|Home | About | Journals | Submit | Contact Us | Français|
Pseudomonas aeruginosa is a major cause of morbidity and mortality in cystic fibrosis patients. This study compared the antimicrobial susceptibilities of 153 P. aeruginosa isolates from the United Kingdom (UK) (n = 58), Belgium (n = 44), and Germany (n = 51) collected from 118 patients during routine visits over the period from 2006 to 2012. MICs were measured by broth microdilution. Genes encoding extended-spectrum β-lactamases (ESBL), metallo-β-lactamases, and carbapenemases were detected by PCR. Pulsed-field gel electrophoresis and multilocus sequence typing were performed on isolates resistant to ≥3 antibiotic classes among the penicillins/cephalosporins, carbapenems, fluoroquinolones, aminoglycosides, and polymyxins. Based on EUCAST/CLSI breakpoints, susceptibility rates were ≤30%/≤40% (penicillins, ceftazidime, amikacin, and ciprofloxacin), 44 to 48%/48 to 63% (carbapenems), 72%/72% (tobramycin), and 92%/78% (colistin) independent of patient age. Sixty percent of strains were multidrug resistant (MDR; European Centre for Disease Prevention and Control criteria). Genes encoding the most prevalent ESBL (BEL, PER, GES, VEB, CTX-M, TEM, SHV, and OXA), metallo-β-lactamases (VIM, IMP, and NDM), or carbapenemases (OXA-48 and KPC) were not detected. The Liverpool epidemic strain (LES) was prevalent in UK isolates only (75% of MDR isolates). Four MDR sequence type 958 (ST958) isolates were found to be spread over the three countries. The other MDR clones were evidenced in ≤3 isolates and localized in a single country. A new sequence type (ST2254) was discovered in one MDR isolate in Germany. Clonal and nonclonal isolates with different susceptibility profiles were found in 20 patients. Thus, resistance and MDR are highly prevalent in routine isolates from 3 countries, with meropenem, tobramycin, and colistin remaining the most active drugs.
Pulmonary infection represents a major cause of morbidity and mortality among cystic fibrosis (CF) patients (1). These patients are therefore regularly exposed to antibiotics for the treatment of infectious exacerbations as well as for the prevention of chronic colonization. Pseudomonas aeruginosa is one of the most prevalent bacterial species, especially in the adult population (2). It is well known for its genetic plasticity and capacity to accumulate resistance mechanisms, including acquisition of foreign genetic material (3). The percentage of patients colonized by P. aeruginosa has decreased in recent years (2), but with improved life expectancy, the absolute number of colonized patients has increased. It has also been proposed that multidrug-resistant (MDR) strains are more frequent in older patients, primarily due to cumulative exposure to antibiotics (2). A further reason for the spread of antibiotic resistance in CF patients is the dissemination of MDR clones. The Liverpool epidemic strain (LES), first described in 1996 (4), has proven particularly successful at acquiring resistance mechanisms over the years (5, 6) and at spreading from the United Kingdom (UK) to other countries, such as Canada, Spain, and Australia (7).
In this study, we compared the antimicrobial susceptibility of P. aeruginosa isolated from CF patients in the UK, where the MDR LES clone is known to be highly prevalent (5), with those of an equivalent number of strains collected in Germany and Belgium, where no specific survey has been published in recent years. We determined the presence of coresistance to unrelated antibiotic classes and its possible association with MDR clones. We found that resistance was high in the three countries but was not related to the dissemination of a specific MDR clone in Germany or Belgium. Carbapenems, tobramycin, and colistin remain the drugs most active against P. aeruginosa respiratory isolates. Importantly, no carbapenemases were detected in these strains.
A total of 153 clinical P. aeruginosa isolates were selected at random among those collected between 2006 and 2012 in 3 CF centers in Belgium (Hôpital des enfants malades Reine Fabiola/Erasme Hospital; n = 44), Germany (University Hospital of Münster; n = 51), and the UK (Queen's University of Belfast; n = 58) during routine visits. The details on the collection are shown in Table 1. When successive strains were collected from a single patient, only those collected at the first occasion were considered. Nevertheless, more than one isolate were analyzed for some patients based on differences in their phenotypic appearance (see Fig. S1 in supplemental material).
The following antibiotics were obtained as microbiological standards (with abbreviations and potencies shown in parentheses): amikacin disulfate (AMK; 74.80%), colistin sulfate (CST; 79.64%), piperacillin sodium (PIP; 94.20%), and ticarcillin disodium salt (TIC; 85.25%) from Sigma-Aldrich, St. Louis, MO; ciprofloxacin (CIP; 85.00%) from Bayer, Leverkusen, Germany; and tobramycin (TOB; 100%) from Teva, Wilrijk, Belgium. The remaining antibiotics were obtained as the corresponding branded product in Belgium for intravenous use and complied with the provisions of the European Pharmacopoeia with respect to content in active agent: ceftazidime as Glazidim (CAZ; 88.20%) from GlaxoSmithKline, Genval, Belgium; imipenem as Tienam (also containing cilastatin, which does not have any antibacterial activity) (IPM; 45.60%) from MSD, Brussels, Belgium; meropenem as Meronem (MEM; 74.00%) from AstraZeneca, Brussels, Belgium; and piperacillin-tazobactam as Tazocin (TZP; 97.00%) from Wyeth, Louvain-La-Neuve, Belgium (now part of Pfizer).
MICs were determined by microdilution in cation-adjusted Mueller-Hinton broth by following Clinical and Laboratory Standards Institute (CLSI) recommendations, using P. aeruginosa ATCC 27853 as a quality control strain (8). Susceptibility was assessed according to the interpretive criteria of both the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (9) and the CLSI (8). Isolates were considered multidrug resistant (MDR) if they were resistant to at least three antibiotic classes among those tested (penicillins/cephalosporins, carbapenems, fluoroquinolones, aminoglycosides, and polymyxins), according to European Centre for Disease Prevention and Control (ECDC) criteria (10).
For all isolates (n = 51) showing MICs of >8 mg/liter for ceftazidime and meropenem, the blaTEM, blaSHV, and blaCTX-M (groups 1, 2, and 9), blaVIM, blaIMP, blaKPC, and blaNDM gene families were detected by real-time multiplex PCR, using group-specific primers (references 11, to ,13 and references therein). Genes encoding OXA (OXA-1, -2, -9, -10, -18, -20, -23, -24, -30, -48, -58, and -198), BEL (BEL-1 to -3), PER (PER-1 to -5 and -7), GES (GES-1 to -18), and VEB (VEB-1 to -7) enzymes were also detected by multiplex PCR.
All MDR isolates in the collection showing coresistance to penicillins and/or cephalosporins and two other classes (n = 56) were characterized by pulsed-field gel electrophoresis (PFGE) analysis (14). In addition, 40 pairs of isolates collected simultaneously and in the same sample from 20 patients (see Fig. S1 in the supplemental material) but differing in their profiles of susceptibility to at least one class of antibiotics were also genotyped by PFGE to determine their genetic relatedness. The pulsotype classification criteria designated a pulsotype by one or two letters, including patterns showing zero to six DNA fragment differences (14). An epidemic pulsotype was defined as a pulsotype recovered from ≥2 patients, while a sporadic pulsotype was recovered only once.
Multilocus sequence typing (MLST) was performed on a representative strain of epidemic pulsotypes detected in ≥3 strains, as previously described (15). The reference LES B58 strain (4) was used as a control. MLST data were uploaded to the P. aeruginosa MLST Database (http://pubmlst.org/paeruginosa) for allele type and sequence type (ST) assignments (16).
Table 2 shows the MIC distribution for 9 antipseudomonal drugs against 153 isolates collected from 118 CF patients originating from three different countries over the period from 2006 to 2012, together with the percentages susceptible and resistant based on both EUCAST and CLSI interpretive criteria. The corresponding MIC cumulative distributions are illustrated in Fig. S2 in the supplemental material. Resistance was high in this collection. Using the EUCAST or the CLSI resistance breakpoints, respectively, the rates of full resistance for the isolates were ≥71% or ≥54% for penicillins (ticarcillin, piperacillin, and piperacillin-tazobactam), 69% or 59% for ceftazidime, 61% or 46% for amikacin, 56% or 27% for ciprofloxacin, ≥20% for carbapenems, and 28 or 16% for tobramycin. Full resistance to colistin was noted for only 8% of the isolates. Strains resistant to ceftazidime and meropenem were screened for the expression of frequent ESBLs, metallo-β-lactamases, and carbapenemases, which returned negative results.
Cross-resistance or coresistance was examined among pairs of antibiotics. Cross-resistance is defined as the presence of a single resistance mechanism that confers resistance to antimicrobial molecules with a similar mechanism(s) of action. It thus describes resistance to an entire class of antibiotics, to different classes of agents with overlapping drug targets, or to different classes of antibiotics that are substrates for the same broad-spectrum efflux system. Coresistance refers to the presence of different mechanisms of resistance in the same bacterial isolate and thus necessarily confers resistance to unrelated antibiotic classes (17). Ninety-four strains were considered MDR according the ECDC (10). The upper right part of Table 3 shows the percentage of strains showing cross-resistance or coresistance to pairs of antibiotics according to EUCAST criteria. About two-thirds of the strains were resistant to both penicillins and ceftazidime and more than 40% were resistant to penicillins and ceftazidime together with amikacin or ciprofloxacin. The rates of coresistance between any studied drug and tobramycin, meropenem, and colistin were lower than 28%, 20%, and 8%, respectively. Of note, only 4 strains in the whole collection were coresistant to meropenem, tobramycin, and colistin (see Fig. S3 in the supplemental material).
The lower left part of Table 3 shows the correlation coefficient between the individual MIC for each pair of antibiotics, with the corresponding multivariate analysis presented in detail in Fig. S4 in the supplemental material. The highest degrees of correlation (>0.75) between individual MICs were observed for ticarcillin versus ceftazidime, piperacillin versus piperacillin-tazobactam, ceftazidime versus piperacillin-(tazobactam), imipenem versus meropenem, and amikacin versus tobramycin, suggesting common mechanisms of resistance between these pairs of antibiotics. Yet differences in the intrinsic potency were nevertheless observed between these pairs of drugs throughout the collection (illustrated in Fig. S4 and associated Table B in the supplemental material): tazobactam reduced the MIC of piperacillin by a factor of 1.5 dilution, while ceftazidime MICs were 0.5 and 1 dilution lower than those of ticarcillin and piperacillin, respectively, and similar to those of piperacillin-tazobactam. Meropenem MICs were 1 dilution lower than those of imipenem, and tobramycin MICs were 3 dilutions lower than those of amikacin.
Among the 94 MDR isolates, most were resistant to penicillins and/or cephalosporins. Only those showing resistance to at least 2 other classes (n = 56) were characterized by PFGE analysis. A high genetic diversity was observed, with 19 sporadic pulsotypes and 9 epidemic pulsotypes (Table 4). With the exception of pulsotype YY recovered for 1 or 2 isolates in the three countries, each epidemic pulsotype remained localized in a single country. The CA epidemic pulsotype found in 3/4 of the UK isolates corresponded to the pulsotype of the LES clone. MLST analysis of epidemic pulsotypes CA, H, and YY showed ST146, ST2254 (new ST), and ST958, respectively (data not shown).
PFGE analysis was also performed on 40 isolates collected as pairs from 20 patients and displaying different susceptibility profiles (see Table S1 in the supplemental material). In 12 patients, the pair of P. aeruginosa isolates had the same pulsotype, while the 8 other patients had isolates with different pulsotypes.
Because of the genetic diversity observed between countries, we then examined the distribution of susceptible, intermediate (when applicable), and resistant isolates classified based on the country where they were collected (Fig. 1). Susceptibility rates differed among countries, with lower resistance in Belgium (significant for all antibiotics except ticarcillin and ciprofloxacin) and higher resistance in Germany and the UK (significant for piperacillin-tazobactam in Germany and for imipenem, ciprofloxacin, and colistin in the UK) than the mean value for the whole collection. There was no significant correlation between the patient's age when the isolate was collected and the number of antibiotic classes to which the isolate was resistant (see Fig. S5 in the supplemental material).
In this study, we examined antibiotic susceptibility within a collection of P. aeruginosa isolates from CF patients in three northern European countries collected during routine examination, which provides a broader view than the majority of previous surveys, which have focused on a single country (18,–20) or a single center (21,–23). A key observation is that resistance rates were high in this population, confirming previous studies with CF patients (2), and notably much higher than that which has been reported for isolates collected in northern Europe from intensive care units (24,–26). Resistance rates were also higher than those previously reported for strains from CF patients in a German survey from the University of Würzburg, except in the case of tobramycin (isolates collected in 2006 ), or in a multicentric study in the UK, except for meropenem and ciprofloxacin (isolates collected in 2000). Moreover, a high degree of cross-resistance or coresistance among antibiotics was observed, which is important from both a pharmacological and a clinical perspective.
From a pharmacological perspective, we noticed, as expected, significant correlations between MICs for antibiotics belonging to the same or similar classes (penicillins and ceftazidime or other penicillins, imipenem and meropenem, and amikacin and tobramycin), but with systematic differences in the potency of each antibiotic within these pairs (see Fig. S4 and associated Table B in the supplemental material). Focusing on β-lactams, the impact of tazobactam on piperacillin activity was modest but of the same order of magnitude as that observed on MIC distribution for wild-type strains reported by EUCAST (29), probably denoting the inhibition by tazobactam of the low basal levels of AmpC produced by the wild-type strains (30, 31). Likewise, a higher potency of ceftazidime than for penicillins and of meropenem than for imipenem is reported in wild-type EUCAST distributions (29). Thus, differences in potency among these pairs of drugs in our collection are likely to reflect differences in intrinsic activity rather than in vulnerability to resistance mechanisms. Remarkably, no carbapenemase production was apparent in this collection. The same finding was reported in two recent investigations of P. aeruginosa isolates collected over the same period as those examined here. The first of these studies was performed in Australia and examined successively a collection of 662 carbapenem-resistant isolates assembled in 2007 to 2009 from diverse CF centers and of 517 isolates collected in a single CF center in 2011 (32). The second study was performed in Brazil and analyzed isolates from 75 patients collected from 2010 to 2011 (19). In contrast, carbapenemases have been detected in 63 out of 217 P. aeruginosa isolates collected from CF patients in China (22). The prevalence of carbapenemase genes could, however, be different in other bacteria infecting CF patients, but there is no large survey published so far for other Gram-negative species (33, 34).
Thus, carbapenem resistance in CF European isolates is probably primarily mediated by the combined effect of AmpC and of a reduced accumulation (porin mutations and/or increased efflux) (35, 53). Of note, however, carbapenem resistance has previously been described for the LES clone (5), but the underlying mechanism(s) has not been investigated to date. For aminoglycosides, the higher potency of tobramycin over amikacin in our collection also reflects what is observed in MIC distributions of wild-type strains assembled by EUCAST (29). Tobramycin has been described as a poorer substrate than amikacin for the efflux pump MexXY-OprM, considered responsible for natural and adaptive resistance to aminoglycosides in P. aeruginosa (36, 37).
Considering our findings from a clinical perspective, a high degree of cross-resistance was observed between penicillins and ceftazidime, which was expected. However, a high degree of coresistance was also apparent between these antibiotics and both ciprofloxacin and amikacin, resulting in 60% of the isolates being categorized as multidrug resistant. In contrast, meropenem, colistin, and, to a lesser extent, tobramycin were active against a large fraction of the isolates, with few strains coresistant to these three antibiotics. Tobramycin and colistin by inhalation are often considered first line for the eradication of early P. aeruginosa infection, and tobramycin is also considered first line for chronic therapies (38,–40). High concentrations delivered by this route of administration may help to overcome resistance (41, 42).
We also noticed an important genetic diversity among multiresistant isolates collected in Belgium and Germany, while those collected in the UK belong in majority to the Liverpool epidemic strain (LES) clone. Global studies of P. aeruginosa population structure concluded that CF isolates present a high genetic diversity but nevertheless belong to a “core lineage” ubiquitous in the natural environment (43), which is highly suggestive of a direct colonization of the patients from the environment. However, a series of epidemic clones have been described (7), among which are the LES clone (4), representing 18 of the 24 MDR isolates collected in the UK in our study, and ST17 (7), which differs by only 1 nucleotide from ST958, found in the three countries we investigated. ST2254, the new ST we describe, was distinct from ST146 (LES clone; 5 alleles different) and ST958 or ST17 (6 alleles different).
We observed that a single patient can be colonized by different strains and, conversely, that clonally related strains isolated at the same time from a single patient can harbor diverse susceptibility profiles. This could be a consequence of the previously described phenotypic variability among isolates with the same colony morphotype and being part of a single clonal lineage (44, 45), as well as of recombination occurring in vivo and generating phenotypic and genetic diversification (46, 47).
Although limited, differences in resistance rates between Belgium and the two other countries are raising questions about segmentation of clone distribution. For strains collected in the UK, higher resistance is clearly related to the high prevalence of the LES clone, which has been described as exhibiting a large proportion of MDR isolates (5). Of interest, we observed different resistance profiles within this clone, which is consistent with the previously described phenotypic variability among LES isolates (6). ST958, represented in the three countries, is also found among the MDR clonal complexes (7). In the German collection, higher resistance is essentially related to the presence of more sporadic MDR clones than in the two other countries. We cannot exclude differences in therapeutic management of patients among these three centers that may influence resistance selection (48), but this specific aspect was not within the scope of our study.
Resistance rates were not higher in the older population than in children and young adults. The interpretation of these data needs to be done with caution because (i) we did not follow the evolution of susceptibility over time in single patients and (ii) we do not know the age of first colonization for each patient. With this limitation in mind, the fact that MDR isolates could be found in young people and susceptible isolates in adults may suggest that resistance depends on the initial susceptibility of the infecting strain. A link between emergence of resistance and early antibiotic use in CF patients is still controversial, even though it was underlined in the last report of the Cystic Fibrosis Foundation (2). A recent study in Australia showed that multiresistance in children is correlated with duration of intravenous antibiotic treatment, which was not the case for adults (18). A correlation with antibiotic usage irrespective of patient age (49) or with time after colonization (6) has also been proposed. In contrast, other studies following the evolution of antibiotic susceptibility in successive isogenic isolates from a single patient suggest that resistance can occur either sporadically (50) or without correlation with the time of isolation (51). In these cases, the presence of mutator variants seems to predetermine the risk of developing resistance over time (6).
Our study has a number of limitations, primarily linked to the fact that samples collected during periodic routine examinations may not correspond to the first P. aeruginosa infections in these patients. Moreover, as we did not have the history of antibiotic use in these patients, we could not determine if there was a potential link between antibiotic usage and subsequent development of resistance. Nevertheless, this collection reflects the situation CF clinicians face daily, where they have to select antibiotics based on susceptibility testing performed on current isolates. In this context, our data may lead to three clinically meaningful conclusions. First, susceptibility testing is important to perform even for newly infected patients, because they can be colonized very early by MDR clones. Second, these tests should be performed on more than one colony (especially if different phenotypes are evidenced on culture plates), because of potential population heterogeneity with respect to susceptibility profiles (52). Third, prudent use of highly active drugs should be promoted in order to preserve their efficacy. This implies the use of optimized doses if administered by conventional routes or administration by inhalation to ensure high local concentrations that could minimize the risk of selection of resistance.
This work was supported by the programs Doctiris of Innoviris (Région bruxelloise) and TEMOEXPAND of the Région wallonne.
M.-H.M. received a PhD fellowship from the Doctiris program of Innoviris and H.C. from the Fonds pour la Recherche dans l'Industrie et l'Agriculture. F.V.B. is senior research associate of the Belgian Fonds de la Recherche Scientifique.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AAC.01046-16.