Introduction

The spread of carbapenemase-producing Enterobacteriaceae (CPE) is alarming and constitutes a major threat to public health worldwide [1, 2]. In most instances, carbapenemases are associated with other β-lactamases and resistance to several other classes of antimicrobials, rendering treatment particularly challenging. Moreover, carbapenemase-encoding genes are easily transferable through plasmids and transposons, and have the propensity to spread amongst Enterobacteriaceae [1].

In Belgium, the number of CPE isolates reported in hospitals has increased dramatically since 2010 [3]. Over the recent years, the epidemiological situation of CPE has worsened in Europe, showing a doubling of the number of countries reporting inter-regional spread or endemic situation for CPE from 2013 to 2015 [4]. However, these data based on voluntary reporting or questionnaires assessment by national experts are not supported by objective epidemiological indicators, which preclude precise comparisons to be made between countries.

The present study, performed as part of the European Survey on Carbapenemase-Producing Enterobacteriaceae in Europe (EuSCAPE), aimed to determine the incidence and the epidemiological characteristics of carbapenemase-producing Escherichia coli and Klebsiella pneumoniae (CP-E/K) in Belgian hospitals.

Materials and methods

Study design

Eleven laboratories serving secondary- and tertiary-care hospitals (median number of 910 beds per hospital) chosen for their representative geographic distribution across Belgium participated in the study. The laboratories were requested to collect carbapenem non-susceptible (CNS) isolates of E. coli and K. pneumoniae (CNS-E/K) from clinical samples between January 2013 and December 2014. Screening specimens (e.g., stools or rectal swabs) were excluded to limit the impact of variations in screening strategies between hospitals. Carbapenem non-susceptibility was defined as decreased susceptibility to at least one carbapenem drug (meropenem, imipenem, or ertapenem) according to the routine methods performed locally using either Clinical and Laboratory Standards Institute (CLSI) or European Committee on Antimicrobial Susceptibility Testing (EUCAST) interpretative guidelines [5, 6]. From November 2013 to April 2014, for each bacterial isolate, the first successive carbapenem-susceptible isolate of the same species was stored as a control isolate as per the protocol of the EuSCAPE structured survey. All isolates were sent to the reference laboratory, accompanied by the following anonymized data: sample origin, sampling date, type of hospital ward/unit, age and gender of the patient, previous history of hospitalization, recent travel history (last 6 months), clinical significance, and nosocomial context of the sampling (hospital stay for two days or more). Institutional activity data (number of admissions and mean length of stay) of the years 2013 and 2014 were collected individually by each hospital for the calculation of incidence rates.

Characterization of the CNS-E/K isolates

All putative CNS-E/K isolates were verified at the central laboratory for species identification using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (Microflex LT, Bruker Daltonics, Bremen, Germany) and were tested for their phenotypic resistance profile by disk diffusion to β-lactams (cefotaxime, ceftazidime, cefepime, aztreonam, piperacillin/tazobactam, temocillin, ertapenem, and meropenem). The production of carbapenemase was sought by the Carba NP test [7] and by multiplex polymerase chain reaction (PCR) targeting bla OXA-48, bla KPC, bla VIM, bla NDM, and bla IMP [8]. In case of phenotypical suspicion of carbapenemase with a negative result for multiplex PCR targeting the five major carbapenemases, additional PCR assays were performed, including the detection of OXA-427, a novel carbapenem-hydrolyzing class D β-lactamase (CHDL) [9]. For confirmed CP-E/K strains, the minimal inhibitory concentrations (MICs) of 14 antibiotics were determined by the broth microdilution method (Sensititre®, TREK Diagnostic Systems, Cleveland, OH, USA), interpreted using EUCAST interpretative criteria [10], except for temocillin [11]. Colistin-resistant CP-E/K isolates were further analyzed by PCR for the presence of plasmid-mediated mcr-1 encoding gene [12]. In addition, carbapenemase-producing K. pneumoniae isolates were characterized by whole-genome sequencing with multilocus sequence typing (MLST) for a 6-month period. The library preparation and the sequencing were performed using the Nextera® XT DNA Library Prep Kit and the Illumina MiSeq® platform (San Diego, CA, USA). The resulting sequence reads were assembled using CLC Genomic workbench (version 8.5.1, Qiagen, Hilden, Germany). Finally, contigs were submitted to the Center for Genomic Epidemiology MLST 1.8 [13] to assign sequence types (STs) based on the closest matches against the MLST database.

Statistical analysis

The statistical analyses were performed with the MedCalc® software (version 16.4.3, Ostend, Belgium). Fisher’s exact test and the Wilcoxon–Mann–Whitney rank-sum test were used to compare proportions and age distributions, respectively. A p-value <0.05 was considered statistically significant. Items with missing data were not included in the analysis.

Results

From January 2013 to December 2014, 125 (45.1 %) of the 277 CNS-E/K isolates collected were confirmed as CP-E/K by the reference laboratory. Based on the total yearly number of admissions to the participating hospitals (361,713 in year 2013 and 358,667 in 2014), the incidence rate of clinical CP-E/K increased significantly from 0.124 (95 % confidence interval [CI]: 0.088–0.161) in 2013 to 0.223 (95 % CI: 0.174–0.272) cases per 1000 admissions in 2014 (p = 0.001) (Table 1). In 2013, 38 K. pneumoniae and seven E. coli isolates were detected in nine hospitals, while in 2014, 66 K. pneumoniae and 14 E. coli were reported by ten hospitals, with OXA-48 as the predominant type of carbapenemase found (Table 2).

Table 1 Number and incidence rates of carbapenemase-producing Escherichia coli and Klebsiella pneumoniae (CP-E/K) isolates (n = 125) reported by 11 Belgian hospitals during the years 2013 and 2014
Full size table
Table 2 Distribution of carbapenemase types among CP-E/K isolates (n = 125) during the years 2013 and 2014
Full size table

During the 6-month EuSCAPE study period, a total of 179,841 admissions and 1,262,758 patient-days were recorded by the 11 participating hospitals. Overall, 35 CP-E/K isolates were found in ten hospitals, resulting in an overall incidence rate of 0.195 (95 % CI: 0.130–0.259) per 1000 admissions and an incidence density rate of 0.028 (95 % CI: 0.019–0.037) per 1000 patient-days. The microbiological characteristics of the 30 K. pneumoniae and five E. coli strains producing a carbapenemase are presented in Table 3. The 13 KPC-producing K. pneumoniae strains detected in three hospitals were grouped into three different clonal backgrounds, dominated by ST512 (n = 7; two hospitals) and ST101 (n = 5; one hospital), all carrying bla KPC-3. The 16 OXA-48-producing K. pneumoniae were detected in seven hospitals and belonged to eight different STs, with ST405 being the most represented clone (n = 6; three hospitals), followed by ST788 (n = 3; one hospital) and ST11 (n = 2; two hospitals). The two OXA-427 carbapenemase-producing strains originated from one single hospital.

Table 3 Clonal relatedness of CP-E/K collected during the six-month EuSCAPE study period
Full size table

The antimicrobial susceptibility testing of 80 CP-E/K strains is summarized in Table 4. 100 % and 97.5 % of CP-E/K isolates were found to be resistant to piperacillin/tazobactam and temocillin, respectively, including high-level temocillin resistance for all OXA-48 producers (MIC range: 64 to >256 mg/L), while the non-susceptibility rate for ertapenem and meropenem were 86.3 % and 45.0 %, respectively. The carbapenem MIC50 values for OXA-48 K. pneumoniae (ertapenem: 4 mg/L, meropenem: 1 mg/L) were lower than those obtained for KPC K. pneumoniae (ertapenem: 64 mg/L, meropenem: 32 mg/L). Six K. pneumoniae (three OXA-48 and three KPC producers) and one OXA-48-positive E. coli isolates were resistant to colistin, but none of these isolates were found to be mcr-1 producers. Three K. pneumoniae (two OXA-48 and one NDM producers) and one OXA-427-producing E. coli strains were intermediately resistant to tigecycline (MIC value: 2 mg/L).

Table 4 Susceptibility testing and minimum inhibitory concentration (MIC) results of CP-E/K isolates detected in 2014 (n = 80)
Full size table

When compared to the 56 control patients (Table 5), CP-E/K-positive patients were more likely to be a male, to have been hospitalized in the last 6 months, or to have stayed in hospital for at least two days before sampling. A higher proportion of bacterial isolates from blood cultures was found in control patients, while, conversely, a higher proportion of lower respiratory tract specimens was documented in CP-E/K patients. One NDM-producing E. coli was recovered from a patient who had traveled to Pakistan and three OXA-48-positive patients had previously stayed in Spain, Tunisia, and Iran, respectively. Data on possible travel importation were lacking for 34.3 % of CP-E/K patients and for 80.4 % of the control patients.

Table 5 Comparison of clinical characteristics of control and CP-E/K-carrying patients
Full size table

Discussion

CPE are rapidly disseminating worldwide, including also in Europe [1]. The European Antimicrobial Resistance Surveillance Network (EARS-Net) is the current European surveillance system designed to monitor crude resistance rates to antimicrobial agents (but not resistance mechanism) used in human medicine. Therefore, the EuSCAPE project was launched in 2012 to gain better insight to the epidemiology of CPE in Europe [14]. This initiative aimed to assess the occurrence of CPE among hospitalized patients using a European-wide standardized sampling frame and to compare calculated epidemiological indicators between countries.

In Belgium, a multicentric prevalence survey performed among 24 Belgian hospitals in 2012 demonstrated an estimated prevalence of 0.28 % of CPE isolates among 4564 screened Enterobacteriaceae isolates (screening samples included in this study). Klebsiella pneumoniae was the most frequent CPE species and OXA-48 the most often encountered carbapenemase [15].

The present study highlighted the significant increase in incidence of CP-E/K in Belgium from 2013 to 2014. A similar rising trend in incidence rates of CNS-E/K was also observed in the National Surveillance report of Health-care-associated infections (NSIH) [16]. Indeed, the overall average incidences of meropenem-I/R E. coli and K. pneumoniae increased from 0.03 and 0.19 cases per 1000 admissions, respectively, in 2013, to 0.15 and 0.36 cases per 1000 admissions in 2014, respectively. However, the results from this particular surveillance program included bacterial isolates from both clinical and screening specimens reported by 120 hospitals country-wide and, like EARS-Net, address mainly carbapenem resistance (intermediately resistant and resistant considered together) and not specifically the carbapenemase resistance mechanism.

In comparison to other EuSCAPE participating countries, the incidence rate of CP-E/K (0.195 per 1000 admissions) in Belgium was higher than in Germany (0.047 CP-E/K per 1000 admissions), but showed a similar distribution of carbapenemase types, mainly represented by OXA-48 (54.3 %) and KPC (37.1 %) [17]. In Italy, nearly all (97 %) of the 187 CP-E/K isolates carried bla KPC, confirming the endemic situation of KPC in this country [18], while in Romania, OXA-48 producers represented a large majority (78.5 %) of the 65 CP-E/K isolates [19]. In the Czech Republic, only two CP-E/K isolates were detected [20]. However, for these three countries, no incidence rates were provided by the authors.

Molecular typing of CP K. pneumoniae strains showed for KPC producers the clonal dissemination of a small number of clusters (mostly ST512 and ST101) in three participating hospitals. The ST512 clone, a single-locus allelic variant of the pandemic ST258 clone [19], and ST101 have been reported as KPC-producing K. pneumoniae causing outbreaks in Italy [21,22]. Our data confirmed the almost complete replacement of ST258 by ST512 and ST101 clones among KPC producers observed since 2011 in Belgium [23].

On the other hand, a more diversified picture was observed for OXA-48-producing K. pneumoniae, which were distributed into eight different STs, supporting the horizontal transfer of the IncL/M-type pOXA-48 conjugative plasmid previously described [24]. Nevertheless, three major STs (ST405, ST788, and ST11) associated with OXA-48 were identified in our study. ST405 is a successful clone of OXA-48 K. pneumoniae already described in Spain [25] and France [26], while ST11 has been reported in Spain [25] and Greece [27]. Further, the clustering of three OXA-48-producing strains belonging to ST788 at one single hospital strongly suggested the local dissemination of this clone.

OXA-427 is a newly described plasmid-mediated CHDL unrelated to OXA-48 (26 % amino acid homology) that displays a broad resistance profile to β-lactams, including expanded-spectrum cephalosporins and carbapenems [9]. The encoding bla OXA-427 gene is carried on an IncA/C plasmid that was detected sporadically in various Enterobacteriaceae species isolates at one Belgian hospital from 2013 onwards.

Antimicrobial susceptibility testing confirmed resistance to piperacillin/tazobactam and temocillin in nearly all CP-E/K isolates, while less than half of the CP-E/K were non-susceptible to meropenem, with lower carbapenem MIC values for OXA-48 producers than those expressing KPC, in line with previous reports [28]. The resistance rate for colistin was 8.7 %, most probably due to chromosomal mutations, as no mcr-1-positive isolates were identified among the seven strains tested. This resistance proportion was much lower than that found in Germany (33.3 %) [17] or Italy (43 %) [18]. The non-susceptibility rate for tigecycline was still low, with only four intermediately resistant strains (5.0 %).

Concerning the risk factor analysis, no clear explanation can be provided for the higher proportion of male subjects among patients with CP-E/K, although the same association was reported in the German study [17]. The higher proportion of bacteremic isolates in control patients could probably be reflected by the tendency of laboratories to store preferentially in collections bacterial isolates recovered from deep-seated infection sites. Recent hospitalization was also identified as a predictor for CPE carriage in Israel [29]. Travel data were missing for the majority of patients included in the study. It is, therefore, difficult to draw conclusions, although the proportion of patients with travel history was higher in the CP-E/K group, similar to the German data [17]. A previous publication further highlighted that the proportion of travel-imported CPE cases could be largely underestimated due to the lack of active investigation in most cases [30].

We acknowledge the fact that our study might have suffered from several limitations. In addition to the incompleteness of the clinical data collected, the EuSCAPE study protocol addressed only K. pneumoniae and E. coli species. In Belgium, at least 20 % of the confirmed CPE isolates belong to Enterobacteriaceae species other than K. pneumoniae and E. coli [16,31]. Therefore, the EuSCAPE structured survey may not give a complete picture of the incidence of clinical CPE. Secondly, the limited participation in this survey could not allow the extrapolation of the results to the whole country. However, the increasing number of CPE reported by hospitals to the national reference laboratory [31] is in line with the trend observed in this survey. Finally, it is difficult to evaluate the impact of independent outbreaks on the rising incidence of CP-E/K between 2013 and 2014. However, the incidence increase in 7 out of the 11 participating hospitals (and only one hospital had decreased incidence) strongly suggests a true global increase in the incidence of CP-E/K.

In summary, this survey highlighted the rapid diffusion of CP-E/K from clinical specimens of CPE in Belgian hospitals between 2013 and 2014, yet the overall incidence remained low (0.195 per 1000 admissions). OXA-48 and KPC are the two major carbapenemases found in Belgium, with noticeable differences in their mode of transmission, mostly clonal for KPC and preferentially through horizontal plasmid transfer for OXA-48 producers.