Introduction

Escherichia coli is known to be a part of the natural human intestinal flora but can also cause intestinal or extraintestinal infections1. In particular, E. coli is a frequent isolate in adult patients with bacteremia2. The most common sources for E. coli bloodstream infections are urinary tract infections, abdominal sepsis and pneumonia3,4.

Antimicrobial treatment of such infections is nowadays complicated by the fact that the prevalence of Enterobacteriales producing extended-spectrum β-lactamases (ESBLs) that came to attention first in the 1980s is on the rise ever since5. The first extended-spectrum β-lactamases described were of the TEM- and SHV-type and were primarily isolated from E. coli and K. pneumoniae6. CTX-M-Group ESBLs were first isolated from E. coli and reported in Germany in 19897. By now, they have become the most common ESBLs worldwide with CTX-M-15 and CTX-M-14 enzymes being most frequently found in human specimens globally8,9.

In this context and considering the rising antimicrobial resistance of E. coli due to extended-spectrum β-lactamases, this study aimed at the molecular characterization of ESBLs detected in E. coli strains obtained from blood cultures of patients of the University Hospital of Leipzig (UKL), Germany. Furthermore, our study provides epidemiological data and resistance patterns of the isolated strains. In order to find indications of an endogenous focus of infection for the E. coli bacteremia, the laboratory database was searched for corresponding stool samples with evidence of ESBL producing E. coli for each patient. Whenever available, resistance patterns from both strains (blood culture and stool sample) of the individual patients were compared.

Materials and methods

E. coli strains

The E. coli strains investigated with this study were all collected between 2015 and 2018 from patients of the UKL. They were either blood culture isolates or strains that were recovered during a continuous risk-adapted screening of stool samples for multiple drug resistant strains.

As part of this screening, stool samples and rectal swabs are examined for the presence of carbapenemase- and β-lactamase-producing bacteria employing the chromogenic media CHROMagar ESBL and CHROMagarTM ESBL (Mast Diagnostica, Germany) as initial test procedure. The stool and rectal samples were collected as part of this UKL screening program and were neither collected by any of the authors nor for the purposes of this study in the first place. Therefore, there were no human participants involved in this study. All strains recovered were identified with MALDI-TOF (BioMerieux, France). Altogether, 117 strains from blood cultures (ECB) with phenotypic evidence of ESBL production and 112 strains recovered from screening (ECS) were considered for further analysis. For 94 of the bacteremic patients ESBL positive stool screening cultures were available. Those as well as all blood culture isolates were further analyzed with this study.

Susceptibility test

All E. coli isolates were submitted to broth microdilution performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Minimum inhibitory concentrations (MICs) for the following antibiotics were determined: ampicillin, ampicillin/sulbactam, piperacillin, piperacillin/tazobactam, ceftazidime, cefotaxime, cefuroxime, aztreonam, imipenem, meropenem, amikacin, gentamicin, tobramycin, ciprofloxacin, levofloxacin, moxifloxacin, colistin, fosfomycin, trimethoprim/sulfamethoxazole and tigecycline. E. coli ATCC 25922 was used as quality control strain.

ESBL detection

Phenotypic testing for ESBL production was performed for all isolates with a MIC of ≥ 1 mg/l for either cefotaxime, ceftazidime or aztreonam. For this purpose, a gradient test system using cefotaxime/clavulanic acid and ceftazidime/clavulanic acid test strips (Etest®, bioMérieux, France) was employed. According to the recommendations of the manufacturer, individual bacterial colonies were transferred to 0.5% NaCl solution until a MC Farland standard of 0.5 was obtained. The suspension was then streaked onto Müller-Hinton agar, the strips were applied, and the tests incubated for 18 to 24 h at 37 °C. All isolates positive in the gradient test were then analyzed for the presence of extended-spectrum beta-lactamase genes CTX-M-14 and CTX-M-15 and the plasmid-mediated ampC gene CMY-2.

ESBL gene detection

Single colonies of individual E. coli isolates were suspended in sterile Protease-, RNase-, and DNase-free water and then submitted to automated DNA extraction by the MagNa Pure 96 system (Roche Diagnostics, France). For the purpose of identifying the target genes (CMY-2, CTX-M-14 und CTX-M-15), the commercial PCR Mix 1 of Streck ARM-D® Kit (Streck, USA), β-Lactamase + MasterMix with fluorescence-marked DNA-probes was used. Real-time amplifications of the target genes were performed by QIAGEN Rotor-Gene Q MDx Thermocycler (QIAGEN, Thermo Fisher Scientific, USA) in accordance with the manufacturer’s instructions.

Data collection and evaluation

Antibiograms and epidemiological data were collected by laboratory IT Hybase (epiNET AG, Germany) and LabCenter (i-solutions health GmbH, Germany). HyBase® as well as Microsoft Excel were used for the following evaluations.

Results

In total, 130 ESBL producing E. coli isolates were preserved and included in the further investigation. Since 13 isolates turned out to be duplicates of the same patient, those were not included in any statistical evaluation leaving 117 isolates for further analysis.

ECB-data

During the time period of 2015 to 2018, 159,074 blood cultures were submitted for microbiological evaluation for patients of the UKL. Out of these, 26,427 yielded bacterial growths (positivity rate: 16.6%). Since multiple cultures were submitted for individual patients with positive culture (n = 9910), a corresponding diagnosis of bacteremia was made in 6.2% of all cases. Altogether, in 1020 cases a bacteremia with E. coli was documented (10.3%) including 153 cases where ESBL producing strains were detected (15%). Thus, out of 9910 patients suffering from bacteremia, ESBL producing E. coli strains were documented in 1.5% of the cases. In total, 130 ESBL producing E. coli isolates were collected and included in the investigations. Among these, 13 isolates were duplicates of individual patients, so they were eliminated from any statistical evaluation leaving 117 isolates for further interpretation.

Susceptibility data for the 117 ECBs are given with Table 1 and MIC distributions are depicted in Fig. 1. The results show that aside from the ß-lactam resistance, 74.4% (87/117) were also resistant to ciprofloxacin.

Table 1 Antimicrobial susceptibility of ECB.
Figure 1
figure 1

MIC distributions of ECB.

In 109 out of the 117 isolates (93.1%), there was at least one of the investigated genes detected, genotype CTX-M-15 being the most prevalent (78/117, 66.7%), followed by genotype CTX-M-14 (30/117, 25.6%) and CMY-2 (4/117, 3.4%). Three isolates (2.6%) were tested positive for the possession of two resistance genes simultaneously (one for CMY-2 and CTX-M-15, one for CTX-M-14 and CMY-2, one for CTX-M-15 and CTX-M-14). Finally, 8 isolates (8/117, 6.8%) were negative for the tested genes. The distribution of the examined genes is shown in Fig. 2.

Figure 2
figure 2

Distribution of detected antimicrobial resistance genes (ESBL producing and plasmid-mediated ampC genes).

Susceptibility testing for CTX-M-15-positive isolates showed that > 97% of the samples (76/78) were resistant to cefotaxime, whereas only two of the tested isolates were susceptible to cefotaxime (2/78, 2.6%). There was a resistance to ceftazidime in > 61% of the CTX-M-15-positive isolates (48/78).

All CTX-M-14-positive isolates (30/30) were resistant to cefotaxime with a MIC > 4 mg/l, whereas 60% were also not susceptible to ceftazidime (18/30). Beside a resistance to ß-lactam antibiotics, a high proportion of strains with an increased ciprofloxacin MIC level (≥ 0.5 mg/l) was detected (89/117, 76.1%), even 74.4% (87/117) with a MIC Level ≥ 1 mg/l.

ECS-data

Of the 117 bacteremic patients included in this study, 112 had submitted stool samples prior to the time of bacteremia. 94 of these specimens tested positive for ESBL producing E. coli (94/112, 83.9%). Retrospectively, a total of 79 (79/94, 84%) E. coli strains found in the stool samples matched with the respective patient’s blood culture isolate phenotypically (MALDI-TOF, antibiogram). The comparison between the stool samples and blood cultures that tested positive for ESBL is shown in Table 2. Considering the ciprofloxacin resistance, 55.4% (62/112) of the patients were colonized with an ESBL positive and ciprofloxacin resistant strain. 15 (15/94, 16%) of the isolates showed no phenotypic match of blood culture isolate and respective stool sample.

Table 2 Comparison between ESBL positive stool samples and blood cultures.

Infection due to ESBL-E. coli

Table 3 shows the epidemiology of multi-resistant E. coli during the investigation period. Between 2015 and 2018 there were 12,346 cases of infection by invasive E. coli, id est isolations of E. coli excluding rectal swabs and stool samples. Overall, 13.1% (1614/12,346) were caused by ESBL-producing isolates, while 8.0% (988/12,346) of these were caused by ESBL-producing strains with additional ciprofloxacin resistance; within ESBL-positive isolates, the proportion of ciprofloxacin resistance (61.2%, 988/1614) was higher than the proportion of ciprofloxacin susceptibility significantly. Further, 1020 cases of bloodstream infections by E. coli were detected during the investigation period. Among the 117 cases included in this study, 74.4% of the isolates were showing a resistance to ciprofloxacin, piperacillin as well as ceftazidime or cefotaxime while being susceptible to imipenem/ meropenem.

Table 3 E. coli epidemiology from 2015 to 2018.

Discussion

Our study focused on patients with bacteremia caused by ESBL producing E. coli strains. The dominant ESBL genotype in our isolates was CTX-M-15 (78/117, 66.7%), followed by genotype CTX-M-14 (30/117, 25.6%) and AmpC CMY-2 (4/117, 3.4%). These results underscore earlier reports that worldwide CTX-M-15 is the most prevalent ESBL found in Enterobacteriaceae10. They are also in line with a previous study in Germany addressing the molecular epidemiology of ESBL producing Enterobacteriaceae among 156 nursing home residents in Bavaria11. Ha YE et al. analyzed bloodstream infections due to E. coli in cancer patients of the Samsung Medical Center in Seoul—they found CTX-M-14 (37.7%), CTX-M-15 (26.1%) or both CTX-M-15 and CTX-M-14 (10.1%) in their isolates12. However, investigators at Nara Medical University, a tertiary care hospital in Japan, reporting on cases of bacteremia with ESBL producing E. coli, found that ESBL-EC CTX-M-27 (33.3%) and CTX-M-14 (30%) were the most prevalent genes13.

Several studies assessed the fecal carriage rate of ESBL producing E. coli, i.e. the rate was reported to be 14.7% in Bavarian nursing home residents and 6.3% in the healthy population11. The prevalence of ESBL producing E. coli in fecal samples of 650 inpatients reported in a study in China, Beijing, was 25.7%14, while a study from Ankara, Turkey, reported the ESBL carriage rate at 34.3% for 1402 outpatients with E. coli strains being the most frequent ones15.

Since an extensive screening program for multidrug resistant microorganisms is in place in our hospital, we were also able to address the fecal carriage of ESBL producing E. coli strains of the patients with bloodstream infections of our study. Interestingly, of the 117 bacteremic patients included in this study, 112 had submitted stool samples prior to time of bacteremia and 94 of these specimens yielded ESBL producing E. coli (94/112, 83.9%). 79 of the E. coli strains found in the stool samples matched with the respective patient’s blood culture isolate phenotypically, suggesting a high incidence of endogenous infections. Reddy et al. reported earlier on an infection control program addressing ESBL producing Enterobacteriales—they found that out of 413 colonized patients 8.5% developed a subsequent bloodstream infection, amounting to 34.3% of all bloodstream infections by ESBL producing Enterobacteriales16. Our data emphasize once more the value of risk adapted screening and—in case of an infection—the necessity to select an empiric therapy accordingly. Moreover, the epidemiology of the underlying ESBL genes should be surveyed on a regular basis and analyzed together with the respective susceptibility data.

With this study E. coli and three genes were addressed, only. This is an obvious shortcoming and we cannot rule out that other ESBL genes were also present in our isolates. However, in 109 out of the 117 isolates (93.1%), there was at least one of the investigated genes detected and the results are in line with the epidemiology reported globally.

Conclusion

The molecular investigation of ESBL producing E. coli collected from blood cultures from patients at the University Hospital of Leipzig, Germany, between 2015 and 2018 showed a distribution of resistance genes that was in accordance with recent studies in Germany as well as worldwide with CTX-M-15 being the most prevalent genotype in our study.

The results of our study provide possible indications of an endogenous focus of infection for an E. coli bacteremia and, therefore, emphasize the importance of screening programs for high-risk patients.