John D. Perry, Department of Microbiology, Freeman Hospital, Freeman Road, High Heaton, Newcastle upon Tyne NE7 7DN, UK. E-mail: firstname.lastname@example.org
To evaluate two chromogenic media, Brilliance CRE and chromID CARBA, with stool samples referred to the Public Health Laboratories Division of the National Institute of Health in Islamabad, and assess the prevalence of carbapenemase-producing Enterobacteriaceae (CPE) in this population.
Methods and Results
One hundred and fifty-two stool samples from patients with diarrhoea were referred to the Microbiology Department and were investigated for the presence of CPE using two chromogenic culture media, Brilliance CRE and chromID CARBA. Thirteen patients (8·6%) were found to be colonized with CPE and all produced NDM-1 carbapenemase. Twelve of these patients (92%) were found to be colonized by culture on chromID CARBA compared with seven (54%) using Brilliance CRE.
If only coloured colonies were considered as presumptive CPE, the sensitivity, specificity and positive predictive value were 54, 23 and 6% for Brilliance CRE and 85, 85 and 36% for chromID CARBA, respectively.
Significance and Impact of the Study
We conclude that Enterobacteriaceae that produce NDM-1 carbapenemase can be found in patients from all major provinces of Pakistan and that chromID CARBA was the most effective of the two chromogenic media in this setting.
The increasing incidence of infections caused by carbapenemase-producing Enterobacteriaceae (CPE) is a major concern. Of particular concern to some is the increasing occurrence of bacteria harbouring NDM-1 carbapenemase, due to the genetic plasticity of the plasmid encoding it (Walsh 2010). Accumulating evidence indicates that bacteria harbouring NDM-1 are endemic in the environment and the population of the Indian subcontinent (Walsh et al. 2011; Walsh and Toleman 2012). Validated methods are urgently needed to screen for carriers of CPE, so that an accurate prevalence can be established and high-risk patients can be screened on admission to hospital. Chromogenic culture media are widely used as convenient tools for screening for other antimicrobial-resistant pathogens (Paniagua et al. 2010; Stamper et al. 2010; Harbarth et al. 2011). Such media are now available for isolation of CPE but there is a need for field trials to validate their performance. The aims of this study were to evaluate two chromogenic media, Brilliance CRE and chromID CARBA, with stool samples referred to the Public Health Laboratories Division of the National Institute of Health in Islamabad, which receives specimens from all regions of Pakistan, and assess the prevalence of CPE in this population.
Material and methods
Culture of stool samples
Between 18 August and 30 September 2011, 152 stool samples from distinct patients were referred to the Microbiology Department, Public Health Laboratories Division, National Institute of Health, Islamabad, Pakistan. The samples were from patients with diarrhoea and were referred by clinicians and/or public health authorities in diverse geographical regions of Pakistan. The age/sex of each patient was recorded as well as their location. Approximately 0·5 ml or 0·5 g of stool sample was emulsified in 0·5 ml of sterile 0·85% saline and 10 μl of the resulting suspension was inoculated onto Brilliance CRE (Oxoid, Basingstoke, UK) and chromID CARBA (bioMérieux, Craponne, France). The inoculum was then spread to obtain isolated colonies. Plates were incubated for 18–20 h at 37°C, and each type of colony was enumerated and stored immediately in a tube containing nutrient agar for further studies.
Phenotypic detection of ß-lactamase genes
All distinct colony types (n =330) recovered on either of the chromogenic media were investigated for evidence of carbapenemase activity using the KPC and MBL Confirm ID kit (Bioconnections, Knypersley, UK). This involved disc susceptibility testing with meropenem alone and with various β-lactamase inhibitors (boronic acid, dipicolinic acid and cloxacillin). A disc containing temocillin (30 μg) was also tested as an indicator of OXA-48 carbapenemase (Glupczynski et al. 2012). Any isolates showing no zone of inhibition with temocillin were tested for OXA-48 using PCR as previously described (Monteiro et al. 2012). A modified Hodge/cloverleaf test was also performed on all isolates, as previously described (Anderson et al. 2007). All isolates recovered on either chromogenic medium were identified using MALDI-TOF mass spectrometry (Bruker, Coventry, UK), with complimentary biochemical tests where appropriate (e.g. to exclude Shigella spp.). All isolates confirmed as Enterobacteriaceae (n =163) were also tested for extended spectrum ß-lactamase (ESBL) and AmpC ß-lactamase using the ESBL/AmpC Screen kit (Bioconnections). For any isolates showing evidence of metallo-ß-lactamase activity, the ESBL and AmpC assays were performed using Mueller–Hinton agar plates supplemented with 500 mg l−1 dipicolinic acid as previously described (Perry et al. 2011).
Investigation of ß-lactamase genes and plasmid replicon typing
For any isolates of Enterobacteriaceae that showed phenotypic evidence of carbapenemase production, blaNDM was sought with primers NDM-For (5′-GGG CAG TCG CTT CCA ACG GT-3′) and NDM-Rev (5′-GTA GTG CTC AGT GTC GGC AT-3′), which amplify a 475-bp product (Perry et al. 2011). Cycling conditions were as follows: 95°C for 5 min; 30 cycles of 95°C for 30 s, 60°C for 40 s and 72°C for 50 s; 72°C for 6 min. All such isolates were also re-inoculated onto both chromogenic media (and nutrient agar) using inocula of 105 and 102 CFU/spot using a multipoint inoculator. PCR analysis for other β-lactamases commonly found in NDM-1 producers, that is, CTX-M-1 group, CMY-2 like and SHV was performed as previously described (Sidjabat et al. 2011). Isolates were also screened for the presence of 16S rRNA methylase genes as previously described (Doi and Arakawa 2007) and also for plasmid replicon types that have been previously reported amongst NDM-1 producers, IncA/C, FII and L/M (Carattoli et al. 2005).
Clonal analysis of NDM-1-producing Escherichia coli and Klebsiella pneumoniae
All E. coli (n =8) and Kl. pneumoniae (n =5) were further analysed for their clonal relationship using semi-automated rep-PCR DiversiLab (bioMérieux, Oakleigh, Australia). DNA preparation and PCR amplification and analysis were performed as described previously (Sidjabat et al. 2010).
Media stability studies and susceptibility testing
Forty-five isolates of non-carbapenemase-producing Enterobacteriaceae that were recovered on either of the media were randomly selected for re-inoculation onto freshly supplied batches of both chromogenic media at the Freeman Hospital laboratory (Newcastle upon Tyne, UK). All isolates were inoculated at 105 CFU/spot on a weekly basis (0–6 weeks) within the stated shelf-life of the media as previously described (Wilkinson et al. 2012). Between tests, both media were stored in the dark at 4°C. Inoculated plates were incubated at 37°C for 24 h and results were recorded.
For four carbapenems (imipenem, meropenem, ertapenem and doripenem), powders of known potency were obtained from their respective manufacturers to determine minimum inhibitory concentrations (MICs) by agar dilution. Each was incorporated into Mueller–Hinton agar plates to produce a concentration range of 64–0·125 mg l−1 for each agent. All isolates of CPE were inoculated at 104 CFU/spot using a multipoint inoculator and plates were incubated overnight at 37°C in air before interpretation of MICs. Antimicrobial-free plates were inoculated in parallel, and E. coli NCTC 10418 was inoculated on every batch of plates as a control. For convenience, the MIC of colistin was measured by Etest in accordance with manufacturer's instructions. Susceptibility testing of 13 other antimicrobials was performed using the EUCAST standardized disc susceptibility method with interpretation of all MIC and disc susceptibility results performed in accordance with EUCAST criteria (EUCAST 2012).
Differences between the efficiencies of the two chromogenic media for isolation of carbapenemase-producing Enterobacteriaceae were compared using McNemar's test with the continuity correction applied.
One hundred and fifty-two patients were sampled in total including 87 women and 65 men with an average age of 21 years (range, 2 months–85 years). A total of 13/152 patients (8·6%) were colonized with CPE. From these 13 patients, 16 isolates of CPE were recovered using a combination of both chromogenic media and all produced NDM-1 carbapenemase (Table 1). All Enterobacteriaceae that were shown to produce a metallo ß-lactamase in phenotypic assays were confirmed as harbouring the NDM-1 gene by PCR. There was no evidence of carbapenemase activity in any other isolates. Nine isolates showed high-level resistance to temocillin and these were tested for OXA-48 enzyme using PCR, but this was not detected. All nine produced ESBL and AmpC ß-lactamase.
Table 1. Isolates recovered from 152 stool samples on two chromogenic media
Metallo ß-lactamase detected by phenotypic methods.
Other non-Enterobacteriaceae (without carbapenemase)
Location data were available for 143 patients and these patients were in 38 distinct towns or districts scattered throughout Pakistan. NDM-1-producing Enterobacteriaceae were recovered from all major provinces including (number of positive patients/number of patients): Balochistan (1/17), Khyber Pukhtoonkhwa (2/50), Punjab (4/19) and Sindh (6/57). Scrutiny of the data suggested the likelihood of local clusters of patients colonized with NDM-1-producing Enterobacteriaceae, for example, such isolates were recovered from 4/7 patients who submitted samples from Bhakkar (a district of Punjab) and three of these patients harboured NDM-1-positive Escherichia coli.
All NDM-1-producing Enterobacteriaceae (n =16) were nonsusceptible to ertapenem and doripenem, two of 16 isolates showed susceptibility to meropenem (MIC = 2 mg l−1 for both isolates) and a single isolate was susceptible to imipenem (MIC = 0·25 mg l−1). All isolates were susceptible to colistin (MICs: 0·125–2 mg l−1) by Etest. Using disc susceptibility testing, all isolates were nonsusceptible to cefotaxime, cefepime, aztreonam, piperacillin/tazobactam, temocillin, ciprofloxacin and gentamicin. In vitro susceptibility was recorded for tigecycline (87%), mecillinam (81%), fosfomycin (69%), amikacin (56%), nitrofurantoin (44%) and cotrimoxazole (6%).
Molecular investigation of CPE
All NDM-1-producing Enterobacteriaceae (n =16) were found to coharbour genes encoding CTX-M 15 (see Fig. 1). Genes encoding SHV ß-lactamase were found mostly in Kl. pneumoniae. CMY ß-lactamases were found in six of 16 isolates. The 16S rRNA methylase genes ArmA and rmtC were found in five of 16 and one of 16 isolates, respectively. Four of six isolates coproducing CMY-2-like genes possessed the A/C plasmid replicon. FII and L/M replicon types were not detected amongst these NDM-1-producing Enterobacteriaceae. This A/C plasmid replicon type has been mostly reported to carry the NDM-1 gene amongst NDM-1-producing Enterobacteriaceae.
Clonal analysis of NDM-1-producing Escherichia coli and Klebsiella pneumoniae
The NDM-1-producing E. coli and Kl. pneumoniae from Pakistan were diverse (similarities <95%). The production of β-lactamases and examination of 16S rRNA were incorporated in this analysis (Fig. 1). Based on the comparison with a reference strain of ST131 E. coli using this analysis, none of these NDM-1-producing E. coli belonged to the pandemic E. coli ST131 clone.
Comparison of chromogenic media
Twelve of 13 positive patients (92%) with NDM-1-producing Enterobacteriaceae were detected using chromID CARBA and seven patients (54%) were detected using Brilliance CRE (P =0·29). All of the NDM-1-producing Enterobacteriaceae grew at high and low inocula when re-inoculated onto both chromogenic media except for one isolate of NDM-1-producing Kl. pneumoniae, which failed to grow on chromID CARBA. A large number of Enterobacteriaceae isolates (n =135) were recovered on Brilliance CRE that did not produce carbapenemase and showed full susceptibility to meropenem. Of these 135 isolates, 91% produced ESBL and 21% produced AmpC ß-lactamase. This contrasted with only seven isolates of carbapenemase-negative Enterobacteriaceae recovered on chromID CARBA (P < 0·0001) – all with ESBL and/or AmpC ß-lactamase. Isolates of non-Enterobacteriaceae were frequently recovered on both media. These were mainly nonfermentative Gram-negatives and many of them showed evidence of metallo-carbapenemase production in phenotypic assays (Table 1). When 45 of the isolates of carbapenemase-negative Enterobacteriaceae were randomly selected for re-inoculation onto fresh batches of both media, all were inhibited on chromID CARBA, but 20% grew on Brilliance CRE. After 6 weeks storage of media at 4°C in a UK laboratory, 9% of isolates were able to grow on chromID CARBA compared with 42% that were able to grow on Brilliance CRE.
Kumarasamy et al. (2010) were the first to provide evidence of widespread dissemination of NDM-1 in India and Pakistan in 2010. A more recent study reported that 13·8% of outpatients attending a military hospital in Rawalpindi had gut colonization with NDM-1-producing Enterobacteriaceae (Perry et al. 2011). These data, supported by other studies in India (Walsh et al. 2011), have been used to estimate a prevalence of 100 million carriers of bacteria harbouring NDM-1 in India alone (Walsh and Toleman 2012). The growing problem of antimicrobial resistance in Pakistan is a serious concern. The emergence of NDM-1 in Pakistan may be attributed to a number of factors, which are likely to include irrational use of antibiotics within public and private sector hospitals and clinics, lack of infection control and antimicrobial surveillance programs in the majority of hospitals, and limited diagnostic facilities for molecular characterization of highly resistant bacteria. The high prevalence of NDM-1 in the Indian subcontinent, including Pakistan, also poses a serious concern for spread of these highly resistant bacteria to other countries (Kumarasamy et al. 2010; Walsh and Toleman 2012). This study provides further evidence of the frequency of gut colonization with NDM-1-producing Enterobacteriaceae in Pakistan. A strength of the study described here is that by using samples referred to a central public health laboratory, we were able to acquire samples sent from all areas of Pakistan. However, a limitation of this study is that samples were exclusively from patients with diarrhoea who are likely to have an altered gut flora, which may not be representative of the general population.
ChromID CARBA showed a better sensitivity than Brilliance CRE for the identification of patients colonized with CPE (92% vs. 54%). Evidence suggests that this was entirely due to the lack of selectivity of Brilliance CRE and the recovery of non-carbapenemase-producing Enterobacteriaceae from most samples. All CPE recovered in this study grew well when re-inoculated onto Brilliance CRE at a low inoculum (100 CFU/spot); however, all stool specimens containing CPE that were not recovered using Brilliance CRE yielded a heavy growth (>100 colonies) of non-carbapenemase-producing Enterobacteriaceae, suggesting that any CPE present may have been overgrown. All isolates of CPE generated coloured colonies on primary isolation on either medium that were either pink/brown (E. coli) or blue/green (other species), except for one isolate of Citrobacter freundii that generated colourless colonies (isolated on chromID CARBA only). The formation of red or blue colonies (e.g. see Fig. 2) allowed differentiation of CPE from the vast majority of non-Enterobacteriaceae – except for Aeromonas spp., which could be excluded using an oxidase test. If only coloured colonies were considered as presumptive CPE, the sensitivity, specificity and positive predictive value were 54, 23 and 6% for Brilliance CRE and 85, 85 and 36% for chromID CARBA, respectively. Our data suggest that NDM-1 producers generally grow well on Brilliance CRE and this medium may perform well in regions where ESBL producers are not so widespread.
Day et al. (2013) performed a parallel study to this one comparing chromID CARBA with Brilliance CRE for the recovery of CPE from 175 patients attending a military hospital in Rawalpindi, Pakistan. Although the study was performed in a different centre and with a completely different patient population, the authors noted the same limitations with Brilliance CRE, which recovered large numbers of Enterobacteriaceae without carbapenemases. As a result of this lack of selectivity, 32 colonized patients were detected using chromID CARBA compared with only 20 patients using Brilliance CRE (P = 0·0015). The authors speculated that the relatively poor performance of Brilliance CRE might be attributable to suboptimal storage conditions during transport of media via courier to Pakistan. In this study, we have demonstrated that selectivity of both media may deteriorate during storage, even when stored under optimal conditions, but this was most pronounced with Brilliance CRE.
Cohen Stuart et al. (2012) evaluated Brilliance CRE with an international collection of 234 Enterobacteriaceae and reported good recovery of CPE (97%) but growth of 46 of 160 (29%) isolates with ESBL or AmpC ß-lactamase that showed increased MICs of ertapenem. Similarly, Wilkinson et al. (2012) reported growth of 24/70 (34%) isolates with ESBL or AmpC ß-lactamase on Brilliance CRE using a low inoculum (100 CFU) compared with 8/70 (11%) able to grow on chromID CARBA. The latter study also evaluated 100 stool samples referred to a UK laboratory for screening for CPE and, although no positive samples were encountered, Brilliance CRE and chromID CARBA both demonstrated good selectivity.
ChromID CARBA (or its prototype, ID CARBA) has been previously evaluated in two other studies to detect faecal colonization with CPE in patients from Greece (Vrioni et al. 2012) and Pakistan (Perry et al. 2011). Sensitivity of the medium was reported to be 92·4% (Greece) and 87·5% (Pakistan). CHROMagar/Colorex KPC has also been evaluated in a number of studies but has been reported to lack sensitivity for some types of carbapenemase producers, for example NDM-1 (Nordmann et al. 2011; Perry et al. 2011).
It is worth emphasizing that there is currently no ‘gold standard’ method for the detection of CPE from stool samples or rectal swabs. Molecular methods (e.g. PCR) have been advocated by some, and some success has been achieved when specific carbapenemase genes have been sought in rectal swabs (Schechner et al. 2009) and spiked stool samples (Naas et al. 2011). There are, however, a number of important limitations that have been articulated in a recent review by Voulgari et al. (2013). These include the need for specialized equipment and/or skills and the need to target an array of carbapenemase genes in areas where a diverse range of carbapenemases may be encountered. Moreover, detection of a carbapenemase gene does not mean that is harboured by species of the Enterobacteriaceae and it does not facilitate recovery of the host species for subsequent analysis (e.g. susceptibility testing). Finally, rectal swabs that have low inocula of CPE may test negative using PCR but may still be recovered by culture (Samra et al. 2008).
In conclusion, we found that 8·6% of patients with diarrhoea who submitted samples to a reference laboratory in Pakistan were colonized with NDM-1-producing Enterobacteriaceae. In this setting, chromID CARBA proved to be a useful medium for isolation of CPE, whereas Brilliance CRE lacked selectivity against ESBL-producing isolates, thus compromising detection of CPE, and making detection highly labour intensive. It should be emphasized that Brilliance CRE is marketed for the detection of ‘carbapenem-resistant Enterobacteriaceae, including those with New Delhi metallo ß-lactamase 1 (NDM-1) mechanism’, rather than being designed for isolation of carbapenemase producers specifically. More extensive studies in different geographical locations are essential to further evaluate chromogenic media for the detection of CPE so that staff in clinical laboratories can make informed choices of the most appropriate medium to use.
The authors are grateful to bioMérieux, La Balme les Grottes, France for provision of chromID CARBA and for providing part-sponsorship of this study. The Freeman Hospital Microbiology Department (represented by K.M.D., M.W.R. and J.D.P.) receives funding from bioMérieux for the development and evaluation of culture media, and J.D.P. has performed paid consultancy work for the same company. The other authors have no conflict to declare.