Clinical and Molecular Epidemiology of Multidrug-Resistant P. aeruginosa Carrying aac(6')-Ib-cr, qnrS1 and blaSPM Genes in Brazil

We described a comprehensive analysis of the molecular epidemiology of multidrug-resistant (MDR) P. aeruginosa. Molecular analysis included typing by Pulsed Field Gel Electrophoresis, identification of genes of interest through PCR-based assays and sequencing of target genes. Case-control study was conducted to better understand the prognostic of patients and the impact of inappropriate therapy in patients with bacteremia, as well as the risk factors of MDR infections. We observed a high rate of MDR isolates (40.7%), and 51.0% of them was independently associated with inappropriate antibiotic therapy. Bacteremia was detected in 66.9% of patients, and prolonged hospital stay was expressive in those resistant to fluoroquinolone. Plasmid-mediated quinolone resistance genes (PMQR), qnrS1 and aac(6’)Ib-cr, were detected in two different nosocomial isolates (5.3%), and the aac(6’)-Ib7 variant was detected at a high frequency (87.5%) in those negative to PMQR. The presence of mutations in gyrA and parC genes was observed in 100% and 85% of selected isolates, respectively. Isolates harboring PMQR genes or mutations in gyrA and parC were not closely related, except in those containing SPM (São Paulo metallo-β-lactamase) clone. In addition, there is no study published in Brazil to date reporting the presence of Pseudomonas aeruginosa isolates harboring both qnrS1 and aac(6’)Ib-cr genes, with alarming frequency of patients with inappropriate therapy.


Introduction
Currently, the increasing incidence of MDR P. aeruginosa is a global problem as a consequence of the ability of this microorganism to develop resistance to almost all antibiotics

Study design and data collection
Active surveillance was conducted from May 2009 to December 2012 and from April to October 2014 for the detection of patients with P. aeruginosa infections resistant to carbapenems and fluoroquinolones at Uberlândia University Hospital (Brazil). In total, 242 episodes of P. aeruginosa infections obtained from 236 patients were included in the study. From this surveillance, two case-control studies were conducted: (i) to determine the risk factors associated with MDR P. aeruginosa infections, and (ii) to determine the risk factors associated with antimicrobial resistance and treatment outcome in patients with P. aeruginosa bacteremia. In both studies, only the first episode of each infection was considered. The demographic, clinical and epidemiological data of the patients were also obtained through review of medical records, following the model of National Healthcare Safety Network (NHSN).

Definitions
According to the Centers for Disease Control and Prevention (CDC), bacteremia was defined as the presence of viable bacteria in the blood, documented by a positive blood culture result [28]. The isolates were considered to be nosocomial if the infection occurred >48 h after admission and no clinical evidence of infection on admission existed [29]. The criteria used for defining MDR phenotype was: non-susceptible to 1 agent in 3 antimicrobial categories [30]. Previous antibiotic use was considered when the patient received therapy with any antibiotic for at least 72 h over a period of 30 days prior to the microbiological infection diagnosis [31]. The antimicrobial therapy was considered to be appropriate if the initial antibiotics, which were administered within 24 h of acquisition of a blood culture sample, included at least one antibiotic that was active in vitro [32]. The 30-day mortality was considered as the number of deaths of patients with infections during hospitalization that occurred within 30 days of the diagnosis of infection [33], and the 5-day mortality, also known as early mortality, was considered as the number of deaths within 5 days of hospitalization [34]. Hospital stays were considered prolonged if they reached or exceeded 45 days [35]. It was considered MIC50 and MIC90 represent the concentration of antimicrobial agent (μg/mL) that inhibited 50% and 90%, respectively, of the isolates tested [36].

Clinical microbiological and antibiotic resistant profile
Microbial identification and antimicrobial susceptibility tests were performed on a VITEK II system (bioMérieux, Brazil) for the following antimicrobials: aminoglycoside (gentamicin, amikacin), carbapenems (imipenem, meropenem), cephalosporin (cefepime), fluoroquinolone (ciprofloxacin) and penicillin plus β-lactamase inhibitors (piperacillin-tazobactam). Qualitycontrol protocols were used according to the standards of the Clinical and Laboratory Standard Institute [37,38]. The isolates with intermediate susceptibility were considered as resistant.
The minimum inhibitory concentration (MIC) and the confirmation test of resistance to imipenem (8 μg/mL) were performed by the E-test 1 method, according to the manufacturer's guidelines (AB Biodisk, Sweden) [38]. In addition, resistance to ciprofloxacin was confirmed by broth microdilution method according to Capuano [39] with modifications, and the interpretations also were made according to CLSI [38], considering resistance to ciprofloxacin 4 μg/mL.

Characterization of strains harboring MBL and PMQR genes
Forty clinical P. aeruginosa fluoroquinolone-resistant isolates were selected, being obtained from 39 patients, with various clinical infections (urinary infection, pneumonia, wound infection, otitis, and bloodstream infection).

Pulsed-Field Gel Electrophoresis (PFGE)
Isolates were typed according to the protocols described by Galetti [40] with modifications, following digestion of genomic DNA with SpeI restriction enzyme (Promega). DNA fragments were separated on 1% (w/v) agarose gels in 0.5x TBE [Tris-borate-ethylene diamine tetra-acetic acid (EDTA)] buffer using a CHEF DRIII apparatus (Bio-Rad, USA) with 6 V/cm, pulsed from 5 s to 40 s, for 21 h at 12°C. Gels were stained with ethidium bromide and photographed under ultraviolet light. Computer-assisted analysis was performed using BioNumerics 5.01 software (Applied Maths, Belgium). Comparison of the banding patterns was accomplished by the unweighted pairgroup method with arithmetic averages (UPGMA) using the Dice similarity coefficient.

Statistical analysis
The Chi-square or Fisher's exact test was used to compare discrete variables. The comparison of two quantitative variables was made using the Mann-Whitney test for nonparametric variables and the Student t test for parametric variables. Two-sided tests were used for all analyses. Multivariate analysis was performed using multiple logistic regression and the values were included when significance was <0.05 in univariate analysis. To determine inappropriate therapy for mortality within 30 days of hospitalization, a multiple logistic regression model was used to control for the effects of confounding variables. All p-value <0.05 was considered statistically significant. The epidemiological data were analyzed through the programs Graph Pad Prism 1 5.0 (La Jolla, USA) and BioEstat 5.0 (Tefé, Brazil).

Ethical considerations
The data and the samples analyzed in the present study were obtained in accordance with the norms and approved by the Federal University of Uberlandia Ethics Committee (UFU), through license number 36601814.7.0000.5152. For this study, samples were collected at the Microbiology Laboratory of the Clinical Hospital, with no contact to the patient and with the permission of the Hospital. Moreover, this study was retrospective and there are no patient identification when performed data collection, so the ethics committee dismissed the informed consent term and clarified.

Results
From May 2009 to December 2012 and from April to October 2014, a total of 236 non-repetitive patients with P. aeruginosa infections at the University Hospital were included in the study. The univariate analysis and independent risk factors associated with MDR P. aeruginosa infections are summarized in Table 1. According to antimicrobial susceptibility testing results, MDR P. aeruginosa infections occurred in 40.7% of the cases. Data from these patients (MDR) were compared with a sensitive P. aeruginosa infections group (non-MDR). In the whole series, prior exposure to carbapenems and inappropriate therapy as well as the co-morbidity condition (diabetes mellitus) were significant in the univariate analysis by MDR P. aeruginosa infections. The results of multivariate analyses showed that factors independently associated with MDR P. aeruginosa were patients who received inappropriate therapy. The Kaplan-Meier cumulative survival estimates (Fig 1) for patients with inappropriate versus appropriate therapy showed that the first group had a lower probability of survival than the group that received appropriate therapy (P = 0.0047). The 30-day mortality rate of the first group was 55.3%, whereas that of the second group was 34.4%. Furthermore, of the total mortality, 5-day mortality rate was 40.6% (54/133) independently of the therapy received. Antimicrobial therapy and clinical outcome of patients with or without bacteremia caused by P. aeruginosa were evaluated and it can be observed that patients with bacteremia caused by isolates resistant to carbapenems, had a high 5-day mortality rate. Moreover, the time of hospital stay was significantly higher for the MDR and fluoroquinolone-resistant groups when compared with susceptible group, and the latter was also independently associated in the multivariate analyses ( Table 2).

Discussion
Recent studies have shown that multi-drug resistance and several virulence determinants are key factors that contribute to the global spread of P. aeruginosa in hospitals [41][42][43]. The present study evaluated the risk factors for the development of infections caused by MDR P. aeruginosa, as well as those associated with antimicrobial resistance and treatment outcome in patients with bacteremia. The independent risk factors for development of MDR P. aeruginosa infections include prior use of antibiotics (carbapenems, fluoroquinolones, broad-spectrum cephalosporins, and aminoglycosides), being bedridden or in the intensive care unit, prolonged hospital stay, P. aeruginosa infection or colonization within a previous period of one year, malignant disease, mechanical ventilation, and history of chronic obstructive pulmonary disease [44,45]. Data of univariate analysis from this study have corroborated some of these risk factors. However, only inappropriate therapy was a risk predictor independent associated for developing MDR P. aeruginosa infections, but it must be considered as such within the epidemiological context [46]. Some studies provide evidences of a general view that the development of MDR can be caused by treatment that is inappropriate, incorrect and widespread use of carbapenems [3,46]. The pressure of carbapenems use has contributed to an explosive increase of KPC (carbapenemase-producing K. pneumoniae), which has been responsible for 70.9% of K. pneumoniae infections in the hospital of this study (data not shown).
Among patients with bacteremia (158 patients), 44.3% had isolates resistant to carbapenems, 42.4% resistant to fluoroquinolones and 42.4% with a multidrug-resistant profile. The association of bacteremia with antimicrobial resistant isolates is common, but few studies have addressed this problem systematically [47][48][49]. In hospitalized patients, the association between bacteremia and fluoroquinolone-resistant P. aeruginosa was observed in our study with a high frequency of patients remained hospitalized longer (56.7%), independently associated with bacteremia. Similar frequencies were also observed for those patients with MDR isolates (58.2%). In the carbapenem-resistant group, was observed a higher early mortality rate (5 days) that was statistically significant with those that had bacteremia infections. These aspects have often been observed in developing countries like Brazil [48,50,51], where the macro and micro-regional differences in relation to the hospitals are extremely significant, as well as the characteristics of the hospitalized population [29,52]. Besides, the lack of the microbiology laboratories and human and financial resources, a result of poor implementation of control practices for prevention of nosocomial infections, favors the intra and inter-hospital transmission of resistant pathogens that exhibit adaptation to the environment [6,52]. Another aspect that must be considered is that the antimicrobial use is commonly abusive, empirical and often less judicious [6,52]. A study performed by Dantas [53] in the same university hospital of this study, showed that the density of antibiotic use was much higher when compared with hospitals with similar size in other countries, allowing MDR strains such as Gram-negative non-fermentative bacteria to emerge and spread quickly [54].
In addition to the multi-drug resistance, special attention was given in this study to P. aeruginosa resistant to carbapenems, considering the significant increase of this resistance in Latin America and widespread of different clones associated with the production enzymes of the MBL type [55][56][57][58]. The predominant MBL-encoding gene in Brazil is bla SPM-1 , which has been disseminated by the MDR P. aeruginosa clone SP/ST277, considered a high-risk clone [7,18,59,60]. The bla SPM-1 gene was first detected in São Paulo, and later in others cities in Brazil, moreover several studies have shown its global spread and pandemic potential, causing important morbidity and mortality in hospital infections [18,43,60,61]. In our study, we observed a high frequency of SPM among the isolates, especially those belonging to clone A. In addition, two isolates harboring bla VIM were detected. An increase in the rates of P. aeruginosa isolates containing the gene bla VIM has been also observed in others hospitals in Brazil [62,63]. The frequent data among different types of metallo-β-lactamase in Brazil, and not in other countries, especially those developed, suggest the spread of specific clones and the knowledge of these facts may contribute to improving the multidrug resistance scenario.
Besides the carbapenem resistance, resistance to fluoroquinolones has become an increasing problem, so far, only a few studies have investigated the occurrence of PMQR in P. aeruginosa [27,29,64,65]. PMQR is an important phenomenon that is being disseminated worldwide and the most relevant PMQR genes to date are the aac(6')-Ib-cr, qnr and genes encoding efflux pumps such as qepA [24]. Surprisingly, the results from this study demonstrated for the first time the presence of PMQR genes in clinical isolates of P. aeruginosa in Brazil (5.3%), as well as a very significant high frequency not shown in other studies of the aac(6')-Ib 7 variant. The PMQR rate in our study was higher than that some reported in the literature [27,65]. According to the study reported by Jiang et al. [27], the frequency of clinical isolates carrying the aac (6')-Ib-cr gene was 1.9% (2/106 isolates), and the total rate of PMQR determinants was 3,8% among P. aeruginosa isolates, while in the Yang and colleagues [65] study, only one in 256 P. aeruginosa isolates (0.4%) showed a PMQR gene. The higher detection frequency of these genes in our study reflects most likely the complexity of epidemiology and resistance mechanisms associated with P. aeruginosa in developing countries. The co-existence of different PMQR genes in the same clinical isolate was not observed throughout this study, although it has been reported by Jiang et al. [27].
Regarding the target site mutations in QRDR of fluoroquinolone resistant P. aeruginosa, we observed mutations consistent with those published previously in all isolates tested [66][67][68]. Of total, 20 Pseudomonas aeruginosa of nosocomial origin were evaluated for mutations in gyrA and parC, and none of them presented PMQR genes. However, according to literature evidences, the chromosomal QRDR mutations in gyrA and parC genes are crucial to fluoroquinolone resistance, and their association with PMQR determinants may have an additional role that contributes to resistance to fluoroquinolones [26].
Another interesting observation from this study was the identification of a mixed wound infection by P. aeruginosa and E. coli in the patient who presented P. aeruginosa harboring the PMQR gene aac(6')-Ib-cr. In a previous evaluation of E. coli infections in the same hospital, 71.4% had aac(6')-Ib-cr gene (data not shown). The epidemiology of P. aeruginosa in this hospital proved to be complex and can be explained by its involvement in the propagation through transference of plasmids to different species as well as by clonal spread.
PFGE results, in this study, suggested that P. aeruginosa isolates harboring PMQR genes or mutations in gyrA and parC were not closely related, except in those containing SPM clone, wherein in the literature showing results with a similarity clonal among isolates of MDR P. aeruginosa [69].
In conclusion, our results confirm previous findings regarding the dissemination of SPMtype clones in Brazil, and contribute whit additional evidence that may indicate that inappropriate therapy may be a crucial factor to the emergence of MDR isolates, besides being related to worse prognosis. Additionally, this study demonstrates for the first time in Brazil the presence of the PMQR determinants in P. aeruginosa, spreading in the hospital. Future follow-up surveillance studies of molecular epidemiology in Brazilian hospitals have crucial importance to infection-control practices and reduce the effects of these infections on hospital patients.
Supporting Information S1 Table. Primer nucleotide sequences and amplicon sizes of the PCR carried out for the detection and/or sequencing of antimicrobial resistance genes in this study.