Increasing proportion of carbapenemase-producing Enterobacteriaceae and emergence of a MCR-1 producer through a multicentric study among hospital-based and private laboratories in Belgium from September to November 2015

Carbapenemase-producing Enterobacteriaceae (CPE) strains have been increasingly reported in Belgium. We aimed to determine the proportion of CPE among Enterobacteriaceae isolated from hospitalised patients and community outpatients in Belgium in 2015. For the hospitalised patients, the results were compared to a previous similar survey performed in the same hospitals in 2012. Twenty-four hospital-based and 10 private laboratories collected prospectively 200 non-duplicated Enterobacteriaceae isolates from clinical specimens. All isolates were screened locally by carbapenem disk diffusion using European Committee on Antimicrobial Susceptibility Testing methodology. Putative CPE strains with inhibition zone diameters below the screening breakpoints were referred centrally for confirmation of carbapenemase production. From September to November 2015, we found a proportion of clinical CPE of 0.55% (26/4,705) and of 0.60% (12/1,991) among hospitalised patients and among ambulatory outpatients respectively. Klebsiella pneumoniae (26/38) and OXA-48-like carbapenemase (28/38) were the predominant species and enzyme among CPE. One OXA-48-producing Escherichia coli isolated from a hospital was found carrying plasmid-mediated MCR-1 colistin resistance. Compared with the 2012 survey, we found a significant increased proportion of clinical CPE (0.55% in 2015 vs 0.25% in 2012; p = 0.02) and an increased proportion of hospitals (13/24 in 2015 vs 8/24 in 2012) with at least one CPE detected. The study results confirmed the concerning spread of CPE including a colistin-resistant MCR-1 producer in hospitals and the establishment of CPE in the community in Belgium.


Introduction
Acquired carbapenemases in Enterobacteriaceae have been reported extensively worldwide [1]. Asymptomatic carriage and infection caused by carbapenemaseproducing Enterobacteriaceae (CPE) isolates currently raise major public health concern for individual therapeutic management and for collective infection control [2]. The prevalence of carbapenem resistance and the types of carbapenemases found in Europe vary between countries [3,4] and since 2010 an increase in the isolation rate of CPE from various European countries has been reported [5]. In Belgium, the last multicentric survey performed among 24 hospitals in 2012 [6] showed an overall proportion of 3.5% of carbapenem non-susceptible Enterobacteriaceae (CNSE) and an estimated proportion of 0.28% of CPE among Enterobacteriaceae isolates using Clinical and Laboratory Standards Institute (CLSI) carbapenems susceptibility breakpoints as screening criteria [7]. Further, one third (8/24) of the participating hospital laboratories had isolated one or several CPE isolates in their institution during the study period [6]. The national surveillance programme of CPE established in January 2012 by the Scientific Public Health Institute together with the National Reference Centre showed a yearly increase from 2012 to 2014 of number and proportion of hospital-based and of private laboratories reporting confirmed CPE [8].
The present cross-sectional survey aimed to determine the evolution of the proportion of CNSE and of CPE among Enterobacteriaceae isolates collected in Belgian hospitals and to gain insight of their epidemiology outside hospitals in specimens collected from community outpatients sent to private laboratories.

Study design, inclusion criteria and testing at participating centres
In 2015, 24 hospital-based laboratories that participated to the survey in 2012 serving around 25% of all acute hospitals in Belgium (hospital size ranging from 288 to 1,803 beds; median number of 747 beds per hospital) were requested to collect consecutively 200 Enterobacteriaceae isolates cultured from clinical samples of hospitalised patients (screening samples such as stools or rectal swabs were excluded) over a period of 2 months (i.e. from September to November). Only the first isolate of the same species per patient was included. The sample collection, culture and bacterial identification were performed using local procedures. In addition, 10 private laboratories (representing around 20% of all private laboratories in Belgium) serving general practitioners participated as well using the same study protocol and were asked to test 200 Enterobacteriaceae isolates cultured from clinical samples of outpatients in the community.
All isolates were tested locally for susceptibility to meropenem (10-µg) and ertapenem (10-µg) by disk diffusion method according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines [9] and all inhibition zone diameters were recorded. All testing materials purchased from the same manufacturer (BioRad, Marnes-la-Coquette, France) had the same manufacture batch number in order to avoid inter-lot variations. Escherichia coli ATCC 25922 was tested as quality control strain in each centre during the survey.
Enterobacteriaceae isolates showing a decreased inhibition zone diameters according to EUCAST screening breakpoints [10] to any of the two carbapenems disks (ertapenem < 25 mm and/or meropenem < 25 mm) were defined as putative CPE (PCPE) and had to be referred to the reference laboratory.

Characterisation of resistance mechanisms and data analysis
All PCPE isolates sent to the reference laboratory were tested for carbapenemase production by the electrochemical imipenem-hydrolysis based BYG Carba test [11] and underwent multiplex PCR targeting blaVIM, blaIMP, blaNDM, blaKPC and blaOXA-48 for the detection of carbapenemase encoding genes [12]. The PCRpositive amplicons were sequenced using an external service company (Macrogen Inc., Seoul, South Korea). The sequence obtained was compared with the genes present in GenBank and aligned with the reference gene cited at the Lahey Clinic [13].
For confirmed CPE isolates, minimal inhibitory concentration (MIC) values of antimicrobial agents including carbapenems were determined by broth microdilution method (Sensititre, Trek Diagnostic Systems, Cleveland, US). The MIC value of temocillin was determined by Etest method (bioMérieux, Marcy-l'Etoile, France). Susceptibility categorisation was interpreted according to EUCAST interpretative criteria [9] for all antimicrobial agents except temocillin (breakpoints according to Fuchs et al.) [14]. Colistin-resistant isolates were further tested by end-point PCR for the presence of the plasmid-mediated colistin resistance mcr-1 gene [15].
The proportion of CNSE as defined by the EUCAST disk diffusion susceptibility breakpoints (ertapenem <25 mm and/or meropenem <22 mm) and of CPE isolates were calculated based on inhibition zone diameters of all isolates provided overall in hospitals or in private laboratories, for each centre and also by species or genus. Following the application of harmonised microbiological criteria (the ertapenem and meropenem discs zone diameters recorded in 2012 were reinterpreted using EUCAST interpretative criteria), results from 2015 were compared with those in 2012 for hospital-based laboratories.

Results
In total, 4,705 and 1,991 non-duplicate Enterobacteriaceae isolates were screened by the 24 hospital-based and by 10 private laboratories respectively from September to November 2015. A median number of 200 isolates per laboratory was achieved and all but one centre reached > 195 isolates screened. The distributions of origins of screened isolates were highly comparable to 2012 with urine (2,642/4,705; 56%) and medical wards (1,818/4,705; 39%) representing the principal origins of the isolates screened and the main sources of CNSE isolates in hospitals (data not shown). The highest proportion of CNSE isolates in hospitals was found in respiratory tract (36/714; 5.0%) and in the intensive care units (30/711; 4.2%). On the other hand, urinary origin (1,902/1,991; 96%) of screened isolates was as expected largely predominant in private laboratories.
The number and the proportions of CNSE according to EUCAST susceptibility criteria and of CPE among tested isolates per species are reported and compared with the 2012 study results in Table 1. Among ambulant patients, a total of 37 isolates from seven private laboratories would be categorised as  (2) C24 OXA-48 (2)     and NDM-type, n = 2). OXA-48 carbapenemase was detected in four E. coli isolates including one coproducing NDM enzyme, three Citrobacter freundii, two C. koseri and two K. oxytoca isolates. One VIM-producing K. oxytoca was also detected by PCR. All PCR-negative PCPE isolates did not hydrolyse imipenem by BYG Carba test, thus excluding the phenotypical expression of undetected carbapenemases by our multiplex PCR assay.
In 2015 among hospitals, while the 16 cases of OXA-48 CPE were distributed in nine different hospitals, the 6 KPC-type CPE cases were grouped within two hospitals. In the community setting, the 12 CPE isolates (dominated by OXA-48; n = 11) were found in isolates from three private laboratories. Based on disc diffusion with EUCAST interpretative criteria, all 38 confirmed CPE isolates in 2015 were intermediately-resistant or resistant to ertapenem disk while 16 (42%; all of OXA-48 type) were categorised as susceptible to meropenem disk. Two OXA-48producing (one E. coli and one C. koseri) isolates had an inhibition zone to meropenem of ≥ 25 mm and would be missed if EUCAST screening breakpoint was applied using meropenem disc alone. One of these two CPE (C. koseri) showed additionally an ertapenem zone diameter of 24 mm and therefore would have been missed as well if the CLSI susceptibility breakpoints (ertapenem <22 mm and/or meropenem <23 mm) used as screening criteria in the 2012 survey were applied.
Origins and microbiological characteristics including MIC results of the 38 confirmed CPE isolates are detailed in Table 3.
Molecular typing by rep-PCR of the 26 carbapenemase-producing K. pneumoniae isolates showed their belonging to 19 different DiversiLab (DL) patterns with four clusters (DL types 3, 5, 16 and 23) of more than one isolates ( Table 3). The three DL type 16 KPC-3 producing isolates were detected in centre C23 and the three DL type 23 OXA-48 producing isolates in centre C17. The other DL types 3 and 5 (all OXA-48 producing) isolates were found in two (one hospital-based and one community-serving) centres each.

Discussion
In this multicentric 2015 survey, we found an overall doubling proportion of 0.55% of clinical CPE among Enterobacteriaceae isolates in hospitals compared with 2012 (0.25%). The microbiological analysis confirmed the predominance of K. pneumoniae and of OXA-48 among CPE isolates in hospitals and in the community in Belgium. Molecular typing of carbapenemase-producing K. pneumoniae confirmed mostly polyclonal distribution although clusters of identical DL types with suspected clonal spread were highlighted as well. Two clusters of different DL types (16 and 23) strongly suggested their intra-hospital clonal dissemination or outbreak. Additionally, the presence of other OXA-48 producing DL types (3 and 5) both in hospitals and in private laboratories that are geographically close (distance < 30 km) raised the possibility of epidemiological links between the two healthcare sectors. Although no patient history regarding travel or hospitalisation in foreign country was available in this study, data from the national surveillance programme and from the previous prevalence study in 2012 support that the proportion of travelimported CPE cases should be limited [17].
The strength of our study relies on the use of a standardised methodology performed within the same hospitals between the two study periods to measure the epidemiological parameters with minimal variations due to technical inter-laboratory variability or samples selection bias. The 24 hospital-based laboratories serving tertiary care and general acute-care hospitals of medium to large size were selected for their representative distribution in the three geographical regions (Flanders, Brussels and Wallonia) across Belgium. Only strains isolated from clinical samples were included in the studies in order to limit recruitment bias due to variations in screening strategies for CPE between hospitals. However, our study has some limitations. First, 18% (31/175) of the PCPE isolates detected locally were not referred to the reference laboratory suggesting that the proportion of CPE of 0.55% should be considered as a minimal estimate. Also, the higher proportion of CPE in hospitals in 2015 compared to 2012 could result from the use of EUCAST screening breakpoints theoretically more sensitive than the CLSI susceptibility criteria applied in 2012. In spite of this, the fact that in the present study, only one OXA-48 producing C. koseri had the two carbapenems inhibition zones above the CLSI susceptibility cut-offs suggests very rare occurrence of these CPE isolates that poorly expressed carbapenemases.
In addition, the higher proportion of colistin resistance among CPE strains detected in hospitals corroborates with the phenomenon of increasing colistin resistance among CPE isolates observed in Belgium and elsewhere for the past years [18,19]. Furthermore the study evidenced the emergence of the first human MCR-1positive CPE in Belgian hospitals as well, similar other countries worldwide, where this has also been recently reported in healthcare [20][21][22].
In Spain, the latest prospective multicentric study in 83 hospitals performed in 2013 reported a similar overall estimated prevalence of 0.3% of CPE [4] (higher than the 0.04% of metallo-beta-lactamase-producing CPE detected in the previous 35-centres survey in 2009) [23], but 20% of the CPE were of faecal screening origin.
The survey showed the same predominance of OXA-48 type CPE, but VIM-producing isolates (the second most frequent) were more widely geographically distributed. Sporadic community-onset infection and post-travel acquisition of CPE had been reported in western countries [24][25][26]. In 2016 a case report in Belgium described a household acquisition of OXA-48 producing K. pneumoniae in a woman living in the same household than a patient previously infected with a similar CPE, raising the possibility of community spread in low-prevalence areas [27]. However to our knowledge, no specific prospective cross-sectional survey addressing clinical isolates in an ambulatory setting has yet been published.
Regarding the smaller proportion of CNSE (mainly due to the significant decreasing proportion of CNSE among Enterobacter spp.) in hospitals in 2015 compared with 2012, no clear explanation could be found. However, a cluster effect could be excluded since the decreased rate of CNSE was documented in 18 of the 24 participating hospitals (data not shown).
In conclusion, this multicentric survey demonstrated a significantly increased proportion of CPE among clinical Enterobacteriaceae isolates in hospitals from 2012 to 2015 with an increased proportion of hospitals-based laboratories detecting at least one CPE. Although the proportion of CPE remained globally low, the steady progression and spread of CPE in Belgian hospitals and the emergence of colistin-resistant (including plasmid-mediated mcr-1 harbouring) strains among CPE raises major concerns. In parallel, the survey in 10 private laboratories highlighted for the first time the establishment of CPE (mainly K. pneumoniae species and of OXA-48 type) in an ambulatory setting and suggested possible epidemiological links with hospitals. We therefore believe that repeated national or regional epidemiological studies using standardised protocols at different healthcare levels are essential to measure more accurately the burden of carbapenem resistance.