Molecular characterization and drug resistance of Escherichia coli strains isolated from urine from long-term care facility residents in Cracow, Poland

Background

The high incidence of infections and high rates of hospitalization, together with institutionalized long-term care (LTC), highlights the important role of management to the decrease antibacterial susceptibility of organisms cultured from elderly patients. The reservoir for these microorganisms may be the bacterial flora in patients receiving long-term antimicrobial therapy or health care personnel who are colonized with resistant strains or infected with the same pathogens. Multidrug-resistant gram-negative rods (MDR-GNR) are rapidly spreading throughout LTCFs around the world, and antibiotics are also commonly prescribed for older adults residing in LTCFs. There are a number of risks associated with the inappropriate use of antibiotics. There is risk of development of multi-drug antibiotic resistance after exposure [1,2] and risk of drug-related adverse effects. Furthermore, the increased use of antibiotics in LTCFs results in significant costs [3]. The rapid spread of antimicrobial-resistant bacteria through health care institutions is considered a serious medical and public health issue [4]. The intensive use of antibiotics promotes a relatively high prevalence of MDR-GNR and pathogens able to produce extended-spectrum β-lactamases (ESBL) in LTCFs [5,6]. LTCFs have older populations with increased prevalence of resistant bacteria isolated from urinary tract infections (UTI). Resistant isolates are more frequent in long-term care populations than in the general population [5,6]. Preceding colonization with such microorganisms is a risk factor for the development of infection, but the presence of virulence factors in the pathogens is equally important. A wide range of virulence factors (VFs) and a persistence of multidrug resistance can make the treatment of infections challenging. VFs for Escherichia coli are widely known. They are essential for the interaction between E. coli and its host, as they facilitate colonization, proliferation, and the transition from uncomplicated to severe infections. Knowledge of which VFs are prevalent in specific clinical situations and populations is important for the targeted prevention of E. coli infections, as revealed by other epidemiological studies [7–9]. Virulence and resistance may influence the outcome of infections. Older populations in long-term care facilities have an increased prevalence of resistant bacteria isolated from UTIs.

Until now, the prevalence of ESBL-positive strains and drug resistance among LTCF-residents in Poland has not been studied. For this reason, the aim of this study was to assess the prevalence of MDR-GNR and ESBL pathogens isolated from asymptomatic bacteriuria and UTIs. The second purpose of this study was to look for a relationship between the phylogeny, antimicrobial resistance, and virulence among E. coli isolates from bacteriuria.

There have been no epidemiological descriptions of the elderly population in Poland. This has resulted in an inability to design feasible infection prevention programs. The proposed methods will be the basis for the development of recommendations for empirical treatment of LTCF residents. The point prevalence study was conducted for cases of asymptomatic bacteriuria and a continuing study is being conducted to measure the rate of infections with ESBL etiology.

Material and Methods

PART I. POINT PREVALENCE STUDY, PPS:

A 1-day prevalence survey (point prevalence study [PPS]) was carried out in October 2009 in 3 LTCFs in Krakow, Poland: 2 residential homes and 1 nursing home. A resident was defined as a person staying in an LTCF during the study day for at least 48 h. A control patient was a person staying in home care (HC), but requiring home visits of a physician. In a nursing home (NH), residents need 24 h/day medical or skilled nurse supervision and intensive health care is provided. In the residential home (RH), residents are unable to live independently and require supervision or assistance with activities of daily living. The study protocol was approved by the Ethics Committee of Jagiellonian University (KBET/57/B/2008), conducted in accordance with the Declaration of Helsinki, and carefully explained to the participants, who then gave their written informed consent. Medical charts were reviewed for patient demographic data, comorbidities, and residency in an LTCF. Medical documentation of residents was analyzed for the presence of chronic diseases and other medical problems. Barthel Index values and Katz Index were obtained from all studied participants. The Barthel Index (BI) is a 10-item measure of disability based on daily activities (e.g., bathing, transfer, dressing, feeding, mobility, stairs, using the toilet, and grooming) [10]. The score corresponds to the sum of all the points obtained, and can range from 0 to 100 points. Lower scores are associated with worse functional status – elderly patients with scores from 0 to 20 are considered to be totally dependent [10]. The Katz Index of Independence in Activities of Daily Living is an instrument used to assess functional status as a measurement of the patient’s ability to perform activities of daily living independently [11]. The Index ranks the adequacy of performance in 6 functions: bathing, dressing, toileting, transferring, continence, and feeding. Patients are scored for independence in each of these functions. A score of 6 indicates full function, 4 indicates moderate impairment, and 2 or less indicates severe functional impairment [11]. Physical dependence (mobility) was classified according to a 5-point scale (1 – independent, 2 – independent with falls, 3 – limitations in mobility, 4 – confined to bed, self-changing body position, 5 – confined to bed, dependent). Data on hospital exposure longer than 7 days in the 3 months preceding enrollment and antibiotics taken in the last month were collected. Infections were defined according to the McGeer criteria and were detected by trained health personnel in cooperation with a project worker [12].

PART II. CONTINUING STUDY:

Continuing prospective infection control was performed between 1 December 2009 and 30 November 2010, using the standard McGeer definition protocol [12]. Cases of infections were registered by a nurse and accompanying physician. Urine samples were taken only from residents when the symptoms of infections occurred. In the period between enrollment and each follow-up, data on potential factors that would increase the likelihood of MDRO acquisition were also collected (ie, antibiotic and hospital exposure, and presence of infections).

The relationship between types of care, socio-demographic characteristics, probability, and epidemiology of ESBL were analyzed with 2 main groups of statistical techniques. If the numerical parameters (e.g., age, length of stay) were compared with the nominal characteristics (e.g., type of care, form of infection), a simple ANOVA test was used. If the distribution of numerical characteristics did not fit the normal distribution, the nonparametric alternative to ANOVA, the Kruskal-Wallis test, was used. Chi-square (χ²) frequency tests and a likelihood ratio were used for the contingency of nominal characteristics. P values <0.05 were considered statistically significant. All analyses were performed using the SAS JMP 7.01 package (JMP^®, Version 7. SAS Institute Inc., Cary, NC, 1989–2007).

Detailed information about epidemiology of infections in the point prevalence study and the continuing study among included LTCFs has been described [13].

BACTERIAL ISOLATES: Samples of urine were routinely collected (clean-catch method) and cultured by standard quantitative methods at the beginning of the study (PPS). When symptoms of UTI in any enrolled resident occurred throughout the study (continuing study), urine samples were also taken. Isolates were identified using polymerase chain reaction (PCR) with species-specific primers [14,15] or phenotypic methods (API ID 32E,BioMerieux). Strains belonging to the Enterobacteriaceae family were further tested.

ANTIMICROBIAL SUSCEPTIBILITY:

All strains belonging to the Enterobacteriaceae family were tested using disk diffusion antimicrobial susceptibility methods on Mueller-Hinton agar plates according to current guidelines of the European Committee on Antimicrobial Susceptibility Testing (Clinical breakpoints tables v. 1.3; http://www.eucast.org v.1.3). Antibiotics used in this study belonged to 6 chemical groups. Antimicrobials included ampicillin (AMP – 10 μg), piperacillin/tazobactam (TZP – 30 μg/6 μg), ceftazidime (CAZ – 10 μg), cefotaxime (CTX – 5 μg), cefepime (FEP – 30 μg), cefuroxime (30 μg)), imipenem (IPM – 10 μg), aztreonam (ATM – 30 μg), ciprofloxacin (CIP – 5 μg), gentamicin (CN – 10 μg), trimethoprim/sulfamethoxazole (SXT – 25 μg), and nitrofurantoin (F – 100 μg).

ESBL activity was detected with a modified double-disk synergy test (DDST) using a combination of cefotaxime (5 μg), ceftazidime (10 μg), cefepime (30 μg), and aztreonam (30 μg) disks, placed 20 mm apart around a disk containing amoxicillin-clavulanate (20 μg/10 μg) [16]. Enhancement of the inhibition zone toward the amoxicillin-clavulanate disks was taken as presumptive evidence of ESBL production. All disks were obtained from Oxoid, Ltd. (Basingstoke, UK). Each isolate was classified as susceptible, intermediate, or resistant using the European Committee on Antimicrobial Susceptibility Testing breakpoints. Criteria for multidrug resistance were defined as resistance or intermediate resistance to 3 or more antimicrobial groups (co-resistance) or 4 or more of the antimicrobials [17].

POLYMERASE CHAIN REACTION (PCR) SCREENING FOR EXTENDED-SPECTRUM β-LACTAMASE GENES AND SEQUENCING OF BLA GENES: All isolates with ESBL activity (according to DDST) were screened with multiplex-PCR for the presence of blaCTX-M, blaSHV, and blaTEM genes using previously published primers [18]. Bacterial DNA was extracted by the boiling method (10 minutes at 100°C). Relevant positive and negative controls were included in each PCR run. The reaction mixture (25 μl) was amplified using the following conditions: 15 min at 95°C, 30 cycles of 30 s at 94°C, 30 s at 55°C, and 2 min at 72°C, with a final extension of 10 min at 72°C. The amplicon sizes were 445, 593, and 747 bp for blaTEM, blaCTX-M, and blaSHV, respectively. Bands were visualized using the UVP GelDocIT Imaging System after 1.5% TAE-agarose electrophoresis (90 min, 100 mV) with ethidium bromide (BioRad). GeneRuler DNA-ladder 100bp (Fermentas) was used as a size marker. The strains with PCR amplicons positive for any of the previously described bla were selected for sequencing. PCR products were purified using a NucleoSpin extract II kit (Macherey-Nagel). Sequencing was performed by an outsourced company (Genomed, Warsaw, Poland). The nucleotide sequences were analyzed with the software available at the National Center of Biotechnology Information (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

VIRULENCE FACTOR SCREENING: E. coli isolates from bacteriuria were checked for the presence of selected virulence genes usually associated with the E. coli strains responsible for extraintestinal infections: papC and papEF (P fimbriae), fimH (type 1 pili), hlyA (hemolysin), iutA (aerobactin), and sfa (S and F1C fimbriae) using PCR with previously published primers [7]. Each VF gene was amplified in a total volume of 25 μl containing 2× Master Mix (A&A Biotechnology, Poland), 0.5 μM primer and DNA. A negative control (containing the same mixture, but with water instead of DNA) was included in each PCR run. The amplification products were separated by electrophoresis in 1.5% agarose gel and visualized using the UVP GelDocIT Imaging System. GeneRuler DNA-ladder 100bp (Fermentas) was used as a size marker. The results were considered positive if the amplification product was of the expected molecular size (papC – 200 bp, papEF – 326bp, fimH – 508bp, hlyA – 1177bp, iutA – 300bp, and sfa – 410bp). The strains were classified according to the Escherichia coli Reference Collection (ECOR) system by the use of the rapid phylogenetic grouping technique described by Clermont et al. [19]. This method is based on a multiplex PCR involving the amplification of 2 genes (chuA and yjaA) and of an anonymous fragment of DNA from E. coli (TspE4C2). The results are interpreting as follows: chuA and yjaA positive indicated group B2; chuA positive and yjaA negative indicated group D; chuA negative and TspE4C2 positive indicated group B1; and chuA negative and TspE4C2 negative indicated group A. Differences were tested using the χ2 test.

Results

The study was conducted in a group of 217 persons (193 residents of LTCFs and 24 from HC as a control group). Of these residents, 86 (39.6%) stayed at RHs, and 107 stayed at NHs. The studied sample corresponded to 2.6% of the total LTCF population in Poland in 2010 [20]. Among the 217 patients participating in this research, 1 patient quit being in a LTCF, 1 was moved to another facility, and 31 patients died during the study.

The population studied was heterogeneous and significantly diverse with regards to: incidence of hospitalization before study, body weight, problems with maintaining personal hygiene (expressed as urinary/stool incontinence and the Katz Index), and expressed necessity of care (Barthel Index, Table 1). The mean value of the Barthel Index was 75.4 in HC (SD: 27.7, 95% CI: 63.7–87.1), 75.6 in RH (SD: 34.5, 95% CI: 68.2–83.0), and 19.5 in NHs (SD: 17.4, 95% CI: 16.1–22.9). In NHs none of the residents had BI higher than 76 points, and 66.7% of HC patients and 69.8% of RH residents had scores above 76 points. The mean value of Katz Index was 4.4 for HC patients (SD: 2.2, 95% CI: 3.4–5.3), 4.7 for RH residents (SD: 2.1, 95% CI: 4.2–5.1), and 1.3 for NH residents (SD: 1.6, 95% CI: 0.9–1.6). Considerably older patients (aged ≥80) stayed in HC, while younger (age <80) patients were more likely to be residents of RHs. The average age for the population (63.2% female) was 76.2 years in RH (standard deviation SD: 10.5, 95% CI: 73.9–78.4), 76.8 years in NHs (SD: 11.1, 95% CI: 74.6–78.9), and 82.9 in HC (SD: 9.9, 95% CI: 78.7–87.0). The average length of LTCF stay was 6.5 years (SD 5.9; 95%CI 4.1–5.9). Differences were also observed in the physical activity of residents – NH residents were more likely to have limited ability to walk independently. The mean mobility was 2.4 for HC patients (SD: 1.6, 95% CI: 1.7–3.1), 1.9 for RH residents (SD: 1.2, 95% CI: 1.7–2.2), and 3.7 for NH residents (SD: 1.1, 95% CI: 3.5–3.9). Independent mobility (with no restrictions) was characteristic for 12 (50.0%) HC patients, 48 (55.8%) RH residents, and 5 (4.7%) NH residents. There were 5 (20.8%) HC residents who were confined to bed and dependent, 4 (4.7%) in RH, and 33 (30.8%) in NHs.

The general prevalence rate of infections was 22.1/100 residents and general incidence density rate was 2.7/1000 residents per day for both groups (residents and home care patients). The general prevalence rate of ESBL was 4.6/100 in residents with bacteriuria and 0.0/100 in the control group. The incidence density of ESBL-producing isolates was 0.3/1000 residents per day (only in NHs and RH). Of the residents, 22 were taking antibiotics in the month preceding PPS.

Gram-negative rods were identified in 101 cultures of urine: 9 from UTIs and 92 from asymptomatic bacteriuria. Bacteriuria was found in 82 (42.5%) residents. The most commonly isolated pathogen was E. coli (58.4%, n=59, including infections). Other isolated species included: Klebsiella pneumonia (13.9%, n=14), Proteus spp. (12.9%, n=13) and others. During the PPS, 1 case of infection was registered.

Accordingly, 60 isolates were ampicillin resistant (58.7% from bacteriuria and 75.0% from UTI) (Figure 1.). Aminoglycoside resistance was present in 11 isolates (10.9%). Resistance to trimethoprim/sulfamethoxazole was identified in 45 isolates (43.5% from bacteriuria and 71.4% UTI). A significant proportion of isolates were resistant to fluoroquinolones, and 33 isolates were resistant to ciprofloxacin (32.7%). The prevalence of pathogens resistant to more than 4 antimicrobials was 16.3% among asymptomatic bacteriuria samples and 33.3% among UTIs. The most common co-resistance pattern was gentamycin and trimethoprim/sulfamethoxazole, present among 9.1% of isolates.

Of the isolates, 14 had ESBL activity (13.9%). ESBL was most prevalent in E. coli (n=6) and Klebsiella spp. (n=6), but there was 1 Citrobacter spp. and 1 Enterobacter aerogenes that were also ESBL-positive. There were no ESBL-positive strains among Proteus spp., Morganella morgani, Serratia spp., or Providencia spp. Asymptomatic bacteriuria was not a risk factor for the occurrence of an ESBL-positive strain (p=0.09). Among residents who were confined to bed, the frequency of isolation of ESBL-positive strains was 3 times higher than in residents able to walk independently. The factors independently associated with ESBL were urinary tract infections (OR 5.4, P=0.0012), urinary and/or fecal incontinence (OR 13.5, P<0.001), being wheelchair-dependent or being confined to bed (OR 6.2, P<0.001), low value on the Barthel (OR 3.6, P<0.001) and Katz Indexes (OR 2.8, P<0.015), and prior hospitalization (OR 3.1, P<0.045) (Table 2). The mortality of residents with ESBL-positive strains was significantly higher (OR 4.1, P<0.001). Any antibiotic used in the preceding month also remained a significant risk factor (OR 3.2, P=0.03) associated with ESBLs.

From the group of LTCF resident isolates, PCR experiments performed on bacterial DNA with primers specific for 3 bla genes showed that blaCTX-M was present in 9 strains (75.0%), blaTEM in 9 strains (75.0%), and blaSHV in 1 (7.1%). Among the isolates, 9 (64.3%) were characterized by the presence of a single β-lactamase gene, whereas 5 (35.7%) had a combination of 2 different bla genes (blaCTX-M and blaTEM). Furthermore, sequencing of the PCR products revealed the presence of 8 blaCTX-M-15 (92.9%) and 1 blaCTX-M-3 (7.1%). The only blaCTX-M-3 β-lactamase was identified in Klebsiella oxytoca, in a sample taken from a case of asymptomatic bacteriuria. All blaTEM were identified as blaTEM-1 and 1 blaSHV as blaSHV-12.

Among E. coli isolates, 13 clustered in the ECOR group A, 2 clustered in the B1 group, 31 in the B2 group, and 10 in the D group (Table 3). Of these, 4 ESBL-positive strains belonged to the B2 group, 1 to the D group, and 1 to the A group. VF genes were detected in strains from all of the ECOR groups, and no relationship between the number of VF genes and the ECOR group was found. The most prevalent virulence factor was fimH, which occurred in 46 isolates (82.1%). Less frequent was iutA, which was found in 32 samples (57.1%). The potential ability of hemolysin production (expressed by the presence of hlyA gene) was characteristic for 4 strains (7.1%) and 3 of these belonged to the B2 group. The papC gene was found in 16 isolates (28.6%), papEF in 6 (10.7%), and sfa/foc was found in 13 (23.2%) strains. E. coli able to produce ESBL possessed up to 2 different virulence factors (median: 1.5, range: 0–2), whereas non-ESBL E. coli possessed up to 6 (median: 2, range: 0–6). Isolates belonging to the B2 group were more likely to be resistant to fluoroquinolones than those from the A or D group (41.9%, 23.1%, and 30.0%, respectively).

Discussion

The presented results are derived from the first Polish study concerning urinary tract infections among long-term care facility residents. The study was performed using 2 independent approaches: a prevalence study (PPS) and an incidence study. These kinds of research programs are carried out in Europe and worldwide [21]

The population of Polish NH and RH residents was slightly different from the population of institutionalized elderly in other countries. The mean age of such residents in Italy is 81 years [22] and 83 in Germany [23]. In Norwegian LTCFs, more that 78% of residents are 81 years or older [24]. The average age of Polish residents was about 5 years younger. Furthermore, the participation of residents confined to bed and residents with low Barthel and Katz Index scores was relatively lower compared to Norwegian or Italian surveys [25]. The use of invasive diagnostic and therapeutic procedures (e.g., catheterization) was also lower in this study than in Italy (18.4% vs. 60.6%). The published literature on the epidemiology of ESBLs in LTCFs is not conclusive; for instance, the incidence density of ESBL-positive Enterobacteriaceae in our study was not comparable with data from a Swiss tertiary care hospital (0.7/1000 patients per day) [26]. This suggests the need for further multicentre studies involving a larger number of residents over the spread of the studied group of microorganisms.

The most commonly isolated pathogen was E. coli (58.4%, n=59, including infections), similarly to other studies, for hospital patients, patients with CAUTI, and outpatients [27].

MDRO are a serious problem in health care institutions such as hospitals and LTCFs. Resistance determined by ESBL-synthesis is one of the most important resistance mechanisms. Patients colonized with ESBL-positive strains and MDRO are considered as reservoirs and are potential sources of infection. It is important to note that colonization often precedes an infection, and about one-fourth of colonized residents may develop an infection with the same factor [17,28]. The prevalence of bacteriuria in residents without chronic long-term indwelling catheters reaches 25–50% for women and 15–40% for men [5], and approximately 40% of nosocomial infections originate in the urinary tract [29]. In this study, bacteriuria was found in 42.5% of residents, and infection developed in 4.6%.

In this study, residents infected with ESBL-positive strains were 2.6 times more frequently transferred from LTCFs to hospitals and treated under hospital conditions. Their attendance at hospital units requires special rules for isolation, common for patients with MDROs (contact isolation). The genes responsible for drug resistance are frequently located on plasmids, which contributes to their fast spread [30]. This study identified a relatively high prevalence of ESBL-positive strains among residents of LTCFs. Another Polish survey, conducted in 13 Polish hospitals, found a 11.1% prevalence of ESBL-positive pathogens [31], which is consistent with our study. The identification of ESBL types revealed the very common occurrence of CTX-M-15 β-lactamase, as far as TEM-1, whereas SHV appeared rarely (only 1 case). The rapid increase in the prevalence of CTX-M producers cannot be questioned and it is the most prevalent ESBL in other countries [32].

Predictors of ESBL-positive strains found by other authors include permanent urinary catheter, pressure sores, and any recent invasive procedures [33,34]. Other strong risk factors that increased the likelihood of ESBL, like being wheelchair-dependent or confined to bed, and low scores on the Barthel or Katz Indexes, have not yet been investigated. Our studies indicate the significance of this element in a long-term care.

Resistance to gentamicin and trimethoprim/sulfamethoxazole frequently occurs among ESBL-positive isolates [17,30,35–37]. Our study confirmed that 35.7% of ESBL-positive strains were resistant to both of these antibiotics. The majority of the studied isolates were resistant to commonly prescribed antibiotics, including ampicillin, trimethoprim/sulfamethoxazole, and ciprofloxacin. This data raises serious concern about the therapeutic options available for physicians treating LTCF residents. The prevalence of pathogens resistant to more than 4 antimicrobials was 16.2%, which is less than in the O’Fallon report from Boston LTCFs [17]. Resistance to piperacillin/tazobactam, which are counted as broad-spectrum antimicrobials, was very low (1.4%). Piperacillin-tazobactam was the most active agent for out-patient urinary isolates in Turkey [37].

It has been shown that the number of VFs is proportional to its pathogenic potential [38]. FimH protein recognizes its receptor on uroplakin, which is disseminated on the surface of uroepithelium [9]. It acts as an adhesin but also as an invasin, because it is necessary to initiate the response signalling pathways of the host, leading to internalization of E. coli[9,39]. Its high frequency shows that it is important in the early stage of urinary tract infection [9]. The prevalence of the fimH gene, which is an adhesin important in the early phase of UTI, was lower than in studies conducted by Narciso et al. [9]. P fimbriae are the second common virulence factor of uropathogenic Escherichia coli, which plays an important role in the pathogenesis of ascending UTIs and pyelonephritis in humans [40,41]. In our study we show that E. coli isolates from bacteriuria are very often equipped in such fimbriae, as the papC gene was found in 28.6% isolates. S fimbriae and F1C fimbriae are implicated in the process of UTI as they show binding to epithelial and endothelial cells [40,42]. The S fimbriae are also associated with E. coli strains that cause sepsis, meningitis, and ascending UTIs [40]. Sfa/foc was found in 23.2% of isolates, which is also a large proportion of the isolates. The reason for the lower prevalence of hlyA in this population than in uroseptic strains could be that those strains came from asymptomatic bacteriuria cases [7]. The studied strains did not possess high levels of virulence, as the selection may favor strains of intermediate resistance.

Conclusions

This study shows that the prevalence of ESBL-producing rods isolated from urine samples from LTCF residents was lower then in other similar studies published recently. The findings derived from this study emphasize the need for further research on epidemiology of MDRO and ESBL in LTCFs, including transmission patterns, rates of infection, and factors associated with infections. It may be necessary to extend the requirements and precautions to MDR-GNR and ESBL-producers. Several potential limitations should be considered in the interpretation of data presented here. First, the number of patients included was very low (only 193 residents from 3 LTCFs). This was due to mistrust on the part of residents (who had to agree to participate) and the insufficiently enthusiastic attitude of personnel. Therefore, the number of isolates was relatively low. The control group was also very small, as these patients also had to agree to participate in the study. However, this is the first Polish surveillance conducted in LTCFs, and further research should be done because it might benefit physicians in Poland in the treatment of LTCF patients. In addition, to develop recommendations for treatment, an extended study is recommended.