibiotic resistance (van der Bij & Pitout, 2012), with the spread of NDM providing just one example of how antibiotic resistance can rapidly disseminate internationally. The increasing recognition of the global extent of the problem posed by resistant pathogens has been reflected in a number of reports from bodies such as the World Health Organization (WHO, 2012) and the European Centre for Disease Prevention and Control (ECDC, 2009) and also by international initiatives such as the formation of a Transatlantic Taskforce on Antimicrobial Resistance between the USA and the European Union (TATFAR, 2012). The problem is all the more pressing, particularly for resistance in Gram-negative bacteria, due to the paucity of new antibiotics in the development pipeline (Livermore, 2011; Wise et al., 2011). As the clinical and public health threat posed by antibiotic resistance clearly now has an international dimension, activities to monitor and control the problem need to be international in scope. Although surveillance activities such as the pan-European surveillance of antimicrobial resistance undertaken by the European Antimicrobial Resistance Surveillance Network are already yielding valuable insight into the epidemiology of resistance in a range of pathogens (ECDC, 2010), routine surveillance and an ability to undertake reliable susceptibility testing are still lacking in many parts of the world, particularly those resource-poor regions which have inadequate infrastructure due to poverty and other factors such as political unrest. Overcoming these difficulties poses a major challenge, and there can be no escaping the fact that international cooperation will be critical in attempts to control the global threat to public health posed by antibiotic resistance.