Characterization of Pseudomonas aeruginosa isolates obtained from patients in Canadian hospitals: Results of the CANWARD 2007 study

1Departments of Medicine and Clinical Microbiology, Health Sciences Centre; 2Department of Medical Microbiology, Faculty of Medicine, University of Manitoba; 3Nosocomial Infections Branch, National Microbiology Laboratory, Winnipeg, Manitoba Correspondence: Dr Andrew Walkty, Health Sciences Centre, Department of Clinical Microbiology, MS673-820 Sherbrook Street, Winnipeg, Manitoba R3A 1R9. Telephone 204-453-3867, fax 204-787-4699, e-mail awalkty@mts.net Pseudomonas aeruginosa is an important cause of nosocomial bloodstream, urinary tract, wound and respiratory infections (1). In recent years, P aeruginosa clinical isolates resistant to multiple classes of antimicrobial agents have become increasingly common (2). Inappropriate initial antimicrobial therapy for infections caused by P aeruginosa has been independently associated with increased mortality (3). It is therefore critical that clinicians have access to current susceptibility CANWARD 2007


Bacterial isolates
Twelve sentinel hospital sites located in major population centres in 7 of the 10 provinces in Canada participated in the Canadian Ward Surveillance Study (CANWARD 2007). These sites were geographically distributed in a population-based fashion. From January through December 2007, inclusive, each study site was asked to submit clinical isolates (consecutive, one per patient per infection site) from inpatients and outpatients with respiratory (n=200), urine (n=100), wound/intravenous (n=50) and bloodstream (n=360) infections. Hospital clinic isolates were subsequently excluded from the current analysis, due to the variable nature of where the isolates came from and their vastly different antimicrobial susceptibility profiles.
Only isolates that were deemed clinically significant were submitted. Limited demographic information was collected for each isolate (patient age, patient sex, hospital ward, specimen site, region of Canada). Isolate identification was performed by the submitting site using local criteria. Where indicated, identification was confirmed at the reference site. Isolates were shipped on Amies semi-solid transport media to the coordinating laboratory (Health Sciences Centre, Winnipeg, Manitoba), where they were then subcultured on appropriate media, and stocked in skim milk at -80°C.

Antimicrobial susceptibilities
Following two subcultures from frozen stock, the in vitro activity of common antipseudomonal antimicrobials was determined by broth microdilution in accordance with the Clinical and Laboratory Standards Institute guidelines (4,5). For several antimicrobial classes (eg, carbapenems, cephalosporins), the in vitro activity of a single representative antimicrobial was assessed due to limited space on the susceptibility panels. Colistin (polymyxin E) susceptibility was evaluated for multidrug-resistant (MDR) isolates. Minimum inhibitory concentration interpretive standards for all antimicrobials were defined according to Clinical and Laboratory Standards Institute breakpoints (4). MDR P aeruginosa isolates were defined as isolates demonstrating resistance to antimicrobials from three or more different classes. The number that appears in the following text and tables after the MDR designation indicates the number of different classes to which isolates were resistant (eg, MDR3 indicates P aeruginosa isolates that were resistant to at least one antimicrobial agent from three different classes). For the purpose of the present report the four antimicrobial classes considered were aminoglycosides (amikacin, gentamicin), fluoroquinolones (ciprofloxacin, levofloxacin), cefepime and piperacillin/tazobactam (considered together as one class), and carbapenems (meropenem). Colistin was not used in the classification of MDR isolates.

Pulsed-field gel electrophoresis
The genetic relationships among MDR P aeruginosa isolates were assessed by pulsed-field gel electrophoresis, developed from a previously described method (6). This method was altered in the following ways; restriction of samples used 40 units of SpeI and switch times were changed to 5.3 s to 34.9 s with a run time of 21 h. Fingerprints were analyzed using BioNumerics version 3.5 software (Applied Maths Inc, USA).

Statistical analysis
Statistical analysis was performed using JMP software, version 7.0 (SAS Institute Inc, USA). A logistic regression model was used to determine whether any of the baseline demographic variables recorded in the present study (specimen source, hospital ward, patient age, patient sex, region of Canada) were associated with either the MDR phenotype or individual antimicrobial resistance (association assessed for hospital ward only). A P of ≤0.05 was considered significant.

ReSultS
In total, 451 P aeruginosa isolates were collected as a part of CANWARD ( . P aeruginosa accounted for 7% (451 of 6471) of all isolates, making it the fourth most common organism obtained from patients in Canadian hospitals (third most common organism from medical wards, fourth from surgical wards and ICUs, fifth from emergency rooms). Susceptibility data for common antipseudomonal antimicrobials are presented in Table 1. The rank order of antimicrobial susceptibility was as follows (percent susceptible): amikacin (93.1%) = piperacillin/tazobactam (93.1%) > meropenem (87.4%) > cefepime (69.4%) > ciprofloxacin (67.2%) > gentamicin (66.1%) > levofloxacin (60.5%). Table 2 presents a breakdown of antimicrobial susceptibility by hospital ward type. Isolates demonstrating reduced susceptibility to cefepime, meropenem and levofloxacin were more frequently obtained from patients in an ICU setting (P<0.05 in all cases).
Thirty-four MDR P aeruginosa isolates were collected as a part of CANWARD (7.5% of all P aeruginosa isolates). The MDR isolate demographics are presented in Table 3. MDR isolates were more commonly obtained from patients in an ICU setting (P=0.003, Table 3) and less commonly obtained from a bloodstream source of infection (P=0.008, Table 3). Excluding colistin, amikacin and piperacillin/tazobactam, followed by meropenem, were the most active antimicrobials evaluated versus the MDR isolates ( Table 1). As a class, fluoroquinolones demonstrated the lowest susceptibility rates against the MDR isolates. None of the 34 MDR isolates were fully susceptible to levofloxacin and only one (2.9%) was fully susceptible to ciprofloxacin. All of the MDR isolates were susceptible to colistin. The various combinations of antimicrobial class resistance that contributed to the MDR phenotype are presented in Table 4. Among MDR3 isolates, resistance was most frequently observed to the combination of fluoroquinolones, aminoglycosides and piperacillin/cefepime (Table 4).
A pulsed-field gel electrophoresis dendrogram for the MDR isolates is provided (Figure 1). Although the majority of MDR isolates were genetically unrelated, three small clusters of related isolates (80% or greater homology) were observed. This included one cluster consisting of two isolates, one cluster containing three isolates, and a larger cluster of six genetically related isolates. Isolates in the latter cluster were collected from bloodstream, respiratory and wound specimens, and were all isolated from patients at a single institution in Ontario (data not shown).

DISCuSSIoN
The data presented here serve to confirm the continued importance of P aeruginosa as a pathogen in nosocomial infections. These data are in agreement with several previously published surveillance studies describing the frequency of occurrence of P aeruginosa among clinical samples obtained from patients in a hospital setting (1,7-9). Among the antipseudomonal antimicrobial agents evaluated here, amikacin, meropenem and piperacillin/tazobactam were the most active, while the fluoroquinolones were the least active. These results are consistent with surveillance data from the United States (10). Variation in the definition of MDR precludes a   None of the MDR isolates evaluated in the current study were resistant to colistin, suggesting that this agent may have a role in the treatment of infections caused by MDR P aeruginosa. This observation is supported by an increasing number of reports demonstrating clinical efficacy of colistin in the treatment of MDR P aeruginosa infections (11).
It is interesting to note that MDR isolates and isolates with reduced susceptibility to certain antimicrobials (cefepime, levofloxacin, meropenem) were more commonly obtained from patients in an ICU setting. One may hypothesize that this observation could relate to higher levels of antimicrobial consumption and/or differences in the specific antimicrobials used in ICUs relative to other hospital locations.
There are several limitations to the data presented here. First of all, limited demographic information was collected on the patients from whom the isolates were obtained. As a consequence of this, it was not possible to assess in detail specific variables associated with the acquisition of MDR P aeruginosa. Additionally, we do not know how many isolates were obtained from patient groups at higher risk of infection with resistant P aeruginosa (eg, cystic fibrosis patients). We cannot exclude the possibility that some of the resistance observed in the present study was being driven by isolates from very specific patient populations. Susceptibility testing was not performed for ceftazidime, tobramycin and imipenem due to lack of space on the susceptibility panels utilized. It is recognized that this data would be beneficial, because these antimicrobials may be a part of many hospital formularies. Up to 2.5% of isolates in the present study may have been submitted from duplicate patients (ie, two different isolates submitted from the same patient). Due to the rarity of this event, it is not believed that inclusion of these isolates had a significant impact on the susceptibility data described here. Finally, we did not investigate the molecular mechanisms conferring antimicrobial resistance among our isolates due to limited time and resources.

CoNCluSIoNS
P aeruginosa is commonly isolated from patients in Canadian hospitals. Of the antimicrobials evaluated, amikacin, meropenem and piperacillin/tazobactam were the most active, while the fluoroquinolones were the least active. Thirty-four isolates (7.5%) were MDR. Isolates with reduced susceptibility to individual antimicrobials and MDR isolates were more frequently obtained from patients in an ICU setting. All of the MDR isolates remained susceptible to colistin.