https://orcid.org/0000-0003-1675-3173
Abstract
The ST131 Escherichia coli clone is associated with the global dissemination of ESBLs. It has been hypothesized that ST131 could take advantage of better colonizing abilities. However, the data on colonization prevalence of ESBL-ST131 in European hospitals are scarce.
To assess the prevalence of the ST131 clone and its microbiological characteristics among colonizing ESBL-producing E. coli (ESBL-Ec) from hospitalized patients in four European hospitals (Berlin, Geneva, Madrid and Utrecht) during the R-GNOSIS study.
ESBL-Ec isolates (n = 688) were obtained from rectal swabs of hospitalized patients from March 2014 to February 2015 using selective media. The ST131 clone and its subclones were sought using PCR and positive isolates were further studied. blaESBL genes were characterized (PCR and sequencing), antibiotic susceptibility testing was performed, clonal relationships were studied by PFGE and fimH allele and O type (PCR) were assessed.
ST131 prevalence was 20.5% (141/688); C1/H30R1 isolates were significantly more prevalent in Geneva (49%) and C2/H30Rx in Madrid (67%). C1/H30R1 isolates showed less resistance to amikacin than C2/H30Rx (4% versus 35%) and all were susceptible to penicillin/inhibitor combinations. CTX-M-15 was the most common enzyme (49%) followed by CTX-M-27 (27%). C1/H30R1 isolates were significantly associated with CTX-M-27 (72%) and all of these isolates belonged to the C1-M27 clade. Moreover, C2/H30Rx isolates and CTX-M-15 were also significantly related (88%).
The predominance of C2/H30Rx-CTX-M-15 in Madrid and C1/H30R1-CTX-M-27 in Geneva demonstrates a changing epidemiology of ESBLs in Europe caused by ST131 subclones; in particular, the emergence of the C1-M27 clade in Europe.
Introduction
Resistance to β-lactam antibiotics due to ESBLs in extraintestinal pathogenic Escherichia coli represents a major public health problem. The worldwide dissemination of the resistance genes encoding these enzymes is mainly associated with E. coli ST131, an MDR pandemic clone causing extraintestinal infections, particularly urinary tract infections and bacteraemia.1–4 Population analysis studies of this clone have shown that the emergence of ST131 E. coli is driven by a specific sublineage named clade C or C/H30, characterized by fluoroquinolone resistance and by the expression of allele H30 of the type 1 fimbriae fimH adhesin.2,3 Price et al.3 described the sublineage H30Rx, subsequently named C2 by Petty et al.,2 nested in clade C and associated with ESBL production, particularly CTX-M-15. In that study, clade H30R1 or C1 included those H30 strains not associated with CTX-M-15 production. Recent studies have described the C1-M27 clade in Japanese, German and French ST131 isolates, characterized by the production of CTX-M-27 and a prophage-like region named M27PP1.5–7 Among non-H30 strains, two other clades, A and B, have been described, associated with alleles H41 and H22, respectively.
Clade C/H30 is the dominant sublineage among ST131 ESBL-producing E. coli (ESBL-Ec) isolates in most studies.8,9 The reason for the successful worldwide dissemination of the ST131 clone and its C/H30R sublineage is not yet completely understood. The transmissibility of ST131 between humans and even among household members including pets has been previously probed in several studies.10–15 The newly described clade C1-M27 is often found colonizing humans, animals and the environment. A study on ESBL transmission dynamics in two rehabilitation wards in a tertiary centre in Israel showed that CTX-M-27-ST131 E. coli had a higher transmission rate than CTX-M-15-ST131.16 Some authors have proposed that this dissemination may be accelerated by the expression of virulence traits that give ST131 strains an advantage over other extraintestinal E. coli to colonize the intestinal tract,2 especially in subjects that have received antibiotics or been frequently admitted to hospital.17–19 Moreover, several studies have demonstrated that ST131 possesses enhanced intestinal colonization capabilities (i.e. binding and invading intestinal epithelial cells and translocation) and these traits seem to be mediated by type-1 fimbriae, with slight differences among ST131 clades.20,21
Therefore, the gastrointestinal niche may act as a reservoir from where the ST131 clone is able to access the bloodstream and cause infection.22,23 However, the data regarding colonization prevalence of ST131 are scarce, are often based on small samples lacking analysis of subclades and seem to vary greatly depending on the host group studied and the selection of the strains (for example, non-selected, ESBL producers or fluoroquinolone resistant).24
A broader observation of the colonization prevalence of the different ST131 clades is needed to better understand the expansion dynamics of this clone and its increasing prevalence in clinical samples. The Resistance of Gram-Negative Organisms: Studying Intervention Strategies (R-GNOSIS) project, which studied strategies for the prevention of the transmission of colonizing ESBL-producing Enterobacteriaceae, provided an excellent opportunity. During this study, a collection of colonizing ESBL-Ec from hospitalized patients from four different European hospitals was obtained. Therefore, we aimed to assess the prevalence, geographical distribution and microbiological characteristics of the ST131 clone and its lineage C/H30 (both C2/H30Rx and C1/H30R1 subsets) and the non-H30 subset in this sample.
Materials and methods
Study design and isolate collection
For this project, a subset of isolates collected during the first year of a larger trial within the R-GNOSIS study group was analysed. Work package 5 (WP5) of this project investigated patient isolation strategies for the prevention of ESBL transmission in medical and surgical wards by evaluation of the rectal colonization by ESBL-producing Enterobacteriaceae of hospitalized patients on admission, weekly for patients staying >7 days and at discharge if the patient’s stay was longer than 3 days. Participating hospitals were based in Berlin (Germany), Geneva (Switzerland), Madrid (Spain) and Utrecht (The Netherlands). This study was conducted between March 2014 and February 2015 in agreement with the declaration of Helsinki and with the guidelines of Good Clinical Practice (ICH-GCP-Guidelines, CPMP/ICH/135/95) issued by the EMA. A waiver of written informed consent from individual patients in the participating wards was granted at all sites.
Swabs were plated on ChromID-ESBL agar plates (bioMérieux, Marcy-l’Étoile, France), which were incubated at 37°C for 48 h for the detection of ESBL-producing isolates. Identification of putative E. coli colonies was performed by MALDI-TOF MS (Bruker Daltonik, Bremen, Germany) and ESBL production was confirmed by the double disc synergy method and the ESBL and AmpC Screen Kit (Rosco Diagnostica, Taastrup, Denmark). In total, 1009 ESBL-Ec isolates from 688 patients were collected. When multiple E. coli isolates were obtained from a patient, the first detected isolate was selected. Finally, 688 ESBL-Ec were selected for further studies (one isolate per patient).
Antibiotic susceptibility testing and characterization of ESBL genes
Antibiotic susceptibility testing was performed by the disc diffusion method using the ADAGIO reader (ADAGIO automated system, Bio-Rad, Hercules, CA, USA) and results interpreted by EUCAST criteria (www.eucast.org). The antimicrobials tested are shown in Table 1. The EUCAST epidemiological cut-off value was used for nalidixic acid (isolates with halo diameters ≥19 mm were considered susceptible). Zone diameter values categorized as intermediate and resistant were considered as non-susceptible. Genes encoding ESBLs (blaTEM, blaSHV and blaCTX-M) were characterized by PCR and sequencing.25–30
Differences in antibiotic non-susceptibility of ST131 isolates by subsets and by the two most commonly produced ESBL enzymes
| Agent | ST131 (n = 141) | Non-H30 (n = 42) | P valuea | ESBL type | P valuea CTX-M-15 versus CTX-M-27 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| C1/H30R1 (n = 47) | C2/H30Rx (n = 52) | C2/H30Rx versus C1/H30R1 | non-H30 versus C2/H30Rx | non-H30 versus C1/H30R1 | CTX-M-15 (n = 69) | CTX-M-27 (n = 38) | ||||
| TZP | 21 (15) | 5 (12) | 0 (0) | 16 (31) | <0.01 | 0.03 | 0.02 | 15 (22) | 0 (0) | <0.01 |
| AMC | 23 (16) | 5 (12) | 0 (0) | 18 (35) | <0.01 | 0.011 | 0.02 | 17 (25) | 0 (0) | <0.01 |
| CAZ | 87 (62) | 28 (67) | 25 (53) | 34 (65) | 0.22 | 0.90 | 0.20 | 52 (75) | 20 (53) | 0.02 |
| FEP | 77 (55) | 32 (76) | 12 (26) | 33 (64) | <0.01 | 0.18 | <0.01 | 50 (72) | 6 (16) | <0.01 |
| NAL | 127 (90) | 30 (71) | 46 (98) | 51 (98) | 1.0 | <0.01 | <0.01 | 66 (96) | 36 (95) | 1.0 |
| CIP | 101 (72) | 3 (7) | 47 (100) | 51 (98) | 1.0 | <0.01 | <0.01 | 52 (75) | 36 (95) | 0.02 |
| AMK | 20 (14) | 0 (0) | 2 (4) | 18 (35) | <0.01 | <0.01 | 0.50 | 18 (26) | 2 (5) | <0.01 |
| TOB | 64 (45) | 15 (36) | 9 (19) | 40 (77) | <0.01 | <0.01 | 0.08 | 51 (74) | 1 (3) | <0.01 |
| GEN | 46 (33) | 15 (36) | 8 (17) | 23 (44) | <0.01 | 0.40 | 0.04 | 34 (49) | 0 (0) | <0.01 |
| SXT | 69 (49) | 18 (43) | 25 (53) | 26 (50) | 0.75 | 0.49 | 0.33 | 34 (49) | 20 (53) | 0.74 |
| Agent | ST131 (n = 141) | Non-H30 (n = 42) | P valuea | ESBL type | P valuea CTX-M-15 versus CTX-M-27 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| C1/H30R1 (n = 47) | C2/H30Rx (n = 52) | C2/H30Rx versus C1/H30R1 | non-H30 versus C2/H30Rx | non-H30 versus C1/H30R1 | CTX-M-15 (n = 69) | CTX-M-27 (n = 38) | ||||
| TZP | 21 (15) | 5 (12) | 0 (0) | 16 (31) | <0.01 | 0.03 | 0.02 | 15 (22) | 0 (0) | <0.01 |
| AMC | 23 (16) | 5 (12) | 0 (0) | 18 (35) | <0.01 | 0.011 | 0.02 | 17 (25) | 0 (0) | <0.01 |
| CAZ | 87 (62) | 28 (67) | 25 (53) | 34 (65) | 0.22 | 0.90 | 0.20 | 52 (75) | 20 (53) | 0.02 |
| FEP | 77 (55) | 32 (76) | 12 (26) | 33 (64) | <0.01 | 0.18 | <0.01 | 50 (72) | 6 (16) | <0.01 |
| NAL | 127 (90) | 30 (71) | 46 (98) | 51 (98) | 1.0 | <0.01 | <0.01 | 66 (96) | 36 (95) | 1.0 |
| CIP | 101 (72) | 3 (7) | 47 (100) | 51 (98) | 1.0 | <0.01 | <0.01 | 52 (75) | 36 (95) | 0.02 |
| AMK | 20 (14) | 0 (0) | 2 (4) | 18 (35) | <0.01 | <0.01 | 0.50 | 18 (26) | 2 (5) | <0.01 |
| TOB | 64 (45) | 15 (36) | 9 (19) | 40 (77) | <0.01 | <0.01 | 0.08 | 51 (74) | 1 (3) | <0.01 |
| GEN | 46 (33) | 15 (36) | 8 (17) | 23 (44) | <0.01 | 0.40 | 0.04 | 34 (49) | 0 (0) | <0.01 |
| SXT | 69 (49) | 18 (43) | 25 (53) | 26 (50) | 0.75 | 0.49 | 0.33 | 34 (49) | 20 (53) | 0.74 |
TZP, piperacillin/tazobactam; AMC, amoxicillin/clavulanate; CAZ, ceftazidime; FEP, cefepime; NAL, nalidixic acid; CIP, ciprofloxacin; AMK, amikacin; TOB, tobramycin; GEN, gentamicin; SXT, co-trimoxazole.
Values are number of isolates (percentage of the column total).
All ST131 isolates were non-susceptible to cefotaxime and susceptible to cefoxitin, tigecycline, imipenem and meropenem.
Intermediate results were considered non-susceptible.
After using the Bonferroni correction method, P values were considered significant if P ≤ 0.02 for comparison across subclones and P < 0.05 for comparison according to ESBL production. P values that are statistically significant are in bold.
Differences in antibiotic non-susceptibility of ST131 isolates by subsets and by the two most commonly produced ESBL enzymes
| Agent | ST131 (n = 141) | Non-H30 (n = 42) | P valuea | ESBL type | P valuea CTX-M-15 versus CTX-M-27 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| C1/H30R1 (n = 47) | C2/H30Rx (n = 52) | C2/H30Rx versus C1/H30R1 | non-H30 versus C2/H30Rx | non-H30 versus C1/H30R1 | CTX-M-15 (n = 69) | CTX-M-27 (n = 38) | ||||
| TZP | 21 (15) | 5 (12) | 0 (0) | 16 (31) | <0.01 | 0.03 | 0.02 | 15 (22) | 0 (0) | <0.01 |
| AMC | 23 (16) | 5 (12) | 0 (0) | 18 (35) | <0.01 | 0.011 | 0.02 | 17 (25) | 0 (0) | <0.01 |
| CAZ | 87 (62) | 28 (67) | 25 (53) | 34 (65) | 0.22 | 0.90 | 0.20 | 52 (75) | 20 (53) | 0.02 |
| FEP | 77 (55) | 32 (76) | 12 (26) | 33 (64) | <0.01 | 0.18 | <0.01 | 50 (72) | 6 (16) | <0.01 |
| NAL | 127 (90) | 30 (71) | 46 (98) | 51 (98) | 1.0 | <0.01 | <0.01 | 66 (96) | 36 (95) | 1.0 |
| CIP | 101 (72) | 3 (7) | 47 (100) | 51 (98) | 1.0 | <0.01 | <0.01 | 52 (75) | 36 (95) | 0.02 |
| AMK | 20 (14) | 0 (0) | 2 (4) | 18 (35) | <0.01 | <0.01 | 0.50 | 18 (26) | 2 (5) | <0.01 |
| TOB | 64 (45) | 15 (36) | 9 (19) | 40 (77) | <0.01 | <0.01 | 0.08 | 51 (74) | 1 (3) | <0.01 |
| GEN | 46 (33) | 15 (36) | 8 (17) | 23 (44) | <0.01 | 0.40 | 0.04 | 34 (49) | 0 (0) | <0.01 |
| SXT | 69 (49) | 18 (43) | 25 (53) | 26 (50) | 0.75 | 0.49 | 0.33 | 34 (49) | 20 (53) | 0.74 |
| Agent | ST131 (n = 141) | Non-H30 (n = 42) | P valuea | ESBL type | P valuea CTX-M-15 versus CTX-M-27 | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| C1/H30R1 (n = 47) | C2/H30Rx (n = 52) | C2/H30Rx versus C1/H30R1 | non-H30 versus C2/H30Rx | non-H30 versus C1/H30R1 | CTX-M-15 (n = 69) | CTX-M-27 (n = 38) | ||||
| TZP | 21 (15) | 5 (12) | 0 (0) | 16 (31) | <0.01 | 0.03 | 0.02 | 15 (22) | 0 (0) | <0.01 |
| AMC | 23 (16) | 5 (12) | 0 (0) | 18 (35) | <0.01 | 0.011 | 0.02 | 17 (25) | 0 (0) | <0.01 |
| CAZ | 87 (62) | 28 (67) | 25 (53) | 34 (65) | 0.22 | 0.90 | 0.20 | 52 (75) | 20 (53) | 0.02 |
| FEP | 77 (55) | 32 (76) | 12 (26) | 33 (64) | <0.01 | 0.18 | <0.01 | 50 (72) | 6 (16) | <0.01 |
| NAL | 127 (90) | 30 (71) | 46 (98) | 51 (98) | 1.0 | <0.01 | <0.01 | 66 (96) | 36 (95) | 1.0 |
| CIP | 101 (72) | 3 (7) | 47 (100) | 51 (98) | 1.0 | <0.01 | <0.01 | 52 (75) | 36 (95) | 0.02 |
| AMK | 20 (14) | 0 (0) | 2 (4) | 18 (35) | <0.01 | <0.01 | 0.50 | 18 (26) | 2 (5) | <0.01 |
| TOB | 64 (45) | 15 (36) | 9 (19) | 40 (77) | <0.01 | <0.01 | 0.08 | 51 (74) | 1 (3) | <0.01 |
| GEN | 46 (33) | 15 (36) | 8 (17) | 23 (44) | <0.01 | 0.40 | 0.04 | 34 (49) | 0 (0) | <0.01 |
| SXT | 69 (49) | 18 (43) | 25 (53) | 26 (50) | 0.75 | 0.49 | 0.33 | 34 (49) | 20 (53) | 0.74 |
TZP, piperacillin/tazobactam; AMC, amoxicillin/clavulanate; CAZ, ceftazidime; FEP, cefepime; NAL, nalidixic acid; CIP, ciprofloxacin; AMK, amikacin; TOB, tobramycin; GEN, gentamicin; SXT, co-trimoxazole.
Values are number of isolates (percentage of the column total).
All ST131 isolates were non-susceptible to cefotaxime and susceptible to cefoxitin, tigecycline, imipenem and meropenem.
Intermediate results were considered non-susceptible.
After using the Bonferroni correction method, P values were considered significant if P ≤ 0.02 for comparison across subclones and P < 0.05 for comparison according to ESBL production. P values that are statistically significant are in bold.
ST131 screening and subclone typing
ST131 screening and subtyping to detect the C/H30 and C2/H30Rx subclones were performed by SNP-based PCR as previously described.9,31 The presence of the M27PP1 marker of the new C1-M27 clade was studied using PCR-based methods, as described previously, in all ST131 isolates and in all CTX-M-27-producing non-ST131 isolates (total n = 150).5,6 The clonal relationship was examined by XbaI-PFGE in all ST131 isolates. PFGE profiles were analysed with Bionumerics version 7.5 (Applied Maths, NV, USA) and a dendrogram was created according to the unweighted pair group method with arithmetic mean based on pairwise Dice similarity coefficients.
O typing and fimH sequencing
O typing was performed by PCR in all ST131 isolates to detect the two most common O types described in this clone.32 All H30 PCR-negative isolates underwent direct sequencing of fimH.33 Isolates with a fimH allele that differed by only one nucleotide in comparison with an established fimH allele reference sequence were considered as belonging to the corresponding fimH allele group.
Statistical analysis
Dichotomous variables were described using frequencies and percentages and were compared using a χ2 or Fisher test, as appropriate. Quantitative variables were described with means and standard deviations or medians and IQRs as appropriate. The criterion for statistical significance was P < 0.05, corrected using the Bonferroni method for multiple comparisons. Data were analysed with STATA version 11 (Stata Statistical Software: Release 11. College Station, TX, USA: StataCorp LP).
Results
ST131 and subclone prevalence
The ST131 detection rate among ESBL-Ec carriers from the four hospitals was 20.5% (141/688), with 16% (46/295) prevalence in Madrid, 18% (31/172) in Utrecht, 23% (31/135) in Berlin and 38% (33/86) in Geneva (P < 0.01).
Subclone typing revealed that although the C/H30 subset comprised the majority (70.2%, 99/141) of the total ST131-ESBL-Ec isolates, all three subsets (non-H30, C1/H30R1 and C2/H30Rx) showed similar prevalence. There was a slight predominance of the C2/H30Rx subclone with 36.9% (52/141) of the isolates, versus the non-H30 subset with 29.8% (42/141) and the C1/H30R1 subclone with 33.3% (47/141) (Figure 1). There were, however, significant geographical differences in subclone prevalence. The C2/H30Rx subclone had the highest prevalence among Madrid isolates (67%, 31/46), whereas the C1/H30R1 subclone predominated in Geneva (49%, 16/33). Conversely, the non-H30 subset represented 48% (15/31) and 42% (13/31) of the ST131 isolates from Utrecht and Berlin, respectively (P < 0.01) (Figure 1).
Prevalence of ST131 subsets among ST131 ESBL-producing E. coli in colonized patients in the four hospitals of the R-GNOSIS-WP5 study.
Of the ST131 isolates screened for the C1-M27 clade marker, 26% (37/141) were positive. Nearly all (95%, 35/37) belonged to the C1/H30R1 subclone. The other positive isolates were two C2/H30Rx isolates. C1-M27 isolates were found in all participating hospitals. None of the nine non-ST131 isolates was positive for the C1-M27 marker.
Of the total ST131 isolates, 75.9% (107/141) and 17% (24/141) corresponded to the O25b and O16 subtypes, respectively. Of the H30-PCR-negative isolates (n = 42), 52% (22/42) encoded allele H41 of the fimH gene, 19% (8/42) were negative for fimH amplification and 29% (12/42) encoded other less common alleles such as H27 (n = 3) and H22 (n = 1) (Figure 2).
XbaI PFGE-based dendrogram of the ST131 ESBL-producing E. coli isolates with O type, fimH alleles, ESBL and subtyping. O type: non, neither O25b nor O26; fimH: NEG, isolate did not amplify for fimH primers; subtyping: subset non-H30, isolates negative for H30 SNP-based PCR and were not H41-O16; A/H41-O16, isolates with allele H41 and O16 O type; C1-M27, isolates positive for the M27PP1 marker; C1-non-M27, isolates positive for H30 SNP-based PCR, negative for both the H30Rx SNP-based PCR and the M27PP1 marker; C2/H30Rx, isolates positive for the H30Rx SNP-based PCR; cluster 1, C/H30; cluster 1A, C1-M27; cluster 2, A/H41-O16. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.
Antibiotic susceptibility and ESBL characterization
Overall, a high proportion of ST131 isolates were non-susceptible to nalidixic acid (90.1%), ciprofloxacin (71.6%), ceftazidime (61.7%), cefepime (54.6%), co-trimoxazole (48.9%) and tobramycin (45.4%) (Table 1). All isolates were non-susceptible to cefotaxime and susceptible to tigecycline, imipenem and meropenem.
Resistance to fluoroquinolones was higher in the C/H30 subset (both C1/H30R1 and C2/H30Rx) than in the non-H30 isolates (nalidixic acid: 98% versus 71%, P < 0.01; ciprofloxacin: 99% versus 7%, P < 0.01). C1/H30R1 isolates were all susceptible to amoxicillin/clavulanate and piperacillin/tazobactam, whereas C2/H30Rx and non-H30 isolates showed a higher resistance prevalence: 35% and 12% for amoxicillin/clavulanate, and 31% and 12% for piperacillin/tazobactam, respectively (C1/H30R1 versus C2/H30Rx: P < 0.01 for both antibiotics; C1/H30R1 versus non-H30: P = 0.02 for both antibiotics).
Similar results were observed for aminoglycosides. C1/H30R1 isolates were more often susceptible to gentamicin and tobramycin than the other ST131 isolates (Table 1). Moreover, C1/H30R1 isolates that belonged to the C1-M27 clade were more often susceptible to all aminoglycosides tested than those that were not included in the clade (91% versus 42%, P < 0.01). Conversely C2/H30Rx isolates were more often resistant to at least two aminoglycosides (77%, 40/52, P < 0.01).
Nearly all (97.2%, 137/141) of the ESBL-Ec-ST131 isolates encoded CTX-M enzymes. The most common enzymes were CTX-M-15 (48.9%, 69/141) and CTX-M-27 (26.9%, 38/141), followed by CTX-M-1 and CTX-M-14 [10.6% (15/142) and 9.9% (14/141), respectively]. Other enzymes detected were SHV-12 (n = 3), TEM-20 (n = 1) and CTX-M-3 (n = 1).
There were differences in ESBL distribution among ST131 subclones. C2/H30Rx isolates mostly encoded CTX-M-15 (88%, 46/52) while C1/H30R1 isolates more often encoded CTX-M-27 (72%, 34/47) (P < 0.01). This association occurred in all four participating hospitals (Figure 3). All CTX-M-27-producing C1/H30R1 isolates belonged to the C1-M27 clade, but also two CTX-M-27-C2/H30Rx and one CTX-M-14-C1/H30R1. Conversely, non-H30 isolates encoded a greater diversity of enzymes, mainly CTX-M-15 (43%, 18/42) and CTX-M-1 (31%, 13/42) (Figure 3). Moreover, non-susceptibility proportions to most antibiotics were higher among isolates encoding CTX-M-15 than those encoding CTX-M-27 (Table 1).
Distribution of ESBL enzymes across ST131 subsets and non-ST131 isolates in each participating hospital of the R-GNOSIS-WP5 study. Enzymes other than CTX-M-15 and -27 included CTX-M-1, CTX-M-3, CTX-M-14, SHV-12 and TEM-20.
To assess the differences in ESBL production between ST131 and non-ST131 isolates, 50 non-ST131 E. coli isolates from Geneva, 50 from Berlin, 48 from Madrid and 57 from Utrecht (total n = 205) were randomly selected from the total ESBL-Ec collection. Among this non-ST131 subset, the most common enzymes were CTX-M-15 (23.4%, 48/205) and CTX-M-14 (22.9%, 47/205) (Table 2). However, the CTX-M-27 ESBL type was only detected in 4.4% of the isolates. CTX-M-14 was the most common ESBL in Madrid and Geneva (42% and 34%, respectively); CTX-M-15 predominated in Utrecht (32%) and CTX-M-1 in Berlin (46%) (P < 0.01).
ESBL-encoding genes among 205 non-ST131 E. coli isolates according to hospital origin
| ESBL group/ESBL | Berlin | Geneva | Madrid | Utrecht | Total isolates |
|---|---|---|---|---|---|
| (n = 50) | (n = 50) | (n = 48) | (n = 57) | (n = 205) | |
| CTX-M (n = 173) | |||||
| CTX-M-15 | 14 (28) | 12 (24) | 4 (8) | 18 (32) | 48 (23.4) |
| CTX-M-1 | 23 (46) | 8 (16) | 4 (8) | 9 (16) | 44 (21.5) |
| CTX-M-3 | 3 (6) | 0 (0) | 1 (2) | 5 (9) | 9 (4.4) |
| CTX-M-55 | 1 (2) | 1 (2) | 2 (4) | 0 (0) | 4 (2) |
| CTX-M-32 | 1 (2) | 0 (0) | 1 (2) | 1 (2) | 3 (1.5) |
| CTX-M-14 | 5 (10) | 17 (34) | 20 (42) | 5 (9) | 47 (22.9) |
| CTX-M-27 | 2 (4) | 3 (6) | 0 (0) | 4 (7) | 9 (4.4) |
| CTX-M-9 | 0 (0) | 0 (0) | 2 (4) | 2 (4) | 4 (2) |
| CTX-M-24 | 0 (0) | 0 (0) | 0 (0) | 1 (2) | 1 (0.5) |
| CTX-M-like | 0 (0) | 2 (4) | 0 (0) | 2 (4) | 4 (2) |
| SHV (n = 22) | |||||
| SHV-12 | 1 (2) | 1 (2) | 13 (27) | 1 (2) | 16 (7.8) |
| SHV (other) | 0 (0) | 1 (2)a | 1 (2)b | 4 (7)c | 6 (2.9) |
| TEM (n = 8) | |||||
| TEM | 0 (0) | 5 (10)d | 0 (0) | 3 (5)e | 8 (3.9) |
| ESBL group/ESBL | Berlin | Geneva | Madrid | Utrecht | Total isolates |
|---|---|---|---|---|---|
| (n = 50) | (n = 50) | (n = 48) | (n = 57) | (n = 205) | |
| CTX-M (n = 173) | |||||
| CTX-M-15 | 14 (28) | 12 (24) | 4 (8) | 18 (32) | 48 (23.4) |
| CTX-M-1 | 23 (46) | 8 (16) | 4 (8) | 9 (16) | 44 (21.5) |
| CTX-M-3 | 3 (6) | 0 (0) | 1 (2) | 5 (9) | 9 (4.4) |
| CTX-M-55 | 1 (2) | 1 (2) | 2 (4) | 0 (0) | 4 (2) |
| CTX-M-32 | 1 (2) | 0 (0) | 1 (2) | 1 (2) | 3 (1.5) |
| CTX-M-14 | 5 (10) | 17 (34) | 20 (42) | 5 (9) | 47 (22.9) |
| CTX-M-27 | 2 (4) | 3 (6) | 0 (0) | 4 (7) | 9 (4.4) |
| CTX-M-9 | 0 (0) | 0 (0) | 2 (4) | 2 (4) | 4 (2) |
| CTX-M-24 | 0 (0) | 0 (0) | 0 (0) | 1 (2) | 1 (0.5) |
| CTX-M-like | 0 (0) | 2 (4) | 0 (0) | 2 (4) | 4 (2) |
| SHV (n = 22) | |||||
| SHV-12 | 1 (2) | 1 (2) | 13 (27) | 1 (2) | 16 (7.8) |
| SHV (other) | 0 (0) | 1 (2)a | 1 (2)b | 4 (7)c | 6 (2.9) |
| TEM (n = 8) | |||||
| TEM | 0 (0) | 5 (10)d | 0 (0) | 3 (5)e | 8 (3.9) |
Values are number of isolates (percentage of the column total). Two enzymes could not be characterized.
SHV-5.
SHV-2a.
Two SHV-2, SHV-11a and SHV-2a.
TEM-52, TEM-214, TEM-78 and two TEM-like.
TEM-52 and two TEM-214.
ESBL-encoding genes among 205 non-ST131 E. coli isolates according to hospital origin
| ESBL group/ESBL | Berlin | Geneva | Madrid | Utrecht | Total isolates |
|---|---|---|---|---|---|
| (n = 50) | (n = 50) | (n = 48) | (n = 57) | (n = 205) | |
| CTX-M (n = 173) | |||||
| CTX-M-15 | 14 (28) | 12 (24) | 4 (8) | 18 (32) | 48 (23.4) |
| CTX-M-1 | 23 (46) | 8 (16) | 4 (8) | 9 (16) | 44 (21.5) |
| CTX-M-3 | 3 (6) | 0 (0) | 1 (2) | 5 (9) | 9 (4.4) |
| CTX-M-55 | 1 (2) | 1 (2) | 2 (4) | 0 (0) | 4 (2) |
| CTX-M-32 | 1 (2) | 0 (0) | 1 (2) | 1 (2) | 3 (1.5) |
| CTX-M-14 | 5 (10) | 17 (34) | 20 (42) | 5 (9) | 47 (22.9) |
| CTX-M-27 | 2 (4) | 3 (6) | 0 (0) | 4 (7) | 9 (4.4) |
| CTX-M-9 | 0 (0) | 0 (0) | 2 (4) | 2 (4) | 4 (2) |
| CTX-M-24 | 0 (0) | 0 (0) | 0 (0) | 1 (2) | 1 (0.5) |
| CTX-M-like | 0 (0) | 2 (4) | 0 (0) | 2 (4) | 4 (2) |
| SHV (n = 22) | |||||
| SHV-12 | 1 (2) | 1 (2) | 13 (27) | 1 (2) | 16 (7.8) |
| SHV (other) | 0 (0) | 1 (2)a | 1 (2)b | 4 (7)c | 6 (2.9) |
| TEM (n = 8) | |||||
| TEM | 0 (0) | 5 (10)d | 0 (0) | 3 (5)e | 8 (3.9) |
| ESBL group/ESBL | Berlin | Geneva | Madrid | Utrecht | Total isolates |
|---|---|---|---|---|---|
| (n = 50) | (n = 50) | (n = 48) | (n = 57) | (n = 205) | |
| CTX-M (n = 173) | |||||
| CTX-M-15 | 14 (28) | 12 (24) | 4 (8) | 18 (32) | 48 (23.4) |
| CTX-M-1 | 23 (46) | 8 (16) | 4 (8) | 9 (16) | 44 (21.5) |
| CTX-M-3 | 3 (6) | 0 (0) | 1 (2) | 5 (9) | 9 (4.4) |
| CTX-M-55 | 1 (2) | 1 (2) | 2 (4) | 0 (0) | 4 (2) |
| CTX-M-32 | 1 (2) | 0 (0) | 1 (2) | 1 (2) | 3 (1.5) |
| CTX-M-14 | 5 (10) | 17 (34) | 20 (42) | 5 (9) | 47 (22.9) |
| CTX-M-27 | 2 (4) | 3 (6) | 0 (0) | 4 (7) | 9 (4.4) |
| CTX-M-9 | 0 (0) | 0 (0) | 2 (4) | 2 (4) | 4 (2) |
| CTX-M-24 | 0 (0) | 0 (0) | 0 (0) | 1 (2) | 1 (0.5) |
| CTX-M-like | 0 (0) | 2 (4) | 0 (0) | 2 (4) | 4 (2) |
| SHV (n = 22) | |||||
| SHV-12 | 1 (2) | 1 (2) | 13 (27) | 1 (2) | 16 (7.8) |
| SHV (other) | 0 (0) | 1 (2)a | 1 (2)b | 4 (7)c | 6 (2.9) |
| TEM (n = 8) | |||||
| TEM | 0 (0) | 5 (10)d | 0 (0) | 3 (5)e | 8 (3.9) |
Values are number of isolates (percentage of the column total). Two enzymes could not be characterized.
SHV-5.
SHV-2a.
Two SHV-2, SHV-11a and SHV-2a.
TEM-52, TEM-214, TEM-78 and two TEM-like.
TEM-52 and two TEM-214.
Clonal analysis
Fingerprinting analysis revealed a cluster (arbitrarily named cluster 1, n = 87) with 72% similarity that grouped most of the C/H30 isolates (87%, 86/99). Nested within cluster 1 was subcluster 1A (n = 37, 80% similarity), which grouped all but one of the isolates belonging to the C1-M27 clade found in our study. See Figure 2.
Conversely, non-H30 isolates were much less related, with the exception of most (82%, 18/22) of the H41-O16 isolates, which grouped in a cluster (cluster 2, n = 23) with 75% similarity.
Discussion
In this study we assessed the prevalence of ST131 and its subsets (non-H30, C1/H30R1 and C2/H30Rx) in a collection of ESBL-Ec from colonized patients in four different European hospitals.
The ST131 prevalence among ESBL-Ec isolates found in this study (20.5%) is lower than the one found in other similar studies, from 26% prevalence found among ESBL-producing isolates from children to 41% among third-generation cephalosporin-resistant E. coli colonizing patients from rehabilitation wards.17,34,35 The differences in ST131 prevalence observed among countries in our study could be related to differences in case mix or other factors such as dietary habits or antibiotic consumption, which may vary between countries.
The C/H30 clade predominated in our study (70%) but all three subclones showed very similar prevalence. However, important differences in subclone prevalence between hospitals were detected. It was particularly striking to observe the predominance of the CTX-M-15-C2/H30Rx and CTX-M-27- C1/H30R1 subclones in Madrid (67%) and Geneva (49%) hospitals, respectively. Although several studies have described the predominance of CTX-M-15-producing-C2/H30Rx isolates among clinical ESBL-Ec from Spain,8,36 to our knowledge this is the first report of the predominance of this subclone among Spanish faecal ESBL-Ec isolates. Curiously, in our study most isolates from Berlin and Utrecht belonged to the non-H30 subset, which is usually a minority subset, particularly among clinical strains.3,37,38
However, the ST131-C1/H30R1 subclone has been increasingly reported in clinical, colonizing and environmental isolates in Europe, particularly in France and Switzerland, especially associated with CTX-M-27 production.39–41 In our study, CTX-M-27 was almost restricted to C1/H30R1 isolates and it was scarcely detected among the non-ST131 subset.
Recently, Matsumura et al.5 described a new ST131 clade, C1-M27, that grouped these CTX-M-27-producing C1/H30R1 strains. Furthermore, this clade was subsequently described in German and French isolates.6,7 Interestingly, we found isolates belonging to this clade in all four participating hospitals, although it was in the Geneva hospital where it predominated. These results represent the first description of this clade in Spain and the Netherlands and confirm the previous reports of German and French isolates. Moreover, PFGE analysis grouped most C1-M27 isolates in a highly clonal cluster. This could suggest a recent introduction of the C1-M27 clade circulating in Europe, as Ghosh et al.7 suggested. Furthermore, these results show a possible shifting of the predominance of CTX-M-9 group enzymes to CTX-M-27 in colonized patients across Europe due to the ST131-C1-M27 clade.
Regarding antibiotic resistance, significant differences were observed between subclones, particularly between C1/H30R1 and C2/H30Rx subclones, despite their phylogenetic relationship. These differences were particularly remarkable in the proportions of resistance to penicillin/β-lactamase inhibitor combinations and to aminoglycosides, which were very low in the C1/H30R1 isolates compared with the C2/H30Rx isolates. Therefore, explanations other than selection by antibiotic pressure should be sought for the emergence of the C1-M27 clade. Patients with extraintestinal infections caused by ESBL-ST131 should not be treated with fluoroquinolones, but aminoglycosides or combinations of penicillins and β-lactamase inhibitors (the latter has already been proposed for the treatment of infections caused by ESBL producers42) could be used depending on the subclone.
A limitation of this study could be related to sampling, with only selected wards in large tertiary hospitals with a mix of community- and hospital-acquired isolates. Moreover, only ESBL-Ec isolates were available for this study and they were obtained from rectal swabs where only one colony was selected for further studies. This could also lead to less abundant populations being missed. This could indicate a detection and reporting bias. However, this study’s aim was primarily descriptive, and it has been able to generate hypotheses about ST131 dissemination dynamics that are being explored in ongoing studies that address these aspects. A strength of our study is the large number of patients included (one of the largest European ESBL-Ec collections studied to date) from very different geographical areas of Europe and the in-depth analysis of the ST131 population to the subclone level, including the newly described C1-M27 clade.
In conclusion, we observed important differences in the prevalence of ESBL-Ec-ST131 colonization among patients admitted to different European hospitals. Moreover, subclone distribution across countries differed in such a way that, together with the association between subclones and particular ESBLs, it led to the predominance of C2/H30Rx-CTX-M-15 in Madrid and C1/H30R1-CTX-M-27 in Geneva, the latter mostly belonging to the new C1-M27 clade. This represents an evolving epidemiology of ESBLs in Europe caused by the pandemic clone ST131 and its subclones and underlines the complex population structure of this clone. Further studies are needed to explain the reasons for the predominance of specific subclones in the different European countries.
Acknowledgements
We thank Mary Harper for English correction of the manuscript.
Other members of the R-GNOSIS Study Group
Marc Desilets, Stéphanie Dul, Fabienne Scherrer-Muller, Benedikt Huttner, Ilker Uçkay, Virginie Prendki and Gesuele Renzi (University of Geneva Hospitals and Faculty of Medicine, Geneva, Switzerland).
Funding
This work was supported by the European Commission, Seventh Framework Program (R-GNOSIS-282512) and the Instituto de Salud Carlos III of Spain Plan Estatal de I+D+I 2013-2016 (grant PI13/02092). The Instituto de Salud Carlos III of Spain also supports the Spanish Network for Research in Infectious Diseases (REIPI) (grants RD12/0015/0004 and RD16/0016/0011). This work was also co-financed by the European Development Regional Fund ‘A Way to Achieve Europe’ (FEDER). I. M. is supported by a research contract from REIPI (grants RD12/0015/0004 and RD16/0016/0011).
Transparency declarations
None to declare.
References
Author notes
Other members are listed in the Acknowledgements section.
